article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
since their introduction @xcite , topological field theories have been responsible for many applications @xcite and are object of continuous investigations .
nowadays they represent an important chapter of quantum field theory .
the original motivation was related to the possibility of describing topological invariants by means of standard field - theory techniques @xcite . in order to give an idea of this framework ,
let us briefly present here the field - theory characterization of one of the most simple and familiar topological invariants , namely , the linking number @xmath3 of two nonintersecting smooth closed oriented curves in @xmath4@xcite : as is well known , the linking number @xmath5 is an integer which counts the number of times that one curve winds around the other .
it is independent from the shape of the curves and can be represented by the gauss integral @xmath6 expression ( [ g - int ] ) is in fact easily seen to be an integer by use of the stokes theorem @xcite .
taking a field theory point of view , the linking number @xmath5 may be obtained by introducing the topological abelian chern
simons action @xcite @xmath7 and by evaluating the correlation function of two loop variables @xmath8 , _
i .. e . , _
@xmath9 that expression ( [ l - cs ] ) reproduces the linking number follows from the observation that the propagator of the gauge field @xmath10 obtained from the chern
simons action ( [ c - s ] ) upon quantization in the landau gauge is precisely the kernel of the gauss integral ( [ g - int ] ) , _
i.e. , _ @xmath11 the correlator ( [ l - cs ] ) may thus be regarded as a field - theory description of the linking @xmath12 the action ( [ c - s ] ) can be suitably extended to higher dimensions , providing a field - theory characterization of the generalizations of the linking number @xcite . moreover , the nonabelian version of the three - dimensional chern simons action ( [ c - s ] ) has been proven to play a very relevant role in knot theory @xcite . although the topological field theories possess their own interests and applications , it is worth underlining here that topological terms appear frequently as parts of more general effective actions useful for the theoretical description of a large number of phenomena in different space - time dimensions . for instance , the effective action corresponding to the bosonization @xcite of relativistic three - dimensional massive fermionic systems at @xmath13 can be written as the sum of the chern - simons term ( [ c - s ] ) and of an infinite series of higher - order terms in the curvature @xmath14 and its derivatives , _
@xmath15 with @xmath16 being a combination of terms of the type @xmath17 this kind of action turns out to be useful in order to study several three - dimensional phenomena such as the fermi
bose transmutation @xcite and the quantum hall effect @xcite .
a second interesting example is provided by the five - dimensional generalization of ( [ eff - cs ] ) , obtained from the ads / cft _ _ _ _ correspondence @xcite , which relates the conformal @xmath18 super - yang - mills theory to type - iib superstring on ads@xmath19 .
in fact , in the conformal case , the dual supergravity on @xmath20 possesses a chern simons term obtained from a @xmath21 doublet of two - forms @xmath22 @xmath23 . in this case
, the relevant effective action for @xmath22 @xmath23 looks like @xcite @xmath24 where the term @xmath25 collects all the higher - order terms in the curvatures @xmath26 .
the correlation function ( [ l - cs ] ) generalizes now to @xmath27 where @xmath28 @xmath29 are appropriate two - surfaces . in view of these applications
, it seems natural to ask ourselves what is the response of a correlator of the type ( [ l - cs ] ) when the corresponding topological field theory is perturbed by the introduction of a nontopological interaction term depending on the curvature .
this is the aim of the present paper .
more precisely , we shall report on the four - loop computation of the correlator @xmath30 when the three - dimensional chern simons action ( [ c - s ] ) is perturbed by a nontopological interaction term of the kind @xmath31 namely _ _ , _ _ expression ( [ p - c ] ) will be evaluated with an effective action @xmath32 given by @xmath33 with @xmath34 and @xmath35 being an arbitrary parameter with negative mass dimension , reflecting the power - counting nonrenormalizability of the perturbation .
in particular , we shall be able to prove that the correlation function ( [ p - c ] ) turns out to be independent from @xmath35 , yielding the linking number @xmath5 of the two curves @xmath1 , @xmath36 although the loop analysis will be worked out only up to the fourth order , this conclusion holds to all orders of perturbation theory and may be easily generalized to any local nontopological interaction term containing arbitrary powers of the curvature @xmath14 as well as to the higher - dimensional cases @xcite as , for instance , the effective action of eq.([ads ] ) .
this result means that the loop correlator ( [ p - c ] ) is stable with respect to the perturbations which can be added to the starting topological action .
in other words , the expression ( [ p - c ] ) will give the linking number of the two curves @xmath1 , @xmath2 , regardless of any @xmath37-dependent perturbation term that can be introduced and of their power - counting nonrenormalizability character .
two remarks are now in order .
first , we will limit here ourselves only to effective actions which are abelian .
second , we shall consider only @xmath37-dependent terms which can be treated as true perturbations .
therefore , we shall avoid in the effective action ( [ seff ] ) the inclusion of a term of the maxwell type @xmath38 where @xmath39 is a mass parameter .
the presence of this term would completely modify the original properties of the model .
in fact , being expression ( [ max ] ) quadratic in the gauge fields , it can not be considered as a perturbation term , as it will be responsible for the presence of massive excitations in the spectrum of the theory @xcite .
rather , the presence of the maxwell term in the effective action ( [ seff ] ) will give rise to the existence of two distinct regimes corresponding to the long and short distance behaviours , respectively . for distances larger than the inverse of the mass parameter @xmath39 ( _
i.e. _ , the low - energy regime ) , the topological term will prevail , _ _
_ _ while the maxwell term will become the relevant one at short distances ( _ i.e. , _ the high - energy regime ) . it is worth mentioning here that these two regimes can be accessed in a very simple way by means of suitable gauge - invariant field redefinitions of the gauge connection @xmath10 @xcite .
however , their full understanding is a difficult and delicate task , which is beyond the aim of the present paper , being under investigation .
we should also underline here that , in the abelian case , the loop variable @xmath8 is gauge invariant for closed curves , and so there is no need to take into account its exponentiation @xmath40 , as it would be required in the nonabelian case .
this feature has a useful consequence .
it allows indeed to avoid the case in which the double - line integral ( [ p - c ] ) has to be taken along the same curve .
this case , usually referred to as the self - linking , would be automatically generated by the perturbative taylor expansion of the exponential @xmath41 . in other words , as far as the abelian case is concerned , the loop variables in eq.([p - c ] ) do not need to be exponentiated . therefore , the two curves @xmath1 and @xmath2 will always refer to two distinct curves which do not intersect each other . as we shall see in the following , this point will be relevant in order to establish the independence from the parameter @xmath35 of the expression ( [ p - c ] ) .
in order to discuss the perturbative expansion of the loop correlator ( [ p - c ] ) , let us first define the gauge - fixed version of the effective action which shall be used throughout the present article , namely , @xmath42 where the lagrange multiplier @xmath43 has been introduced in order to implement the landau gauge .
notice that we have wick - ordered the quartic interaction term , which will allow to rule out tadpole diagrams .
and @xmath44 of the field strength @xmath45 are each gauge invariant . ] as usual in this kind of problem , we shall make use of the configuration space rather than the momentum space .
let us now give the elementary wick contractions which shall be needed for the evaluation of the feynman diagrams .
recalling that @xmath46 from eq.([prop ] ) one obtains @xmath47 and @xmath48 concerning now the perturbative loop expansion , it is easily checked that the first feynman diagram which contributes to the correlation function ( [ p - c ] ) is of two - loop order and can be drawn as follows : in the above figure , the wavy and dashed lines refer respectively to the wick contractions @xmath49 and @xmath50 . the feynman integrals corresponding to the diagram of fig.2 are easily written down by means of eqs.([wc1 ] ) , ( [ wc2 ] ) . however , before computing them , let us spend a few words on the mechanism which is responsible for the independence on the parameter @xmath35 of expression ( [ p - c ] ) . from the structure of the diagram of fig.2 , we observe that the gauge fields @xmath51 and @xmath52 lying on the two curves @xmath53and @xmath54will be always contracted with the @xmath55 s present in the interaction term of the expression ( [ effective ] ) .
therefore , besides contractions of the type @xmath56 , the corresponding feynman integrals will always contain two contractions of the kind @xmath57 .
however , one should remark in the second term of eq.([wc1 ] ) that one of the lorentz indices of the two space - time derivatives corresponds to the vector index of a gauge field lying on either @xmath58or @xmath36 it thus refers to a total derivative with respect to the variable running along one of the closed loops , implying a vanishing contribution .
in other words , the second term of eq.([wc1 ] ) may be neglected . as a consequence ,
all the wick contractions entering the feynman integrals will basically lead to a product of delta functions .
after the introduction of a suitable regularization , the latter can be integrated out , finally resulting in a @xmath59 , where , we remind , @xmath60 and @xmath61 run along each of the two curves , respectively . as these variables never coincide , the whole expression vanishes identically , ensuring the independence from the parameter @xmath35 of the correlator ( [ p - c ] ) .
the same mechanism can be seen to occur at higher loop orders , as it will be explicitly shown later on .
apart from an irrelevant global symmetry coefficient , the diagram of fig.2 therefore corresponds to the following integral : @xmath62 .
\nonumber \\ & & \,\ , \label{int2}\end{aligned}\ ] ] let us analyse the first term of the above expression . making use of the propagators ( [ wc1 ] ) and ( [ wc2 ] ) , we obtain @xmath63 \left [ g_{\nu \gamma } \delta ^3(y - z_2)+\partial _ \nu \partial _
\gamma \frac 1{4\pi \left| y - z_2\right| } \right ] \nonumber \\ & & \quad \times \left [ \varepsilon _ { \beta \rho \lambda } \partial ^\lambda \delta ^3(z_1-z_2)\right ] \left [ \varepsilon ^{\alpha \rho \tau } \partial _
\tau \delta ^3(z_1-z_2)\right ] \left [ \varepsilon ^{\beta \gamma \sigma } \partial _
\sigma \delta ^3(z_1-z_2)\right ] \ , .
\label{second}\end{aligned}\ ] ] as previously mentioned , the terms containing the derivatives @xmath64 and @xmath65 do not contribute , as they correspond to total derivatives on closed curves .
expression ( [ second ] ) then becomes @xmath66 \left [ \varepsilon _ \mu { } ^{\rho \tau } \partial _
\tau \delta ^3(z_1-z_2)\right ] \left [ \varepsilon ^{\beta \sigma } { } _ \nu \partial _
\sigma \delta ^3(z_1-z_2)\right ] \ , .
\label{second2}\end{aligned}\ ] ] in spite of the presence of products of delta functions with the same arguments , the above expression is easily seen to vanish .
let us show this claim in two ways .
first , we observe that there is always a possible order of taking the integrations over the delta functions such that we end up with products of @xmath59 and not of @xmath67 . in the present case
, this would amount to integrate out first the two delta functions with arguments @xmath68 and @xmath69 , which would lead to @xmath70 \left [ \varepsilon _ \mu { } ^{\rho \tau } \partial _
\tau \delta ^3(x - y)\right ] \left [ \varepsilon ^{\beta \sigma } { } _ \nu \partial _
\sigma \delta ^3(x - y)\right ] = 0\ , , \label{second3}\ ] ] since @xmath71 never vanishes .
it is worth remarking that this possibility exists , in fact , for the higher - order diagrams , as will be shown below .
second , we can adopt a more rigorous treatment by regularizing the delta functions with coinciding arguments through the point - splitting procedure already used by polyakov @xcite : @xmath72 more precisely , whenever a product of @xmath73 delta functions with coinciding arguments occurs , it will be understood as @xmath74 ^n=\left [ \delta _ \varepsilon ( z_1-z_2)\right ] ^{n-1}\delta ^3(z_1-z_2)\,,\ ] ] where the limit @xmath75 is meant to be taken at the end of all calculations .
accordingly , expression ( [ second2 ] ) will be replaced by its regularized version , @xmath76 \left [ \varepsilon _ \mu { } ^{\rho \tau } \partial _
\tau \delta _ \varepsilon ( z_1-z_2)\right ] \left [ \varepsilon ^{\beta \sigma } { } _ \nu \partial _
\sigma \delta ^3(z_1-z_2)\right ] \ , . \label{second4}\end{aligned}\ ] ] whatever the order of integration , we get , before taking the limit , an expression containing @xmath59 , which leads to a null result .
the second term of ( [ int2 ] ) follows analogously , so that the two - loop diagram of fig.2 does not contribute to the correlator ( [ p - c ] ) . concerning the higher - order contributions in the perturbation theory , the results are of a similar nature .
the topologically distinct diagrams contributing to the 3- and 4-loop are given in fig.3 and in figs.4 and 5 .
it is sufficient to present here just one typical term of each order .
a notational simplification is convenient .
we define the transverse derivative operator @xmath77 for instance , a typical contraction from fig.3 is proportional to @xmath78 \left [ \tilde{\partial}_{\beta \delta } \delta ^3(z_1-z_3)\right ] \left [ \tilde{\partial}^{\beta \lambda } \delta ^3(z_1-z_3)\right ] \nonumber \\ & & \times \left [ \tilde{\partial}_{\rho \lambda } \delta ^3(z_2-z_3)\right ] \left [ \tilde{\partial}^{\rho \delta } \delta ^3(z_2-z_3)\right ] \nonumber \\ & = & \oint_{\gamma _ 1}dx^\mu \oint_{\gamma _ 2}dy^\nu \int d^3z_3\left [ \tilde{\partial}_{\mu \nu } \delta ^3(x - y)\right ] \nonumber \\ & & \times \left [ \tilde{\partial}_{\beta \delta } \delta _ \varepsilon ( x - z_3)\right ] \left [ \tilde{\partial}^{\beta \lambda } \delta ^3(x - z_3)\right ] \left [ \tilde{\partial}_{\rho \lambda } \delta _ \varepsilon ( y - z_3)\right ] \left [ \tilde{\partial}^{\rho \delta } \delta ^3(y - z_3)\right ] , \nonumber \\ & & \label{third}\end{aligned}\ ] ] while the diagram of fig.4 gives @xmath79 \left [ \tilde{\partial}^{\beta \lambda } \delta ^3(z_1-z_3)\right ] \left [ \tilde{\partial}_{\beta \sigma } \delta ^3(z_1-z_4)\right ] \nonumber \\ & & \times \left [ \tilde{\partial}^{\rho \sigma } \delta ^3(z_2-z_4)\right ] \left [ \tilde{\partial}_{\rho \lambda } \delta ^3(z_2-z_3)\right ] \left [ \tilde{\partial}^{\varphi \omega } \delta ^3(z_3-z_4)\right ] \nonumber \\ & & \times \left [ \tilde{\partial}_{\varphi \omega } \delta ^3(z_3-z_4)\right ] \nonumber \\ & = & \oint_{\gamma _ 1}dx^\mu \oint_{\gamma _ 2}dy^\nu \int d^3z_3\,d^3z_4\,\left [ \tilde{\partial}_{\mu \nu }
\delta ^3(x - y)\right ] \nonumber \\ & & \times \left [ \tilde{\partial}^{\beta \lambda } \delta ^3(x - z_3)\right ] \left [ \tilde{\partial}_{\beta \sigma }
\delta ^3(x - z_4)\right ] \left [ \tilde{\partial}^{\rho \sigma } \delta ^3(y - z_4)\right ] \nonumber \\ & & \times \left [ \tilde{\partial}_{\rho \lambda } \delta ^3(y - z_3)\right ] \left [ \tilde{\partial}^{\varphi \omega } \delta _ \varepsilon ( z_3-z_4)\right ] \left [ \tilde{\partial}_{\varphi \omega } \delta^3(z_3-z_4)\right ] .
\label{fourth}\end{aligned}\ ] ] in order to obtain the above expressions we have followed the same prescription established before for regularizing the delta functions with identical arguments .
notice also that we have integrated out first the two @xmath80-functions whose arguments depend on the points @xmath60 and @xmath61 of the two curves .
all terms in all possible diagrams may then be seen to be proportional to @xmath59 ( or its derivatives ) .
one may easily convince oneself that this mechanism also applies to any order in perturbation theory .
as we always have @xmath81 , these diagrams all amount to a null correction to the basic diagram , so that the correlation function ( [ p - c ] ) for two closed smooth nonintersecting curves @xmath82 gives their linking number to all orders : @xmath83
we have been able to show , in the present article , that the correlation function ( [ p - c ] ) is unaffected by the radiative corrections , provided @xmath82 are two nonintersecting closed curves .
although we have given explicit expressions for the @xmath84 perturbation , the same result may be achieved for any local interaction term of the type @xmath85 .
we may interpret this result as a kind of nonrenormalization property of the linking number , reflecting its stability with respect to any local gauge invariant perturbation of the starting chern - simons action . the conselho nacional de pesquisa e desenvolvimento ( cnpq / brazil ) , the fundao de amparo pesquisa do estado do rio de janeiro ( faperj ) and the sr2-uerj are gratefully acknowledged for financial support .
e. c. marino , phys .
b * * 263 * * ( 1991 ) 63 * * ; * * + e. fradkin and f.a .
schaposnik , phys .
b * * 338 * * ( 1994 ) 253 * * ; * * + f. a. schaposnik , phys . lett .
b * * 356 * * ( 1995 ) 39 * * ; * * + c. p. burgess , c. a. ltken and f. quevedo , phys . lett .
b * * 336 * * ( 1994 ) 18 * * ; * * + d. g. barci , c.d .
fosco and l. e. oxman , phys .
b * * 375 * * ( 1996 ) 267 * * ; * * + r. banerjee and e.c .
marino , nucl .
b * * 507 * * ( 1997 ) 501**. * * t.h .
hansson , a. karlhede and m. rocek , phys . lett .
b * 225 * ( 1989 ) 92**. * * + n. brali and l. vergara , _ fuzzy statistics in covariant quantum field theory , _ int . europhys . conf . on high energy physics , marseille , france , 1993 | the abelian chern simons theory is perturbed by introducing local gauge - invariant interaction terms depending on the curvature .
the computation of the correlation function @xmath0 for two smooth closed nonintersecting curves @xmath1 , @xmath2 is reported up to four loops and is shown to be unaffected by radiative corrections .
this result ensures the stability of the linking number of @xmath1 and @xmath2 with respect to the local perturbations which may be added to the chern
simons action . |
in the last decades several approaches to quantum gravity have been developed , with important results @xcite . among them , group field theories ( gft ) @xcite are , in our opinion , particularly promising .
gfts are quantum field theories defined over a group manifold , representing a meta - spaceof discrete spacetime geometries , and not spacetime itself , as a new algebraic and combinatorial realization of the third quantizationidea @xcite .
moreover , gfts bring together most of the ingredients entering in other _
non perturbative _ and _ background independent _ approaches ( such as loop quantum gravity , spin foam models and simplicial quantum gravity approaches ) @xcite .
very little is known , still , about group field theory models for quantum gravity , in both 3 and 4 dimensions , and a lot of technical work should go into their analysis , for example their classical solutions , which will be one focus of the present work , and their purely field theoretic aspects , which are the subject of attention at present @xcite .
an important issue in many ( non - perturbative ) quantum gravity formalisms is the inclusion and the correct description of matter fields . in the literature we can distiguish two different strategies .
one approach is start from a model describing a quantum spacetime without matter and then add degrees of freedom describing matter , being fields , point particles or extended objects .
this is the standard route both in loop quantum gravity @xcite , spin foam models @xcite and simplicial quantum gravity @xcite . in a gft context
this means , for example , writing down a coupled gft action @xcite for both gravity and matter fields that would produce , in perturbative expansion , a sum over simplicial complexes with dynamical geometry together with feynman graphs for the matter fields living on the simplicial complexes . or it means @xcite defining a gft model whose feynman amplitudes can be understood as state sum models for gravity with appropriate matter field observables .
an alternative approach is to think of matter degrees of freedom as a subset of the same degrees of freedom defining the microstructure of quantum spacetime itself , and emerge as fluctuations around a background configuration of the same .
this idea , pursued for example ( in different ways ) in @xcite , shares some similarities with analog gravity models in condensed matter systems , in which quasi - particles with an effective description as matter fields propagating on curved geometries emerge as fluctuations around stable vacua of the underlying many - body system ( e.g. a bose condensate ) , but are of course collective excitations of the same microscopic degrees of freedom just as well . a strategy for emergent matter within the gft framework , in particular , first studied in @xcite , is based on the following idea : starting from a fundamental gft action , one looks for solutions of the corresponding equation of motion ; then , one considers perturbations around these solutions , obtaining an effective dynamics for the perturbation field .
the task is to identify a class of solutions and of perturbations , such that the effective dynamics for these perturbations takes the form of a matter field theory on an effective spacetime .
it turns out @xcite that the effective matter field theories emerging from gfts following this procedure are noncommutative quantum field theories on non - commutative spacetimes with a lie algebra structure , and with a curved momentum ( group ) manifold .
this only makes this strategy all the more interesting .
in fact , non - commutative geometry has been often advocated as the appropriate language to describe a quantum spacetime , i.e. a regime which , already far from the full non - perturbative quantum gravity dynamics at the ( ultra-)planckian scales , is still semi - classical in that it incorporates somehow the quantum gravity corrections to continuum physics at the effective level .
this can take the form of an effective minimal ( planck ) length scale in matter field theories , of a non trivial commutation relation between position operators in quantum mechanics ( so that the notion of spacetime point , and of spacetime continuum , becomes meaningless ) , or of a generalised uncertainty principle , or of a form of co - gravity , i.e. curvature in momentum space @xcite .
it is actually this sort of effect , a curved momentum sector for particle ( and field ) kinematics , which shows up most naturally in a gft context .
moreover , in recent years much work in the context of quantum gravity phenomenology @xcite was developed .
on the one hand , this starts from the simple idea that several astrophysical systems could work as a magnifying lens to amplify quantum gravity effects even if planck scale suppressed , and make them accessible to experiments .
on the other hand , much of this work relies on effective models of quantum gravity , rather than on the tentative fundamental formalisms currently available , and in particular on effective non - commutative models for matter kinematics and dynamics , one important example being so - called deformed ( or doubly ) special relativity @xcite .
this means that the recent results on the emergence of effective non - commutative matter field theories from gft have the potential to help bridging the gap between the microscopic dynamics they define for quantum space and macroscopic physics and quantum gravity phenomenology . in this paper
we thus follow this strategy , and apply it to a generalised class of gft models introduced in @xcite , focusing on the 3d case .
this generalised gft formalism has been introduced as an attempt to formulate gft models , which on the one hand depended explicitly on metric - related variables , thus allowing a more manifest encoding of simplicial geometry at the level of the gft action , and on the other hand possessed feynman amplitudes with the explicit form of a simplicial gravity path integral , thus clarifying the relation between spin foam and simplicial gravity structures . obviously , the viability and correctness of such a generalised gft formalism is something to be tested .
therefore , we understand the results of this paper in two dual ways : 1 ) as a test of the strategy for the emergence matter from group field theories as perturbations around classical gft solutions ; in fact , we check here whether this strategy works nicely also for ( much ) less trivial models than those it has been applied to so far , and under which assumptions and conditions ; 2 ) as a test of the generalised gft formalism itself ; in fact , we will see that some difficulties encountered in applying this procedure , and some unsatisfactory features of the resulting effective actions for perturbations , can naturally be understood as stemming from limitations of the generalised gft formalism itself , rather than as failures of the emergent matter strategy .
let us also stress that the same identification of classical solutions of the model and of the corresponding perturbative dynamics will be an interesting and highly non - trivial result , from a purely technical perspective .
the outline of the paper is as follows . in the remaining sections of the introduction
, we recall how a non - commutative matter field theory for a scalar field emerges from the perturbations around classical gft solutions in a simple 3d model , and then we introduce the 3d generalised gft model . in section
ii , we consider a simplified version of the latter , identify a class of solutions of the classical gft equations , and then study the corresponding perturbations ; we will see that the basic elements of the complete analysis of perturbations to be done on the full model are already evident , together with the limitations of the same model in this respect . in section iii , we move on to the complete model ; we find some exact and approximate classical solutions of its equations of motion , and extract the effective dynamics for the corresponding perturbations , focusing on the approximate case .
we conclude with a discussion of the results obtained and an outlook on further developments , as suggested also by our results .
in this section we briefly review some results about the emergence of an effective noncommutative field theory in the 3d case @xcite .
we start from the following gft action ( boulatov model ) , based on a group manifold being either @xmath0 or @xmath1 : @xmath2=\frac{1}{2}\int dg_1 dg_2 dg_3\ \phi(g_1 , g_2 , g_3 ) \phi(g_3 , g_2 ,
g_1 ) -\frac{\lambda}{4!}\int dg_1\ldots dg_6\ \phi(g_1,g_2,g_3 ) \phi(g_3,g_4,g_5)\phi(g_5,g_2,g_6)\phi(g_6,g_4,g_1),\ ] ] + where @xmath3 is a real field , that we suppose invariant under diagonal action of the group , @xmath4 .
the corresponding classical equation of motion is @xmath5 + and a class of solutions of the same equation is parametrized by a function @xmath6 , satisfying @xmath7 , and given by : @xmath8 since the boulatov model defines a ( 3rd ) quantization of bf theory , whose only configurations are locally flat geometries on arbitrary topology , such solutions can be interpreted as describing a quantum flat space .
now we want to study classical perturbations around these solutions , in particular those of the form : @xmath9 . for them
we obtain the following effective action : @xmath10&\equiv & s[\phi_f+\psi]-s[\phi_f ] = \int dg \psi(g ) \mathcal{k}_f(g)\psi(g^{-1})-\frac{\mu}{3!}\int dg_1\ dg_2\ dg_3\
\delta(g_1g_2g_3 ) \
\psi(g_1)\psi(g_2)\psi(g_3)+\nonumber\\ & - & \frac{\lambda}{4 ! } \int dg_1\ dg_2\
dg_3\ dg_4\ \delta(g_1\ldots g_4)\ \psi(g_1)\psi(g_2)\psi(g_3)\psi(g_4 ) .
\label{eq : minchia}\end{aligned}\ ] ] + with kinetic term given by @xmath11.\ ] ] + the nature of this action as an effective scalar field theory on a noncommutative space is evident once we interpret the group manifold on which it is defined as momentum space , so that the constraints @xmath12 and @xmath13 represent the momentum conservation . as a curved space ,
this momentum space is dual to a noncommutative configuration space with a lie algebra structure ( given by the @xmath14 ) .
this duality can be made explicit by introducing a fourier transform mapping functions over the group to functions over the corresponding algebra @xcite , using which we can rewrite the above action in term of functions on @xmath15 endowed with a noncommutative @xmath16-product @xcite , which can then be shown to be invariant under the quantum double of @xmath1 , a deformation of the euclidean 3d poincar group .
similar results have been obtained also in the 4d case and in lorentzian signature @xcite , where it has been shown that , starting from a gft model for @xmath17 bf theory , one can obtain an effective non - commutative field theory of the dsr type , i.e. for a scalar field living on the @xmath18-minkowski spacetime and with a de sitter space of momenta .
obviously , this is a model of a more direct interest for quantum gravity phenomenology .
we can conclude that perturbations around a non trivial quantum geometric background , defined by a solution of the gft equations , follow the dynamics of an effective scalar field theory of a noncommutative type .
so we can interpret matter as a phase of quantum geometry , therefore already contained in our quantum gravity models .
moreover , this type of analysis of gft perturbations is useful to expose the deep link between the gft formalism and noncommutative geometry .
a generalised gft formalism has been introduced in @xcite and further developed in both 3d and 4d in @xcite , its purpose being encoding and controlling in a more explicit way simplicial geometry at the level of the gft action , and obtaining feynman amplitude with the manifest form of a simplicial gravity path integral . in turn , this should serve as a complete definition of a quantum dynamics of simplicial structures , to obtain a spin foam model with a clear underlying geometry and to strengthen the links between spin foam and simplicial gravity approaches . in the 3d case , the fundamental variable is a real function of @xmath19 group variables and @xmath19 algebra variables : @xmath20 , which satisfies @xmath21 .
the group and lie algebra variables are interpreted as the discrete triad and connection variables of simplicial bf theory or , equivalently , of discrete 3d gravity .
its classical action is the following : @xmath22 + where we have introduced the operator @xmath23 , with @xmath24 the laplace - beltrami operator on the group manifold @xmath25 , and the notation @xmath26 .
notice that , even though the field is a function of lie algebra triad variables , it is treated in this formalism as an ordinary function , and the lie algebra itself is considered in its vector space aspects only , so that integrations are performed by means of the lebesgue measure and no star product is introduced when multiplying more than one gft field .
this will have important consequences when studying the corresponding perturbations around classical solutions .
for the achievements and limitations of this formalism concerning its original goals , mentioned above , we refer to the literature . here
we will only try to see whether the results on perturbations obtained for the boulatov model extend to this more complicated case , and at the same time use this analysis of perturbations as a test for the formalism , trying to expose further its limitations by making them explicit at the level of the resulting effective dynamics .
let us begin from the analysis of the simpler static - ultralocaltruncation of the model .
it corresponds to the case in which the kinetic operator is `` freezed '' to the identity , imposing @xmath27 in ( [ eq : azione ] ) : @xmath28 + in the following we _ always _ assume that _ solutions depend on the moduli of the vectors @xmath29 _ , and not from its direction , so that the invariance property of the field under the diagonal group action on the lie algebra arguments @xmath30 is trivialized .
moreover , to semplify the notation , we define @xmath31 .
anyway we use the notation @xmath32 , to distinguish integration over all algebra with integration over only the module @xmath33 .
the equation of motion corresponding to the above action is : @xmath34 + and it is easy verify that the following class of functions , a simple generalization of ( [ eq : soluzionesemplice ] ) , define classical solutions : @xmath35 + where the functions @xmath36 , @xmath37 and @xmath38 have to fulfil the following relations : @xmath39 + we also require , for simplicity , the function @xmath40 to be separable in its variables : @xmath41 . let us consider a generic perturbation @xmath42 , and compute the effective action : @xmath43&\equiv&s[\varphi+\epsilon]-s[\varphi]\,=\,\frac{1}{2}\int [ dg]\int [ d\vec{b } ] \ \epsilon_{123}\epsilon_{321}+\nonumber\\ & - & \int [ dg ] dh_1\int [ d\vec{b } ] \psi_1(b_5 ) f_1(g_2h_1)f_2(b_2)f_1(g_4h_1)f_2(b_4)\psi_3(b_1)\ ; \epsilon(h_1^{-1},g_2,g_3;b_1,b_2,b_3)\epsilon(g_3,g_4,h_1^{-1 } ; b_3,b_4,b_5)+\nonumber\\ & -&\frac{1}{2}\int [ dg ] dh_1 dh_2\int [ d\vec{b } ] \psi_1(b_3 ) f_1(g_4h_1)f_2(b_4)\psi_3(b_1)\psi_3(b_5)\psi_1(b_6)f_1(g_4h_2)f_2(b_4)\psi_3(b_6)\times\nonumber\\ & \times & \epsilon(h_2^{-1},g_2,h_1^{-1};b_1,b_2,b_3)\epsilon(h_1^{-1},g_2,h_2^{-1 } ; b_5,b_2,b_6)-\frac{\lambda}{4!(2\pi)^{27}}\int [ dg ] \int [ d\vec{b } ] \
\epsilon_{123}\epsilon_{345}\epsilon_{526}\epsilon_{641}\ , + \nonumber\\ & -&\sqrt{\frac{\lambda}{3!(2\pi)^{27}}}\int [ dg ] dh \int [ d\vec{b } ] \ \psi_1(b_6)f_1(g_4h)f_2(b_4)\psi_3(b_1)\ ; \epsilon(h^{-1},g_2,g_3;b_1,b_2,b_3)\epsilon_{345}\epsilon(g_5,g_2,h^{-1 } ; b_5,b_2,b_6),\nonumber\end{aligned}\ ] ] + where we used the notation @xmath44 , where convenient .
this is the action for a generic perturbation , only assumed to have the sam diagonal invariance of the fundamental gft field , but it simplifies considerably , and can then be interpreted physically , by appropriate simplified choices on the perturbation field . in particular , having in mind the possible interpretation of the @xmath29 and @xmath45 variables as configuration and momentum variables , respectively , we try to reduce the dependence of the perturbation field to either of the two sectors .
* case 1 : _ dynamical group sector _ *
now we show how one can recover the same effective action in @xcite with a particular choice of perturbation field .
we assume a separable dependence on @xmath45 and @xmath29 variables ( this is suggested also by the fact that neither the kinetic nor the interaction term couple the two sectors ) , and the dependence on a single group variable via the combination @xmath46 : @xmath47 .
factoring out the integration over the @xmath29 variables , we get the following effective action for the perturbation field @xmath48 : @xmath49=\int dg\ \phi(g ) \mathcal{k}(g)\phi(g^{-1 } ) -\frac{\eta}{3!}\int [ dg]^3\ \delta(g_1\ldots g_3 ) \psi(g_1)\ldots\psi(g_3 ) - \frac{\mu}{4 ! } \int [ dg]\ \delta(g_1\ldots g_4 ) \psi(g_1)\ldots\psi(g_4).\end{aligned}\ ] ] + here , the kinetic term is given by @xmath50,\ ] ] + which is the same of the action ( [ eq : minchia ] ) , with up to some appropriate redefinition of coupling constants : @xmath51 + of course the perturbation field must be chosen such that this new set of coupling constants does not diverge , with the functional dependence on the b sector to be considered as fixed , with no variation allowed in the classical theory and no fluctuations ( functional integration ) in the quantum theory .
we see that we have recovered , even from the more complicated gft formalism , the expected effective action for the perturbations , that can be interpreted as a noncommutative scalar field theory written in momentum space . even in this drastic simplification of the formalism , we also see , however , that the way we could obtain it was to completely decouple the b and g sector and to leave only this last one as dynamical . * case 2 : _ dynamical lie algebra sector _ * now we focus instead on the @xmath29 sector , and treat the group sector as non - dynamical .
the idea is then to reproduce the same action as above , or a modification thereof , but written now in configuration space , according to the standard interpretation of the effective theory and of the variables appearing in the gft , outlined above . as a regularization
, we introduce a cut - off @xmath52 in the b integration , and the constant @xmath53 , that is the ( truncated ) volume of @xmath15 .
we then consider a perturbation of the form @xmath54 with no dependence on group variables ( notice that this does not lead to any divergence because of the compact nature of the group manifold ) .
the dependence on the three variables @xmath55 has been chosen to be what would have followed from a fourier transform from @xmath45 to @xmath29 variables of the function we used as perturbation in the previous case .
moreover we introduce a new condition on the solution imposing that @xmath56 . with this condition the integration over the group variables kills the cubic interaction and some quadratic terms .
the resulting effective action for this perturbation field is : @xmath57= \frac{1}{2 } \int d\vec{b}\ \psi^2(b ) + \frac{\lambda}{4!}\int d\vec{b}\ \psi^4(b)\ ] ] that is an ordinary local @xmath58 theory with a trivial kinetic term ( i.e. again a static - ultralocal field theory ) .
we can interpret this as a scalar field theory written in configuration space ( given by @xmath15 ) .
the triviality of the kinetic term arises directly from the initial ultra - locality restriction , but could possibly be lifted by some more involved choice of classical solution or perturbation field , at the cost of introducing derivatives in the same .
the point to notice , however , is that we find no sign of the expected underlying non - commutativity in this field theory .
that is , while we find the expected local dependence on @xmath29 variables that matches their interpretation as configuration variables in some sense dual to the group variables , and while we could try to modify our construction to obtain a more complicated ( and interesting ) kinetic term or interaction , it is evident that the conjugate nature of @xmath29 and @xmath45 sectors is lost and can not be recovered . due to the simplicity of the underlying model , it is also clear that this conjugate nature fails to be present in the effective action because it fails to be correctly implemented in the original generalised gft formalism .
we will come back to this point in the following .
in this section we consider the complete action ( [ eq : azione ] ) and the corresponding integro - differential equation of motion : @xmath59 we first try to find an exact solution of these equation and to obtain an effective field theory for the perturbations .
we look for solutions of the type : @xmath60 + where we have imposed the usual invariance under diagonal right action of the group , by explicit projection .
it is then easy to show that any function @xmath61 of the above form is an exact solution of the gft equation of motion if the following conditions are satisfied : 1 .
the functions @xmath62 are solutions of the equation + @xmath63 + so they are eigenfunctions of the operator @xmath64 with eigenvalue @xmath65 .
@xmath37 , @xmath38 and @xmath36 satisfy + @xmath66 we now look for such functions , starting from condition 1 .
we now prove the following : * proposition * : _ let @xmath67 be the set of functions defined on the cartesian product @xmath68 and square integrable with respect to the group variable ( so that we can apply the peter weyl theorem to decompose them in irreducible representations @xmath69 of g ) .
the most general solution , in the set @xmath67 , of the equation ( [ eq : kernel ] ) is of the form _
@xmath70 the proof is straightforward .
because of peter - weyl theorem , every solution of the equation , belonging the set @xmath67 , can be written as @xmath71 , for some coefficient functions @xmath72 . being a solution
, it has to satisfy : @xmath73\tilde{\phi}_j^{nm}(b)\right\}d^j_{nm}(g)=0 $ ] , which implies , since the @xmath74 is a complete orthonormal basis in @xmath67 , that : @xmath75\tilde{\phi}_j^{nm}(b)=0 \quad \forall j , n , m$ ] .
this equation can be satisfied only if the coefficient functions @xmath72 are null for all values such that @xmath76 , while they can be arbitrary otherwise .
@xmath77 therefore , the functions ( [ eq : eigenfunction ] ) , considered as function of @xmath78 are non null only on spheres of radius @xmath79 .
thus the integral over the algebra of these eigenfunctions is obviously zero , because each sphere is a set of null measure .
therefore the conditions 2 and 3 unless the functions @xmath80 are actually distributions on @xmath15 .
the idea is then to define a 3 dimensional delta - like function , i.e. null everywhere , except on the origin and on the spheres of radius @xmath79 , with @xmath81 , where it goes to infinity , and multiply it by a regular function of @xmath29 to define completely the functions @xmath80 .
there are many possibilities to define this distribution with a limit procedure , and we do it in appendix , providing explicit examples . in the following we call that distribution @xmath82 .
its only property we need is : @xmath83 + where both integral and sum have to converge .
let @xmath84 and @xmath85 solutions of ( [ eq : kernel ] ) .
we also introduce the following distributions : @xmath86 + these define solutions too , as they satisfy condition 1 , as it can be easily checked .
now we consider condition 2 , and compute the following integral : @xmath87 + using the above theorem , we know that @xmath88 which results in the condition : @xmath89 which characterizes functions satisfying condition 2 . using the same distribution @xmath82 , we can define a function satisfying condition 3 too .
let @xmath36 be a function of the type [ eq : eigenfunction ] ; we define the distribution @xmath90 , and impose condition 3 , to find @xmath91 + if this sum is convergent then we can normalize @xmath36 so that the result is 1 and condition 3 is satisfied . therefore , using functions satisfying the above conditons , we have found an exact solution of equation ( [ eq : difficile ] ) : @xmath92 now we write down the effective action for a generic perturbation around such solution : @xmath43&=&\frac{1}{2}\frac{1}{(2\pi)^9}\int [ dg]\int [ d\vec{b}]\ , \epsilon_{123}\hat{o}_1\hat{o}_2\hat{o}_3\epsilon_{321}-\frac{\lambda}{4!(2\pi)^{36}}\int [ dg ] \int [ d\vec{b } ] \epsilon_{123}\epsilon_{345}\epsilon_{526}\epsilon_{641}+\nonumber\\ & - & \frac{1}{(2\pi)^9}\int [ dg ] dh_1 dh_2\int [ d\vec{b}]\,\psi_1(g_1h_1,b_1)f(g_2h_1,b_2)f(g_4h_2,b_4)\psi_3(g_5h_2,b_5)\psi(h_1^{-1}h_2 ) \epsilon_{526}\epsilon_{641}\nonumber\\ & -&\frac{1}{2}\frac{1}{(2\pi)^9}\int [ dg ] dh_1 dh_2\int [ d\vec{b } ] \;\psi_1(g_1h_1,b_1)\psi_3(g_3h_1,b_3)\psi_1(g_5h_2,b_5)\psi_3(g_6h_2,b_6)\xi(h_1^{-1}h_2 ) \epsilon_{345}\epsilon_{641}\nonumber\\ & -&\sqrt{\frac{\lambda}{3!(2\pi)^{45}}}\int [ dg]\int [ d\vec{b } ] \psi_1(g_1h , b_1 ) f(g_2h , b_2)\psi_3(g_3h , b_3 ) \epsilon_{345}\epsilon_{526}\epsilon_{641}\,\,\hspace{0.5 cm } \label{eq : long}\end{aligned}\ ] ] + where we have defined : @xmath93 this is a very general result , holding for any solution and any perturbation .
now we would obtain , starting from [ eq : long ] , an action with a kinetic term of the type @xmath94 , either in group or in lie algebra variables , imposing some particular form of the perturbation field .
this turns out to be very difficult , and we have not been able to identify any simple form of the perturbations that would give such simplification .
roughly speaking , it is due to the presence of the terms @xmath95 and @xmath96 , which are not equivalent , because we have not assumed any invariance of the gft field , nor of the corresponding perturbation , under permutations of its arguments .
there is no simple choice of perturbation field that gives a product of the type @xmath97 , for example no field with a dependence on the @xmath29 variables only through the linear combination @xmath98 , or the analogue in the @xmath25 variables .
this is true , unless we assume that the fundamental gft field as well as the classical solution chosen and the corresponding perturbation field are invariant under permutations of their group and lie algebra arguments . imposing this property however ,
would complicate the analysis considerably . leaving aside the issue of perturbations ,
the problem can be seen as due to the non - separability of the solution chosen , in particular the functions @xmath36 and @xmath99 . we can not exclude that a different solution , or a more involved choice of perturbation would lead to a nicer effective action .
however , we also note that , even if one solves somehow this first problem , another trouble comes from the distribution @xmath82 , which appears in the effective action for the perturbations with different powers , leading to additional technical complications .
this suggests us to leave aside for the moment the issue of perturbations around exact solutions of the generalised model .
we turn our attention , instead , to _ approximate solutions _ and their corresponding perturbations .
the basic idea here is to look for eigenfunctions of the laplace - beltrami operator and , at the same time , good approximations of the delta function on the group .
we note that every good approximation of the delta function on the group has to be sharply peaked on the identity .
therefore the laplace - beltrami operator applied to such a function will be close to zero everywhere but in a neighbourhood of the identity element .
for this reason we study now the behaviour of the laplace - beltrami operator on the group @xmath1 which is isomorphic to the 3-sphere @xmath100 , very close to the identity element , which we identify with the north pole .
now we fix a chart in a neighborhood of the identity element , and we use polar coordinates .
the laplace - beltrami operator becomes : @xmath101 + if we apply this operator to functions @xmath36 only dependent on the @xmath102 variable , we can consider only the first term in the previous formula , so we can use the operator : @xmath103 + where the @xmath104 indicates both the approximation in the operator and that we can apply it just on function of @xmath102 considerably different from zero only in a small neighbourhood of @xmath105 .
now we consider the eigenvalue problem for this operator : @xmath106 . among the solutions of this equation ,
one is particular interesting for our purposes @xmath107 , where @xmath108 .
indeed we note that in the limit @xmath109 the function @xmath110 goes to zero everywhere but in @xmath105 where it goes to infinity .
therefore , with appropriate normalization , this is a good representation of the dirac delta function .
thus we define the function @xmath111 + that satisfies the following properties : @xmath112 + of course we can rewrite the @xmath113 function in term of the group variable , using the parametrization : @xmath114 , with @xmath115 and @xmath116 . through this relation
we can interpret the @xmath117 as a @xmath118 on su(2 ) .
now we apply the operator @xmath64 on @xmath118 , with @xmath119 , to find : @xmath120 + of course the approximation improves as @xmath121 tends to zero .
indeed we have supposed that @xmath122 . in order for our solution of the equations of motion to satisfy this condition for all relevant values of @xmath29 , we then assume that it is a function with compact support ( i.e. identically zero outside of some bounded set ) with respect to the @xmath29 variables .
another important point is that we can handle @xmath118 like a true @xmath123 function , as it satisfies @xmath124 .
the last step is to introduce another function @xmath125 on the sphere @xmath100 .
it is defined , for @xmath126 ( i.e. for the the half - sphere corresponding to @xmath0 ) , as @xmath127 + and it is extended to the whole @xmath128 ( i.e. to @xmath129 ) , by requiring @xmath130 , again for @xmath131 .
this function is almost vanishing everywhere but for @xmath132 and satisfies : @xmath133 + in the last two properties the integration is over @xmath1 .
if we restricted the integration only over @xmath0 , the first integral gives @xmath134 and the second one gives @xmath135 . finally using all the above , it is easy to verify that the following function is an approximate solution of ( [ eq : difficile ] ) : @xmath136 + where @xmath62 are functions with compact support , dependent only on the module of @xmath78 , and satisfy the conditions @xmath137 next , we study perturbations around this new approximate solutions .
the effective action for a generic perturbation field @xmath138 is : @xmath139&\propto&\frac{1}{2}\int [ dg]\int [ d\vec{b } ] \ , \epsilon_{123}\hat{o}_1\hat{o}_2\hat{o}_3\epsilon_{321}\,-\,\frac{\lambda}{4!(2\pi)^{27}}\int [ dg ] \int [ d\vec{b } ] \
, \epsilon_{123}\epsilon_{345}\epsilon_{526}\epsilon_{641}\,+ \nonumber \\ & - & \frac{1}{a^6}\int [ dg ] dh\int [ d\vec{b}]\ , \psi_1(b_5)\delta_a(g_2h)f(b_2)\delta_a(g_4h)f(b_4)\psi_3(b_1)\ ; \epsilon(h^{-1},g_2,g_3;b_1,b_2,b_3)\epsilon(g_3,g_4,h^{-1 } ; b_3,b_4,b_5)+\nonumber\\ & -&\frac{1}{2}\frac{1}{a^6}\int [ dg ] dh_1 dh_2\int [ d\vec{b}]\ , \psi_1(b_3)\delta_a(g_4h_1)\psi_3(b_5)\psi_1(b_6)\delta_a(g_4h_2)\psi_3(b_1)\times\nonumber\\ & \times & \epsilon(h_2^{-1},g_2,h_1^{-1};b_1,b_2,b_3)\epsilon(h_1^{-1},g_2,h_2^{-1 } ; b_5,b_2,b_6)+\nonumber\\ & -&\sqrt{\frac{\lambda}{3!(2\pi)^{27}a^6}}\int [ dg ] dh\int [ d\vec{b}]\ , \psi_1(b_6)\delta_a(g_4h)f(b_4)\psi_3(b_1)\ ; \epsilon(h^{-1},g_2,g_3;b_1,b_2,b_3)\epsilon_{345}\epsilon(g_5,g_2,h^{-1 } ; b_5,b_2,b_6)\;\;\;\;\;\;\ ; \label{eq : dupalle}\end{aligned}\ ] ] + here there is a first difference with respect to the previous case : due to the approximation , the linear terms , which would cancel exactly when perturbing around exact extrema of the action , do not cancel exactly but are nevertheless suppressed with @xmath121 and we have not written them for this reason . in order to get a `` good '' effective theory we need again to impose some special conditions on the perturbation field . as in the static - ultralocal theory , we have two sets of variables , group and algebra , therefore we have more room for manouvering with respect to previous models .
the goal is again to obtain a field theory with a kinetic term written in the usual form @xmath94 , and thus with a field depending on a single ( group or algebra ) argument .
moreover we recall that the group and algebra variables have the interpretetion of momentum and configuration variables in the usual matter field theory of noncommutative type .
the first assumption is the separability of the perturbation field : @xmath140 next , having again in mind the possible interpretation of the @xmath29 and @xmath45 variables as configuration and momentum variables , respectively , we try to reduce the dependence of the perturbation field to either of the two sectors , by treating the other sector as non - dynamical , as in the static - ultralocal case . * case 1 : _ dynamical group sector _
* we recall that we have fixed the value of @xmath121 very close to zero .
thus we notice immediatly that in the action ( [ eq : dupalle ] ) we have two dominant quadratic terms .
we restrict our attention to them , i.e. to the free field theory sector .
the dominant terms , comparing powers of @xmath141 , are @xmath43&=&- \frac{1}{(2\pi)^9}\frac{1}{a^6}\int dg_2 dg_3 dg_4 dh_1\int d\vec{b}_1\ldots d\vec{b}_5 \psi_1(b_5)\delta_a(g_2h_1)f(b_2)\delta_a(g_4h_1)f(b_4)\psi_3(b_1)\times\nonumber\\ & \times & \epsilon(h_1^{-1},g_2,g_3;b_1,b_2,b_3)\epsilon(g_3,g_4,h_1^{-1 } ; b_3,b_4,b_5)+\nonumber\\ & -&\frac{1}{2}\frac{1}{(2\pi)^9}\frac{1}{a^6}\int dg_2 dg_4 dh_1 dh_2\int d\vec{b}_1d\vec{b}_2d\vec{b}_3d\vec{b}_5d\vec{b}_6 \psi_1(b_3)\delta_a(g_4h_1)\psi_3(b_5)\psi_1(b_6)\delta_a(g_4h_2)\psi_3(b_6)\times\nonumber\\ & \times & \epsilon(h_2^{-1},g_2,h_1^{-1};b_1,b_2,b_3)\epsilon(h_1^{-1},g_2,h_2^{-1 } ; b_5,b_2,b_6)\nonumber\end{aligned}\ ] ] + now we separate the variables like in ( [ eq : separazione ] ) and we impose the usual condition @xmath142 . after integrating out @xmath29 variables , treating the part of the perturbation field depending on them as non - dynamical , we get @xmath143=\frac{1}{2}\frac{1}{(2\pi)^9}\frac{c}{a^6}\int dg\ \varphi(g)k(g)\varphi(g^{-1}),\ ] ] + where the kinetic operator is
@xmath144 + and we have defined the constant @xmath145 we see that we obtain once more a nice field theory on a single group manifold , interpreted as ( curved ) momentum space , whose dynamics is again entirely determined by the classical gft solution chosen , thus by the quantum spacetime we select when selecting the solution itself , and on which the effective scalar field propagates .
once more , we could fourier transform from group to lie algebra space , playing the role of effective spacetime , introducing the appropriate star product .
* case 2 : _ dynamical lie algebra sector _ * here we restrict to the @xmath1 case . without this restriction
our choice of perturbation field gives a non - local effective action .
this case , from a technical point of view , is very similar to the static - ultralocal limit .
indeed we choose exactly the same type of perturbation field : @xmath146 where @xmath147 is again the ( truncated ) volume of @xmath15 .
after a straightforward calculation , we get @xmath57= \frac{1}{128}\int d\vec{b}\ \psi(\vec{b})k(b)\psi(\vec{b } ) + \frac{\lambda}{4 ! } \int d\vec{b}\ \psi^4(\vec{b})\ ] ] where @xmath148 .
this is a nice local effective action with a rather unusual kinematical term , with an algebraic dependence on the configuration space lie algebra variable .
once more , the specific form of the kinetic term could possibly be modified by different choices of classical ( approximate ) solution or perturbation field , even though it is not straightforward to envisage how to do so .
what can not be modified easily is again the fact that we find no sign of the expected underlying non - commutativity in this field theory , and that the conjugate nature of @xmath29 and @xmath45 sectors is lost .
this confirms that something is dubious in the way the original gft model , i.e. the generalised gft formalism , because we can once more trace back the origin of this limitation in the effective field theory to the way the generalised gft formalism deals with the dependence on the gft field on the lie algebra @xmath29 variables .
let us summarize the results we have obtained .
we have studied the issue of classical solutions of the generalized 3d gft model proposed in @xcite , and of the effective dynamics of the corresponding perturbations , trying to identify specific solutions and perturbations such that the dynamics of the latter can be understood as that of a scalar field on an effective ( noncommutative ) spactime .
the motivations were the following .
first of all , to improve our understanding of such models , at a more technical level .
second , to show that indeed effective matter field theories could be extracted as phases of such generalised models .
this means on the one hand again solving the technical difficulties coming from the more involved structure of this model , as compared to the easier ones which have been dealt with up to now , but on the other hand showing that the idea of matter as emerging from the collective dynamics of the micro - costituents of quantum spacetime is more solid than could be originally thought , holding for a wider class of gft models .
third , and conversely , to use this procedure for extracting effective matter theory from gft as a test for the generalised gft formalism itself .
this means learning from the way this strategy for emergent matter is made to succeed or from the way it fails , in this context , what are the weak and dubious points of the formalism .
our results are indeed mixed . we have managed to identify exact solutions of the gft equations of motion both in the simpler static - ultralocal truncation of the model and in the full , more complicated , one . in the truncated model we could extract rather straightforwardly an effective scalar field theory for the perturbations .
when only the group sector of the model is treated as dynamical , the effective field theory takes indeed the expected form of a non - commutative field theory written in momentum space .
one could then use a non - commutative fourier transform to re - write the theory in configuration space , given by the @xmath14 lie algebra .
because of the presence of both lie algebra and group variables in the fundamental gft model , one could also expect that it should be possible to obtain directly a non - commutative field theory for the perturbations in configuration space , by treating directly the only lie algebra sector of the perturbations as dynamical .
this turns out not to be the case .
one obtains indeed an effective scalar field theory on the same lie algebra , but without the wanted and expected non - commutative structure .
this turn out to be the result both in the truncated model and in the full model , for which we have been able to extract a nice(r ) effective dynamics for perturbations , i.e. one that could be interpreted as a scalar field theory on an emergent 3d spacetime , only within a certain approximation , that is only around approximate solutions of the equations of motion .
moreover , a closer look at the way the solutions of the equations of motions could be found and the effective dynamics for perturbations extracted shows that it has been necessary , also when dealing with the full model , to basically trivialize the kinetic term of the theory to the level of the corresponding equations , i.e. look for solutions for which the non - trivial part of the kinetic term gives a basically trivial contribution to the equations .
while this last point , although somehow disappointing , could be argued to be solvable with more work and some more clever analysis of the same equations , the first point appears to us to be more fundamental , i.e. to suggest a fundamental limitation of the generalized gft formalism itself .
let us be more precise .
the generalised gft formalism of @xcite is based on the idea of enlarging the set of variables on which the gft field depends from group elements alone to include lie algebra elements as well , interpreted as discrete connection and @xmath29 field of a bf - based description of simplicial geometry , which should become manifest at the level of the feynman amplitudes .
the two sets of variables are treated on equal footing and in parallel , while maintaining the gft an ordinary , even if non - local with respect to the combinatorics of arguments , field theory .
this implies two choices : first , treating @xmath14 as an ordinary vector space , thus neglecting its non - commutative nature , and , second , relaxing the conjugate nature ( at the classical level ) of @xmath29 and @xmath45 variables in the definition of the model ( which characterizes bf theory ) , and more specifically in the choice of kinetic term .
some consequences of these choices from the point of view of simplicial geometry are discussed in @xcite .
the results we have presented concerning the effective dynamics of gft perturbations , in particular the negativeones , suggest that the way the lie algebra @xmath29 variables are introduced and treated in the generalised gft formalism is not correct .
the apparent need to trivialize the kinetic term in the choice of classical solution suggests that only when the conjugate nature of the @xmath29 and @xmath45 variables is implemented , and thus a bf kynematics is chosen , the corresponding classical solutions define an effective ( quantum ) flat spacetime as desired .
the lack of noncommutative structure in the effective action in the lie algebra sector suggests instead that the correct way of introducing the @xmath29 variables in the formalism is not to enlarge the domain space of the gft field , but to use from the start a noncommutative fourier transform ( in particular , the one introduced and studied in @xcite to go from group space to lie algebra space , turning the original formulation of gfts into a noncommutative ( and combinatorially non - local ) one .
all these suggestions are indeed taken into account in the new noncommutative formulation of group field theories developed in @xcite , which , on top of solving several difficulties we encountered here in the study of perturbations , brings many additional bonuses and further insights from the point of view of simplicial geometry and fundamental quantum gravity proper .
in this appendix we introduce in a geometrical way the distribution @xmath82 .
the geometric idea is shown in figure [ fig : cilindri ] , where we have suppressed one dimension . over each corona
, we build a cylindrical one such that each one is built around a 2-sphere of radius @xmath79 , and has thickness @xmath149 . just like the dirac delta function
, we then send @xmath150 to zero keeping the volume of each corona fixed .
+ of course it is possible to define our distribution through smoother functions , e.g. gaussian - like functions centered in @xmath161 , and then follow the same procedure showed above , with the same result .
l. freidel , d. oriti , j. ryan , _ a group field theory for 3-d quantum gravity coupled to a scalar field _ ,
arxiv : gr - qc/0506067 ; d. oriti , j. ryan , _ group field theory formulation of 3-d quantum gravity coupled to matter fields _ , class .
quant . grav .
23 , 6543 ( 2006 ) , arxiv : gr - qc/0602010 s. bilson - thompson , f. markopoulou , l. smolin , _ quantum gravity and the standard model _ , class . quant . grav . 24 , 3975 ( 2007 ) , arxiv : hep - th/0603022 ; s. he , y. wan , _ conserved quantities and the algebra of braid excitations in quantum gravity _ , nucl .
b 804 , 286 ( 2008 ) , arxiv : 0805.0453 [ hep - th ] ; f. markopoulou , i. premont - schwarz , _ conserved topological defects in non - embedded graphs in quantum gravity _ , class .
25 , 205015 ( 2008 ) e - print : arxiv:0805.3175 [ gr - qc ] | we identify classical solutions of a generalised group field theory model in 3 dimensions , and study the corresponding perturbations , deriving their effective dynamics .
we discuss their interpretation as emergent matter fields .
this allows us , on the one hand to test the proposed mechanism for emergence of matter as a phase of group field theory , and on the other hand to expose some limitations of the generalised group field theory formalism . |
in 2008 , bern , carrasco , and johansson discovered a novel set of linear relations satisfied by tree - level color - ordered amplitudes in gauge theories@xcite .
they arrived at these relations by writing the tree - level @xmath0-gluon amplitude as a sum over @xmath5 diagrams assembled from cubic vertices = _ i c_i n_i d_i [ cubicdecompintro ] where the color factor @xmath6 associated with the diagram is composed of group theory structure constants @xmath7 , the denominator @xmath8 consists of the product of the inverse propagators associated with the diagram , and the kinematic numerator @xmath9 depends on the momenta and polarizations of the gluons .
all contributions from diagrams with quartic vertices are redistributed among the cubic diagrams . by virtue of the jacobi identity
@xmath10 satisfied by the structure constants , the color factors @xmath6 obey a set of jacobi relations of the form c_i + c_j + c_k = 0 . because of these linear dependences , the kinematic numerators @xmath9 are not uniquely defined , but can undergo _ generalized gauge transformations _ @xmath11 which leave unchanged @xcite .
the authors of ref .
@xcite conjectured that there exists a generalized gauge in which the kinematic numerators satisfy the same algebraic relations as the color factors ; in particular , they can be made to satisfy kinematic jacobi relations n_i + n_j + n_k = 0 .
[ kinematicjacobi ] from this assumption of _ color - kinematic duality _
, they demonstrated the existence of new relations ( subsequently known as bcj relations ) satisfied by the color - ordered amplitudes @xmath12 .
these relations can be derived from the _ fundamental bcj relation _ ( and permutations thereof ) @xcite _ b=3^n ( _ c=1^b-1 k_2 k_c ) a(1 , 3 , , b-1 , 2 , b , , n ) = 0 [ fundbcjintro ] where @xmath13 are the ( outgoing ) momenta of the gluons . besides color - kinematic duality , these relations rely on the properties of the propagator matrix @xcite , constructed from the inverse denominators @xmath14 ( see sec . [
sec : prop ] for a precise definition ) .
specifically , as a consequence of momentum conservation , this @xmath15 matrix has rank @xmath16 , and consequently possesses a set of @xmath17 eigenvectors with eigenvalue zero .
the bcj relations ( [ fundbcjintro ] ) were subsequently proven using string - theory techniques @xcite and bcfw on - shell recursion @xcite , providing evidence for the conjecture of tree - level color - kinematic duality .
bern et al . conjectured that color - kinematic duality also applies to the integrands of loop - level amplitudes @xcite ; while not proven , this conjecture has been tested for @xmath18 supersymmetric yang - mills theory through four loops @xcite , and for pure yang - mills theory through two loops @xcite .
despite the fact that the bcj relations for @xmath0-gluon amplitudes have been definitively established , interest in tree - level kinematic numerators continues , not least because the numerators that are naturally generated by feynman rules generally do not obey the relations ( [ kinematicjacobi ] ) except in the case of four - point amplitudes @xcite .
many approaches have been developed to obtain kinematic numerators that obey color - kinematic duality directly from a lagrangian approach @xcite . in this paper
, we introduce a new set of symmetries obeyed by tree - level gauge - theory amplitudes , associated with each external gluon in the amplitude .
these symmetries act on the color factors @xmath6 of the amplitude , shifting them by momentum - dependent quantities . since color factors do not carry any momentum dependence ,
this is a purely formal operation ; we prove , however , that the tree - level @xmath0-gluon amplitude is invariant under these shifts by writing it in an alternative form known as the _ radiation vertex expansion _
@xcite .
we then show that the bcj relations ( [ fundbcjintro ] ) follow as an immediate consequence of the color - factor symmetry of the @xmath0-gluon amplitude .
although bcj relations have been previously established , our results reveal a more direct connection to the symmetries of the lagrangian formulation of gauge theory and its feynman rules ( gauge and poincar invariance ) and provide a basis for generalizations .
let us describe this symmetry in a bit more detail , reserving a full description for section [ sec : cfs ] . given a tree - level @xmath0-gluon color factor @xmath6 ,
the choice of one of the external gluon legs @xmath19 divides the diagram in two at its point of attachment .
let @xmath20 denote the subset of the remaining legs on one side of this point ; it does not matter which side we choose . the shift of the color factor @xmath6 associated with gluon @xmath19 must satisfy _ a c_i _ c k_a k_c .
[ colorfactorshiftintro ] choosing to sum over the complement of @xmath20 gives the same result ( up to sign ) due to momentum conservation .
the constants of proportionality in are then constrained by requiring that the shifted color factors respect all the jacobi relations satisfied by @xmath6 for any values of the momenta .
consider the case where @xmath19 is one of the legs involved in the jacobi identity ( see fig .
[ fig : one ] ) .
imagine that each of the three graphs in fig .
[ fig : one ] is embedded in a larger tree diagram , the same for each .
denote the color factors associated with each diagram by @xmath21 , where @xmath22 ; these obey the jacobi relation @xmath23 . by
, the shifts of these color factors are _ a = _
( r ) k_a k_(r ) where @xmath24 is the momentum flowing out of each leg . requiring
@xmath25 implies that @xmath26 is independent of @xmath27 , as a result of momentum conservation and masslessness of the gluon .
a more detailed description of the color - factor shifts is given in sec .
[ sec : cfs ] .
the symmetry we have introduced has roots in the radiation symmetry @xcite that underlies the general radiation zero theorem @xcite . in theories with local gauge couplings and spins @xmath28 , all single - photon tree amplitudes vanish if the ratios @xmath29 are all equal , where @xmath13 is the photon momentum , and @xmath30 labels external particles with momentum @xmath31 and charge @xmath32 .
these spin - independent zeros have spin - dependent counterparts where @xmath32 are replaced by numerators @xmath33 , closely related to the kinematic numerators @xmath9 in .
the underlying radiation symmetry refers to invariance under @xmath34 as well as @xmath35 for arbitrary @xmath36 and @xmath37 .
the extension to nonabelian `` charges '' has also been considered and the details behind a nonabelian radiation vertex expansion discussed . the general color - factor symmetry introduced here ,
however , incorporates crucial nonabelian constraints ( jacobi relations ) on @xmath36 , which lead to a complete set of bcj relations , and have not heretofore been developed .
nevertheless , since the color - factor symmetry relies on the presence of massless gauge bosons , we may regard it as a generalized radiation symmetry .
we also introduce in this paper the _ cubic vertex expansion _ of an @xmath0-point amplitude @xmath38 with respect to one of the gluons @xmath19 .
consider the set of cubic diagrams @xmath39 that contribute to the @xmath40-point amplitude of all the particles in @xmath38 except for gluon @xmath19 .
we show that , for any @xmath19 , the amplitude @xmath38 can be written as a triple sum over the legs @xmath27 of the vertices @xmath41 of the cubic diagrams @xmath39 : = _ i _ v 1 _
s=1 ^ 3 d_(a , i , v , s ) . here
@xmath42 is the product of inverse propagators that branch off from leg @xmath27 of vertex @xmath41 of diagram @xmath39 , @xmath43 is the color factor of the @xmath0-point diagram obtained by attaching gluon @xmath19 to leg @xmath27 of vertex @xmath41 of diagram @xmath39 ( as in fig . [
fig : one ] ) , and @xmath44 is the @xmath0-point kinematic numerator associated with that color factor .
the shift of @xmath43 associated with gluon @xmath19 is @xmath45 , where as explained above @xmath46 is independent of @xmath27 .
since the alternative radiation vertex expansion shows that the amplitude @xmath38 is invariant under the color - factor shift , we may conclude from the cubic vertex expansion of @xmath38 that _ i _ v _ ( a , i , v ) _
s=1 ^ 3 d_(a , i , v , s ) n_(a , i , v , r ) = 0 . [ sumoverdeltaintro ]
note that this constraint on the kinematic numerators , less stringent than the kinematic jacobi relations ( which state that @xmath47 for each vertex ) , is nonetheless sufficient to imply the bcj relations ( [ fundbcjintro ] ) .
moreover , unlike the kinematic jacobi relations , the condition ( [ sumoverdeltaintro ] ) is invariant under generalized gauge transformations .
a constraint of precisely the form ( [ sumoverdeltaintro ] ) was derived in refs .
@xcite for the five - gluon amplitude using the monodromy properties of string theory amplitudes .
we show in this paper that more general gauge - theory amplitudes , with both gluons and massless or massive particles in an arbitrary representation of the gauge group and with arbitrary spin @xmath28 , are also invariant under the color - factor symmetry .
consequently , the kinematic numerators of these amplitudes obey a constraint analogous to .
we further show that color - factor symmetry implies bcj relations for the color - ordered amplitudes of a class of @xmath0-point amplitudes involving @xmath48 gluons and a pair of particles in an arbitrary representation of the gauge group and arbitrary spin , as previously conjectured in refs .
@xcite .
bcj relations for the primitive amplitudes of a more general class of amplitudes containing gluons and an arbitrary number of pairs of differently flavored fundamentals ( based on a proper decomposition developed by melia @xcite and johansson and ochirov @xcite ) were conjectured by johansson and ochirov @xcite , and subsequently proven using bcfw on - shell recursion by de la cruz , kniss , and weinzierl @xcite . in a sequel to this paper @xcite , we prove that these bcj relations also follow as a direct consequence of the color - factor symmetry .
the amplitudes of the theory of massless bi - adjoint scalars with cubic interactions@xcite also exhibit invariance under color - factor symmetry , as we show using the cubic vertex expansion . in this case , the color - factor shifts are associated with each massless adjoint scalar in the amplitude . as a consequence ,
we demonstrate the reduced rank of the propagator matrix for the @xmath0-gluon gauge - theory amplitude by deriving the set of its null eigenvectors . finally , we generalize the cubic vertex expansion and color - factor symmetry to loop - level amplitudes containing at least one external gluon .
we exhibit an independent set of shifts that act on the color factors of one - loop @xmath0-gluon amplitudes and which depend on the loop momentum as well as external momenta .
these one - loop amplitudes are invariant under color - factor shifts in theories that admit a color - kinematic - dual representation of numerators .
the color - factor symmetry also implies certain relations among the integrands of one - loop color - ordered amplitudes that were previously uncovered in refs .
@xcite .
the contents of this paper are as follows . in sec .
[ sec : cfs ] we define the color - factor shift for the @xmath0-gluon amplitude and derive the bcj relations as a consequence of the invariance of the amplitude under this shift .
we also introduce the cubic vertex expansion , and use it to derive a gauge - invariant constraint on the kinematic numerators of the @xmath0-gluon amplitude .
we introduce an analogous set of shifts of the kinematic numerators , and show that they correspond to a generalized gauge transformation . in sec .
[ sec : fourgluon ] , we prove the invariance of the four - gluon amplitude under the color - factor symmetry , and in sec . [
sec : rvegluon ] , we extend this to the @xmath0-gluon amplitude by using the radiation vertex expansion . in sec .
[ sec : fund ] , we define the color - factor shift for more general amplitudes , and derive the bcj relations for the class of amplitudes containing @xmath48 gluons and a pair of fundamentals . in sec . [
sec : fourfund ] , we prove the invariance of the four - point amplitude with two gluons and a pair of massive fundamentals of arbitrary spin @xmath28 under the color - factor symmetry , and in sec . [
sec : rvefund ] , we extend this to a general @xmath0-point amplitude containing gluons and fundamentals . in sec .
[ sec : prop ] , we prove the invariance of the amplitudes of the bi - adjoint scalar theory under the color factor symmetry , and derive the null eigenvectors of the propagator matrix . in sec .
[ sec : loop ] , we generalize the cubic vertex expansion and color - factor symmetry to loop - level amplitudes , and derive a constraint on the integrands of one - loop color - ordered amplitudes .
section [ sec : concl ] contains a discussion and conclusions . in appendix
[ sec : app ] , we write the shifts for all the color factors of the five - gluon amplitude , and derive the explicit constraint on the kinematic numerators that follow from the color - factor symmetry .
[ sec : cfs ] we begin this section by introducing the color - factor symmetry in the simplest context , the tree - level four - gluon amplitude _ 4 = c_s n_s s + c_t n_t t + c_u n_u u [ fourpointamplitude ] where c_s = f__1 _ 2 f__3 _ 4 , c_t = f__1 _ 4 f__2 _ 3 , c_u = f__1 _ 3 f__4 _ 1 and @xmath49 , @xmath50 , and @xmath51 are mandelstam variables .
we define the four - point color - factor shift to act as c_s c_s + s , c_t c_t + t , c_u
c_u + u [ fourpointcfs ] where @xmath36 is arbitrary . preserves the jacobi relation @xmath52 by virtue of momentum conservation @xmath53 .
the statement that is invariant under implies the kinematic jacobi relation n_s+
n_t+n_u = 0 .
[ fourpointkinematicjacobi ] it is well - known @xcite that is satisfied in the case of the four - gluon amplitude , and we will show this explicitly in sec .
[ sec : fourgluon ] .
this serves as proof of the invariance of the four - gluon amplitude under the color - factor shift .
recall that the kinematic numerators @xmath9 are not uniquely defined by because a generalized gauge transformation n_s n_s + s , n_t n_t + t , n_u
n_u + u [ fourpointggt ] ( with @xmath37 arbitrary ) leaves unchanged by virtue of the jacobi identity @xmath52 . in the case of the four - gluon amplitude , however , the sum @xmath54 is well - defined : it is invariant under the generalized gauge transformation ( [ fourpointggt ] ) due to momentum conservation .
the four - gluon amplitude can be written in terms of color - ordered amplitudes as _ 4 = c_s a(1,2,3,4 ) - c_u a(1,3,2,4 ) .
[ fourgluonbcj ] invariance of under the shift ( [ fourpointcfs ] ) immediately implies _ 4 = s a(1,2,3,4 ) - u
a(1,3,2,4 ) = 0 which is the four - gluon bcj relation @xcite .
next we turn to tree - level @xmath0-gluon amplitudes with @xmath55 , which may be written as a sum over diagrams composed of cubic vertices ( referred to as cubic diagrams ) @xcite = _ i c_i n_i d_i .
[ cubicdecomp ] associated with each cubic diagram @xmath56 is a color factor @xmath6 obtained by sewing together structure constants @xmath57 . among these color factors @xmath6
we may identify the subset of half - ladder color factors @xmath58 defined by _
,_n-3 f__(1 ) _ ( 2 ) _ 1 f__1 _ ( 3 ) _ 2 f__n-3 _ ( n-1 ) _ ( n ) , s_n .
[ halfladder ] the color factors @xmath6 are not independent but obey a set of jacobi relations . using the procedure outlined in ref .
@xcite , the jacobi identity @xmath59 may be repeatedly applied to reduce each @xmath6 to a linear combination of half - ladder color factors c_i = _ s_n-2 m_i , 1n _ 1 n , _
1 n _ 1 ( 2 ) ( n-1 ) n [ dependent ] where @xmath60 denotes a permutation of @xmath61 .
the @xmath62 half - ladders @xmath63 form an independent set .
alternatively , @xmath64 may be computed by rewriting @xmath6 using @xmath65 t^\tc ) $ ] , reducing the resulting expression to a linear combination of single traces , and then identifying the coefficient of @xmath66 ( see ref .
@xcite ) .
we now define a set of momentum - dependent shifts , associated with each external gluon @xmath19 in the amplitude , that act on the color factors @xmath6 .
the action of the shift @xmath67 associated with gluon @xmath19 is constrained by two requirements : ( i ) that it preserve all the jacobi relations satisfied by @xmath6 , and ( ii ) that it satisfy _ a c_i _ c k_a k_c [ colorfactorshift ] where @xmath20 denotes the subset of the external particles on one side of the point at which @xmath19 is attached to @xmath6
. in particular , if @xmath6 is a color factor in which gluon @xmath19 is attached to an external leg @xmath68 , the shift is proportional to @xmath69 , which is an inverse propagator in the associated feynman diagram .
more generally , is related ( see ) to the propagators in the feynman diagram associated with @xmath6 .
consider the subset of @xmath0-point color factors obtained from a given @xmath40-point cubic diagram @xmath39 by attaching gluon @xmath19 to it in all possible ways .
one of these color factors has gluon @xmath19 attached to external leg @xmath70 of the @xmath40-point diagram ] ; define its shift to be @xmath71 .
one may easily verify ( using the argument in the introduction ) that the conditions ( i ) and ( ii ) above uniquely fix the coefficients of the shifts of all the other color factors in this subset .
the coefficients @xmath72 for different @xmath40-point diagrams are then constrained by jacobi relations among their color factors .
\sigma(b+1 ) \cdots n } $ ] and @xmath73 , width=453 ]
we now demonstrate that there is an @xmath16-parameter family of color - factor shifts associated with each gluon @xmath19 in the @xmath0-gluon amplitude .
first choose @xmath74 , and consider the subset of half - ladder color factors @xmath75 , where @xmath76 denotes a permutation of @xmath77 .
we define the color - factor shift associated with gluon @xmath19 to act on these half ladders as _ a _ 1 a ( 2 ) ( n-1 ) n = _ a , k_a k_1 [ colorshiftdefinition ] where @xmath78 are a set of @xmath16 arbitrary , independent constants ( or functions ) for each @xmath19 .
let @xmath79 \sigma(b+1 ) \cdots n}$ ] denote the color factor shown in fig .
[ fig : two ] ; its shift is proportional to @xmath80 .
this together with and the jacobi relation c_1 ( 2 ) ( b-1 ) [ a(b ) ] ( b+1 ) ( n-1 ) n = _ 1 ( 2 ) ( b-1 ) a ( b ) ( n-1 ) n - _ 1 ( 2 ) ( b ) a ( b+1 ) ( n-1 )
n implies that @xmath81 acts on the independent half - ladder color factors @xmath82 as , replace @xmath83 with @xmath84 , and the sum over @xmath30 should begin with 3 . in the case
@xmath85 , replace @xmath86 with @xmath87 . ] _ a _ 1 ( 2 ) ( b-1 ) a ( b ) ( n-1 ) n = _ a , ( k_a k_1 + _ c=2^b-1 k_a k_(c ) ) , a , b \{2 , , n-1 } , b a [ halfladdershift ] consistent with .
the action on the remaining color factors is given by _ a c_i = _ s_n-2 m_i , 1n _ a _ 1 n [ dependentshift ] which is also consistent with , as may be shown using the procedure described in ref .
@xcite .
the color - factor shifts associated with the gluons @xmath88 are not all independent . in particular , the @xmath16-parameter family of shifts associated with gluon @xmath89 are linear combinations of shifts associated with @xmath90 as a result of momentum conservation . , but we know the result must be true for all @xmath0 because , as we will see in sec . [ sec : prop ] , the color - factor shifts correspond to null eigenvectors of the propagator matrix . since the @xmath91 propagator matrix has rank @xmath16 @xcite there are at most @xmath17 independent null eigenvectors .
] we may also define @xmath16-parameter families of shifts associated with gluons 1 and @xmath0 .
these are also not independent of the others .
thus the dimension of the ( abelian ) group of color - factor shifts is @xmath17 . by using
, the tree - level @xmath0-gluon amplitude ( [ cubicdecomp ] ) may be rewritten in the del duca - dixon - maltoni half - ladder decomposition @xcite = _ s_n-2 _ 1 n a(1 , ( 2 ) , , ( n-1 ) , n ) [ ddm ] where the coefficients a(1 , ( 2 ) , , ( n-1 ) , n ) = _ i m_i , 1n n_i d_i [ colorordered ] are color - ordered amplitudes belonging to the kleiss - kuijf basis @xcite . in the previous subsection
, we defined the action of the color - factor symmetry associated with a given gluon @xmath19 .
the variation of under the shift associated with @xmath92 gives = _ s_n-3 _ 2 , _
b=3^n ( k_1 k_2 + _ c=3^b-1 k_2 k_(c ) ) a(1 , ( 3 ) , , ( b-1 ) , 2 , ( b ) , , ( n-1 ) , n ) . in secs .
[ sec : fourgluon ] and [ sec : rvegluon ] , we prove that @xmath38 is invariant under this shift .
since @xmath93 are arbitrary and independent , this implies that _ b=3^n ( k_1 k_2 + _ c=3^b-1 k_2 k_(c ) ) a(1 , ( 3 ) , , ( b-1 ) , 2 , ( b ) , , ( n-1 ) , n ) = 0 [ fundbcj ] which is the fundamental bcj relation ( [ fundbcjintro ] ) . all other permutations of this relation
can be obtained using the invariance of the amplitude under the color - factor shifts associated with gluons 3 through @xmath89 .
it is known @xcite that the bcj relations reduce the number of independent color - ordered amplitudes from @xmath62 to @xmath94 .
not surprisingly , the difference between these , @xmath17 , is the dimension of the group of color - factor shifts that leave the amplitude invariant . in order to examine the implications of the color - factor symmetry for the kinematic numerators @xmath9 appearing in the cubic decomposition ( [ cubicdecomp ] ) , we introduce in this section the _ cubic vertex expansion _ of the amplitude with respect to one of the gluons .
this expansion is similar to , but distinct from , the radiation vertex expansion @xcite that will be used in secs .
[ sec : rvegluon ] and [ sec : rvefund ] to prove the invariance of @xmath0-point gauge - theory amplitudes under color - factor shifts .
the cubic decomposition ( [ cubicdecomp ] ) is a sum over the @xmath5 cubic diagrams of an @xmath0-gluon amplitude , but for any @xmath95 it can be viewed as a sum over the @xmath96 cubic diagrams of an @xmath40-point function with external legs @xmath97 , to each of which gluon @xmath19 is attached in @xmath98 different ways .
let us label these @xmath40-point cubic diagrams by @xmath39 and their denominators by @xmath99 .
each @xmath40-point cubic diagram @xmath39 has @xmath100 vertices , the set of which we denote by @xmath101 . for each vertex @xmath102
, we can break @xmath99 into three factors @xmath103 , where @xmath42 is the product of propagators that branch off from leg @xmath27 of the vertex . if leg @xmath27 is an external leg of the diagram , then @xmath104 .
we can attach gluon @xmath19 either to one of the @xmath89 external legs or to one of the @xmath105 internal lines of @xmath39 , yielding altogether @xmath98 of the terms in the sum ( [ cubicdecomp ] ) .
let @xmath106 be the momentum running through one of the internal lines of @xmath39 .
attaching gluon @xmath19 to this line will replace the factor @xmath107 in @xmath99 with @xmath108 .
we split the inverse denominator into two terms using the identity = 1 k^2 ( 2 k_a k ) + 1 ( -2 k_a k ) ( k+k_a)^2 [ splitprop ] and we associate each of the terms on the right hand side of the equation with one of the two vertices to which the internal line is connected .
thus , with this doubling of internal line terms , we now have a total of @xmath109 terms for each @xmath39 ; namely , one term for each of the legs of each of the @xmath100 vertices of @xmath39 .
we label this term by @xmath110 , and write the cubic vertex expansion of the @xmath0-gluon amplitude with respect to gluon @xmath19 as = _ i _ v v_(a , i ) 1 _
s=1 ^ 3 d_(a , i ,
v , s ) [ cve ] where @xmath43 are the color factors @xmath21 in fig .
[ fig : one ] associated with each vertex @xmath111 , @xmath44 are the associated kinematic numerators , and @xmath112 denotes the momentum flowing out of leg @xmath27 .
the @xmath43 and @xmath44 are equal to the @xmath6 and @xmath9 in up to signs ( such that @xmath113 ) .
an explicit example of the cubic vertex expansion for the five - gluon amplitude is given in appendix a. the @xmath114 freedom in the definition of @xmath43 is used to make the relative signs in the jacobi relation positive : c_(a , i , v , r ) = 0 .
[ vertexjacobi ] the denominators in each triple also sum to zero k_a k_(a , i , v , r ) = 0 [ vertexmomentum ] by momentum conservation @xmath115 and the masslessness of the gluon @xmath116 . _ a priori _ , however , there is no reason for the kinematic numerators @xmath117 associated with each vertex to sum to zero .
we will see in the next subsection , however , that the color - factor symmetry of the amplitude leads to a constraint on the sum of kinematic numerators .
having introduced the cubic vertex expansion of the @xmath0-gluon amplitude with respect to gluon @xmath19 , we now consider the effect of a color - factor shift @xmath81 on the amplitude . the shift associated with gluon @xmath19 acts on the color factors appearing in the cubic vertex expansion ( [ cve ] ) as _ a c_(a , i , v , r ) = _ ( a , i , v ) k_a k_(a , i , v , r ) where @xmath118 is a linear combination of @xmath78 uniquely determined by the jacobi relations .
these shifts respect by virtue of .
the variation of under this shift gives _
a = 12 _ i _ v v_(a , i ) _ ( a , i , v ) _
s=1 ^ 3 d_(a , i , v , s ) n_(a , i , v , r ) .
[ variation ] we prove that @xmath119 in secs .
[ sec : fourgluon ] and [ sec : rvegluon ] ; hence the color - factor symmetry implies the following constraint on the kinematic numerators _ i _ v v_(a , i ) _ ( a , i , v ) _
s=1 ^ 3 d_(a , i , v , s ) n_(a , i , v , r ) = 0 . [ sumoverdelta ]
now _ if _ the constants @xmath118 were all independent , then we could conclude from this argument that @xmath120 , i.e. that the kinematic numerators necessarily obey jacobi relations .
this would be in conflict with the well - known fact that the kinematic numerators obtained from feynman rules in general do _ not _ satisfy color - kinematic duality , where there is only one term in the sum ( [ sumoverdelta ] ) , may we conclude that @xmath121 . ] ; indeed this is not a valid inference from because the @xmath118 are not independent .
the set of @xmath118 for all the vertices of a given diagram @xmath39 are equal ( up to signs ) because any two adjacent vertices share a common color factor .
in fact , with an appropriate choice of signs for @xmath43 , the @xmath118 may be made independent of @xmath41 .
the @xmath122 for different diagrams @xmath39 are further constrained by the jacobi relations among the color factors of @xmath39 .
while does _ not _ imply that the numerators satisfy the kinematic jacobi relations , it _ does _ impose a set of ( gauge invariant ) conditions that the color - kinematic violations @xmath123 must satisfy .
we wish to emphasize that , while the kinematic jacobi relations @xmath124 are not invariant under generalized gauge transformations ( hence the actual claim of color - kinematic duality is that _ there exists _ a generalized gauge in which they hold true ) , the conditions ( [ sumoverdelta ] ) are invariant under generalized gauge transformations . the argument for this is simple .
a generalized gauge transformation is a transformation @xmath125 that leaves the amplitude ( [ cubicdecomp ] ) unchanged .
hence by starting with @xmath126 and following the steps above ( since the condition @xmath127 is also gauge invariant ) , we obtain the same result ( [ sumoverdelta ] ) except with @xmath44 replaced with @xmath128 . to obtain a more explicit form of , we would need to identify all the linear dependences among the @xmath118 required by the color jacobi relations .
previously , we observed that the number of independent color - factor shifts was @xmath17 , parametrized by constants @xmath78 , where @xmath129 and @xmath76 denotes a permutation of @xmath77 . if we were to write @xmath118 in terms of these independent constants , would yield @xmath17 independent constraints on the @xmath130 . in appendix
a , we carry out this procedure for the five - gluon amplitude .
while the bcj relations ( [ fundbcj ] ) were originally derived as a consequence of the assumption of color - kinematic duality , it was known @xcite from early on that they are equivalent to a set of weaker conditions on the numerators .
the conditions for five - gluon numerators were derived in refs .
@xcite as a consequence of the monodromy properties of string - theory amplitudes .
these conditions are equivalent to , as we show in appendix a. in this section , we have demonstrated that both and the bcj relations ( [ fundbcj ] ) are a consequence of the invariance of the amplitude under the color - factor symmetry .
we have considered the effect on the amplitude of a shift of the color factors .
one may ask what effect an analogous shift of the kinematic numerators would have . considered in refs . @xcite . ]
we show in this section that such a shift is simply a generalized gauge transformation .
let us define the kinematic shift associated with leg @xmath19 , where @xmath74 , on the half - ladder numerator @xmath131 to be _ a _ 1 ( 2 ) ( b-1 ) a ( b ) ( n-1 ) n = _ a , ( k_1 k_a + _ c=2^b-1 k_a k_(c ) ) , a , b \{2 , , n-1 } , b a [ kinematichalfladdershift ] where @xmath76 denotes a permutation of @xmath77 , and @xmath132 are a set of arbitrary constants ( or functions ) .
the action on all other numerators @xmath9 is then defined by _ a n_i = _ s_n-2 m_i , 1n _ a _ 1 n .
[ kinematicdependentshift ] note that we have _ not _ assumed that the @xmath9 obey the jacobi relations ( [ kinematicjacobi ] )
. however , the numerator shifts defined by will satisfy _ a ( n_i + n_j + n_k ) = 0 so if the @xmath9 do satisfy kinematic jacobi relations , the shifted numerators will continue to do so , and if they do not , then the neither will the shifted numerators . now consider the cubic vertex expansion of the @xmath0-gluon amplitude = _ i _ v v_(a , i ) 1 _
s=1 ^ 3 d_(a , i , v , s ) .
the kinematic shift with respect to gluon @xmath19 acts on the numerators appearing in this expression as _
a n_(a , i , v , r ) = _ ( a , i , v ) k_a k_(a , i , v , r ) and therefore on the amplitude itself as _ a = 1 2 _ i _ v v_(a , i ) _ ( a , i , v ) _ s=1 ^ 3 d_(a , i , v , s ) c_(a , i , v , r ) . this vanishes courtesy of ; hence the @xmath0-gluon amplitude is invariant under the shift of kinematic numerators .
this is precisely the definition of a generalized gauge transformation @xcite .
[ sec : fourgluon ] in this section we prove that the tree - level four - gluon amplitude is invariant under the color - factor symmetry . in doing so , we develop some results that will be necessary for our more general proof of the invariance of the @xmath0-gluon amplitude in the next section . the four - gluon amplitude can be constructed from a three - gluon vertex by attaching a fourth gluon to a propagator emanating from each of the legs of the vertex or to the vertex itself . this yields _ 4 = [ fourgluonamplitude ] where c_(1 ) = f__1 _ 4 f__2 _ 3 , c_(2 ) = f__2 _ 4 f__3 _ 1 , c_(3 ) = f__3 _ 4 f__1 _ 2 .
[ fourgluoncolorfactors ] the kinematic numerator @xmath133 receives contributions from the diagram in which gluon 4 is attached to leg @xmath27 as well as from the four - gluon vertex . the color - factor symmetry ( associated with gluon 4 ) acts on the color factors ( [ fourgluoncolorfactors ] ) and the four - gluon amplitude ( [ fourgluonamplitude ] ) as _ 4 = _ 4 k_4 k_r _ 4 _ 4 = 12 _ 4 .
thus , by showing that @xmath134 , we will establish that @xmath135 .
this we now proceed to do .
the three - gluon vertex is t^\tc ) $ ] with @xmath136 , so that @xmath137 = f_{\ta\tb\tc } t^\tc$ ] .
this differs from the standard textbook convention by a factor of @xmath138 .
we use @xmath139 in this paper . ]
- i g 2 f__1 _ 2 _ 3 v^_1 _ 2 _ 3 ( k_1 , k_2 , k_3 ) [ fv ] where v^_1 _ 2 _ 3 ( k_1 , k_2 , k_3 ) = ^_1_2 ( k_2 - k_1)^_3 + ^_2_3 ( k_3 - k_2)^_1 + ^_3_1 ( k_1 - k_3)^_2 [ threegluonvertex ] and @xmath13 are outgoing momenta . in feynman gauge ,
the gluon propagator is @xmath140 .
attaching gluon 4 to leg 1 yields the expression @xmath141 { v_{\nu}^{~\mu_2 \mu_3 } ( k_1 + k_4 , k_2 , k_3 ) } \ , .
{ \nonumber}\end{aligned}\ ] ] the contribution of this diagram to the four - gluon amplitude ( [ fourgluonamplitude ] ) is obtained by contracting with @xmath142 and dividing by @xmath56 .
the first and second terms in the square brackets vanish using @xmath143 .
the third term vanishes due to ( k_1 + k_4)^v_^ _ 2 _ 3 ( k_1 + k_4 , k_2 , k_3 ) _ 2_2 _ 3_3 = 0
[ wardone ] using @xmath144 .
the contribution of the remaining terms of to the kinematic numerator @xmath145 can be written n_(1)|_leg
1_1 v^_2 _ 3 ( k_1 + k_4 , k_2 , k_3 ) _ 2_2 _ 3_3 [ attachleg ] where ( s_r^ ) ^_r
_ = i ( ^_r ^ _ - ^_r ^ _ ) [ s ] are the spin - one angular momentum matrices acting on gluon @xmath27 .
these satisfy the lorentz algebra commutation relations = -i .
attaching gluon 4 directly to the three - gluon vertex , we obtain the four - gluon vertex @xmath146 \ , .
{ \nonumber}\end{aligned}\ ] ] using , this can be recast as v__1 _ 2 _ 3 _ 4^_1 _ 2 _ 3 _ 4 = - i g^2 2 ( c_(1 ) k_1_4 + c_(2 ) k_2_4 + c_(3 ) k_3_4 ) v^_1 _ 2 _ 3 ( k_1 , k_2 , k_3 ) .
[ fourgluonvertexderiv ] the contribution of this vertex to the four - gluon amplitude is obtained by contracting with @xmath142 and dividing by @xmath56 .
this contribution is then parceled out among the three terms in ; the contribution to @xmath145 is n_(1)|_vertex = - g^2 k_4 k_1 _ 4_4 k_1_4 v^_1 _ 2 _ 3 ( k_1 , k_2 , k_3 ) _ 1_1 _ 2_2 _ 3_3 .
[ attachvertex ] the reader may be concerned about the use of for the following reason .
the three - gluon vertex ( [ threegluonvertex ] ) can be rewritten using momentum conservation @xmath147 .
for example , we can eliminate @xmath148 from , writing it as v^_1 _ 2 _ 3 ( k_1 , k_2 , k_3 ) = ^_1_2 ( k_2 - k_1)^_3 + ^_2_3 ( -k_1 - 2 k_2)^_1 + ^_3_1 ( 2k_1 + k_2)^_2 .
[ altthreegluonvertex ] the partial derivatives @xmath149 obtained from differ from those obtained from , so that gives @xmath150 \ , .
\label{altfourgluonvertex}\end{aligned}\ ] ] nonetheless is equal to courtesy of the jacobi relation @xmath151
. simply corresponds to a different way of parceling the four - gluon vertex among the color factors @xmath21 , and the expressions @xmath133 will differ by a generalized gauge transformation ( [ fourpointggt ] ) .
the amplitude ( [ fourgluonamplitude ] ) of course remains unchanged .
using the fact that the three - gluon vertex ( [ threegluonvertex ] ) is linear in momenta , we rewrite using v^_2 _ 3 ( k_1 + k_4 , k_2 , k_3 ) = ( 1 + k_4 k_1 ) v^_2 _ 3 ( k_1 , k_2 , k_3 ) [ taylor ] and then combine to obtain @xmath152 v^{\nu \mu_2 \mu_3 } ( k_1 , k_2 , k_3 ) \ve_{2\mu_2 } \ve_{3\mu_3 } { \nonumber}\\ & \quad\quad + g^2 \ve_{1\mu_1 } \left [ -i \ve_{4\alpha } k_{4\beta } ( s_1^{\alpha\beta } ) ^{\mu_1}_{~~ \nu } \right ] \left ( k_{4\gamma } { \partial \over \partial k_{1\gamma } } \right ) v^{\nu \mu_2 \mu_3 } ( k_1 , k_2 , k_3 ) \ve_{2\mu_2 } \ve_{3\mu_3 } \end{aligned}\ ] ] where the orbital angular momentum operators are defined as l_r^ = i ( k_r^ - k_r^ ) .
[ l ] these satisfy the lorentz algebra commutation relations = -i .
similar expressions are obtained for @xmath153 and @xmath154 . finally , defining the total angular momentum operator ( j_r^ ) ^_r _ = l_r^ ^_r _ + ( s_r^ ) ^_r _ [ j ] we can write all the kinematic numerators as @xmath155 v(k_1 , k_2 , k_3 ) \label{nr}\end{aligned}\ ] ] where we have suppressed the polarization vectors @xmath156 , @xmath157 , and @xmath158 .
the subscripts on @xmath159 and @xmath160 indicate on which polarization indices these operators act .
note that under a gauge transformation @xmath161 of gluon 4 , the last two terms in vanish by virtue of the antisymmetry of @xmath162 and @xmath163 , and the first term gives @xmath164 in accord with .
we observe that the operators @xmath165 and @xmath166 are the same as those appearing in the burnett - kroll form @xcite of the subleading terms of the low soft - photon theorem @xcite as applied to gluons @xcite .
( see refs .
@xcite for recent derivations of the soft - gluon theorem from gauge invariance . ) the first two terms in correspond precisely to the leading and subleading terms in the @xmath167 expansion of the four - gluon amplitude ; the third term is higher order in the soft momentum .
we emphasize , however , that , and the vertex expansion that we will derive in the next section , are exact , and not dependent on taking a soft limit .
we will now show that the sum of the four - gluon kinematic numerators ( [ nr ] ) vanishes .
this result , which has been known at least since 1980 @xcite inspired the conjecture of color - kinematic duality @xcite .
we demonstrate it in a way that will facilitate the proof of color - factor symmetry of the @xmath0-gluon amplitude .
using , we see that @xmath168 the first sum on the right - hand side of vanishes by momentum conservation @xmath169 together with @xmath170 .
the second sum on the right hand side of , which may be written more explicitly as ( j_1^ ) ^_1 _ v^_2 _ 3 ( k_1 , k_2 , k_3 ) + ( j_2^ ) ^_2 _ v^_1 _ 3 ( k_1 , k_2 , k_3 ) + ( j_3^ ) ^_3 _ v^_1 _ 2 ( k_1 , k_2 , k_3 ) [ secondsum ] is the first - order lorentz transformation of the three - gluon vertex .
this vanishes , as may be verified by explicit computation , because @xmath171 is a lorentz tensor .
alternatively , we can define the spin - one angular momentum operator to act on polarization indices @xcite s_r^ = i ( _ r^ - _ r^ ) [ alts ] in which case ( j_r^ ) v(k_1 , k_2 , k_3 ) = i ( k_r^ - k_r^ + _ r^ - _ r^ ) v(k_1 , k_2 , k_3 ) [ lorentzscalar ] where v(k_1 , k_2 , k_3 ) = _ 1 _ 2 _ 3 ( k_2 - k_1 ) + _ 2 _ 3 _ 1 ( k_3 - k_2 ) + _ 3 _ 1 _ 2 ( k_1 - k_3 ) . vanishes because @xmath172 is a lorentz - invariant function of @xmath173 and @xmath174 . the third sum in is proportional to @xmath175 this expression is antisymmetric in @xmath37 and @xmath60 .
since it multiplies @xmath176 in , the whole expression vanishes .
the cancellations that we have just exhibited were originally used in ref .
@xcite to prove the radiation zero theorem .
we have shown that the sum of kinematic numerators vanishes ( the kinematic jacobi relation ) , and thus have demonstrated the invariance of the four - gluon amplitude @xmath177 under the color - factor shift symmetry .
we would like to emphasize that in proving the vanishing of we did not use that @xmath173 were on - shell , nor did we use @xmath178 for @xmath22 .
thus we will be able to use these results in sec .
[ sec : rvegluon ] for an off - shell three - gluon vertex @xmath171 .
[ sec : rvegluon ] we now turn to the proof that the tree - level @xmath0-gluon amplitude is invariant under a color - factor shift associated with any of the gluons . to do so
, we employ a decomposition of the amplitude known as the radiation vertex expansion @xcite .
this is a recursive approach which constructs an @xmath0-point amplitude by attaching a massless vector boson to all possible @xmath40-point diagrams .
consider the set of all tree - level @xmath40-gluon diagrams with external legs @xmath97 constructed using feynman rules .
label these diagrams by @xmath179 .
please note that this set differs from the set of @xmath40-point diagrams appearing in the cubic vertex expansion ( cf .
[ sec : cfs ] ) because it includes not only cubic diagrams but also diagrams with four - gluon vertices .
( in the cubic decomposition ( [ cubicdecomp ] ) , all diagrams containing quartic vertices are redistributed among the cubic diagrams . )
we will construct all possible @xmath0-gluon diagrams by attaching gluon @xmath19 to each @xmath40-gluon diagram @xmath179 in all possible ways .
this includes : ( 1 ) attaching gluon @xmath19 to an external leg , ( 2 ) attaching gluon @xmath19 to an internal line , or ( 3 ) attaching gluon @xmath19 to one of the three - gluon vertices of @xmath179 to make a four - gluon vertex .
we describe these in turn .
first we single out one of the external legs , @xmath68 , of @xmath179 , denoting the resulting expression as _
b_b b^_b__b ( k_b , ) where @xmath180 and @xmath181 are the momentum and polarization vector of gluon @xmath68 , and @xmath182 denotes momenta belonging to gluons @xmath183 . attaching gluon @xmath19 to external leg @xmath68 and using we obtain @xmath184 b_{\tb \nu } ( k_a + k_b , \cdots ) \ , . { \nonumber}\end{aligned}\ ] ]
contracting this with @xmath185 eliminates the first two terms in the square brackets .
the third term is proportional to ( k_a + k_b)^ b _ ( k_a + k_b , ) .
[ wardtwo ] this term does not vanish by itself as it did in the case of the four - gluon amplitude , but when we obtain the total @xmath0-gluon amplitude by including all @xmath40-point diagrams @xmath179 , the sum of such terms vanishes due to gauge invariance ( ward identity )
. therefore we are left with the last two terms in square brackets , which yield the two terms in the following expression - 2 g f__b _ a ( k_a + k_b)^2 _ b_b b _
( k_a + k_b , ) .
[ attachexternalgluon ] we set this expression aside for now .
next we single out one of the internal lines of @xmath179 , which divides the diagram into two subdiagrams @xmath186 and @xmath187 , and splits the external legs @xmath188 into two complementary sets @xmath189 and @xmath190 .
the contribution of the diagram can thus be written as b_^ ( -k , ) ( -i _ ) _
k^2 c_^ ( k , ) where @xmath191 is the momentum running through the line , and the @xmath182 in @xmath186 and @xmath187 denote momenta belonging to @xmath189 and @xmath190 respectively . attaching gluon @xmath19 to the line connecting the two subgraphs yields
b _ ( -k , ) f__a v^_a ( k ,- k - k_a , k_a ) k^2 ( k+k_a)^2 c_(k+k_a , ) .
writing the three - gluon vertex ( [ threegluonvertex ] ) as v^_a ( k , -k - k_a , k_a ) = - ^ k_a^_a + ^_a k^ + ^_a ( k + k_a)^ - 2 ^ k^_a - 2 ( ^_a k_a^- ^_a k_a^ ) we see that the first term vanishes when contracted with @xmath192 .
the second and third terms give terms proportional to k^b _
( -k , ) , ( k+ k_a)^c _ ( k + k_a , ) .
[ wardthree ] again these terms do not vanish by themselves , but when we include all @xmath40-gluon diagrams @xmath179 , @xmath193 will be replaced by the sum over all diagrams containing external legs @xmath189 plus one additional off - shell line , and similarly for @xmath194 , and these expressions will vanish by gauge invariance ( ward identity ) .
we are thus left with - 2 i g b _ ( -k , ) f__a k^2 ( k+k_a)^2 c _ ( k+k_a , ) .
now we use the identity ( [ splitprop ] ) to rewrite this as @xmath195 c_{\tc\nu } ( k+k_a , \cdots ) \right\ } \label{attachinternalgluon } \\ & + \left\ { - \sqrt2 g { f_{\tb \tc \ta_a } \over 2 k_a \cdot k } b_{\tb \mu } ( \cdots , -k ) \big [ \eta^{\mu \nu } \ve_a \cdot k + \left ( \ve_a^\mu k_a^\nu - \ve_a^\nu k_a^{\mu } \right ) \big ] \right \}{-i \over ( k+k_a)^2 } c_{\tc\nu } ( k+k_a , \cdots ) \ , .
{ \nonumber}\end{aligned}\ ] ] we associate each of the terms in this equation with one of the two vertices to which the line is attached . in the previous subsections , we showed that attaching gluon @xmath19 to an @xmath40-gluon diagram yields one term ( [ attachexternalgluon ] ) for each external leg and two terms ( [ attachinternalgluon ] ) for each internal line , or in other words , one term for each leg of each vertex of @xmath179 .
we can therefore reorganize these terms into a sum over the legs of the vertices of each of the @xmath40-gluon diagrams @xmath179 .
this is the radiation vertex expansion @xcite .
first we choose one of the three - gluon vertices @xmath41 of @xmath179 ( if it has any ) .
such a vertex divides the external legs into three non - overlapping subsets @xmath196 , @xmath197 such that @xmath198 .
the contribution of diagram @xmath179 to the @xmath40-gluon amplitude can be expressed -i g 2 f__1_2_3 v^_1 _ 2 _ 3 ( k_1 , k_2 , k_3 ) _
r=1 ^ 3 a^__r _
r ( - k_r , ) [ threegluonfactors ] where @xmath199 is the momentum flowing out of each leg of the vertex , and @xmath182 in @xmath200 denotes momenta belonging to @xmath201 .
if any of the legs is external , then @xmath202 is just @xmath203 .
next we choose one of the four - gluon vertices @xmath41 of @xmath179 ( if it has any ) .
such a vertex divides the external legs into four non - overlapping subsets @xmath208 , @xmath209 such that @xmath210 .
the contribution of the diagram @xmath179 can be expressed as v__1 _ 2 _ 3 _ 4^_1 _ 2 _ 3 _ 4 _
r=1 ^ 4 a^__r _ r ( - k_r , ) .
we now attach gluon @xmath19 to each of the legs of this four - gluon vertex . from
we obtain @xmath211 \eta_{\nu\lambda } v_{\tb \tc_2 \tc_3 \tc_4}^{\lambda \mu_2 \mu_3 \mu_4 } { \nonumber}\\ + & { f _ { \tb \tc_{2 } \ta_a } \over 2 k_a \cdot k_2 } \big [ \eta^{\mu_2 \nu } \ve_a \cdot k_2 + \left ( \ve_a^{\mu_2 } k_a^\nu - \ve_a^\nu k_a^{\mu_2 } \right ) \big ] \eta_{\nu\lambda } v_{\tc_1 \tb \tc_3 \tc_4}^{\mu_1 \lambda \mu_3 \mu_4 } { \nonumber}\\ + & { f _ { \tb \tc_{3 } \ta_a } \over 2 k_a \cdot k_3 } \big [ \eta^{\mu_3 \nu } \ve_a \cdot k_3 + \left ( \ve_a^{\mu_3 } k_a^\nu - \ve_a^\nu k_a^{\mu_3 } \right ) \big ] \eta_{\nu\lambda } v_{\tc_1 \tc_2 \tb \tc_4}^{\mu_1 \mu_2 \lambda \mu_4 } { \nonumber}\\ + & { f _ { \tb \tc_{4 } \ta_a } \over 2 k_a \cdot k_4 } \big [ \eta^{\mu_4 \nu } \ve_a \cdot k_4 + \left ( \ve_a^{\mu_4 } k_a^\nu - \ve_a^\nu k_a^{\mu_4 } \right ) \big ] \eta_{\nu\lambda } v_{\tc_1 \tc_2 \tc_3 \tb}^{\mu_1 \mu_2 \mu_3 \lambda } \bigg ) \ , .
\label{attachlegfour}\end{aligned}\ ] ] one can not attach gluon @xmath19 to the four - gluon vertex itself since there are no five - gluon vertices . thus the contribution to the radiation vertex expansion from the four - gluon vertex is @xmath212 v_{\tb \tc_2 \tc_3 \tc_4}^{\nu \mu_2 \mu_3 \mu_4 } { \nonumber}\\ + & { f _ { \tb \tc_{2 } \ta_a } \over 2 k_a \cdot k_2 } \big [ \delta^{\mu_2 } _
{ ~\nu } \ve_a \cdot k_2 -i \ve_{a\alpha } k_{a\beta } ( s_2^{\alpha\beta } ) ^{\mu_2 } _ { ~~\nu } \big ] v_{\tc_1\tb \tc_3 \tc_4}^{\mu_1 \nu \mu_3 \mu_4 } { \nonumber}\\ + & { f _ { \tb \tc_{3 } \ta_a } \over 2 k_a \cdot k_3 } \big [ \delta^{\mu_3 } _ { ~\nu } \ve_a \cdot k_3 -i \ve_{a\alpha } k_{a\beta } ( s_3^{\alpha\beta } ) ^{\mu_3 } _ { ~~\nu } \big ] v_{\tc_1 \tc_2 \tb \tc_4}^{\mu_1 \mu_2 \nu \mu_4 } { \nonumber}\\ + & { f _ { \tb \tc_{4 } \ta_a } \over 2 k_a \cdot k_4 } \big [ \delta^{\mu_4 } _ { ~\nu } \ve_a \cdot k_4 -i \ve_{a\alpha } k_{a\beta } ( s_4^{\alpha\beta } ) ^{\mu_4 } _ { ~~\nu } \big ] v_{\tc_1 \tc_2 \tc_3 \tb}^{\mu_1 \mu_2 \mu_3 \nu } \bigg ) \ , .
\label{fourgluoncontrib}\end{aligned}\ ] ] to summarize this section , we have expressed an @xmath0-gluon amplitude as a sum over the vertices of all of the @xmath40-gluon diagrams @xmath179 , comprising a term ( [ threegluoncontrib ] ) for each three - gluon vertex of @xmath179 and a term ( [ fourgluoncontrib ] ) for each four - gluon vertex of @xmath179 .
computing the variation of the radiation vertex expansion of the amplitude under a color - factor shift is somewhat more delicate than calculating the variation of the cubic vertex expansion of the amplitude ( as we did in sec .
[ sec : cfs ] ) because each factor @xmath202 in can contain more than one color factor @xmath6 due to the possible presence of four - gluon vertices .
first let us consider the contribution ( [ threegluoncontrib ] ) of a three - gluon vertex to the radiation vertex expansion , and for the moment let us assume that the subdiagrams corresponding to @xmath202 contain only three - gluon vertices .
designate by @xmath213 with @xmath22 the color factor associated with each line of .
thus , for example , @xmath214 is the product of @xmath215 and the structure constants from all the three - gluon vertices in @xmath216 .
these color factors manifestly satisfy @xmath217 .
the variation of @xmath213 under the color - factor shift associated with gluon @xmath19 is _ a c _ = _ ( a,,v ) k_a k_r which preserves @xmath217 .
the variation of under the color - factor shift is therefore proportional to v(k_1 , k_2 , k_3 ) .
[ threesums ] in sec .
[ sec : fourgluon ] , we demonstrated that each of the three terms in vanishes .
if the diagrams corresponding to @xmath202 contain four - gluon vertices , we can use to expand these expressions and then use the argument above to show that each separate contribution will vanish under the color - factor shift .
therefore the contribution of the three - gluon vertices ( [ threegluoncontrib ] ) to the radiation vertex expansion is invariant under the color - factor shift associated with gluon @xmath19 .
second let us consider the contribution ( [ fourgluoncontrib ] ) of a four - gluon vertex to the radiation vertex expansion .
again we begin by assuming that the diagrams corresponding to @xmath202 contain only three - gluon vertices .
we now expand the four - gluon vertices @xmath218 in into several terms , one of which is @xmath219 \left ( \eta^{\nu \mu_3 } \eta^{\mu_2 \mu_4 } -\eta^{\nu \mu_4 } \eta^{\mu_2 \mu_3 } \right ) { \nonumber}\\ + & { f _ { \tb \tc_{2 } \ta_a } f_{\tc_{1 } \tb \td } f_{\td \tc_3 \tc_4 } \over 2 k_a \cdot k_2 } \big [ \delta^{\mu_2}_{~ \nu } \ve_a \cdot k_2 -i \ve_{a\alpha } k_{a\beta } ( s_2^{\alpha\beta } ) ^{\mu_2}_{~~ \nu } \big ] \left ( \eta^{\mu_1 \mu_3 } \eta^{\nu \mu_4 } -\eta^{\mu_1 \mu_4 } \eta^{\nu \mu_3 }
\right ) { \nonumber}\\ + & { f _ { \tb \tc_{3 } \ta_a } f_{b \tc_{4 } \td } f_{\td \tc_1 \tc_2 }
\over 2 k_a \cdot k_3 } \big [ \delta^{\mu_3}_{~ \nu } \ve_a \cdot k_3 -i \ve_{a\alpha } k_{a\beta } ( s_3^{\alpha\beta } ) ^{\mu_3}_{~~ \nu } \big ] \left ( \eta^{\mu_1 \nu } \eta^{\mu_2 \mu_4 } -\eta^{\mu_1 \mu_4 } \eta^{\mu_2 \nu }
\right ) { \nonumber}\\ + & { f _ { \tb \tc_{4 } \ta_a } f_{\tc_{3 } \tb \td } f_{\td \tc_1 \tc_2 }
\over 2 k_a \cdot k_4 } \big [ \delta^{\mu_4}_{~ \nu } \ve_a \cdot k_4 -i \ve_{a\alpha } k_{a\beta } ( s_4^{\alpha\beta } ) ^{\mu_4}_{~~ \nu } \big ] \left ( \eta^{\mu_1 \mu_3 } \eta^{\mu_2 \nu } -\eta^{\mu_1 \nu } \eta^{\mu_2 \mu_3 } \right ) \bigg ) \ , . \label{expandedfourgluoncontrib}\end{aligned}\ ] ] designate by @xmath213 with @xmath220 the color factor associated with each line of , including the structure constants from all the three - gluon vertices in @xmath221 .
these color factors satisfy @xmath222 by virtue of f _ _ 1 _ a f_b _ 2 f__3 _ 4 + f _ _ 2 _ a f__1 f__3 _ 4 + f _ _ 3 _ a f_b _ 4 f__1 _ 2 + f _ _ 4 _ a f__3 f__1 _ 2 = 0 . the variation of @xmath213 under the color - factor shift associated with gluon @xmath19 is _ a c _
= _ ( a,,v ) k_a k_r
. the variation of under the color - factor shift therefore contains two sums .
the first ( _ a k_r ) ( ^_1 _ 3 ^_2 _ 4 -^_1 _ 4 ^_2 _ 3 ) [ variationfourglounleading ] vanishes by momentum conservation , @xmath223 , together with @xmath224 .
the second @xmath225 is the first - order lorentz transformation of the tensor @xmath226 , which vanishes .
the variation under the color - factor shift of the other two terms from the expansion of @xmath218 similarly vanishes .
furthermore , the same argument applies when the diagrams corresponding to @xmath202 contain four - gluon vertices , by expanding these expressions using .
therefore the contribution of the four - gluon vertices ( [ fourgluoncontrib ] ) to the radiation vertex expansion is invariant under the color - factor shift associated with gluon @xmath19 . in fine
, we have shown that each contribution to the radiation vertex expansion is invariant under the color - factor shift associated with gluon @xmath19 , and therefore the entire @xmath0-gluon amplitude is invariant under this shift .
[ sec : fund ] in secs . [ sec : fourgluon ] and [ sec : rvegluon ] , we proved the color - factor symmetry of @xmath0-gluon amplitudes , from which follow the bcj relations for color - ordered amplitudes .
color - factor symmetry is a property of a much larger class of tree - level gauge - theory amplitudes , namely those containing at least one gluon together with massless or massive particles in arbitrary representations of the gauge group with arbitrary spin @xmath28 , with the usual gauge - theory couplings .
we will establish the invariance of this larger class of gauge - theory amplitudes under a color - factor shift in secs .
[ sec : fourfund ] and [ sec : rvefund ] . consider a tree - level @xmath0-point gauge - theory amplitude @xmath38 with gluons as well as particles @xmath227 and @xmath228 , either massless or massive , with spin zero , one - half , or one , in an arbitrary representation of the gauge group .
for convenience , we refer to @xmath227 ( and @xmath228 ) as fundamentals ( and antifundamentals ) , but they can be in any representation .
this amplitude has the cubic decomposition = _ i c_i n_i d_i [ cubicdecompfund ] where we decorate the color factors , kinematic numerators , and denominators with primes to distinguish them from the analogous quantities for @xmath0-gluon amplitudes .
the denominator @xmath229 now consists of the product of inverse propagators for both massless and massive particles .
the color factor @xmath230 associated with each cubic diagram is obtained by sewing together @xmath231 vertices @xmath232 and @xmath233 vertices @xmath234 where @xmath235 denote the generators in the appropriate representation .
contributions from feynman diagrams with quartic vertices ( either @xmath236 or @xmath237 in the case of a scalar or vector @xmath227 ) are parceled out among the cubic diagrams
. the number of cubic diagrams in the sum ( [ cubicdecompfund ] ) will generally be fewer than for the @xmath0-gluon amplitude , as some will be excluded for violating fermion number or flavor symmetry .
just as in the case of the @xmath0-gluon amplitude , the amplitude @xmath38 can be written in a cubic vertex expansion with respect to gluon @xmath19 : = _ i _ v 1 _
s=1 ^ 3 d_(a , i , v , s ) . the only difference between the derivation of this expression and that for the @xmath0-gluon amplitude given in sec .
[ sec : cfs ] is that we must use the modified identity = 1 ( 2 k_a k ) + 1 ( -2 k_a k ) [ ( k+k_a)^2 - m^2 ] [ massivesplitprop ] when gluon @xmath19 is attached to a propagator of a massive field . as usual , the color - factor shift associated with gluon @xmath19
is defined by two requirements : ( i ) that it satisfy all the algebraic symmetries ( jacobi relations ) obeyed by the color factors @xmath230 , and ( ii ) that it satisfy _ a c_i _ c k_a k_c where @xmath20 denotes the subset of the external particles on one side of the point at which @xmath19 is attached to @xmath230 .
these together imply that _ a c_(a , i , v , r ) = _ ( a , i , v ) k_a k_(a , i , v , r ) .
therefore the invariance of the amplitude under the color - factor shift implies the constraint _
i _ v _ ( a , i , v ) _
s=1 ^ 3 d_(a , i , v , s ) n_(a , i , v , r ) = 0 on the sums of kinematic numerators appearing in the cubic decomposition . as in the case of @xmath0-gluon amplitudes ,
color - factor symmetry can be used to derive bcj relations among the color - ordered amplitudes associated with @xmath38 .
bcj relations for @xmath0-point amplitudes with gluons and a single pair of massive fundamentals were conjectured in ref .
@xcite and more generally for amplitudes containing an arbitrary number of pairs of fundamentals in ref .
@xcite , based on the assumption of color - kinematic duality .
a proof of these bcj relations using bcfw on - shell recursion was given in ref .
@xcite . in order to derive these relations , however , it is necessary to write the amplitude in a _ proper _ decomposition @xcite , in terms of an independent set of color factors and gauge - invariant primitive amplitudes . for general amplitudes ,
this is a subtle problem , which was recently solved for the case of multiple pairs of distinct - flavor fundamentals by melia @xcite and johansson and ochirov @xcite . in a sequel to this paper @xcite , we review their solution , and then derive the bcj relations using the color - factor symmetry .
there is one class of amplitudes , however , for which the story is practically identical to the @xmath0-gluon case , namely , @xmath0-point amplitudes with @xmath48 gluons and a single pair of fundamentals .
for that case , an independent set of @xmath62 color factors is given by the half - ladders _1 n ( t^_(2)t^_(3 ) ^_(n-1 ) )
^_1 _ _ n .
all other color factors @xmath238 can be reduced to half ladders c_i = _ s_n-2 m_i ,
1n _1 n by repeatedly applying @xmath239^{\textsf{i}}_{~\textsf{j } } \ , , $ ] similar to the case of @xmath0-gluon amplitudes @xcite .
the coefficients @xmath64 are precisely the same as in the @xmath0-gluon case . the @xmath0-point amplitude can then be written in a proper decomposition @xcite ( _ 1 , g_2 , g_3 , , g_n-1 , _
n ) & = & _ s_n-2 _1 n a ( 1 , ( 2 ) , , ( n-1 ) , n ) [ ddmfund ] where the primitive amplitudes are given by a ( 1 , ( 2 ) , , ( n-1 ) , n ) & = & _
i m_i , 1n n_i d_i .
we define an @xmath16-parameter family of shifts associated with each gluon @xmath74 via @xmath240 where @xmath60 is a permutation of @xmath61 , @xmath241 is a permutation of @xmath77 , and @xmath78 is a set of @xmath16 arbitrary constants for each @xmath19 . as in the case of the @xmath0-gluon amplitude , the dimension of the ( abelian ) group of color - factor shifts is @xmath17 .
we show in secs .
[ sec : fourfund ] and [ sec : rvefund ] that the amplitude @xmath242 is invariant under the color - factors shifts ( [ halfladdershiftfund ] ) . as a consequence , the color - ordered amplitudes defined in obey bcj relations that have the same form ( when expressed in terms of invariants @xmath69 , where @xmath13 is the momentum of a gluon ) as those for the @xmath0-gluon amplitude , namely _
b=3^n ( k_1 k_2 + _ c=3^b-1 k_2 k_(c ) ) a ( 1 , ( 3 ) , , ( b-1 ) , 2 , ( b ) , , ( n-1 ) , n ) = 0 together with all permutations of this equation with @xmath243 replaced by @xmath19 , as conjectured in refs . @xcite .
[ sec : fourfund ] in this section , we prove that the tree - level four - point amplitude with two gluons and two massive particles in the fundamental representation of the gauge group with spin zero , one - half , or one is invariant under the color - factor symmetry .
we will use these results for the proof of the invariance of the more general @xmath0-point amplitude in sec .
[ sec : rvefund ] . the four - point amplitude @xmath244 can be constructed from the @xmath245 vertex by attaching a gluon to a propagator emanating from each of the external legs of the vertex , or ( in the case of a spin - zero or spin - one fundamental ) to the vertex itself .
this yields _ 4 ( _ 1 , _ 2 , g_3 , g_4 ) = [ fourfundamplitude ] where the color factors c_(1 ) = - ( t^_4 t^_3 ) ^_1 _ _ 2 , c_(2 ) = ( t^_3 t^_4 ) ^_1 _ _ 2 , c_(3 ) = f__4 _ 3 ( t^ ) ^_1 _ _ 2 [ fourfundcolorfactors ] obey @xmath246 using @xmath137 = f_{\ta\tb\tc } t^\tc $ ] . the color - factor symmetry ( associated with gluon 4 ) acts on the color factors ( [ fourfundcolorfactors ] ) and the four - point amplitude ( [ fourfundamplitude ] ) as _ 4 = _ 4 k_4 k_r _ 4 _ 4 = 12 _ 4 .
we will establish that @xmath135 by showing that @xmath247 , a result that has long been known @xcite .
we will do this separately for spin zero , spin one - half , and spin one fundamentals .
we begin with the case of a spin - one - half fundamental , which is simpler due to the absence of a @xmath248 vertex .
the @xmath245 vertex . ] and dirac propagator are ( t^_3 ) ^_1 _ _ 2 ^_3 , . attaching gluon 4 to ( fermion ) leg 1 yields the expression @xmath249 \right ] \v^{\mu_3 } \ , .
\label{fermionlegone}\end{aligned}\ ] ] the contribution of this diagram to is obtained by sandwiching between @xmath250 and @xmath251 , contracting with @xmath252 , and dividing by @xmath56 .
the first term in the square brackets vanishes using @xmath253 , leaving n_(1 ) = g^2 @xmath254 satisfy the lorentz algebra commutation relations = -i .
similarly , attaching gluon 4 to ( fermion ) leg 2 yields @xmath255 \right]\,.\end{aligned}\ ] ] using @xmath256 , we obtain n_(2 ) = g^2 n_(3 ) = g^2 _
3_3 |u ( k_1 ) ^ u(-k_2 ) .
the sum of the kinematic numerators is thus @xmath257 the first sum on the right - hand side of this equation vanishes by momentum conservation @xmath169 together with @xmath170 .
the second sum vanishes because ^ ^_3 - ^_3 ^ + ( s_3^ ) ^_3 _ ^ is the first - order lorentz transformation of @xmath258 ( acting on both spinor indices as well as the vector index ) and hence vanishes .
thus the sum of kinematic numerators for the amplitude @xmath244 is zero for spin - one - half fundamentals .
next we turn to the case of spin - zero fundamentals .
the @xmath245 vertex is ( t^_3 ) ^_1 _ _ 2 v^_3 ( k_1 , k_2 , k_3 ) , v^_3 ( k_1 , k_2 , k_3 ) = ( k_1 - k_2)^_3 [ scalarscalargluonvertex ] where @xmath13 are outgoing momenta . the scalar propagator is @xmath259 . thus attaching gluon 4 to ( scalar ) leg 1 yields the expression @xmath260 { v^{\mu_3 } ( k_1 + k_4 , k_2 , k_3 ) } \,.\end{aligned}\ ] ] the contribution of this diagram to is obtained by contracting with @xmath252 and dividing by @xmath56 .
the first term in the square brackets vanishes using @xmath170 , leaving n_(1)|_leg = g^2 _ 4 k_1 _
3_3 v^_3 ( k_1 + k_4 , k_2 , k_3 ) . similarly , attaching gluon 4 to ( scalar ) leg 2 yields n_(2)|_leg = g^2 _ 4 k_2 _ 3_3 v^_3 ( k_1 , k_2+k_4 , k_3 ) . attaching gluon 4 to ( gluon ) leg 3 yields n_(3)|_leg = g^2 _ 3_3 v^ ( k_1 , k_2 , k_3 + k_4 ) .
attaching gluon 4 directly to the @xmath245 vertex , we obtain the @xmath248 vertex ^_3 _ 4 ( \ { t^_3 , t^_4 } ) ^_1 _ _ 2 [ ssgg ] which can be written as - i g^2 2 ( c_(1 ) k_1_4 + c_(2 ) k_2_4 + c_(3 ) k_3_4 ) v^_3 ( k_1 , k_2 , k_3 ) .
[ scalarscalargluongluonvertex ] the rest of the story proceeds exactly as in sec .
[ sec : fourgluon ] , allowing us to write the kinematic numerators as @xmath261 v(k_1 , k_2 , k_3)\end{aligned}\ ] ] where we have suppressed the polarization vector @xmath158 . because two of the legs are scalars ,
we have @xmath262 , whereas @xmath263 is given by . now consider the sum of kinematic numerators @xmath264 where the scalar - scalar - gluon vertex is @xmath265)^{\mu_3}$ ] with @xmath266 arbitrary due to momentum conservation .
the first sum on the right - hand side of this equation vanishes as usual by momentum conservation .
the second sum , which may be written more explicitly as l_1^ v^_3 ( k_1 , k_2 , k_3 ) + l_2^ v^_3 ( k_1 , k_2 , k_3 ) + ( j_3^ ) ^_3 _ v^ ( k_1 , k_2 , k_3 ) [ secondsumscalar ] is the first - order lorentz transformation of @xmath245 vertex .
this vanishes , as may be seen by explicit computation , because @xmath267 is a lorentz tensor .
the third term on the right - hand side of is _ 4 k_4 k_4 ( s_3^)^_3 _ k_3 v^(k_1 , k_2 , k_3 ) = ( k_4 ^ 2 _ 4^_3 - _ 4 k_4 k_4^_3 ) [ thirdtermscalar ] which automatically vanishes due to @xmath268 and @xmath269 .
again , we emphasize that in proving the vanishing of we did not use that @xmath173 were on - shell , nor did we use @xmath178 , for @xmath22 . thus these results remain valid for an off - shell vertex @xmath267 .
finally we consider massive spin - one fundamentals with @xmath270 vertex of the form ( t^_3 ) ^_1 _ _ 2 v^_1 _ 2 _ 3 ( k_1 , k_2 , k_3 ) [ spinonevertex ] where @xmath271 is given by .
we emphasize that although we refer to the vector particles as fundamentals , they could be in any representation , including the adjoint , in which case @xmath272 and is equal to , except that now the vector boson is massive .
the propagator for a massive spin - one particle is , p_(k ) = _ - k_k_m^2 .
[ spinonepropagator ] in the case in which the vector boson gets its mass from a spontaneously - broken symmetry , is the propagator in unitary gauge ; this is most convenient for tree - level calculations as we need not compute contributions involving goldstone bosons . attaching gluon 4 to leg 1 yields the expression v^_1 _ 4 ( k_1 , k_4 , -k_1-k_4 ) p _ ( k_1+k_4 ) v^_2 _ 3 ( k_1 + k_4 , k_2 , k_3 ) .
the contribution of this diagram to is obtained by contracting with @xmath142 and dividing by @xmath56 , giving n_(1)|_leg = g^2 _
1_1 v^_2 _ 3 ( k_1 + k_4 , k_2 , k_3 ) _ 2_2 _ 3_3 where we have used @xmath273 and @xmath274 but we did not use , which is not valid in this case because @xmath275 .
analogous expressions are obtained for @xmath276 and @xmath277 . attaching gluon 4 directly to the @xmath245 vertex
, we obtain the @xmath248 vertex - i g^2 2 ( c_(1 ) k_1_4 + c_(2 ) k_2_4 + c_(3 ) k_3_4 ) v^_1_2_3 ( k_1 , k_2 , k_3 ) .
again , the rest of the story proceeds exactly as in sec .
[ sec : fourgluon ] , allowing us to write the kinematic numerators as @xmath261 v(k_1 , k_2 , k_3)\end{aligned}\ ] ] where we have suppressed the polarization vectors @xmath174 for @xmath22 .
note that the kinematic numerators @xmath278 have exactly the same form as the kinematic numerators for the four - gluon amplitude , even though the masses for particles 1 through 3 can be nonzero .
the proof of the vanishing of the sum of numerators proceeds exactly as in sec .
[ sec : fourgluon ] . in this section
, we have explicitly shown that the sum of kinematic numerators for four - point amplitudes @xmath279 vanishes ( where @xmath227 can have spin zero , one - half , or one ) and thus have demonstrated the invariance of the four - point amplitude under the color - factor symmetry associated with gluon 4 . as the results of the last subsection have shown ,
this result remains valid even when particles 1 through 3 are massive ; only the gluon associated with the color - factor symmetry need be massless .
the results we have derived will be used in the next section to prove a more general result .
[ sec : rvefund ] in this section , we use the radiation vertex expansion to demonstrate the invariance under the color - factor symmetry of tree - level gauge - theory amplitudes containing at least one gluon together with massless or massive particles in arbitrary representations of the gauge group and with arbitrary spin @xmath28 . for concreteness , we focus on the @xmath0-point amplitude @xmath280 with @xmath48 gluons and a pair of fundamentals @xmath227 , but it will be clear that the proof applies to more general amplitudes .
the proof is very similar to that given in sec .
[ sec : rvegluon ] for @xmath0-gluon amplitudes , and so we only highlight the differences .
the radiation vertex expansion constructs the @xmath0-point amplitude @xmath280 by attaching gluon @xmath281 to all possible @xmath40-point diagrams @xmath179 , with two fundamentals and @xmath100 gluons , in all possible ways , and reorganizing this as a sum over all the vertices of the @xmath40-point diagram .
we have already shown in sec .
[ sec : rvegluon ] that the contributions of the three- and four - gluon vertices are invariant under the color - factor symmetry , so we only need to demonstrate the same for vertices involving two fundamentals .
we do this separately for fundamentals with spin zero , one - half , and one .
to derive the contribution of the @xmath245 vertices to the radiation vertex expansion for spin - one - half fundamentals @xmath227 , we examine the effect of attaching a gluon to a fermion leg , either external or internal .
first we single out ( fermion ) leg 1 , denoting the contribution of an @xmath40-point diagram @xmath179 to the amplitude as @xmath282 where @xmath182 denotes momenta belonging to @xmath283 .
attaching gluon @xmath19 to external fermion leg @xmath70 , using @xmath253 , and contracting with @xmath284 , we obtain - 2 g ( t^_a)^_1 _ 2 k_a k_1 |u ( k_1 ) c^ ( k_1+k_a , ) .
[ attachexternalpsibar ] next , we single out ( fermion ) leg @xmath0 , denoting the contribution of the diagram @xmath179 to the amplitude as @xmath285 , where @xmath182 denotes momenta belonging to @xmath286 .
attach gluon @xmath19 to external fermion leg @xmath0 , use @xmath287 , and contract with @xmath284 to obtain + 2 g b _
( k_n + k_a , ) ( t^_a)^ _ _ n 2 k_a k_n u(-k_n ) .
[ attachexternalpsi ] now we single out one of the internal fermion lines of @xmath179 , which divides the diagram into two subdiagrams @xmath288 and @xmath289 , and splits the external legs @xmath290 into two complementary sets @xmath189 and @xmath190 .
the contribution of the diagram @xmath179 can thus be written as b _ ( -k , )
i ^ _ - m c^ ( k , ) where @xmath191 is the momentum running through the line , and the @xmath182 in @xmath186 and @xmath187 denote momenta belonging to @xmath189 and @xmath190 respectively .
attaching gluon @xmath19 to the line connecting the two subgraphs and contracting with @xmath192 , we have - i g2 b _ ( -k , ) 1 - m ( t^_a)^ _ c^ ( k+k_a , ) .
now we use the identity@xcite @xmath291\right ) \over k_a \cdot k } ~-~ { \left ( \ve_a \cdot k + { 1\over 4 } \left [ \epsaslash , \kaslash \right ] \right ) \over k_a\cdot k } { 1 \over \kslash + \kaslash - m } \label{fermionpropidentity}\end{aligned}\ ] ] to rewrite this as @xmath292 c^{\tk } ( k + k_a , \cdots ) \right\ } { \nonumber}\\ & + \left\ { \sqrt2 g { ( t^{\ta_a})^\tj_{~\tk } \over 2 k_a\cdot k } ~b_{\tj } ( -k , \cdots ) \left [ \ve_a \cdot k -i \ve_{a\alpha } k_{a\beta } \sigma^{\alpha\beta } \right ] \right\ } { i \over \kslash + \kaslash - m } c^{\tk } ( k+k_a , \cdots ) \ , .
\label{attachinternalfermion}\end{aligned}\ ] ] each term can be associated with one of the two vertices to which the line is attached .
we now choose one of the @xmath245 vertices @xmath41 of @xmath179 .
such a vertex divides the external legs into three non - overlapping subsets @xmath196 , @xmath197 such that @xmath293 .
the contribution of the diagram @xmath179 to the @xmath40-point amplitude can be expressed as @xmath294 where @xmath199 is the momentum flowing out of each leg of the vertex , and the @xmath182 in @xmath186 , @xmath187 , and @xmath295 denote momenta belonging to @xmath296 , @xmath297 , and @xmath298 respectively . if either fermion leg is external , then @xmath299 or @xmath300 .
if the gluon leg is external , then @xmath301 is @xmath302 .
we now attach gluon @xmath19 to each of the legs of this @xmath303 vertex , either to an external leg or to an internal line .
using the expressions above as well as those in sec .
[ sec : rvegluon ] , we obtain @xmath304 \v^{\mu_3 } + { ( t^{\tc_3 } t^{\ta_a } ) ^{\tj_1}_{~~ \tj_2 } \over 2 k_a \cdot k_2 } \v^{\mu_3 } \big [ \ve_a \cdot k_2 + i \ve_{a\alpha } k_{a\beta } \sigma^{\alpha\beta } \big ] { \nonumber}\\ & + { f _ { \ta_a \tc_{3 } \tb } ( t^{\tb } ) ^{\tj_1}_{~~ \tj_2 } \over 2 k_a \cdot k_3 } \big [ \eta^{\mu_3 \nu } \ve_a \cdot k_3 -i \ve_{a\alpha } k_{a\beta } ( s_3^{\alpha\beta } ) ^{\mu_3\nu } \big ] \v_{\nu } \bigg ) c^{\tj_2 } ( -k_2 , \cdots ) \label{fermioncontrib}\end{aligned}\ ] ] which is the contribution of the @xmath303 vertex to the radiation vertex expansion .
we now wish to show that is invariant under the color - factor symmetry . as in sec .
[ sec : rvegluon ] , we first assume that the subdiagrams corresponding to @xmath295 , @xmath186 , and @xmath187 contain no four - gluon vertices .
designate by @xmath213 with @xmath22 the color factors associated with each of the three terms in , including factors of @xmath57 and @xmath305 in the subdiagrams .
these color factors manifestly satisfy @xmath217 , and the variation of @xmath213 under the color - factor shift associated with gluon @xmath19 is _ a c _ = _ ( a,,v ) k_a k_r [ psicolorfactorshift ] which preserves @xmath217 .
the variation of under is then proportional to ( _ a k_r ) ^_3 - i _ a k_a .
[ variationspinhalf ] the first term vanishes by momentum conservation , and the second by the transformation properties of @xmath306 , as we saw in sec .
[ sec : fourfund ] .
if the subdiagrams @xmath295 , @xmath186 , and @xmath187 do contain four - gluon vertices , we can expand into individual pieces , each of which is invariant under the color - factor symmetry .
together with the result from sec .
[ sec : rvegluon ] that the contributions to the radiation vertex expansion from the three- and four - gluon vertices are also invariant , we have thus shown that the amplitude @xmath280 with spin - one - half fundamentals is invariant under the color - factor symmetry .
in fact , this proof applies to an amplitude with an arbitrary number of pairs of fundamentals , and will be used in the sequel @xcite to prove the bcj relations @xcite for that class of amplitudes . to derive the contribution of the @xmath245 and @xmath248 vertices to the radiation vertex expansion for spin - zero fundamentals @xmath227
, we examine the effect of attaching a gluon to a scalar leg , either external or internal .
first we single out ( scalar ) leg 1 , denoting the contribution of an @xmath40-point diagram @xmath179 to the amplitude as @xmath307 where @xmath182 denotes momenta belonging to @xmath283 . attaching gluon @xmath19 to external scalar leg @xmath70
we obtain @xmath308 c^{\tj } ( k_1+k_a , \cdots ) \,.\end{aligned}\ ] ] contracting this with @xmath284 eliminates the first term in the square brackets , leaving - 2 g ( t^_a)^_1 _ 2 k_a k_1 _ a k_1 c^ ( k_1+k_a , ) .
similarly , attaching gluon @xmath19 to external scalar leg @xmath0 , we obtain + 2 g _ a k_n b _ ( k_n + k_a , ) ( t^_a)^ _ _ n 2 k_a k_n .
next we single out one of the internal scalar lines of @xmath179 , which divides the diagram into two subdiagrams @xmath288 and @xmath289 , and splits the external legs into two complementary sets @xmath189 and @xmath190 .
the contribution of the diagram @xmath179 can thus be written as b _ ( -k , ) i ^_k^2-m^2 c^ ( k , ) where @xmath191 is the momentum running through the line , and the @xmath182 in @xmath186 and @xmath187 denote momenta belonging to @xmath189 and @xmath190 respectively . attaching gluon @xmath19 to the line connecting the two subgraphs yields - i g2 b _ ( -k , ) ( t^_a)^ _ v^_a ( k ,- k - k_a , k_a ) c^ ( k+k_a , ) .
contracting with @xmath192 and using @xmath143 , we have - 2 i g b _ ( -k , ) ( t^_a)^ _ _ a k c^ ( k+k_a , ) .
now we use the identity ( [ massivesplitprop ] ) to rewrite this as @xmath309 we associate each of the terms in this equation with one of the two vertices to which the line is attached .
next we choose one of the scalar - scalar - gluon vertices @xmath41 of @xmath179 ( if it has any ) .
such a vertex divides the external legs into three non - overlapping subsets @xmath196 , @xmath197 such that @xmath293 .
the contribution of the diagram @xmath179 can be expressed as @xmath310 where @xmath199 is the momentum flowing out of each leg of the vertex , and the @xmath182 in @xmath186 , @xmath187 , and @xmath295 denote momenta belonging to @xmath296 , @xmath297 , and @xmath298 respectively . if either scalar leg is external , then @xmath311 or @xmath312 .
if the gluon leg is external , then @xmath301 is @xmath302 .
we now attach gluon @xmath19 to each of the legs of this scalar - scalar - gluon vertex , either to an external leg or to an internal line .
this yields @xmath313 v^{\mu_3 } ( k_1 + k_a , k_2 , k_3 ) + { ( t^{\tc_3 } t^{\ta_a } ) ^{\tj_1}_{~~ \tj_2 } \over 2 k_a \cdot k_2 } \big [ \ve_a \cdot k_2 \big ] v^{\mu_3 } ( k_1 , k_2 + k_a , k_3 ) { \nonumber}\\ & + { f _ { \ta_a \tc_{3 } \tb } ( t^{\tb } ) ^{\tj_1}_{~~ \tj_2 } \over 2 k_a \cdot k_3 } \big [ \eta^{\mu_3 \nu } \ve_a \cdot k_3 + \left ( \ve_a^{\mu_3 } k_a^\nu - \ve_a^\nu k_a^{\mu_3 } \right ) \big ] v_{\nu } ( k_1 , k_2 , k_3 + k_a ) \bigg ) c^{\tj_2 } ( -k_2 , \cdots ) \ , .
\label{attachscalarleg}\end{aligned}\ ] ] we can also attach gluon @xmath19 directly to the scalar - scalar - gluon vertex itself . using , this yields @xmath314 we now use in , and combine as we did in sec .
[ sec : fourfund ] . leaving the @xmath315 index implicit
, we obtain the contribution of the scalar - scalar - gluon vertex to the radiation vertex expansion @xmath316 v(k_1 , k_2 , k_3 ) \label{threescalarcontrib } \\ & + { ( t^{\tc_3 } t^{\ta_a } ) ^{\tj_1}_{~~ \tj_2 } \over 2 k_a \cdot k_2 } \left [ \ve_a \cdot k_2 -i \ve_{a\alpha } k_{a\beta } l_2^{\alpha\beta } \right ] v(k_1 , k_2 , k_3 ) { \nonumber}\\ & + { f _ { \ta_a \tc_{3 } \tb } ( t^{\tb } ) ^{\tj_1}_{~~ \tj_2 } \over 2 k_a \cdot k_3 } \left [ \ve_a \cdot k_3 -i \ve_{a\alpha } k_{a\beta } j_3^{\alpha\beta } -i \ve_{a\alpha } k_{a\beta } k_{a\gamma } s_3^{\alpha\beta } { \partial \over \partial k_{3\gamma } } \right ] v(k_1 , k_2 , k_3 ) \bigg ) c^{\tj_2 } ( -k_2 , \cdots ) { \nonumber}\end{aligned}\ ] ] where @xmath317 and @xmath318 act on the @xmath315 index of @xmath319 . at the end of the last subsection , we discussed the color - factor symmetry acting on the @xmath245 vertex contribution to the radiation vertex expansion . under
, the variation of is proportional to v(k_1 , k_2 , k_3 ) .
[ threesumsfund ] in sec .
[ sec : fourfund ] , we demonstrated that each of the three terms in vanishes .
therefore the contributions of the @xmath245 vertices ( [ threescalarcontrib ] ) to the radiation vertex expansion are invariant under the color - factor shift associated with gluon @xmath19 .
finally we choose one of the scalar - scalar - gluon - gluon vertices @xmath41 of @xmath179 ( if it has any ) .
such a vertex divides the external legs into four non - overlapping subsets @xmath208 , @xmath209 such that @xmath210 . using
, the contribution of diagram @xmath179 can be expressed as @xmath320 we now attach gluon @xmath19 to each of the legs of this scalar - scalar - gluon - gluon vertex , either to an external leg or to an internal line .
this yields @xmath321 \eta^{\mu_3 \mu_4 } + { ( \ { t^{\tc_3 } , t^ { \tc_4 } \ } t^{\ta_a } ) ^{\tj_1}_{~~ \tj_2 } \over 2 k_a \cdot k_2 } \big [ \ve_a \cdot k_2 \big ] \eta^{\mu_3 \mu_4 } { \nonumber}\\ & + { f _ { \ta_a \tc_{3 } \tb } ( \ { t^{\tb } , t^ { \tc_4 } \ } ) ^{\tj_1}_{~~ \tj_2 } \over 2 k_a \cdot k_3 } \big [ \delta^{\mu_3 } _ { ~\nu } \ve_a \cdot k_3 -i \ve_{a\alpha } k_{a\beta } ( s_3^{\alpha\beta } ) ^{\mu_3 } _ { ~~\nu } \big ] \eta^{\nu \mu_4 } { \nonumber}\\ & + { f _ { \ta_a \tc_{4 } \tb } ( \ { t^{\tc_3 } , t^ { \tb } \ } ) ^{\tj_1}_{~~ \tj_2 } \over 2 k_a \cdot k_4 } \big [ \delta^{\mu_4 } _ { ~\nu } \ve_a \cdot k_4 -i \ve_{a\alpha } k_{a\beta } ( s_4^{\alpha\beta } ) ^{\mu_4 } _ { ~~\nu } \big ] \eta^{\mu_3 \nu } \bigg ) c^{\tj_2 } ( -k_2 , \cdots ) \ , . \label{fourscalarcontrib}\end{aligned}\ ] ] we now need to consider the variation under the color - factor symmetry of this contribution to the radiation vertex expansion .
designate by @xmath213 with @xmath220 the color factor associated with each of the terms in , including the factors of @xmath57 and @xmath305 in the subdiagrams .
these color factors satisfy @xmath222 by virtue of @xmath322 the variation of @xmath213 under the color - factor shift associated with gluon @xmath19 is _ a c _
= _ ( a,,v ) k_a k_r .
[ scalarshift ] the variation of under is therefore proportional to ( _ a k_r ) ^_3 _ 4 -i _ a k_a .
[ variationquarticspinzero ] , the first term vanishes by momentum conservation , @xmath223 , and @xmath224 .
the second term is the first - order lorentz transformation of the tensor @xmath323 , which vanishes .
thus each vertex involving scalars that contributes to the radiation vertex expansion is invariant under the color - factor shift associated with gluon @xmath19 . together with the result from sec .
[ sec : rvegluon ] that the contributions from the three- and four - gluon vertices to the radiation vertex expansion are also invariant , we have shown that the full amplitude @xmath280 with spin - zero fundamentals is invariant under the color - factor shift .
in fact , any amplitude built with @xmath245 and @xmath324 vertices for scalar @xmath227 will have the color - factor symmetry . to derive the contribution of the @xmath245 and @xmath248 vertices to the radiation vertex expansion for spin - one fundamentals @xmath227
, we examine the effect of attaching a gluon to a massive vector leg , either external or internal .
first we single out ( vector ) leg 1 , denoting the contribution of an @xmath40-point diagram @xmath179 to the amplitude as @xmath325 where @xmath326 denotes momenta belonging to @xmath283 .
attaching gluon @xmath19 to external vector leg @xmath70 we obtain - g 2 ( t^_a)^_1 _ ( k_a + k_1)^2 - m_1 ^ 2 v^_1 _ a ( k_1 , -k_1 - k_a , k_a ) p _ ( k_1+k_a ) c^ ( k_1+k_a , ) . contracting with @xmath327 , we obtain - 2 g ( t^_a)^_1 _ 2 k_a k_1 _ 1_1 c^_(k_1 + k_a , ) [ attachexternalmassivevector ] where we used @xmath328 and @xmath274 .
we did _ not _ use the vanishing of .
similar expressions result from attaching gluon @xmath19 to the other legs . by comparing , we observe that the expression is the same for a massless or a massive vector particle .
next we single out one of the internal lines of @xmath179 , which divides the diagram into two subdiagrams @xmath186 and @xmath187 , and splits the external legs @xmath188 into two complementary sets @xmath189 and @xmath190 .
the contribution of the diagram can thus be written as b_^ ( -k , ) ( -i ^ _ ) p_(k ) k^2 - m^2 c^ ( k , ) where @xmath191 is the momentum running through the line , and the @xmath182 in @xmath186 and @xmath187 denote momenta belonging to @xmath189 and @xmath190 respectively . attaching gluon @xmath19 to the line connecting the two subgraphs and
contracting with @xmath284 , we obtain - i g2 b_^ ( -k , ) ( t^_a)^ _ p _ ( k ) _
a_a v^_a ( k ,- k - k_a , k_a ) p _ ( k+k_a ) c^ ( k+k_a , ) .
now we use the identity @xcite
@xmath329 [ ( k+k_a)^2-m^2 ] } \label{vectorpropidentity } \\ & = { 1 \over k_a \cdot k } \left\ { { - p_{\mu\lambda } ( k ) \left [ \delta^\lambda_{~\nu } \ve_a \cdot k + \ve_a^\lambda k_{a\nu } - k_a^{\lambda } \ve_{a\nu } \right ] \over k^2-m^2 } + { \left [ \delta_\mu^{~\kappa } \ve_a \cdot k + \ve_{a\mu } k_a^\kappa - k_{a\mu } \ve_a^\kappa \right ] p_{\kappa\nu } ( k+k_a ) \over ( k+k_a)^2 -m^2 } \right\ } { \nonumber}\end{aligned}\ ] ] to rewrite this as @xmath330 c^{\tk}_{\nu } ( k+k_a , \cdots ) \right\ } \\ & + \left\ { \sqrt2 g b_{\tj \mu } ( \cdots , -k ) { ( t^{\ta_a})^\tj_{~\tk } \over 2 k_a \cdot k } \big [ \eta^{\mu \kappa } \ve_a \cdot k + \left ( \ve_a^\mu k_a^\kappa - \ve_a^\kappa k_a^{\mu } \right ) \big ] \right \}{-i p_{\kappa\nu } ( k+k_a ) \over ( k+k_a)^2 -m^2 } c^{\tk\nu } ( k+k_a , \cdots ) \ ,
. { \nonumber}\end{aligned}\ ] ] we did _ not _ use the vanishing of .
we associate each of the terms in curly brackets with one of the two vertices to which the line is attached . for the rest of the discussion
, we can be brief .
the expressions for the contributions to the radiation vertex expansion from @xmath245 and @xmath324 vertices are similar to those for spin - zero fundamentals , except that we must include @xmath159 and @xmath160 terms for @xmath331 and 2 .
the proof that these vertex contributions are invariant under the color - factor symmetry relies on some of the results from sec .
[ sec : rvegluon ] .
thus we have shown that the full amplitude @xmath280 with massive spin - one fundamentals is invariant under the color - factor shift .
in fact , any amplitude built with @xmath245 and @xmath324 vertices with a massive vector particle @xmath227 will have the color - factor symmetry .
we can go even further and state that any amplitude built from @xmath231 and @xmath332 vertices ( with @xmath227 having arbitrary spin @xmath28 ) where not only @xmath227 but also some of the gluons are massive ( through spontaneous symmetry breaking ) will be invariant under the color - factor shift .
the only particle in the amplitude that _ must _ be massless is the gluon associated with the color - factor symmetry .
[ sec : prop ] the symmetry that we have introduced in this paper is possessed not only by gauge - theory amplitudes but also by the amplitudes of the much simpler theory @xcite of massless scalars @xmath333 transforming in the adjoint of the color group @xmath334 .
these bi - adjoint scalars have only cubic interactions of the form f _ f_ ^ ^ ^ where @xmath335 and @xmath336 are the structure constants of @xmath337 and @xmath338 .
the tree - level @xmath0-point amplitude is given by the sum over cubic diagrams _
n = _ i c_i c_i d_i .
[ biadjointamp ] using , the bi - adjoint scalar tree amplitude ( [ biadjointamp ] ) can be written as _
n = _ s_n-2 _ s_n-2 _ 1 n m ( 1 n| 1 n ) _ 1 n [ biadjointprop ] where m ( 1 n| 1 n ) = _ i m_i,1n m_i , 1n d_i [ propagator ] are double - partial amplitudes of the bi - adjoint scalar theory @xcite .
the @xmath339 are also the entries of @xmath15 propagator matrix defined in ref .
vaman and yao argued that the propagator matrix has rank @xmath16 by virtue of momentum conservation , using explicit low @xmath0 examples .
cachazo , he , and yuan confirmed this for general @xmath0 by expressing the double - partial amplitudes as a sum over the @xmath16 solutions of the scattering equations @xcite . the cubic vertex expansion of the @xmath0-point bi - adjoint scalar amplitude with respect to external scalar @xmath19 ( see sec .
[ sec : cfs ] ) is given by _
n = _ i _ v v_(a , i ) 1 _
s=1 ^ 3 d_(a , i , v , s ) [ biadjointvertexexpansion ] where c_(a , i , v , r ) = 0 , _ ( a , i , v , r ) = 0 . [ biadjointjacobi ] the color - factor shift with respect to massless scalar @xmath19 acts on the color factors appearing in as _ a c_(a , i , v , r ) = _ ( a , i , v ) k_a k_(a , i , v , r ) . the variation of under this shift _ a _
n = 12 _ i _ v v_(a , i ) _ ( a , i , v ) _
s=1 ^ 3 d_(a , i , v , s ) _ ( a , i , v , r ) vanishes by virtue of , thus establishing that the amplitudes of the bi - adjoint scalar theory possess the color - factor symmetry .
now we consider the variation of the amplitude ( [ biadjointprop ] ) under the shift ( [ halfladdershift ] ) associated with @xmath92 , @xmath340 using the invariance of @xmath341 , together with independence of @xmath93 and @xmath342 , we obtain _
b=3^n ( k_1 k_2 + _ c=3^b-1 k_2 k_(c ) ) m(1 , ( 3 ) , , ( b-1 ) , 2 , ( b ) , , ( n-1 ) , n | 1
n ) = 0 [ propnullvectors ] i.e. , we have derived a set of @xmath16 null eigenvectors of the propagator matrix .
other sets of null eigenvectors are obtained from the color - factor shifts associated the other massless scalars in the amplitude .
since the @xmath15 propagator matrix is known @xcite to have rank @xmath16 , at most @xmath17 of these null eigenvectors can be independent .
thus , the color - factor symmetry associated with @xmath100 of the massless scalars suffices to guarantee the reduced rank of the propagator matrix .
this nicely explains a result found by one of the present authors in ref .
@xcite , that the propagator matrix persists in having rank @xmath16 even when up to three of the external particles in the amplitude are massive , but if more than three particles are massive , the rank of the propagator matrix is greater .
we now see that for @xmath343 massive and @xmath344 massless particles , the number of null eigenvectors generated by the color - factor symmetry will be @xmath345 , and therefore the rank of the propagator matrix will be @xmath346 for @xmath347 through @xmath348 .
returning now to yang - mills theory , if we assume that the numerators of the @xmath0-gluon amplitude obey color - kinematic duality , i.e. they obey the same jacobi relations as @xmath6 , then they can similarly be expressed in terms of @xmath62 half - ladder numerators @xmath349 via n_i = _ s_n-2 m_i , 1n * n*_1 n .
[ nmn ] using , the color - ordered amplitudes in the kleiss - kuijf basis can be written in terms of the propagator matrix as a(1 , ( 2 ) , , ( n-1 ) , n ) = _ s_n-2 m ( 1 n| 1 n ) * n*_1 n .
[ amn ] thus the null eigenvectors of the propagator matrix ( [ propnullvectors ] ) imply that the color - ordered @xmath0-gluon amplitudes obey the bcj relations ( [ fundbcj ] ) . as we saw in sec .
[ sec : cfs ] , however , it is not necessary to require color - kinematic duality in order to prove the bcj relations .
the bcj relations follow from the weaker constraint ( [ sumoverdelta ] ) , and both and the bcj relations are a consequence of the color - factor symmetry of the amplitude , which is established through the radiation vertex expansion .
[ sec : loop ] given the connection established in this paper between the color - factor symmetry of tree - level gauge - theory amplitudes and color - kinematic duality / bcj relations , it is naturally of great interest to see whether these ideas can be extended to loop level . in this section ,
we generalize the cubic vertex expansion introduced in sec .
[ sec : cfs ] to loop - level amplitudes .
we then define a set of shifts of one - loop color factors that depend on the loop momentum as well as the momenta of external particles , and ask whether the cubic vertex expansion of the one - loop amplitude is invariant under these shifts .
it is obvious from the variation of the cubic vertex expansion ( [ variation ] ) that the tree - level amplitude will be invariant under a color - factor shift if the numerators satisfy color - kinematic duality .
the reader may have noticed , however , that up until now we have taken great pains not to invoke color - kinematic duality to prove color - factor symmetry . we have chosen rather to show that it follows directly from a lagrangian approach . at loop level , we no longer have that luxury , at least at the current stage of development .
instead we will explicitly invoke loop - level color - kinematic duality ( for the theories in which it has been shown to hold @xcite ) in order to demonstrate the color - factor symmetry of the one - loop amplitude .
we then show that the invariance of the amplitude under color - factor shifts implies a set of relations among the integrands of its color - ordered amplitudes .
23}$ ] , @xmath3503}$ ] , @xmath351}$ ] , @xmath352 , @xmath353 , and @xmath354.,width=604 ] to construct the cubic vertex expansion of an @xmath355-loop @xmath0-gluon amplitude with respect to gluon @xmath19 , we begin with the set of @xmath355-loop @xmath40-point cubic diagrams @xmath39 with external legs @xmath356 . for example , for the one - loop four - gluon amplitude , two of the three - point diagrams are shown in fig .
[ fig : three ] ; the rest are obtained from relabelings of the external legs .
then we attach gluon @xmath19 in all possible ways , either to the external legs or to the internal lines of @xmath39 .
for example , attaching gluon 4 to the triangle diagram in fig .
[ fig : three ] in all possible ways , we obtain the diagrams in fig .
[ fig : four ] . the diagrams in fig .
[ fig : four ] correspond to the following terms in the cubic decomposition of the one - loop four - point amplitude @xcite @xmath35723 } \,\ , n_{[14]23 } \over d_{[14]23 } } ~+~ { c_{1[24]3 } \,\ , n_{1[24]3 } \over d_{1[24]3 } } ~+~ { c_{12[34 ] } \,\ , n_{12[34 ] } \over d_{12[34 ] } } { \nonumber}\\ & ~+~ { c_{1234 } \,\ , n_{1234 } \over d_{1234 } } ~+~ { c_{1423 } \,\ , n_{1423 } \over d_{1423 } } ~+~ { c_{1243 } \,\ , n_{1243 } \over d_{1243 } } \bigg ] \label{looptrianglecontrib}\end{aligned}\ ] ] where the denominators in are the products of inverse propagators associated with the diagrams .
( there could be different sets of denominators depending on the mass of the particle circulating in the loop . )
explicit definitions of the color factors are given below in .
the terms obtained by attaching the gluon to an internal line in this case , the last three terms of are split into two by applying the identity ( [ splitprop ] ) or , in the case of massive internal lines , .
the terms are then reorganized into a sum over the vertices of @xmath39 .
for example , the terms in are reorganized into @xmath35823 } \,\ , n_{[14]23 } \over 2 k_4 \cdot k_1 } ~+~ { c_{1234 } \,\
, n_{1234 } \over 2 k_4 \cdot \ell } ~-~ { c_{1423 } \,\
, n_{1423 } \over 2 k_4 \cdot ( \ell+k_1 ) } \right ] { \nonumber}\\ & + { 1 \over \ell^2 ( \ell+k_1)^2 ( \ell - k_3)^2 } \left [ { c_{1[24]3 } \,\ , n_{1[24]3 } \over 2 k_4 \cdot k_2 } ~+~ { c_{1423 } \,\ , n_{1423 }
\over 2 k_4 \cdot ( \ell+k_1 ) } ~-~ { c_{1243 } \,\ , n_{1243 } \over 2 k_4 \cdot ( \ell+k_1+k_2 ) } \right ] \label{loopfourpoint } \\
& + { 1 \over \ell^2 ( \ell+k_1)^2 ( \ell+k_1 + k_2)^2 } \left [ { c_{12[34 ] } \,\ , n_{12[34 ] } \over 2 k_4 \cdot k_3 } ~+~ { c_{1243 } \,\ , n_{1243 }
\over 2 k_4 \cdot ( \ell+k_1+k_2 ) } ~-~ { c_{1234 } \,\ , n_{1234 } \over 2 k_4 \cdot ( \ell+k_1+k_2+k_3 ) } \right ] \bigg\ } { \nonumber}\end{aligned}\ ] ] where @xmath359 in the last term of the last line of differs from the label in fig .
[ fig : four ] by a shift .
if the particle circulating in the loop has mass @xmath360 , the expressions @xmath361 in ( those outside the square brackets ) are all replaced by @xmath362 .
we hasten to remind the reader that @xmath363 is only one part of the one - loop four - gluon amplitude ; similar expressions are obtained by attaching gluon 4 to the other one - loop three - point diagrams @xmath39 .
observe that if we apply a shift to the numerators @xmath36423 } & ~=~ \beta k_4 \cdot k_1 , & \delta_4 n_{1[24]3 } & ~=~ \beta k_4 \cdot k_2 , & \delta_4 n_{12[34 ] } & ~=~ \beta k_4 \cdot k_3 , { \nonumber}\\ \delta_4 n_{1234 } & ~=~ \beta
k_4 \cdot \ell , & \qquad \delta_4 n_{1423 } & ~=~ \beta k_4 \cdot ( \ell + k_1 ) , & \qquad \delta_4 n_{1243 } & ~=~ \beta k_4 \cdot ( \ell + k_1+k_2 ) , \label{loopkinematicshift}\end{aligned}\ ] ] the expression ( [ loopfourpoint ] ) remains unchanged as a result of the jacobi identities 0 = c_[14]23 + c_1234 - c_1423 = c_1[24]3 + c_1423 - c_1243 = c_12[34 ] + c_1243 - c_1234 [ loopcolorjacobi ] which means that corresponds to a generalized gauge transformation .
\cdots}$ ] and @xmath365 , width=264 ] the @xmath0-point generalization of , obtained by attaching gluon @xmath0 to the @xmath40-gon diagram , is given by @xmath366 \cdots } n_{\cdots [ bn ] \cdots } \over 2 k_n \cdot k_b } ~+~ { c_{\cdots b-1,n , b \cdots } n_{\cdots b-1,n , b \cdots } \over 2 k_n \cdot \left ( \ell + \sum_{a=1}^{b-1}k_a \right ) } ~-~ { c_{\cdots b , n , b+1 \cdots } n_{\cdots b , n , b+1 \cdots } \over 2 k_n \cdot \left ( \ell + \sum_{a=1}^{b } k_a \right ) } \right ] \bigg\ } \label{loopnpoint}\end{aligned}\ ] ] where @xmath3673 \cdots n } & ~\equiv~ \sum_{\tb_1,\ldots,\tb_n } f_{\ta_1 \ta_2 \tb_2 } f_{\tb_1 \tb_2 \tb_3 } f_{\tb_3 \ta_3 \tb_3 } \cdots f_{\tb_n \ta_n \tb_1 } \ ,
. \label{loopcolorfactors}\end{aligned}\ ] ] the `` ring '' diagrams @xmath368 are cyclically symmetric and reflection symmetric c_12
n = ( -1)^n c_n21 .
[ reflection ] the color factors appearing in are shown in fig .
[ fig : five ] and satisfy the jacobi relations 0 = c _ + c_b-1,n , b - c_b , n , b+1 .
[ loopnpointjacobi ] again , if a particle of mass @xmath360 is circulating in the loop , we replace terms of the form @xmath369 with @xmath370 .
similar expressions are obtained by attaching gluon @xmath0 to the other one - loop @xmath40-point diagrams @xmath39 , including @xmath40-gon diagrams with other permutations of the external legs @xmath371 , recently , partial fraction identities similar to have been employed to recast one - loop amplitudes into a new form whose denominators contain factors linear in the loop momentum @xcite , somewhat analogous to .
these new expressions are those that naturally emerge from a scattering - equation approach to loop - level amplitudes .
next we consider the behavior of one - loop amplitudes under momentum - dependent shifts of its color factors .
first we must define a set of shifts consistent with the requirements elucidated earlier in the paper .
for the one - loop four - gluon amplitude , the color - factor shift associated with gluon 4 must satisfy _ 4 c_[14]23 = k_4 k_1 , _ 4 c_1[24]3 = k_4 k_2 , _ 4 c_12[34 ] = k_4 k_3 [ loopfourcommshift ] since these diagrams have gluon 4 attached to an external leg . requiring
the color - factor shift to respect the jacobi relations ( [ loopcolorjacobi ] ) implies that _ 4 c_1423 = _ 4 c_1234 + k_4 k_1 _ 4 c_1243 = _ 4 c_1234 + k_4 ( k_1+k_2 ) .
[ loopfourjacobishift ] unlike the tree - level case , however , these requirements alone are not sufficient to fix the values of all the shifts ; one of them , @xmath372 , remains arbitrary . in analogy with , we define the remaining shift to be _ 4 c_1234 = k_4 [ loopfourshift ] where @xmath359 is the loop momentum
. the effect of this shift on is @xmath37323 } + n_{1234 } -n_{1423 } \over \ell^2 ( \ell - k_2 -k_3)^2 ( \ell - k_3)^2 } ~+~ { n_{1[24]3 } + n_{1423 } -n_{1243 } \over \ell^2 ( \ell+k_1)^2 ( \ell - k_3)^2 } { \nonumber}\\ & ~+~ { n_{12[34 ] } + n_{1243 } -n_{1234 } \over \ell^2 ( \ell+k_1)^2 ( \ell+k_1 + k_2)^2 } \bigg ] \ , .
\label{loopsumoverdelta}\end{aligned}\ ] ] similar expressions are obtained for the contributions to the cubic vertex expansion from the other three - point diagrams . a goal consistent with the development in this paper would be to prove by some alternative means ( such as the radiation vertex expansion ) that @xmath374 vanishes under the one - loop color - factor shift .
that would imply the vanishing of the r. h. s. of plus the expressions obtained from the other three - point diagrams , imposing a gauge - invariant constraint on the one - loop kinematic numerators .
but at this point in the development of the subject , we have no such proof , and therefore we will turn the argument around , and use the knowledge that there exist kinematic numerators for the one - loop four - gluon amplitude that obey color - kinematic duality 0 = n_[14]23 + n_1234 - n_1423 = n_1[24]3 + n_1423 - n_1243 = n_12[34 ] + n_1243 - n_1234 [ loopkinematicjacobi ] for pure yang - mills theory ( with only gluons circulating in the loop ) @xcite as well as for theories with other particles circulating in the loop @xcite . in these cases , the kinematic jacobi identities ( [ loopkinematicjacobi ] )
imply that vanishes , as do the other contributions to the cubic vertex expansion .
thus , the one - loop four - gluon amplitude in these theories is invariant under the color - factor shift specified by eqs .
( [ loopfourcommshift ] ) , ( [ loopfourjacobishift ] ) , and ( [ loopfourshift ] ) .
the cubic vertex expansion of the one - loop four - point amplitude of the bi - adjoint scalar theory may be obtained by replacing the kinematic numerators @xmath9 with a second copy of the color factors @xmath375 . since the latter obey the one - loop color jacobi identities ( [ loopcolorjacobi ] ) , the bi - adjoint scalar one - loop four - point amplitude is also invariant under the color - factor shift .
it is known that an independent basis of one - loop color factors are those associated with ring diagrams modulo cyclic permutations and reflections @xcite .
therefore the one - loop @xmath0-gluon amplitude can be written _
n^(1 ) = _ s_n-1/-5pt _ 2 c_1 i(1 , ( 2 ) , , ( n ) ) [ ringbasis ] where @xmath60 is a permutation of @xmath376 , @xmath377 denotes the reflection symmetry @xmath378 , and @xmath379 are the integrands of the one - loop color - ordered amplitudes .
may be regarded as the result of a generalized gauge transformation in which the kinematic numerators associated with the non - ring color factors are set to zero @xcite .
specializing to the one - loop four - gluon amplitude , gives _
4^(1 ) = . for theories
whose numerators respect one - loop color - kinematic duality , the invariance of the one - loop amplitude under the color - factor shift implies the following condition on the integrands 0 = .
[ integrandfourpointconstraint ] relations of this form were first uncovered in refs .
@xcite from the perspective of on - shell recursion relations , and revisited recently using monodromy relations in string theory @xcite .
conversely , if the integrands of the one - loop amplitude of a theory can be shown to satisfy , then it follows that the one - loop amplitude is invariant under the color - factor shift , avoiding the need to invoke color - kinematic duality . it is straightforward to generalize these considerations to one - loop @xmath0-gluon amplitudes .
we may define a color - factor shift associated with any external gluon @xmath19 , but for simplicity of presentation we will focus on gluon @xmath0 .
let @xmath241 denote a permutation of @xmath380 .
the color factor @xmath381 \cdots}$ ] shown in fig .
[ fig : five ] undergoes a shift _
k_n k_(b ) because gluon @xmath0 is attached to gluon leg @xmath382 . requiring the shifts to respect the jacobi relation ( [ loopnpointjacobi ] ) implies _ n c _ ( b ) n ( b+1 ) = _ n c _ ( b-1 ) n ( b ) + _ n c _ we must additionally define the shifts _ n c _ 1 ( 2 ) ( n-1 ) n = _ n , k_n [ additionalshift ] for a set of half ) .
for example for the four - point case @xmath383 but @xmath384 , which is not of the form ( [ additionalshift ] ) .
] of the permutations @xmath241 , where @xmath385 are a set of @xmath386 independent arbitrary constants .
together these conditions imply that the shifts of the ring color factors are given by _
n c_1 ( 2 ) ( b-1 ) n ( b ) ( n ) = _ n , ( k_n + k_n k_1 + _ c=2^b-1 k_n k_(c ) ) , b \{2 , , n-1 } [ loopnpointshift ] and the shifts of all other one - loop color factors are fixed by requiring that they respect the jacobi relations .
thus there is an @xmath386-dimensional family of one - loop color - factor shifts associated with gluon @xmath0 . applying the shift ( [ loopnpointshift ] ) to ,
we obtain _
n , ( n-1)gon^(1 ) _ b=1^n-1 ( n _ + n_b-1,n , b - n_b , n , b+1 ) _ c=1^b ( + _ a=1^c-1 k_a)^2 _ d = b+1^n-1 ( - _ a = d^n-1 k_a)^2 . [ loopsumoverdeltanpoint ] for @xmath18
supersymmetric yang - mills theory , numerators for the one - loop five - gluon @xcite and higher - point @xcite amplitudes have been constructed which satisfy color - kinematic duality 0 = n _
+ n_b-1,n , b - n_b , n , b+1 and which therefore imply that vanishes .
the shifts of the terms in the cubic vertex expansion obtained from other @xmath40-point diagrams @xmath39 similarly vanish .
thus , we have established that these amplitudes possess one - loop color - factor symmetry .
the one - loop @xmath0-point amplitudes of the bi - adjoint scalar theory also possess this symmetry because the second copy of the color factors @xmath375 obey jacobi identities ( [ loopnpointjacobi ] ) . as we did above for the four - gluon amplitude
, we can use this invariance to derive constraints on the integrands of color - ordered amplitudes .
can be rewritten as _
n^(1 ) = _ s_n-2/-5pt _ 2 _ b=2^n c_1 ( 2 ) ( b-1)n ( b ) ( n-1 ) i(1 , ( 2 ) , , ( b-1 ) , n , ( b ) , , ( n-1 ) ) invariance of this expression under together the independence of the parameters @xmath385 yields 0= _ b=2^n k_n ( + k_1 + _ c=2^b-1 k_(c ) ) i(1 , ( 2 ) , , ( b-1 ) , n , ( b ) , , ( n-1 ) ) [ integrandnpointconstraint ] the relations uncovered in refs . @xcite .
conversely , if we were to establish that the integrands of the color - ordered amplitudes of a given theory satisfy , we would have proven that the one - loop amplitude is invariant under the color - factor shift , independently of the assumption of color - kinematic duality .
further study of this alternate path is merited .
[ sec : concl ] in this paper , we have introduced a new set of symmetries of gauge - theory amplitudes , which act as momentum - dependent shifts on the color factors appearing in the cubic decomposition of the amplitude .
these symmetries are intimately linked to the presence of massless gauge bosons in the amplitude ( or massless adjoint scalars in the case of the bi - adjoint scalar theory ) and can be considered generalizations of the radiation symmetry of ref .
we demonstrated that a wide class of tree - level gauge - theory amplitudes are invariant under these shifts , using a representation of the amplitude known as the radiation vertex expansion @xcite .
we also introduced a related but distinct cubic vertex expansion of the amplitude , and used this to derive a set of gauge - independent constraints on the kinematic numerators appearing in the cubic decomposition of the amplitude .
all known bcj relations for tree - level gauge - theory amplitudes @xcite follow as a direct consequence of the color - factor symmetry ( this paper and ref .
@xcite ) . finally , we generalized the cubic vertex expansion and color - factor symmetry to loop level .
we showed that one - loop amplitudes that satisfy color - kinematic duality are invariant under the one - loop color - factor symmetry , and derived a set of relations among the integrands of one - loop color - ordered amplitudes .
let us take a look at the connection between the color - factor symmetry and more fundamental symmetries of the lagrangian , gauge and poincar invariance @xcite .
the color - factor symmetry follows as a result of the vanishing of certain expressions , namely eqs .
( [ threesums ] ) , ( [ variationspinhalf ] ) , and ( [ threesumsfund ] ) , associated with the cubic vertices of a gauge - theory amplitude , and eqs .
( [ variationfourglounleading ] ) , ( [ variationfourglounsubleading ] ) , and ( [ variationquarticspinzero ] ) , associated with the quartic vertices .
it is illustrative to examine the various contributions in a soft expansion in the gluon momentum @xmath13 , even though the color - factor symmetry is exact in @xmath13 .
the leading term in the soft expansion corresponds to the @xmath387 term in each of eqs .
( [ threesums ] ) , ( [ variationfourglounleading ] ) , ( [ variationspinhalf ] ) , ( [ threesumsfund ] ) , and ( [ variationquarticspinzero ] ) .
these are all proportional to @xmath388 where @xmath389 are the momenta flowing out of each leg of the vertex .
this vanishes by @xmath224 together with momentum conservation @xmath390 a result of symmetry under spacetime translations .
the subleading term in the soft expansion corresponds to the @xmath391 term in each of eqs .
( [ threesums ] ) , ( [ variationspinhalf ] ) , ( [ threesumsfund ] ) , and ( [ variationquarticspinzero ] ) , and to .
these expressions are all given by a sum of angular momentum generators @xmath162 , which act as a first - order lorentz transformation on the relevant vertex factors .
they vanish by lorentz invariance .
thus the first two terms in the soft expansion vanish as a result of poincar invariance .
it is a little more difficult to pin down the underlying symmetry responsible for the vanishing of the sub - subleading terms in , together with an analogous expression for spin - one fundamentals .
the @xmath392 term in is proportional to @xmath393 and the @xmath392 term in is proportional to ( s_3^)^_3 _ k_3 v^(k_1 , k_2 , k_3 ) = ( ^_3 ^ -^_3 ^ ) .
neither expression vanishes by itself but both do when contracted with @xmath394 for gluon @xmath19 .
these identities , which go beyond the first - order poincar cancellation and are connected to yang - mills gauge symmetry , are key ingredients contributing to the color - factor symmetry . returning to our discussion in the introduction of the connection of the color - factor symmetry with the photon radiation symmetry and radiation zeros ( for a collection of early references , see refs .
@xcite ) , we have uncovered some additional analogs .
for example , we can write a factorized form for the four - gluon amplitude ( [ fourpointamplitude ] ) _ 4 = s ( c_s s - c_t t ) ( n_s s - n_u u ) which vanishes at @xmath395 const . this is a non - abelian version of the original radiation zero studied almost forty years ago in @xmath396 and @xmath397 reactions , with a zero at @xmath398 constant .
this original radiation factorization and its zero led to the prediction of a measurable experimental dip , which has now been confirmed @xcite .
the analogous zero in the four - gluon amplitude is washed out , however , by the color averaging that must be performed in observable quark - gluon processes . in the generalization to tree - level @xmath0-point amplitudes , the abelian radiation symmetry and existence of zeros for @xmath399 constant ( for photon momentum @xmath13 ) rest on having gauge - theory couplings , as noted earlier . the non - abelian color - factor symmetry uncovered in this paper
can also lead to zeros in @xmath0-point amplitudes , but with an important complication .
the invariance under @xmath400 for the attachment of a gluon with momentum @xmath13 can not be used to systematically cancel out the complete @xmath0-gluon amplitude using an overall common value for @xmath401 .
because of the jacobi relations , that overall common value must vanish .
there are in principle zeros , however , for separate islands of @xmath401 values .
although they are again washed out by color - averaging , the generalized factorization coming from the color - factor symmetry remains useful for theoretical analysis of tree amplitudes . in a different direction , note that the bcj form of the gluon amplitudes has been utilized in the planar zeros studied recently in refs .
@xcite .
the analogs described above are a bridge to a final overall remark .
it has been helpful to think of gluon emission or absorption as effecting a ( first - order ) transformation in both color and kinematic space simultaneously on the graph to which it is attached . in particular , the attachments lead to transformations of the various legs and vertices of the `` parent '' diagram in either momentum or space - time representations .
all the parent wave functions end up transformed , and identities derived from for the different spins , , yield exactly the two terms expected from the propagator with its bilinearity in the wave functions .
the cancellations highlighted throughout this paper arise precisely because we consider those theories whose amplitudes transform covariantly under color and kinematic transformations . adding all possible massless gluon attachments to a complete set of parent graphs leads to a sum of corresponding color shifts that vanishes because of invariance .
such a picture should help in finding directions in the diagrammatic analyses of a variety of extensions of the gauge theories considered in this paper , supersymmetric and otherwise .
this material is based upon work supported by the national science foundation under grants nos .
phy14 - 16123 and pfi : bic 1318206 .
rwb is also supported by funds made available through a cwru institute professorship chair .
sgn gratefully acknowledges sabbatical support from the simons foundation ( grant no .
342554 to stephen naculich ) .
he would also like to thank the michigan center for theoretical physics and the physics department of the university of michigan for generous hospitality and for providing a welcoming and stimulating sabbatical environment .
[ sec : app ] in this appendix , we use the five - gluon amplitude to provide an explicit example of the cubic vertex expansion ( [ cve ] ) introduced in sec . [
sec : cfs ] , and the relations among kinematic numerators ( [ sumoverdelta ] ) resulting from the color - factor symmetry .
the cubic decomposition ( [ cubicdecomp ] ) of the five - gluon amplitude is given by @xmath402 where @xmath58 are half - ladder color factors defined in .
let us recast this amplitude in a cubic vertex expansion with respect to gluon 2 .
we have already arranged the terms in so that each line corresponds to one of the four - gluon diagrams @xmath39 obtained by omitting gluon 2 .
we rewrite the denominator of the third term of the first line as = 1s_45 ( -s_12 - s_23 ) + 1s_13 ( -s_24 - s_25 ) and similarly the denominators of the third terms of the other two lines to obtain @xmath403 which is precisely of the form of the cubic vertex expansion ( [ cve ] ) . the six independent five - gluon half - ladder color factors ( in the notation of ref .
@xcite ) are @xmath404 according to , the color - factor shifts associated with gluon 2 act as @xmath405 where @xmath406 and @xmath407 are arbitrary constants .
the nine remaining five - gluon color factors , and the action of the color - factor shifts thereon , are determined by the jacobi relations to be @xmath408 applying this shift to , we obtain @xmath409 which is precisely of the form of . the color - factor shift with respect to gluon 3 instead
yields @xmath410 since @xmath411 , @xmath412 , @xmath413 , and @xmath414 are independent arbitrary constants , each expression in parentheses vanishes , yielding four independent constraint equations on the kinematic numerators of the five - gluon amplitude .
no additional independent constraints are obtained from the color - factor symmetries associated with the other three gluons .
these `` generalized jacobi relations '' for five - gluon amplitudes were previously derived in refs .
@xcite by using the properties of string theory amplitudes .
d0 collaboration , v. m. abazov et al . , _
`` first study of the radiation - amplitude zero in @xmath419 production and limits on anomalous @xmath420 couplings at @xmath421 = 1.96- tev '' _ , , ` arxiv:0803.0030 ` . | we introduce a new set of symmetries obeyed by tree - level gauge - theory amplitudes involving at least one gluon
. the symmetry acts as a momentum - dependent shift on the color factors of the amplitude . using the radiation vertex expansion , we prove the invariance under this color - factor shift of the @xmath0-gluon amplitude , as well as amplitudes involving massless or massive particles in an arbitrary representation of the gauge group with spin zero , one - half , or one .
the bern - carrasco - johansson relations are a direct consequence of this symmetry .
we also introduce the cubic vertex expansion of an amplitude , and use it to derive a gauge - invariant constraint on the kinematic numerators of the amplitude .
we show that the amplitudes of the bi - adjoint scalar theory are invariant under the color - factor symmetry , and use this to derive the null eigenvectors of the propagator matrix .
we generalize the color - factor shift to loop level , and prove the invariance under this shift of one - loop @xmath0-gluon amplitudes in any theory that admits a color - kinematic - dual representation of numerators .
we show that the one - loop color - factor symmetry implies known relations among the integrands of one - loop color - ordered amplitudes .
bow - ph-162 + mctp-16 - 17 + * bcj relations from a new symmetry of gauge - theory amplitudes * 1.5 cm robert w. brown@xmath1 and stephen g. naculich@xmath2 0.5 cm @xmath1_department of physics + case western reserve university + cleveland , oh 44106 usa _ 0.5 cm @xmath3_department of physics + bowdoin college + brunswick , me 04011 usa _ 0.5 cm @xmath4_michigan center for theoretical physics ( mctp ) + department of physics + university of michigan + ann arbor , mi 48109 usa
_ 1.5 cm |
let @xmath21 be a finite , simple , connected , and undirected graph of order @xmath22 . for a given graph @xmath1 and @xmath23 , we denote by @xmath24 the subgraph induced by @xmath2 . for a vertex @xmath25 ,
the _ open neighborhood of @xmath26 _ is the set @xmath27 and the _ closed neighborhood of @xmath26 _ is the set @xmath28=n_g(v ) \cup \{v\}$ ] .
the _ degree _ @xmath29 of a vertex @xmath25 is the the number of edges incident with the vertex @xmath26 in @xmath1 .
we denote by @xmath16 the _ minimum degree _ of a graph @xmath1 .
the _ distance _ between two vertices @xmath30 , denoted by @xmath31 , is the length of the shortest path between @xmath26 and @xmath32 . for other terminologies in graph theory ,
refer to @xcite .
+ the notion of a zero forcing set , as well as the associated zero forcing number , of a simple graph was introduced by the aim minimum rank special graphs work group " in @xcite to bound the minimum rank of associated matrices for numerous families of graphs .
let each vertex of a graph @xmath1 be given one of two colors , black " and white " by convention .
let @xmath2 denote the ( initial ) set of black vertices of @xmath1 .
the _ color - change rule _ converts the color of a vertex from white to black if the white vertex @xmath33 is the only white neighbor of a black vertex @xmath34 ; we say that @xmath34 forces @xmath33 , which we denote by @xmath35 . and a sequence , @xmath36 , obtained through iterative applications of the color - change rule is called a _ forcing chain_. note that , at each step of the color change , there may be two or more vertices capable of forcing the same vertex .
the set @xmath2 is said to be _ a zero forcing set _ of @xmath1 if all vertices of @xmath1 will be turned black after finitely many applications of the color - change rule .
the _ zero forcing number _ of @xmath1 , denoted by @xmath0 , is the minimum of @xmath37 over all zero forcing sets @xmath23 .
+ since its introduction by the aforementioned aim group " , zero forcing number has become a graph parameter studied for its own sake , as an interesting invariant of a graph . in @xcite
, the authors studied the number of steps it takes for a zero forcing set to turn the entire graph black ; they named this new graph parameter the _ iteration index _ of a graph : from the real world " modeling ( or discrete dynamical system ) perspective , if the initial black set is capable of passing a certain condition or trait to the entire population ( i.e. zero forcing " ) , then the iteration index of a graph may represent the number of units of time ( anything from days to millennia ) necessary for the entire population to acquire the condition or trait . independently ,
hogben et al . studied the same parameter ( iteration index ) in @xcite , which they called _
propagation time_. it s also noteworthy that physicists have independently studied the zero forcing parameter , referring to it as the _ graph infection number _ , in conjunction with the control of quantum systems ( see @xcite , @xcite , and @xcite ) .
more recently in @xcite , the authors initiated a comparative study between metric dimension and zero forcing number for graphs . in @xcite
, the authors also introduced a probabilistic theory of zero forcing in graphs . for more articles and surveys pertaining to the zero forcing parameter ,
see @xcite .
+ chartrand and harary @xcite introduced permutation graphs " ( or generalized prisms " ) .
hedetniemi @xcite introduced the function graph " , which consists of two graphs ( not necessarily identical copies ) with a function relation between them . independently
, drfler @xcite introduced a mapping graph " , which consists of two disjoint identical copies of a graph and additional edges between the two vertex sets specified by a function .
the mapping graph " is rediscovered and studied in @xcite , where it is called a functigraph " . for articles on functigraphs , see @xcite and @xcite .
we recall the definition of the functigraph .
let @xmath5 and @xmath6 be disjoint copies of a graph @xmath1 , and let @xmath7 be a function .
a _ functigraph _
@xmath8 consists of the vertex set @xmath9 and the edge set @xmath38 . in this paper
, we study the zero forcing number of functigraphs . for a graph @xmath1 of order @xmath39
, it is readily seen that @xmath12 for a permutation @xmath13 ; we show that @xmath40 for a function @xmath15 , and the bounds are sharp . we give examples showing that there does not exist a function @xmath17 such that , for every pair @xmath18 , @xmath19 or @xmath20 .
further , we give zero forcing number of functigraphs on complete graphs @xmath41 , and we give bounds for zero forcing number of functigraphs on cycles @xmath42 and on paths @xmath43 .
the _ path cover number _ @xmath44 of @xmath1 is the minimum number of vertex disjoint paths , occurring as induced subgraphs of @xmath1 , that cover all the vertices of @xmath1 .
next , we recall the definition that is stated in @xcite .
a graph @xmath1 is a graph of _ two parallel paths _ if there exist two independent induced paths of @xmath1 that cover all the vertices of @xmath1 and such that @xmath1 can be drawn in the plane in such a way that the two paths are parallel and the edges ( drawn as segments , not curves ) between the two paths do not cross .
a simple path is not considered to be such a graph .
a graph that consists of two connected components , each of which is a path , is considered to be such a graph .
@xcite [ pathcover ] * @xcite for any graph @xmath1 , @xmath45 . * @xcite for any tree @xmath46 , @xmath47 . * @xcite for any unicyclic graph @xmath1 , @xmath48 . @xcite[mindegree ] for any graph @xmath1 , @xmath49 .
@xcite [ idupper ] for any graphs @xmath1 and @xmath50 , @xmath51 , where @xmath52 denotes the cartesian product of @xmath1 and @xmath50 .
[ obsz ] let @xmath1 be a connected graph of order @xmath53 .
then * @xcite @xmath54 if and only if @xmath55 , * @xcite @xmath56 if and only if @xmath1 is a graph of two parallel paths , * @xcite @xmath57 if and only if @xmath58 .
@xcite let @xmath1 be any graph .
then * for @xmath25 , @xmath59 .
* for @xmath60 , @xmath61 .
@xcite [ cutv ] let @xmath1 be a graph with cut - vertex @xmath25 .
let @xmath62 be the vertex sets for the connected components of @xmath63 , and for @xmath64 , let @xmath65 = @xmath66
. then @xmath67-k+1 $ ] .
next , we obtain general bounds for zero forcing number of functigraphs . if @xmath1 is a graph of order @xmath68 , one can easily check that @xmath69 for any function @xmath15 .
so , we only consider a graph @xmath1 of order @xmath11 for the rest of the paper .
notice that @xmath70 forms a zero forcing set for a permutation graph @xmath71 ; this , together with theorem [ mindegree ] , implies the following @xcite[n ] let @xmath1 be a graph of order @xmath11 , and let @xmath72 be a permutation
. then @xmath73 .
[ functibounds ] let @xmath1 be a graph of order @xmath11 , and let @xmath7 be a function
. then @xmath14 .
both bounds are sharp . first , noting that @xmath74 for any function @xmath15 , @xmath75 by ( c ) of theorem [ obsz ] .
next , we show that @xmath76 . if @xmath77 , the result follows by corollary [ n ] .
so , we consider @xmath78 .
assume , to the contrary , that @xmath79 .
then @xmath80 by theorem [ mindegree ] , since @xmath81 .
let @xmath2 be a zero forcing set for @xmath82 with @xmath83 .
then @xmath2 must contain a vertex @xmath84 satisfying @xmath85 , along with all but one vertex in @xmath86 . for each @xmath87 , let @xmath88 be a forcing chain consisting of @xmath89 vertices ; notice that @xmath90 is a zero forcing set for @xmath82 and each vertex in @xmath91 must appear in a forcing chain .
we make the following _ claim : _ at most @xmath92 vertices in @xmath5 are turned black after applying the color - change rule on @xmath2 as long as possible .
_ proof of claim : _ assume , to the contrary , that @xmath93 vertices in @xmath5 are turned black .
let @xmath94 be the first @xmath93 vertices in @xmath5 that are turned black .
since @xmath82 is not a permutation graph , there exists a vertex @xmath95 such that @xmath96 .
notice that @xmath97 ; otherwise , once @xmath98 is turned black , @xmath98 has at least two white neighbors in @xmath5 , and thus at most @xmath92 vertices in @xmath5 are turned black . since there exists a @xmath99 path ( a path connecting @xmath98 and @xmath100 ) in @xmath6 such that @xmath100 is adjacent to a vertex in @xmath101 , for some @xmath102 ( @xmath103 ) and for @xmath104 , even after @xmath100 is turned black , @xmath100 has at least two white neighbors , @xmath105 and a white neighbor in @xmath6 along a @xmath106 path ; thus , it is impossible that @xmath93 vertices in @xmath5 are turned black .
@xmath107 since each vertex in @xmath5 has degree at least @xmath108 , by the claim , no vertex in @xmath5 can force .
so , @xmath2 fails to be a zero forcing set for @xmath91 with @xmath109 , and thus @xmath110 .
+ for the sharpness of the lower bound , take @xmath58 with @xmath111 a permutation ; then @xmath112 and @xmath113 ( see proposition [ perk ] ) . for the sharpness of the upper bound , take @xmath58 with @xmath114 a constant function ; then @xmath115 ( see proposition [ constantk ] ) .
in this section , we give examples of functigraphs showing that there does not exist a function @xmath17 such that , for every pair @xmath18 , @xmath19 or @xmath20 . in @xcite , examples of permutation graphs showing that @xmath117 can be arbitrarily large were given .
here , we first give an example of non - permutation functigraph showing that @xmath118 can be arbitrarily large .
there exists a functigraph such that @xmath118 can be arbitrarily large ( see figure [ zdown2 ] ) ; notice that @xmath119 by ( b ) of theorem [ pathcover ] , and @xmath120 since the solid vertices in figure [ zdown2 ] form a zero forcing set for @xmath91 .
next , we give examples of functigraphs showing that there does not exist a function @xmath17 such that , for every pair @xmath18 , @xmath19 or @xmath20 .
[ refereeq1 ] there does not exist a function @xmath17 such that @xmath19 for every pair @xmath121 .
let @xmath1 be the graph in figure [ zperdownbig ] ( a bouquet of @xmath122 circles ) , @xmath123 , @xmath124 , where @xmath125 .
let @xmath7 be defined by @xmath126 for @xmath127 , @xmath128 for @xmath129 , and @xmath130 for @xmath129 .
note that @xmath131 : @xmath132 by ( a ) of theorem [ pathcover ] , and @xmath133 since @xmath134 is a zero forcing set for @xmath1 . on the other hand , @xmath135 since the solid vertices of figure [ zperdownbig ]
form a zero forcing set for @xmath91 .
[ refereeq2 ] there does not exist a function @xmath17 such that @xmath20 for every pair @xmath18 .
let @xmath136 , @xmath137 , and @xmath138 , where @xmath139 .
let @xmath7 be defined by @xmath140 and @xmath141 , where @xmath142 ( see figure [ refq2 ] ) . notice that @xmath54 by ( a ) of theorem [ obsz ] . on the other hand , @xmath143 since at least a vertex in each @xmath144 ( @xmath129 ) and at least a vertex in each @xmath145 ( @xmath146 ) must belong to a zero forcing set of @xmath91 ; otherwise , a vertex in @xmath147 or a vertex in @xmath148 fails to turn black , after applying the color - change rule as long as possible .
in contrast to the above examples , as an immediate consequence of theorem [ idupper ] , we have the following [ identityupper ] for any graph @xmath1 of order @xmath11 , @xmath149 .
further , we make the folllowing for any graph @xmath1 of order @xmath11 , @xmath150 .
in this section , for @xmath11 , we show that 1 ) @xmath113 for a permutation @xmath13 ; 2 ) @xmath151 for a function @xmath15 with @xmath152 . throughout this section ,
we let @xmath153 and @xmath154 for @xmath155 .
[ perk ] let @xmath58 be the complete graph of order @xmath11 , and let @xmath72 be a permutation
. then @xmath113 .
since @xmath112 , @xmath156 by corollary [ n ] . [ constantk ]
let @xmath58 be the complete graph of order @xmath11 , and let @xmath157 be a constant function . then @xmath115 . without loss of generality
, we may assume that @xmath158 for each @xmath159 ( @xmath160 ) .
if we let @xmath161 , then @xmath162 ; one can view @xmath163 as @xmath50 and @xmath6 being joined at the cut - vertex @xmath164 . by ( c ) of theorem [ obsz ] , @xmath165 and @xmath166 . by theorem [ cutv ] , @xmath167 . by theorem [ functibounds ] , @xmath168
thus , @xmath169 .
theorem [ constantk ] may be generalized as follows . for @xmath170 ,
let @xmath171 and @xmath172 .
let @xmath173 be the graph with @xmath174 and @xmath175 . then @xmath176 . let @xmath58 be the complete graph of order @xmath11 , and let @xmath177 where @xmath178
. then @xmath179 .
let @xmath180 with @xmath181 , where @xmath182 .
without loss of generality , we may assume that @xmath183 such that @xmath184 and @xmath185 .
further , we may assume that @xmath186 , @xmath187 , and @xmath188 ; we adopt the convention that @xmath189 when @xmath190 .
let @xmath2 be a zero forcing set for @xmath191 .
note that @xmath192 is a zero forcing set for @xmath193 with @xmath194 : ( i ) @xmath195 for @xmath196 ; ( ii ) @xmath197 .
so , @xmath198 .
+ next , we show that @xmath199 .
notice that no vertex in @xmath200 can force any vertex in @xmath5 , and no vertex in @xmath5 can force any vertex in @xmath200 .
first , note that at least @xmath201 vertices of @xmath200 must belong to @xmath2 ; otherwise , even after all vertices of @xmath202 are turned black , each vertex in @xmath6 has at least two white neighbors in @xmath6 , and hence it is impossible to turn the entire vertex set of @xmath6 black .
second , we make the following + _ claim 1 . for each @xmath159 ( @xmath196 ) , @xmath203 . _ _ proof of claim 1 : _ assume , to the contrary , that @xmath204 for some @xmath205 ( there exists a @xmath205 since @xmath206 ) , where @xmath196 .
since @xmath207 has at least two white neighbors in @xmath5 , @xmath207 can not force at all .
further , even after all vertices in @xmath208 are turned black , noting that each vertex of @xmath5 has at least two white neighbors in @xmath5 , it is impossible to turn the entire vertex set of @xmath5 black .
this contradicts the assumption that @xmath2 is a zero forcing set .
@xmath107 third , we make the following _ claim 2
. there exists at most one @xmath209 ( @xmath210 ) such that @xmath211 . _ _ proof of claim 2 : _ if @xmath211 for some @xmath209 ( @xmath210 ) , then , by claim 1 , @xmath212 and @xmath213 .
suppose that there are two such @xmath209 s , say @xmath214 and @xmath215 , where @xmath216 .
then , even after all vertices in @xmath217 are turned black , each vertex in @xmath217 has at least two white neighbors in @xmath5 ; similarly , even after all vertices in @xmath218 are turned black , each vertex in @xmath218 has at least two white neighbors in @xmath6 .
so , it is impossible to turn the entire vertex set of @xmath191 black , contradicting the assumption that @xmath2 is a zero forcing set .
so , there exists at most one @xmath209 ( @xmath210 ) such that @xmath211 .
@xmath107 if such a @xmath209 in claim 2 exists , say @xmath219 for some @xmath220 , then either @xmath221 or @xmath222 ; otherwise , each @xmath65 ( @xmath223 ) has at least two white vertices , and thus no vertex in @xmath2 can force , contradicting the assumption that @xmath2 is a zero forcing set for @xmath193 .
if @xmath221 , then @xmath224 ( otherwise , each vertex @xmath225 has two white neighbors , @xmath226 and @xmath227 , and thus @xmath228 . if @xmath222 , then , by claim 1 , @xmath228 . on the other hand , if @xmath229 for each @xmath159 ( @xmath196 ) , then @xmath230
thus , in each case , we have @xmath231 . therefore , @xmath232 for @xmath233 .
in this section , for @xmath11 , we show that 1 ) @xmath234 for a permutation @xmath13 ; 2 ) @xmath235 for a constant function @xmath236 ; 3 ) @xmath237 for a function @xmath15 with @xmath238 .
further , we give examples showing that the bounds of @xmath239 are sharp when @xmath240 . throughout this section ,
we let @xmath153 and @xmath241 ; similarly , let @xmath154 and @xmath242 .
@xcite[zeropc ] for @xmath243 and @xmath244 , @xmath245 . as an immediate consequence of proposition [ zeropc ]
, we have the following [ zerocid ] for @xmath11 , @xmath246 we recall that a graph @xmath1 is _ strongly regular _ with parameters @xmath247 if @xmath248 , @xmath1 is @xmath122-regular ( i.e. , the degree of each vertex in @xmath1 is @xmath122 ) , every pair of adjacent vertices has @xmath249 common neighbors , and every pair of non - adjacent vertices has @xmath250 common neighbors .
@xcite[srg ] if @xmath1 is a strongly regular graph , then @xmath251 .
@xcite[petersen ] the petersen graph @xmath252 ( see figure [ zuppetersen ] ) is strongly regular ; thus , @xmath253 by corollary [ n ] and proposition [ srg ] . as an immediate consequence of corollary [ n ] ,
we have the following let @xmath254 be the cycle of order @xmath11 , and let @xmath72 be a permutation
. then @xmath234 .
@xcite let @xmath254 be the cycle of order @xmath11 , and let @xmath72 be a permutation . then * @xmath255 if and only if @xmath256 ( for any @xmath13 ) ; * @xmath257 if and only if @xmath256 or @xmath258 ( for any @xmath13 ) or @xmath259 is isomorphic to the petersen graph .
let @xmath254 be the cycle of order @xmath11 , and let @xmath260 be a constant function
. then @xmath261 .
without loss of generality , we may assume that @xmath158 for each @xmath159 ( @xmath160 ) .
one can easily check that @xmath262 is a zero forcing set for @xmath263 , and thus @xmath264 .
next , we show that @xmath265 .
if we let @xmath161 , then @xmath266 : @xmath267 since @xmath268 is a zero forcing set for @xmath50 , and @xmath269 by theorem [ mindegree ] . since @xmath270 , noting that @xmath164 is a cut - vertex of @xmath263 , we have @xmath265 by theorem [ cutv ] .
thus , @xmath271 for @xmath11 . [ functicbetween ]
let @xmath254 be the cycle of order @xmath11 .
let @xmath7 be a function with @xmath177 , where @xmath272 .
then @xmath273 , and both bounds are sharp .
let @xmath180 with @xmath181 , where @xmath182 . by theorem [ functibounds ] , @xmath274 .
next , we show that @xmath275 .
we consider two cases .
@xmath276 : _ in this case , no two vertices in @xmath202 are adjacent in @xmath6 ; let @xmath277 , where @xmath278 .
one can readily check that @xmath279 , for @xmath280 with @xmath281 , forms a zero forcing set for @xmath282 with @xmath283 .
@xmath284 : _ in this case , there exist two adjacent vertices in @xmath6 that belong to @xmath202 .
one can readily check that @xmath285 forms a zero forcing set for @xmath282 with @xmath283 .
thus , @xmath275 . for the sharpness of the lower bound , let @xmath286 for @xmath287 and @xmath288 ; then @xmath289 by theorem [ functibounds ] and the fact that @xmath290 is a zero forcing set for @xmath282 . for the sharpness of the upper bound ,
see remark [ csharp ] .
[ csharp ] let @xmath291 , @xmath292 , and @xmath293 , where @xmath125 .
let @xmath294 be defined by @xmath295 , where @xmath296 and @xmath297 ( see figure [ refq ] )
. then @xmath298 and @xmath299 .
let @xmath82 be the functigraph defined in remark [ csharp ] . since @xmath300 is a zero forcing set for @xmath82 with @xmath301 , @xmath302 .
next , we show that @xmath303 .
assume , to the contrary , that @xmath304 .
let @xmath305 and let @xmath2 be a zero forcing set for @xmath82 .
first , notice that @xmath306 ; otherwise , no vertex in @xmath6 can force a vertex in @xmath5 and vice versa .
second , at least two adjacent vertices in @xmath5 must belong to @xmath2 ; otherwise , no vertex in @xmath307 can force any vertex .
suppose that @xmath308 and @xmath309 with @xmath310 .
then at most ( @xmath311 ) vertices in @xmath307 forces their images , and thus at most @xmath312 ( @xmath313 ) vertices in @xmath202 can be black after one global application of the color - change rule .
so , there exists a vertex in @xmath202 , say @xmath314 for some @xmath209 ( @xmath315 ) , that is white unless a vertex in @xmath316 forces @xmath314 after applying the color - change rule on @xmath307 as long as possible . for each @xmath317 ( @xmath297 ) , @xmath318 \not\subseteq s$ ] ; otherwise , @xmath319 , for some @xmath317 , after one global application of the color - change rule .
even if @xmath318 $ ] ( and thus @xmath314 ) are turned black , @xmath314 can not force any vertex in @xmath316 , unless all but one vertex in @xmath316 are black after applying the color - change rule as long as possible .
so , @xmath320 for all but one value @xmath317 ( @xmath297 ) , and at least a vertex in @xmath321 satisfies @xmath322| \ge 3 $ ] .
thus , @xmath323 , contradicting the assumption that @xmath324 for @xmath125 .
thus , @xmath325 , and thus @xmath326 .
in this section , for @xmath11 , we show that 1 ) @xmath327 for a permutation @xmath13 ; 2 ) @xmath328 for a constant function @xmath236 ; 3 ) @xmath329 for a function @xmath15 with @xmath238 .
further , we give examples showing that the bounds of @xmath330 are sharp when @xmath240 . throughout this section ,
we let @xmath153 and @xmath331 ; similarly , let @xmath154 and @xmath332 .
* @xmath336 if and only if @xmath337 ; * @xmath338 if and only if ( i ) @xmath256 and @xmath339 , or ( ii ) @xmath258 and @xmath340 is isomorphic to the permutation graph satisfying @xmath341 , @xmath342 , @xmath343 , and @xmath344 .
@xmath276 : _ in this case , no two vertices in @xmath202 are adjacent in @xmath6 ; let @xmath277 , where @xmath350 .
one can readily check that @xmath351 , for @xmath280 with @xmath352 , forms a zero forcing set for @xmath353 with @xmath354 .
thus , @xmath349 . for the sharpness of the lower bound , let @xmath286 for @xmath287 and @xmath356 ; then @xmath357 by ( b ) of theorem [ obsz ] . for
the sharpness of the upper bound , consider the functigraph @xmath358 with the function @xmath15 in remark [ csharp ] , where @xmath125 ; notice @xmath298 , and one can check that @xmath359 using a similar argument as in the proof of remark [ csharp ] .
_ acknowledgement .
_ the authors greatly appreciate an anonymous referee for valuable comments , suggestions , and corrections on an earlier draft of this paper .
in particular , they thank the said referee for suggesting the problems addressed in remarks [ refereeq1 ] and [ refereeq2 ] , as well as for providing the example in figure [ zperdownbig ] . aim minimum rank - special graphs work group ( f. barioli , w. barrett , s. butler , s.m .
cioab , d. cvetkovi , s.m .
fallat , c. godsil , w. haemers , l. hogben , r. mikkelson , s. narayan , o. pryporova , i. sciriha , w. so , d. stevanovi , h. van der holst , k. vander meulen , a.w .
zero forcing sets and the minimum rank of graphs .
_ linear algebra appl . _
* 428/7 * ( 2008 ) 1628 - 1648 .
f. barioli , w. barrett , s.m .
fallat , h.t .
hall , l. hogben , b. shader , p. van den driessche , and h. van der holst , zero forcing parameters and minimum rank problems , _ linear algebra appl . _
* 433 * ( 2010 ) 401 - 411 .
edholm , l. hogben , m. hyunh , j. lagrange , and d.d .
row , vertex and edge spread of zero forcing number , maximum nullity , and minimum rank of a graph .
_ linear algebra appl . _
( 2010 ) doi:10.1016/j.laa.2010.10.015 .
hedetniemi , on classes of graphs defined by special cutsets of lines .
many facets of graph theory , proc .
western michigan univ . ,
kalamazoo / mi , 1968 , _ lect . notes math .
_ * 110 * ( 1969 ) 171 - 189 . | _ zero forcing number _ , @xmath0 , of a graph @xmath1 is the minimum cardinality of a set @xmath2 of black vertices ( whereas vertices in @xmath3 are colored white ) such that @xmath4 is turned black after finitely many applications of the color - change rule " : a white vertex is converted black if it is the only white neighbor of a black vertex .
zero forcing number was introduced and used to bound the minimum rank of graphs by the
aim minimum rank special graphs work group " .
let @xmath5 and @xmath6 be disjoint copies of a graph @xmath1 and let @xmath7 be a function . then a _ functigraph _
@xmath8 has the vertex set @xmath9 and the edge set @xmath10 . for a connected graph @xmath1 of order @xmath11 , it is readily seen that @xmath12 for any permutation @xmath13 ; we show that @xmath14 for any function @xmath15 , where @xmath16 is the minimum degree of @xmath1 .
we give examples showing that there does not exist a function @xmath17 such that , for every pair @xmath18 , @xmath19 or @xmath20 .
we further investigate the zero forcing number of functigraphs on complete graphs , on cycles , and on paths .
zero forcing set , zero forcing number , permutation graph , generalized prism , functigraph , complete graph , cycle , path 05c50 , 05c38 , 05d99 + |
algorithms for dynamic graphs have traditionally assumed that the graph evolves according to a completely arbitrary sequence of insertions and deletions of graph elements .
this model makes minimal assumptions but often sacrifices efficiency for generality .
for example , real world networks ( router networks , road networks , etc . ) _ do _ change slowly over time .
however , the real dynamism of the networks comes from the frequent _ failure _ of edges / nodes and their subsequent recovery . in this paper
we study connectivity problems in the _ dynamic subgraph model _ , which attempts to accurately model this type of dynamism .
it is assumed that there is some fixed underlying graph whose nodes and edges can be _ off _ ( failed ) or _ on _ ; queries ( connectivity queries , in our case ) then answer questions about the subgraph turned on .
the power of this model ( compared to the fully dynamic graph model ) stems from the ability to preprocess the underlying graph in advance .
there are two natural variants of the dynamic subgraph model . in the @xmath0-failure version failures and recoveries
occur in lockstep : a set @xmath8 of @xmath9 edges / nodes fail together .
our goal is to process @xmath8 , ideally in @xmath10 time , in order to answer connectivity queries in @xmath11 . here
@xmath0 may or may not be a parameter of the algorithm . in the fully dynamic model , node /
edge failures and recoveries are presented one at a time and intermixed with connectivity queries , whereas in the decremental model the updates are restricted to failures .
[ [ results . ] ] results .
+ + + + + + + + we give new algorithms for subgraph connectivity ( aka connectivity oracles ) on undirected planar graphs in the @xmath0-failure model and the decremental model , all of which require linear preprocessing time .
when failures are restricted to edges , we give an especially simple connectivity oracle that processes @xmath0 edge failures ( for any @xmath0 ) in @xmath12 time and subsequently answers queries in @xmath13 time . here
@xmath3 and @xmath4 refer to the time for sorting @xmath0 integers in the universe @xmath14 and @xmath4 for the time for predecessor search , given @xmath1 preprocessing time .
( it is known that @xmath15 deterministically , @xmath16 randomized @xcite , and @xmath17 randomized if @xmath18 @xcite . for predecessor search
the bound is @xmath19 deterministically @xcite and @xmath20 randomized @xcite . )
the problem becomes more complicated when vertices fail since we can not , in general , spend time proportional to their degrees .
our second algorithm is a @xmath0-failure connectivity oracle for edge and vertex failures with the same parameters ( linear preprocessing , @xmath12 to process @xmath0 failures , @xmath13 time per query ) .
it consists of two parts : a solution for triconnected graphs and a generic reduction from @xmath0-failure oracles in general graphs to @xmath0-failure oracles in triconnected graphs .
triconnectivity plays an important role in the algorithm as it allows us to apply barnette s theorem @xcite , which states that every triconnected planar graph contains a degree-3 spanning tree .
it is known @xcite that predecessor search is reducible to the @xmath0-edge / node failure connectivity problem on trees ( and hence planar graphs ) .
our query time is therefore provably optimal .
in particular , ptracu and thorup s lower bound @xcite implies that @xmath20 query time can not be beaten in general , even given @xmath21 time to preprocess the edge / node failures .
our third algorithm is in the decremental model . in triconnected planar graphs we can support vertex failures in @xmath6 amortized time and connectivity queries in @xmath6 time , whereas in general planar graphs both bounds become @xmath7 . of failures ,
i.e. , in triconnected graphs processing @xmath22 failures takes @xmath23 time total . ]
the logarithmic slowdown comes from a new reduction from ( planar ) dynamic subgraph connectivity to the same problem in triconnected graphs .
[ [ prior - work - on - dynamic - connectivity . ] ] prior work on dynamic connectivity .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + before surveying dynamic subgraph connectivity it is instructive to see what type of `` off the shelf '' solutions are available using traditional dynamic graph algorithms .
the best known dynamic connectivity algorithms for general undirected graphs require @xmath24 worst case time per edge update @xcite or @xmath7 time amortized @xcite .
vertex updates are simulated with @xmath25 edge updates . in dynamic planar graphs
the best connectivity algorithms take @xmath6 time per edge update @xcite . [
[ prior - work - on - subgraph - connectivity . ] ] prior work on subgraph connectivity .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the dynamic subgraph model was proposed explicitly by frigioni and italiano @xcite , who proved that in planar graphs , node failures and recoveries could be supported in @xmath26 amortized time per operation and connectivity queries in @xmath26 worst case time .
an earlier algorithm of eppstein et al .
@xcite implies that edge failures , edge recoveries , and connectivity queries in planar graphs require @xmath6 time .
in general graphs , chan , ptracu , and roditty @xcite ( improving @xcite ) showed that node updates could be supported in amortized time @xmath27 , where @xmath28 is the number of edges , and connectivity queries in @xmath29 .
chan et al .
@xcite gave numerous applications of subgraph connectivity to geometric connectivity problems .
the first algorithm with worst - case guarantees was given by duan @xcite , who showed that node updates and queries require only @xmath30 and @xmath31 time , respectively .
in the @xmath0-_edge _ failure model ptracu and thorup @xcite gave a connectivity oracle for general graphs that processes @xmath0 failures in @xmath32 time and answers queries in @xmath33 time .
however , their structure requires _ exponential _ preprocessing time ; a variant constructible in polynomial time has a slower update time : @xmath34 .
duan and pettie @xcite gave a connectivity oracle in the @xmath0-_node _ failure model occupying @xmath35 space that processes failures in @xmath36 time and answers queries in @xmath17 time , where @xmath37 depends on @xmath38 .
they also showed that a @xmath0-edge failure oracle could be constructed in @xmath39 time with @xmath40 update time and @xmath33 query time .
the _ distance sensitivity _ oracles avoiding 1 node failure @xcite or 2 node failures @xcite also , as a special case , answer 1- and 2-failure connectivity queries on _ directed _ graphs in @xmath41 and @xmath6 time , respectively .
these data structures occupy @xmath42 space . [ [ overview ] ] overview + + + + + + + + section [ sec : defsnotres ] reviews notation and terminology . in section
[ sec : multiedge ] we describe our planar @xmath0-edge failure connectivity oracle . in section [ sec : multivertexedge ] we give a @xmath0-vertex failure oracle for triconnected planar graphs and in section [ sec : singlevertexedge ] we extend it to a decremental subgraph connectivity oracle for triconnected graphs . finally , in section [ sect : reduction - to - tri ] we give reductions from general graphs to triconnected graphs , which do not assume planarity .
we assume that all graphs considered are undirected . a _ planar graph _ is a graph that can be drawn in the plane such that no two edges cross .
we refer to such a drawing as a _ plane graph_. a plane graph @xmath43 partitions the plane into maximal open connected sets and we refer to the closure of these sets as the _ faces _ of @xmath44 . if @xmath44 is connected , we define a plane graph , called the _ dual graph _ @xmath45 of @xmath44 , as follows .
associated with each face @xmath22 of @xmath44 is a vertex in @xmath45 which we identify with @xmath22 and which we draw inside @xmath22 . for each edge
@xmath46 in @xmath44 , there is an edge @xmath47 in @xmath45 , where @xmath48 and @xmath49 are the faces in @xmath44 incident to @xmath46 .
we draw @xmath47 such that it crosses @xmath46 exactly once and crosses no other edge in @xmath44 or in @xmath45 .
we identify @xmath47 with @xmath46 .
it can be shown that @xmath45 is also connected and that @xmath50 . in particular , each face in @xmath45 corresponds to a vertex in @xmath44 .
we refer to @xmath44 as the _ primal _ graph . for vertices @xmath51 and @xmath52 in a rooted tree
@xmath53 , we let @xmath54 denote the lowest common ancestor of @xmath51 and @xmath52 in @xmath53 . consider a spanning tree @xmath53 in primal graph @xmath44 .
it is well - known that the edges not present in @xmath53 form a spanning tree @xmath55 of @xmath45 .
we call @xmath55 the _ dual tree _ of @xmath53 .
the following lemma is a well - known result @xcite .
[ lem : primalduallink ] for an edge @xmath46 in primal tree @xmath53 , let @xmath22 and @xmath56 be the faces to either side of @xmath46 in @xmath44 .
then the edges of @xmath57 that connect the two subtrees of @xmath58 are exactly those on the simple path in @xmath55 from @xmath22 to @xmath56 .
a co - path is a sequence of faces that is a sequence of vertices in the dual that form a path .
a co - path _ avoids _ a set of edges @xmath8 if every consecutive pair of faces in the sequence shares an edge not in @xmath8 .
in this section , we develop a data structure that , given a planar undirected graph @xmath43 , an integer @xmath59 , and a dynamic subset of at most @xmath0 _ failed edges _ , supports the following operations : if @xmath51 and @xmath52 are connected in @xmath62 , then there is a @xmath51-to-@xmath52 path @xmath63 in @xmath44 not using any edges in @xmath8 .
this path , viewed as a sequence of vertices in @xmath44 is a sequence of faces , or co - path , in @xmath45 .
since consecutive vertices in @xmath63 are adjacent by way of edges not in @xmath8 , consecutive faces in the identified co - path share an edge not in @xmath8 : this @xmath51-to-@xmath52 co - path avoids @xmath8 .
conversely , let @xmath63 be a @xmath51-to-@xmath52 co - path avoiding @xmath8 .
@xmath63 is a sequence of faces in @xmath45 , and so is a sequence of vertices in @xmath44 .
consecutive faces on @xmath63 in @xmath45 share an edge avoiding @xmath8 and so are adjacent by way of edges not in @xmath8 : @xmath63 is a path in @xmath44 not using any edges in @xmath8 .
consider the subgraph @xmath64 of @xmath45 consisting of the failed edges , @xmath8 .
let @xmath64 inherit the embedding of @xmath45 .
we refer to the faces of @xmath64 as _
each superface corresponds to the union of faces of @xmath45 .
[ lem : superfaces ] let @xmath22 and @xmath56 be faces of @xmath45 .
there is an @xmath22-to-@xmath56 co - path in @xmath45 avoiding @xmath8 iff @xmath22 and @xmath56 are contained in the same superface of @xmath64 .
suppose @xmath22 and @xmath56 are contained in different superfaces ; let @xmath65 be the set of edges that bound the superface containing @xmath22 .
any @xmath22-to-@xmath56 co - path requires a face on either side of @xmath65 and so can not avoid @xmath65 , which is a subset of @xmath8 .
suppose otherwise ; let @xmath66 be the set of faces of @xmath45 that form the superface @xmath67 containing @xmath22 and @xmath56 . since @xmath67 is a face of @xmath64 , there is a curve @xmath68 contained entirely in @xmath67 that starts inside @xmath22 and ends inside @xmath56 .
consider the sequence of faces in @xmath66 that this path visits ; this is an @xmath8-avoiding @xmath22-to-@xmath56 co - path as consecutive faces must share an edge that @xmath68 crosses and this edge can not be in @xmath8 .
fix a rooted spanning tree @xmath53 of the primal graph @xmath44 .
we use @xmath53 to determine the superfaces of @xmath64 containing the faces corresponding to the query vertices @xmath51 and @xmath52 . to do so
, we require constant - time lca query support for @xmath53 @xcite .
the edges @xmath8 are listed in no particular order .
we start by building the planar embedding of the subgraph @xmath64 induced by @xmath8 that is inherited from @xmath45 .
let @xmath69 be the endpoints of @xmath8 ( in the dual sense ) .
for each vertex @xmath70 identify the edges of @xmath8 incident to @xmath52 and their cyclic ordering in @xmath45 .
we can compute these orderings in @xmath1 time .
the boundaries of the superfaces given by @xmath64 can be traversed in @xmath17 time given this combinatorial embedding .
the set @xmath69 is the set of faces of @xmath45 that are along the boundaries of superfaces of @xmath64 .
label a dual face / primal vertex in @xmath69 with the superface that contains it ; mark the vertices of the static tree @xmath53 with these superface labels . to answer a query
` connected?`@xmath61 , suppose we have identified the first and last marked vertices ( if any ) on the @xmath51-to-@xmath52 path in @xmath53 ; call them @xmath71 and @xmath72 . by lemmas [ lem : facepath ] and [ lem : superfaces ] , @xmath51 and @xmath52 are connected in @xmath62 iff @xmath71 and @xmath72 do not exist , or @xmath71 and @xmath72 are labelled with the same superface .
therefore , we need only identify @xmath71 and @xmath72 , if they exist .
lemma [ lem : marked - ancestor ] shows that finding @xmath71 and @xmath72 is reducible to one least common ancestor query and @xmath41 predecessor queries .
[ lem : marked - ancestor ] let @xmath53 be a tree of size @xmath73 and @xmath74 be a set of _ marked _ vertices , with @xmath75 . then after @xmath25 preprocessing ( independent of @xmath76 ) , an @xmath17-size data structure can be constructed in @xmath12 time that answers the following query in @xmath13 time : given @xmath77 , what are the first and last @xmath76-vertices on the @xmath51-@xmath52 path ?
recall that @xmath53 is rooted .
let @xmath78 .
the first marked vertex on the @xmath51-to-@xmath52 path is either the first marked vertex on the @xmath51-to-@xmath79 path or the last marked vertex on the @xmath52-to-@xmath79 path .
thus , we can assume without loss of generality that @xmath52 is an ancestor of @xmath51 .
fix an arbitrary dfs of @xmath53 and let @xmath80 ( resp .
@xmath81 ) be the time when @xmath82 is pushed onto ( resp . popped off ) the stack during dfs . given @xmath76 , we first sort the @xmath83 and @xmath84 indices of its elements , in @xmath12 time , which allows us to label each @xmath85 with the nearest strictly ancestral @xmath76-vertex @xmath86 .
here it is convenient to assume that the root is marked ( honorarily , if not in @xmath76 ) so @xmath86 is defined everywhere except the root .
we build a global predecessor structure on the set @xmath87 and local predecessor structures on the sets @xmath88 , where @xmath85 and @xmath89 is the set of `` immediate '' descendants in @xmath76 connected by a path of non-@xmath76 vertices . to answer a query @xmath90 ( where @xmath52 is ancestral to @xmath51 ) we first find the closest marked ancestors @xmath91 as follows .
let @xmath92 be the predecessor of @xmath93 in @xmath94 . if @xmath95 for some @xmath96 then @xmath82 is an immediate descendant of @xmath97 and @xmath98 .
if @xmath99 then let @xmath100 . see figure [ fig : marked - ancestor](a ) .
it follows that @xmath101 is an ancestor of @xmath51 and that there are no other @xmath76-vertices on the path from @xmath51 to @xmath101 .
if @xmath101 is ancestral to @xmath52 then there are no marked nodes on the @xmath51-@xmath52 path , so assume this is not the case .
in order to find the _ last _ marked vertex on the path from @xmath51 to @xmath52 we search for the predecessor of @xmath93 in @xmath102 , say @xmath103 .
since there is _ some _ marked vertex on the path from @xmath51 to @xmath104 it follows that @xmath51 is a descendant of @xmath97 , which is a descendant of @xmath52 and that there are no other marked vertices on the path from @xmath97 to @xmath52 .
see figure [ fig : marked - ancestor](b ) .
an @xmath105 tree @xmath106 represents how a biconnected graph @xmath44 is assembled form its triconnected components through _ merging _ operations .
all separating pairs of @xmath44 are given implicitly by @xmath106 .
let @xmath107 and @xmath108 be two edge - labeled multigraphs , where @xmath109 and where @xmath110 appears in both @xmath111 and @xmath112 with the same label .
then _ merging _ @xmath107 and @xmath108 results in a new graph @xmath113 .
that is , @xmath46 does not appear in the merged graph but @xmath82 and @xmath97 may still be adjacent in @xmath114 if there were multiple edges between @xmath82 and @xmath97 in either @xmath107 or @xmath108 .
each node @xmath115 is associated with a triple @xmath116 , where @xmath117 is the _ model _ graph , one edge @xmath118 is distinguished as the _ polar edge _ whose endpoints are _
poles_. let @xmath119 be the parent of @xmath51 in @xmath106 and @xmath120 the subtree rooted at @xmath51 .
the graphs @xmath117 and @xmath121 have exactly two vertices in common , namely the endpoints of @xmath122 , and each contains identically labeled copies of @xmath122 .
moreover , for any two @xmath106-nodes on opposite sides of the edge @xmath123 , their model trees do not intersect , except possibly at the poles of @xmath122 .
the graph @xmath124 is formed by merging all model trees in @xmath120 .
that is , @xmath51 has children @xmath125 where @xmath126 and @xmath127 contains the polar edge @xmath128 .
obtain @xmath124 by merging @xmath129 with @xmath117 along the polar edges @xmath130 .
( observe that the polar edge @xmath131 remains in @xmath132 . )
let @xmath133 be @xmath124 without its polar edge @xmath122 .
the tree @xmath106 is conceptually constructed in a top - down fashion as follows .
( linear time triconnectivity algorithms @xcite can be used to construct @xmath106 . ) the root @xmath134 is associated with @xmath135 and an arbitrary polar edge @xmath136 . in general , we are given a @xmath106-node @xmath51 and pair @xmath137 .
we must select a model multigraph @xmath117 and then recursively construct the subtree @xmath120 .
the reader may want to refer to an illustration of a graph and its @xmath105-tree in figure [ fig : spqr ] .
the endpoints of @xmath138 split @xmath132 into @xmath139 components @xmath140 where @xmath141 .
let @xmath127 be @xmath142 augmented with a freshly labeled polar edge @xmath143 and let @xmath117 consist of @xmath139 parallel edges @xmath144 . clearly merging @xmath129 with @xmath117 yields
@xmath124 . for each nontrivial @xmath142 ( nontrivial means having more than one edge ) , create a new child @xmath145 of @xmath51 and recursively compute its subtree in @xmath106 .
trees @xcite trivial graphs are actually assigned to children of @xmath51 . they are called @xmath146-nodes . ]
let @xmath138 and let @xmath147 be the articulation points of @xmath133 .
if there are any such points ( @xmath141 ) let @xmath148 be the partition of @xmath124 into edge - disjoint subgraphs such that @xmath142 and @xmath149 meet at @xmath150 .
let @xmath117 be the cycle @xmath151 and let @xmath127 be @xmath142 augmented with a polar edge @xmath152 ( where @xmath153 ) , whose label matches the corresponding edge in @xmath154 .
for each non - trivial @xmath142 form a new child @xmath145 of @xmath51 and recursively compute its subtree in @xmath106 .
the previous two cases do not apply . a separating pair @xmath156 of @xmath124 splits it into some number of components @xmath157 with @xmath158 .
let @xmath159 and @xmath160 be the union of @xmath161 . call @xmath162 _
maximal _ if there is no separating pair @xmath163 for which @xmath164 is strictly contained in @xmath165 .
let @xmath166 be the set of @xmath167 maximal separating pairs with respect to @xmath122 .
we give @xmath51 children @xmath125 where @xmath145 is given the graph @xmath168 and polar edge @xmath128 .
the model graph @xmath117 is induced by @xmath169 , that is , it is obtained from @xmath124 be replacing @xmath170 with the @xmath145 s polar edge @xmath128 , for all @xmath92 .
the maximality of the separating pairs ensures that @xmath171 is triconnected , also called _
rigid_. the edges in a model graph @xmath117 that do not appear in @xmath44 are _ artificial _ and represent paths in @xmath44 .
our goal is to reduce a connectivity query in @xmath44 to a constant number of connectivity queries in model graphs , so it is important that we identify the set of invalidated artificial edges .
let @xmath172 , where @xmath173 , be the _ most _ ancestral @xmath106-node @xmath51 for which @xmath174 and let @xmath175 .
( note that @xmath176 can appear in an unbounded number of model trees as a pole . )
we begin by coloring _
those @xmath106-nodes @xmath51 for which the endpoints of @xmath122 may be disconnected in @xmath177 . specifically , repeatedly color blue any @xmath51 satisfying one of the following criteria : 1 .
@xmath178 for some @xmath179 .
@xmath51 is an @xmath66-node with at least one blue child .
( @xmath180 is a path , so the removal of one vertex or edge disconnects @xmath51 s poles . )
@xmath51 is either an @xmath155- or @xmath63-node with at least two blue children .
( there are at least two internally vertex - disjoint paths between the endpoints of @xmath122 in @xmath180 , so the removal of just one vertex or edge can not disconnect @xmath51 s poles . ) there are clearly at most @xmath0 blue nodes due to criterion ( 1 ) and at most @xmath182 due to criterion ( 3 ) . since , by observation [ obs : psnodes ] , there are no adjacent @xmath66-nodes , the number due to criterion ( 2 ) is no more than the number due to criteria ( 1 ) and ( 3 )
. they can be found in @xmath17 time as follows .
for each @xmath179 color @xmath172 blue and continue to color any immediate ancestors blue if they satisfy criteria ( 2 ) or ( 3 ) .
the first ancestor not colored blue is noted as having one blue child .
the time per @xmath8-node is @xmath41 plus the number of formerly uncolored ancestors colored blue .
next we color _ purple _ all blue nodes @xmath51 for which the poles of @xmath138 are disconnected in @xmath177 ( that is , ignoring parts of the graph outside @xmath51 s subtree ) , as follows . traverse the blue nodes in @xmath106 in postorder .
if either of @xmath183 or @xmath184 is in @xmath8 then they are trivially disconnected , so color @xmath51 purple .
if @xmath51 is a @xmath63-node then @xmath183 and @xmath184 are still connected if they are connected in _
any _ child of @xmath51 , so color @xmath51 purple if all its children are purple .
if @xmath51 is an @xmath66-node then @xmath183 and @xmath184 are disconnected if the poles of any child @xmath145 of @xmath51 are disconnected , so color @xmath51 purple if any child is purple . finally , if @xmath51 is an @xmath155-node , construct a connectivity oracle @xmath185 for @xmath117 avoiding vertex failures @xmath186 and edge failures @xmath187 .
( the edge failures reflect the fact that the polar edges of purple children are definitely invalidated by the failures and it is not yet known if @xmath122 is invalidated . ) perform one query on @xmath185 to determine if @xmath188 are disconnected in @xmath177 and , if so , color @xmath51 purple .
we also construct connectivity oracles for each colored @xmath66- and @xmath63-node @xmath51 in @xmath189 .
this is trivial if @xmath51 is a @xmath63-node .
( @xmath117 contains two vertices , which are either connected or not . )
if @xmath51 is an @xmath66-node then @xmath117 is a cycle and a connectivity query can be answered with predecessor search , in @xmath2 time .
the time to construct oracles for all colored nodes is @xmath190 .
each colored node @xmath51 can contribute one edge failure to @xmath192 , namely the polar edge @xmath122 .
note that @xmath186 can include at most two additional vertices than @xmath193 , namely the poles of @xmath122 .
thus , the total number of failures is linear in @xmath0 and the number of colored nodes , which are fewer than @xmath181 .
let @xmath194 be the farthest ancestral pole reachable from @xmath173 , or @xmath176 if there is no such pole .
specifically , let @xmath115 be the farthest ancestor of @xmath172 such that @xmath176 is connected to either pole in @xmath122 in @xmath195 .
if there is no such @xmath51 ( @xmath176 is disconnected from @xmath196 ) then @xmath197 and if @xmath176 is connected to both poles @xmath122 then @xmath194 can be either one .
our goal is to preprocess @xmath8 so that we can quickly evaluate @xmath198 .
let @xmath199 be either pole in @xmath196 connected to @xmath176 in @xmath200 , or undefined if there is no such pole .
let @xmath201 be the farthest ancestral pole reachable from @xmath176 via uncolored @xmath106 nodes , defined precisely as follows .
if @xmath202 and @xmath51 is colored then @xmath201 is undefined . if @xmath51 is uncolored and @xmath176 is disconnected from both of @xmath51 s poles ( this can only happen
if they are in the failure set ) then @xmath201 is undefined .
otherwise let @xmath203 be the farthest ancestor of @xmath51 such that the @xmath106-path from @xmath51 to @xmath203 is uncolored and at least one pole of @xmath203 is not in @xmath8 .
then @xmath201 is any non-@xmath8 pole of @xmath203 . by lemma [ lem : marked - ancestor ]
, @xmath106 can be preprocessed in @xmath1 time to support @xmath204 queries in @xmath2 time . 1 . if @xmath179 then @xmath194 is undefined .
2 . if @xmath51 is the root of @xmath106 or is a colored node without a colored ancestor then @xmath194 is a pole of the root of @xmath106 .
3 . if @xmath172 is colored then let @xmath197 if @xmath199 is undefined ( it is disconnected from both poles of @xmath172 ) and @xmath205 otherwise .
4 . if @xmath172 is uncolored then let @xmath197 if @xmath201 is undefined ( it is disconnected from both poles of @xmath172 ) or @xmath201 if @xmath206 is undefined ( @xmath201 is disconnected from both poles of @xmath207 ) or @xmath208 otherwise
. the query asks whether @xmath82 and @xmath210 are connected in @xmath11 , where @xmath211 .
our goal is to reduce this to @xmath41 connectivity queries to the oracles associated with colored @xmath106-nodes .
let @xmath212 if @xmath213 is uncolored and @xmath82 if @xmath213 is colored .
it follows that @xmath214 is colored . using two connectivity queries on @xmath215 s oracle ( between @xmath216 and the poles of @xmath215 ) determine @xmath217 .
if @xmath218 is defined let @xmath219 ; otherwise let @xmath220 .
it follows directly from the definitions of @xmath221 and @xmath198 that @xmath222 is the most ancestral pole connected to @xmath82 , or is @xmath82 if there is no such pole . in any case
let @xmath223 and define @xmath224 analogously , with respect to @xmath210 .
[ lem : ] vertices @xmath82 and @xmath210 are connected in @xmath11 if and only if @xmath225 and @xmath226 and @xmath227 are connected in @xmath11 .
it requires @xmath228 time to answer a connectivity query .
note that by definition of @xmath221 and @xmath198 , @xmath82 is connected to @xmath216 , @xmath216 is connected to @xmath218 ( if it exists ) , and @xmath218 is connected to @xmath226 , hence @xmath82 is connected to @xmath226 .
thus , @xmath82 and @xmath210 are connected if and only if @xmath226 and @xmath227 are connected , which is determined by @xmath229 s oracle if @xmath225 .
( note that @xmath229 is either colored and we have such an oracle , or @xmath229 is the uncolored root and @xmath230 . )
observe that if @xmath229 is not the root , @xmath226 is , by definition of @xmath198 , disconnected from the poles @xmath231 , and therefore also disconnected from _ every _ vertex in @xmath232 .
the same is true of @xmath227 and @xmath233 . if @xmath234 then either @xmath235 or @xmath236 , that is , @xmath226 and @xmath227 must be disconnected .
we prove theorem [ thm : reduction - to - tri - fully - dynamic ] assuming , for notational simplicity , that the graph is biconnected and associated with an @xmath105-tree @xmath106 .
it is easy to extend this algorithm to all graphs .
an obvious strategy is to maintain the structure from section [ sect : bi - to - tri ] , that is , to maintain the ability to evaluate @xmath204 on all vertices , @xmath238 on vertices in colored @xmath106-nodes , and @xmath198 on the poles of colored nodes .
the first difficulty is that a vertex failure @xmath176 at some @xmath172 deep in @xmath106 can cause virtual edge failures in the model graphs of _ every _ ancestor of @xmath172 and potentially change the colored / uncolored status of such nodes .
thus , we can not afford to explicitly maintain connectivity oracles at each colored node .-
nodes this is equivalent to dynamic and - or tree evaluation . ]
if we do not maintain explicit connectivity oracles then evaluating @xmath194 seems impossible since it is a function of @xmath41 @xmath238-queries at _ each _ colored ancestor of @xmath172 .
our solution involves three main ideas .
the first is to partition @xmath106 into heavy paths , a standard technique that ensures every node - to - root path intersects at most @xmath239 distinct paths .
this introduces a @xmath240-factor overhead and reduces the problem to one on paths , yet it does not help us stop changes deep in @xmath106 from propagating upwards .
two consecutive nodes @xmath51 and @xmath119 on a heavy path share two vertices , namely @xmath122 .
our second idea is to make @xmath119 indifferent to _ all _ changes in @xmath241 by having it maintain two candidate connectivity oracles , one where @xmath122 is invalid ( its endpoints are disconnected in @xmath195 ) and one where @xmath122 is valid .
this solves the propagation problem , but we must be able to determine which of @xmath119 s candidate oracles reflects the _ actual _ failure set @xmath8 .
the third idea is to reduce a @xmath198-query to @xmath6 1d product queries over the domain of @xmath242 boolean matrices .
updates and queries to such a structure will take @xmath243 time .
let each nonleaf @xmath115 choose a child @xmath244 maximizing the number of descendants of @xmath244 .
these choices partition the nodes of @xmath106 into a set @xmath245 of _ heavy paths _ such that all leaf - to - root paths intersect at most @xmath240 such paths .
let @xmath246 be the heavy path containing @xmath51 and @xmath247 be the most ancestral node in @xmath246 .
define @xmath248 to be the _ subpolar edge _ of @xmath51 let @xmath249 be the subgraph of @xmath133 induced by @xmath250 , that is , @xmath133 excluding vertices in @xmath251 .
each non - leaf @xmath115 maintains two connectivity oracles for @xmath133 based on the current set @xmath8 of failed vertices .
let @xmath252 be the failed vertices in @xmath253 , excluding the poles @xmath122 , if @xmath51 is not the root of @xmath106 .
let @xmath254 and @xmath255 reflect versions of reality where all or none of the vertices in @xmath244 s subtree fail .
the oracle @xmath256 answers connectivity queries between @xmath257 vertices in the graph @xmath258 whereas @xmath259 is defined with respect to @xmath260 .
if we restrict our attention to connectivity in @xmath257 , at least one of @xmath259 and @xmath256 behaves correctly , with respect to the failure set @xmath261 . in order to determine which one is correct
let us dispense with the old blue / purple / uncolored coloring system - nodes these colors since @xmath51 may be colored purple according to @xmath259 but not @xmath256 . ] from section [ sect : bi - to - tri ] and designate _ all _ @xmath106-nodes _ white _ , _ grey _ , or _
black_. a node @xmath51 is white if the poles @xmath122 are disconnected according to @xmath256 , black if they are connected according to @xmath259 , and gray if they are connected according to @xmath256 but not @xmath259 . in other words , in a gray @xmath51 , all paths between @xmath122 in @xmath262 go through the subpoles @xmath263 .
if @xmath51 is a leaf it has no subpoles and therefore can not be gray , since @xmath264 . before continuing ,
let us note that before any vertex failures occur , all non - leaf @xmath66-nodes are gray ( removing @xmath122 and @xmath263 disconnects the poles in @xmath117 , which is a cycle ) and all @xmath63- and @xmath155-nodes are black since there are at least three edge - disjoint paths between @xmath122 in @xmath117 . in order to determine whether the poles @xmath122 are actually connected in @xmath262 it suffices to maintain a predecessor structure on all non - gray nodes in @xmath246 , which we view as oriented towards @xmath265 .
let @xmath101 be the predecessor of @xmath51 on @xmath246 , that is , the first non - gray descendant of @xmath51 on @xmath246 , which may be @xmath51 itself .
( there must be such a @xmath101 since leaves can not be gray . )
if @xmath101 is black then the poles @xmath122 are connected in @xmath262 , and if white , they are not .
it takes @xmath33 time per query or to delete / insert any node into the predecessor structure , which corresponds to a color transition to / from gray .
consider the failure or recovery of a vertex @xmath176 with @xmath266 and @xmath267 .
this change might cause the color of @xmath51 to change , but , by definition , can not invalidate the color of any other @xmath268 .
now consider @xmath269 , if @xmath270 is not already the root of @xmath106 .
the twin connectivity oracles for @xmath52 entertain the possibility that @xmath271 is invalid or valid , but both must represent the _ actual _ validity of the edge @xmath272 , which may have changed due to the failure / recovery of @xmath176 .
thus , changing the status of @xmath176 may affect @xmath52 but no other nodes on @xmath273 and , in general , affects up to one node on each heavy path traversed from @xmath172 to the root of @xmath106 .
suppose @xmath274 is a heavy path .
if we are only concerned with connectivity between poles of nodes in @xmath63 , each node @xmath145 is effectively in one of 16 states depending on whether each pole in @xmath275 is connected , in the graph @xmath276 , to each pole in @xmath128 .
let @xmath277 be a @xmath242 boolean matrix representing this connectivity .
( in other words , a node s poles are ordered in some consistent fashion to map to rows and columns . )
it follows that the boolean product @xmath278 represents the connectivity between @xmath275 and @xmath279 in the graph @xmath280 .
our algorithm for computing @xmath194 ( and answering connectivity queries ) requires a dynamic data structure for answering various product queries . 1 . `
init`@xmath281 : set @xmath282 , for @xmath283 .
update`@xmath284 : @xmath285 , where @xmath286 is a @xmath242 boolean matrix .
product`@xmath287 : return @xmath288 .
search`@xmath289 : return @xmath290 , given boolean row vector @xmath82 and @xmath242 mask @xmath210 .
( sketch ) each @xmath277 requires 4 bits to represent , so @xmath293 matrices can be packed into one machine word .
we maintain a @xmath293-way tree over the array , where @xmath294 are at the leaves and internal nodes store the product of their descendant leaves .
all operations either need to update or retrieve @xmath41 words of information at each of @xmath295 levels , which can be done in @xmath295 time by tabulating various functions on @xmath293 bits in advance , in @xmath291 time .
define @xmath296 for @xmath297 , to be the pole most ancestral to @xmath172 connected to @xmath176 in @xmath298 , or @xmath176 if there is no such pole . if @xmath296 is not a pole of @xmath270 then clearly @xmath299 .
in general a @xmath194 query is easily reduced to @xmath240 @xmath300 queries , for the heavy paths @xmath63 intersecting the path from @xmath172 to the root of @xmath106 , so we shall focus our attention solely on evaluating @xmath296 , where @xmath301 .
we compute @xmath296 as follows . 1 .
our first task is to determine the boolean vector @xmath82 representing the connectivity between @xmath176 and the poles @xmath128 in @xmath302 . using one predecessor query on the non - gray nodes in @xmath63 we determine which @xmath303 reflects reality , where @xmath304 , then determine @xmath82 with two connectivity queries to @xmath303 .
if @xmath305 then let @xmath306 and halt .
otherwise let @xmath307 ` search`@xmath308 .
let @xmath309 be the poles reachable from @xmath176 in @xmath310 .
it follows that @xmath176 is disconnected from both poles of @xmath311 in @xmath312 , if @xmath313 .
if there is some pole in @xmath314 let @xmath296 be any such pole and halt .
( recall that @xmath315 is the set of failures in @xmath316 _ excluding _ its poles . )
if the procedure has not halted then @xmath317 .
for any index @xmath318 let @xmath319 .
without loss of generality suppose @xmath320 , and that @xmath321 $ ] is the minimum index such that @xmath322 and , if @xmath323 , @xmath324 .
( recall that a pole can appear in an unbounded number of nodes along @xmath63 . )
at least one of the poles in @xmath325 is connected to @xmath176 .
if @xmath326 is disconnected from @xmath176 then @xmath296 can be either pole of @xmath325 connected to @xmath176 .
otherwise @xmath296 is the most ancestral pole among @xmath327 connected to @xmath176 , which can be found as follows .
let @xmath328 `
search`@xmath329 , where the first and second arguments encode that we are only interested in connectivity between @xmath327 . if @xmath330 then @xmath331 , otherwise @xmath332 where @xmath333 is the maximum index such that @xmath334 .
the time bounds claimed in theorem [ thm : reduction - to - tri - fully - dynamic ] follow easily .
it takes linear time to build the @xmath105-tree , decompose it into heavy paths , build the white / black predecessor structure and boolean product structure on each heavy path , and augment it with various pointers , e.g. , pointers from each node to the root of its heavy path .
a vertex update on @xmath176 induces vertex updates in the oracles @xmath335 and @xmath336 , updates to the color and @xmath242 matrix of @xmath172 , which take @xmath243 time , then @xmath6 edge updates in ancestors of @xmath172 together with @xmath6 color and matrix updates . a query on @xmath337 amounts to computing a connectivity query between @xmath338 and @xmath339 .
a @xmath198 evaluation reduces to @xmath240 @xmath300 evaluations , which involve a constant number of queries and operations on the matrix product data structure , for a total time of @xmath340 . | we consider dynamic subgraph connectivity problems for planar undirected graphs . in this model
there is a fixed underlying planar graph , where each edge and vertex is either `` off '' ( failed ) or `` on '' ( recovered ) . we wish to answer connectivity queries with respect to the `` on '' subgraph .
the model has two natural variants , one in which there are @xmath0 edge / vertex failures that precede all connectivity queries , and one in which failures / recoveries and queries are intermixed .
we present a @xmath0-failure connectivity oracle for planar graphs that processes any @xmath0 edge / vertex failures in @xmath1 time so that connectivity queries can be answered in @xmath2 time .
( here @xmath3 and @xmath4 are the time for integer sorting and integer predecessor search over a subset of @xmath5 $ ] of size @xmath0 . )
our algorithm has two discrete parts .
the first is an algorithm tailored to triconnected planar graphs .
it makes use of barnette s theorem , which states that every triconnected planar graph contains a degree-3 spanning tree .
the second part is a generic reduction from general ( planar ) graphs to triconnected ( planar ) graphs .
our algorithm is , moreover , provably optimal .
an implication of ptracu and thorup s lower bound on predecessor search is that no @xmath0-failure connectivity oracle ( even on trees ) can beat @xmath2 query time .
we extend our algorithms to the subgraph connectivity model where edge / vertex failures ( but no recoveries ) are intermixed with connectivity queries . in triconnected planar graphs
each failure and query is handled in @xmath6 time ( amortized ) , whereas in general planar graphs both bounds become @xmath7 . |
in the most studied lattice model for the collapse transition ( also called coil - globule transition ) of polymers @xcite , the chains are represented by self - avoiding walks , so that the bonds of the chain are placed on lattice edges and the monomers are located on the sites .
an attractive interaction between monomers on nearest - neighbor sites which are not linked by bonds is added .
the competition between the repulsive excluded - volume interactions and the attractive interactions leads to a change in the polymerization transition of sasaw ( self - attracting self - avoiding walks ) in a grand - canonical formalism .
experimentally , as the temperature of a polymer solution is lowered , the chain changes its configuration from extended ( coil ) to collapsed ( globule ) , as the temperature crosses a particular value , called the @xmath0-point . for weak attraction ,
the transition between a non - polymerized and a polymerized phase in the monomer fugacity - temperature plane is continuous , becoming discontinuous as the attraction is increased , so that the collapse transition is a tricritical point in this model .
the nature of this transition was studied through a mapping of the polymer model onto a ferromagnetic @xmath1 model in the limit @xmath2 , due to de gennes @xcite .
the contributions to the high - temperature series expansion of the magnetic model are represented by self - avoiding walks on the lattice .
mean field tricritical exponents are found in three dimensions , with logarithmic corrections . in two dimensions ,
non - classical exponents are expected , and a major result is due to duplantier and saleur ( ds ) , who managed to derive the exact tricritical exponents for the sasaw model on a honeycomb lattice @xcite .
the proposal of this model , which requires some fine tuning to allow it to be solved , as a generic result for the collapse transition has been discussed in the literature soon after its proposal , and lead to numerical results which seem to support its robustness @xcite .
another aspect of the problem are the phase diagrams of the variety of models related to the problem of the collapse transition .
even a slight change in the sasaw model , if we assume that the attractive interactions are between polymer bonds on opposite sides of elementary squares of the square lattice , leads to a phase diagram which is different from the one found when the interactions are between monomers on nearest neighbor ( nn ) sites . in this model ,
an additional polymerized phase appears , besides the non - polymerized and the regular polymerized phases , and the critical polymerization line ends at a critical endpoint @xcite .
the self - avoidance constraint may also be changed , allowing for more than one monomer at the same site , but still restricting the number of polymer bonds on each lattice edge to at most one @xcite .
this generalization of saw s , usually called trails , has the distinctive feature that the interactions are now between monomers _ at the same site_. on two - dimensional lattices , the trails may collide or cross at each site , and in the original model , which we will call the isat ( interacting self - avoiding trails ) model , the statistical weights of both configurations are the same . if , the trails are not allowed to cross themselves , so that only collisions of the trails on sites exist , we have a model called visaw ( vertex interacting self - avoiding walks ) , which was exactly solved by blte and nienhuis ( bn ) @xcite and has critical exponents for the collapse transition distinct from the ones found in the ds model .
there has been much discussion in the literature on which of the two distinct sets of exponents ( ds or bn ) is the generic result for the collapse transition of polymers .
the bn exponents seem to be difficult to find in simulations , since numerical results for the exponents of the bn model seem to be closer to the ds values @xcite .
the inclusion of stiffness in the visaw model , associating a bending energy to elementary bends of the chain , leads to an even richer phase diagram @xcite , with the tricritical points from both integrable models ( ds and bn ) residing on a multicritical line .
the robustness of the ds exponents has also been discussed in a recent paper @xcite , where it has been shown that if crossings of the trails are included in the bn model , so that the lattice is no longer planar , the universality class is changed .
although of course the analytic results mentioned above are of inestimable value , details of the phase diagrams of the different models are not always easily found with these techniques , and this is also true for simulations , in particular in the grand - canonical formalism , where the nature of the collapse transition was recognized to be tricritical by de gennes .
it is , therefore , interesting to study these phase diagrams with approximate analytic tools , among them solutions on hierarchical lattices such as the bethe and the husimi lattice .
indeed , the isat model was recently studied on a four - coordinated husimi lattice built with squares and on a four and six - coordinated bethe lattice @xcite .
rich phase diagrams were found in these studies , with the coil - globule transition for four - coordinated cases ( which are mean - field approximations for the square lattice ) being associated with a bicritical point . such behavior was confirmed in a recent study by pretti @xcite , where the isat model was generalized by including an attractive interaction between nn monomers on single occupied sites not linked by a polymer bond and solved on bethe and husimi lattices with @xmath3 . the visaw model ( when the crossings are forbidden and nn interactions vanish ) , the sasaw ( when crossings and collisions are not allowed ) , the model by wu and bradley @xcite ( when collisions and crossings have the same weight ) and the simple isat model ( when the nn interaction vanishes ) are recovered as particular cases .
given the relevance of the semi - flexible extension of the visaw model @xcite in the discussion about the differences between the ds and the bn critical behaviors for the collapse transition , we investigate another generalization of trail models by including an energy associated to elementary bends .
this is done here for bethe lattice , with different statistical weights associated to crossings and collisions , so that semi - flexible visaw and semi - flexible isat are obtained as particular cases of our model .
it is shown that the inclusion of semi - flexibility does not change the nature of the collapse transition when compared with the flexible isat model studied before @xcite , but an additional polymerized phase appears inside the regular polymerized phase , which is both dense and nematic , since all lattice sites are visited and all bonds are in the same direction . for saw ,
the semi - flexible and the self - attracting cases were studied on the bethe lattice @xcite .
when both effects are present , studies on bethe and husimi lattices show the appearance of a second polymerized phase , which is dense and anisotropic in the sense that bonds in one particular direction are favored @xcite .
the nature of the collapse transition is changed when the stiffness is sufficiently high , so that in this case also the appearance of a second polymerized phase signals the change of the nature of the collapse transition .
the model we study is defined in more detail in section [ defmod ] and its solution on a bethe lattice is presented in section [ sbl ] .
final discussions and the conclusion may be found in section [ conc ] .
we consider a semi - flexible generalized self - avoiding trail ( sisat ) model . in this model ,
each lattice edge can be occupied by at most one polymer bond , which has an activity @xmath4 , where @xmath5 and @xmath6 is the chemical potential of a bond .
the bonds connect monomers , which are placed on the sites of the lattice . for a lattice with coordination number @xmath7 , each lattice site
can be occupied by up to @xmath8 ( for even @xmath7 ) and @xmath9 ( for odd @xmath7 ) monomers . in two dimensions ,
walks meeting at a lattice site may either cross or collide , as is apparent in the generalized isat model on a square lattice depicted in fig .
[ figmodrq ] . in higher dimensions
, however , the distinction between collisions and crossings is no longer clear .
we will restrict our attention to lattices with @xmath3 here , whose solutions may be compared with results for the square lattice .
statistical weights @xmath10 and @xmath11 will be associated for each crossing and collision of the chains , respectively .
note that when @xmath12 we recover the classical isat model , while in the case @xmath13 ( crossings forbidden ) the visaw model is obtained . in order to analyze the effects of the polymer stiffness in such models ,
an additional weight @xmath14 is introduced , associated to one polymer bend .
if the energy @xmath15 associated to an elementary bend of the trails is positive ( @xmath16 ) , we say that the walks are semi - flexible .
this is also illustrated in fig .
[ figmodrq ] .
the grand - canonical partition function of the model is given by @xmath17 where @xmath18 , @xmath19 and @xmath20 are the numbers of bonds , collisions and crossings in the system , respectively .
@xmath21 is the number of bends in sites with a single monomer , since the bends in double visited sites are accounted for in the weight @xmath11 , so that @xmath22 , where @xmath23 is the weight of the collision itself , i. e. , of the _ on - site _ monomer - monomer interaction at colliding sites .
we will mostly restrict ourselves in the discussions to @xmath24 , so that the monomer - monomer interaction energy @xmath25 is the same for crossings and collisions , but it is easy to consider @xmath26 , and this will be done in part of the comments below . the sum in eq .
( [ eqpartfunc ] ) is over all allowed configurations of the walks on the lattice we are considering , which will be the four coordinated bethe lattice here . and @xmath10 , respectively , while bends at sites with two incoming bonds are indicated by @xmath27.,width=264 ]
we solve the model on a four coordinated bethe lattice , which corresponds to the core of a cayley tree as shown in fig .
the extremal monomers of each chain are placed on the surface of the tree .
one possible configuration of three chains on a cayley tree with four generations of sites is shown in fig .
[ bl ] .
.,width=264 ] to solve the model on the bethe lattice we consider sub - trees , defining partial partition functions ( ppf s ) for them for a fixed configuration of the root edge @xcite . for the bethe lattice ,
usually only two root configurations are needed , corresponding to the possibilities of empty or occupied ( by chain bonds ) root edges . however , as discussed above , in the semi - flexible case one may expect the appearance of anisotropic phases , which display orientational ( nematic ) ordering , so that bonds in one or more directions are favored . the study of models which present nematic ordering on such hierarchical lattices presents some difficulties , particularly for @xmath28 .
one way to avoid them is to solve these models on other treelike lattices for which the exact solution is the bethe approximation on regular lattices with the same coordination number .
one of such lattices is the random locally treelike layered ( rltl ) lattice introduced by dhar , rajesh and stilck @xcite to study nematic ordering of monodispersed rigid rods . here
we will follow a simpler option , assuming that at each site of the @xmath3 bethe lattice two incoming bonds are in one direction and the two remaining ones are in a perpendicular direction .
actually , one should keep in mind that in the thermodynamic limit this lattice is effectively infinite dimensional , as was shown by baxter @xcite .
thus , for example , to correctly measure the euclidean distance between two sites on an even - coordinated bethe lattice , one may embed it in a hypercubic lattice whose dimension increases with the number of generations @xcite .
therefore , we will define partial partition functions for four root configurations : @xmath29 for a root edge in direction @xmath30 not occupied by a bond , @xmath31 for a empty root edge in direction @xmath32 , and @xmath33 and @xmath34 for subtrees with occupied root edges in directions @xmath30 and @xmath32 , respectively .
we now may obtain recursion relations for the ppf s of a sub - tree with an additional generation , considering the operation of attaching a new root site and edge to three sub - trees of the preceding generation .
the results are : @xmath35 , \\
g'_{1,2 } & = & z_2[g_{0,1}^2g_{1,2}+2\omega g_{0,1}g_{1,1}g_{0,2}+\tau g_{1,1}^2g_{1,2}],\end{aligned}\ ] ] [ rrgb ] where @xmath36 is the only combination of the weights of double occupied sites that appears in the bethe lattice solution ; this will change if longer range correlations are taken into account , such as on the husimi lattice .
we note that we include the possibility of bonds in the two directions having different activities , although we will discuss the thermodynamic behavior of the model only for @xmath37 . here
, @xmath38 and @xmath39 are ppf s of sub - trees with @xmath40 and @xmath41 generations , respectively . usually , the ppf s diverge in the thermodynamic limit ( when @xmath42 ) .
thus , it is suitable to define the ratios : @xmath43 which should remain finite for non - dense phases , where a finite fraction of the lattice sites is empty .
the recursion relations for these ratios are : @xmath44 [ rrrb ] we find four distinct fixed points for these recursion relations when @xmath37 , which are stable in distinct regions of the parameter space @xmath45 .
they are : * a non - polymerized ( * np * ) phase : @xmath46 , @xmath47 . * a regular polymerized ( * p * ) phase : @xmath48 and finite . * a dense polymerized ( * dp * )
phase : @xmath49 . * a dense anisotropic and nematic ( * an * ) phase : @xmath50 and @xmath51 or @xmath52 and @xmath53 . in the dense phases ,
the edges corresponding to the direction of the ratio which diverges are all occupied by bonds .
it is useful to define the reciprocal ratios @xmath54 to study the fixed points which are associated to these phases . for the * dp * phase we may rewrite the recursion relation eqs .
( [ rrrb ] ) as : @xmath55 so that the * dp * fixed point is now located at the origin @xmath56 . for the anisotropic * an * phase , there are two equivalent fixed points where the chains occupy bonds in one of the two directions .
if we consider the fixed point with bonds in the 1 direction , the recursion relations may be written in terms of the variables @xmath57 and @xmath58 , so that the fixed point again is located at the origin .
the result is : @xmath59 we notice that the product @xmath60 attains a finite value at the * an * fixed point , given by : @xmath61 the region of the parameter space where each fixed point is stable may be found by studying the eigenvalues of the jacobian of the recursion relations : @xmath62 where the @xmath63 s are the appropriate ratios in each case . in the * np * fixed point @xmath64 , the secular equation of the recursion relation ( [ rrrb ] ) is @xmath65 so that this fixed point is stable for : @xmath66 the secular equation associated to the * dp * fixed point @xmath67 is : @xmath68 and thus the region where this phase is stable will be : @xmath69 the secular equation for the * an * fixed point @xmath70 , @xmath46 is : @xmath71 from which it follows that this fixed point is stable if @xmath72 since @xmath73 ( as well as @xmath74 and @xmath27 ) are non - negative , the * an * phase is stable for allowed values of @xmath73 if @xmath75 , so that it exists only when @xmath76 .
this is expected since the chains should be sufficiently stiff to generate nematic order .
for the isotropic polymerized fixed point * p * , where @xmath77 , the elements of the jacobian are @xmath78 and @xmath79 , with : @xmath80 where the squared ratio of ppf s is given by : @xmath81 and thus the stability condition for this fixed point will be @xmath82 [ slp ] we find that this condition is obeyed in the region of the parameter space between the surfaces @xmath83 , @xmath84 , and @xmath85 , which correspond to the stability limits of the * np * , * dp * , and * an * fixed points , respectively .
note that @xmath86 when @xmath87 , so that the * p * phase disappears in this limit . as will be shown below ,
although the order parameter is discontinuous at the * p - an * transition , the two stability limits meet at the transition surface . for @xmath88 , the two critical surfaces * np - p * and *
dp - p * meet at the bicritical line , located at : @xmath89 [ bcl ] before proceeding , we will discuss the possibility of regular polymerized phases with nematic order , with distinct and finite ratios in both directions .
summing and subtracting the fixed point equations ( @xmath90 in eqs .
( [ rrrb ] ) ) we obtain : @xmath91=0 , \\
( r_1-r_2)[(2\omega+z\tau-1)p - z(1 - 2\omega)+1]=0,\end{aligned}\ ] ] where we recall that @xmath60 .
if the phase is polymerized and nematic , the second factors of both equations have to vanish .
this leads to : @xmath92 we notice that the conditions @xmath93 and @xmath94 must be satisfied , and also the three parameters of the model are not independent in the nematic phase with finite ratios , since they are related by the equation : @xmath95 which happens to be equivalent to the stability limit of the * an * phase above , eq .
( [ sldn ] ) .
thus , on this surface of the parameter space , we have a continuous set of marginally stable ( with @xmath96 ) fixed points @xmath97 , which includes the regular polymerized fixed point @xmath98 and the * an * fixed point @xmath50 , @xmath51 .
therefore , we have a discontinuous transition between these two phases , but the transition surface is not between two spinodal surfaces of the two phases which coexist
. incidentally , we notice that the value of the product of ratios in the * an * phase [ eq .
( [ p ] ) ] reduces to eq .
( [ prod ] ) on the stability limit of this phase .
this rather unusual feature of the * an*-*p * transition , which is critical but has a discontinuous order parameter , has been discussed in the literature before .
one simple situation of this kind is the one - dimensional ising model with interactions decaying with the distance between spins as @xmath99 @xcite . at zero field , in this model a discontinuous magnetization is found at the critical point .
the one - dimensional ising model with nearest - neighbor interactions has also been studied via exact renormalization - group transformations by nelson and fisher @xcite , and , although the phase transition there is degenerate , since it happens at zero temperature , it may be interpreted also as a critical discontinuous transition .
the possibility of such transitions in the framework of the renormalization group was discussed in general by fisher and berker @xcite .
this unusual critical behavior was also found in the stationary behavior of non - equilibrium models associated to the ising model and in the threshold contact process @xcite .
the partition function of the model on the whole cayley tree is obtained considering the operation of connecting four subtrees to the central site .
the result is : @xmath100 where the first term corresponds to the configuration with no bond incident on the central site , the next three have two incident bonds and in the last all four edges are occupied .
this expression may also be written as @xmath101 , with @xmath102 the densities of bonds in both directions , and bends ( in non - colliding sites ) , collisions and crossings at the central site are given respectively by : @xmath103 as discussed before , the total density of bends is @xmath104 . in the * np * phase all densities vanish . in the * an * phase with @xmath50 and @xmath51 , @xmath105 and all other densities
vanish . in the * dp * phase , @xmath106 , @xmath107 , @xmath108 , and @xmath109 . finally , in the * p * phase , @xmath110/[2z(1 + 2\omega)-1-\tau z^2]$ ] , @xmath111 , @xmath112 , and @xmath113 ,
where the ratio @xmath114 is given by eq .
( [ rfp ] ) .
the bulk free energy per site on the bl , which differs from the one obtained before in eq .
( [ pfct ] ) by the contribution of the surface of the cayley tree , may be found following an ansatz by gujrati @xcite .
the result is : @xmath115 from the recursion relations ( eqs .
( [ rrgb ] ) ) we have that : @xmath116 where @xmath117 is the partition function calculated with ppf s of subtrees with an additional generation with respect to the unprimed ppf s and the thermodynamic limit is implicit . in the
* np * fixed point , @xmath29 and @xmath31 are dominant over the other terms in the partition function eq .
( [ pfct ] ) , so that we may rewrite eq .
( [ phibi ] ) as : @xmath118 ^ 3}=0 , \label{phinp}\ ] ] where we have used the recursion relations in eqs .
( [ rrgb ] ) .
this result is consistent with the fact that this phase corresponds to an empty lattice . in the * dp * phase , the last term of the partition function dominates over the others ,
so that : @xmath119 , \label{phidp}\end{aligned}\ ] ] where we recall that in this phase four bonds are incident on each site . in the * an * phase , supposing that the bonds are in the 1 direction , the second term of the partition function dominates , so that : @xmath120 and again this confirms that in this phase each site has two bonds in direction 1 incident on it . finally , in the regular polymerized phase *
p * , where @xmath77 , we may rewrite eq .
( [ phibi ] ) as : @xmath121z^2}{1 - 2(1 + 2\omega)z+\tau z^2}.\end{aligned}\ ] ] beyond the critical surfaces ( continuous for the * np - p * transition and * p - dp * and discontinuous for * p - an * transition ) , we note that the * np * and * dp * phases coexist in the region @xmath122 and @xmath123 . the discontinuous * np - dp
* transition is located at the surface where the bulk free energies per site of the two phases are equal and , from eqs .
( [ phinp ] ) and ( [ phidp ] ) , we find it at @xmath124 .
this surface ends at the bicritical line ( eqs .
[ bcl ] ) . as expected , at all other transition surfaces the respective bulk free energies of the involved phases
are also equal . as discussed above , for @xmath125 the * an * phase is not present in the phase diagrams , see an example in fig .
[ pd](a ) for @xmath126 , which is qualitatively similar to the one obtained in the flexible case ( @xmath127 ) @xcite . for @xmath128 ,
the thermodynamic behavior is still the same , except for the presence of the * an * phase , as well as the critical discontinuous * p - an * surface .
diagrams for @xmath129 and @xmath130 are shown in figs .
[ pd](b ) and ( c ) , respectively , where one sees that by decreasing @xmath27 the region occupied by the * p * phase decreases , whilst that by the * an * phase increases . , ( b ) @xmath129 , ( c ) @xmath131 and ( d ) @xmath132 .
regular continuous transitions are shown as solid lines , discontinuous transitions are represented as dashed lines and the dash - dotted line corresponds to the critical discontinuous transitions .
the dot in ( d ) the endpoint of the bicritical line , where the three discontinuous transition lines meet.,title="fig : " ] , ( b ) @xmath129 , ( c ) @xmath131 and ( d ) @xmath132 .
regular continuous transitions are shown as solid lines , discontinuous transitions are represented as dashed lines and the dash - dotted line corresponds to the critical discontinuous transitions .
the dot in ( d ) the endpoint of the bicritical line , where the three discontinuous transition lines meet.,title="fig : " ] , ( b ) @xmath129 , ( c ) @xmath131 and ( d ) @xmath132 .
regular continuous transitions are shown as solid lines , discontinuous transitions are represented as dashed lines and the dash - dotted line corresponds to the critical discontinuous transitions .
the dot in ( d ) the endpoint of the bicritical line , where the three discontinuous transition lines meet.,title="fig : " ] , ( b ) @xmath129 , ( c ) @xmath131 and ( d ) @xmath132 .
regular continuous transitions are shown as solid lines , discontinuous transitions are represented as dashed lines and the dash - dotted line corresponds to the critical discontinuous transitions .
the dot in ( d ) the endpoint of the bicritical line , where the three discontinuous transition lines meet.,title="fig : " ] indeed , in the limit of rigid trails @xmath133 ( @xmath134 ) , only the two dense phases appear , besides the * np * phase , since the * p * phase between them is absent .
the roots of the secular equation related to the * an * phase [ eq .
( [ sedn ] ) ] are @xmath135 and @xmath136 in this case , so that the stability limit of this phase for @xmath93 is @xmath137 , which coincides with the stability limit of the * dp * phase [ eq .
( [ sldp ] ) ] ( since there are no bends , @xmath138 ) . also , the stability limits of the * an * and * np * phases meet at @xmath139 .
as always , the * np * and the * dp * phases coexist on the line @xmath140 .
the three transition lines meet at @xmath141 , as may be seen in eq .
( [ bcl ] ) . as @xmath133 , at the discontinuous * an*-*np * transition line , located at @xmath139 , the * np*-*p * critical surface and the * p*-*an * discontinuous critical surface meet , while the * dp*-*p * critical surface and the * p*-*an * discontinuous critical surface meet at the * dp*-*an * discontinuous transition line , located at @xmath137 .
the phase diagram in this limit is shown in fig . [ pd](d ) . actually , it is quite simple to obtain the free energy of the model on the square lattice in this limit of rods , and the same phase diagram is obtained .
this calculation is presented in the appendix . for @xmath142 and ( a ) @xmath129 , ( b ) @xmath131 and ( c ) @xmath132.,title="fig:",width=272 ] for @xmath142 and ( a ) @xmath129 , ( b ) @xmath131 and ( c ) @xmath132.,title="fig:",width=272 ] for @xmath142 and ( a ) @xmath129 , ( b ) @xmath131 and ( c ) @xmath132.,title="fig:",width=272 ] in this subsection
, we investigate the behavior of the densities at the different transitions .
we start noting that the total density of bonds @xmath143 assumes the values : @xmath144 in the * np * phase , @xmath145 in the * dp * phase , @xmath146 in the * an * phase , and @xmath147 in the * p * phase
. it may be useful to recall that the order parameter of the polymerization transition is actually @xmath148 , as is shown by the mapping of the problem onto the magnetic @xmath149-vector model in the limit @xmath2 @xcite .
moreover , we may define a nematic order parameter as : @xmath150 which is @xmath151 in the * an * phase and @xmath152 otherwise , indicating that any transition to * an * phase is discontinuous .
indeed , this is confirmed in fig .
[ rhoz ] , which shows the variation of @xmath153 with @xmath74 , for @xmath142 and several values of @xmath27 .
close to the * np*-*p * transition we have @xmath154 , consistent with an mean field exponent @xmath155 for the order parameter . along the * p - an * transition surface
, we have @xmath156 for the * p * phase , which increases with @xmath74 , from @xmath157 ( at @xmath158 ) to @xmath159 ( for @xmath160 ) . between these limits
, there exists a line , given by @xmath161 , where @xmath153 is continuous ( i.e. , @xmath162 ) , but @xmath63 ( and so , the transition ) is still discontinuous . still on the * p - an * critical surface , the infinite set of marginally stable solutions @xmath163 have densities @xmath164 and @xmath165 .
thereby , for fixed @xmath27 and @xmath74 , @xmath166 increases ( @xmath167 decreases ) from 0 to 1 ( from 1 to 0 ) when @xmath168 changes from @xmath169 to @xmath170 . in both limits , we recover the * an * result and , when @xmath171 the * p * phase is obtained .
we note that in this phase , @xmath172 for all values of @xmath27 , @xmath74 and @xmath168 , except at the line @xmath161 , where @xmath146 regardless the value of @xmath168 .
let us discuss in some more detail the * an*-*p * transition by studying the behavior of the appropriate susceptibility close to it .
given the activities @xmath173 and @xmath174 of bonds in each direction , we may define @xmath175 and @xmath176 the activity @xmath177 is the appropriate field - like variable conjugated to the nematic order parameter , so that we define the nematic susceptibility as : @xmath178 where the nematic order parameter @xmath63 is defined in eq .
( [ eqnop ] ) . to obtain an expression for @xmath179 in the * p * phase , we start with the fixed point equations which follow from the recursion relations eqs .
( [ rrrb ] ) , remembering that @xmath180 and @xmath181 . differentiating these equations with respect to @xmath177 , we obtain a system of linear equations for the derivatives of the ratios , whose solution is ( for @xmath182 ) : @xmath183 where we have @xmath184(r_1 ^ 2+r_2 ^ 2)+\nonumber \\ & & 8\omega(1-\tau z^2)r_1r_2 + 3[4\omega^2-(1-\tau z)^2]r_1 ^ 2r_2 ^ 2 . \label{eq_diffrzbar}\end{aligned}\ ] ] now we may obtain an expression for the susceptibility as a function of the parameters @xmath74 , @xmath73 , and @xmath27 , as well as of the ratios @xmath168 and @xmath58 and their derivatives with respect to @xmath177 .
the expression is too long to be given here , but if we particularize it to the * p * phase , where @xmath77 , with @xmath114 given by eq .
( [ rfp ] ) it simplifies to : @xmath185[1-(1+\tau-4\omega^2)z+\tau z^2]}. \label{chi}\ ] ] at the * p*-*an * transition @xmath186 , as expected the denominator in eq .
( [ chi ] ) vanishes , and thus we may write this equation as : @xmath187(\tau-\tau_{critical})}. \label{chi1}\ ] ] we thus conclude that , in agreement with the findings of fisher and berker @xcite for discontinuous critical transitions , the transition from the regular polymerized to the nematic phase is characterized by the critical exponents @xmath188 , since the nematic order parameter is discontinuous , and @xmath189 , as is clear in eq .
( [ chi1 ] ) . of course
, the divergence may also be seen if we cross the critical line in another direction , such as parallel to the axis which represents the bond activity @xmath74 , as long as we really cross it and do not touch the critical line tangentially . in fig .
( [ chiws ] ) some of these curves are shown close to the transition activity . in the * p *
phase as a function of the bond activity @xmath74 for @xmath190 , close to the * p*-*an * transition .
the curves , from left to right , correspond to @xmath191 ( blue ) , @xmath192 ( red ) and @xmath129 ( black ) .
dashed lines indicate the corresponding critical activities.,width=264 ]
semi - flexible trails on a bethe lattice with coordination number equal to 4 show a very rich phase diagram in the parameter space defined by the activity of a bond ( @xmath74 ) , the statistical weights of a crossing and a collision ( @xmath11 and @xmath10 ) , and the statistical weight of an elementary bend in the trail ( @xmath27 ) . for sufficiently flexible chains ( @xmath193 ) the phase diagrams are qualitatively similar to the one found in the flexible case ( @xmath127 ) , studied in @xcite , with non - polymerized ( * np * ) , regular polymerized ( * p * ) and dense polymerized ( * dp * ) phases meeting at a bicritical point .
when the boltzmann factor of bends is smaller than 1/2 , an additional polymerized phase appears inside the * p * phase . in this phase
all lattice sites are visited by the trails and all bonds are in one of the two possible directions , thus characterizing it as anisotropic and nematic ( * an * ) .
the nature of the * p*-*an * transition is quite unusual : while the nematic order parameter is discontinuous , it also has a critical nature , characterized by the fact that the stability limits of both phases coincide with the transition line .
this type of criticality was studied in the framework of the renormalization group by fisher and berker @xcite , and for the present case we have verified their result that the susceptibility critical exponent @xmath194 should be equal to 1 . in the limit of rigid chains @xmath195 the * p * phase is no longer stable , and three coexistence lines ( * an*-*np * , * an*-*dp * , and * np*-*dp * ) meet at a triple point which is the endpoint of the bicritical line
. some features of the results presented here may be due to the particular lattice on which the model is solved . on the bethe lattice ,
since there are no closed paths , any collision may be replaced by a crossing and vice - versa , so that the two statistical weights associated to these configurations only appear in the combination @xmath196 ; this should no longer be true on a lattice with closed paths .
also , the phases * dp * and * an * are totally frozen in the bethe lattice solution : all lattice edges are occupied by bonds in the former , the same happening for all edges in one of the two directions in the latter .
this also should change on lattices which are closer to regular ones .
it is thus of interest to study this problem on the husimi lattice , where small loops are present , and we are presently working on this . against @xmath27 for isat and visaw models.,width=264 ]
it is interesting to compare the behavior of the isat model , where we choose @xmath197 and @xmath198 , and the visaw model , for which @xmath13 and @xmath198 . in the former ,
the bicritical line is located at @xmath199 , while for the visaw model it is at @xmath200 .
these behaviors are compared in fig .
[ isatxvisaw ] . in the isat model , as the stiffness of the chains is increased , the collapse transition becomes easier , since @xmath23 decreases as @xmath27 decreases .
for the visaw model , we notice an opposite behavior , with @xmath201 when @xmath87 , so that an infinite ( on - site ) monomer - monomer interaction is needed to collapse the chains .
this is expected , since crossings are forbidden in visaw and , thus , the stiffness will make the collapse more difficult .
moreover , this suggests that the globule phase for visaw is similar to the one in the sasaw model . on the other hand ,
the results for isat s show that its collapsed phase is quite different from the one found in previous models .
again we could expect that such behavior should change on lattices with closed paths , as suggested by the results for the semi - flexible visaw on the square lattice @xcite .
we acknowledge support of the brazilian agencies cnpq and fapemig , particularly from the former through project pve 401228/2014 - 2 , which made tp s stay in brazil possible .
we thank profs .
mrio j. de oliveira and ron dickman for helpful discussions regarding discontinuous critical transitions .
tp acknowledges support by epsrc grant ep / l026708/1 .
jfs thanks the queen mary university of london for hospitality .
the particular case of the model where bends are forbidden ( @xmath195 ) and therefore there are also no collisions ( since @xmath202 ) as well , allows for a simple solution on the square lattice , which is a generalization of the case with no crossings discussed in @xcite .
we may use a transfer matrix calculation to calculate the free energy , although this may also be done using combinatorial arguments .
we start defining the model on a strip of the square lattice in the @xmath203 plane with finite width @xmath204 and length @xmath205 , as shown in fig .
[ strip ] .
boundary conditions are periodic in both directions .
the states of the transfer matrix @xmath209 will be defined by the set of @xmath204 horizontal edges between two successive vertical lines of the lattice .
we therefore have @xmath210 possible configurations , which may be defined by a vector @xmath211 , where @xmath212 ( 1 ) if the edge @xmath213 is empty ( occupied ) .
the grand partition function will be @xmath214 where @xmath215 are the eigenvalues of the transfer matrix . in the thermodynamic limit @xmath216 , the partition function
is dominated by the leading eigenvalue @xmath217 , so that the free energy per site will be : @xmath218 by construction , the transfer matrix is diagonal , and its eigenvalues are @xmath219 where @xmath220 is the number of horizontal rods and @xmath173 and @xmath174 are the activities of horizontal and vertical rods , respectively .
the two contributions in the right hand side of eq .
( [ eqlambatm ] ) correspond to having or not a vertical rod between the two horizontal edge sets , respectively .
the number of states with @xmath221 rods is @xmath222 $ ] .
the functions @xmath223 are convex , so that the maxima should be located at @xmath224 or @xmath225 .
now @xmath226 , and @xmath227 .
the transition between these two phases occurs at @xmath228 , which reduces to @xmath229 in the symmetric case @xmath37 .
thus , even in one dimension ( finite values of @xmath204 ) , a discontinuous phase transition happens between a phase * a * with no horizontal rods ( @xmath230 ) and a density @xmath231 for vertical ones , and a phase * b * where all horizontal edges are occupied by bonds ( @xmath232 ) and the density of vertical bonds is @xmath233 $ ] .
* non - polymerized * np * , where the densities of rods vanish ( @xmath236 ) , if @xmath237 . * anisotropic nematic ( * an * ) , where all horizontal edges and no vertical edge is occupied ( @xmath232 and @xmath238 ) , if @xmath239 ; or all vertical and no horizontal edge is occupied ( @xmath230 and @xmath240 ) , if @xmath241 . * a dense polymerized ( * dp * ) phase , where all edges are occupied ( @xmath232 and @xmath240 ) , if @xmath242 . in the symmetric case , when @xmath37 , the * np * phase is stable for @xmath243 , the * an * phase is stable in the region of the parameter space where @xmath244 and @xmath245 and finally the stability region of the * dp * phase is given by @xmath246 . recalling that * np * and * dp * phases came out from @xmath247 and @xmath248 , respectively , they coexist at @xmath140 .
these results turn out to be identical to the ones for the bethe lattice solution ( in this particular case , with @xmath195 ) , leading to the phase diagram obtained above and depicted in fig .
[ pd](d ) .
b. duplantier and h. saleur , phys .
lett * 59 * , 539 ( 1987 ) ; ibid * 60 * , 1204 ( 1988 ) ; ibid * 62 * , 1368 ( 1989 ) .
see also a. coniglio , n. jan , i. majid , and h. e. stanley , phys .
b * 35 * , 3617 ( 1987 ) .
j. f. stilck , k. d. machado , and p. serra , phys . rev
* 76 * , 2734 ( 1996 ) ; k. d. machado , m. j. de oliveira , and j. f. stilck , phys .
e * 64 * 051810 ( 2001 ) ; p. serra , j. f. stilck , w. l. cavalcanti , and k. d. machado , j. phys .
a * 37 * 8811 ( 2004 ) .
h. w. j. blte and b. nienhuis , j. phys .
a * 22 * , 1415 ( 1989 ) ; m.t .
batchelor , b. nienhuis and s.o .
warnaar , phys .
62 * , 2425 ( 1989 ) ; b. nienhuis , int . j. mod .
b * 4 * , 929 ( 1990 ) ; s.o .
warnaar , m.t .
batchelor and b. nienhuis , j. phys .
a : math . gen . * 25 * , 3077 ( 1992 ) .
m. j. de oliveira j. f. f. mendes , and m. a. santos , j. phys .
a * 26 * , 2317 ( 1993 ) ; m. j. de oliveira , physical review e * 67 * , 066101 ( 2003 ) ; e. f. da silva and m. j. de oliveira , j. phys .
a * 44 * , 135002 ( 2011 ) . | we consider a model of semi - flexible interacting self - avoiding trails ( sisat s ) on a lattice , where the walks are constrained to visit each lattice edge at most once .
such models have been studied as an alternative to the self - attracting self - avoiding walks ( sasaw ) to investigate the collapse transition of polymers , with the attractive interactions being on site , as opposed to nearest - neighbor interactions in sasaw .
the grand - canonical version of the sisat model is solved on a four - coordinated bethe lattice , and four phases appear : non - polymerized ( * np * ) , regular polymerized ( * p * ) , dense polymerized ( * dp * ) and anisotropic nematic ( * an * ) , the last one present in the phase diagram only for sufficiently stiff chains . the last two phases are dense , in the sense that all lattice sites are visited once in * an * phase and twice in * dp * phase . in general ,
critical * np*-*p * and * dp*-*p * transition surfaces meet with a * np*-*dp * coexistence surface at a line of bicritical points .
the region in which the * an * phase is stable is limited by a _ discontinuous critical _ transition to the * p * phase , and we study this somewhat unusual transition in some detail . in the limit of rods , where the chains are totally rigid , the * p * phase is absent and the three coexistence lines ( * np*-*an * , * an*-*dp * , and * np*-*dp * ) meet at a triple point , which is the endpoint of the bicritical line . |
one of the hallmarks of flat - spectrum radio - loud active galactic nuclei ( agns ) is their tendency to display large flux density variations over a wide range of wavelengths .
studies of complete agn samples ( e.g. , @xcite ) have shown that the degree of variability at radio wavelengths is well - correlated with the prominence of a bright , flat - spectrum core component that is thought to harbor the supermassive black hole and accretion disk that power the agn .
parsec - scale images made with vlbi techniques show that this core is usually located at the base of a highly collimated , relativistic outflow .
the extremely high speeds of agn jets are responsible for numerous biases in the observed properties of samples selected on the basis of core flux density , due to special relativistic effects ( e.g. , @xcite ) .
for example , the distribution of source orientations in such samples will be heavily weighted towards jets seen nearly end - on , due to relativistic beaming of radiation in the direction of motion .
these highly beamed sources ( also known as blazars ) will also tend to have more variable light curves than typical radio - loud agns , since any intrinsic flux variations associated with the relativistic outflow will have their timescales shortened and amplitudes boosted in the observer s frame .
previous theoretical studies of observational biases associated with relativistic beaming in agns ( e.g. , @xcite ) have considered only steady - state fluxes , and have not taken into account possible additional biases due to variability . in most flux - limited samples of astronomical objects ,
source variability is not usually a major source of bias , since its effects tend to be statistically averaged out in the sampling process .
the situation is different , however , for samples of objects whose flux densities are relativistically beamed , such as flat - spectrum agns and gamma - ray bursts .
as we will show in this paper , the observed variability duty cycles of these objects ( i.e. , the fraction of time spent above a particular flux level ) can be strongly affected by beaming , and should therefore be considered as a possible source of bias in flux - limited samples . since these types of samples are often used to infer general properties of the parent population , it is important to ascertain the strength of this effect . in the first part of this paper
we develop a simple shot - noise variability model to examine the effects of relativistic beaming on the radio light curves and duty cycles of agns .
we describe the parameters of our model in 2 , and compare our simulations to observed agn light curves in 3 . in
4 we show how highly beamed agns are able to reach very high flux levels , but will in fact be observed to spend most of their time in low flaring states . in the remaining portion of the paper we use monte carlo simulations to model the beaming and variability properties of the blazar parent population , and determine the degree to which core - selected agn samples are additionally biased by source variability . in 5
we show that despite the dramatic range of _ observed _ variability levels in blazars , their overall parent population likely consists of weakly beamed sources with stable flux levels that arise from the superposition of many long - term flares .
we find that as a result , the `` quiescent '' flux levels of these sources ( identified in most unified models as radio galaxies ) are significantly higher than their `` true '' quiescent ( i.e. , non - flaring ) levels .
we further demonstrate how the predicted orientation bias in medium - sized , flat - spectrum agn samples is slightly weakened in the case of a flaring source population .
we summarize our main findings in 6 . throughout this paper
we use a freidmann cosmology with zero cosmological constant , @xmath0 , and a hubble constant @xmath1 measured in units of @xmath2 . over the redshift range of our simulations
, this choice of cosmology gives luminosity distances that are within @xmath3% of those calculated using more contemporary @xmath4 , @xmath5 models ( e.g. , @xcite ) .
the study of variability in blazars at radio wavelengths is hindered by complications arising from relativistic beaming .
the true jet speeds and orientations of individual objects are in most cases poorly known , which makes it difficult to disentangle possible beaming effects from their observed light curves . as a result , there are relatively few observational constraints on the intrinsic amplitudes and timescales of individual flares that are responsible for the majority of flux density variations in radio - loud agns . these considerations have led us to develop a relatively simple stochastic variability model for this study that contains a minimal number of free parameters , yet still reproduces the major variability characteristics of agns at radio wavelengths . in this section
we describe the parameters of this model , which we will use to create simulated light curves for agns having various jet speeds and orientations .
long - term flux monitoring studies at cm- and mm - wavelengths at the u. michigan @xcite and metshovi @xcite observatories have shown that the flux density variations of agns tend to be stochastic ( i.e. , there are very few instances of periodicity ) . rather than being completely random , however , agn light curves display a flare - like behavior whose spectral properties are consistent with a shot - noise process @xcite . one way to reproduce
shot noise is through a superposition of a series of identical impulses , occurring at intervals dictated by poisson statistics . in a poisson process , the overall rate of events is statistically constant , yet the starting times of individual events are independent of all previous ones .
the time intervals between events follow an exponential distribution .
we will use such a process in our agn variability model , by assuming a constant flare rate @xmath6 , such that the probability that a second flare will occur within a time interval @xmath7 after the first one is @xmath8 .
long - term monitoring of bright agns in the mm - wave regime at the metshovi radio observatory has shown that on average , most sources experience one major flare per year ( e. valtaoja , private communication ) . after correcting for the mean redshift of the metshovi monitoring sample ( @xmath9 ; @xcite ) ,
this translates into a source rest frame flare rate of @xmath10 . for simplicity
we will assume this rate for all our simulated light curves .
vlbi monitoring of agn jets on parsec scales has shown that flaring events are often accompanied by the emergence of a bright component from the base of the jet ( e.g. , @xcite ) . for our purposes
we will assume that the source of an agn flare is a luminous blob moving down a relativistic jet at an angle @xmath11 to the line of sight with velocity @xmath12 in units of the speed of light .
the doppler factor of the blob is @xmath13 , and the lorentz factor is @xmath14 .
standard relativistic transformations ( e.g. , @xcite ) can be used to convert quantities in the blob frame to the observer frame . in particular , the intrinsic time scale of the flare is multiplied by a factor of @xmath15 , while its flux density @xmath16 is boosted by a factor @xmath17 , where we have assumed a spectral index @xmath18 . for convenience , we will parameterize both the amplitude of the flare and the total flux density of the source in units of the observed quiescent ( non - flaring ) flux density @xmath19 .
we will assume that the latter has spectral index of zero and arises in a continuous jet having the same doppler factor as the blob .
its flux density is therefore boosted by a factor @xmath20 ( the flare emission has a larger doppler boost due to its finite lifetime see , e.g. , appendix b of @xcite ) .
the observed source flux density at time @xmath21 since the start of the flare is then the sum of the quiescent and flaring components : @xmath22 , where @xmath23 represents a flare profile that ranges from 0 to @xmath24 times the quiescent flux density in the source frame .
@xcite have found that most major agn flares in the mm - wave regime have profiles of the form @xmath25 for the rise and decay portions of the flare event .
fitting to more recent data by @xcite has shown that the decay time is typically a factor of @xmath26 times longer than the rise time . in figure [ flareshape ]
we display this profile , which has the form : @xmath27 , & \mbox{$t \in [ 0,\tau/2.3]$}\\ { p\over(k-1)}\left[k^{{2.3\over1.3}(1-t/\tau)}-1\right ] , & \mbox{$t \in [ \tau/2.3 , \tau]$ } \\ 0 , & \mbox{$t > \tau$ } , \end{array } \right.\ ] ] where @xmath7 is the time scale of the flare and @xmath28 is a constant .
one aspect of this profile is that for low values of @xmath28 , there are large exponential `` wings '' where the flux density remains quite low compared to the peak . in order to make @xmath7 more representative of the true time scale of the flare , we minimize these wings by arbitrarily setting @xmath29 .
it is difficult to obtain good observational constraints on the range of intrinsic timescales and amplitudes of agn flares at radio wavelengths , since isolated flares are exceedingly rare , and the beaming factors are usually not well known .
the majority of blazar light curves are composed of many different flares having a variety of timescales , so that often what appears to be a large isolated flare is actually a superposition of several smaller events . given these uncertainties , we let the flare lengths in our model be uniformly distributed over a range [ @xmath30 , @xmath31 in the rest frame of the jet . since our estimate of the flare rate ( @xmath32 ) is based on bi - weekly monitoring data , and does not account for any flares with observed timescales @xmath33 month we set @xmath34 y. we derive a lower limit on @xmath35 by noting that there are no sources in either of the two major agn monitoring campaigns ( u. of michigan or metshovi ) that show a tendency to return to a quiescent baseline level in between flaring events .
this implies that flares are of sufficient duration in the source frame so that agns are rarely observed in a non - flaring state . in our shot noise model , the probability of this occurring is approximately @xmath36 where @xmath37 is the mean flare time scale in the source frame .
we can estimate an upper limit on @xmath38 for the agn parent population based on measurements of superluminal motion in samples of highly beamed agns ( i.e. , blazars ) .
@xcite find component velocities in the jets of gamma - ray loud blazars that range up to @xmath39 .
the relevant constraints from superluminal motion equations ( @xmath40 and @xmath41 ; see @xcite ) imply a maximum doppler factor of @xmath42 , assuming @xmath1 . by choosing a reasonably low probability for being in a non - flaring state ( @xmath43 ) and substituting @xmath44 , @xmath45 and @xmath46 in eq .
[ fractime ] , we obtain @xmath47 y. although this value is only a lower limit , we will show in 4 that higher values of @xmath35 will simply increase the mean steady - state flux level of the overall parent population , and will not affect our main conclusions .
given the observational uncertainties regarding the intrinsic flare amplitudes of agns , we will make the simple assumption that all flare amplitudes are the same fraction @xmath24 of a source s rest frame quiescent flux level .
although this is likely a gross oversimplification in the case of real agns , our study is not concerned with low - amplitude flares , since they are not likely to affect the overall selection statistics of flux limited samples .
furthermore , we will show in
3 that a broad intrinsic flare amplitude distribution is not needed to reproduce the basic properties of agn light curves , since the effects of flare superposition and beaming can conspire to create a large range of observed flare amplitudes .
we constrain our flare amplitude parameter @xmath24 by examining the range of observed flux density in the long term ( @xmath48 yr ) light curves in the u. michigan agn sample . of the 69 agn light curves analyzed by @xcite , the highest absolute peak - to - minimum flux ratio of any source at 4.8 ghz was @xmath49 for the bl lac object 0235 + 164 . through repeated simulations of twenty - year light curves with various values of @xmath24
, we find that for @xmath50 , the probability of finding a peak - to - minimum ratio exceeding 30 in a highly beamed source ( @xmath44 ) is @xmath51 .
we therefore adopt @xmath50 in the simulations that follow .
having established the general parameters of our model , we are now able to compare our simulated light curves to those of actual agns . in figure [ data_lc ]
we show the 20 year light curves of three agns at 4.8 ghz from the university of michigan agn monitoring program . in each panel
the flux densities have been divided by the mean for comparison purposes .
the top , middle and bottom panels show the light curves of the radio galaxy 3c 380 , the quasar 3c 345 , and the bl lac object 0235 + 164 , respectively . in order to make comparisons with our simulated light curves , estimates of the doppler factors of these sources
are needed .
it is possible to use the measured apparent speed in the inner jet of 3c 380 ( @xmath52 ; @xcite ) to obtain a crude estimate of its doppler factor if we assume that the jet is viewed at an angle which maximizes its apparent velocity .
this gives @xmath53 .
the doppler factor of 3c 345 has been well - constrained by @xcite , who modeled the three - dimensional motion of two components in its inner jet .
they were able to constrain the viewing angle to @xmath54 and the lorentz factor of the jet to be @xmath55 , which implies @xmath56 .
the bl lac object 0235 + 164 is the one of the most variable blazars in the u. michigan monitoring sample , and is therefore likely to have a doppler factor at the high end of the possible range ( @xmath57 ; see [ amptime ] ) .
this source has also been constrained by @xcite to have @xmath58 based on equipartition arguments and the ratio of observed inverse - compton to synchrotron flux densities . in figure [ sim_lc ]
we show three simulated light curves with doppler factors and redshifts corresponding to the three agns in figure [ data_lc ] .
these light curves were calculated after allowing the sources to achieve a steady - state flaring level .
the simulated flux densities have been divided by the mean in each case .
we note that the source redshift will affect the timescales but not the amplitudes of our simulated light curves , since the flux densities are measured in units of the observed quiescent level .
our simulated curves reproduce the general variability characteristics of all three sources fairly well , even with our simplistic assumption of equal flare amplitudes .
a further comparison with the data can be made by examining the first order structure function of our simulated light curves .
this function is defined as @xmath59 ^ 2\right>$ ] , where @xmath60 is the flux density at time @xmath21 and @xmath7 is the time lag .
structure functions are commonly used as a means of characterizing the distribution of power in a stochastic process on different time scales ( e.g. , @xcite ) .
the slope @xmath61 is a general indicator of the type of stochastic process , with @xmath62 and @xmath63 corresponding to flicker and shot noise , respectively . as pointed out by @xcite , a slope of unity at long time lags
is indicative of a shot - noise process of the type we are simulating .
these authors found that the majority of the agns in the u. michigan monitoring survey have structure function slopes that cluster near unity , with values ranging from approximately 0.7 to 1.65 . in figure [ sf ]
we show the structure function for the simulated light curve in the middle panel of figure [ sim_lc ] with @xmath64 and @xmath65 , corresponding to the properties of 3c 345 .
we find good agreement in the structure function slope of our model ( @xmath66 ) and that measured by @xcite at 4.8 ghz .
the slope of our simulated structure function for 3c 345 is slightly steeper than the value of unity expected for a pure shot noise process , due to the doppler contraction of flaring timescales in the observer s frame , and the fact that we are modeling only major flares , which have a relatively low rate of occurence .
the light curves of other agns may include longer - term , low amplitude flares that serve to flatten the slope of the structure function .
we do not include these flares in our model , as they simply add a slowly varying component to the quiescent level , and do not have a major impact on the duty cycle of the source .
an important quantity in the statistics of flaring source samples is the variability duty cycle , which we define to be the fraction of time the source spends above a particular flux density level . in this section
we derive numerical estimates of agn duty cycle functions by simulating light curves for sources with a wide range of doppler factors and redshifts .
a list of our individual model parameters is given in table [ parameters ] .
for each set of parameters we simulate 500 twenty - year light curves , and generate a cumulative histogram that gives the fraction of time spent above a particular flux density @xmath67 , measured in units of the observed quiescent flux density ( in our case , @xmath68 ) .
we plot the simulated duty cycles for sources at @xmath69 in figure [ dutycycle1 ] .
it is apparent that relativistic beaming can have a large effect on the observed duty cycles of agns , especially for sources with @xmath70 . as expected , these highly beamed jets tend to have a much wider range of observed variability amplitude than the more weakly beamed sources .
the cosmological redshifts of agns will stretch their intrinsic flux variations in time by a factor of @xmath71 , and should therefore have an impact on their observed duty cycles as well . in figure [ dutycycle3 ]
we show the duty cycle functions of two simulated sources , one with @xmath72 ( upper panel ) , and the other with @xmath44 ( lower panel ) .
the three curves in each panel represent the duty cycle of the source as it would appear at redshifts of 0 , 1 , and 3 .
for sources located at high redshifts , cosmological time dilation increases the probability that flares will overlap in the observer frame .
this tends to increase the median flux density level of the source and shift the duty cycle curves to the right in figure [ dutycycle3 ] .
a similar effect occurs when the maximum flare time scale is increased ( figure [ dutycycle2 ] ) , since this will also increase the likelihood of overlapping flares .
the net effect of raising @xmath35 is to boost the overall flux density levels of both strongly and weakly beamed sources alike .
although this will have implications on the intrinsic luminosity function of the parent population , it will not affect the statistics of flux - limited samples .
our duty cycle simulations reveal a somewhat paradoxical property of agn variability : although the flux densities of highly beamed sources can reach very high levels due to boosting , their flaring timescales are severely contracted , so that these sources are observed to spend most of their time in relatively low flaring states .
consequently , it is the _ weakly beamed _ sources that are more likely to be found in a state that is much higher than their quiescent level , due to the overlap of many long term flares .
as we will show in the following section , this tends to slightly weaken the predicted effects of doppler bias in flaring samples selected on the basis of relativistically beamed flux .
the statistics of flux - limited samples should not be strongly influenced by flux variability as long as the variability properties of the parent population are relatively uniform . however
, if some objects have a markedly different variability duty cycle than the rest of the population , this can affect the likelihood that they will appear in a flux - limited sample , thereby creating a selection bias . in this section
we investigate possible variability biases in flat - spectrum agn samples by performing monte carlo simulations of both flaring and non - flaring beamed jet populations .
we compare the observed properties of flux - limited samples drawn from these parent populations to determine to what degree core - selected agn samples are additionally biased by source variability . in a previous study ( @xcite )
, we successfully modeled the observed radio properties of a complete core - selected agn sample ( the caltech - jodrell flat - spectrum survey ; cj - f ) using a simulated population of two - sided , relativistically beamed jets .
we found that the apparent jet velocity distribution of the cj - f was best fit by a parent population of jets with a power - law lorentz factor distribution that was weighted toward low speeds .
this also appears to be the case for several other flat - spectrum agn samples ( e.g. , @xcite ) whose apparent velocity distributions are similar to that of the cj - f . for the purposes of this study we adopt the best - fit model of @xcite for the cj - f , which has no dependence of intrinsic luminosity on jet speed and @xmath73 , with @xmath74 .
the jets have random spatial orientations and are distributed with a constant co - moving space density out to @xmath75 .
their unbeamed 5 ghz luminosity function follows a fit to that of powerful ( fr - ii ) radio galaxies , which incorporates pure exponential luminosity evolution @xcite .
since few constraints exist on the size of the cj - f parent population , the lower cutoff of the parent luminosity function ( @xmath76 ) is an important parameter .
we can obtain an estimate of @xmath76 by examining the properties of the radio galaxy ngc 3894 , which has the lowest 5 ghz luminosity of any source in the cj - f sample .
its structure consists of a two - sided jet on parsec - scales , which has been monitored extensively by @xcite . by using the apparent expansion and flux density ratio of the jet and counter - jet
, these authors found the viewing angle and jet speed to be @xmath77 and @xmath78 , respectively . assuming equal intrinsic luminosities and spectral index of zero for the jet ( @xmath79 ) and counter - jet ( @xmath80 )
, this implies that the radio luminosity of ngc 3894 is boosted by a factor of @xmath81 , where we have assumed @xmath82 , and @xmath83 and @xmath84 .
given that the observed 5 ghz luminosity of ngc 3894 is @xmath85 @xcite , we will adopt a lower luminosity function cut - off of @xmath86 for our simulated non - flaring parent population . in our monte carlo procedure
we iteratively generate jets according to the above parameters until we obtain a sample of 293 sources with @xmath87 jy .
these values correspond to the sample size and flux density cutoff of the cj - f .
approximately @xmath88 parent objects are needed in a typical monte carlo run . in @xcite
we required a substantially larger parent population ( @xmath89 objects ) to fit the cj - f , since we used a smaller value of @xmath76 and a jet lorentz factor distribution that only extended up to @xmath90 .
we show the viewing angle and doppler factor distributions for a single monte carlo run in figure [ distribs ] . as expected , the simulated cj - f sample is heavily biased towards sources with low viewing angles since these tend to have the highest amount of doppler boosting .
the predicted doppler factors of the cj - f sources range up to the maximum possible value of @xmath91 . in order to simulate a flaring jet population , only slight modifications to our monte carlo model are necessary .
we first assume that all of the parent objects have the same intrinsic variability characteristics given by the model parameters in table [ parameters ] .
although this is not likely true of the actual agn population , this simplification makes it easier to isolate the effects of beaming on the statistics of flaring samples .
furthermore , the fact that radio variability is well - correlated with several other statistical beaming indicators ( e.g. , @xcite ) suggests that the observed variability characteristics of agns are more dependent on beaming than on intrinsic factors . as in our previous monte carlo simulations
, we simulate quiescent flux levels for the parent objects , but we now also assign a flaring level to each source based on its duty cycle function . we accomplish this by first binning the source into one of four redshift bins at @xmath92 = 0 , 0.5 , 1 , 2 , or 3 , and then further binning it into one of thirteen sub - bins based on its doppler factor .
we then randomly generate a flare level according to the duty cycle ( i.e. , flaring probability ) curve for the appropriate redshift / doppler factor bin .
the flux density of the source is then simply the quiescent flux density times the flaring level .
if this flux density is fainter than the cj - f cutoff , the source is discarded .
we continue generating simulated sources in this manner until we obtain the necessary 293 sources with @xmath93 jy .
we note that it is necessary in these flaring simulations to modify the lower cutoff of the parent luminosity function , since we showed in 4 that weakly beamed flaring sources are likely to be observed at levels considerably higher than their quiescent ( non - flaring ) level . recalling that we used the least luminous source in the cj - f ( ngc 3894 ) to estimate @xmath76 in the non - flaring case
, we estimate from our duty - cycle curves that its typical flaring level is approximately 7.4 times its quiescent level , given its redshift and doppler factor .
we therefore decrease @xmath76 by this factor and adopt a lower cutoff of @xmath94 for the quiescent luminosities of the simulated flaring population . in figure [ cumhist1 ]
we show cumulative histograms of the predicted doppler factor distribution for the cj - f sample in the case of flaring ( dashed line ) and non - flaring ( solid line ) populations .
the differences in the two cases are rather small , with the flaring sample being slightly more biased towards lower doppler factors .
there is also slightly less orientation bias present in the flaring sample ( not shown ) , with relatively more sources found at higher viewing angles .
we have verified these results using multiple monte carlo runs that all display the same effect .
these findings are independent of our flaring amplitude parameter @xmath24 , since increasing this parameter merely increases the overall flux density level of the entire parent population .
this will not affect the selection statistics since increasing @xmath24 also requires that @xmath76 be adjusted downward accordingly .
although we have indeed found a bias in the statistics of flaring agn samples , its predicted effects are rather small in medium - sized surveys such as the cj - f .
the main reason for this lies with the lorentz factor distribution of the agn population , which is weighted toward slow jets .
the majority of the cj - f parent objects are low - doppler factor sources with jet axes inclined at @xmath95 from the line of sight . as we showed in
4 , the duty - cycle functions of these weakly - beamed objects are very similar , and are roughly symmetric about the mean steady - state level ( fig .
[ dutycycle1 ] ) . a sample composed predominantly of weakly - beamed flaring sources
will therefore be statistically similar to a non - flaring sample , since the effects of variability will be averaged out .
another important factor that reduces the size of the variability bias is the limited range of flux levels seen in individual agns at radio wavelengths ( maximum peak / trough ratio @xmath49 ) . by comparison ,
the entire parent population likely spans a flux density range of at least ten orders of magnitude , given the wide range of intrinsic luminosities , cosmological distances , and beaming factors that are present .
variability will therefore only influence the selection statistics of a relatively small fraction of the parent population that have fluxes near the survey limit .
in large core - selected samples having a faint flux cutoff , these sources will tend to be weakly beamed , due to the predicted decrease in mean doppler factor with flux density ( see @xcite ) .
as a result , the effects of variability will be statistically averaged out as described above .
those sources whose duty cycles are affected by beaming ( with @xmath70 ) will tend to have beamed flux densities that are well above the survey limit , and will always be selected regardless of their flaring level .
we have verified these effects by performing simulations using the same parent population as the cj - f , only with a survey cutoff ten times fainter .
we find virtually no statistical differences in the flaring and non - flaring samples , which contained approximately @xmath96 sources .
we have developed a simple shot - noise variability model to examine how relativistic beaming affects the radio light curves and selection statistics of flat - spectrum agns .
we summarize our main findings as follows : \1 .
relativistic beaming will preferentially boost the amplitude of a flaring event with respect to an agn s quiescent flux level in the observer frame , and will also shorten its apparent time scale . as a result , highly beamed agns ( i.e. , with doppler factors @xmath97 ) are able to reach very high flux levels , but will in fact be observed to spend most of their time in relatively low flaring states
the fact that the most highly beamed blazars are not observed to return to a quiescent level between flares suggests that the intrinsic timescales of individual flares are rather long , given the large lorentz factors ( @xmath98 ) of these jets that are inferred from superluminal motion studies .
we find that the flare amplitudes must be a small fraction ( @xmath99 ) of the quiescent flux level in the source frame in order to reproduce the relatively small peak - to - trough flux density ratios seen in the most highly variable agns .
the unbeamed counterparts of blazars should therefore have very stable radio light curves that are made up of a steady - state superposition of many long - term flares .
the apparent `` quiescent '' flux density levels of these sources ( identified in many unified models as radio galaxies ) will be many times greater than their `` true '' quiescent ( i.e. , non - flaring ) levels .
we have used the monte carlo beaming model of @xcite to model the parent population of the caltech - jodrell flat - spectrum survey ( cj - f ) using both flaring and non - flaring jets .
we find that the standard orientation bias toward highly beamed jets with small viewing angles is predicted in both cases . in the case of a flaring population , however , the amount of orientation bias is slightly reduced due to the fact that the highly - beamed sources have a higher probability of being found in low flaring states .
the magnitude of this effect is rather small in moderately - sized samples such as the cj - f ( @xmath100 sources ) due to the fact that the majority of the highly beamed sources in the parent population lie well above the survey flux limit and will be selected regardless of their flaring levels .
we find that for larger samples with fainter cutoffs , any selection biases associated with variability should be negligible .
the author thanks a. p. marscher and the referee , p. a. hughes for useful comments .
this research has made use of data from the university of michigan radio astronomy observatory , which is supported by the national science foundation and by funds from the university of michigan .
kellermann , k. i. , vermeulen , r. c. , zensus , j. a. , cohen , m. h. 2000 , in astrophysical phenomena revealed by space vlbi , eds .
h. hirabayashi , p. g. edwards , & d. w. murphy ( sagamihara : institute of space & astronautical science ) , 159 lll @xmath6 & flaring rate & 2 @xmath101 + @xmath30 & minimum flare time scale & 1 month + @xmath35 & maximum flare time scale & 620 y + @xmath24 & peak - to - quiescent flux ratio of flare & 1.05 + @xmath38 & doppler factor & 1 , 2 , 4 , 6 , 10 , 20 , 30 , 40 , 60 , 70 , 80 , 90 + @xmath92 & redshift & 0 , 0.5 , 1 , 2 , 3 + | we discuss the impact of special relativistic effects on the observed light curves and variability duty cycles of radio - loud active galactic nuclei ( agns ) .
we model the properties of agn light curves at radio wavelengths using a simulated shot noise process in which the occurrence of major flaring events in a relativistic jet is governed by poisson statistics .
we show that flaring sources whose radiation is highly beamed toward us are able to reach very high flux levels , but will in fact spend most of their time in relatively low flaring states .
this is primarily due to relativistic doppler contraction of flaring time scales in the observer frame .
the fact that highly beamed agns are not observed to return to a steady - state quiescent level between flares implies that their weakly beamed counterparts should have highly stable flux densities that result from a superposition of many long - term , low - amplitude flares .
the `` apparent '' quiescent flux levels of these weakly beamed agns ( identified in many unified models as radio galaxies ) will be significantly higher than their true quiescent ( i.e. , non - flaring ) flux levels .
we have also performed monte carlo simulations to examine how relativistic beaming and source variability bias the selection statistics of flat - spectrum agn samples .
we find that in the case of the caltech - jodrell flat - spectrum survey ( cj - f ) , the predicted orientation bias towards jets seen end - on is weakened if the parent population is variable , since the highly beamed sources have a stronger tendency to be found in low flaring states .
this effect is small , however , due to the fact that highly beamed sources are relatively rare , and in most cases their flux densities will be boosted sufficiently above the survey limit such that they will be selected regardless of their flaring level .
we find that for larger flat - spectrum agn surveys with fainter flux density cutoffs , variability should not be an appreciable source of selection bias . |
in the last few years , nuclear reactions involving weakly bound nuclei became a subject of interest due to the observation of flux enhancement for processes like nucleon transfer and breakup . through the study of these processes @xcite , it is possible to obtain information about nuclear structure , such as single - particle states and nuclear cluster structure , as well as information about the influence of continuum states in the nuclear reaction dynamics @xcite . additionally , the investigation on how these properties change from the stability line to regions far from the stability valley can also be addressed .
in this context , the elastic scattering measurement and coupled channel analysis @xcite are very important tools to investigate nucleon transfer and breakup , as they appear as competing mechanisms in the reproduction of the measured angular distributions . in this work , elastic scattering cross section measurements were performed for the @xmath3be+@xmath1c reaction to study anomalies @xcite in the extracted optical parameters values and the contribution of inelastic channels , transfer and compound nucleus formation in the elastic scattering process .
the experimental data show an enhancement in the elastic cross sections at intermediate and backward angles . this behavior is typically observed in systems where projectile and target present the same core structure @xcite .
this effect can be understood in terms of a @xmath4he transfer process , assuming that @xmath1c has a @xmath4he+@xmath0be cluster structure .
the elastic scattering angular distributions were analyzed in a four steps procedure .
distinctively of previous works , our experimental elastic scattering data have been compared to optical model predictions using a empirical double folding potential @xcite . in this process
, the normalization parameters for the real and imaginary potentials were adjusted to describe the data at forward angles ( @xmath5@xmath680@xmath7 ) .
the normalization of the real part of the potential shows a decrease as a function of the bombarding energy , and the normalization of the imaginary part of the potential is approximately constant . during the following two steps
, we investigate the importance of the coupling to inelastic and transfer channels respectively , and finally in the fourth step we analyse the compound elastic contribution .
the experiment was performed at the university of so paulo physics institute .
the @xmath3be beam was delivered by the 8ud pelletron accelerator with energies e@xmath2=13.0 , 14.5 , 17.3 , 19.0 and 21.0 mev ( e@xmath8=7.4 , 8.3 , 9.9 , 10.8 and 12.0 mev respectively ) and hit a 40 @xmath9g/@xmath10 thick @xmath1c target .
the charged particles produced in the @xmath3be+@xmath1c reaction were detected by means of 13 triple telescopes @xcite separated by @xmath11 = 10@xmath7 in the reaction plane , which covered the angular range from @xmath12 = 10@xmath7 to @xmath12 = 140@xmath7 .
the triple telescopes were composed of an ionization chamber , filled with 20 _ torr _ of isobutane gas , followed by a 150 @xmath9 m silicon detector and a 40 mm csi scintillator crystal .
the identification of the elastic events was done by means of two - dimensional spectra like the one shown in fig .
[ biparametrico]-a .
the elastic yields were obtained by projecting the z = 4 region on the e@xmath13 axis ( energy measured by the silicon detector ) and identifying the elastic processes as shown in fig .
[ biparametrico]-b .
as ilustraded in this figure , we can see a considerable number of events due to target contamination . in order to subtract the contributions in the energy spectra from this target contamination
, we have assumed rutherford scattering for the involved cross sections .
this is a good approximation in our experimental conditions .
the uncertainties in the differential cross section were estimated considering the statistical uncertainty in the yield and the systematic uncertainty of 5% in the target thickness .
in the present work , the experimental data were analyzed using the fresco code @xcite in the framework of the so paulo potential ( spp ) @xcite due to the recent success of this approach in describing nuclear reactions involving weakly bound nuclei @xcite .
clearly , there are other options for the scattering potential , as shown in recent works by a. t. rudchick _ et .
_ @xcite and j. carter _ et .
@xcite who have used a woods - saxon shape and a double folding potential , respectively , and the effects of coupled channels to describe the @xmath3be+@xmath1c scattering data . for the so paulo potential ,
the radial dependence is a folding potential and the energy dependence takes non - local effects into account , in the form @xmath14 where @xmath15=v / c , @xmath16 is the local relative velocity between the two nuclei and @xmath17 is a folding potential obtained by using the matter distributions of the nuclei involved . in detail
, the folding potential depends on the matter densities in the form @xmath18 where @xmath19 is a physical nucleon - nucleon interaction given by @xmath20 with @xmath21 mevfm@xmath4 and the usage of a delta function corresponds to the zero range approach .
extensive systematics were performed in ref .
@xcite , to provide a good description of matter and charge distribution . in the present work
, we adopt the matter and charge diffuseness @xmath23 fm and @xmath24 fm respectively , and the matter and charge distribution radius given by @xmath25 and @xmath26 respectively . in this first step of the analysis
, we considered the real and imaginary potential normalizations as free parameters that are adjusted in order to describe the forward angles of the angular distribution ( @xmath5 @xmath6 80@xmath7 ) . in this case the nuclear potential is given by the equation @xmath27 where the normalization coefficients @xmath28 and @xmath29 are reaction energy dependent .
the results for the first step are shown as solid lines in fig .
[ elastico ] .
we observe a good agreement for the forward angular region .
however , a pronounced disagreement at backward angles for 17.3 , 19.0 and 21.0 mev , is an indication of the importance of other reaction mechanisms , not taken into account in the optical potential . for 13.0 and 14.5 mev
, the optical model description is reasonable . finally , fig .
[ elastico ] also shows an angular distribution for e@xmath2=19.0 mev extracted from ref .
these data are in good agreement with our data and with the optical model calculation at forward angles .
the @xmath28 and @xmath29 energy dependence is depicted in fig .
the uncertainties are determined by a @xmath30 analysis .
one can notice that n@xmath31 increases in the vicinity of the coulomb barrier energy ( e@xmath8 @xmath32 9 mev @xcite ) .
the values of n@xmath33 are approximately constant for all energies , and they are in agreement with the results obtained in ref .
@xcite .
the increase of the n@xmath31 parameter near the coulomb barrier suggests the presence of a threshold anomaly @xcite .
however , no strong statement about this anomaly can be made due to the constant behavior of n@xmath33 .
the results from the first step of the analysis were used to perform the coupling of the inelastic and transfer channels presented in the next sections .
be + @xmath1c system at e@xmath2 = 13.0 , 14.5 , 17.3 , 19.0 and 21.0 mev .
the solid lines correspond to optical model fits , the dashed and dash dotted lines represent the same optical potential including the inelastic and elastic transfer mechanisms respectively ( color online ) . ] and n@xmath33 as a function of the bombarding energy obtained from fits using the so paulo potential ( equation [ sppotential ] ) for the @xmath0be + @xmath1c system .
the lines are just to guide the eyes ( color online ) . ] with the optical potentials previously obtained , we are able to take into account the effects of other channels . the second step of the analysis consisted in calculating the effects of inelastic channels in the theoretical elastic cross section , by considering 5/2@xmath34 and 7/2@xmath34 states of @xmath35be and 2@xmath36 of @xmath1c . in our study the transitions to excited states of @xmath35be and @xmath1c
were calculated using the rotational model approach , where the coupling interaction v@xmath37(*r * ) of the multipole @xmath38 is @xmath39 the coulomb excitations are included by considering deformations on the charge distributions in the form @xmath40 where @xmath41 therefore , to take into account the effects of inelastic process in the theoretical elastic cross section , we take from the literature @xcite the nuclear deformation parameters @xmath42 and the reduced transition probabilities b(e@xmath38:j@xmath43@xmath44j@xmath45 ) , in order to calculate the deformations on the symmetrical central potential .
the parameters for each channel included in the calculations and the reduced transition probabilities are presented in table [ tabela1 ] .
p2 cm p2 cm p2 cm p2 cm p2 cm nucleus & transition & @xmath38 & b(e@xmath38 ) & @xmath42 + @xmath3be & @xmath46 @xmath44 @xmath47 & 2 & [email protected] & 2.4 + + @xmath3be & @xmath46 @xmath44 @xmath49 & 2 & 33@xmath48 1 & 2.4 + + @xmath1c & @xmath50 @xmath44 @xmath51 & 2 & 42@xmath48 1 & 1.52 + + + [ tabela1 ] the results are shown as dashed lines in fig .
[ elastico ] .
in general , we observed that the inclusion of these channels decreases the theoretical cross section when compared to the results obtained only with the optical potential . comparing with the experimental data
, we observed that the curves show a good description at forward angles . for intermediate and backward angles ,
the theoretical prediction underestimates the cross section .
this demonstrates that the inclusion of these inelastic channels is not sufficient to explain the experimental results at intermediate and backward angles . in order to improve the description of experimental elastic distributions at intermediate and backward angular region , a coupling to
the @xmath52he transfer was included in the third step of the analysis .
p3 cm p1.75 cm p1.5 cm p1.5 cm systems & v@xmath53(mev ) & r@xmath53(fm ) & a@xmath53(fm ) + + @xmath3be+@xmath0be@xmath54 & 189.3 & 1.0 & 0.63 + + @xmath4he+@xmath3be & 55.8@xmath55 & 1.35 & 0.65 + + + p3 cm p1.75 cm p1.5 cm p1.5 cm a = c+v & j@xmath56 & @xmath57 & a + + @xmath1c=@xmath0be @xmath58 @xmath52he & 0@xmath59 & 2p@xmath60@xmath61 & 1.224@xmath61 + + [ parametrospotencialoptico ] + .optical parameters for dwba calculations and spectroscopic factor for a a = c+v system .
[ cols= " < " , ] + the calculation was performed using the hauser - feshbach statis code @xcite .
the nuclear level density has been described by means of a level density expression given by lang @xcite , and the transmission coefficients were determined by an internal fermi parametrization @xcite .
the levels considered in this calculations are listed on table [ tabela3 ] and are quite similar to those used in the ref .
@xcite .
= 19.0 and 21.0 mev .
the calculated points , were normalized to agree with behavior of data from @xcite ( color online ) . ]
finally the angular distributions are normalized in such a way that the total cross section agree with the results presented in the excitation function from ref .
@xcite ( figure [ secaototalcompound ] ) .
when the obtained angular distributions are incoherently added to the @xmath52he transfer results , we can see that the contribution of this reaction mechanisms is not relevant , even for our highest energies , as shown in fig .
[ total ] for the two highest energies e@xmath2 = 19.0 and 21.0 mev .
be + @xmath1c system at e@xmath2 = 19.0 and 21.0 mev .
the dot lines are the ce angular distributions from hf calculations .
we can notice that the contribution of ce to differential cross section enhancement at backward angles is not significante ( color online ) . ]
in this work we measured elastic scattering angular distributions for the @xmath35be+@xmath1c light system at bombarding energies ranging from 13.0 mev to 21.0 mev . the double folding so paulo potential was used in the analysis that was performed in four steps . in the first one we considered n@xmath53 and n@xmath43 as free parameters for the fits to the angular distributions at the forward angular region .
the angular distributions calculated with the optical model have shown a good agreement with the 13.0 and 14.5 mev experimental data . for 17.3 , 19.0 and 21.0 mev
the agreement is reasonable at forward angles .
however , the description at backward angles is not good , suggesting that the coupling to other mechanisms is important . in the second step of the analysis , no evidence of the coupling to inelastic channels
was observed . using the spectroscopic factors extracted from the literature for @xmath1c=@xmath4he@xmath58 @xmath3be , in the third step we took into account the @xmath4he elastic transfer channel . for the @xmath4he transfer
we see a pronounced improvement in the description of the data for 17.3 , 19.0 and 21.0 mev , which suggests that this process is important at intermediate and backward angles .
finally in the fourth step , corresponding to the compound elastic calculation , the results suggest that this mechanism is not an important process at the energies studied in this work . finally , to obtain information about the effects of the elastic transfer process and check the values of spectroscopic factors
, it would be important to perform a more carefully analysis and measurements of the elastic scattering angular distributions at more backward angles . | in this work , angular distribution measurements for the elastic channel were performed for the @xmath0be+@xmath1c reaction at the energies e@xmath2=13.0 , 14.5 , 17.3 , 19.0 and 21.0 mev , near the coulomb barrier .
the data have been analyzed in the framework of the double folding so paulo potential .
the experimental elastic scattering angular distributions were well described by the optical potential at forward angles for all measured energies . however , for the three highest energies , an enhancement was observed for intermediate and backward angles .
this can be explained by the elastic transfer mechanism .
@xmath0be+@xmath1c , elastic scattering , so paulo potential 24.10 |
many porous media related practical problems lead to the numerical approximation of the pressure equation @xmath0 where @xmath1 is the part of the boundary of the domain @xmath2 ( denoted by @xmath3 ) where the dirichlet boundary condition is imposed . in case
the measure of @xmath1 ( denoted by @xmath4 ) is zero , we assume the compatibility condition @xmath5 . on the above equation
we have assumed without loss of generality homogenous boundary conditions since we can always reduce the problem to that case .
the domain @xmath2 is assumed to be a convex polygonal region .
we note however that this convexity is not required for the discretization , it is required only when regularity theory of partial differential equations ( pdes ) is considered for establishing the a priori error estimates .
in multi - phase immiscible incompressible flow , @xmath6 and @xmath7 are the unknown pressure and the given phase mobitity of one of the phases in consideration ( water , oil or gas ) ; ( see e.g. , @xcite ) . in general
, the forcing term @xmath8 is due to gravity , phase transitions , sources and sinks , or when we transform a nonhomogeneous boundary condition problem to a homogeneous one .
the mobility phase in consideration is defined by @xmath9 , where @xmath10 is the absolute ( intrinsic ) permeability of the porous media , @xmath11 is the relative phase permeability and @xmath12 the phase viscosity of the fluid . efficiently and accurately solving the equations like ( [ eq : problem1 ] ) governing fluid flow in oil reservoirs as well as in groundwater modeling and simulation of flow linked to advective / convective transport phenomena ( e.g. , @xcite ) is very challenging because of the complex porous media environment and the intricate properties of fluid phases .
a key ingredient on the transport phenomena in porous media and related real - life applications is precisely the well - known darcy law in which linked to equations in ( [ eq : problem1 ] ) is a fundamental pde with a wide spectrum of relevance .
it is fundamental in applied mathematics @xcite and for modeling fluid flow through porous media @xcite as well as of a benchmark prototype model for proof - of - concept , efficient implementation and rigorous analysis for the design and development of new finite element approaches , as the one discussed here , but also for other novel procedures , for instance msfem @xcite , virtual finite elements @xcite , classical mixed finite elements @xcite . indeed , discontinuous galerkin ( dg ) formulations have become an increasingly popular way to discretize the darcy flow equations , either in the mixed finite element dg @xcite or in the stabilized mixed dg @xcite framework , just to name a few of the relevance of model problem ( [ eq : problem1 ] ) from different perspectives .
the field of fluid flow simulation in petroleum reservoirs @xcite as well as the groundwater modeling and simulation of flow @xcite linked to several transport phenomena have seen significant advances in the last few decades ( see , e.g. , @xcite ) due to novel discretizations associated do darcy problem ( [ eq : problem1 ] ) , along with the challenges in modeling : flow and transport .
we emphasize the challenges in the construction of new methodologies into a reservoir simulation should have into account the following issues : * local mass conservation properties ; * stable - fast solver , and ; * the flexibility of re - use of the novel technique into more complex models ( such as to nonlinear time - dependent equations ) .
also , the convergence of the method should be studied for various kinds of heterogeneities linked to flow in porous media transport problems , as such incompressible immiscible two - phase flow @xcite and incompressible immiscible three - phase flow @xcite ; see also the references cited therein .
the impact of porous media heterogeneity on darcy flow is very relevant .
a review of studies on such topic over the recent past decades can be found in @xcite .
even with modern novel techniques simulation of darcy flows through a heterogeneous porous medium with fine - scale features can be computationally expensive if the flow is fully resolved . moreover , the unstable displacement of fluids with different viscosities , or viscous fingering provides a powerful mechanism to increase fluid - fluid interfacial area and enhance mixing that in linked to the mobility phase in consideration ; ( see also @xcite ) .
thus , fast multiscale reservoir simulations using darcy flow reduced - order models based on the model problem ( [ eq : problem1 ] ) is still up to date @xcite .
the main goal of our work is to obtain conservative solution of the equations above when they are discretized by high order continuous piecewise polynomial spaces .
the obtained solution satisfies some given set of linear restrictions ( may be related to subdomains of interest ) .
our motivations come from the fact that in some applications it is to have some conservative properties represented as conservations of total flux in control volumes .
for instance , if @xmath13 represents the approximation to the flux ( in our case @xmath14 where @xmath15 is the approximation of the pressure ) , it is required that @xmath16 here @xmath17 is a control volume that does not cross @xmath1 from a set of controls volumes of interest , and here and after @xmath18 is the normal vector pointing out the control volume in consideration .
if some appropriate version of the total flux restriction written above holds , the method that produces such an approximation is said to be a conservative discretization .
several schemes offer conservative discrete solutions .
these schemes depend on the formulation to be approximated numerically . among the conservative discretizations for the second the order formulation the elliptic problem we mention the finite volume ( fv ) method , some finite difference methods and some discontinuous galerkin methods . on the other hand , for the first order formulation or the darcy system we have the mixed finite element methods and some hybridizable discontinuous galerkin ( hdg ) methods . in this paper , we consider methods that discretize the second order formulation .
working with the second order formulation makes sense especially for cases where some form or high regularity holds .
usually in these cases the equality in the second order formulation is an equality in @xmath19 so that , in principle , there will be no need to weaken the equality by introducing less regular spaces for the pressure as it is done in mixed formulation with @xmath19 pressure . among the method mentioned above , and especially for second order problems ,
a very popular conservative discretization is the fv method .
the classical fv discretization provides and approximation of the solution in the space of piecewise linear functions with respect to a triangulation while satisfying conservation of mass on elements of a dual triangulation .
when the approximation of the piecewise linear space is not enough for the problem at hand , advance approximation spaces need to be used ( e.g. , for problems with smooth solutions some high order approximation may be of interest ) .
however , in some cases , this requires a sacrifice of the conservation properties of the fv method . here in this paper , we design and analyze conservative solution in spaces of high order piecewise polynomials .
we follow the methodology @xcite that imposes the total flux restrictions by employing lagrange multiplier technique .
we note that fv methods that use higher degree piecewise polynomials have been introduced in the literature .
the fact that the dimension of the approximation spaces is larger than the number of restrictions led the researchers to design some method to select solutions : for instance , in @xcite to introduce additional control volumes to match the number of restrictions to the number of unknowns .
it is also possible to consider a petrov - galerkin formulation with additional test functions rather that only piecewise constant functions on the dual grid .
another approaches have been also introduced , see for instance @xcite and references therein . here in this paper
, we consider a ritz formulation and construct a solution procedure that combines a continuous galerkin - type formulation that concurrently satisfies mass conservation restrictions .
we impose finite volume restrictions by using a scalar lagrange multiplier for each restriction .
this is equivalently to a constraint minimization problem where we minimize the energy functional of the equation restricted to the subspace of functions that satisfy the conservation of mass restrictions .
then , in the ritz sense , the obtained solution is the best among all functions that satisfy the mass conservation restriction .
another advantage of our formulation is that the analysis can be carried out with classical tools for analyzing approximations to saddle point problems @xcite .
we carried out and abstract analysis and present a detailed example for the case of second order piecewise polynomials .
an important finding of these paper is that we where able to obtain optimal error estimates in the @xmath20 norm as well as the @xmath19 norm .
as far as we are informed , optimal @xmath19 approximation is obtained only for specially collocated dual meshes .
the optimal @xmath19 approximation is obtained adding the lagrange to the approximation @xmath21 by an aubin - nitsche trick ; ( see @xcite ) .
the @xmath19 bound is a theoretical advantage of using a symmetric formulation for a conservative method .
the rest of the paper is organized as follows . in section
[ sec : con ] we present the lagrange multipliers formulation or our problem . in section [ sec : dis ] we introduce the saddle point approximation for which and abstract analysis is presented in section [ sec : analysis ] . in section
[ sec : smooth ] we present the particular cases of high - order continuous finite element spaces . for this last case we present some numerical experiments in section [ sec : num ] . to close the paper we present some conclusions in section [ sec : conclusions ] .
denote @xmath22 as the subspace of functions in @xmath23 which vanish on @xmath1 . in case @xmath24 , @xmath22 is the subspace of functions in @xmath23 with zero average on @xmath2 .
the variational formulation of problem ( [ eq : problem1 ] ) is to find @xmath25 such that @xmath26 where the bilinear form @xmath27 is defined by @xmath28 and the functional @xmath29 is defined by @xmath30 in order to consider a general formulation for porous media applications we let @xmath31 in problem ( [ eq : problem ] ) to be uniform elliptic and bounded , however , in certain parts of the paper when analysis and regularity theory are required , we assume @xmath32 . the problem ( [ eq : problem ] ) is equivalent to the minimization problem : find @xmath33 and such that @xmath34 where @xmath35 in order to deal with mass conservation properties we adopt the strategy introduced in @xcite .
let us introduce the meshes we are going to use in our discrete problem .
let the primal triangulation @xmath36 be made of elements that are triangles or squares and let @xmath37 be the number of elements of this triangulation .
we also have a dual triangulation @xmath38 where the elements are called control volumes , and @xmath39 is the number of control volumes .
figure [ figure1 ] illustrates a primal and dual mesh made of squares when @xmath40 , and in this case @xmath41 is equal to the number of interior vertices of the primal triangulation . in general it
is selected one control volume @xmath42 per vertex of the primal triangulation when the measure @xmath43 . in case @xmath24
, @xmath39 is the total number of vertices of the primal triangulation including the vertices on @xmath3 . in order to ensure the mass conservation ,
we impose it as a restriction ( by using lagrange multipliers ) in each control volume @xmath44 .
we mention that our formulation allows for a more general case where only few control volumes , not necessarily related to the primal triangulation , are selected .
let us define the linear functional @xmath45 , @xmath46 .
we first note @xmath47 is not well defined for @xmath48 . to fix that , note that @xmath49 , therefore , let us define the hilbert space @xmath50 with norm @xmath51 .
it is easy to see by using integration by parts with the function @xmath52 that @xmath53 is a continuous linear functional on @xmath54 .
let @xmath6 be the solution of and define @xmath55 , @xmath56 .
the problem ( [ eq : unconstraint ] ) is also equivalent to : find @xmath57 such that @xmath58 where @xmath59 problem ( [ eq : problem - with - restriction ] ) above can be view as lagrange multipliers min - max optimization problem .
see @xcite and references therein .
then , in case an approximation of @xmath6 , say @xmath21 is required to satisfy the constraints @xmath60 , @xmath61 , we can do that by discretizing directly the formulation . in particular
, we can apply this approach to a set of mass conservation restrictions used in finite volume discretizations . in order to proceed with the associate lagrange formulation ,
we define @xmath62 to be the space of piecewise constant functions on the dual mesh @xmath63 .
the lagrange multiplier formulation of problem ( [ eq : problem - with - restriction ] ) can be written as : find @xmath64 and @xmath65 that solve : @xmath66 here , the total flux bilinear form @xmath67 is defined by @xmath68 the functional @xmath69 is defined by@xmath70 the first order conditions of the min - max problem above give the following saddle point problem : find @xmath71 and @xmath72 that solve : @xmath73 see for instance @xcite .
note that if the exact solution of problem ( [ eq : unconstraint ] ) satisfies the restrictions in the saddle point formulation above we have @xmath74 and we get the uncoupled system @xmath75 also observe that the second equation above corresponds to a family of equations , one for each triangulation parametrized by @xmath76 , all of them have the same solution .
recall that we have introduced a primal mesh @xmath36 made of elements that are triangles or squares .
we also have given a dual mesh @xmath77 where the elements are called control volumes .
figure [ figure1 ] illustrates a primal and dual mesh made of squares .
let us consider @xmath78 the space of continuous and piecewise polynomials of degree @xmath79 on each element of the primal mesh , and @xmath80 ( which are the functions in @xmath81 that vanish in @xmath82 ) .
let @xmath62 be the space of piecewise constant functions on the dual mesh @xmath63 .
we mention here that our analysis may be extended to different spaces and differential equations .
see for instance @xcite where we consider gmsfem spaces instead of piecewise polynomials .
the discrete version of ( [ eq : saddlepoint ] ) is to find @xmath83 and @xmath65 such that @xmath84 let @xmath85 be the standard basis of @xmath86 .
we define the matrix @xmath87 \quad \mbox { where } a_{ij } = \int_{\omega } \lambda \nabla\varphi_i \cdot \nabla\varphi_j.\ ] ] note that @xmath88 is the finite element stiffness matrix corresponding to finite element space @xmath86 .
introduce also the matrix @xmath89 \quad \mbox { where } \overline{a}_{ij } = \int_{\partial v_k } -\lambda \nabla\varphi_j \cdot \textbf{n}.\ ] ] with this notation , the matrix form of the discrete saddle point problem is given by , @xmath90 \left [ \begin{array}{c } u^h\\ \lambda^h \end{array } \right ] = \left [ \begin{array}{c } f\\ \overline{f } \end{array } \right]\ ] ] where the vector @xmath91 is defined by , @xmath92 \quad \mbox { with } f_{i } = \int_{\omega } q\ , \varphi_i & & \overline{f } = [ \overline{f}_k]_{k=1}^{n_h^ * } = \int_{v_k } q.\end{aligned}\ ] ] for instance , in the case of the primal and dual triangulation of figure [ figure2 ] and polynomial degree @xmath93 , the finite element matrix @xmath94 is a sparse matrix with 19 diagonals .
also , for a control volume @xmath95 that has no empty interception with , there are at most @xmath96 supports of basis functions @xmath97 , see figure [ figure2 ]
. then , matrix @xmath98 is also sparse . basis function . ]
note that matrix @xmath98 is related to classical ( low order ) finite volume matrix .
matrix @xmath98 is a rectangular matrix with more columns than rows .
several previous works on conservative high - order approximation of second order elliptic problem have been designed by `` adding '' rows using several constructions .
one can proceed as follows : 1 . construct additional control volumes and test the approximation spaces against piecewise constant functions over the total of control volumes ( that include the dual grid element plus the additional control volumes ) .
we mention that constructing additional control volumes is not an easy task and might be computationally expensive .
we refer the interested reader to @xcite for additional details .
2 . use as additional the basis functions the basis functions that correspond to nodes other than vertices to obtain an fv / galerkin formulation .
this option has the advantage that no geometrical constructions have to be carried out . on the other hand
, this formulation seems difficult to analyze . also , some preliminary numerical tests suggest that the resulting linear system becomes unstable for higher order approximation spaces ( especially for the case of high - contrast multiscale coefficients ) .
3 . use the ritz formulation with restrictions . note that if the polynomial degree @xmath99 , in the linear system , the restriction matrix corresponds to the usual finite volume matrix .
this matrix is known to be invertible . in this case , the affine space @xmath100 is a singleton . moreover , the only function @xmath101 satisfying the restriction is given by @xmath102 .
the ritz formulation reduces to the classical finite volume method .
then , in the ritz sense , the solution of is not worse that any of the solutions obtained by the methods 1 . or 2 . mentioned above . furthermore , the solution of the associated linear system , which is a saddle point linear system can be readily implemented using efficient solvers for the matrix @xmath88 ( or efficient solvers for the classical finite volume matrix @xmath98 ) ; see for instance @xcite . additionally , we mention that the analysis of the method can be carried out using usual tools for the analysis of restricted minimization of energy functionals and mixed finite element methods .
the numerical analysis of our methodology is under current investigation and it will presented elsewhere .
we show next that imposing the conservation in control volumes using lagrange multipliers does not interfere with the optimality of the approximation in the @xmath20 norm .
as we will see , imposing constraints will result in non optimal @xmath19 approximation but we were able to reformulate the @xmath19 approximation to get back to the optimal approximation by using the discrete lagrange multiplier as a corrector . before proceeding we introduce notation to avoid proliferation of constants .
we use the notation @xmath103 to indicate that there is a constant @xmath104 such that @xmath105 .
if additionally there exist @xmath106 such that @xmath107 we write @xmath108 .
+ denote @xmath109 for all @xmath110 and let us remind that @xmath111 , and set @xmath112 .
we present a concrete example of the norms of @xmath113 and @xmath114 in the next section , see ( [ eq : norm - for - p2 ] ) and ( [ eq : norm - for - mh ] ) , respectively .
+ * assumption a : * there exist norms @xmath115 and @xmath116 for @xmath113 and @xmath114 , respectively , such that 1 . augmented norm : @xmath117 forall @xmath118 .
continuity : there exists @xmath119 such that @xmath120 3 .
inf - sup : there exists @xmath121 such that @xmath122 the inf - sup condition above can be replaced by : there exists @xmath121 such that @xmath123 we have the following result .
assume that @xmath124 is the solution of ( [ eq : saddlepoint ] ) and @xmath125 the solution of ( [ cela ] ) .
we have the following result .
the proof uses classical approximation techniques for saddle point problems .
[ thm : t1 ] assume that `` assumption a '' holds .
then , there exists a constant @xmath126 such that @xmath127 note that in both problems , and , @xmath12 belongs to the finite dimensional subspace @xmath128 . also , the exact solution of the lagrange multiplier component of is @xmath74 .
now we derive error estimates following classical saddle point approximation analysis .
define @xmath129 and @xmath130 first we prove @xmath131 the inf - sup above in implies that @xmath132 ( as well as @xmath133 ) is not empty .
take any @xmath134 and solve for @xmath135 the problem , @xmath136 since @xmath27 is elliptic there exists a unique solution and therefore @xmath137 where @xmath15 is the solution of .
we have from and and using that @xmath138 then , by using the elipticity of @xmath27 , we have @xmath139 then @xmath140 so that holds true .
we now show that @xmath141 take any @xmath142 .
the inf - sup condition implies that there exists a unique @xmath143 such that @xmath144 then we have that @xmath145 , @xmath146 and therefore @xmath147 note that we have used the continuity of @xmath148 in the extended norm @xmath149 .
put @xmath150 then @xmath151 therefore we have that @xmath152 .
moreover , @xmath153 combining and we get the result .
as showed up in the numerical experiments the error @xmath154 is not optimal but according to the next result if we correct @xmath15 to @xmath155 we do get optimal approximation .
the proof of the following results follows a duality argument similar to that of the aubin - nitsche method ; see @xcite .
since we will use regularity theorem , we will assume from now on that @xmath156 ( identity ) .
in this case , @xmath157 , and we will show in the next section that `` assumptions a and b '' hold and @xmath158 and @xmath159 .
assume that the problem is regular ( see @xcite ) and `` assumptions a and b '' hold .
then , @xmath160 for @xmath161 define @xmath162 and @xmath163 as the solution of @xmath164 analogously , define @xmath165 as the solution of @xmath166 observe that @xmath167 , @xmath168 and @xmath169 . according to our previous result in theorem
[ thm : t1 ] combined with standard regularity and approximation results ( @xcite ) we have @xmath170 recall that , @xmath171 by using the definition of @xmath172 , @xmath173 and @xmath174 in and we get @xmath175 in the last step we have used . replacing the last inequality in we get our result .
in this section we consider a regular mesh made of squares . see figure [ figure1 ] .
define @xmath176 that is , @xmath177 is the interior interface generated by the dual triangulation . for @xmath178 define @xmath179 $ ] on @xmath177 as the jump across element interfaces , that is , @xmath179|_{\partial v_k \cap \partial v_{k^{\prime}}}= \mu_k-\mu_{k^{\prime}}$ ] .
note that for @xmath180 @xmath181.\end{aligned}\ ] ] for each control volume @xmath42 , denote by @xmath182 the set of element of the primal mesh that intersect @xmath42 .
note that in each control volume we have @xmath183 to motivate the definition of the norms we study the continuity of the bilinear form @xmath184 .
observe that , @xmath185\right)^2 & \leq \left ( h\int_{\partial v_k \cap \partial v_{k^\prime } } ( \nabla p \cdot \mathbf{n } ) ^2 \right ) \
, \left ( \frac{1}{h } \int_{\partial v_k \cap \partial v_{k^\prime } } [ \mu]^2 \right).\end{aligned}\ ] ] and therefore by applying cauchy inequality and adding up we get , @xmath186 ^ 2 \right ) ^{1/2}.\ ] ] using an inverse inequality ( @xcite ) we get that @xmath187 now we are ready to define the norm @xmath188 note that if @xmath189 then @xmath190 .
also , if @xmath191 we have @xmath192 by using inverse inequality . also define the discrete norm for the spaces of lagrange multipliers as @xmath193 ^ 2.\ ] ] we have shown above that the form @xmath184 is continuous , that is , there is a constant @xmath126 such that , latexmath:[\ ] ] this finish our proof .
* assumption b : * let @xmath222 . then following approximation holds @xmath223 we mention that , for regular meshes and @xmath224 finite element spaces and the norm @xmath225 defined in , assumption b follows from classical finite element approximation and stability results of finite element interpolation operators .
see @xcite .
consider the norms for @xmath113 and @xmath114 defined in ( [ eq : norm - for - p2 ] ) and ( [ eq : norm - for - mh ] ) , respectively . if @xmath226 and `` assumptions a and b '' hold , then @xmath227 and if the problem is regular @xmath228 we finish this section by mentioning that our analysis is valid for high order finite element on regular meshes made of triangles since a similar analysis holds in this case .
we consider the dirichlet problem and employ the meshes depicted in figure [ figure1 ] with a variety of mesh sizes .
we impose conservation of mass as described in the paper by using lagrange multipliers . for this paper
, we solved the saddle point linear system by lu decomposition .
several iterative solvers can be proposed for this saddle point problem but this will be considered in future studies and not here .
consider * @xmath229\times [ 0,1]$]*. we consider a regular mesh made of @xmath230 squares .
the dual mesh is constructed by joining the centers of the elements of the primal mesh .
we performed a series of numerical experiments to compare properties of fem solutions with the solution of our high order fv formulation ( to which we refer from now on as fv solution ) . the fv formulation with correction we denote by fv + @xmath231 .
we selected the following forcing term and exact solution , @xmath232 first we implemented the case of @xmath233 elements that corresponds to the classical finite element and classical finite volume methods .
we compute @xmath19 and @xmath20 errors .
we present the results in table [ tabl22q1 ] and displayed graphically in figures [ logl22q1 ] and [ logh12q1 ] .
we observe here optimal convergence of both strategies .
.table of * fem * and * fv * @xmath19 and @xmath20 errors for numerical solutions of example 1 , using @xmath234 discretization , calculated over @xmath96 different values of mesh norm , @xmath235 . [ cols="^,^,^,^,^",options="header " , ]
as we previously announced , we have emphasized the challenges in the construction of new methodologies into a reservoir simulation should have into account the following issues : 1 ) local mass conservation properties , 2 ) stable - fast solver and 3 ) the flexibility of re - use of the novel technique into more complex models ( such as to nonlinear time - dependent transport equations equation for the convection dominated transport equation ) . for darcy - like model problems with very high contrasts in heterogeneity , the discretization of model ( [ eq : problem1 ] ) alone may be very hard to solve numerically due to a large condition number of the arising stiffness matrix .
moreover , the situation in even more intricate for modeling non trivial two- @xcite and three - phase @xcite transport convection dominated phenomena problems for flow through porous media ( see also other relevant works @xcite ) .
in addition , existence , uniqueness and regularity issues for such problems at a fine level out of reach .
thus , numerical simulation of fluid dynamics provide insight into the numerical simulation of fluid flow linked to theory , numerics and applications . with this in mind , as a future perspective is to plug our novel high - order conservative finite element method in a numerical simulator . for the purpose of such study
it is necessary to describe a coupled pressure - velocity ( elliptic ) and convection dominated transport ( parabolic with a strong hyperbolic character ) conservative solution and stable - fast algorithm to model essentially the interface capturing problems for multiphase flows where neither of the phases can be regarded as dominant .
while the volume fractions ( or saturation ) are gained by solving the pertinent transport equations , fluid flow velocity is solved by solving the darcy pressure problem ; although the need of the pressure as primary unknown , is just the darcy velocity that we need to plug into the convection dominated transport system .
thus , to achieve a sufficiently coupling between the volume fractions ( or saturation ) and the pressure - velocity , the full problem can be treated along with a fractional - step numerical procedure @xcite ; we point out that we are aware about the very delicate issues linked to the discontinuous capillary - pressure ( see @xcite and the references therein ) ; such issues must be considered no matter how is the time discretization of the convection dominated transport , namely , explicitly or implicitly . before continuing , remember that figure [ figure1 ] illustrates a primal @xmath236 and dual @xmath63 mesh made of squares .
thus , a feasible time marching algorithm ( see , e.g. @xcite ) to the full set of nonlinear differential model saturation - pressure - velocity equations is given by solving for the saturation , in hyperbolic -parabolic sub - steps , and the total darcy velocity , in the velocity - pressure sub - steps . the approximation of elliptic pressure equation will take advantage of our high - order conservative finite element method while the transport equation can be handle by fast - accurate finite - volume method , by combining primal and dual meshes as we are using here . indeed
, the fluxes ( darcy velocities ) are smooth at the vertices of the cell defining the integration volume in the dual triangularization , since these vertices are located at the centers of non - staggered cells , away from the jump discontinuities along the edges .
this facilitates the construction of second - order and high - order approximations linked to the hyperbolic - parabolic model problem .
this also means that the pertinent spatial integrals can be approximated in a straightforward manner .
for instance , the finite volume differencing @xcite , unlike upwind differencing , bypasses the need for riemann solvers , yielding simplicity , avoiding dimensional splitting in multi - dimensional problems .
in particular , such framework also allows for the extension of the scheme to hyperbolic systems by component - wise application of the scalar framework ( for more details see @xcite and the references cited therein ) .
moreover , at the same vertices of the integration volume , the novel high - order conservative finite element method gives the best accurate velocity field even in the presence of highly variable permeability fields , providing useful insights into the numerical simulation of fluid flow .
this gives some of the benefits of staggering between primal and dual mesh triangulation by combining our novel high - order conservative finite element method with finite volume for hyperbolic - parabolic conservation laws modeling fluid flow in oil reservoirs and groundwater modeling as well as associated hazards of contamination and transport by ground water , co2 geological sequestration and storage , acid mine drainage remediation , just to name a few up to date subsurface hydrology and groundwater modeling , and geothermal energy associated to petroleum science and engineering .
e. abreu ( p.i . ) thanks financial support fapesp through grants no .
2014/03204 - 9 and cnpq through grants no .
mcti / cnpq / universal 445758/2014 - 7 .
ciro diaz thanks capes for a graduate fellowship .
marcus sarkis thanks in part by nsf - mri 1337943 and nsf - mps 1522663 d. braess , finite elements ( iii edition ) ( theory , fast solvers , and applications in elasticity theory , translated from the german by larry l. schumaker ) , cambridge university press , cambridge 2007 .
h. brezis and f. analysis , sobolev spaces and partial differential equations , springer science business media , 2010 , 600 pages .
j. galvis , y. efendiev , domain decomposition preconditioners for multiscale flows in high contrast media .
reduced dimension coarse spaces , siam j. multiscale modeling and simulation 8 ( 2010 ) 16211644 .
m. ghasemi , y. yang , e. gildin , y. efendiev and v. calo , fast multiscale reservoir simulations using pod - deim model reduction .
paper 173271-ms presented at the spe reservoir simulation symposium , houston , usa , 22 - 25 february 2015 .
m. presho and j. galvis , a mass conservative generalized multiscale finite element method applied to two - phase flow in heterogeneous porous media , journal of computational and applied mathematics , 296 ( 2016 ) 376 - 388 . | a new high - order conservative finite element method for darcy flow is presented .
the key ingredient in the formulation is a volumetric , residual - based , based on lagrange multipliers in order to impose conservation of mass that does not involve any mesh dependent parameters .
we obtain a method with high - order convergence properties with locally conservative fluxes .
furthermore , our approach can be straightforwardly extended to three dimensions .
it is also applicable to highly heterogeneous problems where high - order approximation is preferred . |
the spectacular hickson compact group 31 ( hickson 1982 ) shows a wide range of indicators of galaxy interaction and merging : tidal tails , irregular morphology , complex kinematics , vigourous star bursting ( e.g. rubin et al . 1990 ) and possible formation of tidal dwarf galaxies ( e.g. hunberger et al .
all the objects belonging to the group are embedded in a common large hi envelope ( williams et al .
the group is formed by members a+c , b , e , f , g , q ( rubin et al . 1990 ) .
two scenarios have been put forward to explain the nature of the central system a+c : it is either two systems that are about to merge ( e.g. rubin et al . 1990 ) or a single interacting galaxy ( richer et al .
2003 ) . in this letter
we use our new fabry - perot maps and deep imaging from gemini - n to revisit the important problem of the merger nature of the central object of the group , a+c , and , in addition , we investigate the internal kinematics of the tidal dwarf galaxies , in an attempt to identify if they are self gravitating objects or not . we adopt a distance of 54.8 mpc , from the redshift z=0.0137 ( hickson et al . 1992 ) and using h@xmath3=75 km s@xmath4 mpc@xmath4 , hence 1 " @xmath5 0.27 kpc .
observations were carried out with the fabry - perot instrument cigale ( gach et al .
2002 ) attached to the eso 3.6 m telescope in august 2000 .
interferograms were obtained with a high order ( p=1938 ) fabry - perot scanning interferometer , giving a free spectral range of @xmath6 with a _ finesse = 24 _ leading to a spectral resolution of @xmath7 .
the pixel size is 0.405 arcsec ; the total exposure time was 72 min ( 6 cycles of 12 min each , 48 scanning steps per cycle ) and the fwhm of the interference filter centered around 6651 @xmath8 was 20 @xmath8 .
the velocity sampling was 3 @xmath9 @xmath10 and the relative velocity accuracy is @xmath11 over the whole field where the s / n is greater than 3 .
reduction of the data cubes were performed using the cigale / adhocw software ( boulesteix , 2002 ) .
the data reduction procedure has been described e.g. in blais - ouellette et al .
( 1999 ) and garrido et al .
for the adopted distance of the group , one pixel corresponds to @xmath5 0.11 kpc . in august 2003 , we obtained two images with gmos at gemini - n in g and r with exposure times of 1200 and 900 seconds respectively .
these images have a pixel size of 0.14 `` and typical seeing of 0.75 '' .
a color map of the group is presented in fig .
1a . it shows a wealth of star forming regions and the large extent of the optical diffuse light which envelopes the group .
for the first time , regions @xmath2 and @xmath0 , to the east of complex a+c , are seen in great detail and depth .
d show the h@xmath12 monochromatic map of the group , the velocity map and several zoom panels showing the typical velocity profiles in selected regions of the group .
the velocity field was corrected from free spectral range ambiguity using previous kinematic observations . at a first look
, galaxies a and c appear to be a single kinematic entity , as their velocity fields show no discontinuity . under this assumption
, we can infer a rough mass for the a+c complex , within a radius of @xmath5 18 arcsec ( @xmath5 4.9 kpc ) of @xmath13 .
the following parameters and assumptions were used for this mass determination : a maximum rotational velocity of 70@xmath14 ( only from the nw side of the system given that the se side is too disturbed ) , a kinematic inclination and position angle of 51@xmath15 and 130@xmath16 degrees respectively and the assumption that the measured motions are due to disk rotation . nevertheless , multiple profiles ( see fig .
1d ) , evident almost everywhere in the main body of a+c , strongly suggests that a+c is not a single entity and therefore this rough determination provides only an order of magnitude for the mass .
2 shows the velocity gradients of the objects @xmath2 , @xmath0 , e and f , situated around the pair a+c ( see fig 1a ) and which have been thought to be candidate tidal dwarf galaxies ( hunsberger et al . 1996 , iglesias - pramo & vlchez 2001 and richer et al . 2003 ) .
the curves are not corrected for inclination and the central position and velocity were chosen such that the curves were as symmetric as possible , with both sides matching , when possible .
the object with the highest velocity gradient is e , with ordered velocities which range from 3950 @xmath17@xmath4 to 4000 @xmath17@xmath4 .
object @xmath0 also shows ordered motion , with velocities going from 4125 to 4175 @xmath17@xmath4 .
surprisingly , object f , thought to be the best tidal dwarf galaxy candidate in the group , shows a completely flat rotation curve , as does also the smaller object to the northeast of a+c , fragment @xmath2 .
in addition , @xmath2 and f have low gaseous velocity dispersions .
these results will be discussed in the next section .
we obtained the map of the gaseous velocity dispersion at each pixel of the image ( amram et al .
2004 , in preparation ) assuming the profiles are well represented by a single gaussian .
the value for the velocity dispersion ranges from 10 to 30 @xmath17@xmath4 throughout the group . in particular , objects @xmath2 , @xmath0 , e and f show typical velocity dispersions of 15 @xmath17@xmath4 , which in some isolated regions can reach up to 25 @xmath17@xmath4 .
the highest values for the velocity dispersion lie in the overlapping region between a and c , mainly due to double components .
it is noticeable that these highest values do not match the most intense star forming regions everywhere in the galaxies but particularly where disk a and c overlap , implying that the line broadening and the multiple components are not directly linked to star formation triggered by interaction with another galaxy but specifically by the merging of a+c .
hickson & menon ( 1985 ) reached a similar conclusion , analyzing a radio continuum map of the group .
they found that the 20 cm peak of emission comes from the overlapping regions of a and c , indicating additional evidence of recent excessive starburst activity in this region .
richer et al . ( 2003 ) also presented fabry - perot velocity maps of hcg 31 .
there is fairly good general agreement between their velocity field and ours .
the spectral resolution of our maps is nevertheless six times higher ( 7900 vs 45900 ) and the detection limit several magnitudes fainter .
several authors supported that a and c are separate entities in an ongoing merging phase ( e.g. vorontsov - velyaminov and arhipova , 1963 ; rubin et al . ,
1990 ) while richer et al .
( 2003 ) supported the scenario that a+c is a single interacting spiral galaxy .
the new piece of evidence in support of the merging scenario reported in this paper is the presence of double velocity components throughout the system a+c .
in fact , our higher resolution velocity field shows kinematic structures which are not naturally explained by a single disk but by the merging of two disks .
although sc galaxies are , in general , transparent objects ( bosma 1995 ) , moderate amounts of molecular gas ( as traced by the co ) and cold dust may make some regions opaque . the co emission is weak in hcg 31 but the brightest co peak occurs in the overlapping region between galaxies a and c ( yun et al . 1997 ) where the broader h@xmath12 profiles are observed .
if the co belongs to the foreground galaxy , multiple components are observed in regions optically thick and then could not be observed if disk a and c are two separate galaxies seen in projection , i.e. chance alignment .
multiple gaseous components are observed in the same disk plane when they are not in equilibrium ( e.g. stlin et al .
this occurs when two different entities merge , or when the feedback gas due to star formation interacts with the ism .
we observe in hcg 31 the signature of both mechanisms , the second one being probably a consequence of the first one .
the general pattern of the velocity field of hcg31 a+c is somewhat similar to that of ngc 4038/9 ( the antennae , amram et al . , 1992 )
in which a continuity in the isovelocities between both galaxies is also observed in the overlapping region .
the merging stage of the antennae is slightly less advanced than that for a+c . in the antennae , the two galaxies are clearly separate entities and their bodies , which are not yet overlapping , each display an increasing velocity gradient , which is roughly parallel and run from the ne to the sw ( amram et al .
it is likely that when the disks of ngc 4038 and of ngc 4039 overlap , the velocity field will also present total continuity , as observed in a+c .
galaxy c ( mrk 1089 ) has been classified as a double nucleus markarian galaxy , the two nuclei being separated by 3.4 arcsec ( mazzarella & boroson , 1993 ) and it is difficult to explain the existence of the two nuclei without invoking a merging scenario , as shown by numerical simulations ( e.g. barnes & hernquist 1992 ) . to reproduce the double line profile in ngc 4848 , vollmer et al ( 2001 )
have used numerical simulations .
they interpret them as the consequence of infalling gas which collides with the ism within the galaxy .
this gives rise to an enhanced star formation observed in the h@xmath12 and in the 20 cm continuum map .
hcg 31 is most probably a group in an early phase of merger , growing through slow and continuous acquisition of galaxies from the associated environment .
moreover , several evidences for interaction with the other galaxy components of the group , namely b , q and g ( e.g. the group is completely embedded in a large hi envelope showing a local maximum on g to the sw and another one around q to the ne ) indicate that the complex a+c is most probably accreting the surrounding galaxies . several papers in the past ( hunsberger et al .
1996 , johnson and conti 2000 , richer et al .
2003 and iglesias - pramo & vlchez 2001 ) have mentionned the possibility that tidal dwarf galaxies in hcg 31 were formed .
the best candidates are objects e and f , for which metallicities were measured and they were determined to be similar to that of the complex a+c , suggesting a tidal origin for these objects ( richer et al . , 2003 ) . as shown in fig .
2 , we detect ordered motions only for objects @xmath0 and e and not for @xmath2 and f. it might be suspected that the internal velocity motions measured in @xmath0 and e could be due to streaming motions in incipient tidal tails in formation ( see fig .
this is , however , not the case because these objects are counterrotating with respect to the main body of a+c .
the discontinuity of the isovelocities can be clearly seen from fig .
1c : the velocities go from high ( northeast ) to low ( southwest ) in a+c , towards object e. then , along the body of e they go in the opposite sense .
similarly , for object @xmath0 , it presents counterrotation with respect to its immediate neighbor to the west : galaxy a. these objects may fall back onto their progenitor . in fact , from their velocity differences with respect to the a+c complex ( + 115 @xmath17@xmath4 and -60 @xmath17@xmath4 respectively ) and from their relative projected distances ( 6.7 kpc and 5.4 kpc respectively ) and assuming a total mass for the a+c complex of @xmath13 , we could determine that these two objects will indeed , most probably , fall back onto a+c .
the same is true for object f , which although more distant from a+c , has a very small radial velocity difference of @xmath5 60 @xmath17@xmath4 .
we note , however , that given the fact that we measure radial velocities ( and not the velocity component in the plane of the sky ) , the observed internal velocities of the tidal fragments are lower limits .
region f , which was previously thought to be the best example of a tidal dwarf galaxy in hcg 31 , is indeed part of the main merger , following the same kinematic pattern of the parent galaxy and presenting no rotation nor significant internal velocity dispersion .
region f was found to have low or inexistent old stellar population by johnson and conti ( 2000 ) . if there is no old stellar population the velocity dispersion of the gas is mainly indicating the dynamics of the cloud .
the range of values derived for the internal velocity dispersion of f , 15 - 25 @xmath17@xmath4 , is too close to the natural turbulence of the gas and/or the expanding velocity due to starburst winds .
the lack of rotation and the continuity of the kinematics between the main body of the merger and object f suggest it is simply a tidal debris , although , considering its projected distance from a+c ( 16 kpc ) and the large amount of fuel available in the whole area of the merger ( 2.1 @xmath18 of hi gas ) , it could perhaps develop into a tidal dwarf galaxy in the future , by accretion of infalling material .
there is also the possibility that regions @xmath2 and f present no velocity gradient due to their rotation pattern be along the line of sight .
although this could be a possibility for the smaller and rounder region @xmath2 , it is less likely the case for region f , given its elongated morphology .
our two main results are : \1 ) we measure multiple kinematic components throughout the body of a+c which we interpret as a strong indication that this complex is an ongoing merger . the double photometric nucleus has been identified in several previous images of hcg 31 including the spectacular hst image published by johnson and conti ( 2000 ) and in fig .
the double kinematic component is shown here for the first time in fig .
\2 ) f and @xmath2 present flat rotation velocity profiles and insignificant velocity dispersions . in contrast ,
@xmath0 and e are structures counterrotating with respect to the parent a+c complex .
we conclude that hcg 31 is a merging group which is probably going to soon end up as a field elliptical galaxy
. it would be very valuable to compare our new , high spectral resolution maps to simulations of groups to investigate which interaction and merger parameters fit these data .
the authors thank olivier boissin for help during the observations , jorge iglesias - pramo for kindly providing the calibrated h@xmath12 image and to acknowledge financial support from the french - brazilian pics program .
cmdo , ls and esc would like to thank the brazilian pronex program , fapesp , cnpq .
cmdo deeply acknowledges the funding and hospitality of the mpe institut in garching , where this work was finalized .
cc and oh acknowledge support from fqrnt , qubec and nserc , canada .
amram p. , marcelin m. , boulesteix j. et al 1992 , a&a , 266 , 106 amram p. , mendes de oliveira c. , boulesteix j. et al .
1998 , a&a , 330 , 881 blais - ouellette s. ; carignan c. , amram p. et al , 1999 , aj , 118 , 2123 boulesteix j. , 2002 , adhocw red.package , www.oamp.fr/adhoc/ bosma a. , 1995 in nato advanced science inst .
c , 469 , 317 gach j .-
l . , hernandez o. , boulesteix j. et al .
2002 , pasp , 114 , 1043 garrido o. , marcelin m. , amram p. et al , 2002 , a&a , 387 , 821 hickson p. 1982 , apj , 255 , 382 hickson p. , mendes de oliveira c. , huchra j.p .
1992 , apj 399,353 hickson p. & menon t.k . 1985 ,
apj 296 , 60 hunsberger s.d .
, charlton j.c . & zaritsky d. 1996 , apj 462 , 50 iglesias - pramo j. & vlchez j.m . 2001 , apj 550 , 204 johnson k.e . &
conti p.s .
2000 , aj 119 , 2146 stlin g. , amram p. , bergvall n. et al .
2001 , a&a 374 , 800 richer m.g .
, georgiev l. , rosado m. et al . , 2003 , a&a , 397 , 99 rubin v.c . ,
hunter d.a . ,
ford w.k.jr .
1990 , apj , 365 , 86 vollmer b. , braine j. , balkowski c. et al . , 2001 , a&a,374,824 vorontsov - velyaminov b.a . &
arhipova , trudy g. astron .
sht . , 33 , 1 , 1963 williams b.a . , mcmahon p.m. , van gorkom j.h . , 1991 ,
aj , 101 , 1957 yun m.s . , verdes - montenegro l. , del olmo a. et al . , 1997 , apj , 475l,21 . | we have obtained high spectral resolution ( r = 45900 ) fabry - perot velocity maps of the hickson compact group hcg 31 in order to revisit the important problem of the merger nature of the central object a+c and to derive the internal kinematics of the candidate tidal dwarf galaxies in this group .
our main findings are : ( 1 ) double kinematic components are present throughout the main body of a+c , which strongly suggests that this complex is an ongoing merger ( 2 ) regions @xmath0 and e , to the east and south of complex a+c , present rotation patterns with velocity amplitudes of @xmath1 and they counterrotate with respect to a+c , ( 3 ) region f , which was previously thought to be the best example of a tidal dwarf galaxy in hcg 31 , presents no rotation and negligible internal velocity dispersion , as is also the case for region @xmath2 .
hcg 31 presents an undergoing merger in its center ( a+c ) and it is likely that it has suffered additional perturbations due to interactions with the nearby galaxies b , g and q. |
within the last decade , a new class of compact stellar systems has been discovered : the so - called `` ultra - compact dwarf galaxies '' ( ucds , phillips et al .
2001 ; see also hilker et al .
1999 ; drinkwater et al .
2000 & 2003 ; mieske et al .
2004 , 2006 , 2007 ; haegan et al .
ucds are characterised by old stellar populations , absolute magnitudes @xmath1 mag , half - light radii @xmath2 and masses @xmath3 .
hence they are larger , brighter and more massive than the biggest milky way globular clusters ( gcs ) , but at the same time significantly more compact than typical dwarf galaxies of comparable luminosity .
there are three main hypotheses on the origin of ucds : 1 .
they are remnant nuclei of tidally stripped dwarf galaxies ( zinnecker et al .
1988 ; bassino et al . 1994 ; bekki et al .
2 . ucds are merged star cluster complexes created in galaxy mergers ( fellhauer & kroupa 2002 and 2005 ) .
ucds are genuine compact dwarf galaxies originating from small - scale peaks in the primordial dark matter power spectrum ( drinkwater et al . 2004 ) . in the latter case
, ucds may be expected to still be dark matter dominated .
a powerful way to investigate the nature of ucds is to measure their dynamical mass - to - light ratios ( m / l ) . contrasting the measurements with predictions from stellar population models
, one can estimate the amount of unseen mass in ucds , potentially constraining their formation scenarios . in the virgo cluster , haegan et al .
( 2005 ) have found several ucds with m / l ratios well above the ranges predicted by stellar population models that assume a canonical imf ( kroupa or chabrier ) .
evstigneeva et al . (
2007 ) investigate a disjunct sample of 6 virgo ucds and find that within the errors , their m / l ratios are compatible with canonical ssp models .
hilker et al .
( 2007 ) study the internal dynamics of 5 fornax ucds and the 6 virgo ucds from evstigneeva et al .
they arrive to the conclusion that the ucd m / l ratios tend to be above predictions from ssp models assuming a canonical kroupa or chabrier imf , but are consistent with a salpeter imf , which is characterised by being more bottom - heavy than the canonical imf ( kroupa 2001 ) . in fig .
[ mass_ml ] we plot the available literature m / l@xmath4 measurements for compact stellar systems vs. their mass , ranging from low - mass globular clusters up to the most massive ucds .
this sample spans a mass range of 4 decades . clearly , starting at a mass of @xmath52@xmath6 m@xmath7 ,
the data show a trend of increasing m / l with increasing mass ( this trend has also been noted by evstigneeva et al .
2007 and rejkuba et al . 2007 ) . in this context
it is important to stress that stellar population models predict an increase of m / l@xmath4 ratio with increasing metallicity , as noted previously in several studies ( e.g. haegan et al .
2005 , hilker et al .
2007 , evstigneeva et al . 2007 ) . therefore
, a possible increase of metallicity with mass among the sample of compact objects investigated will naturally create a trend of m / l@xmath4 with mass as seen in fig . [ mass_ml ] . in order to remove such a possible effect from the data , we normalise the m / l@xmath4 ratio values in fig . [ mass_ml ] to the same metallicity .
the result of this exercise is shown in fig .
[ mass_ml_norm ] . for normalising , we use the m / l predictions as a function of [ fe / h ] for a 13 gyr population from the two model sets of bruzual & charlot ( 2003 , chabrier imf ) and maraston ( 2005 , kroupa imf ) ( see also dabringhausen et al .
2008 ) . to each of the two sets of model predictions we fit a function of the form @xmath8)\hspace{0.5 cm } \label{form_norm}\ ] ]
the parameters of the fits are shown in table 1 .
for each of the two model sets , we then normalise the literature m / l values in the following way : @xmath9 where @xmath10 is the theoretical prediction for [ fe / h]=0 , i.e. solar metallicity .
.parameters of the exponential fit ( see equation [ form_norm ] ) to the relation of m / l as f([fe / h ] ) for a 13 gyr population with a canonical imf , shown for the two sets of literature models .
[ table_1 ] [ cols="<,>,>,>",options="header " , ] + @xmath11 rejkuba et al .
[ fe / h ] from broad - band colours , see text .
@xmath12 haegan et al .
[ fe / h ] from broad - band colours , see text .
@xmath13 evstigneevae et al .
[ fe / h ] from spectroscopy , see text .
@xmath14 hilker et al .
[ fe / h ] from spectroscopy for ucd1 , ucd3 , ucd4 , from ( v - i ) for ucd2 and ucd5 . in fig .
[ mass_ml_norm ] we show the normalised m / l ratios , in separate plots for the two model sets chosen for normalisation .
we note that the normalisation to solar metallicity is of course an arbitrary choice , but we prefer this particular choice in order to have a well defined reference that is independent on the actual metallicity of the investigated population . in fig .
[ mass_ml_norm ] , a change of the normalisation would change nothing but the scale of the y - axis .
the location of the data with respect to the m / l ratios expected from the models is independent on the choice of normalisation .
for both model choices plotted in fig .
[ mass_ml_norm ] it is clear that the distribution of mass vs. m / l ratios does exhibit a very interesting separation at about 2@xmath1510@xmath16 m@xmath7 : sources with lower masses have m / l ratios in line with canonical stellar populations , whereas sources with higher masses have m / l ratios that tend to be above those for canonical stellar populations .
the ratio of the average m / l ratio between low- and high - mass sources is 2.3 @xmath17 0.2 for both choices of normalisation .
the occurence of this break at 2@xmath1510@xmath16 m@xmath7 is in line with previous findings of breaks in terms of size ( haegan et al . 2005 , mieske et al .
2006 ) and metallicity ( mieske et al .
2006 ) at about the same mass .
it hence strengthens the proposed separation between canonical globular clusters and ultra - compact dwarf galaxies at this mass ( haegan et al .
2005 , mieske et al .
interestingly , the separation mass corresponds to the mass range where the relaxation time of compact stellar systems is comparable to a hubble time ( dabringhausen et al .
the high m / l ratios for ucds can be explained by the presence of dark matter , which would have to be at very high densities ( @xmath18 pc@xmath19 within 20 pc ) .
this is comparable to central density values expected for cuspy dark matter halos ( walker et al .
2007 ) . however , it is incompatible with recent claims of cored dark matter halos with low central densities in dwarf spheroidal galaxies ( e.g. gilmore et al . 2007 ) , which are the most dark matter - dominated systems known ( see however metz & kroupa 2007 for an entirely different interpretation )
. investigating the nature of the high m / l ratios in ucds therefore offers important constraints on the phase space properties of dark matter particles .
here , we describe a method to test whether a bottom heavy stellar initial mass function ( imf ) causes the high m / l@xmath4 ratios of ucds . such steep low - mass stellar imfs may result if the radiation field , stellar winds and supernova explosions in ucds were so intense during their formation , that pre - stellar cloud cores were ablated before they could fully condense to stars ( kroupa & bouvier 2003 ) . by confirming such an overabundance of low - mass stars with respect to a canonical imf ( kroupa 2001 )
, we would for the very first time have unambiguous evidence for a radically different star - formation process under extreme physical conditions when ucds formed .
by excluding this scenario we would know that the high m / l@xmath4 ratios are probably due to dark matter .
this could be in the form of stellar remnants , perhaps as a result of a top - heavy imf in primordial and pristine star formation regions ( pop iii , jimenez & haiman 2006 , tornatore et al .
2007 , dabringhausen et al . 2008 ) .
alternatively , it could be due to non - baryonic dark matter , which in turn would open new avenues towards studying the phase - space properties of dark matter particles ( see above ) .
we note that tidal heating ( fellhauer & kroupa 2006 ) may increase the measured m / l ratio of individual ucds after a very close passage ( @xmath20pc ) next to the host galaxy s center .
however , this can certainly not explain the existence of the general and very clear trend of m / l with mass ( fig .
[ mass_ml_norm ] ) . for diagnosing a bottom heavy imf , we propose to study a portion of the spectrum where the hypothetical overabundant population of low - mass ( main sequence ) stars contributes significantly to the integrated spectrum . in this region
, the shape of the spectral energy distribution ( sed ) emitted by these unevolved dwarf stars must be different to that of the evolved giant stars .
this is the case for the near - infrared wavelength region of the co band ( @xmath21 , see fig .
[ cospectra ] ) . for intermediate to high metallicities , late type giant stars have a very strong co absorption feature , while late type dwarf stars have a much weaker feature ( see figs .
[ cospectra ] and [ coindex ] ) , independent of metallicity ( e.g. frogel et al . 1978 ) . at a given metallicity
, the strength of the co absorption feature anti - correlates with the fraction of low - mass stars present in a stellar population . one thus expects a weaker co index for high m / l ratio sources , if the high m / l ratios are caused by low - mass stars .
our working hypothesis is that the stellar population of a high m / l ratio ucd consists of a canonical part with a normal m / l and imf , _ plus _ an additional population of low mass stars .
kroupa & gilmore ( 1994 ) present a detailed analysis of such a scenario .
they quantify which luminosity fraction @xmath22 of the composite overall luminosity is contributed by this additional population of low - mass stars in the co wavelength region , which is close to the k - band . in the context of this paper this fraction can be written in terms of luminosities as @xmath23 obviously , @xmath24 for a normal stellar population without additional low - mass stars . in terms of masses and m / l ratios we re - write
@xmath25 the term @xmath26 ranges between 1.4 and 0.66 for [ fe / h ] between -2 and 0 dex , assuming a 12 gyr population ( worthey et al .
1994 ) . introducing the ratio @xmath27 ,
we get @xmath28 the ratio @xmath29 can be expressed as @xmath30 .
as @xmath31 we assume the mean normalised m / l ratio of compact stellar system less massive than 2@xmath1510@xmath16 m@xmath7 ( see fig .
[ mass_ml ] ) , which is 2.75 .
for example , we get @xmath32 for the highest @xmath33 ratio ucd ( object s999 with normalised @xmath33=17 ) .
the term @xmath34 in @xmath22 depends on the slope @xmath35 and cutoff mass @xmath36 in the mass distribution of the additional low - mass - star population . from kroupa & gilmore ( 1994 )
we find a range of @xmath37 for the parameter space @xmath38 and @xmath39 .
we can now express the co index as a function of @xmath22 , and hence as a function of @xmath33 , @xmath40 from fig .
[ cospectra ] and kroupa & gilmore ( 1994 ) we adopt @xmath41 mag .
the choice of @xmath42 requires a brief discussion .
the co index of old stellar populations with canonical imf is dominated by giant stars . since the dominant stellar type in the integrated light changes towards earlier types for lower metallicitiy , and because the co index anticorrelates with surface temperature , the co index anticorrelates with integrated metallicity ( e.g. frogel et al .
we use the linear calibration between [ fe / h ] and co by frogel et al .
( 2001 ) ( see also fig . [ coindex ] ) derived from galactic globular clusters in order to determine @xmath42 at a given [ fe / h ] . for each of the compact stellar systems from fig .
[ mass_ml_norm ] , we can now calculate the expected co index , under the hypothesis that a m / l ratio offset above the average value for galactic globular clusters is explained by an additional population of low - mass stars . for this
we assume the mean of the co indices corresponding to the two extreme values of @xmath34 from above .
the result of this exercise is shown in fig .
[ coindex ] . in the left panel
we plot [ fe / h ] against the expected co index .
we show both the photometric index ( fig . [ cospectra ] ) and the spectroscopic one ( feature band between 2.291 and 2.302@xmath43 , see frogel et al .
2001 for the exact pass - band definitions ) , using the well defined linear relation between both indices ( frogel et al .
we overplot the calibration relation between [ fe / h ] and co from frogel et al .
( 2001 ) .
there can be substantial offsets @xmath44 mag due to the higher m / l ratios , corresponding to 30% of the equivalent with of the co absorption feature .
as aid for the eye we also indicate in fig .
[ cospectra ] the offset relation for two different fractions of additional low - mass stars .
the dashed curve corresponds to a mass fraction of 55% contributed by additional low - mass stars , which is representative of the average m / l ratio offset between ucds and gcs in fig .
[ mass_ml ] .
the dotted curve indicates a mass fraction of 75% , corresponding to the most extreme observed ucds . in the right panel of fig .
[ coindex ] , we plot the expected offset from the frogel et al .
[ fe / h]@xmath45co relation vs. mass of the compact stellar systems .
the largest offsets are expected for high masses , given
the increase of m / l ratios in the ucd regime . in the next section
we assess the observational detectability of these offsets .
looking at fig .
[ coindex ] , there are three minimum requirements for detecting a drop in the co index _ in a single source _ due to a high m / l ratio .
first , one requires an accurate calibration between co and [ fe / h ] , applicable to @xmath50.1 dex precision .
second , one requires an [ fe / h ] measurement of a single source accurate to @xmath50.1 dex .
third , one requires measurement of the co index to an accuracy of [email protected] mag , or 5 - 10% in terms of equivalent width .
we note that the existing calibration between co and [ fe / h ] by frogel et al .
( 2001 ) is performed with galactic gcs , and has a cluster - to - cluster dispersion of the order @xmath50.15 dex .
it is reasonable to assume that future calibration attempts including population synthesis models ( e.g. brodie et al .
2007 ) will reach accuracy levels down to @xmath50.10 dex .
still , even with an accurate calibration , an offset in a single object measurement may be detected with very accurately measured co and [ fe / h ] only for the largest expected co offsets . the more efficient way to detect a change in co are relative measurements of statistically significant samples ( @xmath4610@xmath4520 objects ) covering a broad range in metallicities .
sources with m / l ratios consistent with canonical imfs have to be observed in parallel to ucds with high m / l ratios .
the aim is to arrive at a comprehensive sample of precise measurements of m / l , co and [ fe / h ] , obtained with identical or similar instrument settings .
this changes the requirements on [ fe / h ] from @xmath50.1 dex accuracy to @xmath50.1 to 0.2 dex relative precision .
this is well possible with moderately high s / n optical spectroscopic measurements of [ fe / h ] ( e.g. mieske et al .
2006 ) , requiring on - source integration times of a few hours at 6 - 10 m class telescopes . to cover a substantial portion of the known ucd population , only a few pointings of multi - object spectroscopy with moderately large fov ( 0.25 to 0.5 degree ) would be required .
see a map of all known fornax and virgo ucds in fig . [ map ] .
that is , within a few nights of observations , the required precision of [ fe / h ] can be reached for the known ucd population .
what about the necessary integration time for precisely measuring the co feature depth ?
in the right panel of fig .
[ cospectra ] we show a smoothed nir spectrum around the co regions , indicating the 1@xmath47 envelope for a s / n=150 per 0.04@xmath43 resolution element at 2.2@xmath43 . integrating over the feature and continuum regions , the precision of measuring the co index is about 0.01 mag .
to calculate the integration time needed to achieve this s / n , we use the exposure time calculator ( etc ) of isaac on the vlt ( version 3.2.1 ) .
we assume v=20.0 mag as object magnitude , which is at the faint end of the observed ucd magnitude range in fornax and virgo , and thus gives an upper limit on the required integration time . as input for the isaac etc
we adopt @xmath48=2.5 ( worthey 1994 for a 12 gyr population of [ fe / h]=-0.7 dex ) , hence k=17.5 mag . assuming a k - band seeing of 0.7@xmath49 , a slit - width of 1@xmath49 , a black - body curve of 4000k , and the low - resolution spectroscopy mode
, we require @xmath57 hours on - source integration time to achieve s / n=150 per 0.04@xmath43 at 2.2@xmath43 .
as comparison objects with canonical m / l ratios , one will want to choose _ bright _ objects , requiring much shorter integration times .
those could be massive gcs in more nearby galaxies , for which m / l ratios and metallicities are available ( e.g. gcs in ngc 5128 , see rejkuba et al .
2007 ) . the ideal way to measure the co index for a comprehensive sample of ucds certainly is wide - field imaging with purpose - build narrow - band ( nb ) filters covering the co feature and continuum , each of width @xmath50.1 @xmath43 ( fig . [ cospectra ] ) . in fig . [ map ]
we indicate the projected distribution of ucds in the fornax and virgo cluster relative to the respective central cluster galaxies . for comparison
we show the fov of available ir imagers on 4 - 8 m class telescopes .
this plot shows that most of the ucds in either cluster are contained within one wfcam or vista pointing .
for the above example of a ucd with k=17.5 mag , we can estimate the required amount of integration time necessary to achieve s / n=100 for a point - source image in the co continuum and feature pass - band .
for this we use the vista etc ( http://www.ast.cam.ac.uk/vdfs/etc ) and calculate the required time for s / n=100 with the ks filter , which is centred on 2.15@xmath43 with fwhm=0.3@xmath43 . for k=17.5 mag ,
the required integration time using this ks filter is about 5400 seconds , or 1.5 hours . to convert this into the time required for the same s / n with a nb co filter , two things need to be taken into account
first , the co feature bands are 3 times narrower than the ks filter .
second , the sky brightness in the co feature band ( 2.35@xmath43 ) is about 2.5 times higher than at 2.15@xmath43 ( see isaac etc ) .
taking these two differences into account , one requires on - source integration of 5.6 hours for the co continuum band and 27 hours for the co feature band .
we conclude that measuring the depth of the co feature ( 2.35@xmath43 ) in the near - ir sed of old extragalactic systems is a promising tool to constrain the low - mass slope of the imf , provided that precise metallicity estimates ( @xmath50.15 dex ) are available . for the case of ucds ,
such observations offer an important test as to whether their high m / l ratios are caused by very bottom heavy imfs . given that an alternative explanation for these high m / l ratios is dark matter at very high densities ,
the study of the imf in ucds will allow important constraints on the phase space properties of dark matter particles . to demonstrate the feasibility of the method proposed here , the first step is to obtain low resolution spectroscopy of the ucds with largest expected drop in co , and reference sources of comparable metallicity with normal m / l ratios .
as a next step , wide - field nir imaging with appropriate narrow - band filters will allow to probe large representative samples of ucds within a few nights of telescope time . with such observations it will also be possible to investigate spatial variations in the imf in early - type giant and dwarf galaxies ,
provided that spatially resolved information on their [ fe / h ] abundances are available from ifu or long - slit spectroscopy . in the context of the tidal stripping scenario ,
an important aspect in that context will be to compare the co index in nuclear regions of dwarf galaxies with those in ucds .
barmby , p. , huchra , j. p. , brodie , j. p. et al . 2000 , aj , 119 , 727 bassino , l. p. , muzzio , j. c. , & rabolli , m. 1994 , apj , 431 , 634 bekki , k. , couch , w.j . ,
drinkwater , m.j . , shioya , y. , 2003 , mnras , 344 , 399 brodie , j. p. , cenarro , a. j. , beasley , m. , strader , j. , & cardiel , n. 2007 , noao proposal i d # 2007a-0203 bruzual , g. , & charlot , s. 2003 , mnras , 344 , 1000 dabringhausen , j. , hilker , m. , & kroupa , p. 2008 , submitted to mnras drinkwater m.j . , jones j.b . , gregg m.d .
, phillipps s. , 2000 , pasa 17 , 227 drinkwater , m.j . ,
gregg , m.d . ,
hilker , m. _ et al . _ , 2003 ,
nature , 423 , 519 drinkwater m.j .
, gregg m.d . , & couch w.j .
_ et al .
_ 2004 , pasa , 21 , 375 evstigneeva , e. a. , gregg , m. d. , drinkwater , m. j. , hilker , m. aj , 133 , 1722 fellhauer , m. , & kroupa , p. 2002 , mnras , 330 , 642 fellhauer , m. , & kroupa , p. 2005 , mnras , 359 , 223 fellhauer , m. , & kroupa , p. 2006 , mnras , 367 , 1577 frogel , j. a. , persson , s. e. , matthews , k. , & aaronson , m. 1978 , apj , 220 , 75 frogel , j. a. , stephens , a. , ramirez , s. , & depoy , d. l. 2001 , aj , 122 , 1896 gilmore , g. , wilkinson , m. , i. ; wyse , r. f. g. , kleyna , j. t. , koch , a. , evans , n. w. , grebel , e. k. 2007 , apj , 663 , 948 goldader , j. d. , joseph , r. d. , doyon , r. , & sanders , d. b. 1997 , apjs , 108 , 449 haegan , m. , jordn , a. , ct , p. _ et al .
_ ( vcs team ) 2005 , apj , 627 , 203 hilker , m. , infante , l. , vieira , g. , kissler - patig , m. , & richtler , t. 1999 , a&as , 134 , 75 hilker , m. , baumgardt , h. , infante , l. , drinkwater , m. , evstigneeva , e. , & gregg , m. 2007 , a&a , 463 , 119 ivanov , v. d. , rieke , g. h. , groppi , c. e. , alonso - herrero , a. , rieke , m. j. , & engelbracht , c. w. 2000 , apj , 545 , 190 jimenez , r. & haiman , z. 2006 , nature , 440 , 501 kroupa , p. , & gilmore , g. f. 1994 , mnras , 269 , 655 kroupa , p. , 2001 ,
mnras , 322 , 231 kroupa , p , , & bouvier , j. 2003 , mnras , 346 , 369 lancon , a. , & rocca - volmerange , b. 1992 , a&as , 96 , 593 maraston , c. 2005 , mnras , 362 , 799 mclaughlin , d. e. , van der marel , r. p. 2005 , apjs , 161 , 304 metz , m. , & kroupa , p. 2007 , mnras , 376 , 387 mieske , s. , hilker , m. , infante , l. 2004 , a&a , 418 , 445 mieske , s. , hilker , m. , infante , l. , & jordn , a. 2006 , aj , 131 , 2442 mieske , s. , hilker , m. , jordn , a. , infante , l. , & kissler - patig m. 2007 , a&a , 472 , 111 phillipps s. , drinkwater m.j . , gregg m.d . , jones j.b . , 2001 , apj 560 , 201 rejkuba , m. , dubath , p. , minniti , d. , & meylan , g. 2007 , a&a , 469 , 147 tornatore , l. , ferrara , a. , & schneider , r. 2007 , mnras in press , arxiv:0707.1433 walker , m. et al .
2007 , apj , 667l , 53 worthey , g. 1994 , apjs , 95 , 107 zinnecker , h. , keable , c. j. , dunlop , j. s. , cannon , r. d. , & griffiths , w. k. , 1988 , proceedings of the 126th symposium of the international astronomical union , edited by jonathan e. grindlay and a. g. davis philip .
kluwer academic publishers , dordrecht , 1988 . | a new type of compact stellar systems , labelled `` ultra - compact dwarf galaxies '' ( ucds ) , was discovered in the last decade .
recent studies show that their dynamical mass - to - light ratios ( m / l ) tend to be too high to be explained by canonical stellar populations , being on average about twice as large as those of galactic globular clusters of comparable metallicity .
if this offset is caused by dark matter in ucds , it would imply dark matter densities as expected for the centers of cuspy dark matter halos , incompatible with cored dark matter profiles . investigating
the nature of the high m / l ratios in ucds therefore offers important constraints on the phase space properties of dark matter particles .
here we describe an observational method to test whether a bottom - heavy imf may cause the high m / l ratios of ucds .
we propose to use the co index at 2.3@xmath0 which is sensitive to the presence of low - mass stars to test for a bottom heavy imf . in the case that the high m / l ratios are caused by a bottom - heavy imf , we show that the equivalent width of the co index will be up to 30% weaker in ucds compared to sources with similar metallicity that have canonical imfs .
we find that these effects are well detectable with current astronomical facilities in a reasonable amount of time ( a few hours to nights ) .
measuring the co index of ucds hence appears a promising tool to investigate the origin of their high m / l ratios . |
to date , seven papers have been published providing listings and finding charts for faint uv - bright sources identified on plates from the sandage two - color survey of the galactic plane .
the most recently published paper , paper vii ( @xcite ) , included proper motions for many of the sources found using information extracted from the usno - b1.0 catalog @xcite .
the combination of the magnitude , color and proper motion values available for these sources provides important additional data that may lead to the detection of previously unidentified white dwarfs .
the usno - b1.0 catalogue was not available prior to publication of paper vii .
therefore , it was considered worthwhile to go back and examine all previously identified lanning uv - bright sources in hopes of finding additional evidence to identify the best candidates for new white dwarfs in the published catalogues . as noted in the previous papers devoted to this survey project , the sandage two - color survey consists of more than 100 plates measuring 6.6 degrees on a side and centered on the galactic plane .
plates were double - exposed with one image taken in the uv and the second image in the blue separated by 12 arcseconds . while the original goal of the sandage project was to identify the optical counterparts to x - ray sources found by the
x - ray satellite , the uv - bright sources found during this analysis do not necessarily correspond to previously discovered x - ray sources .
they do represent a comprehensive survey of the entire 43 square degree field on each schmidt plate examined .
the usno - b1.0 catalog contains data obtained from scans of 7435 schmidt plates taken for various surveys during a 50 year timespan @xcite .
proper motions tabulated in the catalog were determined from observations at all available epochs on the scanned plates .
the usno - b1.0 also provides optical blue magnitudes from the iiiaj plates ( b@xmath3 ) , optical red magnitudes from iiiaf plates ( r@xmath4 ) , and infrared magnitudes from the ivn plates ( i@xmath5 ) .
the usno - b1.0 catalog was searched for possible high proper motion counterparts to the lanning sources .
objects with a recorded usno - b1.0 proper motion larger than 10 mas yr@xmath1 and located within 10@xmath6 of the lanning sources were selected as possible counterparts .
the positions of the proper motion objects were extrapolated back to the mean epoch of the sandage two - color survey plates ( circa 1970 ) , prior to the match .
a total of 213 possible high proper motion counterparts were found .
figure 1 shows the difference in right ascension and declination between these and their corresponding lanning sources .
all possible high proper motion usnob-1.0 counterparts are found to be within 5@xmath6 of the quoted positions of the lanning sources , with 194 found within only 2@xmath6 .
this excellent agreement is conclusive evidence that the usno - b1.0 proper motion stars are true counterparts to the lanning sources .
the uv sources thus found to exhibit a large proper motion are listed in table i. source numbers in the table follow the convention established by the simbad database for paper i and have been sorted in numerical order .
the format of table i consists of 1 ) the lanning source number , 2 ) the sandage plate i d ( plate center in galactic coords . ) , 3 ) right ascension ( equinox j2000 ) , 4 ) declination ( equinox j2000 ) , 5 ) estimated photographic blue magnitude , 6 ) estimated u - b color difference , 7 ) proper motion in right ascension @xmath7ra , 8) proper motion in declination @xmath7decl , 9 ) photographic blue ( b@xmath8 ) magnitude from usno - b1.0 , 10 ) photographic red ( r@xmath9 ) magnitude from usno - b1.0 , 11 ) photographic infrared ( i@xmath10 ) magnitude from usno - b1.0 , 12 ) the spectral type or estimated spectral class see 3 below , and 13 ) general notes , including identifications with tycho-2 stars or previously known objects .
note that for 15 of the sources , the usno - b1.0 does not report a value for the i@xmath10 magnitude . for those stars
, we estimated a rough magnitude from a direct visual inspection of digitized sky survey scans .
entries for right ascension , declination , @xmath11 magnitude and u@xmath12b color are taken from the original papers in the series .
corrections to coordinate errors found following publication have been incorporated here and on the web site devoted to the survey ( see below ) .
all positions were measured directly from images extracted from the gss plates at stsci .. estimates for the u@xmath12b color for paper i @xcite sources ( nos . 1 - 82 )
were only given in three ranges , strong sources were bluer than @xmath120.8 , moderate sources were between @xmath120.5 and @xmath120.8 and marginal sources were between 0.0 and @xmath120.5 and are represented as such in the table .
many of the sources listed in the table were referenced as tycho-2 entries in the usno - b1.0 catalog ; their tycho-2 catalog numbers are noted in the comments . in each case , the proper motion values quoted in columns 7 - 8 are from the tycho-2 catalog @xcite instead of from usno - b1.0 .
proper motions in the tycho-2 catalog were determined based upon tycho-2 observations at a mean epoch of 1991.25 and observations from 143 transit circle and photographic catalogues between 1905 and 1991.25 , and are more accurate than the usno - b1.0 values .
this uv source is the star sao 161359 , an o8 - 9v star which is a probable member of the galactic star forming region m17 @xcite .
we found its usno - b1.0 counterpart to be listed with a proper motion vector @xmath13 mas yr@xmath1 .
the star is also listed in the tycho-2 catalog ( catalog number tyc 6265 - 1440 - 1 ) as having a proper motion @xmath14 mas yr@xmath1 .
these values are , however , inconsistent with an m17 membership .
the star forming region is at an estimated distance of d=@xmath15 kpc @xcite ; at such a distance , a star with a proper motion of 31 mas yr@xmath1 would have a transverse velocity @xmath16 km s@xmath1 which is inconsistent with a young cluster membership .
the large , quoted proper motion is most likely in error .
indeed , we find that the star is listed in the tycho reference catalog @xcite with a much smaller proper motion @xmath17 mas yr@xmath1 , a value which is much more reasonable . we have thus not included this star in our final list of high proper motion uv sources .
this uv - bright source is one component in a common proper motion pair with a 7@xmath6 angular separation .
the lanning source is the west component of the pair .
the single object listed in the usno - b1.0 catalog appears to be a compound of the two stars . while the two components have about the same brightness on the b@xmath8 plate , the lanning source is significantly fainter on the i@xmath10 plate .
the magnitudes quoted in table 1 were estimated visually from digitized sky survey scans .
the pair is most likely a main sequence + white dwarf binary , with lanning 140 being the white dwarf member .
the quoted usnob-1.0 proper motion for this source is @xmath18=@xmath1230 mas yr@xmath1 , @xmath19=+20 mas yr@xmath1 .
however , an examination of digitized sky survey scans shows that this proper motion is most certainly wrong , as the star is clearly seen moving in the southeast direction , and with a much larger proper motion than quoted in usno - b1.0 .
a more accurate proper motion measurement performed with the superblink software @xcite indicates a proper motion @xmath18=+113 mas yr@xmath1 , @xmath19=@xmath1266 mas yr@xmath1 , which we adopt in place of the quoted usnob-1.0 value .
many of the the lanning sources , with their large uv excess , are blue stars with large effective temperatures .
stars with large proper motions tend to be relatively nearby , which means that any relatively faint lanning source with a significant proper motion is most likely to be a white dwarf .
brighter lanning sources with moderate proper motions , on the other hand , are most likely to be more distant main sequence stars .
it is possible to identify the sources which are most likely to be white dwarfs with the use of a reduced proper motion diagram .
we use the photographic magnitudes of the usno - b1.0 counterparts ( b@xmath3 , r@xmath4 , i@xmath10 ) and their recorded proper motion ( @xmath7 ) , to build a diagram of the reduced proper motion h@xmath20=r@xmath4 + 5@xmath21 + 5 versus the b@xmath22i@xmath10 color .
the resulting diagram is shown in figure 2 .
the reduced proper motion diagram is analogous to the standard color magnitude diagram in that white dwarfs tend to populate the lower left side of the plot , corresponding to a large h@xmath20 and a blue b@xmath22i@xmath10 color ; the main sequence stars generally populate a region above and to the right of the white dwarfs , and subdwarf stars are generally found in between @xcite .
this pattern occurs because h@xmath20 is correlated with the absolute magnitude of the source .
the location in the reduced proper motion diagram of sources with known spectral types clearly demonstrates the separation between main sequence stars and white dwarfs .
we have searched for counterparts of the lanning sources in the catalog of spectroscopically identified white dwarfs @xcite and found 16 matches .
these are shown as open triangles in figure 2 .
we also searched the simbad database for sources with published spectral types ; we found 8 stars with main sequence spectral types from b to g5 ; these are denoted with open square symbols in fig.2 .
we also found two stars formally classified as hot subdwarfs ( sdo ) ; these are plotted as inverted open triangles .
the spectral types of these objects are also noted in table 1 .
we separate the reduced proper motion diagram into 3 regions which correpond to an increasing probability for a source to be a white dwarf .
stars falling in each group are assigned a spectral class corresponding to the region where they fall .
the first region contains all stars for which : @xmath23 and which are most probably main sequence stars . to these objects we assign a spectral class `` ms '' .
the second region includes all objects for which : @xmath24 and which probably contain a mix of white dwarfs and subdwarfs , with perhaps some main sequence stars as well ; these stars are assigned a spectral class `` wd / sd '' .
finally , the third region includes all stars with : @xmath25 which are the objects most likely to be white dwarfs , and to which we assign a spectral class `` wd '' .
the tentative spectral classes are given in table 1 for all stars which do not have formal spectral classification at this time .
these are meant to be used as guides to future spectroscopic follow - up surveys .
we have determined that 213 uv - bright , lanning sources have counterparts with significant proper motions , as recorded in the usno - b1.0 catalog . for a majority of the sources , in particular the faint blue objects
, the large proper motion strongly suggests that they are white dwarfs .
the location of the lanning sources in the reduced proper motion diagram confirms this impression , and indicates which objects are most likely to be white dwarfs .
spectroscopic follow - up of these sources should uncover a significant number of formerly uncatalogued white dwarfs in the solar vicinity . the assistance of dr .
stephen levine , united states naval observatory , flagstaff , in extracting the proper motion information via batch processing was most appreciated .
the original photographic survey was supported in part by the national aeronautics and space administration under grant ngr 09 - 140 - 009 .
this continuing research project has been supported by funding from the stsci , which is operated by the association of universities for research in astronomy , inc . under nasa contract nas5 - 26555 and by the nasa adp contract po # s-92513-z .
sl is supported by the national science fundation under nsf grant # ast 0607757 .
sl also acknowledges support from mr .
hilary lipsitz , and from the american museum of natural history .
this research has made use of the usnofs image and catalogue archive operated by the united states naval observatory , flagstaff station .
this research has also made use of the simbad reference database and of the vizier catalogue service , both maintained by the centre de donnes astronomiques de strasbourg .
chini , r. _ et al .
_ 2003 , asp conference series , vol .
287 , p. 415
hanson , m. m. , & conti , p. s. 1995 , , 448 , l45 hg , e. , _ et al .
_ 1998 , , 335 , l65 hg , e. , _ et al .
_ 2000 , , 355 , l27 lanning , h. h. 1973 , , 85 , 70 lanning , h. h. , and meakes , m. 2004 , , 116 , 1039 ( paper vii ) lpine , s. , rich , r. m. & shara , m. m. 2005 , , 125 , 1598 lpine , s. & shara , m. m. 2005 , , 129 , 1483 mccook , g. p. , & sion , e. m. 2003 , vizier online data catalog , 3235 , 1 monet , david g. , _ et al .
_ 2003 , , 125 , 984 salim , s. & gould , a. 2002 , , 575 , l83 llrrrrrrrrrrl 004 & 123 + 6 & 01 26 08.8 & 69 01 55.3 & 16 & @xmath26 - 0.8 & -22 & 8 & 15.9 & 16.3 & 16.5 & wd & + 005 & 134 + 0 & 02 13 23.0 & 62 04 43.6 & 20 : & @xmath26 - 0.8 & 22 & -48 & 16.7 & 16.6 & 17.0 & wd & + 012 & 193 + 0 & 06 09 08.6 & 16 42 16.6 & 19 : & @xmath26 - 0.8 & 2 & -10 & 17.4 & 17.8 & 17.4 & wd / sd & + 013 & 193 + 0 & 06 19 28.9 & 18 38 01.2 & 19 : & @xmath26 - 0.8 & -4 & -42 & 17.2 & 16.9 & 17 : & wd & + 018 & 33.2 + 0 & 18 47 39.1 & 01 57 39.1 & 12 & @xmath26 - 0.8 & -2 & -74 & 12.9 & 13.1 & 12.3 & da2 & + 019 & 53 + 0 & 19 33 49.9 & 18 52 03.1 & 15.5 & @xmath26 - 0.8 & 10 & -10 & 15.4 & 15.7 & 15.8 & wd / sd & + 020 & 53 + 0 & 19 43 31.2 & 18 24 35.2 & 12 & @xmath26 - 0.8 & 3.3 & -43.8 & 12.1 & 12.5 & 12.7 & sdo & tyc 1606 - 367 - 1,als 10448 + 021 & 60.1 + 0 & 19 44 59.1 & 22 45 49.2 & 16 & @xmath26 - 0.8 & -8 & -16 & 15.4 & 16.2 & 17.4 & wd & + 023 & 111.1 + 0 & 22 49 52.4 & 58 34 31.9 & 14 & @xmath26 - 0.8 & 8 & -14 & 14.3 & 14.6 & 14.8 & da1 & + 024 & 123 + 6 & 01 24 05.1 & 69 13 20.3 & 19 : & -0.5 to -0.8 & 94 & -10 & 17.0 & 17.2 & 16.3 & wd & + 029 & 148 - 6 & 03 17 51.6 & 53 04 20.0 & 17 & -0.5 to -0.8 & -14 & -16 & 17.4 & 17.6 & 17.5 & wd & + 032 & 172 + 0 & 05 20 00.1 & 36 34 06.7 & 15.7 & -0.5 to -0.8 & -10 & -8 & 15.4 & 15.6 & 15.6 & da & wd 0516 + 365 + 034 & 193 + 0 & 06 08 23.9 & 20 30 11.1 & 18 : & -0.5 to -0.8 & 10 & -6 & 17.6 & 17.1 & 17.5 & wd / sd & + 036 & 193 + 0 & 06 18 45.6 & 15 16 52.2 & 13 & -0.5 to -0.8 & -32 & 22 & 12.7 & 13.1 & 13.3 & wd / sd & + 048 & 53.4 - 6 & 20 03 17.4 & 16 18 44.4 & 12 & -0.5 to -0.8 & -2.2 & -18.3 & 12.5 & 12.6 & 12.7 & ms & tyc 1617 - 1006 - 1 + 050 & 53.4 - 6 & 20 07 36.5 & 17 42 15.4 & 15.3 & -0.5 to -0.8 & 102 & -10 & 15.2 & 15.0 & 14.9 & dape & wz sge + 051 & 96.1 + 0 & 21 18 56.7 & 54 12 34.3 & 12.5 & -0.5 to -0.8 & -85.4 & 195.2 & 12.6 & 12.9 & 13.2 & da3.5 & tyc 3953 - 480 - 1 + 057 & 53.4 - 6 & 03 10 03.2 & 53 09 24.5 & 17 & 0.0 to -0.5 & 28 & -70 & 16.6 & 16.2 & 17.2 & wd & + 062 & 53.4 - 6 & 03 31 18.1 & 53 03 52.0 & 17 & 0.0 to -0.5 & -18 & -2 & 16.8 & 17.0 & 17.0 & wd / sd & + 079 & 18 - 3 & 18 28 27.7 & -17 33 01.3 & 11 & 0.0 to -0.5 & 12.7 & -4.5 & 11.7 & 11.6 & 11.5 & ms & tyc 6270 - 1710 - 1 + 082 & 96.1 + 0 & 21 26 24.9 & 55 13 29.3 & 14 & 0.0 to -0.5 & 268 & 192 & 15.7 & 15.0 & 14.8 & da4 & + 083 & 88.0 + 6 & 20 11 15.0 & 49 10 37.0 & 20.5 & @xmath121.3 & 150 & 86 & 17.6 & 17.3 & 17.8 & wd & + 088 & 88.0 + 6 & 20 17 21.9 & 45 52 50.5 & 20.5 & @xmath121.2 & 18 & 22 & 16.6 & 17.4 & 18.2 & wd & + 089 & 88.0 + 6 & 20 17 55.0 & 47 51 43.5 & 20.5 & @xmath121.4 & 10 & 34 & 16.2 & 17.5 & 17.9 & wd & + 091 & 88.0 + 6 & 20 24 35.2 & 45 05 28.7 & 18.5 & @xmath121.0 & 16 & 2 & 16.7 & 17.5 & 18.1 & wd & + 094 & 88.0 + 6 & 20 32 28.9 & 48 01 48.1 & 19 & @xmath120.6 & -40 & -38 & 17.2 & 16.5 & 16 : & wd / sd & + 097 & 96 + 6 & 20 40 45.1 & 53 03 46.1 & 20.5 & @xmath120.7 & 10 & 18 & 19.0 & 18.6 & 18.0 & wd / sd & + 098 & 96 + 6 & 20 42 20.0 & 51 42 14.4 & 20.5 & -0.5 & 10 & 4 & 17.7 & 17.2 & 17.0 & ms & + 100 & 96 + 6 & 20 43 51.7 & 53 13 25.8 & 21.5 & @xmath120.9 & -14 & -24 & 19.4 & 18.6 & 18 : & wd & + 101 & 88.0 + 6 & 20 42 51.2 & 47 09 06.8 & 20.5 & @xmath121.0 & 6 & -36 & 17.5 & 17.0 & 17.4 & wd & + 108 & 96 + 6 & 20 54 12.3 & 55 04 09.1 & 20 & @xmath120.5 & 12 & -6 & 18.7 & 18.5 & 18.5 & wd / sd & + 109 & 96 + 6 & 20 55 21.0 & 50 14 07.4 & 21 & @xmath121.0 & -8 & 22 & 18.2 & 18.0 & 17.8 & wd / sd & + 112 & 96 + 6 & 21 00 31.4 & 50 51 19.2 & 15.5 & @xmath120.4 & -54 & -100 & 16.0 & 15.0 & 14.9 & da5 & + 113 & 96 + 6 & 21 02 46.4 & 55 30 18.5 & 20.5 & @xmath121.4 & 20 & 10 & 18.8 & 18.3 & 18 : & wd & + 114 & 96 + 6 & 21 01 54.5 & 51 38 56.6 & 21 & @xmath120.6 & -36 & 6 & 17.9 & 18.2 & 17.9 & wd & + 115 & 96 + 6 & 21 04 35.3 & 55 57 39.4 & 21 & @xmath121.2 & -14 & -40 & 18.4 & 17.1 & 17.4 & wd / sd & + 116 & 96 + 6 & 21 04 48.8 & 55 35 15.9 & 8 & @xmath120.2 & 12.6 & 22.6 & 8.2 & 7.8 & 7.7 & am & tyc 3956 - 803 - 1 , hd 201033 + 119 & 96 + 6 & 21 12 23.0 & 55 55 24.6 & 20.5 & @xmath120.5 & -24 & 10 & 16.4 & 16.6 & 17.2 & wd & + 120 & 96 + 6 & 21 14 05.7 & 49 51 17.0 & 19 & moderate & 16 & -8 & 17.0 & 17.9 & 17.6 & wd & + 122 & 96 + 6 & 21 18 56.3 & 54 12 45.4 & 13 & @xmath120.6 & -85.4 & 195.2 & 12.6 & 12.9 & 13.2 & da3.5 & tyc 3953 - 480 - 1 + 123 & 117 + 6 & 23 26 58.9 & 71 00 19.1 & 20.5 & -1.5 & -10 & 8 & 17.0 & 16.4 & 15.7 & ms & + 126 & 117 + 6 & 23 39 10.5 & 66 51 45.6 & 19.5 & @xmath120.8 & 44 & -8 & 16.8 & 17.3 & 18.2 & wd & + 127 & 117 + 6 & 23 39 45.5 & 67 46 54.9 & 20.5 & @xmath121.0 & 16 & -2 & 22.0 & 18.1 & 19.0 & ms & + 128 & 117 + 6 & 23 47 07.2 & 65 34 18.7 & 20 & -0.6 & 10 & 6 & 18.6 & 16.9 & 16.0 & ms & + 131 & 117 + 6 & 00 01 50.6 & 66 07 01.4 & 21.5 & @xmath120.3 & -26 & -2 & 18.6 & 18.0 & 18.7 & wd & + 133 & 117 + 6 & 00 12 24.8 & 65 28 53.6 & 21 & @xmath120.6 & -36 & -28 & 17.6 & 18.1 & 18.4 & wd & + 136 & 134 + 6 & 02 01 31.7 & 69 42 52.6 & 17.8 & @xmath120.8 & -14 & 4 & 18.0 & 18.4 & 19.0 & wd & + 138 & 134 + 6 & 02 31 45.1 & 69 23 32.0 & 11.0 & @xmath120.3 & 1.3 & -25.9 & 11.1 & 10.6 & 10.5 & ms & tyc 4316 - 2117 - 1 + 140 & 134 + 6 & 02 23 57.1 & 68 49 08.7 & 16.0 & @xmath120.3 & -44 & -60 & 15.5 & 16.0 & 16.0 & wd & + 142 & 134 + 6 & 02 28 50.0 & 68 35 37.9 & 17.2 & @xmath120.7 & 14 & -50 & 17.6 & 18.0 & 18.3 & wd & + 143 & 134 + 6 & 03 03 27.8 & 68 29 54.1 & 16.5 & @xmath120.3 & 50 & -18 & 16.8 & 16.8 & 17.4 & wd & + 144 & 134 + 6 & 02 30 35.7 & 68 29 36.5 & 16.7 & @xmath120.5 & 18 & -12 & 18.3 & 18.5 & 17.8 & wd / sd & + 146 & 134 + 6 & 02 38 07.8 & 66 57 36.9 & 17.8 & @xmath120.5 & -6 & -16 & 19.4 & 18.7 & 18.7 & wd / sd & + 147 & 134 + 6 & 02 05 37.9 & 65 59 36.0 & 18.0 & @xmath120.3 & 14 & 24 & 20.1 & 18.9 & 18.9 & wd / sd & + 148 & 134 + 6 & 02 59 20.1 & 65 05 45.1 & 10.5 & @xmath120.2 & 59.2 & 9.2 & 11.1 & 10.5 & 10.3 & ms & tyc 4056 - 913 - 1 + 150 & 134 + 6 & 02 15 34.6 & 64 53 22.2 & 16.0 & @xmath120.5 & 44 & -68 & 17.8 & 17.0 & 17.5 & wd & + 151 & 134 + 6 & 03 00 56.7 & 64 10 54.3 & 17.5 & @xmath120.6 & -14 & -14 & 18.0 & 18.9 & 18 : & wd & + 153 & 134 + 6 & 02 29 19.2 & 64 03 30.3 & 16.5 & @xmath120.4 & -10 & -14 & 17.7 & 16.9 & 16.4 & ms & + 155 & 128 + 6 & 01 18 55.7 & 71 20 11.4 & 18 : & -0.3 & 10 & 2 & 17.1 & 16.2 & 16.0 & ms & + 157 & 128 + 6 & 02 18 21.4 & 70 47 48.1 & 18.0 & @xmath120.8 & 6 & -12 & 17.8 & 17.6 & 18.8 & wd & + 158 & 128 + 6 & 02 19 06.3 & 70 08 39.8 & 16.5 & @xmath121.0 & 36 & -40 & 16.6 & 16.6 & 17.1 & wd & + 159 & 128 + 6 & 01 55 10.9 & 69 42 41.3 & 17.0 & @xmath120.3 & -112 & -6 & 17.4 & 17.1 & 17.3 & wd & + 160 & 128 + 6 & 01 21 57.1 & 69 24 50.6 & 17.6 & @xmath120.4 & 18 & -12 & 18.1 & 18.6 & 19.0 & wd & + 161 & 128 + 6 & 01 19 41.4 & 68 51 10.8 & 18.0 & @xmath120.5 & -26 & 10 & 17.8 & 17.2 & 16.1 & ms & + 163 & 128 + 6 & 01 33 38.3 & 68 03 32.7 & 16.5 & @xmath120.5 & -34 & -18 & 17.6 & 16.8 & 16.9 & wd / sd & + 165 & 128 + 6 & 01 51 09.9 & 67 39 31.3 & 13.5 & @xmath120.6 & 136 & -6 & 14.1 & 14.6 & 15.2 & da2 & + 166 & 128 + 6 & 01 56 06.2 & 67 14 45.5 & 18 : & @xmath120.6 & -22 & -18 & 16.6 & 17.7 & 18.5 & wd & + 167 & 128 + 6 & 02 18 18.9 & 67 02 25.4 & 17.6 & @xmath120.8 & -14 & -4 & 18.0 & 18.1 & 18.5 & wd & + 169 & 128 + 6 & 01 32 42.9 & 66 35 47.3 & 16.5 & @xmath120.2 & 32 & -28 & 17.1 & 16.1 & 16.4 & wd / sd & + 171 & 128 + 6 & 02 14 52.9 & 65 57 59.0 & 17.0 & @xmath120.5 & 32 & -48 & 18.4 & 17.9 & 18.1 & wd & + 174 & 148 + 6 & 03 41 17.1 & 62 42 03.1 & 16.8 & @xmath120.7 & 2 & -18 & 17.1 & 17.2 & 17.6 & wd & + 175 & 148 + 6 & 03 59 51.1 & 62 29 01.3 & 19.0 & @xmath120.6 & 30 & -38 & 18.6 & 18.6 & 18 : & wd & + 176 & 148 + 6 & 03 56 51.0 & 62 26 20.0 & 12.5 & @xmath120.2 & 10 & -18 & 12.6 & 11.2 & 10.8 & ms & + 177 & 148 + 6 & 03 51 15.7 & 61 52 46.4 & 17.0 & @xmath120.5 & 84 & -70 & 16.8 & 17.3 & 17.6 & wd & + 183 & 148 + 6 & 04 23 08.3 & 60 02 39.8 & 11.4 & @xmath120.4 & -6.7 & -11.8 & 11.7 & 11.3 & 11.1 & ms & tyc 4065 - 34 - 1 + 184 & 148 + 6 & 04 20 00.2 & 60 05 21.8 & 11.0 & @xmath120.4 & -25.6 & -49.5 & 11.2 & 10.4 & 10.0 & ms & tyc 4065 - 2 - 1 + 188 & 148 + 6 & 04 16 02.5 & 59 44 02.4 & 17.5 & @xmath120.6 & 6 & -22 & 17.8 & 18.2 & 18.5 & wd & + 191 & 148 + 6 & 04 04 23.8 & 58 58 36.2 & 16.8 & @xmath120.3 & -14 & 8 & 17.6 & 17.5 & 17.8 & wd & + 193 & 148 + 6 & 04 22 46.1 & 58 22 11.4 & 16.5 & @xmath120.4 & 6 & -28 & 16.8 & 16.1 & 15.5 & ms & + 197 & 123 - 6 & 00
53 01.7 & 59 59 42.6 & 15.0 & @xmath120.6 & -76 & 8 & 16.4 & 16.2 & 15.8 & wd & + 209 & 123 - 6 & 00 54 23.3 & 59 27 27.8 & 12.0 & @xmath120.2 & 25.6 & -7.6 & 11.6 & 11.3 & 11.2 & ms & tyc 3680 - 887 - 1 + 212 & 123 - 6 & 00 54 33.9 & 59 16 49.3 & 12.3 & @xmath120.2 & -14.9 & -3.7 & 11.4 & 10.7 & 10.4 & ms & tyc 3680 - 539 - 1 + 225 & 123 - 6 & 00 43 10.5 & 58 14 20.8 & 18.5 & @xmath120.6 & 16 & -16 & 18.5 & 18.0 & 17.8 & wd / sd & + 226 & 123 - 6 & 01 12 23.8 & 58 07 50.5 & 10.5 & @xmath120.5 & -67.6 & -1.4 & 9.9 & 9.2 & 8.9 & g0 & tyc 3681 - 121 - 1,hd 236653 + 234 & 123 - 6 & 01 12 50.9 & 57 16 43.4 & 11.5 & @xmath120.3 & -19.7 & 0.3 & 10.3 & 9.6 & 9.3 & ms & tyc 3677 - 1423 - 1 + 239 & 123 - 6 & 01 09 17.2 & 57 07 27.6 & 12.5 & @xmath120.3 & 4.2 & -11.7 & 11.7 & 11.2 & 11.0 & ms & tyc 3677 - 442 - 1 + 241 & 123 - 6 & 01 13 07.1 & 57 05 58.3 & 11.8 & @xmath120.4 & -12.3 & -18.2 & 10.8 & 10.0 & 9.7 & ms & tyc 3677 - 1939 - 1 + 245 & 123 - 6 & 01 12 26.8 & 56 39 59.7 & 12.2 & @xmath120.3 & 49.8 & 5.1 & 11.9 & 11.2 & 10.9 & ms & tyc 3677 - 1474 - 1 + 247 & 123 - 6 & 01 10 29.1 & 56 37 40.2 & 10.8 & @xmath120.5 & 11.8 & -4.7 & 10.6 & 10.2 & 10.1 & ms & tyc 3677 - 2235 - 1 + 248 & 123 - 6 & 00 53 28.9 & 56 38 36.7 & 11.2 & @xmath120.2 & -4.3 & -12.5 & 10.7 & 10.6 & 10.6 & b & tyc 3676 - 2052 - 1 + 253 & 123 - 6 & 00 33 40.7 & 55 51 47.0 & 14.8 & @xmath120.6 & 14 & -14 & 16.6 & 16.5 & 15.6 & wd / sd & + 260 & 123 - 6 & 01 09 20.6 & 55 10 43.3 & 12.5 & @xmath120.3 & 9.4 & -15.6 & 11.3 & 10.7 & 10.4 & ms & tyc 3673 - 1035 - 1 + 272 & 141 + 0 & 02 57 45.1 & 60 34 27.4 & 14.0 & @xmath120.5 & 10 & -12 & 15.4 & 15.4 & 15.9 & wd / sd & + 273 & 141 + 0 & 03 08 18.4 & 60 35 30.1 & 17.0 & @xmath120.6 & 14 & -74 & 17.6 & 17.3 & 17.2 & wd & + 280 & 141 + 0 & 02 48 41.3 & 59 16 12.4 & 16.5 & @xmath120.7 & 44 & -118 & 16.3 & 16.1 & 16.6 & wd & + 284 & 141 + 0 & 03 04 17.9 & 58 44 05.2 & 17.0 & @xmath121.2 & -36 & -22 & 16.5 & 16.6 & 17.0 & wd & + 288 & 141 + 0 & 03 07 17.2 & 57 50 29.5 & 18.5 & @xmath120.6 & 26 & -24 & 17.4 & 17.8 & 18.3 & wd & + 290 & 141 + 0 & 03 01 09.0 & 57 39 56.4 & 19.5 & @xmath120.5 & 78 & -38 & 18.5 & 18.4 & 18.0 & wd & + 291 & 141 + 0 & 03 03 31.9 & 57 19 14.8 & 19.5 & @xmath120.8 & 12 & 0 & 17.9 & 17.0 & 16.6 & ms & + 293 & 141 + 0 & 03 09 21.8 & 57 05 21.5 & 19.5 & @xmath120.5 & 18 & -12 & 18.4 & 18.7 & 17.8 & wd / sd & + 294 & 141 + 0 & 03 21 05.6 & 56 54 43.5 & 19.0 & -0.6 & 10 & -10 & 18.6 & 18.2 & 18 : & wd & + 295 & 141 + 0 & 03 06 16.1 & 56 51 43.3 & 18.5 & @xmath120.6 & 2 & -18 & 17.7 & 18.1 & 18.0 & wd & + 296 & 141 + 0 & 03 04 13.4 & 56 15 55.9 & 19.5 & @xmath120.5 & 0 & 0 & 18.4 & 18.3 & 18 : & wd & + 297 & 141 + 0 & 03 10 05.5 & 56 03 18.8 & 19.0 & @xmath120.6 & 22 & -22 & 18.4 & 18.2 & 18.0 & wd & + 298 & 141 + 0 & 02 55 16.3 & 55 53 25.2 & 17.5 & @xmath120.4 & 20 & 46 & 17.5 & 16.8 & 16.4 & wd / sd & + 300 & 141 + 0 & 03 25 21.9 & 55 00 08.9 & 16.5 & @xmath120.5 & -8 & -90 & 16.9 & 16.7 & 17.0 & wd & + 303 & 141 + 0 & 03 00 59.1 & 54 32 42.3 & 17.0 & @xmath120.5 & 154 & -170 & 17.2 & 16.5 & 16.0 & wd & + 311 & 148 + 0 & 03 47 21.0 & 55 51 52.1 & 17.0 & @xmath120.4 & -16 & -88 & 16.6 & 17.0 & 17.2 & wd & + 313 & 148 + 0 & 04 08 11.1 & 55 50 12.8 & 12.0 & @xmath120.3 & 0.1 & 12.0 & 12.7 & 12.5 & 12.5 & ms & tyc 3722 - 650 - 1 + 316 & 148 + 0 & 03 39 21.7 & 55 25 29.6 & 18.5 & -0.6 & 10 & 0 & 18.1 & 18.2 & 18.1 & wd / sd & + 318 & 148 + 0 & 04 11 26.4 & 55 11 39.8 & 11.0 & @xmath120.2 & 16.4 & -30.1 & 11.3 & 10.6 & 10.3 & ms & tyc 3722 - 789 - 1 + 321 & 148 + 0 & 03 57 42.3 & 54 04 22.6 & 18.0 & @xmath120.5 & 18 & -48 & 16.9 & 16.9 & 17.3 & wd & + 328 & 148 + 0 & 03 36 41.2 & 53 02 30.3 & 12.0 & @xmath120.3 & 14.2 & -26.7 & 11.5 & 11.1 & 10.9 & ms & tyc 3716 - 334 - 1 + 337 & 148 + 0 & 03 39 58.3 & 52 09 53.9 & 17.5 & @xmath120.8 & 24 & 42 & 16.8 & 17.3 & 17.4 & wd & + 339 & 148 + 0 & 03 52 06.9 & 51 40 03.5 & 12.0 & @xmath120.1 & -33.4 & 39.4 & 11.1 & 10.5 & 10.2 & ms & tyc 3338 - 1104 - 1 + 351 & 148 + 0 & 04 06 07.1 & 54 31 33.9 & 14.5 & @xmath120.6 & -90 & 16 & 15.6 & 15.5 & 15.5 & wd & + 353 & 80.4 + 0 & 20 54 27.3 & 42 42 44.2 & 18.0 & -0.3 & -2 & -10 & 18.3 & 17.8 & 17.5 & wd / sd & + 356 & 80.4 + 0 & 20 24 33.0 & 41 23 37.2 & 17.0 & @xmath120.4 & 16 & 12 & 17.3 & 17.1 & 17.6 & wd & + 358 & 80.4 + 0 & 20 33 52.7 & 40 56 47.3 & 18.3 & -0.3 & 10 & 2 & 18.8 & 18.3 & 18.7 & wd / sd & + 359 & 80.4 + 0 & 20 50 29.7 & 40 20 33.5 & 12.0 & @xmath120.3 & 15.9 & -55.1 & 10.7 & 10.0 & 9.7 & ms & tyc 3171 - 452 - 1 + 360 & 80.4 + 0 & 20 38 22.6 & 40 17 34.4 & 18.5 & @xmath120.2 & 14 & 4 & 18.6 & 18.3 & 18.7 & wd & + 361 & 80.4 + 0 & 20 48 08.3 & 39 51 38.3 & 14.0 & @xmath120.9 & -36 & -34 & 14.8 & 14.5 & 14.2 & da & + 364 & 80.4 - 6 & 20 56 37.2 & 43 13 26.3 & 17.0 & @xmath120.2 & 2 & 10 & 17.8 & 16.5 & 15.9 & wd & + 366 & 80.4 - 6 & 20 54 27.0 & 42 42 47.7 & 18.0 & -0.4 & -2 & -10 & 18.3 & 17.8 & 17.5 & wd / sd & + 367 & 80.4 - 6 & 21 04 02.6 & 42 47 03.5 & 15.5 & @xmath120.3 & 32 & 22 & 16.8 & 16.7 & 16.7 & wd & + 369 & 80.4 - 6 & 21 23 12.8 & 42 01 52.4 & 10.6 & @xmath120.3 & -11.6 & -35.3 & 11.1 & 10.3 & 10.0 & ms & tyc 3190 - 1956 - 1 + 370 & 80.4 - 6 & 21 24 19.1 & 41 58 41.3 & 10.7 & @xmath120.4 & 14.3 & 3.5 & 11.1 & 10.7 & 10.5 & ms & tyc 3190 - 1682 - 1 + 371 & 80.4 - 6 & 21 23 28.8 & 41 46 33.3 & 10.5 & @xmath120.3 & 17.2 & 2.0 & 10.8 & 10.3 & 10.1 & ms & tyc 3190 - 2428 - 1 + 372 & 80.4 - 6 & 20 55 47.9 & 41 37 35.4 & 11.8 & @xmath120.3 & 16 & -14 & 12.5 & 11.4 & 10.9 & ms & + 374 & 80.4 - 6 & 21 26 24.0 & 41 14 20.0 & 10.8 & @xmath120.3 & -12.5 & -10.5 & 12.0 & 11.5 & 11.3 & ms & tyc 3186 - 1289 1 + 375 & 80.4 - 6 & 21 25 08.0 & 40 53 16.9 & 10.5 & @xmath120.2 & -0.8 & -10.4 & 10.9 & 10.4 & 10.1 & ms & tyc 318 - 1900 - 1 + 377 & 80.4 - 6 & 20 54 36.3 & 40 33 18.8 & 10.8 & @xmath120.3 & -9.0 & -10.1 & 11.5 & 11.1 & 10.9 & ms & tyc 3171 - 685 - 1 + 378 & 80.4 - 6 & 20 56 34.2 & 40 19 09.9 & 10.7 & @xmath120.3 & 27.7 & 29.6 & 11.0 & 10.3 & 10.1 & ms & tyc 3171 - 1120 - 1 + 379 & 80.4 - 6 & 21 23 31.8 & 40 16 05.9 & 10.8 & @xmath120.2 & -14.8 & -52.9 & 12.2 & 11.1 & 10.6 & ms & tyc 3186 - 2361 - 1 + 380 & 80.4 - 6 & 21 24 46.9 & 39 42 29.9 & 10.6 & @xmath120.2 & 5.4 & 10.1 & 11.3 & 10.6 & 10.3 & ms & tyc 3186 - 1551 - 1 + 390 & 80.4 - 6 & 21 11 17.7 & 38 43 34.7 & 11.0 & @xmath120.3 & 9.1 & -15.6 & 11.4 & 11.0 & 10.8 & ms & tyc 3169 - 2148 - 1 + 392 & 80.4 - 6 & 21 13 40.5 & 38 13 02.8 & 17.5 & @xmath120.6 & 12 & 10 & 17.4 & 17.5 & 17.5 & wd / sd & + 398 & 100 + 4 & 22 03 47.1 & 60 56 37.6 & 10.8 & @xmath120.1 & -0.5 & -12.3 & 11.3 & 10.5 & 10.1 & ms & tyc 4263 - 1549 - 1 + 401 & 100 + 4 & 21 16 59.3 & 60 40 30.3 & 11.6 & @xmath120.1 & 18 & 38 & 13.7 & 12.8 & 12.0 & ms & + 402 & 100 + 4 & 21 29 15.4 & 60 52 45.9 & 19.0 & @xmath120.5 & 28 & 8 & 18.4 & 17.8 & 17.8 & wd / sd & + 403 & 100 + 4 & 21 33 14.4 & 60 48 30.4 & 17.3 & @xmath120.6 & 24 & 16 & 17.7 & 17.7 & 17 : & wd / sd & + 406 & 100 + 4 & 21 24 44.9 & 60 35 17.5 & 12.4 & @xmath120.1 & 8 & 12 & 14.2 & 13.0 & 11.9 & ms & + 407 & 100 + 4 & 21 19 48.2 & 60 21 01.3 & 18.8 & @xmath120.8 & 28 & 38 & 16.9 & 16.6 & 17.3 & wd & + 413 & 100 + 4 & 22 04 38.4 & 59 59 43.0 & 17.8 & -0.7 & -10 & -10 & 17.3 & 17.4 & 16.9 & wd / sd & + 414 & 100 + 4 & 21 40 10.9 & 59 55 38.3 & 11.2 & @xmath120.1 & -13.6 & -23.5 & 11.5 & 11.0 & 10.8 & f8 & tyc 3979 - 143 - 1 , trumpler 37 1790 + 416 & 100 + 4 & 21 28 10.9 & 59 27 13.4 & 19.0 & -1.0 & -10 & -10 & 18.4 & 17.3 & 15.9 & ms & + 417 & 100 + 4 & 21 38 53.1 & 59 29 08.6 & 18.5 & @xmath120.6 & 22 & -4 & 18.5 & 18.3 & 18 : & wd & + 419 & 100 + 4 & 21 58 39.5 & 59 21 57.8 & 10.5 & @xmath120.2 & 29.8 & 7.0 & 11.3 & 10.4 & 10.1 & ms & tyc 3980 - 719 - 1 + 421 & 100 + 4 & 22 02 43.0 & 58 44 55.1 & 10.8 & @xmath120.2 & 27.2 & 18.5 & 11.9 & 11.3 & 11.0 & ms & tyc 3981 - 5 - 1 + 422 & 100 + 4 & 21 43 56.4 & 58 56 35.9 & 18.8 & @xmath120.8 & 20 & 0 & 17.7 & 17.5 & 17.8 & wd & + 423 & 100 + 4 & 21 42 14.7 & 58 54 01.7 & 16.5 & @xmath120.4 & -14 & -2 & 16.5 & 15.3 & 14.2 & ms & + 425 & 100 + 4 & 21 17 48.3 & 58 43 33.5 & 10.9 & @xmath120.2 & -34.8 & -71.8 & 11.4 & 10.9 & 10.8 & f8 & tyc 3965 - 1604 - 1 , trumpler 37 1196 + 427 & 100 + 4 & 21 21 55.9 & 58 37 47.3 & 11.1 & @xmath120.1 & 56.8 & 19.7 & 11.7 & 10.9 & 10.6 & g2 & tyc 3978 - 1304 - 1 , trumpler 37 1196 + 430 & 100 + 4 & 21 20 42.1 & 58 19 25.4 & 17.0 & @xmath120.5 & 14 & -136 & 16.5 & 15.9 & 15.8 & da6 & + 432 & 100 + 4 & 21 58 32.5 & 58 04 33.4 & 18.5 & @xmath120.6 & -6 & -70 & 17.4 & 17.2 & 16.9 & wd & + 435 & 100 + 4 & 21 33 29.0 & 58 11 17.2 & 17.5 & @xmath120.7 & 0 & 12 & 17.2 & 17.3 & 17.4 & wd / sd & + 439 & 100 + 4 & 21 28 22.9 & 57 38 04.7 & 19.6 & @xmath120.6 & 26 & 14 & 18.5 & 17.0 & 15.5 & ms & + 444 & 100 + 4 & 21 35 20.5 & 56 10 49.5 & 17.0 & @xmath120.9 & 66 & 20 & 17.6 & 16.9 & 16.7 & wd / sd & + 445 & 100 + 4 & 21 37 40.3 & 56 10 14.0 & 18.8 & @xmath120.8 & 70 & -26 & 18.5 & 17.4 & 17.1 & wd / sd & + 448 & 100 + 4 & 21 39 22.4 & 55 59 32.7 & 11.5 & @xmath120.2 & -5.0 & -12.6 & 12.2 & 12.1 & 12.0 & ms & tyc 3971 - 1001 - 1 + 450 & 100 + 4 & 22 01 22.5 & 55 34 52.4 & 11.0 & @xmath120.2 & 3.7 & -42.7 & 11.5 & 10.8 & 10.6 & ms & tyc 3973 - 1619 - 1 + 454 & 100 + 4 & 21 41 41.1 & 55 27 50.2 & 11.4 & @xmath120.5 & 102.1 & -13.9 & 10.9 & 10.1 & 9.8 & ms & tyc 3971 - 891 - 1 + 458 & 104.1 + 6 & 22 45 16.4 & 66 08 13.3 & 18.0 & @xmath120.5 & 22 & 10 & 18.8 & 19.0 & 19 : & wd & + 459 & 104.1 + 6 & 22 05 38.2 & 62 24 35.7 & 16.0 & @xmath120.8 & -38 & 10 & 17.1 & 17.5 & 17.7 & da & + 460 & 141 + 6 & 03 12 40.1 & 66 44 32.3 & 11.3 & @xmath120.1 & -4.0 & -19.4 & 11.6 & 10.9 & 10.6 & ms & tyc 4061 - 302 - 1 + 461 & 141 + 6 & 02 54 14.2 & 66 30 35.1 & 17.8 & @xmath121.1 & 80 & -48 & 17.5 & 17.6 & 17.2 & wd & + 463 & 141 + 6 & 03 27 08.8 & 66 30 59.3 & 10.4 & @xmath120.2 & 58.9 & -37.5 & 11.1 & 10.3 & 10.0 & ms & tyc 4074 - 1144 - 1 + 465 & 141 + 6 & 03 11 02.3 & 66 06 04.9 & 10.5 & @xmath120.1 & -12.6 & -2.0 & 10.6 & 10.0 & 9.8 & ms & tyc 4061 - 89 - 1 + 468 & 141 + 6 & 03 42 22.6 & 65 35 43.1 & 10.2 & @xmath120.1 & -59.5 & 3.8 & 10.9 & 10.1 & 9.7 & ms & tyc 4070 - 398 - 1 + 469 & 141 + 6 & 03 09 45.0 & 65 39 34.3 & 19.5 & @xmath121.0 & 26 & -42 & 18.1 & 18.5 & 19 : & wd & + 470 & 141 + 6 & 03 31 53.8 & 65 31 48.1 & 19.0 & @xmath120.6 & -28 & 60 & 18.0 & 18.1 & 18.0 & wd & + 472 & 141 + 6 & 03 23 53.9 & 65 08 56.8 & 20.5 & @xmath121.0 & 48 & -94 & 18.7 & 18.1 & 18.6 & wd & + 476 & 141 + 6 & 02 53 34.7 & 64 30 37.3 & 10.5 & @xmath120.1 & -16.8 & 6.8 & 10.6 & 9.9 & 9.6 & ms & tyc 4056 - 704 - 1 + 477 & 141 + 6 & 02 54 56.1 & 64 22 55.3 & 16.8 & @xmath120.7 & -14 & -20 & 16.8 & 16.6 & 16.0 & wd / sd & + 478 & 141 + 6 & 03 11 30.4 & 64 25 14.4 & 19.2 & @xmath120.4 & 6 & -58 & 18.2 & 18.5 & 18.1 & wd & + 480 & 141 + 6 & 03 29 05.7 & 64 04 42.9 & 11.3 & @xmath120.9 & 29.6 & -11.2 & 13.2 & 13.2 & 13.2 & sdo & tyc 4070
- 2419 - 1,als 7220 + 484 & 141 + 6 & 02 58 15.1 & 63 09 26.3 & 11.5 & @xmath120.4 & -29.0 & -54.7 & 11.3 & 10.6 & 10.2 & ms & tyc 4052 - 1027 - 1 + 486 & 141 + 6 & 03 24 31.3 & 62 50 53.8 & 15.3 & @xmath120.7 & -24 & 0 & 15.8 & 16.1 & 16.5 & wd & + 489 & 141 + 6 & 03 00 37.2 & 62 28 27.9 & 16.8 & @xmath120.5 & 6 & 30 & 16.3 & 16.6 & 16.9 & wd & + 491 & 141 + 6 & 03 19 15.4 & 62 05 02.3 & 18.3 & @xmath120.7 & 36 & -26 & 17.3 & 17.5 & 17.9 & wd & + 494 & 141 + 6 & 02 57 22.3 & 61 45 09.5 & 19.0 & @xmath121.0 & 18 & -18 & 18.0 & 18.3 & 18.1 & wd & + 495 & 141 + 6 & 03 09 31.9 & 61 26 20.9 & 17.8 & @xmath120.6 & 58 & -8 & 17.3 & 17.4 & 17.9 & wd & + 497 & 141 + 6 & 03 21 14.3 & 61 05 26.8 & 17.5 & @xmath120.9 & -14 & 16 & 17.1 & 16.9 & 16.5 & wd / sd & + 499 & 141 + 6 & 03 18 34.8 & 65 00 03.9 & 16.0 & @xmath120.8 & 100 & -158 & 16.4 & 16.4 & 16.8 & da3 & + 500 & 164 - 6 & 04 23 12.9 & 45 04 51.2 & 17.1 & @xmath120.5 & -16 & -96 & 17.6 & 17.5 & 17.6 & wd & + 501 & 164 - 6 & 04 15 52.6 & 45 01 29.8 & 18.5 & @xmath120.5 & 24 & -28 & 17.9 & 17.4 & 17.6 & wd & + 503 & 164 - 6 & 04 47 13.4 & 44 24 34.8 & 10.6 & @xmath120.2 & 17.4 & -21.1 & 11.0 & 10.5 & 10.3 & ms & tyc 2905 - 1267 - 1 + 513 & 164 - 6 & 04 39 22.8 & 42 48 01.8 & 19.0 & @xmath120.6 & 18 & -46 & 18.4 & 18.9 & 18.4 & wd & + 515 & 164 - 6 & 04 14 36.9 & 42 22 49.0 & 11.0 & @xmath120.4 & 12.6 & -37.4 & 10.7 & 10.1 & 9.8 & k5 & tyc 2886 - 1029 - 1,hd
276118 + 521 & 164 - 6 & 04 32 36.4 & 42 06 11.9 & 20.0 & @xmath120.5 & 8 & -22 & 18.2 & 17.5 & 16.7 & ms & + 525 & 164 - 6 & 04 45 19.3 & 41 58 40.7 & 10.5 & @xmath120.3 & 34 & -14 & 10.2 & 12.3 & 14.8 & ms & + 526 & 164 - 6 & 04 12 03.8 & 41 37 19.3 & 18.5 & @xmath120.2 & 113 & -66 & 17.4 & 16.5 & 15.8 & wd / sd & + 531 & 164 - 6 & 04 36 27.6 & 40 47 08.6 & 16.5 & @xmath120.5 & -42 & 44 & 16.9 & 16.6 & 16.5 & da6 & + 532 & 164 - 6 & 04 12 42.9 & 40 41 26.5 & 16.0 & @xmath120.7 & 14 & -44 & 16.4 & 16.9 & 17.0 & wd & + 533 & 164 - 6 & 04 31 15.6 & 41 18 11.9 & 13.0 & @xmath120.6 & 62 & -32 & 18.0 & 18.1 & 18.1 & wd & + 534 & 164 - 6 & 04 19 51.4 & 40 22 46.8 & 15.5 & @xmath120.7 & -24 & -24 & 16.2 & 16.4 & 16.6 & da & + 535 & 164 - 6 & 04 22 59.8 & 40 16 18.5 & 20.0 & @xmath120.7 & 12 & -34 & 19.4 & 19.6 & 19 : & wd & + 536 & 164 - 6 & 04 26 28.1 & 40 13 54.2 & 19.0 & @xmath120.6 & 22 & -22 & 18.8 & 19.0 & 18.5 & wd & + 537 & 164 - 6 & 04 42 10.0 & 39 59 03.8 & 11.0 & @xmath120.3 & -22.5 & 0.6 & 11.8 & 10.3 & 9.6 & g5 & tyc 2897 - 262 - 1,hd 276727 + 540 & 164 - 6 & 04 16 39.2 & 40 02 33.9 & 17.0 & @xmath121.0 & -72 & 4 & 16.3 & 15.5 & 16.4 & wd & + 541 & 164 - 6 & 04 13 08.9 & 38 56 21.2 & 19.5 & @xmath120.6 & 8 & -16 & 19.0 & 19.2 & 19 : & wd & + 542 & 164 - 6 & 04 38 39.3 & 41 09 34.4 & 15.0 & @xmath120.6 & -8 & -102 & 14.9 & 15.1 & 14.9 & db3 & + 545 & 164 + 0 & 04 57 38.6 & 45 14 59.1 & 11.0 & @xmath120.5 & 14.0 & -19.5 & 11.0 & 10.3 & 10.0 & ms & tyc 3344 - 1396 - 1 + 548 & 164 + 0 & 05 07 43.8 & 45 14 05.6 & 10.5 & @xmath120.5 & 1.8 & -13.5 & 11.8 & 11.2 & 11.0 & ms & tyc 3345 - 2493 - 1 + 554 & 164 + 0 & 04 46 47.3 & 42 15 00.7 & 19.0 & @xmath120.5 & 14 & -14 & 18.4 & 19.1 & 18.2 & wd & + 556 & 164 + 0 & 05 13 07.7 & 41 48 50.7 & 19.5 & @xmath120.5 & 16 & -24 & 18.6 & 18.6 & 18.6 & wd & + 558 & 164 + 0 & 04 48 43.7 & 41 21 43.9 & 18.5 & @xmath120.6 & 14 & -28 & 18.3 & 18.7 & 18.8 & wd & + 562 & 164 + 0 & 05 00 46.1 & 40 34 13.5 & 18.5 & -0.3 & 2 & -10 & 17.8 & 16.7 & 16.1 & ms & + 563 & 164 + 0 & 04 46 12.2 & 40 23 40.2 & 19.0 & -0.3 & 8 & -10 & 17.3 & 16.0 & 15.6 & ms & + 567 & 164 + 0 & 05 15 00.9 & 39 44 11.0 & 19.0 & @xmath120.6 & 20 & -30 & 18.7 & 18.8 & 19 : & wd & + 568 & 164 + 0 & 05 16 58.5 & 39 34 42.3 & 18.8 & @xmath120.6 & -28 & 38 & 18.0 & 18.4 & 18.3 & wd & + 570 & 164 + 0 & 04 55 04.5 & 39 29 29.8 & 18.5 & @xmath120.6 & 18 & 4 & 18.0 & 18.0 & 18.4 & wd & + 572 & 164 + 0 & 04 44 50.1 & 39 15 19.7 & 17.0 & @xmath120.5 & -26 & -44 & 17.1 & 17.5 & 17.5 & wd & + | proper motions with values @xmath0 mas yr@xmath1 or @xmath2 mas yr@xmath1 have been extracted from the usno - b1.0 and tycho ii catalogues for all lanning uv - bright sources identified in the sandage two - color survey of the galactic plane and presented in papers i - vi . of the 572 sources examined , we find at least 213 which exhibit a significantly large proper motion . based on the location of the sources in a reduced proper motion diagram , we demonstrate that about two thirds of the high proper motion sources are likely or very likely to be heretofore unidentified white dwarfs . |
a ternary algebra or triple system is a vector space @xmath1 endowed with a ternary law of composition @xmath2 which is a linear mapping with respect to each its argument , and we will call this mapping a ternary multiplication or triple product of a ternary algebra @xmath1 .
hence a ternary algebra is an algebra which closes under a suitable triple product .
obviously any binary algebra which closes under double product can be considered as a ternary algebra if one defines the ternary multiplication as twice successively applied binary one , and in this case the ternary multiplication is generated by a binary one .
however there are ternary multiplications which can not be obtained as twice successively applied binary multiplication .
for instance , pure imaginary numbers or elements of grading one of a superalgebra closes under triple product .
a well known example of a ternary matrix algebra is the vector space @xmath3 of @xmath4 matrices endowed with the ternary multiplication @xmath5 , where @xmath6 and @xmath7 is transpose of the matrix @xmath8 . since lie algebras play a fundamental role in physics , particular attention
was given to ternary algebras when they were shown to be building blocks of ordinary lie algebras .
given ternary algebra one can construct a lie algebra by using the method proposed by kantor in @xcite .
this method was extended to super lie algebras in @xcite and later was applied by the same authors in @xcite to construct a gauge field theory by introducing fundamental fields associated with the elements of a ternary algebra .
a skew - symmetric bilinear form is an important component in the large class of algebraic structures such as lie algebras , grassmann algebras and clifford algebras .
for example , the lie brackets @xmath9:{\frak l}\times{\frak l}\to { \frak l}$ ] of a lie algebra @xmath10 is the skew - symmetric bilinear form , and the multiplication @xmath11 of a grassmann algebra @xmath12 restricted to the subspace of odd elements is the skew - symmetric bilinear form .
a skew - symmetry of a bilinear form can be interpreted by means of the faithful representation of the symmetric group @xmath13 , where @xmath14 is the identity permutation , as follows : a bilinear form @xmath15 is skew - symmetric if @xmath16 .
making use of this interpretation we can construct a ternary analog of a skew - symmetric bilinear form replacing @xmath17 by @xmath18 with its faithful representation by cubic roots of unity @xmath19 , i.e. @xmath20 , where @xmath14 is the identity permutation and @xmath21 are the cyclic permutations , as follows : a trilinear form @xmath22 is called @xmath0-skew - symmetric if for any elements @xmath23 of a vector space @xmath1 it satisfies @xmath24 the notion of a @xmath0-skew - symmetric form can be assumed as a basis for a ternary analog of grassmann , clifford and lie algebras .
these ternary structures were developed in @xcite and applied to construct a ternary analog of supersymmetry algebra in @xcite . in this paper
we study algebras with ternary law of composition . in section 2
we consider partially and totally associative ternary algebras of first and second kind . we show that a triple product of a ternary algebra induces three binary multiplications and find the sufficient and necessary condition a triple product of a ternary algebra must satisfy in order to induce the associative binary algebra . assuming the vector space underlying a ternary algebra to be a topological space and a triple product to be continuous mapping we consider the trivial vector bundle over a ternary algebra and
show that a triple product induces a structure of binary algebra in each fiber of this vector bundle .
the sufficient and necessary condition a ternary multiplication must satisfy in order to induce a structure of associative binary algebra in each fiber is given in terms of the vector bundle over a ternary algebra .
the relations for different kinds of partial and total associativity of a ternary algebra and induced by it binary algebras are found in the terms of the structure constants of a ternary algebra .
it should be pointed out that the cohomologies of a ternary algebra of associative type are studied in @xcite . in section 3
we consider an algebraic structure consisting of two bimodules over unital associative algebras with involution and construct a ternary algebra by means of this algebraic structure .
choosing different modules , unital associative algebras and homomorphisms we show that this structure allows to construct a large class of ternary algebras including a ternary algebra of rectangular matrices and ternary algebras of sections of a vector bundle over a smooth finite dimensional manifold .
we end the section 3 by constructing the binary lie algebra of matrices whose entries are the elements of bimodules and unital associative algebras .
it should be mentioned that there are @xmath25-ary generalizations of lie algebra which include the concepts such as @xmath25-ary algebra of lie type enclosing @xmath25-ary nambu algebra , @xmath25-ary nambu - lie algebra .
the concept of @xmath25-ary hom - algebra structure generalizing previously mentioned @xmath25-ary generalizations of lie algebra is introduced and studied in @xcite .
a good and detailed survey on the theory of ternary algebras can be found in @xcite .
it is well known that a large class of associative algebras can be constructed by means of square matrices and their multiplication .
though the rectangular matrices can be successfully used to construct a ternary algebra we think that probably more appropriate objects to construct ternary algebras are the cubic matrices .
our aim in section 4 is to construct ternary algebras of cubic matrices and to study their structures .
we find four different totally associative ternary multiplications of second kind of cubic matrices and prove that these are the only totally associative ternary multiplications of second kind in the case of cubic matrices .
it is worth mentioning that our search for associative ternary multiplications of cubic matrices has shown that there is no totally associative ternary multiplication of first kind in the case of cubic matrices .
i section 5 we describe the ternary analog of lie algebra of cubic matrices of second order by finding all commutation relations of generators of this algebra with respect to @xmath0-commutator .
in this section we remind a notion of a ternary algebra and its partial or total associativity of first or second kind . holding fixed one argument of a ternary multiplication
we get the binary multiplications and study the relation between the associativity of a ternary multiplication and the associativity of induced binary multiplication .
we propose to use a vector bundle approach to describe the family of binary algebras induced by a ternary algebra .
let @xmath26 be complex vector spaces , and @xmath27 be a @xmath28-valued trilinear form .
we will call @xmath22 a ternary law of composition or ternary multiplication on @xmath1 if @xmath22 is a @xmath1-valued trilinear form .
the pair @xmath29 is said to be a ternary algebra or triple system if @xmath1 is a complex vector space , and @xmath30 is a ternary law of composition on @xmath1 .
it is obvious that relation analogous to binary associativity in the case of ternary law of composition should contain at least five elements of @xmath1 .
there are three different ways to apply twice a ternary multiplication @xmath22 to ordered sequence of five elements @xmath31 which lead us to the following relations defining a notion of partial associativity for ternary multiplication : @xmath32 hence we have three different kinds of partially associative ternary algebra @xmath33 which will be called -partially associative ternary algebra ( [ lr - kind ] ) , -partially associative ternary algebra of first kind ( [ lc - kind ] ) and -partially associative ternary algebra of first kind ( [ cr - kind ] ) .
a ternary algebra @xmath33 is said to be totally associative ternary algebra of first kind if its ternary multiplication @xmath22 satisfies any two of the relations ( [ lr - kind])([cr - kind ] ) .
it is obvious that in the case of totally associative ternary algebra of first kind a ternary multiplication @xmath22 satisfies the relations @xmath34 where @xmath31 .
the notion of totally associative ternary algebra of first kind can be viewed as a direct ternary generalization of classical associativity @xmath35 , where @xmath36 are the elements of an algebra @xmath37 with a binary law of composition @xmath38 , when one applies twice algebra multiplication ( binary or ternary ) to ordered sequence of elements of algebra successively shifting the first ( interior ) multiplication from left to right and setting equal obtained products . in this sense
the notion of -partial associativity can be considered as most similar to classical associativity whereas the notion of -partial or -partial associativity can be defined for the first time only in the case of ternary multiplication because in the case of binary multiplication there is no central group of two elements in the middle of a sequence @xmath39 .
since the notion of -partial or -partial associativity appears for the first time in the case of ternary multiplication there is no reason to keep the requirement of fixed order of a sequence @xmath31 looking for a possible analogs of associativity in the case of ternary algebras .
it turns out that we get a useful notion of ternary associativity giving up the requirement of fixed order of elements in a sequence @xmath31 .
this means that unlike the case of ternary associativity of first kind we not only successively shift the first ( interior ) multiplication inside a sequence of elements @xmath31 from left to right but at the same time permute the elements @xmath40 in the middle of sequence .
obviously we should use non - cyclic permutation in order to get the initial order of a sequence @xmath41 on the second step .
this reasoning leads us to the following relations : @xmath42 a ternary algebra @xmath29 is said to be -partially associative ternary algebra of second kind if ternary multiplication @xmath22 satisfies ( [ lc - secondkind ] ) and -partially associative ternary algebra of second kind if @xmath22 satisfies ( [ cr - secondkind ] ) .
a ternary algebra @xmath29 is said to be totally associative ternary algebra of second kind if it is -partially associative and either -partially associative of second kind or -partially associative of second kind .
hence in the case of totally associative ternary algebra of second kind we have @xmath43 a ternary multiplication @xmath22 of ternary algebra @xmath33 has three arguments @xmath44 , where @xmath45 , and if we fix one of them then @xmath22 induces the binary multiplication on @xmath46 .
it is obvious that this allows us to split the ternary multiplication @xmath22 into three binary ones .
we can study the structure of ternary multiplication @xmath22 from this point of view by making use of known concepts and methods of the theory of binary algebras . given an element @xmath47 a ternary multiplication @xmath22 induces three binary multiplications @xmath48 on @xmath1 defined as follows : @xmath49 where @xmath50 .
the binary multiplications @xmath51 are not independent because of the relations @xmath52 a vector space @xmath1 equipped with the binary multiplication @xmath53 becomes the binary algebra which will be denoted by @xmath54 .
considering an element @xmath55 in @xmath54 as a parameter ranging within a vector space @xmath1 we have three families of binary algebras @xmath56 induced by a ternary multiplication @xmath22 .
the family of binary algebras @xmath54 is said to be an associative family of binary algebras induced by a ternary algebra @xmath33 if for any @xmath31 it holds @xmath57 taking @xmath58 in the previous relation we see that each associative family of binary algebras @xmath54 is the family of associative binary algebras .
it is useful to describe the above mentioned families of binary algebras in terms of vector bundle . for this purpose
we will assume that @xmath46 is a topological vector space , and a ternary multiplication @xmath2 is a continuous mapping . let us consider the direct product @xmath59 as the trivial vector bundle over the base space @xmath46 with the fiber @xmath1 and the projection @xmath60 defined as usual @xmath61 , where @xmath62 .
any fiber @xmath63 of @xmath64 is isomorphic to @xmath46 , and we will denote this isomorphism at a point @xmath55 of the base space @xmath1 by @xmath65 , i.e. @xmath66 and @xmath67 , where @xmath68 .
let @xmath69 be two points of the base space of a vector bundle @xmath64 .
then @xmath70 is the isomorphism between two fibers . in order to apply the constructed vector bundle @xmath71 to describe the families of binary algebras induced by a ternary algebra @xmath33 within the framework of a single structure we assume that the base space @xmath1 of this bundle is equipped with a ternary multiplication @xmath22 . for any point @xmath47 of the base space a fiber @xmath72 at this point
is endowed with one of the binary multiplications @xmath48 which we carry over from the family of binary algebras to fibers of @xmath71 by requiring @xmath65 to be an isomorphism of algebras , i.e. @xmath73 where @xmath74 .
if each fiber of @xmath71 is endowed with a binary multiplication @xmath75 then in order to emphasize this algebraic structure of fibers we will denote the corresponding vector bundle by @xmath76 .
thus @xmath77 , where the base space ( the first factor in the direct product ) is a ternary algebra @xmath33 , and a fiber @xmath63 is the binary algebra @xmath54 .
we will call @xmath76 , where @xmath78 , the vector bundle of binary algebras over a ternary algebra @xmath33
. a section @xmath79 of the vector bundle @xmath76 is a continuous mapping @xmath80 satisfying @xmath81 , and the vector space of continuous sections will be denoted by @xmath82 .
evidently this vector space equipped with the binary multiplication @xmath83 where @xmath84 , is the binary algebra .
the notion of an associative family of binary algebras defined by ( [ associativefamily ] ) can be described in the terms of vector bundle @xmath76 .
let @xmath85 be two points of a vector bundle @xmath76 and @xmath47 .
a vector bundle of binary algebras @xmath76 is said to be an associative vector bundle of binary algebras over a ternary algebra @xmath33 if for any @xmath85 and @xmath47 it holds @xmath86 particularly ( [ associative_family ] ) implies the associativity of a fiber @xmath63 for any @xmath47 if we take @xmath87 in ( [ associative_family ] ) , i.e. any associative vector bundle of binary algebras @xmath76 is a vector bundle of associative binary algebras whereas the converse is generally not true .
now it is natural to pose a question concerning the associativity of induced binary algebras @xmath54 provided a ternary algebra @xmath33 is partially or totally associative of first or second kind .
[ lr - associativity ] a ternary algebra @xmath33 is -partially associative ternary algebra if and only if @xmath88 is the associative vector bundle of binary algebras over a ternary algebra @xmath33 . particularly if a base space @xmath33 is -partially associative ternary algebra then each fiber of the vector bundle @xmath88 is an associative binary algebra and the binary algebra of sections of this bundle @xmath89 is associative algebra .
indeed the left side of ( [ associative_family ] ) can be transformed as follows : @xmath90 analogously for the right side of ( [ associative_family ] ) we have @xmath91 and this proves the -partial associativity of a ternary algebra @xmath33 .
it is well known that any associative binary algebra is the lie algebra with respect to the commutator defined with the help of a binary multiplication of this algebra , and the associativity of a binary multiplication implies the jacoby identity for the commutator .
it follows from the proposition [ lr - associativity ] that if a ternary algebra @xmath92 is -partially associative ternary algebra then each fiber @xmath63 , where @xmath93 , of the vector bundle @xmath88 is the lie algebra with respect to the commutator @xmath9_a$ ] defined by @xmath94_a=\binc(p , q)-\binc(q , p)\ ] ] where @xmath74 . clearly
the associative binary algebra of sections @xmath89 is the lie algebra under the commutator @xmath95(a)=[\xi(a),\eta(a)]_a$ , \quad where\quad $ \xi,\eta\in \gamma({\frak e}_2)\ ] ] a ternary algebra @xmath33 is said to be a ternary algebra of lie type of first kind ( of second kind ) if for any @xmath96 it holds @xmath97 where @xmath98 is the symmetry group of third order and @xmath99 is its cyclic subgroup .
clearly any ternary algebra of lie type of second kind is a ternary algebra of lie type of first kind whereas the converse is generally not true . from ( [ grassmann_type ] )
it follows that any element @xmath55 of a ternary algebra of lie type ( of first or second order ) satisfies @xmath100 .
it is pointed out in the introduction that we can construct a ternary analog of the notion of skew - symmetry by means of a faithful representation of @xmath99 by cubic roots of unity .
let @xmath101 be the primitive cubic root of unity .
a ternary multiplication @xmath22 of a ternary algebra @xmath33 is said to be @xmath0-skew - symmetric if @xmath102 where @xmath45 .
if a ternary multiplication @xmath22 of @xmath33 is @xmath0-skew - symmetric then @xmath33 is a ternary algebra of lie type of second order . indeed in this case
we have @xmath103 we see that the notion of @xmath0-skew - symmetric ternary multiplication is based on the faithful representation of the cyclic group @xmath99 by cubic roots of unity . given a ternary algebra @xmath33 we can make it the ternary algebra of lie type of second order by endowing it with the ternary @xmath0-brackets or the ternary @xmath0-commutator which is defined by @xmath104=\tau(a , b , c)+j\,\tau(b , c , a)+j^2\,\tau(c , a , b)\ ] ] if @xmath1 has an involution @xmath105 then @xmath22 will be called hermitian if it satisfies @xmath106 .
hence a hermitian @xmath0-skew - symmetric ternary multiplication @xmath22 satisfies ( [ cyclic j - symmetric ] ) and @xmath107 let us suppose that @xmath108 is a -partially nonassociative ternary algebra , i.e. in general we have @xmath109 , where @xmath110 .
a ternary algebra @xmath108 is said to be a ternary algebra of jordan type if its ternary multiplication @xmath111 satisfies the following identities : @xmath112 where @xmath45 .
it is easy to see that if @xmath108 is a ternary algebra of jordan type then for any @xmath47 the binary algebra @xmath113 is the jordan algebra . indeed in this case
the identities ( [ first_identity],[second_identity ] ) take on the form @xmath114 if @xmath33 is -partially associative ternary algebra then the ternary algebra @xmath108 , where @xmath115 is the ternary algebra of jordan type .
we see that having fixed one variable in a triple product @xmath44 of a ternary algebra @xmath33 we can study the structure of a ternary multiplication @xmath22 by splitting it into three binary ones .
what kind of structures induces a ternary multiplication of @xmath33 if one fixes two variables in @xmath44 ? obviously fixing two variables we get the linear operator acting on @xmath46 , and this is the second way for studying the structure of a ternary multiplication .
let @xmath116 be the algebra of linear operators of the vector space @xmath1 .
given a pair @xmath117 we define the linear operators @xmath118 , where @xmath78 , as follows : @xmath119 actually these operators are not independent because for any @xmath120 we have the relations @xmath121 it is easy to see that for every @xmath122 the linear operator @xmath123 is bilinear with respect to its variables @xmath124 , and therefore the family of linear operators @xmath125 determines the bilinear mapping @xmath126 . on the other hand if there is a vector space @xmath1 equipped with a bilinear mapping @xmath127 then one can construct the ternary algebra @xmath33 by letting @xmath128 .
now we can introduce an analog of identity element for a ternary algebra @xmath33 by means of the bilinear mappings @xmath129 , @xmath78 . indeed given an element @xmath47 we define @xmath130 by @xmath131 where @xmath132 is the identity operator .
a pair @xmath133 is said to be an identity @xmath122-pair for an element @xmath55 if @xmath134 .
let @xmath135 and @xmath136 .
if @xmath137 then we will call an element @xmath138 an identity @xmath122-pair , and similarly if @xmath139 then we will call an element @xmath140 an identity pair of a ternary algebra @xmath33 .
let us now assume that the vector space @xmath1 of a ternary algebra @xmath141 is a finite dimensional vector space , i.e. @xmath1 is an @xmath142-dimensional vector space and @xmath143 is a basis for @xmath144 then for any element @xmath145 we have @xmath146 , and the triple product of elements @xmath120 can be expressed as follows : @xmath147 where @xmath148 are the structure constants of a ternary algebra @xmath33 defined by @xmath149 if @xmath150 is another basis for a vector space @xmath151 and @xmath152 where @xmath153 is the transition matrix then @xmath154 where @xmath155 are the structure constants of a ternary algebra @xmath141 with respect to a basis @xmath156 , i.e. @xmath157 , and @xmath158 is the inverse matrix of @xmath159 .
if we require a ternary algebra @xmath33 to be a partially or totally associative ternary algebra either of first or second kind then this requirement leads to the relations the structure constants of @xmath33 have to satisfy .
these relations for different kinds of associativity of first kind have the following form : @xmath160 it follows from the above relations that if @xmath141 is a totally associative ternary algebra of first kind then the structure constants satisfy @xmath161 in the case of ternary associativity of second kind we have the following relations : @xmath162 the structure constants of a totally associative ternary algebra of second kind satisfy @xmath163 for any @xmath47 a ternary algebra @xmath33 induces three binary algebras @xmath164 , with the binary multiplications defined by relations @xmath165 .
the structure constants @xmath166 of binary algebra @xmath167 defined by @xmath168 can be expressed in terms of the structure constants of a ternary algebra @xmath33 as follows : @xmath169 where @xmath170 .
in this section we propose few different methods for constructing ternary algebras and apply these methods to construct a ternary algebra of vector fields on a smooth finite dimensional manifold and a ternary algebra of rectangular matrices .
particularly the curvature of an affine connection determines the structure of a ternary algebra on the module of vector fields on a smooth manifold .
given two modules over the algebras with involutions we construct a ternary algebra which is used to construct a lie algebra .
our approach generalizes the approach proposed in @xcite , where the authors use the rectangular complex matrices .
let @xmath1 be a vector space over the complex numbers @xmath171 and @xmath172 be the dual space .
given a @xmath171-multilinear mapping @xmath173 we construct the ternary algebra @xmath33 by defining the ternary multiplication @xmath22 as follows : @xmath174 where @xmath175 .
particularly given a @xmath171-bilinear mapping @xmath127 , where @xmath116 is the algebra of linear operators of a vector space @xmath1 , we define @xmath176 and applying ( [ dual_t ] ) we get the ternary algebra @xmath33 whose ternary multiplication @xmath22 can be described implicitly by the formula @xmath177 applying this construction to a module over an associative unital algebra we can construct a ternary algebra by means of ( [ dual_t ] ) or ( [ operator_way ] ) . indeed if @xmath1 is a left @xmath178-module , where @xmath178 is a binary unital associative complex algebra , @xmath172 is the dual module and @xmath179 is an @xmath178-multilinear mapping then @xmath33 is the ternary algebra with the ternary multiplication defined by ( [ dual_t ] ) . similarly given an @xmath178-module @xmath1 and a @xmath178-bilinear mapping @xmath127 , where @xmath116 is the algebra of @xmath178-linear operators of a module @xmath1 , then @xmath46 is the ternary algebra @xmath33 with the ternary multiplication @xmath22 defined by ( [ operator_way ] )
. we can use ( [ dual_t],[operator_way ] ) to construct the ternary algebras by means of well known structures of differential geometry on a smooth manifold .
let @xmath180 be a vector bundle over a smooth finite dimensional manifold @xmath181 , @xmath182 be the algebra of smooth functions on a smooth manifold @xmath181 , and @xmath183 be the module of smooth sections of @xmath184 .
given a @xmath182-multilinear mapping @xmath185 , where @xmath186 is the dual bundle , we obtain the ternary algebra @xmath187 of smooth sections of a vector bundle @xmath184 with the ternary multiplication @xmath188 , where @xmath189 are sections of @xmath184 , and @xmath190 . particularly if @xmath181 is a smooth manifold
, @xmath191 is the tangent bundle , @xmath192 is the cotangent bundle , @xmath193 is the module of vector fields , @xmath194 is the module of 1-forms , @xmath195 is an affine connection on @xmath181 , and @xmath196 } , \quad x , y\in { \frak d}(m)\ ] ] is the curvature of @xmath195 then we have the @xmath182-multilinear mapping @xmath197 induced by the curvature @xmath198 and this mapping induces the structure of the ternary algebra @xmath199 on the module of vector fields with the ternary multiplication @xmath200 let @xmath201 be ( binary ) unital associative algebras over @xmath171 with involutions respectively @xmath202 and @xmath203 , where @xmath204 . let @xmath205 be a @xmath206-bimodule .
we suppose that @xmath207 is an abelian group which is isomorphic to the abelian group @xmath205 , where @xmath208 is the corresponding isomorphism .
then @xmath207 can be endowed with the structure of @xmath209-bimodule if we define the right and left multiplication by elements of algebras @xmath210 , @xmath211 as follows @xmath212 let @xmath213 , @xmath214 be the tensor products of modules where the first tensor product has the structure of @xmath178-bimodule , and the second has the structure of @xmath215-bimodule .
it is clear that the algebras @xmath210 , @xmath215 can be viewed respectively as @xmath178-bimodule and @xmath215-bimodule .
we also assume that there are two homomorphisms @xmath216 , @xmath217 respectively of @xmath178-bimodules and @xmath215-bimodules which satisfy @xmath218 where @xmath219 .
evidently @xmath205 has the structure of vector space over @xmath171 .
we define the ternary law of composition @xmath22 on @xmath220 by the formula @xmath221 @xmath222 is -partially associative ternary algebra .
[ lr - associativity - of - m ] indeed for any quintuple of elements @xmath223 of @xmath205 we have @xmath224 on the other hand @xmath225 and this ends the proof . from this proposition and the proposition [ lr - associativity ]
it follows that for any @xmath226 the binary algebra @xmath227 , where @xmath228 , is an associative algebra .
let us define the ternary multiplication @xmath111 on @xmath220 by @xmath229 , where @xmath219 .
the ternary algebra @xmath230 is the ternary algebra of jordan type , and the ternary multiplication @xmath111 of this algebra satisfies the identity @xmath231 where @xmath232 .
we use the ternary algebra @xmath233 to construct a lie algebra which will be constructed by means of defining commutation relations .
we denote the set of generators of this lie algebra by @xmath234 , where @xmath235 , i.e. we assign to each element @xmath236 the generator @xmath237 , to each element @xmath238 the generator @xmath239 , and to each pair @xmath240 the generator @xmath241 .
let @xmath242 be the linear spans induced by the corresponding generators and @xmath243 .
we define @xmath244 = s_{mn},\quad\;\;\;\;\;\;\;\ ; [ s_{pq},u_{m } ] = u_{\sigma(p , q , m ) } \label{brackets-1}\\ [ s_{pq},u_{\bar n } ] = -u_{\overline{\sigma(q , p , n)}},\;\ ; [ s_{mn},s_{pq } ] = s_{\sigma(m , n , p)\,q}-s_{p\,\sigma(n , m , q ) } \label{brackets-2}\end{gathered}\ ] ] the vector space @xmath245 endowed with the brackets defined by ( [ brackets-1 ] ) , ( [ brackets-2 ] ) is the lie algebra , and the identity ( [ identity ] ) leads to the jacoby identity for the brackets ( [ brackets-1 ] ) , ( [ brackets-2 ] ) .
we can construct a matrix representation for the lie algebra @xmath245 if we consider the set @xmath246 of all @xmath247 square matrices of the type @xmath248 given two such matrices @xmath249 we define their product as follows @xmath250 the vector space of matrices @xmath246 endowed with the multiplication ( [ product of matrices ] ) is a unital associative ( binary ) algebra with the unity element @xmath251 where @xmath14 is the unity element of @xmath178 and @xmath252 is the unity element of @xmath215 .
now we construct the matrix representation for the lie algebra @xmath245 as follows @xmath253 and the matrix representation of @xmath241 can be found by explicit calculation .
having constructed the lie algebra @xmath245 and its matrix representation we can go further and construct a gauge field theory based on ternary algebra @xmath230 .
for this purpose we take a vector bundle @xmath184 over a smooth finite - dimensional manifold @xmath181 with the fiber @xmath230 .
a section of this bundle is a gauge field of our theory may be called ternon . now in each fiber we construct the lie algebra @xmath245 and this leads us to the vector bundle of lie algebras .
next we construct the lie group corresponding to @xmath245 by means of exponential mapping and campbell - hausdorf series .
we get the principal fiber bundle and the we proceed in constructing the gauge field theory as usual .
in this section we consider a vector space of cubic matrices , where by cubic matrix we mean a quantity @xmath254 with three subscripts @xmath255 each running some set of integers .
we use this vector space to construct a ternary algebra by means of triple product of cubic matrices . a triple product or ternary multiplication of cubic matrices
is constructed in analogy with the classical product of two rectangular matrices by means of summation which is taken over certain system of subscripts of three cubic matrices .
our aim in this section is to find all totally associative ternary multiplications of first or second kind , and we prove that there are four ternary multiplications of cubic matrices each yielding the associative ternary algebra of second kind .
let @xmath256 , where @xmath257 and @xmath258 are integers satisfying @xmath259 , @xmath260 , @xmath261 .
we will call @xmath159 a complex @xmath262space matrix provided that its entries @xmath263 are arranged in the vertices of a 3dimensional lattice and this structure is shown in particular case of a cubic matrix on the figure below .
let us denote the set of all such matrices by @xmath264 , i.e. @xmath265 the set of @xmath262space matrices is the vector space if we define the addition of space matrices and multiplication by complex numbers as usual @xmath266 our main concern in this paper is a special case of space matrices when @xmath267 . in this case we will call @xmath254 , where @xmath268 , a complex @xmath269-cubic matrix and denote the vector space of such matrices by @xmath270 .
particularly , if @xmath271 is a cubic matrix of third order then we will place its entries into the vertices of @xmath272dimensional lattice as follows @xmath273\ar@{.}'[d]'[dd][ddd ] & & & a_{123 } \ar@{.}[rrr]\ar@{.}'[d]'[dd][ddd ] & & & a_{133 } \ar@{.}[ddd ] \\ & a_{112 } \ar@{-}[ur ] \ar@{-}[rrr ] \ar@{-}'[d][ddd ] & & & a_{122 } \ar@{-}[ur ] \ar@{-}[rrr ] \ar@{-}'[d][ddd ] & & & a_{132 } \ar@{-}[ur ] \ar@{-}[ddd ] \\ a_{111 } \ar@{-}[ur ] \ar@{-}[rrr ] \ar@{-}[ddd ] & & & a_{121 } \ar@{-}[ur ] \ar@{-}[rrr ] \ar@{-}[ddd ] & & & a_{131 } \ar@{-}[ur ] \ar@{-}[ddd ] \ar@{-}[ddd ] \\ & & a_{213 } \ar@{.}'[r]'[rr][rrr ] \ar@{.}'[d]'[dd][ddd ] & & & a_{223 } \ar@{.}'[r]'[rr][rrr ] \ar@{.}'[d]'[dd][ddd ] & & & a_{233 } \ar@{.}[ddd ] \\ & a_{212 } \ar@{-}[ur ] \ar@{-}'[rr][rrr ] \ar@{-}'[d][ddd ] & & & a_{222 } \ar@{-}[ur ] \ar@{-}'[rr][rrr ] \ar@{-}'[d][ddd ] & & & a_{232 } \ar@{-}[ur ] \ar@{-}[ddd ] \\ a_{211 } \ar@{-}[ur ] \ar@{-}[rrr ] \ar@{-}[ddd ] & & & a_{221 } \ar@{-}[ur ] \ar@{-}[rrr ] \ar@{-}[ddd ] & & & a_{231 } \ar@{-}[ur ] \ar@{-}[ddd ] \\ & & a_{313 } \ar@{.}'[r]'[rr][rrr ] & & & a_{323 } \ar@{.}'[r]'[rr][rrr ] & & & a_{333 } \\ & a_{312 } \ar@{-}[ur ] \ar@{-}'[rr][rrr ] & & & a_{322 } \ar@{-}[ur ] \ar@{-}'[rr][rrr ] & & & a_{332 } \ar@{-}[ur ] \\ a_{311 } \ar@{-}[rrr]\ar@{-}[ur ] & & & a_{321 } \ar@{-}[rrr]\ar@{-}[ur ] & & & a_{331 } \ar@{-}[ur ] } \nonumber\end{aligned}\ ] ] the above @xmath272dimensional lattice clearly shows that if one fixes a value of subscript @xmath274 in @xmath275 with @xmath276 ranging from 1 to 3 then the corresponding entries of cubic matrix @xmath159 form the square matrix of order @xmath272 . hence we get three square matrices of order @xmath272 which all together give us a cubic matrix of third order @xmath159 .
therefore any cubic matrix @xmath159 of third order can be represented as the set of square matrices of order @xmath272 as follows @xmath277 \nonumber\end{aligned}\ ] ] now our aim is to construct a multiplication of space matrices .
we will do this for cubic matrices of order @xmath269 because constructed multiplication can be extended to a vector space of @xmath262space matrices in an obvious way .
if @xmath278 is a cubic matrix of order @xmath269 then it induces two mappings as follows : 1 .
@xmath279 2 .
@xmath280 where @xmath281 is the vector space of linear operators acting on @xmath282 .
these mappings determined by a cubic matrix of order @xmath269 show that if we wish to construct a multiplication of cubic matrices of order @xmath269 which is based on composition of linear mappings as in the calculus of rectangular matrices then we should take three cubic matrices in order to close a corresponding multiplication in the sense that the product of three cubic matrices of order @xmath269 will be the cubic matrix of order @xmath269 . given two cubic matrices @xmath283 we can form two products of corresponding mappings @xmath284 and @xmath285 which are not close with respect to composition because the first product is the linear mapping @xmath286 , and the second is the linear mapping @xmath287 .
it is obvious than we can close a procedure of taking compositions of this sort of mappings by adding one more cubic matrix @xmath288 . in this case
the triple product @xmath289 closes the operation of taking compositions giving the mapping @xmath290 .
this kind of reasoning suggests a possible way of constructing the cubic matrix of order @xmath269 from three given ones by means of summation which is taken over the certain pair of subscripts . given three cubic matrices of order @xmath269 we have nine subscripts , and because the number of possible combinations of subscripts is finite we can use the methods of computer algebra to find all associative ternary multiplications either of first or second kind .
our analysis shows that there is no total associative ternary multiplication of first kind , and all total associative ternary multiplications of second kind are described by the following theorem .
there are only four different triple products of complex cubic matrices of order @xmath269 which obey the total ternary associativity of second kind .
these are 1 .
@xmath291 2 .
@xmath292 3 .
@xmath293 4 .
@xmath294 [ theorem ]
now our aim is to construct a ternary analog of lie algebra with the help of ternary multiplication of cubic matrices and ternary analog of lie bracket . a notion of ternary analog of lie bracket
is based on a faithful representation of cyclic group @xmath295 by cubic roots of unity .
we construct a ternary analog of lie algebra which may be considered as an analog of the lie algebra generated by pauli matrices . in our construction
we use the cubic matrices of second order with certain symmetries with respect to the subscripts as generators of our algebra and find all commutation relations . in this section
we use the ternary multiplication of cubic matrices which has the property that any cyclic permutation of the matrices in the product is equivalent to the same permutation on the subscripts and this multiplication is studied in @xcite .
let @xmath296 @xmath297 @xmath298 be cubic matrices of order @xmath269 .
we define the triple product @xmath299 by the following formula @xmath300 it is easy to see that any cyclic permutation of the matrices in the product is equivalent to the same permutation on the subscripts , i.e. @xmath301 it should be mentioned that the ternary multiplication ( [ multiplication ] ) is neither partially nor totally associative . the ternary algebra of cubic matrices of order @xmath269 with respect to the multiplication ( [ multiplication ] )
can be decomposed into direct sum of subspaces of cubic matrices with certain symmetries according to the irreducible representation of the symmetry group @xmath302 it should be mentioned that we have the similar decomposition in the case of square matrices of order @xmath303 where the algebra of square matrices can be decomposed into the direct sum of subspaces of symmetric and skew - symmetric matrices according to representation of the permutation group @xmath304 the symmetry group @xmath98 possesses a full and faithful representation on the complex plane , which can be generated by two elements representing a cyclic and an odd permutation
. this representation can be constructed by assigning the operator of multiplication by the cubic root of unity @xmath305 to the cyclic permutation @xmath306 and assigning the operator of complex conjugation to the odd permutation @xmath307 . in order to construct the subspace of cubic matrices with certain symmetries we use the cyclic part of this representation where the cyclic group @xmath295 is represented by cubic roots of unity .
a cubic matrix @xmath308 of order @xmath269 is said to be a @xmath0-skew - symmetric if it satisfies @xmath309 similarly a cubic matrix @xmath310 of order @xmath269 is said to be a @xmath311-skew - symmetric if it satisfies @xmath312 it can be shown that the space of cubic matrices of order @xmath269 can be decomposed into the direct sum of the subspace of @xmath0-skew - symmetric matrices , the subspace of @xmath311-skew - symmetric matrices and the subspace of symmetric matrices , where under symmetric matrix we mean a cubic matrix @xmath313 of order @xmath269 which satisfies @xmath314 .
each of subspaces of @xmath0-skew - symmetric matrices and of @xmath311-skew - symmetric matrices has the dimension @xmath315 the subspace of symmetric matrices can be decomposed into the direct sum of the subspace of diagonal cubic matrices ( all entries excepting diagonal ones are equal to zero ) and the subspace of symmetric matrices with zeros on diagonal .
obviously the dimension of the subspace of diagonal cubic matrices is equal to @xmath269 and the subspace of symmetric matrices with zeros on diagonal is equal to @xmath315 let @xmath316 be a basis for the subspace of @xmath0-skew - symmetric matrices , @xmath317 be a basis for the subspace of @xmath311-skew - symmetric matrices , and @xmath318 be a basis for the subspace of symmetric matrices with zeros on diagonal , where @xmath319 a basis for the subspace of diagonal cubic matrices will be denoted by @xmath320 where @xmath321 thus we have @xmath322 now our aim is to construct a ternary analog of lie algebra by means of the ternary @xmath0-commutator which we define as follows @xmath323=a\circledcirc b\circledcirc c+jb\circledcirc c\circledcirc a+j^2c\circledcirc a\circledcirc b,\quad a , b , c\in \mbox{cmat}_{n}({\mathbb c } ) \label{commutator}\ ] ] where the ternary multiplication @xmath324 is defined by ( [ multiplication ] ) .
it is easy to see that the ternary @xmath0-commutator ( [ commutator ] ) may be viewed as an analog of the binary lie commutator @xmath325=xy - yx,$ ] where @xmath326 @xmath327 are the elements of an associative binary algebra , because we replace the symmetric group @xmath328 and its representation @xmath329 respectively by the cyclic group @xmath295 and its representation @xmath330 indeed in the case of the binary lie commutator we have the property @xmath325=-[y , x].$ ] for the ternary @xmath0-commutator we have the similar property @xmath331=j[b , c , a]=j^2[c , a , b].$ ] in analogy with the binary lie commutator for any @xmath332 we have @xmath333=0.$ ] let us consider the space of cubic matrices of order @xmath334 i.e. each subscript runs from 1 to 2 . in this case
all subspaces mentioned above are 2-dimensional , which means that basis for each subspace consists of two matrices , i.e. @xmath335 we construct the basis for the subspace of diagonal cubic matrices @xmath336 by choosing @xmath337 we choose the basis for the subspace of symmetric cubic matrices with zeros on diagonal by fixing @xmath338 @xmath339 with other entries are equal to zero .
we get the basis for the subspace of @xmath0-skew - symmetric cubic matrices by taking @xmath340 @xmath341 with other entries equal to zero and the basis for the subspace of @xmath311-skew - symmetric cubic matrices if we put @xmath342 @xmath343 with other entries equal to zero .
the space of cubic matrices of order 2 equipped with the ternary @xmath0-commutator can be considered as a ternary analog of matrix lie algebra and all commutation relations of this algebra in the basis @xmath344 are given in the following table : @xmath345 & = -\orhob , & [ \rhob , \rhoa , \rhob ] & = -\orhoa & [ \orhoa , \orhob , \orhoa ] & = 2\orhob , & [ \orhob , \orhoa , \orhob ] & = 2\orhoa \\ [ \omegaa , \omegab , \omegaa ] & = -\orhob , & [ \omegab , \omegaa , \omegab ] & = -\orhoa & [ \etaa , \etab , \etaa ] & = 0 , & [ \etab , \etaa , \etab ] & = 0 \\ \\ [ \rhoa , \orhoa , \rhoa ] & = 3\orhoa , & [ \rhob , \orhob , \rhob ] & = 3\orhob , & [ \orhoa , \rhoa , \orhoa ] & = 0 , & [ \orhob , \rhob , \orhob ] & = 0 \\ [ \rhoa , \orhob , \rhoa ] & = -\orhob , & [ \rhob , \orhoa , \rhob ] & = -\orhoa , & [ \orhoa , \rhob , \orhoa ] & = -\orhob , & [ \orhob , \rhoa , \orhob ] & = -\orhoa \\ [ \rhoa , \orhoa , \rhob ] & = 2\orhob , & [ \rhoa , \orhob , \rhob ] & = 2\orhoa , & [ \rhoa , \orhoa , \orhob ] & = -\orhob , & [ \rhob , \orhob , \orhoa ] & = -\orhoa \\ \\ [ \rhoa , \omegaa , \rhoa ] & = 0 , & [ \omegaa , \rhoa , \omegaa ] & = 3\orhoa , & [ \rhob , \omegab , \rhob ] & = 0 , & [ \omegab , \rhob , \omegab ] & = 3\orhob \\ [ \rhoa , \omegab , \rhoa ] & = 2\orhob , & [ \omegab , \rhoa , \omegab ] & = 2\orhoa , & [ \rhob , \omegaa , \rhob ] & = 2\orhoa , & [ \omegaa , \rhob , \omegaa ] & = 2\orhob \\ [ \rhob , \rhoa , \omegaa ] & = -\orhob , & [ \omegab , \rhob , \rhoa ] & = -\orhoa , & [ \omegaa , \rhoa , \omegab ] & = -\orhob , & [ \omegaa , \rhob , \omegab ] & = -\orhoa \\ \\ [ \orhoa , \omegaa , \orhoa ] & = 3\orhoa , & [ \omegaa , \orhoa , \omegaa ] & = 0 , & [ \orhob , \omegab , \orhob ] & = 3\orhob , & [ \omegab , \orhob , \omegab ] & = 0 \\ [ \orhoa , \omegab , \orhoa ] & = -\orhob , & [ \omegab , \orhoa , \omegab ] & = -\orhoa , & [ \orhob , \omegaa , \orhob ] & = -\orhoa , & [ \omegaa , \orhob , \omegaa ] & = -\orhob \\ [ \orhoa , \omegaa , \orhob ] & = -\orhob , & [ \orhoa , \omegab , \orhob ] & = -\orhoa , & [ \orhoa , \omegaa , \omegab ] & = 2\orhob , & [ \omegaa , \omegab , \orhob ] & = 2\orhoa \\ \\ [ \rhoa , \etaa , \rhoa ] & = \orhoa , & [ \etaa , \rhoa , \etaa ] & = 0 , & [ \rhob , \etab , \rhob ] & = \orhob , & [ \etab , \rhob , \etab ] & = 0 \\ [ \rhoa , \etab , \rhoa ] & = \orhob , & [ \etaa , \rhob , \etaa ] & = \orhob , & [ \rhob , \etaa , \rhob ] & = \orhoa , & [ \etab , \rhoa , \etab ] & = \orhoa \\ [ \rhoa , \rhob , \etaa ] & = \orhob , & [ \etab , \rhoa , \etab ] & = \orhoa , & [ \etaa , \rhoa , \etab ] & = 0 , & [ \etaa , \rhob , \etab ] & = 0 \\ \\ [ \orhoa , \etaa , \orhoa ] & = \orhoa , & [ \etaa , \orhoa , \etaa ] & = 0 , & [ \orhob , \etab , \orhob ] & = \orhob , & [ \etab , \orhob , \etab ] & = 0 \\ [ \orhoa , \etab , \orhoa ] & = \orhob , & [ \etaa , \orhob , \etaa ] & = \orhob , & [ \orhob , \etaa , \orhob ] & = \orhoa , & [ \etab , \orhoa , \etab ] & = \orhoa \\ [ \orhoa , \etaa , \orhob ] & = \orhob , & [ \orhoa , \etab , \orhob ] & = \orhoa , & [ \etaa , \orhoa , \etab ] & = 0 , & [ \etaa , \orhob , \etab ] & = 0 \\ \\ [ \etaa , \omegaa , \etaa ] & = 0 , & [ \etaa , \omegab , \etaa ] & = \orhob , & [ \etab , \omegaa , \etab ] & = \orhoa , & [ \etab , \omegab , \etab ] & = \orhob \\ [ \omegaa , \etaa , \omegaa ] & = \orhoa , & [ \omegaa , \etab , \omegaa ] & = \orhob , & [ \omegab , \etaa , \omegab ] & = \orhoa , & [ \omegab , \etab , \omegab ] & = \orhob \\ [ \etaa , \omegaa , \etab ] & = 0 , & [ \etaa , \omegab , \etab ] & = 0 , & [ \etaa , \omegaa , \omegab ] & = \orhob , & [ \etab , \omegab , \omegaa ] & = \orhoa \\ \\ [ \rhoa , \orhoa , \omegaa ] & = 0 , & [ \omegaa , \rhoa , \orhob ] & = 2\orhob , & [ \rhoa , \orhoa , \omegab ] & = -\orhob , & [ \rhoa , \rhob , \omegab ] & = -\orhoa \\ [ \etaa , \rhoa , \orhoa ] & = \orhoa , & [ \etaa , \rhoa , \orhob ] & = \orhob , & [ \orhoa , \etab , \rhoa ] & = \orhob , & [ \rhoa , \orhob , \etab ] & = \orhoa \\ [ \orhoa , \omegaa , \rhob ] & = -\orhob , & [ \rhob , \orhob , \omegaa ] & = -\orhoa , & [ \omegab , \rhob , \orhoa ] & = 2\orhoa , & [ \rhob , \orhob , \omegab ] & = 0 \\ [ \etaa , \orhoa , \rhob ] & = \orhob , & [ \rhob , \etaa , \orhob ] & = \orhoa , & [ \etab , \rhob , \orhoa ] & = \orhoa , & [ \etab , \rhob , \orhob ] & = \orhob \\ [ \etaa , \orhoa , \omegaa ] & = \orhoa , & [ \etaa , \omegab , \orhoa ] & = \orhob , & [ \omegaa , \etab , \orhoa ] & = \orhob , & [ \etab , \orhoa , \omegab ] & = \orhoa \\ [ \etaa , \orhob , \omegaa ] & = \orhob , & [ \orhob , \etaa , \omegab ] & = \orhoa , & [ \orhob , \omegaa , \etab ] & = \orhoa , & [ \etab , \orhob , \omegab ] & = \orhob \\ [ \etaa , \rhoa , \omegaa ] & = \orhoa , & [ \rhoa , \etaa , \omegab ] & = \orhob , & [ \omegaa , \etab , \rhoa ] & = \orhob , & [ \rhoa , \etab , \omegab ] & = \orhoa \\ [ \omegaa , \etaa , \rhob ] & = \orhob , & [ \omegab , \etaa , \rhob ] & = \orhoa , & [ \omegaa , \etab , \rhob ] & = \orhoa , & [ \etab , \rhob , \omegab ] & = \orhob\end{aligned}\ ] ] .1 cm we would like to point out that the above table of commutation relations demonstrates a peculiar property of the algebra of cubic matrices of second order in the basis @xmath344 : the @xmath0-commutator of any three generators is proportional to one of the generators @xmath346 .
moreover taking @xmath311-commutator @xmath323^{*}=a\circledcirc b\circledcirc c+j^2\,b\circledcirc c\circledcirc a+j\,c\circledcirc a\circledcirc b,\quad a , b , c\in \mbox{cmat}_{n}({\mathbb c } ) , \label{con - commutator}\ ] ] which can be considered as a conjugate commutator to @xmath0-commutator ( [ commutator ] ) , we get the similar table of commutation relations ( we do not demonstrate it here in order not to overburden the paper ) which clearly shows that the @xmath311-commutator of any three generators is proportional to one of the generators @xmath347 . finally considering a ternary analog of anti - commutator defined by @xmath348 and applying it to the cubic matrices of second order
we obtain the table of commutation relations for the generators @xmath349 @xmath350 @xmath351 @xmath352 @xmath353 @xmath354 @xmath355 , @xmath356 which can be shortly described as follows : the ternary anti - commutator of any three generators is a linear combination of the generators @xmath357 .
we conclude this section by pointing out an analogy between the generators of our ternary algebra of cubic matrices of second order and the pauli matrices @xmath358 indeed applying the ternary @xmath0-commutator to pauli matrices we get @xmath359=2i\sigma_2 , \quad[i\sigma_2,i\sigma_1,i\sigma_2]=2i\sigma_1 \nonumber\end{aligned}\ ] ] in the table of commutation relations of cubic matrices of order 2 we have @xmath360 = 2 \orhob,\quad\;\;\;\ ; [ \orhob , \orhoa , \orhob]=2\orhoa \nonumber\end{aligned}\ ] ] which clearly demonstrates a striking analogy between the @xmath311-skew - symmetric generators of ternary algebra of cubic matrices of order 2 and the skew - hermitian matrices @xmath361 @xmath362 @xmath363 where @xmath364 @xmath365 @xmath366 are pauli matrices .
v. abramov , o. liivapuu and s. shitov gratefully acknowledge the financial support of their research by the estonian science foundation under the research grant etf 7427 .
v. abramov , r. kerner and o. liivapuu would like to thank the french - estonian scientific program `` g. f. parrot '' for financial support under the travel grant p-2/2007 . | we study partially and totally associative ternary algebras of first and second kind . assuming the vector space underlying a ternary algebra to be a topological space and a triple product to be continuous mapping we consider the trivial vector bundle over a ternary algebra and
show that a triple product induces a structure of binary algebra in each fiber of this vector bundle .
we find the sufficient and necessary condition for a ternary multiplication to induce a structure of associative binary algebra in each fiber of this vector bundle .
given two modules over the algebras with involutions we construct a ternary algebra which is used as a building block for a lie algebra .
we construct ternary algebras of cubic matrices and find four different totally associative ternary multiplications of second kind of cubic matrices .
it is proved that these are the only totally associative ternary multiplications of second kind in the case of cubic matrices .
we describe a ternary analog of lie algebra of cubic matrices of second order which is based on a notion of @xmath0-commutator and find all commutation relations of generators of this algebra . 17a40 , 20n10 .
[ firstpage ] |
semiconductor quantum dots ( qds ) coupled to optical microcavities have been extensively studied as a solid - state system for cavity quantum electrodynamics ( c - qed ) and as a promising platform to implement quantum information processing protocols involving single photons and single electron spins@xcite .
efficient single photon sources have been demonstrated@xcite , and the generation of polarization entangled photons has been proposed@xcite and demonstrated@xcite .
quantum information protocols employing c - qed require the implementation of a spin - photon interface , to optically read - out the spin state@xcite and to deterministically generate spin - photon entanglement@xcite . for these schemes and others , the cavity needs to be polarization degenerate and there needs to be good mode - matching between the input light field and the field radiated by the atom@xcite .
planar photonic crystal ( phc ) microcavities are widely studied because of their small mode volumes , @xmath10 , and high quality factors , q @xmath11@xcite .
in - plane confinement is achieved by the photonic band gap introduced by the air holes etched in a high - index membrane , while out - of - plane confinement relies on total internal reflection at the membrane - air interface .
historically , phc cavities are not designed for good mode - matching ; however , far - field optimization of phc devices has been recently an active area of research@xcite . in particular , the h1 defect@xcite , formed by removing one hole from a triangular lattice of air holes , is a promising candidate for the applications described because of its orthogonally - polarized , spectrally - degenerate dipole modes .
although fabrication imperfections lift the degeneracy of the two dipole modes , some post - fabrication techniques have been implemented to restore the degeneracy@xcite . here
we show that h1 phc microcavities can be optimized for high extraction efficiency while maintaining a high q. we optimize the far - field emission patterns of our devices and suggest a compromise design in which we can achieve a collection efficiency of @xmath12 into a numerical aperture of 0.8 while maintaining a high q of 15,000 in simulation .
the maximum coupling of this cavity mode to a single mode fiber is @xmath13 .
we demonstrate experimentally , for the first time to our knowledge , the far - field emission pattern optimization and the improved coupling efficiency of h1 phc devices . this paper is organized as follows : in sec .
[ sims ] , we discuss our cavity design optimization and numerical simulation results ; in sec . [ results ] ,
we discuss our experimental results , including the far - field and coupling efficiency measurements ; and in sec .
[ conclude ] , we conclude and provide an outlook of this work .
m. ( c ) characteristic photoluminescence measurement of a fabricated device , illustrating the splitting of the h and v modes due to fabrication imperfections .
( d ) fdtd simulated near - field amplitude components for the h dipole mode ( top row ) and the v dipole mode ( bottom row ) . plotted for each mode are re(@xmath14 ) and re(@xmath15 ) .
scale bar in all plots : 250 nm .
( e ) fdtd simulated far - field radiation intensities of @xmath14 and @xmath15 components for the h dipole mode ( top row ) and the v dipole mode ( bottom row ) .
the white concentric circles correspond to na @xmath16 and @xmath17,scaledwidth=80.0% ] the h1 phc cavity design is a candidate structure for polarization degenerate light emission , and by optimizing its structural properties , we are able to dramatically improve the coupling efficiency to the cavity modes .
figure [ fignfff](a ) gives a sketch of the h1 cavity , and an sem image of a fabricated device in gaas is shown in fig .
[ fignfff](b ) .
the h1 cavity , being a point - like defect on a triangular lattice , has a @xmath18 symmetry , i.e. the geometry remains the same rotating the structure by @xmath19 around the center of the cavity .
in such a geometry , if a mode is an eigenstate of the wave equation , then any rotation of the mode by @xmath19 is also an eigenstate .
if we consider a dipole - like mode , which returns to itself only under a @xmath20 rotation , this gives six possible eigenstates , which are in general not linearly independent .
it can be shown@xcite that only two modes are linearly independent and that if @xmath21 is a dipole - like eigenstate , then , under the approximation of scalar fields , the orthogonal linearly - independent dipole - mode is : @xmath22\ ] ] where @xmath23 indicates a rotation of @xmath24 around the cavity center .
these two degenerate orthogonally - polarized dipole - modes are perfectly suited for polarization - degenerate light emission .
these considerations are valid in the ideal case of a perfect triangular lattice ; in practice , however , fabrication imperfections lift the ideal structure symmetry and remove the degeneracy of the two dipole modes .
this mode splitting is illustrated in the characteristic photoluminescence measurement in fig .
[ fignfff](c ) .
these two modes , labeled h and v , each have both @xmath24 and @xmath25 e - field components , as shown in fig .
[ fignfff](d ) . for the h mode
, @xmath15 has an antinode at the center of the cavity region , whereas the less intense @xmath14 has a node at the center , so a perfectly centered qd would primarily couple to the @xmath15 component of the h mode .
in addition , the @xmath14 far - field pattern is most intense at large angles and so would be virtually undetected by a microscope objective with an na of 0.8 or smaller , as shown in fig . [ fignfff](e ) .
therefore , we can consider the h mode to be primarily @xmath25-polarized . for the v mode , the reverse is true , and it can therefore be considered to be primarily @xmath24-polarized .
we use finite - difference - time - domain ( fdtd ) software@xcite to simulate the electromagnetic field in our phc cavities , and we perform a near- to far - field transformation@xcite using the e - field components at a plane 10 nm above the membrane surface to calculate the far - field emission profiles of the cavity modes . simulation parameters for the near - field and far - field patterns shown in fig .
[ fignfff](d , e ) are the following : lattice constant @xmath26 nm , radius @xmath27 , and membrane thickness @xmath28 nm . in fabricated devices , the phc membrane is separated from a distributed - bragg - reflector ( dbr ) or the bottom substrate by an air gap .
this bottom reflector , which significantly affects the far - field patterns@xcite and improves collection efficiency of light emitted out of the bottom surface of the membrane , is modeled in our simulations as a dielectric substrate of refractive index @xmath29 , with an air gap @xmath30 . to improve the quality factors of the modes , as well as to improve collection efficiency through modification of the far - field profiles , optimization of the lattice structure
was performed by varying the outward shift ( @xmath31 ) and smaller radius ( @xmath32 ) of the six holes nearest to the cavity and by varying the air gap separation ( @xmath30 ) between the membrane and the bottom reflector . while these optimization techniques have been employed by other groups to study different phc cavity structures@xcite
, we combine the optimization of several parameters and demonstrate the optimization experimentally in our fabricated h1 devices . here
we will primarily discuss the role of the bottom reflector and the effect of tuning @xmath33 for the h dipole mode .
a discussion of the influence of tuning @xmath34 can be found in appendix [ fdtd ] .
simulation and experimental results for the orthogonally polarized v dipole mode can be found in appendix [ vmode ] . .
simulations without a bottom reflector are represented by the blue circles ( labeled ` no r ' ) , and simulations with a bottom reflector separated by an air gap of @xmath35 nm are represented by the red squares ( labeled ` with r ' ) .
( b ) simulated collection efficiency as a function of @xmath33 for na @xmath36 0.5 ( in blue ) , 0.8 ( in red ) , and 0.9 ( in green ) .
simulations with ( without ) a bottom reflector are represented by squares ( circles ) . for simulations without the bottom reflector ,
the collection efficiency maximum is 50@xmath0 ; however , with the bottom reflector , the maximum collection efficiency is 100@xmath0 .
the ideal compromise value , @xmath37 , is indicated by the cyan ellipse , where the q is 15,000 and the coupling efficiency for an objective with na of 0.8 is 80@xmath0.,scaledwidth=55.0% ] from our fdtd results , we calculate coupling efficiency to a microscope objective and to a single - mode fiber . for a given far - field pattern ,
the collection efficiency by a microscope objective of a given na is found by dividing the power radiated within a certain angle by the total power radiated into the half - space above the membrane .
for example , a microscope objective with an na of 0.8 will collect light emitted within an angle of @xmath38 from the @xmath39 point in k - space . for simulations without the bottom reflector ,
the collection efficiency maximum is 50@xmath0 because only the light emitted from the top of the membrane is collected , with the other half being lost out the bottom of the membrane . in the case with the bottom reflector , however , all light can be collected above the membrane , so the maximum collection efficiency is 100@xmath0 .
we find the single - mode fiber coupling by calculating the mode overlap of the far - field phc mode with a gaussian of specified mode diameter , which is the guided mode of the fiber . the simulated quality factor and collection efficiency for cavities with different outward shift values are plotted in fig .
[ figqsh ] , indicating the trade - off between these figures of merit . as the collection efficiency
is improved , the q is expected to decrease as more light is being coupled to radiation modes rather than remaining confined in the dielectric slab .
the collection efficiency increases for large outward shifts ( @xmath33 greater than 0.1 ) , but the q reaches a maximum for @xmath40 .
we find that @xmath41 gives a good compromise value , demonstrating an improved collection efficiency of @xmath12 for na @xmath42 while still maintaining a high q of 15,000 .
figure [ figqsh ] compares simulations without a bottom reflector ( labeled ` no r ' ) and with a bottom reflector separated by an air gap of @xmath35 nm ( labeled ` with r ' ) . in general , the q factor of the mode can be enhanced or degraded by the inclusion of a bottom reflector , which is discussed extensively in ref .
the data indicate that for the specific air gap separation of 925 nm , the q is virtually unaffected when this bottom reflector is introduced ; however , the collection efficiency changes significantly .
this is partially due to the fact that the collection efficiency maximum is 50@xmath0 for simulations without a bottom reflector , while it can be as high as 100@xmath0 when the bottom relfector is taken into account . for the optimized @xmath41 device
, we performed a series of simulations to optimize the air gap separation , with results shown in fig .
[ figair1 ] . in this case
, we consider the trade - off between q and maximum coupling to a single mode fiber , noting that the fiber mode diameter for maximum coupling may be different for each device .
far - field radiation patterns for a selection of devices are shown in the bottom panel of fig [ figair1 ] .
we find the best compromise air gap to be @xmath43 nm , which corresponds to @xmath44 .
the maximum coupling between this device and a fiber mode with a beam waist of 1.5 @xmath45 m is 92@xmath0 .
the relevant parameter , @xmath46 will be different for each device fabricated on a wafer ; however , we chose to use @xmath35 nm for our fabricated samples . .
the best compromise , indicated by the red ellipse , occurs for @xmath35 nm .
bottom : simulated far - field radiation patterns for devices of different @xmath30 values .
each profile is normalized by its own maximum intensity , with a scale bar shown at the far right . as the air gap separation increases ,
the mode profiles become more gaussian , which show better coupling to a single - mode fiber .
the white concentric circles correspond to na = 0.2 , 0.4 , 0.6 , 0.8 , and 1.0.,scaledwidth=50.0% ] the effect of the bottom reflector on the radiation pattern can be understood by considering the interference of light that reflects multiple times between the phc membrane and the bottom reflector .
following the treatment of kim et al.@xcite , let the membrane be approximated as a uniform dielectric slab with an effective refractive index , @xmath47 , and @xmath30 and @xmath48 are the air gap thickness and the slab thickness , respectively .
waves initially propagate upward and downward from the resonant mode of wavelength @xmath49 in the membrane ; the downward propagating wave will eventually be redirected upward and will interfere with the upward propagating wave .
if we consider radiation into an angle @xmath50 , the phase change through the phc slab , @xmath51 , is given by @xmath52 and the round trip phase in the air gap is @xmath53 , where @xmath54 let @xmath55 be the sum of all waves detected upward which initially propagate downward .
it can then be shown that@xcite @xmath56 where @xmath57 is the phase changes at the bottom reflector , and @xmath58 and @xmath59 are coefficients of amplitude reflection and transmission , respectively , for a single dielectric interface .
an interference between @xmath55 and the wave initially propagating upward is characterized by @xmath60 , and the resultant radiation intensity is given by @xmath61 in fig .
[ figair2 ] , the intensity @xmath62 is plotted as a function of na and @xmath46 for the following parameters : @xmath63 , @xmath64 nm , @xmath28 nm , and @xmath65 . ) as a function of na , for different values of @xmath46 .
the calculation is done for the following paramaters : @xmath63 , @xmath64 nm , @xmath28 nm , and @xmath65 . in the bottom panel , line cuts are taken for @xmath46 = 0.93 ( left , in pink ) , 0.96 ( middle , in green ) , and 1.06 ( right , in cyan ) . plotted are a gaussian ( blue curve ) and the gaussian convolved with the function @xmath66 ( red curve ) .
the experimental air gap of @xmath30 = 925 nm , for @xmath64 nm , is represented by the middle green curve.,scaledwidth=100.0% ]
our samples were grown by molecular beam epitaxy ( mbe ) on a gaas wafer , and included a 130 nm gaas membrane layer with inas qds embedded in the middle of the membrane , a 925 nm algaas sacrificial layer , and a 10 layer algaas / gaas distributed - bragg - reflector ( dbr ) located below the sacrificial layer .
phc h1 cavities were fabricated using the following standard recipe : e - beam lithographic patterning to define the patterns in a photoresist layer , inductively - coupled plasma dry etching to transfer the pattern into the semiconductor membrane , and hydrofluoric acid wet etching to undercut the sacrificial layer , leaving a free - standing membrane .
devices were made with systematic variations of small hole sizes and outward shifts , according to the simulated far - field optimization .
experimental qs of our cavities were typically 3,000 to 8,000 .
we experimentally measure the far - field profiles of our cavities by imaging the photoluminescence emission at the back focal plane of the objective onto an intensified ccd@xcite .
we use a @xmath67 imaging setup , where @xmath68 cm is the focal length of the imaging lens ; our objective has an na of 0.8 . using a ccd exposure time of 30 to 120 seconds , spectrally filtered images
were collected of the far - field radiation pattern .
a background image , taken with a slightly tilted interference filter , was subtracted to improve the signal - to - noise ratio .
figure [ figshcomp ] shows simulated and experimental k - space far - field profiles of the h dipole mode for different outward shift values . for small @xmath33 values , the radiation intensity is spread over a large range of angles ; as @xmath33 is increased , the emitted light becomes more intense in the central region of the light cone .
it is clear that the bottom reflector ( middle row of fig .
[ figshcomp ] ) modifies the far - field profiles significantly and shows much better agreement with the experimental results than the simulations of the membrane only ( top row of fig .
[ figshcomp ] ) . as demonstrated in sec .
[ sims ] , it is sufficient to approximate the dbr below the air gap as a simple dielectric of constant refractive index for determining the ideal fabrication parameters for collection efficiency and q factor .
however , here we simulate the full 10-layer dbr structure to provide the best possible comparison with the experimental results .
small discrepancies between the experimental and simulated far - field patterns could be due to fabrication imperfections or the finite grid size of the numerical simulations . for devices with increasing outward shift values ,
the experimental far - field measurements show that more of the light is emitted at small angles , in good agreement with simulations . .
the profiles are cut off at na @xmath36 0.8 to allow for better comparison with experimental results .
middle row : fdtd simulations , including the air gap and dbr , for the same @xmath33 values as above . for increasing shift values ,
the vertical beaming of the radiation clearly improves .
bottom row : experimental far - field profiles , for one row of devices , showing a nice comparison with simulation results.,scaledwidth=80.0% ] by modifying the far - field profiles to become gaussian - like , the cavity modes are expected to be better mode - matched to an external input field using a microscope objective of reasonable numerical aperture .
this is important in c - qed applications based on the interference between the input field and the field re - radiated by the emitter in the cavity@xcite . when performing measurements with a quantum dot ensemble , rather than a single quantum dot , coupled to the cavity mode , it is difficult to extract the coupling efficiency directly from the signal intensity .
instead , the improved coupling efficiency was experimentally demonstrated using resonant reflection spectroscopy .
a tunable laser was scanned across the cavity resonance , and the reflected light was collected by the same microscope objective and passed through a polarizer before arriving at the detector .
the polarizer was adjusted to select one of the two linearly - polarized modes . in the case of perfect mode - matching
, one would expect a dip in the reflection spectrum corresponding to the cavity resonance . in the case of quantum information schemes based on dipole - induced reflection ,
the bare cavity reflectivity on resonance should be ideally zero , so that the whole reflected signal can be given by the emitter coupled to the cavity .
our experimental results are shown in fig .
[ figfano ] . for the un - optimized cavities we could not get a detectable dip above the noise level , while for far - field optimized cavities we could get a high - contrast reflection dip , with contrast up to 60@xmath0 .
the contrast of the dip gives a quantitative measure of the mode - matching between the cavity mode and the gaussian laser beam input field . in this way
, we demonstrate improved coupling efficiency for our far - field optimized devices . 0.100 , 0.115 , 0.131 , and 0.146 .
data points are in blue , and the red curves are fano lineshapes as predicted by our model , with @xmath69 = 0.95 , and the mode - matching being the only free parameter .
larger dips are seen for cavities that were optimized for gaussian - like far - field profiles .
the measured far - field profiles ( both h and v ) for each of the cavities are shown below the reflection curves.,scaledwidth=100.0% ] the reflection dips typically exhibit a fano lineshape , which have been observed previously in phc cavities@xcite and can be explained using the temporal coupled - mode theory developed by s. fan et al.@xcite let us describe the photonic crystal membrane cavity as a resonator with central frequency @xmath70 and lifetime @xmath71 , coupled to two ports , as illustrated in fig .
[ figmodel](a ) . in structures with mirror symmetry , the relation between the field in port-2 and port-1 is given by : @xmath72 = \left [ \begin{array}{cc } \sigma_{11 } & \sigma_{12 } \\
\sigma_{21 } & \sigma_{22 } \end{array } \right ] \left [ \begin{array}{c } e_1 \\
e_2 \end{array } \right]\ ] ] with the following scattering matrix : @xmath73 - \frac{(r+jt)/\tau}{j(\omega-\omega_0)+1/\tau } \left [ \begin{array}{cc } 1 & 1 \\ 1 & 1 \end{array } \right]\ ] ] the scattering matrix is composed of two contributions .
the first matrix describes the direct coupling between input and output modes , given by direct reflection or transmission by the membrane ( coefficients @xmath74 and @xmath75 , respectively ) .
the second matrix , on the other hand , describes the coupling through the resonator mode .
the reflectivity of the structure described by eqs .
( [ dip1 ] ) and ( [ dip2 ] ) is : @xmath76 in the case of @xmath77 ( or @xmath78 ) , we just see the cavity resonance , with a lorentzian lineshape . in all other cases ,
the interference between the resonance - assisted path and the path directly reflected by the membrane gives rise to a characteristic asymmetric fano lineshape . in our case
the membrane is @xmath79 nm thick , which is approximately half the central wavelength of the resonator .
a @xmath80-membrane is completely transmissive and the cavity resonance should have a lorentzian lineshape . however , our system is complicated by the presence of a dbr , placed at a distance @xmath30 from the bottom of the photonic crystal membrane ( see fig . [ figmodel](b ) ) . the dbr couples the input and output channels for port-2 : @xmath81 where @xmath69 is the fresnel coefficient of the dbr stack . combining with @xmath82
, we get a relation between @xmath83 and @xmath84 : @xmath85 and we can reduce the system to a one - port device with a reflectivity @xmath86 .
the effective reflectivity is : @xmath87 the predicted reflection spectra for our experimental parameters are plotted with our data in fig .
[ figfano ] , with the only free parameter being the mode - matching .
the predicted asymmetric reflection dips match well with our experimental observations .
in conclusion , we have performed extensive simulations to optimize the far - field emission profiles of h1 phc microcavities , and our experimental devices agree well with the simulation and theoretical results .
we have identified phc designs that provide excellent vertical beaming of the photoluminescence signal , as well as high contrast reflection dips up to 60@xmath0 . in our best compromise structure , we find that the simulated q remains high at 15,000 and the collection efficiency is @xmath12 into a microscope objective with an na of 0.8 .
the maximum mode matching of this device to a single - mode fiber is found to be 92@xmath0 .
these results demonstrate that this platform is well - suited for realizing quantum information protocols in the solid state .
, where @xmath32 represents the radius of the six nearest holes , and @xmath74 represents the radius of the other holes in the phc lattice .
( b ) simulated collection efficiency as a function of @xmath34 for na @xmath36 0.5 ( in blue ) , 0.8 ( in red ) , and 0.9 ( in green ) .
the ideal compromise value , @xmath88 , is indicated by the red ellipse , where the q is 17,000 and the coupling efficiency for an objective with na of 0.8 is 40@xmath0.,scaledwidth=40.0% ] while the main text focused on the roles of the @xmath33 parameter and the bottom reflector in tuning the far - field profiles , we will discuss here the role of @xmath34 , which can also be varied to optimize our devices .
collection efficiency and the quality factor for simulated cavities with different @xmath34 values are plotted in fig .
[ figqrsm ] , indicating the trade - off between these figures of merit .
these simulation results consider only the dielectric membrane and the surrounding air , without including the bottom reflector . as such , the maximum collection efficiency is @xmath89 , as only light emitted out of the top of the membrane can be collected by the objective .
collection efficiency increases for smaller @xmath34 values , while the q reaches a maximum for @xmath90 0.72 .
we find that @xmath90 0.64 provides a good compromise , with a collection efficiency of 40@xmath0 for a microscope objective with na @xmath36 0.8 and a q of 17,000 .
simulated far - field emission patterns are shown in fig .
[ figrsmcomp1 ] and fig .
[ figrsmcomp2 ] .
these data indicate that for small @xmath34 values , the radiation pattern is well centered within the light cone , whereas for large @xmath34 values , most of the light is emitted at large angles , with a less intense bright spot at the center of the light cone .
the bottom reflector , approximated here as a dielectric substrate of refractive index @xmath29 , again plays a crucial role in accurately simulating the far - field patterns . experimentally measured far - field patterns for two rows , @xmath91 and @xmath92 , of devices with different @xmath34 values are presented in fig .
[ figrsmcomp1 ] and fig .
[ figrsmcomp2 ] , respectively .
these data show good qualitative agreement with simulations , as the emitted light becomes more intense for small na values as the radius of the small holes is decreased .
discrepancies between the simulated and experimental far - field patterns could be due to fabrication imperfections or the approximations made in the numerical simulations .
the profiles are cut off at na @xmath36 0.8 to allow for better comparison with experimental results . for decreasing small hole size
, the radiation becomes more concentrated at small angles .
middle row : fdtd simulations , including the bottom reflector , for the same r@xmath93/r values as above .
bottom row : experimental far - field profiles , for row @xmath91 of devices , showing a nice comparison with simulation results.,scaledwidth=70.0% ] /r .
the profiles are cut off at na @xmath36 0.8 to allow for better comparison with experimental results . for decreasing small hole size
, the radiation becomes more concentrated at small angles .
middle row : fdtd simulations , including the bottom reflector , for the same r@xmath93/r values as above .
bottom row : experimental far - field profiles , for row @xmath92 of devices , showing a nice comparison with simulation results.,scaledwidth=70.0% ]
the fdtd simulation optimization and the experimental results for the v dipole mode are similar to those for the h dipole mode . as indicated in fig .
[ figshiftv ] , the far - field patterns of the v dipole mode are rotated by 90@xmath94 with respect to the h dipole mode patterns .
we measure experimentally the far - field profiles of the v dipole mode using the imaging procedure described in the main text .
these data for devices with variation in outward shift are shown and compared with simulation in fig .
[ figshiftv ] . for these comparisons ,
the bottom reflector is approximated in the simulations as a dielectric substrate of refractive index @xmath29 .
as with the h dipole mode , the experimental far - field profiles agree well with the patterns from simulations including the bottom reflector . .
the profiles are cut off at na @xmath36 0.8 to allow for better comparison with experimental results . for increasing shift values
, the vertical beaming of the signal improves . middle row : fdtd simulations , including the bottom reflector , for the same @xmath33 values as above . bottom row : experimental far - field profiles , for one row of devices , showing a nice comparison with simulation results.,scaledwidth=70.0% ]
the authors would like to thank michiel de dood for useful discussions .
this work was supported by nsf nirt grant no .
0304678 , marie curie ext - ct-2006 - 042580 and fom / nwo grant
no . 09pr2721 - 2 .
a portion of this work was carried out in the ucsb nanofabrication facility , a part of the nsf funded nnin network . | hybrid quantum information protocols are based on local qubits , such as trapped atoms , nv centers , and quantum dots , coupled to photons .
the coupling is achieved through optical cavities . here
we demonstrate far - field optimized h1 photonic crystal membrane cavities combined with an additional back reflection mirror below the membrane that meet the optical requirements for implementing hybrid quantum information protocols . using numerical optimization
we find that 80@xmath0 of the light can be radiated within an objective numerical aperture of 0.8 , and the coupling to a single - mode fiber can be as high as 92@xmath0 .
we experimentally prove the unique external mode matching properties by resonant reflection spectroscopy with a cavity mode visibility above 50@xmath0 .
[ 1][1]#1 10 k. j. vahala , `` optical microcavities , '' nature * 424 * , 839846 ( 2003 ) .
b. lounis and m. orrit , `` single - photon sources , '' rep .
. phys . * 68 * , 11291179 ( 2005 ) .
s. strauf , n. g. stoltz , m. t. rakher , l. a. coldren , p. m. petroff , and d. bouwmeester , `` high - frequency single - photon source with polarization control , '' nat . photon . * 1 * , 704708 ( 2007 ) .
s. reitzenstein and a. forchel , `` quantum dot micropillars , '' j. phys .
d : appl .
phys . * 43 * , 033001 ( 2010 ) .
m. larqu@xmath1 , t. karle , i. robert - philip , and a. beveratos , `` optimizing h1 cavities for the generation of entangled photon pairs , '' new j. phys .
* 11 * , 033022 ( 2009 ) .
p.
k. pathak and s. hughes , `` cavity - assisted fast generation of entangled photon pairs through the biexiton - exiton cascade , '' phys .
rev .
b * 80 * , 155325 ( 2009 ) .
s. m. de vasconcellos , a. calvar , a. dousse , j. suffczy@xmath2ski , n. dupuis , a. lema@xmath3tre , i. sagnes , j. bloch , p. voisin , and p. senellart , `` spatial , spectral , and polarization properties of coupled micropillar cavities , '' appl .
. lett . * 99 * , 101103 ( 2011 ) .
r. hafenbrak , s. m. ulrich , p. michler , l. wang , a. rastelli , and o. g. schmidt , `` triggered polarization - entangled photon pairs from a single quantum dot up to 30 k , '' new j. phys .
* 9 * , 315 ( 2007 ) .
a. dousse , j. suffczy@xmath2ski , a. beveratos , o. krebs , a. lema@xmath3tre , i. sagnes , j. bloch , p. voisin , and p. senellart , `` ultrabright source of entangled photon pairs , '' nature * 466 * , 217220 ( 2010 ) .
s. t. yilmaz , p. fallahi , and a. imamo@xmath4lu , `` quantum - dot - spin single - photon interface , '' phys .
rev .
lett . * 105 * , 033601 ( 2010 ) . c. y. hu , a. young , j. l. obrien , w. j. munro , and j. g. rarity , `` giant optical faraday rotation induced by a single - electron spin in a quantum dot : applications to entangling remote spins via a single photon , '' phys .
rev .
b * 78 * , 085307 ( 2008 ) . c. y. hu , w. j. munro , j. l. obrien , and j. g. rarity , `` proposed entanglement beam splitter using a quantum - dot spin in a double - sided optical microcavity , '' phys .
rev .
b * 80 * , 205326 ( 2009 ) . c. bonato , f. haupt , s. oemrawsingh , j. gudat , d. ding , m. p. van exter , and d. bouwmeester , `` cnot and bell - state analysis in the weak - coupling cavity qed regime , '' phys .
rev .
lett . * 104 * , 160503 ( 2010 ) .
e. waks and j. vu@xmath5kovi@xmath6 , `` dipole induced transparency in drop - filter cavity - waveguide systems , '' phys .
rev .
lett .
* 96 * , 153601 ( 2006 ) .
a. auff@xmath7ves - garnier , c. simon , j .-
m .
g@xmath1rard , and j .-
p .
poizat , `` giant optical nonlinearity induced by a single two - level system interacting with a cavity in the purcell regime , '' phys .
rev .
a * 75 * , 053823 ( 2007 ) .
d. englund , a. faraon , i. fushman , n. stoltz , p. petroff , and j. vu@xmath5kovi@xmath6 , `` controlling cavity reflectivity with a single quantum dot , '' nature * 450 * , 857861 ( 2007 ) .
y. akahane , t. asano , b .- s .
song , and s. noda , `` high - q photonic nanocavity in a two - dimensional photonic crystal , '' nature * 425 * , 944947 ( 2003 ) .
s .- h .
kim , s .- k .
kim , and y .- h .
lee , `` vertical beaming of wavelength - scale photonic crystal resonators , '' phys .
rev .
b * 73 * , 235117 ( 2006 ) .
n .- v .- q .
tran , s. combri@xmath1 , and a. d. rossi , `` directive emission from high - q photonic crystal cavities through band folding , '' phys
. rev .
b * 79 * , 041101 ( 2009 ) .
n .- v .- q .
tran , s. combri@xmath1 , p. colman , t. mei , and a. d. rossi , `` vertical high emission in photonic crystal nanocavities by band - folding design , '' phys .
rev .
b * 82 * , 075120 ( 2010 ) .
s. l. portalupi , m. galli , c. reardon , t. f. krauss , l. ofaolain , l. c. andreani , and d. gerace , `` planar photonic crystal cavities with far - field optimization for high coupling efficiency and quality factor , '' opt .
express * 18 * , 1606416073 ( 2010 ) .
d. pinotsi , j. m. sanchez , p. fallahi , a. badalato , and a. imamo@xmath8lu , `` charge controlled self - assembled quantum dots couple to photonic crystal nanocavities , '' photon .
nanostruct .
: fundam . appl .
* 10 * , 256262 ( 2012 ) .
s. haddadi , l. le - gratiet , i. sagnes , f. raineri , a. basin , k. bencheikh , j. a. levenson , and a. m. yacomotti , `` high quality beaming and efficient free - space coupling in l3 photonic crystal active nanocavities , '' opt .
express * 20 * , 1887618886 ( 2012 ) .
m. shirane , s. kono , j. ushida , s. ohkouchi , n. ikeda , y. sugimoto , and a. tomita , `` mode identification of high - quality - factor single - defect nanocavities in quantum dot - embedded photonic crystals , '' j. appl . phys . * 101 * , 073107 ( 2007 ) .
k. hennessy , c. h@xmath9gerle , e. hu , a. badalato , and a. imamo@xmath4lu , `` tuning photonic nanocavities by atomic force microscope nano - oxidation , '' appl . phys . lett .
* 89 * , 041118 ( 2006 ) .
i. j. luxmoore , e. d. ahmadi , b. j. luxmoore , n. a. wasley , a. i. tartakovskii , m. hugues , m. s. skolnick , and a. m. fox , `` restoring mode degeneracy in h1 photonic crystal cavities by uniaxial strain tuning , '' appl .
. lett . * 100 * , 121116 ( 2012 ) .
s .- h .
kim and y .- h .
lee , `` symmetry relations of two - dimensional photonic crystal cavity modes , '' ieee j. quantum electron .
* 39 * , 10811085 ( 2003 )
. _ lumerical solutions , inc .
_ j. vu@xmath5kovi@xmath6 , m. lon@xmath5ar , h. mabuchi , and a. scherer , `` optimization of the q factor in photonic crystal microcavities , '' ieee j. quantum electron . * 38 * , 850856 ( 2002 ) . c. bonato , j. hagemeier , d. gerace , s. m. thon , h. kim , l. c. andreani , p. m. petroff , m. p. van exter , and d. bouwmeester , `` far - field emission profiles from l3 photonic crystal cavity modes , '' photon .
nanostruct .
: fundam .
appl .
doi : 10.1016/j.photonics.2012.07.003 ( 2012 ) .
m. t. rakher , n. g. stoltz , l. a. coldren , p. m. petroff , and d. bouwmeester , `` externally mode - matched cavity quantum electrodynamics with charge - tunable quantum dots , '' phys .
rev .
lett . * 102 * , 097403 ( 2009 ) .
m. galli , s. l. portalupi , m. belotti , l. c. andreani , l. ofaolain , and t. f. krauss , `` light scattering and fano resonances in high - q photonic crystal nanocavities , '' appl .
phys .
lett . *
94 * , 071101 ( 2009 ) .
m. w. mccutcheon , p. b. deotare , y. zhang , and m. lon@xmath5ar , `` high - q transverse - electric / transverse - magnetic photonic crystal nanobeam cavities , '' appl .
phys .
lett . * 98 * , 111117 ( 2011 ) .
s. fan , w. suh , and j. d. joannopoulos , `` temporal coupled - mode theory for the fano resonance in optical resonators , '' j. opt .
soc .
am .
a * 20 * , 569572 ( 2003 ) . |
since stars are formed within the coolest molecular material of the interstellar medium ( ism ) , the star - formation rate ( sfr ) should be determined simply by the gas reservoir and by the free - fall time @xmath0 of molecular clouds ( @xcite ) .
this , however , raises a conflict between the ism conditions and observed sfrs in the sense that @xmath0 for a typical molecular cloud density of 100 @xmath1 amounts to 10@xmath2 sec , i.e. @xmath3 yrs . for
the total galactic molecular mass of @xmath4 the sfr should then amount to about 100 to 1000 , what is by orders of magnitudes larger than observed and the gas reservoir within the milky way would have been used up today .
this means , that the sf timescale must be stretched with respect to collapse or dynamical timescale by introducing a sf efficiency ( sfe ) @xmath5 and its definition could read : @xmath6 . already in 1959
schmidt argued that the sfr per unit area is related to the gas column density @xmath7 by a power law with exponent @xmath8 .
kennicutt ( 1998 ) derived from the @xmath9 luminosity of spiral galaxies @xmath10 a vertically integrated and azimuthally averaged sfr ( in @xmath11 ) , i.e. of gas disks in rotational equilibrium , and found a correlation with @xmath12 holding over more than 4 orders of magnitude in @xmath7 with a drop below a density threshold at 10 @xmath13 . while this relation establishes an equilibrium sfr ,
it is not surprising that the slope varies for dynamically triggered sf , as in starburst and merger galaxies and for high - z galaxies , when the disks form by gas infall , for the latter reaching @xmath14 ( @xcite ) . the ordinary kennicutt - schmidt ( ks ) relation can be understood by the simple analytical assumption allowing for the @xmath15 relation and for a uniform state of the ism .
its equilibrium on disk scales requires that heating processes counteract to the natural cooling of plasmas . besides the possible heating processes from dissipation of dynamical effects , as there are the differential rotation of the disk , gas infall , tidal interactions , shocks , etc .
, to the feedback by freshly produced stars , not for all of them it is obvious , how effectively they influence the sf by the expected self - regulation or vice versa trigger it .
unfortunately , the issue of a general ks - law is confused by the similarity of slopes under various stellar feedback strengths ( see e.g. sect . 2 in @xcite ) .
@xcite demonstrated already that the sfr achieves a dependence on @xmath16 , if the stellar heating is compensated by collisional - excited cooling radiation ( e.g. @xcite ) .
the coefficient of this relation determines @xmath17 .
obviously , the sf in galaxies , therefore , depends on both the gas content and the energy budget of the ism .
since the most efficient stellar energy power is exerted by supernovae ( sne ) , and here particularly by the explosions of the shortly living massive stars as type ii sne , their feedback to the ism is of fundamental relevance for the sf . in this paper , we overview and discuss the regulation of the ism by the sf feedback thru sne and more pronounced by their cumulative effect as superbubbles .
although the expression _ feedback _ of sne also includes their release of freshly produced elements , here we only focus on the dynamical and energetic issues and refer the reader interested on the chemical evolutionary consequences to @xcite .
as sn explosions were since long known to expel vehemently expanding hot gas , @xcite predicted this hot ism phase to expand from the galactic disk where it can not be bound or pressure confined to form a hot halo gas .
not before the 70 s and with the aid of observations which made formerly unaccessible spectral ranges ( from the fir / submm to x - rays ) available , the existence a hot gas component within our milky way was manifested and later also perceived in other disk galaxies . although the ism in its cool molecular component is conditioned for sf , hot gas regulates its dynamics as driver of shocks and turbulence as well as its energetics by heating thru cooling radiation and heat conduction , by this , exerting negative stellar feedback .
after these facts became internalized , a significant amount of mostly analytical explorations were dedicated to a first understanding of the ism structure , its temporal behaviour with respect to the hot ism ( comprehensively reviewed by @xcite ) , to volume filling factors and mass fractions of the gas phases . for this purpose , randomly distributed and temporally exploding sne were considered with cooling and according expansion ( @xcite ) and with transitions to / from the warm / cool gas ( @xcite ) or as non - dynamically interchanging gas phases ( see e.g. @xcite ) . in general , the action of sne on the ism is multifacetted ( @xcite ) and to a significant part affecting the sf . because sne type ii happen on much shorter timescales than type ia
, all the ism and sf feedback studies focus mostly on their energy deposit and dynamical issues . while these latter studies include arbitrary sn rates with a description of the temporal size evolution of individual supernova remnants ( snrs ) , the sf dependence on the physical state is not yet properly treated . with a toy model consisting of 6 ism components and at least 10 interchange processes @xcite also included the formation of stars from giant molecular clouds . those can only form from cool clouds which , on the other hand , are swept - up and condensed in sn shells , so that sf and sn explosions together with different gas phases and interchange processes form a consistent network . as a natural effect of this local consideration
the sf can oscillates with the corresponding timescales .
first dynamical approaches to the structure evolution of the ism and gas disks , aiming at understanding the excitation of turbulence , were performed by @xcite . as heating sources they took the energy of massive stellar winds only into account , which is at the lower range of stellar power to the ism because of their very low energy transfer efficiencies ( @xcite ) .
nevertheless , their models already demonstrate the compression of gas filaments and the expulsion of gas vertically from the disk even with the inclusion of self - gravity .
indeed , the sf recipe is not self - consistent and the vertical boundary conditions seem to affect the vertical gas dynamics artificially . in more recent numerical investigations of the ism evolution in the context of sf and sn explosions @xcite performed studies of the influence of sn feedback , self - gravity , and stars as an additional gravitational component on the sfr .
the main issues can be summarized as that feedback enhances the ism porosity , increases the gas velocity dispersion and the contrasts of t and @xmath18 , so that smaller and more pronounced structures form . since the sfr depends on most of these variables , more importantly ,
the sn feedback models reach sfrs by a factor of two higher than without feedback . at the same time , @xcite simulated the structure evolution of the solar vicinity in a box of @xmath19 size and identified the local bubble and its neighbouring loop i in their models .
in addition , filamentary neutral gas structures , resolved down to 0.625 pc , become visible and the vertical matter cycle . however , because of the lack of self - consistent sf they applied an analytical recipe of random sn events ( @xcite ) to their models that obviously underestimates the production of superbubbles from massive ob associations and , moreover , can not reasonably achieve issues on the feedback of sne on the sf .
the ism processes also do not include self - regulation effects by radiation or heat conduction , which lead to negative sf feedback .
sn and stellar wind - driven bubbles sweep up surrounding gas , condense it , and are , by this , expected to excite sf in a self - propagating manner ( abbreviated spsf ) as a positive feedback . in a cornerstone paper @xcite
have demonstrated by a parameter study , how snr expansion can lead to structure formation in galactic disks by spsf .
their main aim , however , to reproduce the general spiral arm structure thru spsf by means of the coherence of different timescales can not be reached and most galaxies develop patchy structures randomly . with a salpeter imf . the galactic disk is vertically composed of the three - phase interstellar medium , cool , warm and hot phase , respectively , with ( central density @xmath20 $ ] ; temperature t [ _ in k _ ] ; vertical scaleheight h [ _ in pc _ ] ) of ( @xmath21 ; 150 ; 100 ) , ( @xmath22 ; 9000 ; 1000 ) , ( @xmath23 ; @xmath24 ; 4000 ) .
the density varies from almost @xmath25 , in the densest part of the shell to @xmath26 in the darkest bubble interiors . _
( from @xcite ) _ , width=226 ] such sf trigger by the condensation of swept - up gas in sn or more efficiently in superbubbles ( see fig .
1 ) can be explored in detail by the investigation of the fragmentation timescale ( see e.g. @xcite ) .
ehlerova et al .
compared a self - similar analytical solution with the results of 3d numerical simulations of superbubble expansions in homogeneous media .
the amount of energy supply from the final number of young stars in an ob association , the value of the sound speed , the stratification and density of the ambient medium , the galactic differential rotation , and the vertical gravitational force in the galactic disk , all these influence the fragmentation .
the typical superbubble radius at which shells start to fragment decreases from almost 700 pc at an ambient gas density n of 1 @xmath1 to 200 pc at 10 @xmath1 .
while in thick disks like they exist in dwarf galaxies ( dgs ) nearly the whole shell fragments , in thin disks it is restricted to the galactic equator only .
the sf process itself can not be resolved in these studies and the assumption follows the line , that unstable fragments may become molecular and trigger the formation of molecular clouds in which new stars are formed .
the main conclusion is that in dgs the sf may propagate in all directions possibly turning the system as the whole into a starburst , while in spiral galaxies the sf propagates within a thin strip near the symmetry plane only .
since the applied thin shell approximation is reasonably only a 0th - order approximation , in a recent paper @xcite ( see also this volume ) clarify that the shell thickness and the environmental pressure influences the fragments in the sense that their sizes become smaller for higher pressure .
nevertheless , the deviations from the thin - shell approximations are not large .
yet as a drawback the influences of magnetic fields is not taken into account .
the perception of sf trigger in sn or superbubble shells sounds reasonable from the point of view of numerical models because ( as shown in fig .
1 ) sufficient ambient ism mass is swept up , preferably in the gas disk itself , and is capable to cooling and gravitational instabilities .
shell - like distributions of young stars , are e.g. found in g54.4 - 0.3 , called _ sharky _ ( @xcite ) , in the orion - monoceros region ( @xcite ) , more promising in the orion - eridanus shell ( @xcite ) , and in several superbubbles in the large magellanic cloud , as e.g. henize 206 ( @xcite ) . also the formation of gould s belt stellar associations in the shell of a superbubble is most probable ( @xcite ) , however , still debated ( @xcite ) .
another possible feedback effect by sn is caused , when the ultra - fast snr shock overruns a dense interstellar cloud , so that the clouds are quenched and stars are expected to be formed instantaneously .
although such cloud crushing is numerically modelled its effect on the necessary collapse of sub - clumps is not resolved simultaneously ( @xcite ) . since in several dgs excessively high sfrs
are observed ( @xcite and further more ) , it is a matter of study whether starbursts are the result of a sf self - trigger mechanism or the issue of an external process because the objects are obviously linked to large enveloping gas reservoirs from which gas infall must be invoked ( @xcite ) .
how the hot sn gas that transits from superbubbles to galactic winds interacts with infalling cold gas is yet unexplored but an attractive challenge from the present to the early universe ( @xcite ) .
high - velocity clouds ( hvcs ) in our milky way on their passage through the hot halo gas are expected to be easily disrupted due to kelvin - helmholtz ( kh ) instability .
in contrast , model clouds of self - gravitating clouds including the effect of saturated heat conduction survive over almost 100 myrs because they are mostly stabilized against kh instability ( @xcite ) .
while their compression in the stratified halo gas should not be able to avoid sf , its absence in hvcs is still a mistery , so that its understanding would provide a further insight into the positive vs. negative feedback effect of sn gas on sf in clouds .
because of their enormous power in various forms , the majority of galaxy evolution models usually take the energy deposit of sneii explosions into account as the only heating source for the ism .
although it is generally agreed that the explosive energy of an individual sn lies around 10@xmath27 ergs with significant uncertainties ( or an intrinsic scatter ) , however , of one order of magnitude , the energy deposit as turbulent and consequently as thermal energy to the ism is still more than unclear , but it is one of the most important ingredients for our understanding of galaxy formation ( e.g. @xcite ) . as a similar study , the energy release of massive stars as radiation - driven and wind - blown regions can be considered .
analytical estimates for purely radiative regions yielded an energy transfer efficiency @xmath28 of the order of a few percent ( @xcite ) .
although the additional stellar wind power @xmath29 can be easily evaluated from models and observations , its fraction that is transferred into thermal and turbulent energy is not obvious from first principles .
transfer efficiencies for both radiative and kinetic energies remain much lower from detailed numerical simulations than analytically derived and amount to only a few per mil ( @xcite , and references therein ) . nevertheless , as valid for regions and because massive stars do not disperse from the sf site , sneii explode within the stellar associations .
while massive stars contribute significantly to the ism structure formation as e.g. by cavities and holes , in the gas and chimneys of hot gas , on large scales the energy release by massive stars triggers the matter circulations via galactic outflows from a gaseous disk and galactic winds . by this , also the chemical evolution is affected thru the loss of metal - enriched gas from a galaxy ( for observations see e.g. @xcite , for models e.g. @xcite ) .
sn explosions as an immediate consequence of sf stir up the ism by the expansion of hot bubbles , deposit turbulent energy into the ism , thereby , heat the ism and regulate the sf again ( @xcite ) .
this negative energy feedback is enhanced at low gravitation because the sn energy exceeds easily the galactic binding energy and drives a galactic wind .
although investigations have been performed for the heating ( or energy transfer ) efficiency @xmath30 of sne ( @xcite ) , superbubbles ( e.g. @xcite ) , and starbursts ( @xcite ) they are yet too simplistic and mostly spatially poorly resolved to account for quantitative results .
thornton et al . derived an efficiency @xmath30 of 0.1 from 1d sn simulations as already applied by chemo - dynamical galaxy models ( @xcite ) , while unity is also used in some galaxy models ( see sect .
3 ) , but seems far too large .
although numerical experiments of superbubbles and galactic winds are performed , yet they only demonstrate the destructive effect on the surrounding ism but lack of self - consistency and a complex treatment .
simulations of the chemical evolution of starburst dgs by @xcite , 2006 ) , that are dedicated to reproduce the peculiar abundance patterns in these galaxies by different sf episodes , found , that @xmath30 can vary widely .
a superbubble expanding from a stellar association embedded in a thick hi disk has , at first , to act against the surrounding medium , by this , is cooling due to its pressure work and radiation , but compresses the swept - up shell material and implies turbulent energy to the ism . here
the superbubble expansion is efficiently hampered , but depends on the hi disk thickness ( @xcite ) and the energy loss by radiative cooling .
in fig.2 the temporal behaviour of the above - mentioned starburst dg model by @xcite ) is displayed .
comparison of the various kinds of energy contents reveal an heating efficiency @xmath30 of to about 18% . while the accumulated snii energy release reaches @xmath31 ergs after 20 myrs , the thermal energy content starts at @xmath32 drops and increases successively but not above @xmath33 ergs . for the subsequent snia explosions , always single events , the accumulated energy release is clearly discernible while the thermal energy decreases again and varies below @xmath32 ergs so that here @xmath34 . moreover ,
if a closely following sf episode pushes its sneii into the already existing chimney of a preceding superbubble , the hot gas can easily escape without any hindrance and thus affects the ism energy budget much less .
@xcite found that depending on the external density the chimneys do not close before a few hundred myrs .
numerical simulations of galaxy evolution are only feasible with intermediate - scale spatial resolution so that sf can not be resolved and must already be prescribed by reasonable recipes .
numerous papers have implemented sf criteria , such as e.g. threshold density @xmath35 with sf if @xmath36 , excess mass in a specified volume with respect to the jeans mass , i.e. m@xmath37 , convergence of gas flows ( @xmath38 ) , cooling timescale @xmath39 , temperature limits t@xmath40 t@xmath41 , toomre s q parameter , temperature dependent sfr , but to some extent also already under the assumption of the ks law ( @xcite ) .
this latter , however , is unjustified for non - equilibrium situations , because the sf should converge to this relation due to self - regulation .
if at least some of these conditions are fulfilled , as a further step , the gas mass which is converted into stars , i.e. the sfe , has to be set e.g. by fixing an empirical value @xmath5 from observations , i.e. @xmath42 , where @xmath43 is e.g. the mesh volume in a grid code .
if the numerical timestep @xmath44 is smaller than the dynamical timestep @xmath0 , @xmath45 has to be weighted by this time ratio ( @xcite ) . since @xmath5 must inherently depend on the local conditions so that it is high in bursting sf modes , as requested for the globular cluster formation , but of percentage level in the self - regulated sf mode , numerical simulations often try to derive the realistic sfe by comparing models of largely different @xmath5 with observations , as e.g. to reproduce gas structures in galaxy disks and galactic winds ( e.g. @xmath5 = 0.05 and 0.5 in @xcite , 2008 ) .
in addition , @xmath30 by them is fixed to 10@xmath27 erg per 55 @xmath46 of formed stars ( @xmath47 erg @xmath48 by @xcite ) , so that these results can be treated indicatively but not yet quantitatively , since they also mismatch with the ks relation .
theoretical studies by @xcite achieved a dependence of @xmath5 on the external pressure , while @xcite explored a temperature dependence of the sfr .
furthermore , most galaxy evolutionary models at present lack of the appropriate representation of the different ism phases allowing for their dynamics and their direct interactions by heat conduction , dynamical drag , and dynamical instabilities thru forming interfaces , not to mention resolving the turbulence cascade .
the dominating influence of sn explosions and superbubbles on structure , dynamics , and energy budget of the ism are obvious and agreed . signs and strengths of these feedback effects are , however , widely uncertain . whether the feedback is positive ( trigger ) or negative ( suppression ) can be understood analyticly from first principles , but because of the non - linearity and the complexity level of the acting plasmaphysical processes clear results can not be quantified reliably .
in addition , the temporal behaviour varies by orders of magnitude because of the changing conditions . in summary ,
the energy transfer efficiencies of sne and superbubbles to the ism are much below unity and depend on the temporal and local conditions , but must not be overestimated .
spatially and temporarily resolved simulations of sn and superbubbles in an extended environment with varying conditions of the ism are necessary in order to connect large - scale effects on sf clouds with the existing detailed simulations on the star - forming scales .
the author is grateful to simone recchi for substantial contributions to this topic and providing fig .
the attendance of the symposium was funded by the key programme
computational astrophysics of the university of vienna under project no . fs538001 . | supernovae are the most energetic stellar events and influence the interstellar medium by their gasdynamics and energetics . by this ,
both also affect the star formation positively and negatively . in this paper , we review the development of the complexity of investigations aiming at understanding the interchange between supernovae and their released hot gas with the star - forming molecular clouds . commencing from analytical studies
the paper advances to numerical models of supernova feedback from superbubble scales to galaxy structure .
we also discuss parametrizations of star - formation and supernova - energy transfer efficiencies .
since evolutionary models from the interstellar medium to galaxies are numerous and apply multiple recipes of these parameters , only a representative selection of studies can be discussed here .
2/pc^2 a&a |
quantum chromodynamics at low energy remains enigmatic chiefly because of an almost complete lack of knowledge of the properties of soft glue .
glue must certainly be understood if phenomena such as color confinement , mass generation , and dynamical symmetry breaking are to be understood .
the discovery and explication of hadrons with excited gluonic degrees of freedom is clearly an important step in this process .
furthermore , the search for nonperturbative glue , in particular as manifested in hybrid mesons , would be greatly facilitated by a rudimentary knowledge of the hybrid spectrum and decay characteristics .
although it appears that lattice estimates of light quenched hybrid masses are forthcoming@xcite , hadronic decays remain difficult to calculate on the lattice .
thus one is forced to rely on model estimates of the couplings of hybrids to ordinary mesons .
historically , there have been two approaches to such estimates .
the first assumes that hybrids are predominantly quark - antiquark states with an additional constituent gluon@xcite and that decays proceed via constituent gluon dissociation@xcite .
the second assumes that hybrids are quark - antiquark states moving on an adiabatic surface generated by an excited flux tube " configuration of glue@xcite .
decays then proceed by a phenomenological pair production mechanism ( the @xmath3 model " ) coupled with a flux tube overlap@xcite .
an important feature of this model is that the quark pair creation vertex is uncorrelated with the gluonic modes of the hybrid .
a third possibility for hybrid decay has been recently introduced @xcite .
this model also assumes flux tube hybrids but employs a different decay vertex .
the vertex is constructed by using the heavy quark expansion of the coulomb gauge qcd hamiltonian to identify relevant operators .
the gluonic portion of these are then evaluated using a slightly extended version of the flux tube model of isgur and paton@xcite .
the essential new feature is that the gluon field operator is expressed in terms of the nonperturbative phonon modes of the flux tube model rather than traditional plane waves .
this paper begins with a review of the development of the decay model of ref .
@xcite and describes in detail several issues which arise in converting the amplitudes to decay widths .
we then summarize the main general features of the model and compare these with the flux tube decay model of isgur , kokoski , and paton ( hereafter referred to as ikp ) .
the main portion of this work is a comprehensive review of the decay modes of all low lying isovector , isoscalar , @xmath4 , @xmath5 , and @xmath2 @xmath6 , @xmath7 , @xmath8 , and @xmath9 hybrids .
a detailed discussion of interesting features in the phenomenology of these states follows .
the first step in the construction of any hybrid decay model is determining what is meant by a hybrid .
we stress that choosing a model of hybrids with the correct degrees of freedom is crucial because decays probe the internal structure of the participating particles .
thus for example , in the flux tube model low lying vector hybrids must have the quarks in a spin singlet and this implies that vector hybrids may not decay to a pair of spin zero mesons ( see below for further discussion of this point ) .
however , this need not be true in a model which assigns hybrid quantum numbers differently ( for example , it is possible to construct spin one vector hybrids in constituent glue models ) . in this work ,
we choose to employ a slightly modified version of the flux tube model hybrids of isgur and paton , as described in refs .
recent lattice calculations of adiabatic hybrid potential surfaces show that the flux tube model does a good job of describing the level orderings and degeneracies apparent in the data ( although it does not reproduce many details)@xcite .
thus one may be confident that the model captures the essential features of ( heavy ) hybrid structure necessary for the construction of a viable decay model . the flux tube model of isgur and paton @xcite
is extracted from the strong coupling limit of the qcd lattice hamiltonian .
the hamiltonian is first split into blocks of distinct topologies " ( in reference to possible gauge invariant flux tube configurations ) and then adiabatic and small oscillation approximations of the flux tube dynamics are made to arrive at an n - body discrete string - like model hamiltonian for gluonic degrees of freedom .
this is meant to be operative at intermediate scales @xmath10 where the strong coupling is of order unity .
the lattice spacing is denoted by @xmath11 , the string tension by @xmath12 , and there are @xmath13 `` beads '' ( or links ) evenly spaced between the @xmath14 pair . diagonalizing the flux tube hamiltonian yields phonons , @xmath15 , which are labelled by their color ( @xmath11 ) , mode number ( @xmath16 ) , and polarization ( @xmath17 ) .
a hybrid may be built of @xmath18 phonons in the mth mode with polarization @xmath19 .
in particular , hybrid states with a single phonon excitation are constructed as @xmath20 spin and flavor indices have been suppressed and color indices are explicit .
the factor @xmath21 in the hybrid wavefunction projects onto states of good parity and charge conjugation .
the quantum numbers of these states are given by @xmath22 and @xmath23 where @xmath24 and @xmath25 .
these expressions differ from isgur and paton@xcite because we have adopted the standard definitions for the polarization vectors and the wigner rotation matrix , following the jacob - wick conventions .
we shall consider low - lying hybrids only so that @xmath26 in what follows .
it remains to specify the structure of the decay operator .
to leading order in the hopping parameter and strong coupling expansion , one can show that the operator for producing a @xmath27 pair has the following structure @xcite , @xmath28 the dependence on the relative distance , @xmath29 , comes from integrating @xmath30 products of the link operators from the kinetic term @xmath31 over a straight line in direction of @xmath32 .
the prefactor @xmath33 can be identified with the schwinger tunneling factor for pair production in an external field of the parent @xmath34 meson . in our picture , hybrids are characterized by excitations of the gluonic field .
we will therefore assume that hybrid decays can proceed through local deexcitation of this field rather then by quark tunneling in the external field of the meson source .
thus the expectation value of the gluon operator in @xmath35 between excited ( hybrid ) and deexcited ( low lying meson decay products ) is used to obtain the effective @xmath34 production operator . in @xcite the chromoelectric @xmath36 and @xmath37 fields
have been mapped onto the flux tube space of gluon excitations described by the phonon operators .
using these expressions together with @xmath38 one then obtains @xmath39 this yields the following effective decay operator @xmath40 where the @xmath41 are polarization vectors orthogonal to @xmath42 .
the integral is defined along the @xmath43 axis only .
integration over the transverse directions yields the factor @xmath44 which may be interpreted as the transverse size of the flux tube .
note that the phonon operators represent gluonic excitations which are perpendicular to the @xmath43 axis .
although this appears problematical in traditional perturbation theory , it is required here because , in the adiabatic limit , the gluonic field configuration must be defined in terms of the quark configuration and therefore the field expansion of the vector potential depends on the quark state under consideration .
the decay amplitude for a hybrid @xmath45 into mesons @xmath46 and @xmath47 is then given by : @xmath48\end{aligned}\ ] ] where @xmath49 is the matrix element of the pauli matrices between quark spin wavefunctions , @xmath50 this amplitude should be multiplied by the appropriate flavor overlap and symmetry factor .
the evaluation of the matrix elements is greatly facilitated by performing the angular integrals analytically .
this may be achieved through use of the relations @xmath51 and @xmath52 the resulting expression completely factorizes from the radial and flux tube integrals except for a trivial dependence on the wave in the final channel , greatly simplifying the algebra .
we note the following general properties of the decay amplitude .
the operator is nonzero only along the hybrid @xmath43 axis as follows from the structure of the interaction hamiltonian .
thus @xmath34 creation occurs on a line joining the original @xmath43 quarks , smeared over the transverse size of the flux tube .
this is in contrast to the model of ikp which has transverse extent and a node along the @xmath43 axis .
furthermore the spin operator contracts with the flux tube phonon polarization vector , which is absent in the ikp model . finally , the decay amplitude vanishes when the final mesons are identical due to the nodal structure in the vector potential .
this is true for any single - phonon hybrid in an odd mode .
thus one obtains the selection rule : low - lying hybrids do not decay to identical mesons .
this subsumes the selection rule of ikp so that none of their qualitative conclusions are changed .
however we also predict , for example , that hybrids do not decay to pairs of identical p - wave mesons .
this rule has recently been shown to be more general than specific models @xcite .
the preferred decay channels are to @xmath53 wave pairs@xcite .
we stress that the selection rule forbidding @xmath54 wave final states no longer operates if the internal structure or size of the two s wave states differ @xcite .
another rule , the spin selection " rule , exists : if the @xmath55 in either hybrid or conventional mesons are in a net spin singlet configuration then decay into final states consisting only of spin singlet states is forbidden .
this rule follows because pair creation is spin - triplet .
it appears to be a universal feature in all non relativistic decay models . for @xmath56 states this selection rule distinguishes between conventional vector mesons which are @xmath57 or @xmath58 states and hybrid vector mesons where
the @xmath55 are coupled to a spin singlet .
for example , it implies that in the decay of hybrid @xmath59 , the channel @xmath60 is forbidden whereas @xmath61 is allowed ; this is quite opposite to the case of @xmath62 conventional mesons where the @xmath61 channel is relatively suppressed and @xmath60 is allowed @xcite .
the extensive analysis of data in ref .
@xcite revealed the clear presence of @xmath63 @xcite with a strong @xmath61 mode but no sign of @xmath60 , in accord with the hybrid situation .
there are a number of amplitudes that vanish for the sho wave functions employed here in addition to those governed by the selection rules above .
some of these decays vanish simply due to quantum numbers , e.g. @xmath64 to two vector mesons ( see the proof in appendix 1 of ref .
@xcite ) . some amplitudes vanish in both this work and the ikp model .
these include all f wave amplitudes for hybrid decay to two s wave mesons , and all g wave amplitudes .
also , @xmath65 hybrid decays to two vector mesons vanishes .
in addition , the decays @xmath66 and @xmath67 ; @xmath68 ; and @xmath69 vanish . alternatively , in the ikp model @xmath70 and
@xmath71 vanish .
the final step is to calculate hybrid widths .
this involves choosing prescriptions for evaluating the decay phase space , the vertex coupling @xmath72 , and wavefunction parameters .
the choice of the appropriate phase space is , unfortunately , a difficult issue to resolve ( it is discussed extensively in @xcite ) .
for example , in our conventions standard relativistic phase space evaluates to @xmath73 where @xmath74 is the energy of meson @xmath46 in the final state .
this can differ substantially > from the nonrelativistic version : @xmath75 especially when pions are in the final state . finally , we mention a third possibility employed by kokoski and isgur @xcite , called the ` mock meson ' method .
the authors use @xmath76 where @xmath77 refers to the ` mock meson ' mass of a state .
this is defined to be the hyperfine - splitting averaged meson mass . in practice ,
the numerical result is little different from the relativistic phase space except for the case of the pion , where a mock mass of @xmath78 gev is used .
the net effect on low lying meson decays is to enhance the decay for processes with pions in the final state by a factor of @xmath79 for each pion in the final state .
this procedure improved the fit to experimental data substantially .
in fact , it is generally true that the @xmath3 model ( with relativistic phase space ) fits the data quite well except for the case where pions are in the final state .
we have adopted a different approach to phase space which also solves this problem and which we believe is better physically motivated .
we suggest @xcite that the root of the problem lies in the goldstone boson nature of the pion .
this implies that a pion is not a simple @xmath14 state , but rather is collective in nature .
an explicit way to incorporate this physics into a constituent quark model has been suggested by several groups@xcite .
the method relies on constructing a nontrivial vacuum for qcd which breaks chiral symmetry .
the pion may then be manifested as a goldstone mode by using the random phase approximation ( rpa ) to construct it .
the point of interest to the current discussion is that in the random phase approximation the pion wavefunction contains backward moving pieces .
these pieces allow new contributions to meson decay diagrams when pions are in the final state . in the chiral limit , the net result is quite simple : amplitudes with two pions in the final state should be multiplied by 3 ( over the naive quark model result ) , while those with a single pion in the final state should be multiplied by 2 .
the efficacy of this prescription is illustrated in table 1 .
ccccc & @xmath80 & @xmath81 & @xmath82 & @xmath83 no rpa & 0.71 & 0.53 & 0.46 & 0.42 rpa & 0.24 & 0.26 & 0.23 & 0.21 [ tablei ] as can be seen , the improvement is dramatic .
precisely the same argument applies to hybrid decays .
thus our prescription is as follows : use relativistic phase space and the rpa pion factors mentioned above to arrive at the final decay amplitudes .
the work of ikp was greatly expanded in close and page@xcite ; since one of the purposes of this work is to compare this model with ikp , we have both quoted the results of close and page below and have used their meson and hybrid meson wavefunction parameters as our standard parameters " ( these are discussed in the appendix ) .
note that in order to calculate the ikp model predictions given below , we use the same normalization as in ref .
@xcite , which corresponds to the @xmath3 pair creation parameter @xmath84 favored for mock meson phase space @xcite .
although @xmath85 is preferred for relativistic phase space @xcite , ref .
@xcite used @xmath86 for high mass meson resonances .
we simply choose to retain @xmath84 .
the normalization of this model is fixed to give the same average width as the ikp model for the decays of isovector hybrids to @xmath87 and @xmath88 with the `` standard parameters '' .
this yields a coupling of @xmath89 gev@xmath90 .
these particular decay modes were chosen because the two models can analytically be shown to mimic the predictions of each other in decays to two ground state or radially excited s wave final states .
thus decays to these final states may be regarded as `` model invariant '' . finally , as discussed above , we note that the absolute widths in the ikp model could be up to @xmath91 times bigger than the widths quoted here . furthermore , since phase space conventions and absolute magnitude conventions have changed since former ikp model calculations @xcite care should be taken with comparisons . indeed , the authors of ikp state that a model error of ( an additional ) factor of 2 should be allowed for in their predicted widths . to make contact with the original development of this model@xcite and to illustrate the parameter dependence of the model predictions
, we also employ the parameters of ref .
@xcite as an alternative parameter " set .
this set was normalized to the experimental decay pattern of the hybrid meson candidate @xmath92 , yielding @xmath93 gev@xmath90 .
these parameters are also listed in the appendix .
simple harmonic oscillator ( sho ) wavefunctions are used throughout for the final state mesons .
this is typical of decay calculations and it has been demonstrated that using coulomb + linear wavefunctions does not change the results significantly@xcite . we have taken the following masses for the @xmath94 , @xmath95,@xmath96 , and @xmath97 hybrids : 1.8 , 2.0 , 4.1 , and 10.7 gev respectively .
masses for known mesons are taken from ref .
@xcite and otherwise > from ref .
the quark model assignments for the mesons are those of the pdg tables @xcite .
the @xmath98 is assumed to be the scalar @xmath99 state .
we assume the @xmath100 @xmath95 mesons to be @xmath101 respectively .
thus @xmath98 denotes a generic scalar state at 1.37 gev , containing either light quarks or @xmath95 , depending on the context .
the flavor structure of the @xmath102 is taken to be @xmath103 at 547 mev and @xmath104 is @xmath105 at 958 mev .
the @xmath106 and @xmath107 are assumed to be @xmath108 and @xmath95 respectively , with @xmath107 the second @xmath109 peak at 1490 mev .
@xmath110 is not well established .
@xmath111 denotes the pdg state @xmath112 . @xmath113 and @xmath114 are the low and high mass @xmath115 states respectively .
the high mass state can be identified with the pdg state @xmath116 .
as stated earlier , we employ relativistic phase space and rpa pion factors and work in the narrow resonance approximation .
we also extend the rpa prescription to kaons and @xmath102s ; but not to the @xmath117 .
decay modes include all possible charge combinations , e.g. @xmath88 means @xmath118 and @xmath119 . in the following tables we present the dominant widths for hybrid @xmath120 for various @xmath121 hybrids in partial wave @xmath122 .
column 1 indicates the @xmath121 of the hybrid , column 2 the decay mode and column 3 , @xmath122 . in columns 4 , 5 , 6 and 8
we indicate predictions of this model .
column 6 uses the `` standard parameters '' used throughout the text and defined in the appendix .
column 5 uses the same parameters , except that all hybrids are assumed to be @xmath123 gev heavier ( and the @xmath96 hybrids @xmath124 gev heavier to put them above the @xmath125 thresholds at approximately @xmath126 gev ) . column 4 uses the `` alternative parameters '' .
columns 4 and 6 should hence be compared to estimate parameter sensitivity of our predictions . for hybrid decays to two ground state s wave mesons
we indicate the `` reduced width '' in column 8 .
this is the width divided by the dimensionless ratio @xmath127 , where @xmath128 is the inverse radius of the sho wave function @xcite .
it gives a measure of how strong the decay is with the difference of the wave functions explicitly removed . in column 7
we give ikp model predictions for the `` standard parameters '' , so that columns 6 and 7 should be compared when this model is compared with the ikp model .
as stated earlier , we omit f wave amplitudes for hybrid to two s wave mesons , and all g wave amplitudes , since these vanish in both models . we do not list decays with two s wave mesons in the final state which have identical wave functions ( e.g. @xmath129 ) , since these amplitudes vanish due to the `` s+s '' selection rule .
the symbol `` '' indicates that an amplitude is exactly zero , not only numerically small .
finally , a dash indicates that a decay mode is below threshold ( recall that we work in the narrow width approximation ) .
.isovector hybrid decay modes [ cols="<,^,^,^,^,^,^,^",options="header " , ] [ table6 ]
we proceed to discuss the phenomenology of mainly isovector hybrids made from @xmath130 flavored quarks for each @xmath121 , as these are expected to be the easiest to isolate experimentally . 0.1 in it was argued in refs .
@xcite that the @xmath63 and the @xmath132 can not be accommodated within the phenomenologically successful @xmath3 decay model as conventional mesons
a hybrid component is needed .
this conclusion depends strongly on the results of the influential data analysis of ref .
the central problem is that the substantial experimental @xmath133 mode @xcite can not be accommodated along with other modes of @xmath63 if the state is @xmath134 or @xmath58 quarkonium .
however , if the experimental @xmath133 width of 190 mev @xcite can be reduced by 50% , the @xmath63 can be fitted as @xmath135 @xcite .
the ikp model predicted that @xmath133 would be the largest decay mode of a hybrid , consistent with the data .
it is of interest to examine these conclusions here . for an isovector @xmath131 state at 1.5 gev
we calculate for `` standard parameters '' the widths xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx @xmath136@xmath137@xmath138@xmath133 + this work62.615mev + ikp model51.343mev + where both models predict @xmath139 and @xmath140 to vanish . for an isoscalar state at 1.5 gev xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx @xmath88@xmath141@xmath138 + this work201.6mev + ikp model171.3mev + where both models predict @xmath142 and @xmath143 to be negligible .
the predictions for the models are very similar , except that the @xmath133 mode of the isovector state is smaller in this model .
however , the ordering of modes according to their relative sizes remains the same , and @xmath133 remains the dominant channel .
it is clear that it becomes difficult to support the huge experimental @xmath133 mode in both models . in the light of this
we urge quantification of this mode at da@xmath144ne and jlab ( and at a coupled channel analysis currently in progress at crystal barrel@xcite ) .
if the @xmath63 has indicated the existence of the vector hybrid nonet , then we need to establish which of the other seven multiplets expected nearby should also be visible .
states whose couplings are predicted to be strong , with highly visible decay channels and moderate widths relative to the @xmath145 candidate , _ must _ be seen if hybrids are to be established .
conversely , channels where no signals are seen should be those with signals which are predicted to be weak . 0.2 in the clearest signature for a hybrid meson is the appearance of a flavored state with exotic @xmath121 .
it was noted in the ikp model @xcite that the isovector @xmath146 width is predicted to be large ( over 600 mev according to table [ table2 ] ) . here
the width is 100 250 mev depending on parameters , making the state narrower .
however , as shown in table [ table2 ] , if the mass of the state increases , the width may increase dramatically .
there are accordingly two likely reasons why this state has not yet been observed : ( i ) its mass is higher than 1.8 gev , making it very wide .
this possibility is underpinned by recent lattice gauge theory calculations supporting a mass difference of @xmath147 gev between @xmath146 and the lowest lying @xmath148 hybrid @xcite .
( ii ) its decay modes are idiosyncratic .
it can be seen from the table that decays are only to @xmath53 wave states , most likely to @xmath149 and @xmath150 .
however , @xmath151 , @xmath152 , and @xmath153 are broad states , making the @xmath146 difficult to isolate .
the isovector @xmath154 was predicted to be broad in the ikp model ( @xmath155 mev ) @xcite .
this is especially true if the mass of the state increases , as indicated by lattice gauge theory calculations , which suggest a mass difference of @xmath156 gev between @xmath154 and @xmath148 levels @xcite .
however , in this model we discover a radically different result : @xmath154 is @xmath157 mev wide and rises to only @xmath158 mev at 2 gev .
the total width of the @xmath154 hence forms a strong test for the model .
part of the difficulty to detect the @xmath154 may be that decays to @xmath159 wave states only occur in d wave , and that decay modes like @xmath160 and @xmath150 contain broad p
wave states .
however , in view of the possible narrowness of this state , we urge experimenters to allow for the exotic @xmath154 wave in partial wave analyses .
particularly , @xmath161 should be studied .
an excellent opportunity for isolating exotic hybrids occurs in the @xmath148 wave .
recently , there has been several experimental claims for @xmath148 signals , most notably by brookhaven and ves , in two distinct mass regions : ( i ) refs .
@xcite sees a broad structure in the mass region @xmath162 gev in @xmath163 , which is suggestive of being a composite of two objects at @xmath164 and @xmath165 gev .
it is the latter that appears to have a resonant phase though they admit that more data is required for a firm conclusion .
( ii ) ref .
@xcite claims a resonance at @xmath166 mev with width @xmath167 mev and ref .
@xcite a `` preliminary '' resonance at @xmath168 gev with width @xmath169 gev .
we hence study model predictions for @xmath148 states at 1.6 and 2.0 gev .
our expectations for a @xmath170 hybrid at 2.0 gev are ( in mev ) [ bnl2 ] xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx @xmath143@xmath171@xmath172@xmath88@xmath173@xmath163@xmath174@xmath175 + this work4333271612107 7 + ikp model5875211612381319 + where we have neglected @xmath176 , and @xmath177 modes which are predicted to be smaller than 5 mev in both models .
furthermore , the @xmath178 , and @xmath179 modes are all negligible in both models . because of the substantially increased phase space available relative to a 1.6 gev hybrid candidate , @xmath180 channels are dominant .
the model has several modes suppressed relative to the ikp model .
also note in addition to the important @xmath143 channel , @xmath171 emerges as prominent channel , leading us to suggest the search channel @xmath181 . for a @xmath148 state at 1.6 gev
one has [ bnl1 ] xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx @xmath143@xmath88@xmath163@xmath172@xmath138 + this work2495 2.8mev + ikp model598141.4mev + where both models predict @xmath182 and @xmath183 to be 0 mev .
superficially , the main effect of this model is to make the @xmath180 modes of a more similar size to the @xmath159 modes than they are in the ikp model , in agreement with the clear presence of the experimental state in @xmath88 .
however , this conclusion is parameter dependent ( compare columns 4 and 6 in table [ table2 ] ) .
nevertheless we emphasize the importance of searching for the hybrid in @xmath88 , as well as in the @xmath143 and @xmath163 channels . also , both models concur that @xmath143 should be primarily focused upon .
such a search has been proposed and conditionally approved at jlab @xcite .
although both models underpredict the total experimental width at @xmath184 mev , we do not consider this significant at the level of accuracy expected of this model , especially in view of the fact that not all possible decay modes have been calculated .
the strong dependence of the partial widths on the hybrid mass is displayed in fig .
note that the s+p " selection rule forces this to be true for any hybrids in the 2 gev mass range because decays may only occur to final states near threshold .
[ figrho ] it is significant that there is no experimental evidence for hybrids in the 1.6 2 gev region in @xmath185 @xcite which is consistent with the predictions of both models , and is in fact model
independent due to a relativistic symmetrization selection rule @xcite . in this context , searches in @xmath185 at jlab and bnl @xcite could be disappointing .
more experimental work is needed to clearly establish whether both @xmath148 signals are solid , and more detailed knowledge of branching ratios are necessary in order to compare our predictions with experiment .
0.2 in 0.2 in 0.1 in an important model distinction emerges for @xmath186 hybrids : we predict widths of approximately 100 mev , while the ikp model predicts widths larger than @xmath187 mev .
we shall argue below that the experimental evidence for the @xmath188 indicates that if it is regarded as a single resonance , then it is not a @xmath186 hybrid . within this model _ either _ both the conventional meson and the hybrid are produced , with the hybrid weaker , _ or _
the @xmath186 hybrid is higher than 2 gev in mass , which would push its width to more than 200 mev . in either case
we expect the dominant decay channel in this model to be to @xmath189 or @xmath190 ( and if phase space allows @xmath177 ) .
another experimental challenge would be considering the possibility of two resonances in the @xmath191 gev mass region .
we now argue that the experimental evidence for the @xmath188 is consistent with it being a conventional meson . here
we assume for simplicity that the @xmath188 is a single resonance , independent of the channel it is observed in .
current experimental data does not allow us to go beyond this assumption .
it was noted in ref .
@xcite that the large d wave to s wave ratio for @xmath88 amplitudes found by ves is consistent with expectations for a @xmath192 conventional meson .
it is clear from table [ table2 ] that the large d
wave is not explicable for a hybrid in this model or in the ikp model .
@xcite also predicted a @xmath88 width of @xmath193 mev for @xmath192 , while we expect a @xmath88 width of @xmath194 mev for a @xmath195 gev state .
this is consistent with the @xmath192 being strongly produced via the @xmath88 production vertex sampled at ves .
this , together with the stronger @xmath163 width of the @xmath192 ( @xmath196 mev ) , is consistent with the state observed in @xmath163 @xcite being the @xmath192 .
ves also reported possible evidence for the @xmath197 mode . since the predicted @xmath197 width of @xmath192 quarkonium is 2 mev and that of a hybrid is 0 mev , this supports the weakness of the mode observed .
recently , ves has reported the observation of a structure in @xmath198 at 1.8 gev that can be identified with the @xmath188 , coupling to the @xmath143 and @xmath199 channels @xcite . in both the current model and the ikp model ,
this is inconsistent with the hybrid interpretation , as the coupling of the hybrid to @xmath143 and @xmath199 is expected to vanish .
in fact , ves reports an absence of @xmath199 s wave @xcite , inconsistent with the hybrid interpretation where the s wave dominates the d wave ( see table [ table2 ] ) , but consistent with the @xmath200 model prediction that the s
wave should be zero ( see eq .
( a53 ) of ref .
@xcite ) . moreover , the @xmath163 channel is dramatically suppressed for the hybrid in this model in contrast to the ikp model . in summary , if we assume that @xmath188 is a single resonance , it is consistent with being a conventional meson . within this assumption
, it should be counted as one of the successes of this approach that we can explain the non observation of the @xmath186 hybrid in a way the ikp model can not . 0.2 in 0.1 in it is clear from table ii that the predictions of this model and the ikp model are very similar , except for @xmath197 which can vary substantially depending on parameters .
@xcite concluded that the @xmath92 can not be understood as a conventional meson in the @xmath3 model .
@xcite concluded that the @xmath92 can be interpreted as a hybrid meson in the ikp model .
the current work does not change these conclusions .
@xcite contains a calculation of the widths of the @xmath92 in this model which include below threshold decays to @xmath202 of 85 mev .
mode predicted in this model is expected to couple to @xmath203 via @xmath204 final state interactions , which are known be substantial experimentally , so that this model estimate is actually less than 85 mev .
] it is useful to correlate the decay modes to experimentally known ratios .
specifically , using the ves experimental branching ratios in s wave is assumed to arise solely from @xmath202 . ] @xcite and correcting for decays of particles into the specific channels observed by ves @xcite , we obtain [ bnl2 ] xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx @xmath202@xmath197@xmath88@xmath138@xmath199 + experiment@xmath205 @xmath206 @xmath207 @xmath208@xmath209 + this [email protected] + where the model widths evaluated for the `` alternative parameters '' have been scaled by a common factor to allow comparison to the experimental ratios deduced in ref .
the correspondence is remarkable .
we emphasize that although @xmath88 is suppressed in the data , we expect the resonance to have a non negligible coupling to this channel .
the total width is expected to be @xmath211 mev and is consistent with the experimental width , since the decay modes @xmath212 and @xmath213 that are known to occur @xcite have not been computed here and are experimentally known to give substantial additional contributions @xcite .
one inconsistency with ves data is the @xmath199 mode .
it is significant that the resonance in @xmath199 has a mass @xmath214 gev , shifted significantly downward from the usual @xmath92 mass parameters , and that there are indications of the presence of a broad @xmath201 wave @xcite .
this may signal the presence of @xmath215 light quark state expected at 1.88 gev @xcite with dominant decay to @xmath199 @xcite , removing the apparent inconsistency with the hybrid interpretation of @xmath92 .
important tests are now that there should be a measureable coupling to the @xmath88 channel with only a small @xmath216 or @xmath138 contribution . 0.2 in 0.1 in we expect both isovector and isoscalar @xmath66 hybrids to be narrow , and they should hence be seen .
the difference between the predictions of our approach and the ikp model does not appear to be substantial , especially when parameters are allowed to vary ( see table [ table2 ] ) .
the most striking difference between the models is the isovector @xmath66 decay to @xmath143 , which this model finds exactly zero .
however , it is fairly small in the ikp model too .
> from the selection rule forbidding the decay of a spin singlet meson into pairs of spin singlets , it follows that the decay of @xmath217 is prevented .
hence the @xmath143 channel may not be a strong discriminant between hybrid and conventional @xmath66 , as previously suggested @xcite .
recent ves data on the @xmath66 in @xmath143 does appear to indicate a structure at @xmath218 gev , but no firm conclusions are possible at this stage @xcite .
the phenomenology of the @xmath66 discussed in refs .
@xcite suffices at this stage : isovector decays to @xmath88 and @xmath183 and isoscalar decays to @xmath140 remains the dominant signature .
ves noted a @xmath66 structure @xmath219 at @xmath220 gev with width @xmath221 mev coupling strongly to @xmath197 but absent in @xmath183 and @xmath222 @xcite , although an earlier experiment by accmor reported the state in @xmath223 and @xmath197 @xcite .
a similar excess may exist in e852 data @xcite .
theory expects a second radially excited quarkonium state at 2.13 gev @xcite . in the isoscalar sector
, evidence exists for a @xmath66 resonance at @xmath224 gev .
there are three plausible possibilities for its interpretation as a conventional quarkonium state : \(i ) light quark @xmath225 : the light quark @xmath225 state @xmath226 has most likely already been isolated by crystal barrel @xcite and wa102 @xcite , as interpreted in ref .
@xcite .
\(ii ) @xmath227 : this would be a natural assignment for a @xmath228 gev state , based on the predicted mass of @xmath229 gev @xcite .
however , this assignment appears troublesome if we consider the fact that it has only been observed in final states not resonance , the data appear to indicate an enhancement at @xmath230 gev in the @xmath66 partial wave produced in @xmath231 ( fig . 2 of ref .
since the production process may enhance @xmath95 above light quark production , lass may have evidence for the @xmath95 nature of the enhancement . ] containing strangeness .
moreover , there is evidence from crystal ball and cello for an isovector partner at @xmath228 gev ( see the detailed discussion in ref .
@xcite ) , in contradiction with the @xmath95 assignment .
however , the isovector partner is not seen in recent analyses from argus @xcite and l3 @xcite .
it is expected that e852 would have more to contribute on this subject in the @xmath88 @xcite , @xmath163 and @xmath232 channels @xcite .
\(iii ) light quark @xmath233 : as observed above , these states are expected at much higher masses than @xmath228 gev , and there is already evidence for an isovector @xmath66 in the correct mass region .
if future experimental work determines that none of these three possibilities are viable interpretations for the 1.8 gev state , there is a strong possibility that the @xmath228 gev isoscalar state is a hybrid meson .
this is because it is unlikely to be a glueball which is predicted by lattice gauge theory at @xmath234 gev @xcite .
we also do not expect a molecule or four quark state in this region , although the state may contain a long range @xmath235 component due to its nearness to the @xmath235 threshold @xcite .
it is hence of interest to determine whether data on the state is consistent with decays calculated in this work .
recently , the wa102 collaboration reported evidence for two @xmath66 states in central @xmath236 collisions at 450 gev , which were absent in previous analyses by wa76 and wa91 at 85 , 300 and 450 gev @xcite .
the upper @xmath66 state is found at @xmath237 mev with a width of @xmath238 mev .
the observed decay mode is @xmath140 , in accordance with the predictions of this model and the ikp model .
the crystal ball collaboration reported some time ago a state with undecided @xmath121 ( claimed to be @xmath66 ) at @xmath239 mev , with a width of @xmath240 mev , decaying equally to @xmath140 and @xmath241 @xcite .
similar conclusions were drawn by the cello collaboration @xcite .
a doubling of isoscalar @xmath66 peaks has also been reported by crystal barrel , in the isoscalar sector in @xmath242 @xcite .
masses and widths of @xmath243 mev and @xmath244 mev have been reported for the upper @xmath66 state .
the high mass state @xmath245 has been seen only in @xmath246 ( only 50 mev above threshold ) , and no evidence of it is found in @xmath247 , @xmath248 , or @xmath249 .
the absence of the state in @xmath249 is consistent with the hybrid interpretation ( see column 5 of table [ table3 ] ) .
however , the non appearance of the state in @xmath140 appears disasterous at first glance .
we would like to point out here that this is in fact not the case .
experimentally , @xmath250 employing branching ratios from ref .
@xcite and theoretical widths yields @xmath251 in both this model and the ikp model for a 1.875 gev hybrid .
the mean value was obtained for the `` standard parameters '' and the error corresponds to the `` alternative parameters '' we find a ratio of 1.0 @xcite and for a @xmath233 a ratio of 0.7 , all evaluated for a meson at 1.875 gev . ] .
equality is reached in the narrow resonance approximation .
the ratio appears to be consistent with the large errors estimated from experiment .
we conclude that although @xmath245 can be @xmath227 ; it is equally consistent with the hybrid interpretation
. a critical discriminant between these possibilities would be the experimental confirmation of an isovector partner @xcite since the hybrid candidate consists of light quarks .. 0.2 in strangeonium hybrids could be studied by intense photon beams at jlab , due to the strong affinity of the photon for @xmath95 .
vector and @xmath252 hybrids have non
negligible @xmath253 couplings which could form a good search channel .
moreover , we note that some non exotic hybrids are substantially narrower than their quarkonium partners , e.g. for @xmath254 the hybrid has a width of @xmath255 mev in both models compared to the prediction for @xmath58 quarkonium of 650 mev @xcite .
this generates the prospect of photoproduction of vector states beyond the well known @xmath256 .
when the total widths of all @xmath257 , @xmath258 and @xmath95 hybrids listed in table [ table4 ] are computed , we find that for `` standard parameters '' the average total widths of the three flavor varieties are very similar in both models ( although @xmath258 are about @xmath259 narrower ) .
this dispells a popular misconception that @xmath95 hybrids should be narrower than light quark hybrids .
0.2 in the widths of charmonium hybrids are suppressed below @xmath260 threshold , where only @xmath261 and @xmath262 modes are allowed , since these are the only open charm combinations where the wave functions the two final states are different .
widths in table [ table5 ] are in the 1 - 20 mev range , and hence surprisingly narrow for charmonia at such high masses . however , when the hybrids are allowed to become more massive than the @xmath260 threshold , the total widths increase drastically ( see figure 2 ) to @xmath263 mev for 4.4 gev hybrids ( see column 5 in table [ table5 ] ) . however , in this model ( but not in the ikp model @xcite ) the @xmath154 exotic remains narrow at @xmath264 mev .
[ figccbar ]
we have explored the implications of the hybrid decay model constructed in ref .
the model assumes the validity of the flux tube description of hybrids .
the hybrid decay vertex is motivated by the heavy quark limit of the qcd hamiltonian .
it is essentially given by transverse gluon dissociation into a @xmath265 pair .
thus , the decay model is similar to earlier @xcite hybrid decay models which assumed that constituent gluons produced @xmath34 pairs in the standard perturbative manner .
the main difference is that the hybrid and the decay mechanism have been written in terms of the degrees of freedom appropriate to the flux tube model ( ie .
, phonons ) . in this sense ,
the model presented here is similar to @xmath57 " meson decay models whereas the ikp model is similar to @xmath3 models .
this similarity extends to amplitude ratios .
amplitude ratios serve as a sensitive probe of the decay vertex and may be used to test models . for example , @xmath266 amplitude ratios tend to be significantly smaller in @xmath3 meson decay models than in @xmath57 models due to details of momentum routing . because of this it has been shown that @xmath3 models are heavily favored by the data@xcite .
a similar situation exists between this model and that of ikp . for example , the @xmath266 amplitude ratio for @xmath267 is roughly 2 in the ikp model while it is 250 in this model . similarly the @xmath266 ratio for @xmath268 is 5 in the ikp model and 40 in this model .
one can envision a time when these ratios may be experimentally determined and the models distinguished .
hybrid states that have small total widths should be accessible experimentally . we find that for `` standard parameters '' the total width of the @xmath269 and @xmath95 @xmath66 hybrids are less than 100 mev in both models .
moreover , the same is true for @xmath258 @xmath270 and @xmath95 @xmath154 .
the stability of these narrow widths in both models is significant , and neccesitates experimental examination of these states .
there are also states which are less than 100 mev wide in this model , but not in the ikp model .
these are the @xmath257 and @xmath258 @xmath154 , the @xmath258 and @xmath95 @xmath148 , the @xmath258 @xmath201 and @xmath146 . in general the ikp model and
this one give similar decay widths ( in large part because both obey the spin and s+p selection rules ) .
however they differ dramatically in a few places .
the most obvious is the anomalously narrow width of exotic @xmath154 hybrids predicted by this model ( less than 10 mev ) .
this surprising result needs to be accounted for in experimental searches and partial wave analyses .
the channel @xmath271 @xmath272 is especially important in this regard .
other differences are in the total widths of the @xmath273 and @xmath274 hybrids , which we predict to be roughly 200 mev , while ikp predict values 3 times larger .
a larger discrepancy is in the @xmath275 state which we predict to be 50 mev wide , while ikp predict 450 mev . among the conclusions of our survey of interesting hybrid candidates
were the following .
the @xmath63 remains enigmatic and further experimental study of this state is vital .
this is especially true of the @xmath276 mode which appears to be anomalously large . amongst quantum number - exotic hybrids
, the isovector @xmath146 appears to be very wide and thus may be difficult to detect .
alternatively , there is growing evidence for ( several ) @xmath148 states .
we stress the importance of exploring the @xmath143 and @xmath163 channels as well as @xmath277 and , if the hybrid is heavy enough , @xmath171 .
in fact the latter mode is expected to be the largest if the hybrid is heavier than 2.1 gev .
the @xmath92 is difficult to accomodate as a conventional meson and makes a likely hybrid candidate .
indeed , the experimental branching ratios agree spectacularly with our predictions .
alternatively , it appears likely that the @xmath188 is a @xmath278 quarkonium state due to the small s - wave @xmath277 mode and the strong @xmath163 channel .
finally , we conclude that the @xmath245 can be an @xmath0 @xmath225 state or a hybrid .
searching for an isovector partner for this state would therefore be especially interesting .
all @xmath1 and @xmath2 hybrids are very narrow if they lie within their expected mass ranges .
since the heavy quarkonium spectrum is well understood , searches for these hybrids are especially interesting .
in general , all hybrid widths depend strongly on available phase space so that care should be exercised when employing our results .
furthermore , there can be substantial parameter dependence in the predicted widths .
the standard and alternative data sets typically led to predictions differing by 50% and sometimes as much as 100% .
finally , the overall scale is not well known and may change substantially as new information emerges .
we look forward to the day when hybrids and their decays are experimentally well established since this is doubtlessly an important step in developing an understanding of the mechanics of strong qcd and low energy glue . financial support of the doe under grants de - fg02 - 96er40944 ( ess ) and de - fg02 - 87er40365 ( aps ) is acknowledged .
prp acknowledges a lindemann fellowship from the english speaking union to visit jlab , where part of this work was completed .
the `` standard parameters '' are as follows .
all @xmath128 s are those of ref .
@xcite , i.e. for @xmath94 , @xmath95 , @xmath96 , @xmath97 hybrids 0.27 , 0.30 , 0.30 , 0.34 gev , for @xmath279 , @xmath280 , @xmath281 , @xmath282 , @xmath283 , @xmath284 , @xmath98 , @xmath285 , @xmath286 0.34 gev , for @xmath151 , @xmath63 , @xmath287 0.35 gev , for @xmath288 , @xmath289 0.37 gev , for @xmath290 , @xmath291 , @xmath292 , @xmath293 0.38 gev , for @xmath294 , @xmath145 , @xmath295 , @xmath296 , @xmath297 0.39 gev , for @xmath47 , @xmath298 , @xmath299 , @xmath300 , @xmath98 , @xmath301 0.41 gev , for @xmath302 0.42 gev , for @xmath35 , @xmath303 0.43 gev , for @xmath304 0.45 gev , for @xmath256 0.46 gev , for @xmath102 , @xmath104 0.47 gev and for @xmath305 0.54 gev . in the case of hybrid decays to s wave mesons the widths are zero for @xmath306 .
the width divided by @xmath127 remains finite , and is called the `` reduced width '' . for hybrid decays to s wave mesons we calculate the actual width by multiplying the reduced width by @xmath127 , but this time we take the @xmath128 s to be those of ref .
@xcite , i.e. for @xmath294 0.75 gev , @xmath307 0.74 gev , @xmath308 0.45 gev , @xmath305 0.51 gev , @xmath35 0.71 gev , @xmath303 0.48 gev , @xmath296 0.66 gev , @xmath297 0.54 gev , @xmath47 0.64 gev and @xmath298 0.57 gev .
we assume that the quarks that are created may have different mass than the initial quarks .
specifically , the mass of the @xmath309 quarks are assumed to be @xmath310 gev .
we assume @xmath311 and @xmath312 ( high mass @xmath115 state ) to have masses of 2.40 and 2.45 gev respectively .
the wave functions are taken to be s.h.o .
wave functions except for the hybrid , where a radial prefactor of @xmath313 , with @xmath314 is assumed @xcite .
the @xmath315mixing is @xmath316 @xcite in the p wave kaon sector .
@xmath317 / @xmath312 mixing is @xmath318 .
the `` alternative parameters '' ( also employed in ref .
@xcite ) change from the preceding as follows .
@xmath128 of all hybrids are @xmath124 gev .
@xmath128 of @xmath319 are 0.54 , 0.31 , 0.31 , 0.53 , 0.36 , 0.43 , 0.45 , 0.37 , 0.43 , 0.40 gev respectively@xcite .
other mesons have @xmath320 gev @xcite .
we allow the final states to have different @xmath128 s .
all other conventions are the same as for the `` standard parameters '' .
a. le yaouanc , l. oliver , o. pne , and j .- c .
raynal , phys .
rev . * d29 * , 1233 ( 1984 ) ; a. le yaouanc , l. oliver , s. ono , o. pne , and j .- c .
raynal , phys .
rev . * d31 * , 137 ( 1985 ) ; s. adler and p. davis , nucl . phys . * b244 * , 469 ( 1984 ) .
m. acciarri _
( l3 collab . ) , cern - ppe-97 - 068 ; eps 1997 summer conference , jerusalem ( abstract no .
474 ) ; s. hou , _ proc . of hadron 97 _
( upton , ny , august 1997 ) , p. 745 , eds .
chung and h.j .
willutzki .
d. antreasyan _ et al . _
( crystal ball collab . ) , _ z. phys . _
* c48 * ( 1990 ) 561 ; k. karch _
( crystal ball collab . ) , _ z. phys . _ * c54 * ( 1992 ) 33 ; m. feindt , _ proc . of @xmath324 ichep _
( singapore , 1990 ) , p. 537 | the phenomenology of a newly developed model of hybrid meson decay is developed .
the decay mechanism is based on the heavy quark expansion of qcd and the strong coupling flux tube picture of nonperturbative glue .
a comprehensive list of partial decay widths of a wide variety of light , @xmath0 , @xmath1 , and @xmath2 hybrid mesons is presented .
results which appear approximately universal are highlighted along with those which distinguish different hybrid decay models .
finally , we examine several interesting hybrid candidates in detail .
# 1 |
semiconductor nanowires are almost ideal objects for studying quantum effects and electron interference phenomena .
the use of the bottom - up approach for nanowire growth simplifies the preparation substantially and allows us to create novel confinement schemes , such as axial or radial heterostructures.@xcite the large surface - to - volume ratio of nanowires means that surface properties are crucial for discussions of transport properties , so that low band - gap semiconductors , e.g. inas , inn , or insb , are particularly interesting . in these systems ,
the fermi level at the surface is pinned inside the conduction band,@xcite and an accumulation layer is formed .
this guarantees that the conductance is sufficiently large even at low nanowire radius .
the presence of the surface accumulation layer means that a tubular conducting channel is formed , and this shape of the conductor has important implications for the magnetoconductance of the nanowires .
an example is the theoretical prediction and experimental confirmation of flux - periodic oscillations in nanowires with a magnetic field applied along the wire axis .
@xcite the electronic states of a cylindrical two - dimensional electron gas in a transverse magnetic field were calculated by ferrari _
et al._,@xcite while magarill _ et al . _
@xcite discussed the kinetics of electrons in a tubular conductor .
many concepts have been developed for planar semiconductor layer systems that make use of the spin degree of freedom for device structures .
the best - known example is the spin field - effect transistor,@xcite which uses the gate - controlled spin - precession induced by the rashba effect .
@xcite the rashba spin - orbit coupling originates from a macroscopic electric field in an asymmetric quantum well.@xcite meanwhile , research activities have been extended to planar quasi one - dimensional structures , which promise a superior spin control.@xcite the energy spectrum and spin precession in these structures are governed by the interplay between confinement and energy splitting due to spin - orbit coupling .
@xcite only a few theoretical investigations have dealt with the effect of spin - orbit coupling in cylindrical conductors on the electronic states and on the quantum transport.@xcite the spin - dynamics in curved two - dimensional electron gases was discussed by trushin and schliemann @xcite while the weak antilocalization effect in cylindrical wires was studied by wenk and kettemann.@xcite the presence of spin - orbit coupling was confirmed for inn @xcite and inas semiconductor nanowires by measuring the weak antilocalization effect .
@xcite the various possibilities of spin control in two - dimensional electron gases and planar wire structures opened up by the rashba effect have inspired us to analyze theoretically the spin dynamics in tubular conductors .
we have used a cylindrical inas nanowire with a surface two - dimensional electron gas as a model system , but our findings also apply to other systems , e.g. inn or insb nanowires . in sect .
[ sec : ii ] we analyze the electronic states , focusing on spin properties , and we discuss the conditions under which a spin precession can be observed in tubular nanowires at zero magnetic field ( sect .
[ sec : iii ] ) and in an axial magnetic field ( sect .
[ sec : iv ] ) . in sect .
[ sec : v ] , we comment on the suitability of tubular conductors for spin electronic devices .
electrons confined in a cylinder move along the axis with a linear momentum @xmath0 ( @xmath1 real ) and around the axis with an angular momentum @xmath2 ( @xmath3 integer ) .
as long as the translational and rotational symmetries of the cylinder are not perturbed these momenta are conserved quantities . the wave function of an electron @xmath4 is a product of exponential functions in @xmath5 , the coordinate along the axis and the azimuthal angle around the axis respectively , and a radial distribution function @xmath6 .
the distribution is determined by internal forces produced by the cylinder material . in our case
, we took a planar 2-dimensional electron gas ( 2deg ) at the surface of inas as a reference,@xcite i.e. assuming a surface state charge density of @xmath7 @xmath8 , a background @xmath9-doping of @xmath10 @xmath11 and an effective electron mass of @xmath12 .
the calculations were done for a cylinder radius @xmath13 nm .
a schematic illustration of the nanowire is depicted in fig .
[ fig:1 ] ( upper inset ) .
electrons of atoms at the surface may find energetically more favorable states in the conduction band . due to the coulomb attraction between the electrons and the ions remaining at the surface
the electrons get trapped in a layer close to the surface forming a 2deg.@xcite the potential @xmath14 resulting from the charge density of occupied electron states @xmath15 , of ions at the surface and of dopants @xmath16 @xmath17 is shown in fig .
[ fig:1 ] .
@xmath18 is the elementary charge , @xmath19 the spin index .
@xmath20 is the bulk dielectric constant of inas.@xcite it takes the polarization charges of the medium into account .
the potential profile is determined by poisson s equation which is solved in cylindrical symmetry analytically @xmath21 equations ( [ dens ] ) and ( [ vau ] ) are solved self - consistently . starting from the potential of a homogeneous distribution of electrons in the cylinder
the distribution is recalculated using the schrdinger equation given below [ see eq .
( [ schroe ] ) ] and eq .
( [ dens ] ) .
the iteration procedure converges monotonically .
we assumed an interface barrier of infinite height . , the spinor components @xmath22 and @xmath23 and potential profile @xmath14 as a function of the normalized radius @xmath24 .
the upper inset shows a schematic illustration of the nanowires , including the relevant electric and magnetic fields .
the lower inset shows the spin orientation along the circumference for @xmath25 .
[ fig:1 ] ] due to the electric field @xmath26 across the surface of the cylinder the spin @xmath27 of the electron is coupled to its orbital motion @xmath28
\frac{\gamma}{\hbar } \nonumber \\ & = & \gamma v^{\prime } \left [ \left ( \begin{array}{cc } 0 & \imath\ e^{-\imath\phi } \\ -\imath\ e^{\imath\phi } & 0 \\ \end{array } \right ) \frac{\partial}{\imath\partial z } \right . \nonumber \\ & & \left
. + \left ( \begin{array}{cc } 1 & 0 \\ 0 & -1 \\
\end{array } \right ) \frac{\partial}{r\imath\partial\phi } \right ] \ ; . \label{hso}\end{aligned}\ ] ] the coupling - strength @xmath29 is determined by the band structure of the cylinder material ( @xmath30 nm@xmath31 for inas).@xcite the second part of eq .
( [ hso ] ) expresses @xmath32 in terms of pauli matrices for @xmath33 acting on a 2-component ( spinor ) wave function @xmath34 .
the off - diagonal terms in @xmath35 raise ( lower ) the value of the orbital angular momentum @xmath36 of @xmath37 ( @xmath38 ) by @xmath39 . the stationary states are eigenstates of the total angular momentum @xmath40 ( @xmath41 ) with eigenvalues @xmath42 .
the spinor is of the form : @xmath43 where @xmath44 are real functions and solve the differential equations @xmath45 here , @xmath46 contains the contributions of the centrifugal force and the diagonal spin - orbit term , @xmath47 is the energy without the axial kinetic energy . at the wire boundary we assumed a barrier of infinite height.@xcite the influence of an external magnetic field @xmath48
is not included yet .
dispersion at @xmath49 .
the dashed line indicates the axial kinetic energy left out , which crosses the bands at @xmath50 .
the pair of dots represent states forming the superpositions states @xmath51 and @xmath52 , while the square indicates the state @xmath53 .
the dotted line illustrates the gaussian wave packet of width @xmath54 .
[ fig:2 ] ] in fig .
[ fig:2 ] the energy @xmath55 is plotted for several @xmath56-bands at @xmath49 .
the parabola indicates the axial kinetic energy left out .
it crosses the bands at the fermi - momentum @xmath50 , i.e. states with energy below the parabola are occupied . at @xmath57
the coupling between @xmath3 and @xmath58 vanishes [ cf .
( [ hso ] ) ] .
classification with respect to @xmath3 is possible .
the splitting between the second and third band ( @xmath59 ) is caused by the diagonal part of @xmath35 and increases proportional to @xmath3 for the higher states .
due to the mirror symmetry @xmath60 states with angular momentum and spin reversed have the same energy .
therefore , all bands are twofold degenerate . the solution @xmath61 of eq .
( [ schroe ] ) for @xmath25 at @xmath50 is shown in fig .
[ fig:1 ] .
the spin - orbit coupling increases linearly with @xmath1 , i.e. at @xmath50 with @xmath62 there is the strongest spin - orbit coupling .
the spin density attains a sizable tangential component @xmath63 the component along the wire axis is @xmath64 the radial component is zero . the spin orientation around the cylinder for @xmath25 is illustrated in fig .
[ fig:1 ] ( lower inset ) . according to eq .
( [ spidi ] ) the spin turns to the axial direction .
this is shown for different values of @xmath56 in the plot of the spin densities @xmath65 and @xmath66 in fig .
[ fig:3 ] . as can be seen here , the spin is oriented exclusively tangentially and along the axial direction . when averaged over the cylinder plane @xmath67 for each state
@xmath68 the spin components @xmath69 and @xmath70 are zero , while a finite contribution @xmath71 remains along the @xmath72-direction .
of the lower energy states for total angular momenta @xmath73 and @xmath74 .
the spin is oriented only tangentially and along the @xmath72-axis .
[ fig:3 ] ]
for each @xmath1 and @xmath56 there are two solutions of eq .
( [ schroe ] ) @xmath75 .
the ( + ) -state originates from @xmath76 , the ( @xmath77)-state from @xmath78 , the solutions at @xmath57 , with @xmath79 .
they are orthogonal to each other and have opposite spin direction ( @xmath80 ) .
they have different energies @xmath55 and therefore different @xmath50 .
their superposition @xmath81 yields @xmath82 the contributions of the basis states ( @xmath80 ) almost cancel each other and are neglected further on .
the interference between the states is constructive due to orthogonality and leads to @xmath83 an oscillation of the average spin along the cylinder axis with a wavelength @xmath84 .
the spin components @xmath85 , @xmath86 in the cylinder plane are both zero .
superpositions of eigenstates with different @xmath56 s form states with a non - zero average spin component in the cylinder plane , e.g. @xmath87 .
as one can easily retrace , these states originate from states with the same angular momentum @xmath3 .
the interference term gives the only @xmath88-independent contribution to the densities of @xmath89 . with @xmath90 \left ( \begin{array } { c } \tilde{f}\left(r\right ) \\
\imath e^{\imath\phi}\tilde{h}\left(r\right ) \end{array } \right ) \ ; , \nonumber \\ \end{aligned}\ ] ] the averages are @xmath91 the @xmath92 contribution is small and does not depend on @xmath72 .
in particular , for the superposition @xmath93 of the lowest two states @xmath92 is zero . for @xmath51 , the superposition of @xmath94 and @xmath95 , the spin precesses counterclockwise in the cylinder plane along the cylinder axis , as illustrated in fig .
[ fig:4](a ) . at the fermi energy
constituted of the states @xmath94 and @xmath95 for a propagation along the wire axis from @xmath96 to @xmath97 .
( b ) spin orientation of the sum of the contribution shown in ( a ) and the corresponding clockwise contribution @xmath52 being a superposition of @xmath98 and @xmath99 .
( c ) spin oscillations resulting from the combinations of the two lower energy superposition states @xmath100 and @xmath101 .
( d ) spin variation for a wave - packet of width @xmath102 centered between the @xmath50 s of the states @xmath100 and @xmath101 ( cf . fig .
[ fig:2 ] ) .
[ fig:4],title="fig : " ] at the fermi energy constituted of the states @xmath94 and @xmath95 for a propagation along the wire axis from @xmath96 to @xmath97 .
( b ) spin orientation of the sum of the contribution shown in ( a ) and the corresponding clockwise contribution @xmath52 being a superposition of @xmath98 and @xmath99 .
( c ) spin oscillations resulting from the combinations of the two lower energy superposition states @xmath100 and @xmath101 .
( d ) spin variation for a wave - packet of width @xmath102 centered between the @xmath50 s of the states @xmath100 and @xmath101 ( cf . fig .
[ fig:2 ] ) .
[ fig:4],title="fig : " ] at the fermi energy constituted of the states @xmath94 and @xmath95 for a propagation along the wire axis from @xmath96 to @xmath97 .
( b ) spin orientation of the sum of the contribution shown in ( a ) and the corresponding clockwise contribution @xmath52 being a superposition of @xmath98 and @xmath99 .
( c ) spin oscillations resulting from the combinations of the two lower energy superposition states @xmath100 and @xmath101 .
( d ) spin variation for a wave - packet of width @xmath102 centered between the @xmath50 s of the states @xmath100 and @xmath101 ( cf . fig .
[ fig:2 ] ) .
[ fig:4],title="fig : " ] at the fermi energy constituted of the states @xmath94 and @xmath95 for a propagation along the wire axis from @xmath96 to @xmath97 .
( b ) spin orientation of the sum of the contribution shown in ( a ) and the corresponding clockwise contribution @xmath52 being a superposition of @xmath98 and @xmath99 .
( c ) spin oscillations resulting from the combinations of the two lower energy superposition states @xmath100 and @xmath101 .
( d ) spin variation for a wave - packet of width @xmath102 centered between the @xmath50 s of the states @xmath100 and @xmath101 ( cf . fig .
[ fig:2 ] ) .
[ fig:4],title="fig : " ] here , we assumed an initial spin orientation along the @xmath103 direction , which in practice can be realized by spin injection from a spin - polarized electrode .
for @xmath52 constituted of the opposite states @xmath98 and @xmath99 the spin precession is clockwise .
both precessions have the same period of @xmath104 .
their energy is degenerate . due to their exactly inverse precession sense the combination of these states results in an oscillatory behavior of the net spin orientation , as depicted in fig .
[ fig:4](b ) . for an initial spin orientation along the @xmath103 direction
the spin oscillates in the @xmath105-plane .
superposition of the respective opposite states restores the left - right symmetry and eliminates spin precession .
the oscillation period @xmath106 of @xmath107 depends on @xmath56 . for smaller @xmath108 ,
e.g. @xmath100 the corresponding difference in @xmath109 and @xmath110 becomes smaller so that the period @xmath106 is enlarged , as one can infer from fig .
[ fig:4](c ) as compared to fig .
[ fig:4](b ) .
the superposition state @xmath53 constituted of the two lowest lying energy states @xmath111 ( cf .
[ fig:2 ] , square ) shows no precession at all , because here @xmath109 and @xmath110 are identical .
figure [ fig:4](d ) shows the spin variation for a gaussian wave packet of width @xmath54 centered between the @xmath50 s of the states @xmath100 and @xmath101 . in position space
this corresponds to a distribution of width @xmath112 .
the oscillation deviates from a purely harmonic oscillation , as shown in fig .
[ fig:4](c ) , due to the contributions of the other states at the fermi energy .
this effect is also increasing with decreasing @xmath108 when the @xmath50 s get closer to each other .
the electron spin is usually injected from a spin - polarized electrode in all states at the fermi energy @xmath113 having the correct spin direction .
thus , if only the direction of the spin is fixed by the electrode , all states are likely to transport electrons through the cylinder and a definite precession will not be observed .
the total spin will only vary in the plane which is defined by the initial spin orientation and the @xmath72-axis , similar to the situation illustrated in fig .
[ fig:4 ] . in order to observe spin precession about the cylinder axis , a selection mechanism which breaks the left - right symmetry of the system must be adopted . as it will be discussed in the next section ,
this is achieved by applying a longitudinal magnetic field @xmath114 .
the vector potential @xmath115 of a longitudinal magnetic field introduces a paramagnetic ( zeeman- ) and diamagnetic ( landau- ) term into eq .
( [ schroe ] ) .
@xmath116 is extended to @xmath117 with @xmath118 the gyromagnetic - factor of the electron spin ( @xmath119 for inas@xcite ) .
the paramagnetic ( second ) term in @xmath120 raises @xmath55 for states with @xmath121 and lowers @xmath55 for states with @xmath122 .
the energy difference increases @xmath123 for @xmath124 ( cf . fig .
[ fig:5 ] ) . for larger @xmath48
@xmath55 increase @xmath125 due to the diamagnetic ( third ) term . in the linear range
the influence of the @xmath126-dependence of the third term is negligible .
the densities do not change significantly .
dispersion ( left panel ) at @xmath57 and energy vs. @xmath1 dispersion ( right panel ) at @xmath127 t. the dashed line indicates the axial kinetic energy left out , which crosses the bands at @xmath50 .
the pairs of dots indicate the states forming the superposition states @xmath51 and @xmath52 at @xmath50 with a net spin in the cylinder plane .
the two states @xmath128 and @xmath99 with @xmath129 are marked by triangles .
[ fig:5 ] ] the main effect of @xmath48 is the energetic separation of the @xmath130-states .
it opens possibilities of observing spin dynamics in electronic transport .
this will be demonstrated in the following at @xmath127 t. figure [ fig:5 ] shows the @xmath48-dependence at @xmath57 up to @xmath127 t and the @xmath1-dependence at @xmath127 t of @xmath55 for states from @xmath131 to @xmath132 .
again , the parabola marks the fermi edge .
superpositions with spin in the cylinder plane according to eq .
( [ spiprae ] ) , @xmath107 are marked as pairs in fig .
[ fig:5 ] .
the lower pair corresponds to @xmath51 depicted in fig .
[ fig:4](a ) . as illustrated in fig .
[ fig:6](a ) , it shows the same counter - clockwise precession . at @xmath127 t. ( b ) corresponding spin precession for the state @xmath52 .
( c ) spin orientation and magnitude of the sum of the contributions shown in ( a ) and ( b ) for a propagation from @xmath96 to @xmath97 .
the arrow indicates the direction of the initially injected spin .
[ fig:6],title="fig : " ] at @xmath127 t. ( b ) corresponding spin precession for the state @xmath52 .
( c ) spin orientation and magnitude of the sum of the contributions shown in ( a ) and ( b ) for a propagation from @xmath96 to @xmath97 .
the arrow indicates the direction of the initially injected spin .
[ fig:6],title="fig : " ] at @xmath127 t. ( b ) corresponding spin precession for the state @xmath52 .
( c ) spin orientation and magnitude of the sum of the contributions shown in ( a ) and ( b ) for a propagation from @xmath96 to @xmath97 .
the arrow indicates the direction of the initially injected spin .
[ fig:6],title="fig : " ] in contrast to the zero field case , now the superposition state @xmath52 has a larger @xmath50-difference , i.e. a shorter precession length [ cf .
[ fig:6](b ) ] .
consequently , the precessions of @xmath51 and @xmath52 are not exactly opposite .
in contrast to the case at @xmath49 , the spin still rotates following the state with the faster precession , when both states are superposed .
this is illustrated in fig .
[ fig:6](c ) , where one finds that in addition to the oscillation of the spin amplitude its orientation is also changed during propagation .
thus , by applying a magnetic field a spin precession can be achieved . along the wire axis for the superposition states
@xmath133 with @xmath134 and @xmath135 at @xmath127 t. ( b ) illustration of the oscillation of the average spin along the wire axis for the superposition state @xmath136 .
( c ) modulation of the total spin orientation @xmath71 ( dashed line ) resulting from a combination of the @xmath137 and @xmath136 states at @xmath127 t. [ fig:7 ] ] in the previous section , we already pointed out that the superposition state @xmath133 with equal total angular momentum but opposite spin orientation result in an oscillation of the average spin along the cylinder axis . in fig .
[ fig:7](a ) and ( b ) these oscillations of @xmath138 are shown for different values of @xmath56 at @xmath127 t. one finds that for larger total angular momentum values the oscillation period is shorter owing to the larger difference of fermi wave vectors . in fig .
[ fig:5 ] the states contributing to @xmath136 are marked by triangles . compared to the previously discussed @xmath107 states , here the difference in the fermi vectors is relatively large , leading to a faster oscillation compared to the spin precession period shown in fig . [
fig:6](b ) .
once again the application of an axial magnetic field breaks the symmetry of the @xmath139 states . as can be inferred from fig .
[ fig:7](c ) , a different oscillation period is found for the @xmath136 and @xmath137 states .
thus , when these states are combined a beating in the oscillation of the average spin appears .
in the previous two sections we learned that an injected spin is strongly modulated while propagating through a cylindrical nanowire . for a spin injection along the wire axis ,
e.g. by a ferromagnetic electrode , the spin is carried by superposition states with equal total angular momenta . in analogy to the spin field - effect transistor based on a planar 2deg,@xcite a transistor structure can be realized by placing a second magnetic electrode at the opposite terminal of the nanowire as a spin detector .
control of the spin orientation can achieved by manipulating the strength of the rashba effect by means of a gate electrode . by applying a bias voltage to the gate
, the strength of the electric field @xmath140 in the surface 2deg is adjusted . in order to obtain a uniform control within the channel , a so - called wrap - around gate should be preferred.@xcite usually , in a realistic situation a larger number of states with different total angular momenta @xmath56
is occupied .
as we observed , for each superposition state @xmath133 different oscillation periods are found .
this leads to a rather complex modulation of the spin along the axial direction .
an obvious strategy for simplification is to reduce the number of occupied states , i.e. by depleting the channel by means of a gate .
another possibility might be to only occupy certain states by means of @xmath1-selective filters .
this might be realized by means of an injection through a single or a resonant tunneling barrier . as pointed out in sect .
[ sec : iii ] , one possible way to model this situation is to assume the formation of a state with a gaussian distribution around the average momentum .
in addition to a spin injection and detection along the wire axis it is also possible to inject spins in transversal direction . here
, the spins are carried by superposition states @xmath107 constituted of states with different total angular momenta @xmath56 . as long as no magnetic field
is applied the spin is exclusively modulated in the plane spanned by the injection orientation and the wire axis . here
, the output signal in a spin field - effect transistor is gained by gate - modulating the spin orientation along or opposite to a detector electrode , which is polarized parallel or antiparallel to the injector . by applying an axially oriented magnetic field , spin precession about the wire axis can be achieved .
this additional feature might be an interesting option to implement more complex functionalities in spin electronic devices . in conclusion
, we have shown that semiconductor nanowires affected by rashba spin - orbit coupling are promising candidates for future nanowire - based spin electronic devices .
the complex spin dynamics in these cylindrically - shaped conductors provide many opportunities to tailor the device functionality .
we thank n. demarina ( forschungszentrum jlich ) for fruitful discussions regarding the schrdinger - poisson solver in cylindrical systems and u. zlicke ( massey university , new zealand ) and r. winkler ( northern illinois university , usa ) on the rashba effect at semiconductor interfaces .
furthermore , we acknowledge the support of or work by s. blgel and d. grtzmacher ( forschungszentrum jlich ) . this work was supported by the deutsche forschungsgemeinschaft through for 912 . | the spin precession in a cylindrical semiconductor nanowire due to rashba spin - orbit coupling has been investigated theoretically using an inas nanowire containing a surface two - dimensional electron gas as a model .
the eigenstates , energy - momentum dispersion , and the energy - magnetic field dispersion relation are determined by solving the schrdinger equation in a cylindrical symmetry . the combination of states with the same total angular momentum but opposite spin orientation results in a periodic modulation of the axial spin component along the axis of the wire .
spin - precession about the wires axis is achieved by interference of two states with different total angular momentum .
because a superposition state with exact opposite spin precession exists at zero magnetic field , an oscillation of the spin orientation can be obtained .
if an axially oriented magnetic field is applied , the spin gains an additional precessing component . |
big bang nucleosynthesis ( bbn ) makes a clean prediction for the primordial helium abundance @xmath0 , depending on only one parameter ( the baryon - to - photon ratio @xmath3 ) and on that only weakly .
a precise measurement of @xmath0 is therefore necessary to test bbn even in the present situation where measurements of other predicted quantities constraining @xmath3 , such as the primordial lithium abundance , deuterium abundance and the cosmic baryon density , are not yet so precise ( e.g. walker 1991 , smith 1993 , sarkar 1996 , fields 1996 , hogan 1997ab ) .
even though the bulk of the helium of the universe originates in the big bang , the additional helium enrichment by stars can not be ignored in estimating the primordial abundance from observations of present - day helium . unlike the case of deuterium
, we do not have the option of measuring directly the nearly primordial abundance of 4 at high redshift .
lyman-@xmath4 absorption by he@xmath5 at high redshift now gives a rough direct estimate of primordial abundance ( hogan , anderson & rugers 1997 ) , but require uncertain ionization corrections to estimate @xmath2 .
the high precision needed for strong cosmological tests can only be attained with high signal - to - noise measurements of nebular emission lines in well characterized nearby regions ( pagel 1992 , izotov , thuan , & lipovetsky , 1994 , 1997 , skillman 1994 , 1997 ) , necessarily requiring measurements in gas with a certain amount of stellar helium superimposed on the primordial helium . under these circumstances , what is the best way to estimate the primordial helium abundance @xmath0 ?
it certainly helps to find regions with as little stellar helium as possible , as indicated both by their helium abundances and by the abundances of other heavier elements .
the best studied example is i zw 18 ( pagel 1992 , skillman & kennicutt 1993 , izotov , thuan , & lipovetsky , 1997 ) , with an average estimated helium abundance from five independent measurements @xmath6 ( olive , skillman , & steigman 1997 ) , and a metallicity 1/50th of solar . in the analysis
below , we will use samples drawn from the complete sample of 62 individual regions ( including two in i zw 18 ) compiled by olive , skillman , & steigman ( 1997 ) from the data in pagel ( 1992 ) , izotov , thuan , & lipovetsky , ( 1994 , 1996 ) , and skillman ( 1994 , 1997 ) .
this sample represents all published @xmath2 measurements of regions with o / h @xmath7 and n / h @xmath8 ( for comparison ( o / h)@xmath9 and ( n / h)@xmath10 ) . one statistical technique to extract a primordial helium abundance from such a sample was introduced by peimbert & torres - peimbert ( 1974 ) and is still widely used . for each galaxy in the sample , a metallicity @xmath11 ( either o / h or n / h ) and helium abundance @xmath2 are measured ; one then fits a linear relation between @xmath2 and @xmath11 and extrapolates to zero metallicity to find the primordial value .
this technique offers insights into the nuclear evolution of these systems , and for relatively large enrichments the extrapolation is empirically well founded . in olive , steigman & walker ( 1991 ) and in olive & steigman ( 1995 ) , it was shown that these two quantities as measured in extragalactic regions are in fact strongly correlated .
the stability of the resulting fits was tested by a statistical bootstrap in olive & scully ( 1996 ) and in olive , skillman , & steigman ( 1997 ) showing that the results were not particularly sensitive to any individual data point . however , one of the limitations of this method is the need to assume a linear relation between @xmath2 and @xmath11 , which is not well motivated .
for example , helium and oxygen are not produced in the same stars ; while helium is produced primarily in intermediate mass stars , oxygen is produced only in stars with masses @xmath12 .
physically realistic enrichment models often include quadratic terms , as well as stochastic variations in enrichment history , including different slopes @xmath13 .
it is not clear that complicating the fits to include such effects results in a value for primordial @xmath0 with any greater statistical significance or reliability .
moreover most of the information on the primordial abundance is contained in the lowest metallicity points , where the correlation is not very reliably established ; this information is not being efficiently used in regression fits dominated by highly enriched regions .
another approach has been to simply take the lowest , best measured points and use them as estimates of ( or at least limits on ) the primordial abundance .
indeed , it has been argued that it may be sufficient to determine the primordial 4 abundance from even a single well studied low metallicity region such as i zw 18 ( kunth 1994 ) , although to avoid bias it should be chosen on the basis of low metallicity and not low helium abundance .
although these approaches yield results consistent with the linear fit to the data , caution is needed to avoid subtle biases if one starts off by selecting a sample by intentionally choosing the lowest points . moreover in a sample of more than just one point we need a statistical method to combine the data , which recognizes the real spread in stellar helium enrichment and still recovers an unbiased estimate of the primordial value in spite of the obvious `` bias '' that the nonprimordial contributions are always additive .
we explore here a simple but systematic bayesian approach to these issues , which we believe is the simplest recipe to extract an unbiased primordial abundance estimate from the data , free from detailed assumptions about metal enrichment .
we aim to assume as little as possible only that there is some universal primordial helium abundance , and that all subsequent evolution has increased the abundance .
we encapsulate these assumptions mathematically with a bayesian `` prior '' and derive statistical constraints on the primordial abundance which explicitly recognize the bias introduced by stellar enrichment .
inspiration for our approach and more background on the statistical methods is given by press ( 1996 ) application of bayesian arguments to estimates of the hubble constant .
suppose we have a sample @xmath14 consisting of a series of abundance measurements @xmath15 in a set of galaxies . in each case
@xmath16 there is some true abundance @xmath17 which we do not measure , because of the measurement error .
we seek to evaluate the relative probability ( or likelihood ) of obtaining the data in sample @xmath14 , given a primordial abundance @xmath0 : @xmath18 where @xmath19 denotes the probability of obtaining the measurement @xmath15 given the primordial value @xmath0 . for this in turn
we write @xmath20 the first term in the integrand is just the distribution of measurement errors in each case . for the second term we need to explicitly construct a bayesian prior @xmath21 .
this `` enrichment probability function '' encodes our assumptions about what the distribution of helium abundances in galaxies in our sample ought to be , given a primordial abundance . what do we know about @xmath21 ?
the most important property , supported by all models of nuclear evolution , is that there is no net destruction of helium : stars can only increase the primordial abundance .
thus nowhere can the true abundance be less than @xmath0 : @xmath22 for @xmath23 . at a more detailed level
, the shape of this prior depends on what we think about the chemical histories of the galaxies in the sample under study ; there will be some range of @xmath0 of width @xmath24 for which @xmath25 .
fortunately as we shall demonstrate , within reasonable limits the detailed shape for @xmath26 does not much affect the statistical limits on @xmath0 : the most important thing is its asymmetry about @xmath27 . within this framework
we can now explore estimates of @xmath28 , and consequent statistical constraints on @xmath0 . to start with , ( 1 ) assume that the observational errors are normally distributed with variances @xmath29 , and ( 2 ) assume the simplest form for @xmath21 , a uniform distribution or top - hat function of width @xmath24 , @xmath30 . the likelihood function is then given by @xmath31 where the product is over the @xmath32 sample data points .
the integral in ( [ prob1 ] ) can be performed analytically , resulting in @xmath33 the normalization constants @xmath34 are chosen so that each of the individual functions @xmath35 , when integrated over @xmath0 , yield unity . to a very good approximation , @xmath36 . for sufficiently large @xmath24 , the first error function is approximately unity . we choose to leave @xmath24 as a free parameter and maximize the likelihood ( [ prob2 ] ) in both @xmath0 and @xmath24 that is , we estimate @xmath24 from the sample itself . for each choice of @xmath37
we compute the likelihood @xmath38 ; the maximum of the likelihood gives the best values of these parameters .
the relative likelihood of other choices then allow us to derive constraints on @xmath0 and @xmath24 .
this can be illustrated in contour plots showing the equal likelihood contours representing 1 , 2 and 3 standard deviations .
these are determined by comparing the log of the likelihood to its peak value = _ peak - s^2/2 [ s ] for @xmath39 standard deviations .
the contours in the figures represent 1 , 2 , and 3 standard deviations from the peak value as determined by ( [ s ] ) .
if a large number of points contribute , the @xmath40 contour translates in the usual way to approximately 95% confidence . as an illustrative exercise it is useful first to consider a sample of just one galaxy , the lowest metallicity galaxy i zw 18 where @xmath2 has been measured independently in two distinct regions , yielding an average helium abundance @xmath41 .
because the measured values of @xmath0 for the two regions in i zw 18 are consistent with each other , the likelihood function ( [ prob2 ] ) is peaked at @xmath42 and prefers a value @xmath43 , equal to the average of the two points ( see figure 1a ) .
the full power of the method becomes clearer when we employ a larger statistical sample . in figures 1b-1e
, we show the effect of progressively adding more 4 data to our sample .
the data is ordered by metallicity so that for example , the 11 point set includes the 11 extragalactic regions with the lowest values of o / h . except for sample selection ,
the oxygen abundance is not explicitly utilized .
we show the likelihood distributions up to what was described olive , skillman , and steigman ( 1997 ) as set c corresponding to 32 data points . in this case , the linear extrapolation with respect to o / h gave @xmath44 .
the cuttoffs for the various data sets considered here correspond to o / h @xmath4535 , 45 , 60 , and 85 @xmath46 for the 7 , 11 , 18 , and 32 point sets respectively . , the correlation between @xmath2 and o / h can not be neglected . even in the 32 point set considered here ,
the correlation is statistically significant .
notice in table 1 that the likely value of @xmath24 increases from the 18 to the 32 point set ; larger samples contain more information on enrichment but obscure the information on @xmath0 . ]
as soon as one goes beyond a 2-point data set , the dispersion in the data ( normally associated with enrichment correlated with metallicity ) produces a likelihood distribution which is peaked at a non - zero value for @xmath24 yielding an estimate of the stellar helium enrichment of the sample .
consider for example the 11-point data set .
the peak of the distribution occurs at @xmath47 and at that width , @xmath48 .
the individual helium abundances in this sample range from 0.225 to 0.251 , with a weighted mean 0.237 @xmath49 0.002 .
the value for @xmath0 determined by the likelihood distribution is significantly below the mean value of the data .
this bias is familiar from studies of the malmquist effect : the best estimate of the distribution from which points are drawn is in general quite different from the distribution of the values for the points themselves . in the present context
it is the asymmetry of the prior which leads to the apparent bias .
it is interesting to note that at the 2@xmath50 level , the data are consistent with a single primordial value with no subsequent enrichment @xmath51 .
this corresponds to a conservative 2@xmath50 upper limit , @xmath52 as the number of data points in the sample increases , the peak of the likelihood distribution shifts to higher values of @xmath0 .
the 2@xmath50 upper limit however remains at @xmath42 , and its value is relatively insensitive to the data set .
these results are summarized in table 1 .
the most important arbitrary step in the above analysis is in the choice of prior , and to calibrate its effect we consider a range of physically plausible possibilities .
in fact the top - hat function is already quite conservative towards allowing the largest possible upper limits on @xmath0 that is , since very low @xmath2 systems in the present universe , even in the current samples of low metal galaxies , are quite rare , it appears that stellar helium enrichment is seldom very small .
we could however try to bias the result by choosing a different shape biased _ a priori _ towards smaller ( or larger ) nonprimordial enrichments .
consider for example a sawtooth function prior , @xmath53 for @xmath17 between @xmath0 and @xmath54 and 0 otherwise .
this stipulates _ a priori _ that it is most likely that there has been very little helium enrichment , and no enrichment greater than @xmath24 , thus @xmath55 rises abruptly to some value at @xmath27 , linearly decreasing for @xmath56 , and zero for @xmath57 .
we will refer to this prior as the negative bias ( named for the slope with respect to @xmath17 ) .
alternatively , we might choose @xmath58 for @xmath17 between @xmath0 and @xmath54 and 0 otherwise .
this assigns a greater probability for larger helium enrichment but still no enrichment greater than @xmath24 ; thus @xmath55 rises linearly from @xmath0 to some value at @xmath59 .
this is indeed the case for typical galaxy samples , where low enrichment systems are hard to find , so this is perhaps the more realistic prior .
we refer to this prior as the positive bias . taken together , these priors span the range of reasonable possibilities .
likelihood contours for these two additional choices for priors are shown in figures 2 and 3 for the same subsamples discussed above .
for all three priors , and for all the samples considered , the goodness of fit of the model the effective reduced @xmath60 is less than unity , reflecting that the model is an adequate description of the data ( and that the random observational errors have typically been overestimated . )
the differences in the parameter estimates from the top - hat case are small , reflecting the fact that the most critical assumption affecting the results is the one we are most confident in , namely the assignment of @xmath61 for @xmath62 . as is evident from the figures , the likelihood contours
are shifted to the right for the negative bias prior , yielding a higher value for @xmath0 and to the left for the positive bias , favoring a lower @xmath0 .
however , the @xmath63 upper limit is found to be nearly independent of our choice of prior .
these results are summarized in tables 2 and 3 .
the significant detection of nonzero @xmath24 reveals that even in these samples of metal - poor galaxies , there is evidence for a spread in @xmath2 produced by stars .
the best fit to the data is for values of the primordial abundance in the range @xmath64 , depending on the specific model and subsample .
a 2@xmath50 upper limit , @xmath65 , holds for all cases ; this bound is virtually independent of the stellar helium enrichment model .
there are enough data points entering so that even if individual errors are nongaussian , this limit corresponds approximately to a 95% confidence level .
the statistical evidence from this sample indicates that the internally estimated errors are if anything overly conservative , so the quoted limits are also generously estimated .
the only reasonable way to reconcile the data with a higher value of @xmath0 would be some systematic error in common among all the points that is , an in - common mistake biasing the results of all the measurements .
of course , errors of this kind can not be corrected by any purely statistical technique .
even with quite simple and conservative assumptions , our analysis yields a final limit which is sufficiently precise to overconstrain the big bang picture when combined with other data .
our limit of @xmath65 corresponds to a limit on the baryon / photon ratio @xmath66 and a predicted deuterium abundance @xmath67 , in conflict with some recent claims ( e.g. tytler 1996 ) but in accord with others ( e.g. songaila 1996 ) .
it is also consistent with bbn predictions of the li abundance ( fields & olive 1996 , fields 1996 ) .
for the observed microwave background temperature , the baryon density is predicted to be @xmath68 , which begins to conflict with some recent estimates based on models of quasar lyman-@xmath4 absorption ( e.g. , rauch 1997 , weinberg 1997 ) , but accords with other estimates of baryon density ( e.g. , hogan 1997ab ) .
an overall concordance of the big bang picture is certainly possible but will depend on which of these datasets `` gives way '' .
it is clear that the reliability of @xmath0 estimates would improve significantly with comprehensive studies of even one more region similar to i zw 18 .
we would like to thank e. skillman and g. steigman for helpful conversations .
we are grateful to the institute for nuclear theory , seattle , funded by doe , for sponsoring the 1996 workshop on nucleosynthesis in the big bang , stars , and supernovae , where this work was started .
this work was also supported at the university of washington by nasa and nsf , and at the university of minnesota in part by doe grant de - fg02 - 94er-40823 . | we introduce a new statistical method to estimate the primordial helium abundance @xmath0 from observed abundances in a sample of galaxies which have experienced stellar helium enrichment . rather than using linear regression on metal abundance
we construct a likelihood function using a bayesian prior , where the key assumption is that the true helium abundance must always exceed the primordial value . using a sample of measurements compiled from the literature we find estimates of @xmath0 between 0.221 and 0.236 , depending on the specific subsample and prior adopted , consistent with previous estimates either from a linear extrapolation of the helium abundance with respect to metallicity , or from the helium abundance of the lowest metallicity region , i zw 18 .
we also find an upper limit which is insensitive to the specific subsample or prior , and estimate a model - independent bound @xmath1 at 95% confidence , favoring a low cosmic baryon density and a high primordial deuterium abundance .
the main uncertainty is not the model of stellar enrichment but possible common systematic biases in the estimate of @xmath2 in each individual hii region .
# 1 # 1 -.2 in |
with the exponential growth in machine learning ( ml ) datasets sizes and increasing sophistication of ml techniques , there is a growing need for systems that can execute ml algorithms efficiently in parallel on large clusters .
unfortunately , based on our comprehensive survey , we find that existing popular high level parallel abstractions , such as mapreduce @xcite and dryad @xcite , do not efficiently fit many ml applications .
alternatively , designing , implementing , and debugging ml algorithms on low level frameworks such as openmp @xcite or mpi @xcite can be excessively challenging , requiring the user to address complex issues like race conditions , deadlocks , and message passing in addition to the already challenging mathematical code and complex data models common in ml research . in this paper
we describe the culmination of two years of research in collaboration with ml , parallel computing , and distributed systems experts . by focusing on machine learning we designed graphlab , a domain specific parallel abstraction @xcite that fits the needs of the ml community , without sacrificing computational efficiency or requiring ml researchers to redesign their algorithms . in @xcite
we first introduced the graphlab multi - core api to the ml community . in this paper
we build upon our earlier work by refining the graphlab abstraction and extending the graphlab api to the distributed setting .
we provide the first formal presentation of the streamlined graphlab abstraction and describe how the abstraction enabled us to construct a highly optimized c++ api for the distributed setting .
we conduct a comprehensive performance analysis on the amazon elastic cloud ( ec2 ) cloud computing service .
we show that applications created using graphlab outperform equivalent hadoop / mapreduce@xcite implementations by 20 - 60x and match the performance of carefully constructed and fine tuned mpi implementations .
our main contributions are the following : -3pt asurvey of common properties of machine learning algorithms and the limitations of existing parallel abstractions .
[ sec : needforgraphlab ] ) a formal presentation of the graphlab abstraction and how it naturally represents ml algorithms .
[ sec : graphlababstraction ] ) two efficient distributed implementations of the graphlab abstraction ( sec .
[ sec : impldetails ] ) : @xmath0 * chromatic engine : * uses graph coloring to achieve efficient sequentially consistent execution for static schedules . *
locking engine : * uses distributed locking and latency hiding to achieve sequential consistency while supporting prioritized execution .
implementations of three state - of - the - art machine learning algorithms using the graphlab abstraction .
[ sec : applications ] ) an extensive evaluation of graphlab using a 512 processor ( 64 node ) ec2 cluster , including comparisons to hadoop and mpi implementations .
[ sec : experiments ] )
[ sec : needforgraphlab ] * computation model * & * sparse depend . * & * async . comp . * & * iterative * & * prioritized ordering * & * sequentially consistent @xmath1 * & * distributed * + mpi@xcite & messaging & yes & yes & yes & n / a @xmath2 & n / a @xmath2 & yes + mapreduce@xcite & par . data - flow & no & no & extensions@xmath3 & n / a & n / a & yes + dryad@xcite & par .
data - flow & yes & no & extensions@xmath4 & n / a & n / a & yes + pregel@xcite / bpgl@xcite & graphbsp@xcite & yes & no & yes & n / a & n / a & yes + piccolo@xcite & distr .
map@xmath5 & n / a @xmath5 & yes & yes & no & accumulators & yes + pearce et.al.@xcite & graph visitor & yes & yes & yes & yes & no & no + * graphlab * & * graphlab * & yes & yes & yes & yes@xmath6 & yes & yes + the graphlab abstraction is the product of several years of research in designing and implementing systems for statistical inference in probabilistic graphical models .
early in our work @xcite , we discovered that the high - level parallel abstractions popular in the ml community such as mapreduce @xcite and parallel blas @xcite libraries are unable to express statistical inference algorithms efficiently .
our work revealed that an efficient algorithm for graphical model inference should explicitly address the _ sparse dependencies _ between random variables and adapt to the input data and model parameters .
guided by this intuition we spent over a year designing and implementing various machine learning algorithms on top of low - level threading primitives and distributed communication frameworks such as openmp @xcite , cilk++ @xcite and mpi @xcite . through this process , we discovered the following set of core algorithmic patterns that are common to a wide range of machine learning techniques .
following , we detail our findings and motivate why a new framework is needed ( see table [ table : abstrcomptable ] ) .
* sparse computational dependencies : * many ml algorithms can be factorized into local * dependent * computations which examine and modify only a small sub - region of the entire program state . for example , the conditional distribution of each random variable in a large statistical model typically only depends on a small subset of the remaining variables in the model .
this computational sparsity in machine learning arises naturally from the statistical need to reduce model complexity .
parallel abstractions like mapreduce @xcite require algorithms to be transformed into an embarrassingly parallel form where computation is * independent*. unfortunately , transforming ml algorithms with computational _ dependencies _ into the embarrassingly parallel form needed for these abstractions is often complicated and can introduce substantial algorithmic inefficiency @xcite .
alternatively , data flow abstractions like dryad @xcite , permit directed acyclic dependencies , but struggle to represent cyclic dependencies common to iterative ml algorithms .
finally , graph - based messaging abstractions like pregel @xcite provide a more natural representation of computational dependencies but require users to explicitly manage communication between computation units .
* asynchronous iterative computation : * from simulating complex statistical models , to optimizing parameters , many important machine learning algorithms iterate over local computation kernels .
furthermore , many iterative machine learning algorithms benefit from @xcite and in some cases require @xcite asynchronous computation . unlike * synchronous * computation , in which all kernels are computed simultaneously ( in parallel ) using the previous values for dependent parameters , * asynchronous * computation requires that the local computation kernels use the most recently available values .
abstractions based on bulk data processing , such as mapreduce @xcite and dryad @xcite were not designed for iterative computation .
while recent projects like mapreduce online @xcite , spark @xcite , twister @xcite , and nexus @xcite extend mapreduce to the iterative setting , they do not support asynchronous computation .
similarly , parallel graph based abstractions like pregel @xcite and bpgl @xcite adopt the bulk synchronous parallel ( bsp ) model @xcite and do not naturally express asynchronous computation .
* sequential consistency : * by ensuring that all parallel executions have an equivalent sequential execution , sequential consistency eliminates many challenges associated with designing , implementing , and testing parallel ml algorithms .
in addition , many algorithms converge faster if sequential consistency is ensured , and some even require it for correctness . however , this view is not shared by all in the ml community . recently ,
@xcite advocate soft - optimization techniques ( e.g. , allowing computation to intentionally race ) , but we argue that such techniques do not apply broadly in ml . even for the algorithms evaluated in @xcite , the conditions under which the soft - optimization techniques work are not well understood and may fail in unexpected ways on different datasets .
indeed , for some machine learning algorithms sequential consistency is strictly required .
for instance , gibbs sampling @xcite , a popular inference algorithm , requires sequential consistency for statistical correctness , while many other optimization procedures require sequential consistency to converge ( fig .
[ fig : synchronouspmf ] demonstrates that the prediction error rate of one of our example problems is dramatically better when computation is properly asynchronous . ) . finally , as @xcite demonstrates , the lack of sequential consistency can dramatically increase the time to convergence for stochastic optimization procedures . by designing an abstraction which enforces sequentially consistent computation , we eliminate much of the complexity introduced by parallelism , allowing the ml expert to focus on algorithm design and correctness of numerical computations .
debugging mathematical code in a parallel program which has random errors caused by non - deterministic ordering of concurrent computation is particularly unproductive .
the discussion of sequential consistency is relevant only to frameworks which support asynchronous computation .
piccolo @xcite provides a limited amount of consistency by combining simultaneous writes using accumulation functions .
however , this only protects against single write races , but does not ensure sequential consistency in general . the parallel asynchronous graph traversal abstraction by pearce et .
@xcite does not support any form of consistency , and thus is not suitable for a large class of ml algorithms . ) comparing prediction error when running sequentially consistent asynchronous iterations vs inconsistent asynchronous iterations over a five node distributed cluster .
consistent iterations converge rapidly to a lower error while inconsistent iterations oscillate and converge slowly.,scaledwidth=25.0% ] * prioritized ordering : * in many ml algorithms , iterative computation converges asymmetrically .
for example , in parameter optimization , often a large number of parameters will quickly converge after only a few iterations , while the remaining parameters will converge slowly over many iterations @xcite .
if we update all parameters equally often , we could waste substantial computation recomputing parameters that have effectively converged .
conversely , by focusing early computation on more challenging parameters first , we can potentially reduce computation .
adaptive prioritization can be used to focus iterative computation where it is needed . the only existing framework to support this is the parallel graph framework by pearce et .
al . @xcite .
the framework is based on the visitor - pattern and prioritizes the ordering of visits to vertices .
graphlab however , allows the user to define _ arbitrary ordering _ of computation , and our implementation supports efficient fifo and priority - based scheduling .
* rapid development : * machine learning is a rapidly evolving field with new algorithms and data - sets appearing weekly . in many cases these algorithms
are not yet well characterized and both the computational and statistical properties are under active investigation .
large - scale parallel machine learning systems must be able to adapt quickly to changes in the data and models in order to facilitate rapid prototyping , experimental analysis , and model tuning . to achieve these goals ,
an effective high - level parallel abstraction must hide the challenges of parallel algorithm design , including race conditions , deadlock , state - partitioning , and communication .
[ sec : graphlababstraction ] using the ideas from the previous section , we extracted a single coherent computational pattern : _ asynchronous parallel computation on graphs _ with a _ sequential _ model of computation .
this pattern is both sufficiently expressive to encode a wide range of ml algorithms , and sufficiently restrictive to enable efficient parallel implementations .
the graphlab abstraction consists of three main parts , the data graph , the update function , and the sync operation .
the data graph ( sec . [ sec : data_graph ] ) represents user modifiable program state , and both stores the mutable user - defined data and encodes the sparse computational dependencies .
the update functions ( sec . [ sec : update_functions ] ) represent the factorized user computation and operate on the data graph by transforming data in small overlapping contexts called scopes . finally , the sync operation ( sec . [ sec : sync_operation ] ) is used to maintain global aggregate statistics of the data graph .
we now present the graphlab abstraction in greater detail . to make these ideas more concrete
, we will use the pagerank algorithm @xcite as a running example . while pagerank is not a common machine learning algorithm , it is easy to understand and shares many properties common to machine learning algorithms .
[ ex : pagerank_intro ] the pagerank algorithm recursively defines the rank of a webpage @xmath7 : @xmath8 in terms of the ranks of those pages that link to @xmath7 and the weight @xmath9 of the link as well as some probability @xmath10 of randomly jumping to that page . the pagerank algorithm , simply iterates eq . ( [ eqn : pagerank ] ) until the individual pagerank values converge ( i.e. , change by less than some small @xmath11 ) .
[ sec : data_graph ] the graphlab abstraction stores the program state as an undirected graph called the * data graph*. the data graph @xmath12 is a container which manages the user defined data @xmath13 .
here we use the term data " broadly to refer to model parameters , algorithmic state , and even statistical data .
the user can associate arbitrary data with each vertex @xmath14 and edge @xmath15 in the graph . since some machine learning applications require directed edge data ( e.g. , weights on directed links in a web - graph ) we provide the ability to store and retrieve data associated with directed edges . while the graph data is mutable ,
the graph structure is _ _ static _ _ and can not be changed during execution . the data graph for pagerank is directly obtained from the web graph , where each vertex corresponds to a web page and each edge represents a link .
the vertex data @xmath16 stores @xmath17 , the current estimate of the pagerank , and the edge data @xmath18 stores @xmath19 , the directed weight of the link .
the data graph is convenient for representing the state of a wide range of machine learning algorithms .
for example , many statistical models are efficiently represented by undirected graphs @xcite called markov random fields ( mrf ) .
the data graph is derived directly from the mrf , with each vertex representing a random variable . in this case
the vertex data and edge data may store the local parameters that we are interested in learning .
[ sec : update_functions ] computation is encoded in the graphlab abstraction via user defined update functions . an * update function * is a stateless procedure which modifies the data within the * scope * of a vertex and schedules the future execution of other update functions .
the scope of vertex @xmath7 ( denoted by @xmath20 ) is the data stored in @xmath7 , as well as the data stored in all adjacent vertices and edges as shown in fig .
[ fig : scope ] . of vertex @xmath21 .
each of the gray cylinders represent a block of user defined data and is associated with a vertex or edge .
the * scope * of vertex @xmath21 is illustrated by the region containing vertices @xmath22 .
an * update function * applied to vertex @xmath21 is able to read and modify all the data in @xmath23 ( vertex data @xmath24 , @xmath25 , @xmath26 , and @xmath27 and edge data @xmath28 , @xmath29 , and @xmath30 ) .
, title="fig : " ] [ fig : scope ] a graphlab update function takes as an input a vertex @xmath7 and its scope @xmath20 and returns the new version of the scope as well as a set of tasks @xmath31 which encodes future task executions .
@xmath32 after executing an update function the modified scope data in @xmath20 is written back to the data graph .
each * task * in the set of tasks @xmath31 , is a tuple @xmath33 consisting of an update function @xmath34 and a vertex @xmath7 .
all returned task @xmath31 are executed _ eventually _ by running @xmath35 following the execution semantics described in sec .
[ sec : execution_model ] . rather than adopting a message passing or data flow model as in @xcite , graphlab allows the user defined update functions complete freedom to read and modify any of the data on adjacent vertices and edges .
this simplifies user code and eliminates the need for the users to reason about the movement of data . by controlling
what tasks are added to the task set , graphlab update functions can efficiently express adaptive computation .
for example , an update function may choose to reschedule its neighbors only when it has made a substantial change to its local data .
the update function mechanism allows for _ asynchronous computation _ on the _ sparse dependencies _ defined by the data graph .
since the data graph permits the expression of general cyclic dependencies , _ iterative computation _ can be represented easily .
many algorithms in machine learning can be expressed as simple update functions .
for example , probabilistic inference algorithms like gibbs sampling @xcite , belief propagation @xcite , expectation propagation @xcite and mean field variational inference @xcite can all be expressed using update functions which read the current assignments to the parameter estimates on neighboring vertices and edges and then apply sampling or optimization techniques to update parameters on the local vertex . the update function for pagerank ( defined in alg .
[ alg : pagerank_updatefunction ] ) computes a weighted sum of the current ranks of neighboring vertices and assigns it as the rank of the current vertex .
the algorithm is adaptive : neighbors are listed for update only if the value of current vertex changes more than a predefined threshold .
[ alg : pagerank_updatefunction ] @xmath36 @xmath37 ( ) @xmath38}$ ] @xmath39 [ sec : sync_operation ] in many ml algorithms it is necessary to maintain global statistics describing data stored in the data graph .
for example , many statistical inference algorithms require tracking of global convergence estimators .
alternatively , parameter estimation algorithms often compute global averages or even gradients to tune model parameters . to address these situations ,
the graphlab abstraction expresses global computation through the * sync operation * , which aggregates data across all vertices in the graph in a manner analogous to mapreduce .
the results of the sync operation are stored globally and may be accessed by all update functions . because graphlab is designed to express iterative computation , the sync operation runs repeated at fixed user determined intervals to ensure that the global estimators remain fresh .
the sync operation is defined as a tuple @xmath40 consisting of a unique key , three user defined functions , an initial accumulator value , and an integer defining the interval between sync operations .
the sync operation uses the @xmath41 and @xmath42 functions to perform a _
global synchronous reduce _ where @xmath41 aggregates vertex data and @xmath42 combines intermediate @xmath41 results .
the @xmath43 function performs a transformation on the final value and stores the result .
the ` key ` can then be used by update functions to access the most recent result of the sync operation .
the sync operation runs periodically , approximately every @xmath44 update function calls .
we can compute the second most popular page on the web by defining the following sync operation : @xmath45 \end{aligned}\ ] ] where the accumulator taking on the initial value as the empty array @xmath46 . the function @xmath47 " returns the two pages with the highest pagerank in the set @xmath48 .
after the global reduction , the @xmath49 array will contain the top two pages and @xmath50 $ ] in @xmath43 extracts the second entry .
we may want to update the global estimate every @xmath51 vertex updates .
[ sec : execution_model ] the graphlab execution model , presented in alg .
[ alg : execution_model ] , follows a simple single loop semantics .
the input to the graphlab abstraction consists of the data graph @xmath52 , an update function @xmath53 , an initial set of tasks @xmath31 to update , and any sync operations . while there are tasks remaining in @xmath31 , the algorithm removes ( line [ ln : pop ] ) and executes ( line [ ln : update ] ) tasks , adding any new tasks back into @xmath31 ( line [ ln : push ] ) .
the appropriate sync operations are executed whenever necessary . upon completion , the resulting data graph and synced values
are returned to the user .
[ alg : execution_model ] the exact behavior of @xmath54 ( line [ ln : pop ] ) is up to the implementation of the graphlab abstraction .
the only guarantee the graphlab abstraction provides is that @xmath55 removes and returns an update task in @xmath31 . the flexibility in the order in which @xmath55 removes tasks from @xmath31 provides the opportunity to balance features with performance constraints .
for example , by restricting task execution to a fixed order , it is possible to optimize memory layout .
conversely , by supporting _
prioritized ordering _
it is possible to implement more advanced ml algorithms at the expense run - time overhead . in our implementation
( see sec .
[ sec : impldetails ] ) we support fixed execution ordering ( chromatic engine ) as well as fifo and prioritized ordering ( locking engine ) . the graphlab abstraction presents a rich _ sequential model _ that is automatically translated into a _ parallel execution _ by allowing multiple processors to remove and execute update tasks simultaneously . to retain the same _ sequential _ execution semantics
we must ensure that overlapping computation is not run simultaneously .
however , the extent to which computation can _ safely _ overlap depends on the user defined update function . in the next section we introduce several * consistency models * that allow the runtime to optimize the parallel execution while maintaining consistent computation .
[ sec : consistency ] a parallel implementation of graphlab must guarantee sequential consistency @xcite over update tasks and sync operations .
we define sequential consistency in the context of the graphlab abstraction as : [ dfn : sequential_consistency ] for every parallel execution of the graphlab abstraction , there exists a sequential ordering on all executed update tasks and sync operations which produces the same data graph and synced global values .
a simple method to achieve sequential consistency among update functions is to ensure that the scopes of concurrently executing update functions do not overlap .
we refer to this as the * full consistency * model ( see fig . [
fig : scoperings ] ) .
full consistency limits the potential parallelism since concurrently executing update functions must be at least two vertices apart ( see fig .
[ fig : scopetradeoff ] ) .
even in moderately dense data graphs , the amount of available parallelism could be low .
depending on the actual computation performed within the update function , additional relaxations can be safely made to obtain more parallelism without sacrificing sequential consistency .
we observed that for many machine learning algorithms , the update functions do not need full read / write access to all of the data within the scope .
for instance , the pagerank update in eq .
( [ eqn : pagerank ] ) only requires read access to edges and neighboring vertices . to provide greater parallelism
while retaining sequential consistency , we introduced the * edge consistency * model . if the edge consistency model is used ( see fig .
[ fig : scoperings ] ) , then each update function has exclusive read - write access to its vertex and adjacent edges but read only access to adjacent vertices .
this increases parallelism by allowing update functions with slightly overlapping scopes to safely run in parallel ( see fig .
[ fig : scopetradeoff ] ) . finally , for many machine learning algorithms there is often some initial data pre - processing which only requires read access to adjacent edges and write access to the central vertex . for these algorithms
, we introduced the weakest * vertex consistency * model ( see fig .
[ fig : scoperings ] ) .
this model has the highest parallelism but only permits fully independent ( ` map ` ) operations on vertex data .
while sequential consistency is essential when designing , implementing , and debugging complex ml algorithms , an adventurous user @xcite may want to relax the theoretical consistency constraints .
thus , we allow users to choose a weaker consistency model at their own risk .
[ fig : consistency ]
in our prior work @xcite we implemented an optimized shared memory graphlab runtime using pthreads . to fully utilize clouds composed of multi - core instances , we implemented distributed graphlab on top of our shared memory runtime . as we transitioned to the distributed setting , we had to address two main design challenges : @xmath0 * distributed graph : * to manage the data graph across multiple machines we needed a method to efficiently load , distribute , and maintain the graph data - structure over a potentially varying number of machines .
* distributed consistency : * to support the various consistency models in the distributed setting , we needed an efficient mechanism to ensure safe read - write access to the data - graph .
we first implemented a data graph representation that allows for rapid repartitioning across different loads cluster sizes .
next , we implemented two versions of the graphlab engine for the distributed setting , making use of asynchronous communication implemented on top of tcp sockets .
the first engine is the simpler chromatic engine ( sec .
[ sec : chromaticengine ] ) which uses basic graph coloring to manage consistency .
the second is a locking engine ( sec .
[ sec : lockingengine ] ) which uses distributed locks . efficiently implementing the data graph in the distributed setting
requires balancing computation , communication , and storage . to ensure balanced computation and storage
, each machine must hold only a small fraction of the data graph . at the same time we would like to minimize the number of edges that cross partitions , to reduce the overall state that must be synchronized across machines .
finally , the cloud setting introduces an additional challenge . because the number of machines available may vary with the research budget and the performance demands
, we must be able to quickly load the data - graph on varying sized cloud deployments . to resolve these challenges
, we developed a graph representation based on two - phased partitioning which can be efficiently load balanced on arbitrary cluster sizes .
the graph is initially over - partitioned by an expert , or by using a graph partitioning heuristic ( for instance metis @xcite ) into @xmath56 parts where @xmath56 is much greater than the number of machines ( see fig .
[ fig : graphcut ] ) .
each part is stored as a different file possibly on a distributed store ( hdfs , amazon s3 ) . the connectivity structure of the @xmath56 parts is then represented as a * meta - graph * with @xmath56 vertices .
each vertex of the * meta - graph * represents a partition , and is weighted by the amount of data it stores .
each edge is weighted by the number of edges crossing the partitions . distributed loading is accomplished by performing a fast balanced partition of the meta - graph into @xmath57 parts , and each machine constructs its local partition of the graph by merging its assigned files . to facilitate communication
, each machine also stores the * ghost * of its local partition : the set of vertices and edges adjacent to the partition boundary ( see fig .
[ fig : atoms ] ) .
the ghosts conveniently act as local caches for their true counterparts across the network , and cache coherence is managed using versioning @xcite .
the two - stage partitioning technique allows one graph partition to be reused for different numbers of machines without incurring a repartitioning step .
a study on the quality of the two stage partitioning scheme is beyond the scope of this paper , though simple experiments using graphs obtained from @xcite suggest that the performance is comparable to direct partitioning .
the graphlab * engine * emulates the _ execution model _ defined in sec . [ sec : execution_model ] and is responsible for executing update task and sync operations , maintaining the update task set @xmath31 , and ensuring sequential consistency with respect to the appropriate consistency model ( see sec .
[ sec : consistency ] ) . as discussed earlier in sec .
[ sec : execution_model ] , variations in how @xmath31 is maintained and the order in which elements are removed is up to the implementation and can affect performance and expressiveness . to evaluate this trade - off
we built the low - overhead * chromatic engine * , which executes @xmath31 in a fixed order , and the more expressive * locking engine * which executes @xmath31 in an asynchronous prioritized order .
the * chromatic engine * imposes a static ordering on the update task set @xmath31 by executing update task in a canonical order ( e.g. , the order of their vertices ) .
a classic technique @xcite to achieve a sequentially consistent parallel execution of a set of dependent tasks is to construct a vertex coloring .
a vertex coloring assigns a color to each vertex such that no adjacent vertices share the same color .
given a vertex color of the data graph , we can satisfy the _ edge consistency model _ by executing , _ in parallel _
, all update tasks with vertices of the same color before proceeding to the next color .
the sync operation can be run safely between colors .
we can satisfy the other consistency models simply by changing how the vertices are colored .
the stronger _ full consistency model _ is satisfied by constructing a second - order vertex coloring ( i.e. , no vertex shares the same color as any of its distance two neighbors ) .
conversely , we can trivially satisfy the _
vertex consistency model _ by assigning all vertices the same color . in the distributed
setting it is necessary to both prevent overlapping computation , and also synchronize any changes to ghost vertices or edge data between colors .
a full communication barrier is enforced between color phases to ensure completion of all data synchronization before the next color begins . to maximize network throughput and to minimize time spent in the barrier , synchronization of modified data
is constantly performed in the background while update functions are executing .
approximate graph coloring can be quickly obtained through graph coloring heuristics .
furthermore , many ml problems have obvious colorings . for example
, many optimization problems in ml are naturally expressed as bipartite graphs ( sec .
[ sec : applications ] ) , while problems based upon templated bayesian networks @xcite can be easily colored by the expert through inspection of the template model @xcite .
the simple design of the chromatic engine is made possible by the explicit communication structure defined by the data graph , allowing data to be pushed directly to the machines requiring the information .
in addition , the cache versioning mechanism further optimizes communication by only transmitting modified data .
the advantage of the chromatic engine lies its predictable execution schedule .
repeated invocations of the chromatic engine will always produce identical update sequences , regardless of the number of machines used .
this property makes the chromatic engine highly suitable for testing and debugging purposes .
we provide a distributed debugging tool which halts at the end of each color , allowing graph data and scheduler state to be queried . even though the chromatic engine is a complete implementation of the graphlab abstraction as defined in sec .
[ sec : graphlababstraction ] , it does not provide sufficient scheduling flexibility for many interesting applications .
here we describe an implementation which directly extends from a typical shared memory implementation to the distributed case . in the shared memory implementation of graphlab , the consistency models were implemented by associating a readers - writer lock with each vertex .
the vertex consistency model is achieved by acquiring a write lock on the central vertex of each requested scope .
the edge - consistency model is achieved by acquiring a write lock on the central vertex , and read locks on adjacent vertices .
finally , full consistency is achieved by acquiring write locks on the central vertex and all adjacent vertices . the main execution loop in the shared memory setting uses worker threads to pull tasks from the scheduler , acquire the required locks , evaluate the task , and then release the locks .
this loop is repeated until the scheduler is empty .
the sync operation is triggered by a global shared - memory task counter .
periodically , as sync operations become ready , all threads are forced to synchronize in a barrier to execute the sync operation . in
the distributed setting , the same procedure is used . however , since the graph is partitioned , we restrict each machine to only run updates on vertices it owns .
the ghost vertices / edges ensure that the update will always have direct memory access to all information in the scope , and distributed locks are used to ensure that the ghost is up to date .
finaly , the scheduling flexibility permitted by the abstraction allow the use of efficient approximate fifo / priority task - queues .
distributed termination is evaluated using a multi - threaded variant of the distributed consensus algorithm described in @xcite . since the distributed locking and synchronization introduces substantial latency ,
we rely on several techniques to reduce latency and hide its effects @xcite .
first , the ghosting system provides caching capabilities eliminating the need to wait on data that has not changed remotely .
second , all locking requests and synchronization calls are _ pipelined _ allowing each thread to request multiple scope locks simultaneously and then evaluate the update tasks only when the locks are satisfied . the _
lock pipelining _ technique we implemented shares similarities to the continuation passing method in @xcite .
[ sec : applications ] to evaluate the performance of the distributed graphlab runtime as well as the representational capabilities of the graphlab abstraction , we implemented several state - of - the - art ml algorithms .
each algorithm is derived from active research in machine learning and is applied to real world data sets with different structures ( see table [ table : appinp ] ) , update functions , and sync operations .
in addition , each application tests different features of the distributed graphlab framework .
the source and datasets for all the applications may be obtained from [ * http://graphlab.org * ] .
data * & * edge data * & * update complexity * & * shape * & * partition * & * engine * + netflix & 0.5 m & 99 m & @xmath58 & @xmath59 & @xmath60 & bipartite & random & chromatic + coseg & 10.5 m & 31 m & @xmath61 & @xmath62 & @xmath63 & 3d grid & frames & locking + ner & 2 m & 200 m & @xmath64 & @xmath65 & @xmath63 & bipartite & random & chromatic + [ table : appinp ] [ sec : netflix ] the netflix movie recommendation task @xcite uses _ collaborative filtering _ to predict the movie ratings of users , based on the ratings of similar users . the * alternating least squares * ( * als * ) algorithm is commonly used in collaborative filtering .
the input to als is a sparse users by movies matrix @xmath66 containing the movie ratings of each user .
the algorithm proceeds by computing a low rank approximate matrix factorization : @xmath67{als_eqn}\ ] ] where @xmath68 and @xmath69 are rank @xmath70 matrices .
the als algorithm alternates between computing the least - squares approximation for @xmath68 or @xmath69 while holding the other fixed .
the quality of the approximation depends on the magnitude of @xmath70 , as shown in fig .
[ fig : alsaccuracy ] .
while als may not seem like a graph algorithm , it can be represented elegantly using the graphlab abstraction .
the _ sparse _ matrix @xmath71 defines a bipartite graph ( see table [ table : appinp ] ) connecting each user with the movie he / she rated .
the edge data contains the rating for a movie - user pair .
the vertex data for users and movies contains the corresponding row in @xmath68 and column in @xmath69 respectively .
the als algorithm can be encoded as an update function that recomputes the least - squares solution for the current movie or user given the neighboring users or movies .
each local computation is accomplished using blas / lapack linear algebra library for efficient matrix operations .
the bipartite graph is naturally two colored , thus the program is executed using the chromatic engine with two colors .
a sync operation is used to compute the prediction error during the run . due to the density of the graph , a random
partitioning was used .
[ sec : coseg ] video cosegmentation automatically identifies and clusters spatio - temporal segments of video ( see fig .
[ fig : videoframe ] ) that share similar texture and color characteristics .
the resulting segmentation ( see fig . [
fig : cosegframe ] ) can be used in scene understanding and other computer vision and robotics applications .
previous cosegmentation methods @xcite have focused on processing frames in isolation . as part of this work , we developed a joint cosegmentation algorithm that processes all frames simultaneously and therefore is able to model _ temporal _ stability .
we preprocessed @xmath72 frames of high - resolution video by coarsening each frame to a regular grid of @xmath73 rectangular * super - pixels*. each super - pixel stores the color and texture statistics for all the raw pixels in its domain .
the coseg algorithm predicts the best label ( e.g. , sky , building , grass , pavement , trees ) for each super pixel using * gaussian mixture model * ( * gmm * ) @xcite in conjunction with * loopy belief propagation * ( * lbp * ) @xcite .
the gmm estimates the best label given the color and texture statistics in the super - pixel .
the algorithm operates by connecting neighboring pixels in time and space into a large three - dimensional grid and uses lbp to smooth the local estimates .
we combined the two algorithms so that coseg alternates between lbp to compute the label for each super - pixel given the current gmm and then updating the gmm given the labels from lbp .
the graphlab data graph structure for video cosegmentation is the three dimensional grid graph ( see table [ table : appinp ] ) .
the vertex data stores the current label distribution as well as the color and texture statistics for each super - pixel .
the edge data stores the parameters needed for the lbp algorithm .
the parameters for the gmm are maintained using the sync operation .
the graphlab update function executes the lbp local iterative update .
we implement the state - of - the - art adaptive update schedule described by @xcite , where updates which are expected to change vertex values significantly are prioritized .
we therefore make use of the locking engine with the approximate priority ordering task queue .
furthermore , the graph has a natural partitioning by slicing across frames .
this also allows feature processing of the video to be performed in an embarrassingly parallel fashion , permitting the use of hadoop for preprocessing .
[ sec : coem ] named entity recognition ( ner ) is the task of determining the type ( e.g. , ` person ` , ` place ` , or ` thing ` ) of a * noun - phrase * ( e.g. , _ obama _ , _ chicago _ , or _ car _ ) from its * context * ( e.g. , _
`` president _ _ '' _ , _ `` lives near _ _ . '' _ , or _
`` bought a _ _ . '' _ ) .
ner is used in many natural language processing applications as well as information retrieval . in this application
we obtained a large crawl of the web and we counted the number of occurrences of each noun - phrase in each context . starting with a small seed set of pre - labeled noun - phrases , the coem algorithm @xcite labels the remaining noun - phrases and contexts ( see table [ tbl : coemwords ] ) by alternating between estimating the best assignment to each noun - phrase given the types of its contexts and estimating the type of each context given the types of its noun - phrases .
the graphlab data graph for the ner problem is bipartite with vertices corresponding to each noun - phrase on one side and vertices corresponding to each context on the other .
there is an edge between a noun - phrase and a context if the noun - phrase occurs in the context .
the vertex for both noun - phrases and contexts stores the estimated distribution over types .
the edge stores the number of times the noun - phrase appears in that context .
the ner computation is represented in a simple graphlab update function which computes a weighted sum of probability tables stored on adjacent vertices and then normalizes .
once again , the bipartite graph is naturally two colored , allowing us to use the chromatic scheduler . due to the density of the graph , a random partitioning
since the ner computation is relatively light weight and uses only simple floating point arithmetic ; combined with the use of a random partitioning , this application stresses the overhead of the graphlab runtime as well as the network .
[ sec : otherapplications ] in the course of our research , we have also implemented several other algorithms , which we describe briefly : * gibbs sampling on a markov random field .
* the task is to compute a probability distribution for a graph of random variables by sampling .
algorithm proceeds by sampling a new value for each variable in turn conditioned on the assignments of the neighboring variables .
strict sequential consistency is necessary to preserve statistical properties @xcite .
* bayesian probabilistic tensor factorization ( bptf ) .
* this is a probabilistic markov - chain monte carlo version of alternative least squares that also incorporates time - factor into the prediction . in this case , the tensor @xmath71 is decomposed into three matrices @xmath74 which can be represented in graphlab as a tripartite graph .
in addition , graphlab has been used successfully in several other research projects like clustering communities in the twitter network , collaborative filtering for bbc tv data as well as non - parametric bayesian inference .
[ sec : experiments ] we evaluated graphlab on the three applications ( netflix , coseg and ner ) described above using important large - scale real - world problems ( see table [ table : appinp ] ) .
we used the chromatic engine for the netflix and ner problems and the locking engine for the coseg application .
equivalent hadoop and mpi implementations were also tested for both the netflix and the ner application .
an mpi implementation of the asynchronous prioritized lbp algorithm needed for coseg requires building an entirely new asynchronous sequentially consistent system and is beyond the scope of this work .
experiments were performed on amazon s elastic computing cloud ( ec2 ) using up to 64 high - performance cluster ( hpc ) instances ( ` cc1.4xlarge ` ) .
the hpc instances ( as of february 2011 ) have 2 x intel xeon x5570 quad - core nehalem architecture with 22 gb of memory , connected by a low latency 10 gigabit ethernet network .
all our timings include data loading time and are averaged over three or more runs .
our principal findings are : @xmath0 _ graphlab is fast ! _ on equivalent tasks , graphlab outperforms hadoop by 20x-60x and is as fast as custom - tailored mpi implementations .
graphlab s performance scaling improves with higher computation to communication ratios .
when communication requirements are high , graphlab can saturate the network , limiting scalability .
the graphlab abstraction more compactly expresses the netflix , ner and coseg algorithms than mapreduce or mpi .
[ sec : scaling ] in fig .
[ fig : speedupmain ] we present the parallel speedup of graphlab when run on 4 to 64 hpc nodes .
speedup is measured relative to the 4 hpc node running time . on each node
, graphlab spawned eight shared memory engine threads ( matching the number of cores ) .
of the three applications , coseg demonstrated the best parallel speedup , achieving nearly ideal scaling up to 32 nodes and moderate scaling up to 64 nodes .
the excellent scaling of the coseg application can be attributed to its sparse data graph ( maximum degree 6 ) and a computationally intensive update function .
while netflix demonstrated reasonable scaling up to 16 nodes , ner achieved only modest 3x improvement beyond 16x or more nodes .
we attribute the poor scaling performance of ner to the large vertex data size ( 816 bytes ) , dense connectivity , and poor partitioning ( random cut ) which resulted in substantial communication overhead per iteration .
[ fig : networkpernode ] shows for each application , the average number of bytes transmitted by each node per second as the cluster size is increased . beyond 16 nodes ,
it is evident that ner saturates the network , with each machine sending at a rate of over 100 mb per second .
note that fig .
[ fig : networkpernode ] plots the _ average _ transmission rate and the peak rate could be significantly higher .
to better understand the effect of the ratio between computation and communication on the scaling of the graphlab abstraction , we varied the computational complexity of the netflix experiment .
the amount of computation performed in netflix can be controlled by varying @xmath70 : the dimensionality of the approximating matrices in eq .
( [ eqn : als ] ) . in fig .
[ fig : compcomplex ] we plot the speedup achieved for varying values of @xmath70 and the corresponding number of _ instructions per byte _ ( ipb ) of data accessed .
the speedup at 64 nodes increases quickly with increasing ipb indicating that speedup is strongly coupled .
[ sec : netflixandner ] in this section , we compare our graphlab implementation of the netflix application and the ner application to an algorithmically equivalent hadoop and mpi implementations .
* mapreduce / hadoop . *
we chose to compare against hadoop , due to its wide acceptance in the machine learning community for large scale computation ( for example the mahout project @xcite ) .
fair comparison is difficult since hadoop is implemented in java while ours is highly optimized c++
. additionally , to enable fault tolerance , hadoop stores interim results to disk . in our experiments , to maximize hadoop s performance , we reduced the hadoop distributed filesystem s ( hdfs ) replication factor to one , eliminating fault tolerance .
a significant amount of our effort was spent tuning the hadoop job parameters to improve performance .
[ fig : alshadoop ] shows the running time for one iteration of netflix application on graphlab , hadoop and mpi ( @xmath75 for all cases ) , using between 4 and 64 nodes .
the hadoop evaluation makes use of the of sebastian schelter contribution to the mahout project , while the mpi implementation was written from scratch .
we find that graphlab performs the same computation between * 40x-60x * times faster than hadoop .
[ fig : coemhadoop ] plots the running for one iteration of the ner application on graphlab , hadoop and mpi .
the hadoop implementation was aggressively optimized : we implemented specialized binary marshaling methods which improve performance by 5x over a baseline implementation .
[ fig : coemhadoop ] shows that the graphlab implementation of ner obtains a 20 - 30x speedup over hadoop .
part of the performance of graphlab over hadoop can be explained by implementation differences , but it is also easy to see that the graphlab representation of both ner and als is inherently more efficient .
for instance the case of ner , when implemented in hadoop , the map - function , normally the cornerstone of embarrassing - parallelism in mapreduce essentially does no work .
the map only serves to emit the vertex probability table for every edge in the graph , which corresponds to over 100 gigabytes of hdfs writes occurring between the map and reduce stage . the cost of this operation can easily be multiplied by factor of three if hdfs replication is turned on .
comparatively , the graphlab update function is simpler as users do not need to explicitly define the flow of information from the map to the reduce , but simply modifies data in - place . in the case of such iterative computation , graphlab s knowledge of the dependency structure
allow modified data to be communicated directly to the destination .
overall , we can attribute graphlab s superior performance over hadoop to the fact that the abstraction is a much better fit .
it presents a simpler api to the programmer and through the data graph , and informs graphlab about the communication needs of the program .
* in order to analyze the cost of using a higher level abstraction we implemented efficient versions of the netflix and ner applications using mpi . the implementations made use of synchronous mpi collective operations for communication .
the final performance results can be seen in fig .
[ fig : alshadoop ] and fig .
[ fig : coemhadoop ] .
we observe that the performance of mpi and graphlab implementations are similar and conclude that graphlab does not impose significant performance penalty .
graphlab also has the advantage of being a higher level abstraction and is easier to work with .
the coseg application makes use of dynamic prioritized scheduling which requires the locking engine ( sec .
[ sec : lockingengine ] ) .
to the best of our knowledge , there are no other abstractions which provide the dynamic asynchronous scheduling as well as the sequentially consistent sync ( reduction ) capabilities required by this application . in fig .
[ fig : speedupmain ] we demonstrate that the locking engine can provide significant scalability and performance on the large 10.5 million vertex graph used by this application , achieving a 10x speedup with 16x more machines .
we also observe from fig .
[ fig : weakscalability ] that the locking engine provides nearly optimal weak scalability .
the runtime does not increase significantly as the size of the graph increases proportionately with the number of processors .
we can attributed this to the properties of the graph partition where the number of edges crossing machines increases linearly with the number of processors , resulting in low communication volume . in fig .
[ fig : lockevaluation ] we further investigate the properties of the distributed lock implementation described in sec .
[ sec : lockingengine ] .
the evaluation is performed on a tiny 32-frame ( 192k vertices ) problem on a 4 node cluster .
two methods of cutting the graph is explored .
the first method is an optimal partition " where the frames are distributed evenly in 8 frame blocks to each machine .
the second method is a `` worst case partition '' where the frames are striped across the machines ; this is designed to stress the distributed lock implementation since every scope acquisition is forced to acquire a remote lock .
the maximum number of lock requests allowed in the pipeline is varied ( ` maxpending ` ) .
the baseline evaluation is optimal partitioning with the ` maxpending ` set to zero .
[ fig : lockevaluation ] demonstrates that on well - partitioned models , increasing the maximum number of pending locks from @xmath76 to @xmath77 increases performance significantly .
however , we observe diminishing returns as ` maxpending ` is further increased to @xmath78 . on the other hand
when the partitioning is poor , increasing the number of pending locks to @xmath78 improves performance significantly . to help put costs in perspective , we plot a price - performance curve for the netflix application in fig . [
fig : ec2cost ] .
the curve shows the cost one should expect to pay to obtain a certain desired performance level . to ensure interpretability of the curve , the cost assumes fine - grained billing even though amazon ec2 billing rounds up utilization time to the nearest hour .
the curve has an `` * *
l * * '' shape implying diminishing returns : as lower runtimes are desired , the cost of attaining those runtimes increases faster than linearly . as a comparison
, we also provide the price - performance curve for hadoop on the same application .
it is evident that for the netflix application , graphlab is about two orders of magnitude more cost - effective than hadoop . fig . [ fig : ec2accuracy ] is an interesting plot which the cost required to attain a certain degree of accuracy ( lower rmse is better ) on the netflix task using 32 hpc nodes .
similarly the curve demonstrates diminishing returns : the cost of achieving lower test errors increase quickly .
the lower bound of all four curves inform the reader with the cheapest " value of @xmath70 which attains the desired accuracy .
[ sec : relatedwork ] perhaps the closest approach to our abstraction is * pregel * @xcite , which also computes on a user - defined data graph .
the most important difference is that pregel is based on the bsp model @xcite , while we propose an asynchronous model of computation .
parallel bgl @xcite is similar .
* piccolo * @xcite shares many similarities to graphlab on an implementation level , but does not explicitly model data dependencies . sequential consistency of execution is also not guaranteed .
* mapreduce * @xcite and * dryad * @xcite are popular distributed data - flow frameworks which are used extensively in data mining .
the use of mapreduce for ml in multi - core setting was first made popular by chu et .
@xcite in 2006 .
such data - flow models can not express efficiently sparse dependencies or iterative local computation .
there are several extensions to these frameworks , such as * mapreduce online * @xcite , * twister * @xcite , * nexus * @xcite and * spark * @xcite , but none present a model which supports sparse dependencies with asynchronous local computation .
most notably , * surfer * @xcite extends mapreduce with a special primitive _ propagation _ for edge - oriented tasks in graph processing , but this primitive is still insufficient for asynchronous local computation . recently ,
work by * pearce * et .
al @xcite proposed a system for asynchronous multithreaded graph traversals , including support for prioritized ordering .
however , their work does does not address sequential consistency or the distributed setting .
finally , * optiml * @xcite , a parallel programming language , for machine learning .
we share their approach of developing domain specific parallel solutions .
optiml parallelizes operations on linear algebra data structures , while graphlab defines a higher level model of parallel computation .
[ sec : conclusion ] many important ml techniques utilize sparse computational dependencies , are iterative , and benefit from asynchronous computation .
furthermore , sequential consistency is an important requirement which ensures statistical correctness and guarantees convergence for many ml algorithms .
finally , prioritized ordering of computation can greatly accelerate performance .
the graphlab abstraction we proposed allows the user to explicitly represent structured dependent computation and extracts the available parallelism without sacrificing sequential consistency .
furthermore , graphlab s sync operation allows global information to be efficiently aggregated even as an asynchronous iterative algorithm proceeds .
since the graph representation of computation is a natural fit for many ml problems , graphlab simplifies the design , implementation , and debugging of ml algorithms .
we developed a highly optimized c++ distributed implementation of graphlab and evaluated it on three state - of - the - art ml algorithms using real data : collaborative filtering on the netflix dataset , named entity recognition , and video cosegmentation .
the evaluation was performed on amazon ec2 using up to 512 processors in 64 hpc nodes .
we demonstrated that graphlab outperforms hadoop ( a popular framework in the ml community ) by 20 - 60x , and is competitive with tailored mpi implementations .
future work includes supporting dynamic and implicitly represented graphs , as well as support for graphs in external storage .
the current implementation provides limited support for external storage through the use of * mmap*ed memory for vertex and edge data .
there are interesting possibilities for the intelligent placement and caching of graph data to maximize performance @xcite .
while the current graphlab implementation does not provide fault tolerance , relatively simple modifications could be made to support snapshotting capabilities .
in particular , a globally consistent snapshot mechanism can be easily performed using the sync operation .
additionally , we plan to extend graphlab to other architectures including gpus and supercomputers . | machine learning ( ml ) techniques are indispensable in a wide range of fields .
unfortunately , the exponential increase of dataset sizes are rapidly extending the runtime of sequential algorithms and threatening to slow future progress in ml . with the promise of affordable large - scale parallel computing ,
cloud systems offer a viable platform to resolve the computational challenges in ml .
however , designing and implementing _ efficient _ , _ provably correct _ distributed ml algorithms is often prohibitively challenging .
to enable ml researchers to easily and efficiently use parallel systems , we introduced the graphlab abstraction which is designed to represent the computational patterns in ml algorithms while permitting efficient parallel and distributed implementations . in this paper
we provide a formal description of the graphlab parallel abstraction and present an efficient distributed implementation .
we conduct a comprehensive evaluation of graphlab on three state - of - the - art ml algorithms using real large - scale data and a 64 node ec2 cluster of 512 processors . we find that graphlab achieves orders of magnitude performance gains over hadoop while performing comparably or superior to hand - tuned mpi implementations . |
understanding the dynamics of chemical kinetics is a question of fundamental character in nonequilibrium statistical mechanics , apart from being of broad importance in applications to other sciences . indeed , many models in chemistry @xcite , biochemistry @xcite , ecology @xcite , and biology @xcite use stoichiometric relations as theoretical first principles describing some phenomenon . the simplest description of the time evolution of a set of @xmath0 reacting species is probably given by the mean - field equations , an @xmath1dimensional dynamical system representing the concentrations , or total number of molecules of the reacting species .
one of the advantages of this approach is that it allows us to use the powerful machinery of dynamical systems theory @xcite , raising the possibility of identifying stationary states with fixed points , periodic behavior with limit cycles , etc . of course
, as with every theory in physics , the mean - field approximation has a range of validity .
as it ignores fluctuations , its description of the chemical system might be accurate for short times ; however , long - time dynamics will be affected , dramatically in some cases , by rare events .
it is possible to study exactly the system evolution as a time - continuous markov process , the probability distribution of which is given by the solution of an adequate master equation @xcite .
the full analytical solution of the master equation is , in most situations , a formidable problem , and it yields , usually , much more information than that needed in applications .
the development of approximate theories is thus clearly justified .
one of the most common approximations is the fokker - planck equation , obtained from the master equation by means of a kramers - moyal or van kampen size expansion @xcite .
this theory assumes that the number of reagents is very large , so we can consider the implicit stochasticity of the process as small gaussian fluctuations around the mean - field behavior .
definitely , rare events do not belong to this category . if one wants to deal with fluctuations that are comparable with , or even greater than , the mean value , any approximation scheme must not rely on assumptions such as an asymptotically large value of the number of reagents .
it is possible to construct such a theory , for instance , by taking advantage of the relation among chemical kinetics and quantum mechanics @xcite .
once the master equation is formulated as a quantum problem , it is possible to develop eikonal @xcite or wkb approximations @xcite , or matched asymptotic expansions of the spectral formulation @xcite , able to tackle rare events . in this work
we will concentrate on the eikonal approximation introduced in @xcite .
one of the advantages of this approach is that it reduces the problem again to a dynamical system of @xmath2 dimensions for @xmath0 reacting species .
the additional @xmath0 degrees of freedom are the conjugate `` momenta '' corresponding to each of the concentrations and represent a measure of the size of the fluctuations . due to the hamiltonian symmetry of this system , it can be effectively reduced to a @xmath3dynamical system on some riemannian manifold .
so , if there is only one chemical species , the case in which we will concentrate here , we have a reduced number of dynamical scenarios . in particular , only fixed points will be of physical relevance , and they will denote the possible stationary states in which the system can be found .
contrary to the mean - field situation , rare events can drive the system for one stationary state to another , a fact that is reflected by the existence of connections between the fixed points of the dynamical system .
these connections are not the unique , but the optimal way in which a system evolves from one state to the other @xcite , and , as they are reminiscent of instantons in quantum mechanics @xcite , we will denote them as instanton connections . its importance is huge : the web of connections , having the fixed points at their intersections , encodes the qualitative behavior of the chemical system @xcite , and its topology serves as a principle for the classification of nonequilibrium phase transitions in reaction - diffusion models @xcite .
all these works have been focused on the dynamics of chemical kinetics happening at constant rates . while this assumption is reasonable in many cases ,
there are situations in which we should go beyond it and consider the explicit time variation of the reaction rates . periodically illuminated chemical reactions @xcite or seasonal variation in population dynamics @xcite serve as examples of nontrivial behavior generated by a temporal forcing .
not much attention seems to have been paid to eikonal approximations of time - dependent chemical kinetics , and when they are considered , nonautonomous perturbations are usually treated within the quasistationary approximation @xcite .
it is our goal to extend the existent approaches and consider arbitrary frequency , albeit small , perturbations .
the concrete problem under study is how the phase portrait of the eikonal dynamical system is modified when time - dependent perturbations enter into play : do the instanton connections survive or do they disappear changing the qualitative behavior of the system ? giving a totally general answer to this question is a difficult task , but we will see that these connections seem to be very robust and persistent when they are subject to a small periodic forcing .
this will be shown with the help of particular models , for which rigorous results ( for the eikonal approximation ) are easily proven , and then we will extend them to the general setting when possible .
for the shake of clarity , we will illustrate the problem of instanton persistence with a particular reaction , branching and annihilation of identical particles , but we will state general results for arbitrary reaction sets @xcite .
let us start considering a single species of identical particles @xmath4 , which annihilate in pairs and undergo binary branching @xmath5 the master equation describing the probability distribution of having @xmath6 reagents at time @xmath7 reads @xmath8+\frac{\lambda(t)}{2 } [ ( n+2)(n+1)p_{n+2}(t)- n(n-1)p_n(t)].\ ] ] it can be represented by a partial differential equation ( pde ) by introducing the generating function @xmath9 which provides us with the time - dependent hamiltonian @xmath10 and the imaginary time schrdinger equation @xmath11 the eikonal approximation proposes the reduction of the problem to hamilton equations @xcite , which in this case become the nonautonomous dynamical system ( note , however , that we could reduce the problem to one time - dependent reaction rate by means of the time reparametrization @xmath12 ) @xmath13 suppose for a moment that both @xmath14 and @xmath15 are time independent .
in this case the system exhibits three lines with zero energy : the invariant lines @xmath16 , @xmath17 , and @xmath18 furthermore , we know that these three lines connect the fixed points @xmath19 , @xmath20 , and @xmath21 in the @xmath22 plane , see fig.([fase1 ] ) ( or fig.(3 ) in @xcite ) . now we can try to understand what happens when we let the reaction rates vary in time .
it is easy to see that the lines @xmath16 and @xmath17 remain invariant zero - energy lines of the system ; however , the explicit time dependence of the hamiltonian prevents the conservation of the energy @xmath23 and , in general , forces the disappearance of the third zero - energy line .
we can generalize this fact for an arbitrary reaction hamiltonian .
given any set of reactions with time - dependent rates , we know that the hamiltonian necessarily fulfills @xmath24 due to the conservation of probability @xcite .
so this means that _ for any set of ( time - dependent ) reaction rules , the @xmath16 line is an invariant , zero - energy line of the dynamical system_. since this line describes the mean - field dynamics of the system , we will call it the _ mean - field line _ @xcite . some systems
possess an absorbing state when they contain zero particles ; this happens if all reactions of the type @xmath25 are absent in the dynamics . in this case
, the hamiltonian must obey the condition @xcite @xmath26 so we can claim that _ any ( nonautonomous ) system without particle production from the vacuum has the invariant , zero - energy line @xmath17_. and thus , when it is present , we will call it _ the absorbing - state line_. these properties can be clearly seen from the general form of the hamiltonian term representing the reaction @xmath27 it is @xcite @xmath28 in the autonomous situation , the set of zero - energy lines determines the physics of the problem : the fixed points obtained when these lines cross represent the possible states in which the system can be found and the connections among them the possible transition paths . together with the mean - field line (
the global minimum of the action ) and the absorbing - state line one finds other lines with zero energy : _ the instanton lines _
we know that both the mean - field line and the absorbing - state line persist if we let the reaction rates vary in time , but however , it is not so obvious to see what happens with the instanton lines . these are defined in terms of zero energy if the system is not explicitly dependent on time , but when this is not the case , the definition loses its meaning since energy is no longer conserved . due to this fact and because the physical role of the instanton lines is to be optimal paths between different states , what we would like to know at this point is if both fixed points and connections among them survive after the nonautonomous forcing is switched on
. we can be sure that hyperbolic fixed points persist to a small periodic time - dependent forcing , after possible relocation of their position , but do the instanton connections persist ? to address this question
we will use the method developed by melnikov @xcite ( see also @xcite ) . consider an autonomous two - dimensional dynamical system @xmath29 @xmath30 , @xmath31 , with two hyperbolic fixed points @xmath32 and @xmath33 linked by a heteroclinic connection , which is parametrized by the system solution @xmath34 for initial time @xmath35 .
assuming that the motion goes from @xmath32 to @xmath33 , this connection is formed by a branch of the unstable manifold of @xmath32 , which totally overlaps with a branch of the stable manifold of @xmath33 .
let us now consider the perturbed version of this problem , @xmath36 where the perturbation @xmath37 is time periodic with period @xmath38 , amplitude @xmath39 small enough and sufficiently regular .
the dynamics of this nonautonomous system is given by the associated poincar map @xmath40 , which maps every initial condition point @xmath41 with the corresponding value of the solution @xmath42 after one period has elapsed @xcite .
the hyperbolic fixed points of the unperturbed system , @xmath32 and @xmath33 , are hyperbolic fixed points of the poincar map @xmath43 . since the map @xmath40 is a perturbation of @xmath43 , these points have a continuation , @xmath44 and @xmath45 , as hyperbolic fixed points of @xmath40 , and their invariant manifolds vary continuously with respect to @xmath39 .
the perturbed system will not , in general , maintain the coincidence between the branches of the unstable and stable manifolds of @xmath32 and @xmath33 , respectively : now these branches might intersect , preserving the existence of the heteroclinic connection ( see fig.([fase2 ] ) ) or might not , destroying it , like in figs.([fase3 ] ) and ( [ fase4 ] ) .
the distance between the stable and unstable branches is given by @xmath46 , with @xmath47 where @xmath48 denotes the wedge product of vectors @xmath49 and @xmath50 .
equation ( [ melnikov ] ) defines the so - called melnikov function , which yields the first - order approximation in @xmath39 of the distance between the stable and unstable manifolds measured along a direction that is perpendicular to the unperturbed connection at the point @xmath51 .
a change of sign of @xmath52 means that there exists some @xmath35 such that @xmath53 , implying the existence of a solution @xmath54 of eq .
( [ perturbed ] ) defining a heteroclinic connection among the two hyperbolic fixed points @xmath44 and @xmath45 of the poincar map corresponding to eq .
( [ perturbed ] ) , say , @xmath55 let us now consider the branching and annihilating system , eqs .
( [ brann1 ] ) and ( [ brann2 ] ) , with constant reaction rates .
the instanton connection ( the separatrix linking @xmath21 to @xmath19 ) is parametrized by the system solution @xmath56 if we assume that the perturbation of the reaction rates is given by @xmath57 and the new rates are @xmath58periodic and regular enough , we obtain the system @xmath59 , \\ \dot{q } & = & \sigma(2p-1)q - \lambda pq^2 + \epsilon \left [ \sigma_1(t)(2p-1)q - \lambda_1(t ) pq^2 \right ] , \label{q1}\end{aligned}\ ] ] which is of type ( [ perturbed ] ) . the associated poincar map @xmath40 still has the origin as hyperbolic fixed point . another hyperbolic fixed point of this map , corresponding to @xmath21 when @xmath60 , is @xmath61 , where @xmath62 is obtained through the solution of eq .
( [ q1 ] ) for @xmath16 .
this is nothing but a bernoulli differential equation , which can be straightforwardly integrated to get @xmath63}{1+q(0)\int_0^t \exp \left[\int_0^s \sigma(\tau)d\tau \right ] \lambda(s ) ds}.\ ] ] we just need to apply the condition @xmath64 to this solution to find @xmath65 - 1}{\int_0^t \exp \left [ \int_0^s \sigma(\tau)d\tau \right]\lambda(s)ds}.\ ] ] to determine if the unstable manifold of @xmath61 intersects the stable manifold of @xmath19 we substitute the corresponding values of @xmath49 and @xmath50 into the melnikov function ( [ melnikov ] ) , @xmath66 \left [ ( 1-p)(1-p - p^2)q^2 \right]({\bf x}_h(t - t_0))dt= \\ \int_{-\infty}^\infty h(t)\phi(t - t_0)dt=\int_{-\infty}^\infty h(s/\sigma + t_0)\tilde{\phi}(s)ds,\end{aligned}\ ] ] after the change of variables @xmath67 and where @xmath68({\bf x}_h(t ) ) , \\
\tilde{\phi}(s ) & = & -\frac{4 \sigma^2}{\lambda^2}\frac{e^{2s}[1 + 2 \mathrm{sh}(s)]}{(1+e^s)^3(2+e^s)^2},\end{aligned}\ ] ] the symbol @xmath69 standing for the hyperbolic sine .
it is easy to see that @xmath70 has zero mean , @xmath71}{(1+e^s)^3(2+e^s)^2 } ds= \frac{2 \sigma^2}{\lambda^2}\left [ \frac{e^s}{(1+e^s)^2(2+e^s ) } \right]_{-\infty}^\infty=0,\ ] ] a fact that will prove its usefulness later on . since @xmath72 is @xmath38 periodic and continuous , we can expand it in fourier series , @xmath73 and substitute it into the melnikov function to obtain @xmath74}{(1+e^s)^3(2+e^s)^2 } e^{2 \pi
i n s/(\sigma t ) } ds,\ ] ] which is the fourier decomposition of the melnikov function .
we can check the validity of the fourier series ( [ fourier ] ) after contrasting that the norm @xmath75}{(1+e^s)^3(2+e^s)^2 } \right| ds = \frac{5\sqrt{5}-11}{2}\ ] ] is finite .
this representation allows us to calculate the mean value @xmath76 where we have used @xmath77 and @xmath78 denotes the kronecker delta .
so we know that the melnikov function is periodic , continuous ( from its very definition in eq .
( [ melnikov ] ) it only depends on @xmath49 , @xmath50 , and @xmath79 ) , and with zero mean , implying that it either crosses zero or vanish identically . in the first case , we can claim that the point @xmath61 is connected to the origin , in the second , we can not , due to the possible presence of small terms @xmath80 .
however , we can see that the second case only happens when @xmath72 is constant .
indeed , integrating by parts and changing variables @xmath81 we obtain @xmath82}{(1+e^s)^3(2+e^s)^2 } e^{2 \pi i n s/(\sigma t ) } ds= \frac{i \pi n}{\sigma t } \int_{-\infty}^{\infty } \frac{e^s}{(1+e^s)^2(2+e^s)}e^{2 \pi i n s/(\sigma t)}ds= \\ \frac{i \pi n}{\sigma t } \int_{0}^{\infty } \frac{x^{2 \pi i n/(\sigma t)}}{(1+x)^2(2+x)}dx.\end{aligned}\ ] ] this last integral can be obtained by means of contour integration in the complex plane .
noticing that for a complex variable @xmath83 @xmath84 we can use the keyhole contour choosing the logarithm branch cut as the positive real axis , and by employing the residue theorem we obtain @xmath85+\mathrm{res}[f(z),-2 ] \},\ ] ] where @xmath86 we can now compute the residues @xmath87 & = & \exp \left [ -\frac{2 \pi^2 n}{\sigma t}+\frac{2 \pi i n}{\sigma t}\ln(2 ) \right ] , \\ \mathrm{res}[f(z),-1 ] & = & \left .
\frac{d}{dz}\frac{z^{2 \pi i n/(\sigma
t)}}{(2+z ) } \right|_{-1}= -\left ( 1+\frac{2 \pi i n}{\sigma t } \right ) \exp \left ( -\frac{2 \pi^2 n}{\sigma t } \right),\end{aligned}\ ] ] to conclude @xmath82}{(1+e^s)^3(2+e^s)^2 } e^{2 \pi
i n s/(\sigma t ) } ds= \\
\frac{2 \pi^2 n/(\sigma t)}{1-\exp[-4 \pi^2 n/(\sigma t)]}\exp \left ( -\frac{2 \pi^2 n}{\sigma t } \right ) \left ( 1+\frac{2 \pi i n}{\sigma t } -\exp \left [ \frac{2 \pi i n}{\sigma t}\ln(2 ) \right ] \right),\end{aligned}\ ] ] and one can see that it only vanishes if @xmath88 or @xmath89 , for @xmath14 and @xmath38 positive finite real numbers . hence , the melnikov function is identically zero if and only if @xmath90 for all @xmath91 in eq .
( [ phifourier ] ) or , what is the same , if and only if @xmath72 is constant .
we still have to analyze what is the meaning of the condition @xmath92 for some constant @xmath93 .
if @xmath94 , then the system can be reduced to the unperturbed one by means of a reparametrization of the time variable : @xmath95d\tau.\ ] ] in this case the geometry of the phase space is exactly preserved ; only the parametrization of the system solution along the separatrices might change . if @xmath96 and @xmath97 are chosen constant , then for any value of @xmath93
, the phase - space topology is the same , as a small retuning of the hamiltonian parameters can not modify it .
in these two particular cases we know that the melnikov function is identically zero because the connection is still given by the complete superposition of a branch of the stable manifold of one of the fixed points and a branch of the unstable manifold of the other . in the general case we know that a perturbation fulfilling @xmath98 is a combination of these two , translation and reparameterization , @xmath99 , \\
\lambda(t ) & = & ( \lambda + \epsilon \lambda_1)[1+\epsilon \phi(t ) ] , \label{rep2}\end{aligned}\ ] ] for some function @xmath100 , up to terms @xmath80 .
this means that , in order to know if the distance between the invariant manifolds of the fixed points is identically zero , we would have to use the extension of the melnikov method to higher orders @xcite
. it would be highly desirable to extend two properties of this example to general systems , say , the possibility of expressing the melnikov function as a fourier series , and to show that it has zero mean .
we will start with the second claim assuming that the first one is true , and we will check its validity afterwards .
the most general reaction hamiltonian can be built adding the generic terms ( [ genreact ] ) , @xmath101 and it allows us to express the dynamical system as @xmath102 where all the rates @xmath103 are supposed to have the same period @xmath38 ( the case of perturbations with different periods will be discussed below ) .
assuming the form of the perturbation @xmath104 allows us to split the hamiltonian into two terms @xmath105 .
so we can write the integrand of the melnikov function as the set of poisson brackets @xmath106 and obtain @xmath107 where we have used the fact that instanton lines link fixed points lying on zero - energy invariant lines .
since @xmath108 we can integrate by parts to get @xmath109 this allows us to rewrite , at least formally , the melnikov function as a fourier series , @xmath110 after the change of the integration variable @xmath111 .
its mean value is thus given by @xmath112 in order to check the validity of fourier series ( [ melfou ] ) we need to bound the norm @xmath113 this is not a difficult task because both @xmath114 and @xmath115 must be continuously differentiable along the connection and with finite limits when @xmath116 , which must coincide with their values at the fixed points . furthermore , the derivatives @xmath117 and @xmath118 decrease to zero exponentially at infinity due to the hyperbolic character of the fixed points .
this implies that the norm ( [ normfou ] ) is necessarily finite .
it is actually simpler to show that the melnikov function has zero mean .
equation ( [ mj ] ) can be cast into the form @xmath119 which , together with the periodicity of the functions @xmath120 , directly provides the desired result .
however , the fourier series setting favors the identification of the conditions under which the melnikov function vanishes , as eq .
( [ condition ] ) in the last section , which can not be straightforwardly extracted from this expression .
so we believe that the fourier series representation will be more helpful in applications to concrete reaction schemes . in conclusion
, we have shown that the melnikov function of a perturbed instanton connection can always be expanded in fourier series , provided that it links hyperbolic fixed points .
furthermore , it is always a zero mean function , which implies that it either crosses zero or vanish identically .
so for any reaction hamiltonian we have the following result : given an instanton connection linking two hyperbolic fixed points , any small time - periodic perturbation of the rates will preserve the existence of the connection if the corresponding melnikov function is not identically zero . in the other case , we would have to study the behavior of the melnikov distance at higher orders , in order to be able to give a rigorous conclusion @xcite .
to finish , let us note the difficulties in extending this result and proving a general theorem about the persistence of the instanton connections even when the melnikov function is identically zero . in this case , we could recall the expansion of the melnikov distance in terms of higher - order melnikov functions @xcite .
one would be tempted to use an equivalent argument to that of the present section to try to show that an arbitrary - order melnikov function either has zero mean or vanish identically .
this would imply in turn that either some function in the expansion is nonzero , and thus the connection is preserved by means of a transversal crossing of the stable and unstable manifolds , or the whole series becomes identically zero .
while this can suggest , at first sight , that the melnikov distance is also identically zero in this case , we can not rely on a rigorous argument to prove so , since the perturbative expansion is not analytic in the small parameter in general situations .
furthermore , we can always find a family of transformations ( for instance , by continuing the chain of reparametrizations in eqs .
( [ rep1 ] ) and ( [ rep2 ] ) ) able to nullify an arbitrary order melnikov function .
so a full proof would need to handle these special perturbations , and this is complicated by the lack of analyticity of the series expansion , despite the probable fact that they are built by combinations of reparametrizations , constant shifts of the parameter values , or other trivial transformations preserving the phase - space topology . also , proving that the higher - order melnikov functions have zero mean is not a trivial fact .
these imply nonlinear combinations of the external perturbations and the corresponding mixture of fourier modes that complicates the development of a simple form like eq .
( [ melfou ] ) in these cases .
we illustrate this situation with the second - order melnikov function in the appendix .
anyway , perturbations requiring a higher - order melnikov analysis do not constitute the generic case , and we believe that the majority of the physical situations could be analyzed within the first - order , or at most second - order , formalism .
so far we have concentrated in showing that the instanton connections persist when the chemical system is temporally forced by a weak perturbation .
we devote this section to explain the physical consequences of this fact .
as we already argued , the topology of the phase space describes the physics of the system , so if its topology is preserved when the system is forced , this means that the qualitative behavior of the system is preserved too .
the disappearance of an instanton connection would mean that this qualitative behavior is modified , but how strongly ? these connections are optimal paths linking different physical states ( given by the fixed points of the dynamical system ) , but the absence of one connection communicating two states does not mean that the system can not go from one to the other .
it only means that _ there is not an optimal way of going_. we will show the importance of this fact using a toy model of biological relevance in plankton modeling .
suppose we have the reactions @xmath121 occurring at the same constant rate @xmath122 .
this set of equations has been used to model plankton patchiness in some occasions @xcite .
we can straightforwardly write the schrdinger equation for the generating function @xcite , @xmath123 so the reaction hamiltonian reads @xmath124 the corresponding dynamical system is highly degenerate , with two invariant lines @xmath16 and @xmath17 , with all their points being fixed .
this means that if we start with an initial condition @xmath125 , for some @xmath126 , then this will be the solution for all times .
as is totally clear , there are no instanton lines linking the mean - field line with the absorbing - state line .
let us now concentrate directly in the exact pde ( [ plankton ] ) instead of the approximate dynamical system .
this is a linear , first - order pde which can be easily solved along characteristics @xcite @xmath127,\ ] ] and due to normalization , we have the long - time asymptotics @xmath128 = g_0(1 ) = 1,\ ] ] which corresponds exactly to the probability distribution @xmath129 .
this means that in the infinite - time limit , regardless of the initial condition , the system will be in the absorbing state with zero particles .
so , as we have seen , _ the absence of instanton connections does not stop the system from undergoing an extinction transition_. one can figure out how this happens representing the process ( [ plnktnrel ] ) with the it stochastic differential equation @xcite @xmath130 where @xmath131 denotes a wiener process .
this equation describes brownian motion with state - dependent diffusion and tells us that the transition to extinction is very different in this case : no optimal trajectory is chosen ; instead , the state with zero particles is reached after performing a random walk from the initial condition .
this will not necessarily happen in every situation ; actually , state - dependent brownian motion is one of the simplest possibilities .
more complicated hamiltonians with powers of the momentum above the second exhibit properties corresponding to non - gaussian statistics @xcite . in general , the random walk will be more complex than simple brownian motion , but all the cases will have in common the absence of an optimal way of going from one physical state to another .
the instanton connections are a useful tool that allows us to calculate the frequency of rare fluctuations that drive the system among different states . the mean transition time is obtained , at exponential order , computing the action along the instanton connection @xcite . if our perturbation is pure time reparametrization , as the second perturbation in eqs .
( [ rep1 ] ) and ( [ rep2 ] ) , and the function @xmath100 periodic , a straightforward computation reveals that the action along the heteroclinic orbit is exactly the same as for the unperturbed system .
more general perturbations are not tractable analytically , and we would have to rely on a numerical treatment of the problem .
the heteroclinic connection can be obtained , for instance , by means of a shooting method and the action computed as a numerical integral on it . in any case
, we expect that small , well - behaved , periodic perturbations will yield transition times of the same order of magnitude as the unperturbed system .
instanton persistence is important as it allows approximating this sort of nonautonomous systems by their autonomous counterparts in the first instance and serves as a justification of the quasistationary approximation for slow signals .
perturbations that are not periodic might have a stronger effect on the system , and in particular , singular enough perturbations could change the phase - space topology ; however , it is rather difficult to conceive physical situations that give rise to such a perturbation . also , the mathematical framework changes considerably , since the absence of a poincar map does not allow the definition of the instanton as a connection among fixed points of this map , like in the periodic case . in any case , even if the instanton connections were broken , but the physical states did not drift apart form each other , transitions will be possible if we wait long enough , due to the uncorrelated fluctuations the system is subject to @xcite , but however , there will be no optimal paths linking those states .
in this work we have studied the persistence of instanton connections in chemical systems , when we promote the reaction rates from constants to functions of time .
a set of chemical reactions can be mapped , under the eikonal approximation , onto a dynamical system , the fixed points of which denote the possible states where the chemical system can be found .
connections among fixed points denote optimal transition paths communicating different states .
the persistence of these instanton lines indicates the robustsness of the physical processes that are taking place in the reactor and shows us that the qualitative behavior of the system will be similar when subject to small nonautonomous perturbations .
we have shown our results with one particular model , and we have stated them in a more general setting when it has been possible . the main theoretical tool that we have employed is the melnikov function , which has proven itself very useful in dealing with this type of problems .
one of the directions in which we would like to extend the theory is the problematic of connections linking nonhyperbolic , or one nonhyperbolic and one hyperbolic , fixed points .
the physical motivation comes from the recent classification of phase transitions in reaction - diffusion models , which identifies critical points of the physical system with the nonhyperbolic fixed points at a bifurcation threshold of the dynamical system @xcite . while it is very difficult to see what happens in the general situation , we know that the mean value of the melnikov function is zero if the mean value of all the external perturbations is zero , as is shown by eq .
( [ mvalue ] ) , provided it is well defined in this case .
this suggests to us that in a broad number of situations the topology of the phase space will remain unchanged if the mean value of all the external perturbations vanishes , even when some of its fixed points were nonhyperbolic .
it is , however , very difficult to prove a result in that direction , since perturbating a system in a critical state could have unpredictable consequences .
furthermore , the existence of the fourier series expansion of the melnikov function is no longer guaranteed , as the loss of hyperbolicity implies that we do not have an exponential decay at infinity of the absolute value in eq .
( [ normfou ] ) .
if some of the perturbations have a nonzero mean , then the problem becomes much more difficult .
the system will be moved out of criticality at some of its points , and its behavior will be more similar to that of the ( partially ) noncritical phase corresponding to the autonomous system with the parameters retuned according to the mean value of the external perturbations .
unfortunately , we can not map this situation to that described in sec .
[ ip ] , because in this case we have no control on the amplitude of the perturbation , which may be comparable to the distance to some of the critical points . so the problem becomes genuinely nonperturbational , and we can no longer employ a technique like the melnikov function . in this case
it is difficult to deal even with very slow perturbations , since this type of settings favors the appearance of soft modes , which rules out the possibility of treating the external signals adiabatically @xcite .
another problem we would like to deal with is the case of several perturbations with different periods .
the simplest case is that in which the ratio of the periods of all possible different pairs of perturbations that are affecting the system is a rational number . in this case
, we say that the perturbations are commensurable , and we can find a period which is common to all of them .
once obtained , we have reduced our problem to the one studied in sec [ ip ] .
the case of incommensurable perturbations is , of course , more complex , as it implies that the dynamics of the perturbed system can not be reduced to the dynamics of a periodic poincar map .
the fixed points of the unperturbed system give rise to quasiperiodic solutions of the perturbed problem , with their invariant manifolds being quasiperiodic as well @xcite .
there are many other questions that remain to be answered .
one is the possible appearance of chaotic behavior induced by internal stochasticity .
as noted in @xcite , for two reacting species we have a four - dimensional eikonal hamiltonian , which will give rise , in general , to a three - dimensional dynamical system on some riemannian manifold . in this case , chaos , unlike in the two - dimensional mean - field dynamical system , is indeed possible .
but chaos is also possible in the case of one reacting species obeying reaction rules with time - dependent rates . in this situation ,
`` energy '' is no longer conserved and the nonautonomous dynamical system , without integrals of motion , becomes effectively three dimensional .
indeed , situations like the one plotted in fig . (
[ fase2 ] ) , that is , when the heteroclinic connection is preserved due to the intersection of the stable and unstable manifolds , give rise to complex dynamical scenarios .
these include the appearance of smale horseshoes and even the presence of strange attractors related to the creation and destruction of such horseshoes @xcite .
although the generation of chaotic behavior of this type can be attractive from a nonlinear dynamics point of view , we are not aware , at this point , of its possible physical meaning . of course , studying multispecies reactions , both autonomous and nonautonomous , is a very interesting problem that deserves further efforts .
the nonautonomous situation implies the technical difficulty of extending the melnikov function theory to a higher dimensionality .
the ideas developed for three - dimensional systems @xcite could perhaps be adapted for studying four- or higher - dimensional problems and to try to obtain in this way the results that we would need to understand the more general multispecies reactions . also , as we have already pointed out , the melnikov function is a perturbative result .
it allows us to treat arbitrary frequency but necessarily small perturbations .
to fully understand the dynamics of nonautonomous chemical reactions , even in the one species case , we would need a nonperturbational result . with it
we could try to address questions such as the forcing of systems near criticality or the appearance of soft modes .
as we can see , chemical kinetics presents a very complex phenomenology , and it is challenging from a mathematical point of view .
a deep understanding of the physics of these nonequilibrium systems will presumably imply the parallel development of powerful methodological techniques .
the authors gratefully acknowledge input from ernest fontich and carles sim .
this work has been partially supported by the ministerio de educacin y ciencia ( spain ) through projects nos .
ex2005 - 0976 , fis2005 - 01729 , and mtm2005 - 02094 .
consider the two - dimensional dynamical system @xmath132 where @xmath30 , @xmath31 , @xmath133 , and @xmath134 are sufficiently regular functions .
the second - order melnikov function reads @xmath135dt , \label{m2}\end{aligned}\ ] ] where @xmath136 and @xmath137 denote the hessian and jacobian of @xmath49 and @xmath50 , respectively , @xmath138 is the solution parametrizing some given heteroclinic connection , and @xmath139 is the solution to the variational equation @xmath140 we now assume that this dynamical system is a reaction hamiltonian system and both @xmath50 and @xmath141 come from small nonautonomous perturbations of the hamiltonian parameters , just like in sec .
then , because eq . ( [ variation ] ) is linear , its solution will depend linearly on the fourier modes of the time - dependent parameters evaluated at @xmath35 .
but this solution appears quadratically in eq .
( [ m2 ] ) and also multiplying the jacobian of @xmath50 , which depends linearly , by assumption , on the same fourier modes .
this quadratic dependence complicates recasting the second - order melnikov function into a form similar to eq .
( [ melfou ] ) , which would help enormously for calculating its mean value .
s. chow , j. hale , and j. mallet - paret , j. diff .
equations * 37 * , 3 ( 1980 ) .
s. smale , bull .
* 73 * , 747 ( 1967 ) . l. mora and m. viana , acta math . * 171 * , 1 ( 1993 ) .
a. pumario and j. .
rodrguez , _ coexistence and persistence of strange attractors _ , lecture notes in mathematics , vol .
1658 ( springer - verlag , berlin , 1997 ) . | the evolution of a system of chemical reactions can be studied , in the eikonal approximation , by means of a hamiltonian dynamical system .
the fixed points of this dynamical system represent the different states in which the chemical system can be found , and the connections among them represent instantons or optimal paths linking these states .
we study the relation between the phase portrait of the hamiltonian system representing a set of chemical reactions with constant rates and the corresponding system when these rates vary in time .
we show that the topology of the phase space is robust for small time - dependent perturbations in concrete examples and state general results when possible .
this robustness allows us to apply some of the conclusions on the qualitative behavior of the autonomous system to the time - dependent situation . |
density perturbations at the epoch of recombination are imprinted onto the cosmic microwave background ( cmb ) , leaving temperature anisotropy that has now been well - studied on a wide range of angular scales . on scales of several arcminutes and smaller , corresponding to multipole moments of @xmath5 , the level of cmb anisotropy power from primordial fluctuations
is strongly suppressed by photon diffusion , and secondary sources of power , including the sunyaev - zeldovich effect ( sze ) , are expected to play a significant role ( e.g. , * ? ? ?
the sze results from the inverse compton scattering of cmb photons by the hot electron gas within clusters of galaxies @xcite .
this interaction leaves a small spectral distortion in the cmb which produces anisotropy power on scales of @xmath6 , detectable as a decrement in the cmb intensity at 30 ghz .
the amplitude of this power is extremely sensitive to the history of structure formation and , specifically , to the value of @xmath7 , the normalization of the matter power spectrum ( e.g. , * ? ? ?
evidence for anisotropy above that expected from the primary cmb on these small angular scales has been detected at 30 ghz by the cosmic background imager ( cbi ) and berkeley illinois maryland association ( bima ) experiments @xcite .
recent constraints on excess power at these scales from the arcminute cosmology bolometric array receiver ( acbar ) at 150 ghz indicate that the reported excess power is inconsistent with thermal cmb fluctuations but is consistent with sze fluctuations @xcite . taken together
, these measurements indicate a level of sze anisotropy power consistent with a value of @xmath7 somewhat greater than those preferred by other contemporary measurements of the parameter @xcite . in this paper
, we describe a new , high - sensitivity measurement of power in the cmb on scales ranging from @xmath8 made at 30 ghz with the sunyaev - zeldovich array ( sza ) .
the cmb anisotropy data presented here were obtained with the sza , an eight - element interferometer located at caltech s owens valley radio observatory near bishop , california .
the sza antennas are equipped with sensitive , wide - bandwidth receivers operating at 30 ghz and 90 ghz . for these observations we used the 30 ghz receivers , tuned to detect sky frequencies of 27@xmath935 ghz .
the receivers are based on low - noise , cryogenic high electron mobility transistor amplifiers @xcite , with characteristic receiver temperatures @xmath10 . including atmospheric and other noise contributions , the typical system temperatures were 35@xmath945 k. the 3.5 m sza antennas have a primary beam that is well described by a gaussian with a fwhm of 11 at the center of the 30 ghz band .
cross - correlations of the signals from pairs of the eight antennas ( _ visibilities _ ) are formed in a digital correlator , which processes the 8 ghz if bandwidth in 16 bands of 500 mhz , each of which is further subdivided into 17 channels of 31 mhz , allowing rejection of narrow - band interference . for the anisotropy measurements reported here , the sza antennas were arranged in a hybrid configuration to provide simultaneous sensitivity to the arcminute - scale structure of the sze signal from galaxy clusters and the finer - scale contamination from radio sources ( details of the configuration can be found in * ? ? ?
six of the eight antennas were packed closely together ( spacings of [email protected] m ) , yielding 15 baselines with typical projected lengths of 400@xmath91400@xmath11 , corresponding to arcminute angular scales ; the window function for our anisotropy measurement is determined by the baselines formed from combinations of these six antennas . by convolving the distribution of projected baseline lengths in the unflagged data with the autocorrelation of the antenna illumination pattern , we obtain the @xmath12-space window function shown in figure [ fig : win_ps ] .
this is the filter through which we observe the power spectrum ; multiplying this function by @xmath13 and integrating over @xmath12 yields our measurement , where @xmath14 is the cmb angular power spectrum .
the sensitivity - weighted mean value of the window is @xmath15 , with 68% of the area encompassed by the interval @xmath16 .
in addition to the central six antennas , two antennas were positioned @xmath1750 m from the array center .
the 13 baselines involving these antennas are sensitive to @xmath18 scales , corresponding to multipoles of @xmath19 , and can be used to identify radio sources that contaminate the short - baseline data . between 2005 november and 2007 june , we devoted 1340 hr of observations to 44 sza deep fields , comprising 2 deg@xmath20 .
these fields were selected to pass nearly overhead at the @xmath21 latitude of the telescopes , and arranged in right ascension to accommodate other large sza observing programs . within these constraints ,
the fields were placed at locations with low galactic emission in the @xmath22 100 @xmath23 m survey .
no consideration was given to the presence of radio sources to avoid any bias that might be introduced by the correlation between radio sources and galaxy clusters .
one of the 44 fields was found to contain a 700 mjy radio source ( @xmath24 ) ; unable to exclude this source from our data with sufficient precision , we are forced to omit this field from our analysis . using simulated sze sky maps we have verified that even in the pessimistic case in which this single field contains the most power of any of our fields , the bias introduced by excluding it is 10%
. observations were designed to allow subtraction of ground contamination .
fields were observed at constant declination in groups of four , spaced by four minutes of right ascension .
we observed the fields in 16 minute blocks between observations of a nearby phase calibrator .
each field , starting with the westernmost field of the group , was observed for just under 4 minutes at a time , so that after including time lost to slews and calibration we tracked all four fields through precisely the same azimuth / elevation path .
this ensures that each sees the same contribution from the ground , which can be removed by subtracting the mean of the four fields ( in practice we implement this subtraction by means of a constraint matrix , described in section [ sec : lh ] ) . at an integration time of 20
s , we obtain 10 integrations per field in each cycle .
the total integration time per field was @xmath25 hr .
of this , approximately 20% of the data were discarded due to hardware problems , antenna shadowing , excessive noise , or jumps in calibrator phase or system temperature which occurred on a time - scale faster than our calibrations .
an additional @xmath175% of the data were flagged for showing unexpected correlations among baselines and bands or statistically unlikely noise behavior on a single antenna .
these cuts significantly improved the results of the jackknife tests discussed in section [ sec : jk ] ; the data quality cuts were refined using only the jackknifed data sets to avoid biasing ourselves against signal . the first member of each group of four sza fields is listed in table [ tab : fieldtab ] , along with its position , the total , unflagged , on - source integration time , and the achieved rms noise level for the field , both with and without data from the long baselines .
lccccc cmb07 & 02:07:37.00 & + 34:00:00.0 & 22.9 & 0.18 & 0.12 + cmbcc1 & 02:11:31.30 & + 33:27:43.0 & 21.9 & 0.19 & 0.14 + cmba1 & 02:12:00.00 & + 33:00:00.0 & 19.7 & 0.21 & 0.14 + cmbi1 & 02:12:00.00 & + 32:37:08.2 & 22.8 & 0.21 & 0.14 + cmbr1 & 02:12:15.60 & + 32:11:24.8 & 32.3 & 0.18 & 0.12 + cmby1 & 02:12:00.00 & + 31:51:24.4 & 23.0 & 0.18 & 0.12 + cmbdd1 & 14:18:40.10 & + 35:01:42.0 & 18.0 & 0.26 & 0.17 + cmbee1 & 14:18:39.24 & + 35:31:52.3 & 20.9 & 0.19 & 0.14 + cmbxxb & 21:28:50.60 & + 24:59:35.0 & 18.2 & 0.21 & 0.15 + cmbaa1 & 21:24:38.70 & + 25:29:37.0 & 19.6 & 0.22 & 0.16 + cmbbb1 & 21:24:38.10 & + 25:59:24.0 & 18.8 & 0.21 & 0.15 [ tab : fieldtab ] the absolute calibration of the flux density scale was derived from bimonthly observations of mars .
we use the @xcite model to predict the brightness temperature of mars as a function of frequency and time .
since the planet is partially resolved by our longest baselines , a strong , unresolved source is used to transfer the calibration from the short baselines .
the absolute calibration was cross - checked by comparing sza observations of jupiter to those of _
wilkinson microwave anisotropy probe _
( _ wmap _ ) and cbi @xcite . based on these measurements
, we conservatively estimate that our absolute flux scale is accurate to better than 10% ( 20% in power ) .
away from the galactic plane , the most significant contributor of non - cmb power on arcminute angular scales at 30 ghz is compact radio sources .
since the power from these sources is constant as a function of @xmath12 , their contribution to @xmath26 can be quite large at small angular scales . while the brightest sources at 30 ghz can be detected by the long baselines of the sza , sources near our noise level ( and hence undetectable in our data ) can still contribute substantial power to the measurement .
we therefore supplement our data with a higher - sensitivity search for radio sources in our fields using the nrao very large array ( vla ) .
the large disparity between the vla ( 25 m ) and sza ( 3.5 m ) antenna sizes makes it impractical to survey the sza fields at a frequency near 30 ghz .
instead , we elected to conduct the vla survey at 8 ghz , a compromise between survey speed , which decreases toward higher frequency as the vla primary beam shrinks , and spectral extrapolation of the detected sources . in 2006 november , we collected 36 hr of data with the vla in its c configuration , targeting the 24 fields that the sza had observed at that point . in 2008
august , an additional set of observations covering all but two of the remaining fields were obtained in the more compact d configuration , which better matches the angular resolution of the sza long baselines . because the vla primary beam at 8 ghz is only 4.9 across
, it was necessary to make mosaic observations with 19 pointings to cover the area of the sza s larger primary beam .
we reached a noise level of @xmath27jy at the center of this mosaic pattern , less than 1/6 of the noise level in our sza data .
the integration time is tapered at pointings away from the center of the mosaic , such that the achieved sensitivity in the 8 ghz maps has a profile that matches that of the 30 ghz data , set by the sza primary beam pattern .
the vla data were calibrated and mosaicked in aips .
the map sensitivity @xmath28 was determined from the mosaic images using the task rmsd .
sources were then extracted from the maps using the task sad , with a limiting significance of 5@xmath28 , yielding more than 180 sources .
due to weather and scheduling limitations , two of the sza fields were not observed with the vla at 8 ghz .
we omit the entire group of four fields that includes this pair from the analysis that follows , in addition to the single field that contained a bright radio source .
this leaves us with 39 of the 44 observed fields to include in our analysis .
as an interferometer , the sza directly measures the fourier components of the sky brightness .
since it is precisely the variance of these components that the power spectrum describes , interferometers are well suited for measuring the power spectrum , without the intervening stage of map making ( e.g. , * ? ? ?
we have used a maximum likelihood method to extract measurements of the power spectrum from our data .
following @xcite , we construct a vector @xmath29 of @xmath30 visibilities and a likelihood estimator that is a function of @xmath29 and the cmb power @xmath31 , assumed to be constant across the range of angular scales probed by the sza .
we expect our data , and thus the likelihood function , to be gaussian , so that the likelihood is given by : @xmath32 where n is the length of the data vector @xmath29 , and the covariance @xmath33 of the data is written as the sum of a contribution from the cmb , the diagonal instrument noise , and a set of constraints ( see below ) : @xmath34 if our visibilities were independent of one another , the contribution to the covariance matrix from the cmb signal , @xmath35 , would be the identity matrix times the level of cmb power , @xmath31 .
we assume a flat band power , i.e. , constant @xmath26 .
however , in practice , visibilities can be correlated , depending on the separation of the fourier - space ( ) coordinates they sample .
a visibility corresponding to a given coordinate is in fact an integral over a small patch of the fourier plane , weighted by the antenna s aperture autocorrelation function ; the overlap integral of this function centered on two neighboring coordinates yields the correlation between adjacent measurements .
we calculate these overlap integrals for each of our visibilities and insert the resulting matrix into the total covariance matrix as indicated in equation ( [ eqn : covar ] ) .
visibilities that are more than 95% correlated are averaged together .
the instrumental contribution to the noise , @xmath36 , dominates the weak signal from the cmb anisotropy and must therefore be accurately determined .
we estimate the noise variance from the scatter of visibilities taken on 20 s timescales within a 4 minute scan . from the 10 visibilities within each scan
, we subtract the mean of the real and imaginary parts from each sample and then compute a single sample variance from the 20 mean - subtracted numbers .
all averages within the analysis are uniformly weighted , as weighting by variances derived from just 20 samples introduces substantial biases .
the resulting noise estimates are independent of one another , so @xmath36 in equation ( [ eqn : covar ] ) is diagonal .
we have verified through simulation that variances measured in this way produce an unbiased estimate of the noise , and , hence of the cmb power level .
we expect the sky signal to be dominated by radio sources that are point - like at the sza s arcminute resolution , principally active galactic nuclei emitting synchrotron radiation at radio frequencies . through the use of constraint matrices
@xcite , we can eliminate the modes of the data that are corrupted by these sources , provided that we know their positions .
the constraint matrix is computed by writing a template vector of visibilities that describes the contribution of the contaminant in question to each component ; in the case of a point source at position @xmath37 with respect to the interferometer s phase center , the contribution @xmath38 to the visibility measured with component @xmath39 is given by : @xmath40 where @xmath41 is an unknown amplitude . by forming the outer product of this vector with itself ,
@xmath42 , we find the covariance created among our @xmath30 visibilities by a source at this position ; the amplitude of this covariance remains unknown , but its shape is fixed by the ( known ) source location .
each constraint matrix is then multiplied by a large prefactor ( as large as possible without causing the matrix to become poorly conditioned for inversion ) and added to the total covariance matrix in our likelihood function , effectively setting to zero the weight of the mode corresponding to the source .
the unknown amplitude is therefore unimportant ; the operation is equivalent to marginalizing over the source s unknown flux .
we form such a matrix for each source that we wish to eliminate from our data ; because each matrix is created from a single vector , each has rank one and eliminates 1 degree of freedom .
radio sources in our data may have a wide range of spectral indices and can not in general be excluded by a single constraint .
instead , we project 3 modes of different spectral index for each source ; we therefore lose three degrees of freedom for each projected source .
we have verified in simulation that a linear combination of these components can effectively remove the contribution of a source of any reasonable spectral index .
we remove a component common to the fields within a group of four ( ground or antenna crosstalk ) using a similar constraint .
this constraint identifies each visibility among a group of four with its three counterparts in different fields , all of which are measured at the same antenna position .
projecting this constraint is equivalent to subtracting the mean of the four visibilities from each of them , and therefore reduces the sensitivity by 25% .
the radio source positions are derived from several data sets .
the brightest sources at 30 ghz are identifiable directly in the sza data . for purposes of source detection , we combine the long and short - baseline data ( typical noise of @xmath43jy ) , providing 30% better sensitivity to compact sources than would be obtained using only the short baselines from which our anisotropy measurement is determined . above a significance of @xmath44 in the 30 ghz data , 42 of 44 detected sources have counterparts in the 8 ghz vla survey ; the remaining two sources were found to be extended even at the 20 resolution of the long sza baselines and are likely heavily resolved by the vla in its c configuration ( 3 resolution ) .
both of these sources have 1.4 ghz counterparts in the nrao vla sky survey ( nvss ; @xcite ) , which has angular resolution intermediate to that of the long and short sza baselines , giving us confidence that they are real , extended sources rather than noise peaks in our 30 ghz maps . as sources below our 30 ghz detection threshold
can still contribute significant anisotropy power , we constrain 180 sources detected at 5@xmath28 in the 8 ghz survey , as well as the two sources detected at 30 ghz that are too resolved to be seen with the vla .
the effect of using different detection thresholds in the two data sets can be seen in table [ tab : pntsrc ] .
clearly , radio sources contribute the bulk of the detected anisotropy ; even sources just above the 5@xmath28 threshold at either frequency contribute substantial power .
we also note that the two sources without 8 ghz counterparts contribute a sizable fraction of our final power ; the inclusion of these two sources is therefore important to our final result .
lccc 6@xmath28 & & @xmath45&33 + 5@xmath28 & & @xmath46&44 + & 20@xmath28 & @xmath47 & 81 + & 15@xmath28 & @xmath48 & 98 + & 8@xmath28 & @xmath49 & 141 + & 5@xmath28 & @xmath50 & 180 + 5@xmath28 & 5@xmath28 & @xmath51 & 182 for each field , we assemble the noise and constraint components of the covariance matrix and vary the level of cmb power , evaluating the likelihood of the data according to equation ( [ eqn : lh ] ) over a range of values for the power @xmath31 . treating our fields as independent samples
, we take the product of all of their likelihood curves to form a global likelihood for the experiment ; we report the maximum of this curve as the most likely power , and we use the points that correspond to 16% and 84% of the cumulative likelihood to define the 68% confidence interval .
the likelihood is allowed to extend below zero power .
we tested this analysis pipeline extensively with simulated data , including tests in which the data generation and power spectrum analysis were performed by different parties , using independent software packages , and found consistent results in all cases .
processing the data as described above , we can compute the relative likelihood of different levels of cmb power , given our data .
we now subject the data to three jackknife tests to rule out contaminating power from non - astronomical sources or from inaccuracies in our noise variance estimation . in the jackknife tests ,
the data are split into halves , each of which measures very nearly the same set of spatial fourier components .
the data are taken in such a way that three of these splits are possible : ( 1 ) a frequency jackknife , wherein we separate alternating 500 mhz frequency bands , as neighboring bands sample very nearly the same points , ( 2 ) a jackknife in which the data are separated into halves by time , and ( 3 ) a jackknife in which the data are separated into even and odd days .
the latter two jackknife tests take advantage of the fact that the sza observed the same fields in the same way for many days . after dividing the data into halves
, we difference the matching pairs of visibilities .
emission from astronomical sources , which is common to both halves , is removed , while any contamination that varies with time or frequency will remain in the differenced ( jackknifed ) data .
note that the requirement that there be matching , unflagged visibilities in both halves for each test results in some additional loss of data and degradation of the measurement sensitivity , particularly in the two time - jackknife tests .
we compute variances from the differenced visibilities , and compute the power in each jackknifed data set just as for our unjackknifed data , with the expectation that the measured power will be consistent with zero .
the jackknife tests also verify our noise model ; if the noise in our data has been misestimated , we expect that the jackknifed data will be poorly described by our computed covariance matrices .
the jackknife tests provide evidence that we detect a signal correlated with the orientation of our antennas relative to the ground .
this signal may originate from antenna crosstalk or ground emission . as discussed in
[ sec : szadata ] , the observations were designed to allow removal of such ground contamination " by subtracting the average of each group of four observations via an additional constraint .
we find that without removing this component ( but including strict cuts on data quality discussed in [ sec : szadata ] ) the frequency jackknife test fails badly , with a probability to exceed the jackknifed @xmath52 of @xmath53 . after implementing the ground subtraction , the three jackknife tests all pass , as seen in table [ tab : jk ] , with probabilities to exceed the @xmath52 greater than 0.1 .
we therefore employ ground subtraction to determine anisotropy power in all the analyses presented in this paper .
our analysis procedure is blind to the final result , as we have refined our data cuts and analysis treatment based only on the jackknife tests .
lccc frequency & @xmath54 & 0.27 + first
second half & @xmath55 & 0.56 + even odd days & @xmath56 & 0.59 + unjackknifed data & @xmath57 & 0.10
repeating the analysis on the unjackknifed data , including ground - subtraction and constraining radio sources detected at @xmath58 in both the 8 ghz vla data and the 30 ghz sza data , leads to a residual power of @xmath59 . to set constraints on secondary cmb anisotropy
, we must estimate the residual power contributions to this measurement from other astronomical sources . in this section ,
we first consider contributions from primary cmb anisotropy , undetected radio sources , and diffuse galactic emission , and we end with our constraint on the level of secondary anisotropy .
we note that contributions from undetected radio sources and diffuse galactic emission are , in principle , distinguishable from the cmb by their spectral signature across the 25% fractional bandwidth of the sza data .
however , it is not possible to distinguish a preferred spectral index given the low signal - to - noise ratio of the data . at a multipole of @xmath60 ,
the wmap5 cosmology @xcite predicts no significant primary cmb anisotropy ( figure [ fig : win_ps ] ) .
however , our window function has limited sensitivity to the larger scales at which the primary cmb signal is strong .
we estimate this contribution and its variance with simulated observations .
we generate cmb skies according to wmap5 cosmological parameters using cmbfast @xcite .
we use 35 sets of 39 fields with identical noise realizations but different cmb realizations and sample them according to the sza coverage and noise . from these simulations , we determine that the primary cmb contributes a mean power of @xmath61 to our measurement . integrating the cmb primary power spectrum multiplied by our window function yields a very similar estimate . to assess the level of residual fluctuations from sources that are undetected at 8 ghz and 30 ghz at our 5@xmath28 threshold ( see [ sec : lh ] )
, we repeat our analysis using simulated data that include a 30 ghz selected source population extending below our detection thresholds .
this population s source density as a function of flux is derived from the sza blind cluster survey @xcite , which provides source counts down to 1 mjy ; we extrapolate to lower flux levels according to a power law .
this survey is the deepest available at this frequency . to assign 8 ghz fluxes to our simulated population of sources
, we use the spectral index distribution of @xcite measured for 200 sources between 5 and 30 ghz .
we simulate sza observations of fields containing these source populations , constraining those that appear above our detection thresholds at either 8 or 30 ghz . we find that undetected sources in measurements of 39 fields contribute @xmath2 to the detected power . at lower frequencies ,
emission from the galaxy via synchrotron and free - free radiation contribute significantly to the sky brightness , as can be seen in the _ wmap _ 22 ghz all - sky map @xcite .
we lack sensitive , high - resolution measurements of these components and are forced to estimate their contribution from lower - resolution maps and forecasts derived from them .
@xcite used degree - resolution templates to predict that this emission should contribute less than @xmath62 at @xmath63 . renormalizing their model using an analysis of foregrounds measured with wmap @xcite increases this estimate to @xmath64 at 30 ghz .
an additional source of potential foreground contamination in our power spectrum measurement comes from galactic dust .
while the predictions of @xcite at 30 ghz imply @xmath65 contribution for thermal dust emission , there has been evidence that dipole radiation from small , spinning dust grains may dominate the total dust emission at 30 ghz .
such emission is expected to peak at frequencies of tens of ghz and to be tightly correlated with galactic dust @xcite .
dust - correlated emission within this frequency range has been observed on various angular scales @xcite . during the sza blind cluster survey , increased map noise
was indeed observed in fields with more obvious _
iras _ structure .
to estimate the contribution of spinning dust to the sza measurement , we calculate the power spectrum of the _ iras _
@xmath66 @xmath67 maps of our fields , fit it to @xmath68 , and extrapolate it to the smaller angular scales that we are sensitive to , but to which _ iras _ is not .
we then take the highest measured emissivities of dust - associated emission at 30 ghz , as tabulated by @xcite , and scale this power spectrum to match the observations . multiplying by the window function shown in figure [ fig : win_ps ] and integrating over @xmath12
, we get a contribution for each of the observed emissivities lying in the range from @xmath69 to @xmath70 .
note that the highest emissivity @xcite , derived from a very localized portion of the galactic plane , is substantially higher than other measurements , and so the corresponding @xmath70 is likely a substantial overestimate of the contribution of this foreground to our measurement , made in a different part of the sky with less galactic emission .
we now estimate the contribution from secondary cmb anisotropy .
we account for the expected contributions from primary cmb anisotropy and undetected radio sources .
we do not consider the contributions from sources of diffuse galactic emission , since even the conservative estimates discussed above are subdominant to the contributions from primary cmb and undetected radio sources .
subtracting the residual power for undetected radio sources and including our 20% calibration uncertainty , we are left with @xmath71 for the combined primary and secondary anisotropy result .
the constraint on the level of secondary cmb anisotropy after subtracting the expected contribution from primary cmb anisotropy is @xmath72 .
applying a prior of positive power and integrating the likelihood curve results in an upper limit to the level of secondary cmb power in our maps of @xmath73 at 95% cl ( @xmath74 at 68% ) .
the estimated anisotropy contributions and the resulting constraint on the level of secondary anisotropy are tabulated in table [ tab : foregrounds ] . in figure
[ fig : win_ps ] , we plot the sza anisotropy measurement corrected only for residual power from radio sources but including the contribution from primary cmb , along with previous measurements of small - scale cmb power at 30 ghz from cbi and bima ( * ? ? ?
* respectively ) .
lc primary cmb & @xmath75 + undetected radio sources & @xmath76 + galactic synchrotron and free - free & 2 + galactic spinning dust & @xmath77 + secondary cmb ( sze ) & @xmath72 + and upper limit at 95% cl & 155 + [ tab : foregrounds ]
to understand the implications of our measurement for the value of @xmath7 , we compare it to mock observations of simulated sze maps .
several groups have carried out large - scale cosmological structure simulations and converted three - dimensional , simulated universes into two - dimensional projections of the compton @xmath78-parameter , the frequency - independent measure of the magnitude of the sze .
we take these maps as inputs to mock sza observations , and attempt to estimate the mean level of power that would be measured by the sza , as well as the scatter in these measurements .
we use 60 maps , each @xmath79 , from the hydrodynamical simulation of @xcite , generated with @xmath80 , including cooling and feedback .
@xcite provide 360 maps from dark matter simulations with @xmath81 and gas pasted into cluster haloes after the simulation is complete ; @xcite follow a similar method , producing 100 maps with @xmath82 .
finally , @xcite use the pinocchio algorithm of @xcite to generate halo distributions and merging histories , resulting in 900 maps over a range of @xmath7 . for each set of simulated maps , we form 50 groups of 39 fields ;
we pick the maps from the available set randomly with replacement . for each map , we simulate sza observations , reproducing the coverage and noise properties of the actual data .
we process these mock observations just as we do the real data , and produce a maximum likelihood power for each set of 39 fields .
we compute the mean and scatter of these powers from the 50 independent realizations .
results for each set of simulated maps are shown in figure [ fig : sim_power ] .
the strong dependence of the detected power on the value of @xmath7 is clear , as are significant systematic differences among simulations .
the maximum likelihood power for the secondary cmb anisotropy measured by the sza is shown by a shaded horizontal bar in the figure .
the shaded bar indicates the 68% confidence interval , including the uncertainty in the absolute calibration and the subtraction of power from primary cmb anisotropy and undetected radio sources .
figure [ fig : sim_power ] shows that the range of @xmath7 values consistent with the sza anisotropy measurement is large , reflecting significant differences among the simulations .
the figure shows there is no tension between the sza anisotropy measurement with @xmath83 determined from recent measurements using other techniques , while values higher than 0.9 are disfavored .
we note that the power spectra derived from simulated @xmath78-maps have greater sample variance than maps of gaussian noise with the same rms power .
while we can measure the variance in simulation by measuring the scatter of many realizations of our experiment , in the actual data we assume gaussianity in calculating our error bars .
we are therefore underestimating the sample variance in our measurement . using the simulated @xmath78-maps , we find that the magnitude of this underestimate has a weak dependence on the value of @xmath7 , but for @xmath84 the actual confidence region of our measurement is likely 1.5 times broader than the range reported here , which assumes gaussian sample variance
. there is observational evidence that radio sources are spatially correlated with clusters ( e.g. , @xcite ; @xcite ; and references therein ) .
because the short sza baselines are sensitive to both compact radio sources and the cluster sz signal , the projection of cluster - correlated radio sources could , in principle , remove significant secondary anisotropy along with the radio source power , biasing our measurement low .
we examine the magnitude of this effect by simulating our measurement .
we simulate observations of sets of 39 compton @xmath78-maps from @xcite including appropriate instrumental noise .
for these fields we generate radio source populations according to the prescription of section [ sec : radsrcs ] and include constraints for those that would be detected in our 8 or 30 ghz data .
we introduce a pessimistic degree of correlation between clusters and radio sources by increasing the radio source density in the inner 0.5 of the simulated clusters according to the excess observed toward massive ( @xmath85 ) clusters by @xcite .
we find that the mean measured power in these maps is reduced by 50% .
this should be a significant overestimate of the effect in our observations as the signal we seek comes from clusters an order of magnitude less massive than those observed by @xcite , with proportionately fewer radio sources per halo .
although the short sza baselines are unable to discriminate between radio sources and clusters , the long baseline data are wellsuited to making this distinction and could be used to recover the power lost to projection of correlated sources .
our simulations show that the inclusion of long baselines in our analysis restores an average of 25% of the lost power .
however , for any particular realization of the data the difference between the power measured with and without the long baselines may be positive or negative , with only a slight preference for increased power for our noise level and given the @xmath86 mean power in the simulated maps .
we therefore conclude that for a single realization of our observation ( the real data ) , a comparison of the anisotropy with and without the long baselines is not a useful predictor of the fraction of power lost to source - cluster correlation .
previous measurements at 30 ghz by cbi @xcite and bima @xcite have suggested power at high @xmath12 in excess of the primary cmb anisotropy , implying a value of @xmath7 inconsistent with other contemporary measurements of the parameter if the excess power is attributed to sz signal .
the cbi experiment reported @xmath87 for their high-@xmath12 bin spanning @xmath88 , of which 80@xmath990 @xmath3 is attributed to primary cmb anisotropy .
the bima experiment reported @xmath89 for @xmath90 , at which there is no significant primary cmb contribution .
the sza data presented here do not show evidence of excess power .
however , the discrepancy between the sza result and the cbi and bima results is not highly significant after accounting for measurement uncertainties and particularly the non - gaussian sample variance discussed above . here
, we investigate whether the discrepancies may also be partially accounted for by the different prescriptions used to account for residual radio source power . for the cbi experiment , @xcite used nvss to remove sources with a limiting 1.4 ghz flux of 3.4 mjy from their 30 ghz anisotropy data .
they estimated the residual contribution from undetected sources to be @xmath91 of their highest-@xmath12 bandpower , or @xmath92 , with an uncertainty of @xmath93 .
repeating our sza analysis using the same prescription , i.e. , constraining only sources detected at @xmath94 mjy in the nvss survey , results in an increase of our measured power by @xmath95 , raising our total residual power to @xmath96 .
extrapolating this to the @xmath97 of the cbi measurement according to the respective window functions of the two experiments implies a residual power of @xmath98 for the cbi result , only slightly higher than reported and consistent within the uncertainties .
the discrepancy between the measurement presented here and that of cbi is therefore unlikely to be due to the different prescriptions for estimating residual point source power .
the bima experiment @xcite used a 4.8 ghz vla survey , with depth comparable to our 8 ghz survey , to generate source constraints for their anisotropy measurement .
based on the flattening of measured power versus vla detection threshold , they estimated that radio sources contribute negligible residual power .
we note that the flattening of the measured power with detection threshold could indicate other sources of contamination not constrained by their observational strategy , such as ground pickup .
the low significance of the bima result , especially in light of the large sample variance resulting from the small sky area covered by the measurement , implies no significant tension with the result presented here .
we present results from 30 ghz measurements of the cmb angular power spectrum with the sza , on scales where the secondary anisotropy from the sze is expected to dominate . for a broad bin centered at multipole 4066
, we find @xmath0 , of which @xmath1 is the expected contribution from primary cmb anisotropy and @xmath2 is the expected contribution from undetected radio sources .
the resulting constraint on secondary anisotropy is @xmath72 , implying an upper limit of @xmath73 at 95% cl ( @xmath74 at 68% ) .
the sza results indicate lower secondary anisotropy power than previously reported at 30 ghz by cbi @xcite and bima @xcite .
the discrepancies between the sza result and the cbi and bima results , however , are not highly significant after accounting for measurement uncertainties and non - gaussian sample variance .
we show that the level of sze anisotropy power implied by the sza measurement is in good agreement with expectations based on simulated compton @xmath78-maps made with @xmath4 , but disfavor values of @xmath99 . the differences among various simulations , however , prevent a more quantitative determination of @xmath7 .
we thank john cartwright , ben reddall , and marcus runyan for their significant contributions to the construction and commissioning of the sza instrument .
we thank the staff of the owens valley radio observatory and carma for their outstanding support .
we also thank tom crawford , kyle dawson , nils halverson , bill holzapfel , ryan keisler , tom plagge , tony readhead , and jonathan sievers for helpful discussions , and david s. meier and eric greisen for assistance with the vla data reduction .
we gratefully acknowledge the james s. mcdonnell foundation , the national science foundation and the university of chicago for funding to construct the sza .
the operation of the sza is supported by nsf division of astronomical sciences through grant ast-0604982 .
partial support is provided by nsf physics frontier center grant phy-0114422 to the kavli institute of cosmological physics at the university of chicago , and by nsf grants ast-0507545 and ast-05 - 07161 to columbia university .
am acknowledges support from a sloan fellowship , sm from an nsf astronomy and astrophysics fellowship , and cg , sm , and ms from nsf graduate research fellowships . | we present 30 ghz measurements of the angular power spectrum of the cosmic microwave background ( cmb ) obtained with the sunyaev - zeldovich array .
the measurements are sensitive to arcminute angular scales , where secondary anisotropy from the sunyaev - zeldovich effect ( sze ) is expected to dominate . for a broad bin centered at multipole 4066
we find @xmath0 , of which @xmath1 is the expected contribution from primary cmb anisotropy and @xmath2 is the expected contribution from undetected radio sources .
these results imply an upper limit of 155@xmath3 ( 95% cl ) on the secondary contribution to the anisotropy in our maps .
this level of sze anisotropy power is consistent with expectations based on recent determinations of the normalization of the matter power spectrum , i.e. , @xmath4 . |
a _ legendrian knot _ in @xmath1 is a smooth embedding of @xmath2 that is always tangent to the @xmath0planes of the so - called _ standard contact structure _ @xmath3 , and a _ link _ is a finite disjoint union of knots .
we consider legendrian knots up to _ legendrian isotopy _ , that is homotopy through legendrian knots .
the most basic , `` classical '' invariants of legendrian isotopy are the thurston bennequin number @xmath4 , rotation number @xmath5 , and smooth type .
( of course , any invariant of smooth isotopy is also an invariant of legendrian isotopy . ) in this paper , we will study the so - called ruling invariants , introduced by chekanov and pushkar @xcite and independently fuchs @xcite .
the beginning of legendrian knot theory dates back to 1984 when bennequin @xcite proved the famous inequality @xmath6 where @xmath7 is an arbitrary legendrian knot and @xmath8 is its ( smooth ) seifert genus .
our goal is to strengthen the relationship between the genus and legendrian invariants by proving the following : [ thm : canon ] the genus of any @xmath0graded ruling of the legendrian knot @xmath7 is less than or equal to the canonical genus of @xmath7 .
rulings and the idea that a genus may be associated to them are central to this paper .
however the discussion of these notions is deferred to section [ sec : rul ] .
recall that the _ canonical genus _
@xmath9 is the minimum of the genera of spanning surfaces obtained using seifert s algorithm on diagrams of the given knot .
the proof of theorem [ thm : canon ] , given toward the end of section [ sec : dan ] , is a trivial application of a result of rutherford @xcite and morton s inequality @xcite that the @xmath10degree of the homfly polynomial bounds the canonical genus from below . because for homogeneous knots ( a class of knots that includes all alternating and positive knots ) the canonical genus agrees with the genus @xcite , the following is immediate .
[ kov ] the genus of any @xmath0graded ruling of the homogeneous legendrian knot @xmath7 is less than or equal to the genus of @xmath7 .
in addition to homogeneous knots , the statement of the corollary holds for all prime knots up to @xmath11 crossings ; see section [ sec : jim ] .
it also holds for connected sums of the aforementioned by the additivity of genus and proposition [ pro : connsum ] .
it does not , however , hold in general .
[ ellenpelda ] there exist ( prime ) knots whose genus is less than the maximum of the genera of their @xmath0graded rulings .
two examples are shown in figure [ fig:14n22180 ] .
one of these ( @xmath12 in the hoste thistlethwaite table ) is a whitehead double of the trefoil , but the other ( @xmath13 ) is not a satellite knot ( thus it is hyperbolic ) .
both are of genus @xmath14 and both possess rulings with genera up to @xmath0 .
the genus @xmath0 ruling is unique and @xmath15graded for both .
this shows that corollary [ kov ] fails in general for non - homogeneous knots even if we restrict the statement to @xmath15graded rulings .
a strengthening of our statements to the @xmath16ball genus is not conceivable either .
there exist slice knots , for example the alternating knot @xmath17 in rolfsen s table , with positive genus @xmath0graded rulings
. the paper is organized as follows . in section [ sec : rul ] , we review rulings , their gradedness and genera , and establish some relations of these to the topology of the ruling . in section [ sec : dan ] , we review rutherford s work and prove theorem [ thm : canon ] .
section [ sec : jim ] details how the two knots of example [ ellenpelda ] were found .
acknowledgements : the phenomena described in this paper are related to lagrangian surfaces in @xmath18 that a legendrian in @xmath19 bounds . thus this paper can be thought of as a prelude of forthcoming joint work @xcite with tobias ekholm and ko honda .
i had many inspiring discussions with them .
i also thank jim hoste for the tremendous insight gained from the computer searches of knot tables that he conducted for me .
my conversations with francis bonahon and dan rutherford were very useful , too .
the _ front diagram _ of a legendrian link @xmath20 is its projection onto the @xmath21plane . for generic @xmath7
, it has an equal number of so - called _ left and right cusps _ ( with respect to the @xmath22direction ) , and the rest of the diagram consists of strands that connect a left cusp to a right cusp .
maslov potential _
@xcite is a locally constant @xmath23valued , up to sign , of a legendrian link is defined as the greatest common divisor of the rotations of its constituent components . ]
function that is defined along these strands with the following property : at each cusp , out of the two strands that join there , the one that is ( locally ) upper has potential that is @xmath14 higher than that of the other strand . recall that the _ index _ of a crossing in a front diagram is the difference ( in @xmath23 ) of the maslov potentials of the two intersecting strands .
( the potential of the lower strand is to be subtracted from that of the upper one . )
we adopt the convention that when a front diagram is oriented , strands traveling to the right ( increasing @xmath22 ) have even maslov potential .
this implies that even for multi - component links , a crossing has even index if and only if it is positive . for the following definition @xcite
, we assume that the front diagram of the legendrian is generic in that no pair of crossings share the same @xmath22coordinate .
[ def : rul ] a _ ruling _ is a partial smoothing of a front diagram @xmath24 where certain crossings , called _ switches _ , are replaced by a pair of arcs as in figure [ fig : switch ] so that the diagram becomes a union of standard unknot diagrams , called _
eyes_. ( an eye is a pair of arcs connecting the same pair of left and right cusps that contain no other cusps and that otherwise do not meet , not even at switches . ) we assume the so - called _ normality condition _ : in the vertical ( @xmath25 ) slice of the diagram through each switch , the two eyes that meet at the switch fit one of the three configurations in the middle of figure [ fig : switch ] . the notion above is also known as an _
ungraded ruling_. if we assume that all switches are of even index , we get _
@xmath0graded rulings_. the set of these objects will be denoted by @xmath26 . finally , _
@xmath15graded rulings _ are those in which each switch has index @xmath27 . the eyes bound disks and with the addition of twisted bands at the switches , we may consider the ruling as a surface . of @xcite . ] this may be referred to as the _ surface associated to the ruling _ , but usually we will simply identify the ruling and the surface .
the _ genus _ of a ruling is the genus of its associated surface .
this surface does not come with an embedding into @xmath28 though .
( in @xcite , we will associate to it an immersed exact lagrangian in @xmath18 . ) in fact it follows from the results of this paper that such an embedding often does not exist . but
first , let us make several simple observations .
[ iranyit ] the surface associated to a @xmath0graded ruling is orientable .
if an ungraded ruling of a legendrian link is topologically an orientable surface , then the link has an orientation so that the ruling is @xmath0graded .
consider the maslov potential on the front diagram .
the entire top arc of each eye in a @xmath0graded ruling is either on odd or on even potential and the entire bottom arc has the opposite parity .
switches connect odd arcs to odd arcs and even arcs to even arcs .
hence if we give ` odd top ' eyes one orientation of the projection plane and to ` even top ' eyes the opposite orientation , then ( by an examination of figure [ fig : switch ] ) this will extend continuously over the bands . for the other claim , simply induce an orientation of the legendrian from that of its ruling . doing so , all switches will be positive crossings .
now the following is obvious : for front diagrams of ( single - component ) knots , a ruling is @xmath0graded if and only if its associated surface is orientable .
ungraded , that is not - necessarily - orientable versions of theorem [ thm : canon ] and corollary [ kov ] fail even for alternating knots .
the knot @xmath29 ( the positive @xmath30 torus knot ) bounds an embedded mbius band , thus has unorientable genus @xmath14 , and it also has a ruling of oriented genus @xmath0 ( i.e. , unoriented genus @xmath16 ) .
if a ruling is topologically a disk , then it is @xmath15graded . in particular
its boundary legendrian has @xmath31 .
the eyes of such a ruling are connected by the switches in a tree - like fashion .
consider a leaf of this tree and notice that because of the way the maslov potential is defined , its single switch must be of index @xmath27 .
then remove this eye from the diagram and induct on the number of eyes .
in fact , from the existence of a @xmath0graded ruling it already follows that @xmath31 @xcite . on the other hand
, there exist legendrians with @xmath31 and ungraded rulings , hence with maximal @xmath4 in their smooth type @xcite , yet without a single @xmath0graded ruling .
one such example is the untwisted double of the negative trefoil shown in figure [ untwdblnegtref ] , produced using the method of @xcite on a @xmath32adequate diagram of the knot . from rutherford s results , to be reviewed in the next section
, it follows that no legendrian representative of this knot can have a @xmath0graded ruling .
we may classify rulings according to their genera .
in fact the quantity @xmath33 introduced in @xcite , is just the euler characteristic of the surface associated to the ruling @xmath34 .
the counts of ungraded , @xmath0graded , and @xmath15graded rulings with a given genus ( or euler characteristic ) are legendrian isotopy invariants @xcite .
the sequence of these counts for all genera is called the _ complete ruling invariant_. in the @xmath15graded case , the complete ruling invariant is very effective in distinguishing legendrian knots with the same classical invariants .
by contrast , the ungraded and @xmath0graded complete ruling invariants are determined by the smooth type and the thurston
bennequin number .
this is true by the two main theorems of @xcite .
in particular , in the @xmath0graded case rutherford proves the following .
[ thm : dan ] let @xmath35 denote the homfly polynomial of the legendrian knot @xmath7 , and let @xmath36 be the coefficient of @xmath37 in @xmath35 . then @xmath38 [ max ] on the other hand , the smallest @xmath39degree in the homfly polynomial , which we will denote by @xmath40 , is a well - known strict upper bound on @xmath4 @xcite . by the above
, the existence of a @xmath0graded ruling implies that this bound is sharp ( that is , @xmath41 ) , i.e. that the legendrian in question has maximal thurston
bennequin number within its smooth isotopy class .
conversely , @xmath41 implies that our legendrian does have @xmath0graded rulings .
in theorem [ thm : dan ] , our replacement of rutherford s @xmath42 with @xmath43 is valid by proposition [ iranyit ] .
note that we also replaced the variable @xmath44 in his formulation of the homfly polynomial by @xmath45 . for a knot type @xmath46
, we define the _ ruling genus of @xmath46 _ , denoted by @xmath47 , to be the supremum of the genera of all @xmath0graded rulings of all legendrian representatives of @xmath46 .
this is either a finite value ( and in that case a maximum ) or @xmath48 .
note that by theorem [ thm : dan ] and remark [ max ] , any front diagram with smooth type @xmath46 and maximal thurston
bennequin number may be used to determine @xmath47 . in particular , if we know that @xmath49 , then we may directly read @xmath50 off of the homfly polynomial . on the other hand , deciding whether @xmath51 is essentially equivalent to knowing the maximum thurston bennequin number of @xmath46 . at the time of this writing , that is still a hard problem in general
. with this new terminology , corollary [ kov ] simply asserts that @xmath52 whereas by theorem [ thm : canon ] , @xmath53 from rutherford s theorem above , it is obvious that the ruling genus @xmath50 is at most half of the @xmath10degree of the homfly polynomial , which in turn is a well known lower bound ( * ? ? ? * theorem 2 ) for the canonical genus of the knot .
we close this section with one last simple observation , which essentially follows from ( * ? ? ?
* proposition 11.2 ) .
[ pro : connsum ] the ruling genus is additive for connected sums : @xmath54 choose front diagrams for @xmath55 and @xmath56 with maximal thurston
bennequin number .
one may form their connected sum as in figure [ fig : connsum ] .
assume without loss of generality that the maslov potentials near the two merging cusps match .
then it is plain to see that with any of the three notions of definition [ def : rul ] , the set of rulings for the new diagram is the cartesian product of the two old sets .
the claim follows at once .
one finds several sufficient conditions for @xmath57 in terms of polynomial knot invariants . for all knots .
whitehead doubles , having genus @xmath14 , are natural candidates for a counterexample ; see example [ ellenpelda ] .
however i initially dismissed them , having misunderstood a claim in @xcite . ] these are listed in the two propositions below . as before , let @xmath40 denote the minimum @xmath39degree of the homfly polynomial @xmath58 of @xmath46 .
versions of and on the following list were also published in @xcite .
a. [ lenny ] `` khovanov beats homfly '' : @xmath59 , where @xmath60 is the khovanov homology group of @xmath46 in bigrading @xmath61 .
b. [ kauff ] `` kauffman beats homfly '' : the minimum @xmath39degree of the kauffman polynomial of @xmath46 is less than @xmath40 . c. [ negative ] `` negative counts '' :
there is an integer @xmath62 so that the coefficient @xmath63 of @xmath64 in the homfly polynomial is negative .
d. [ subset ] `` subset failure '' : the minimum @xmath39-degree in the kauffman polynomial is @xmath40 or more , but there is an integer @xmath62 so that if @xmath65 is the coefficient of @xmath66 in the dubrovnik version of the kauffman polynomial , then @xmath67 fails to hold . in the first two cases , we separate @xmath68 and @xmath40 by inequalities .
claim is a direct consequence of the so - called khovanov bound of the thurston
bennequin number @xcite .
claim follows in the same straightforward way from the so - called kauffman bound @xcite . in the other two cases , we assume @xmath41 and find a contradiction with rutherford s results .
claim is obvious from theorem [ thm : dan ] . to prove the last statement
, one needs to compare theorem [ thm : dan ] with rutherford s other main theorem in @xcite , which says that the coefficients @xmath69 count ungraded rulings of some given ( @xmath70 , to be exact ) euler characteristic .
now the contradiction is obvious from the fact that @xmath0graded rulings are also ungraded .
let @xmath71 denote the maximum @xmath10degree that appears in the homfly polynomial of @xmath46 in a monomial that also contains @xmath72 . with this , either @xmath73 or @xmath74 .
( it is difficult to tell which one is true in the case of an arbitrary knot , even if none of the conditions listed in proposition [ pro : noruling ] holds . at present ,
the best one can do is to try and construct a legendrian representative with @xmath41 .
if such a front is found , then we know that @xmath73 . ) there is nothing to prove if @xmath74 , so we will assume that @xmath73 .
that implies @xmath41 for those legendrian representatives that have @xmath0graded rulings .
now , the bennequin test is so called because it is obvious from : @xmath78 using the conditions listed above , jim hoste kindly conducted a computer search of the table of knots for potential examples with @xmath80 .
knots with @xmath81 ( i.e. , @xmath82 or @xmath48 ) were also ruled out . among prime knots up to @xmath11 crossings ,
only two candidates were found : @xmath83 and @xmath84 .
it turns out that these knots do have @xmath0graded rulings ; in fact @xmath85 for both , with unique genus @xmath0 rulings . in both cases
, i was able to add a @xmath14handle to that ruling , realize the result as an embedded surface , and then find a gabai disk decomposition @xcite .
thus both knots have @xmath86 .
however at crossing number @xmath87 , the phenomenon of example [ ellenpelda ] occurs .
y. v. chekanov and p. e. pushkar : _ combinatorics of fronts of legendrian links and arnold s 4conjectures _ ( russian ) uspekhi mat .
nauk 60 ( 2005 ) , no .
1(361 ) , 99154 ; english translation in russian math .
surveys 60 ( 2005 ) , no . 1 , 95149 . | each ruling of a legendrian link can be naturally treated as a surface . for knots ,
the ruling is @xmath0graded if and only if the surface is orientable . for @xmath0graded rulings of homogeneous ( in particular , alternating ) knots ,
we prove that the genus of this surface is at most the genus of the knot . while this is not true in general
, we do prove that the canonical genus ( a.k.a .
diagram genus ) of any knot is an upper bound for the genera of its @xmath0graded rulings . |
gravity and dark energy govern the structure in the universe at the largest scales but complex baryonic processes like cooling , heating , self - gravity and star formation are important at galactic scales ( e.g. , @xcite ) .
numerical simulations have made tremendous progress in understanding galaxy formation , starting from pure gravitational n - body simulations to the current models which try to model the aforementioned complex processes .
modeling the gravitationally - interacting dark matter is straightforward in principle , and only limited by the available computing power . but
the modeling of baryonic processes is rather involved .
in particular , there is no consensus on which baryonic processes are important and how they should be implemented numerically .
given the dynamic range of scales , from large scale structure ( 10s of mpc ) to an individual star forming cloud ( @xmath4pc ) , simulations have to resort to unresolved ` subgrid ' models for star formation and feedback due to star formation ( e.g. , @xcite and references therein ) .
while different star formation recipes seem to give similar star formation histories and stellar mass distributions , provided molecular clouds are resolved ( e.g. , @xcite ) , simulations are quite sensitive to the various feedback prescriptions ( e.g. , thermal feedback due to supernovae , momentum injection via dust absorbing / scattering photons produced by massive stars and supernovae ) even with high resolution .
stars form in clusters and super - star - clusters ( 100s to @xmath5 stars ) of various sizes and in different environments , ranging from low density galactic outskirts to dense galactic centers ( see @xcite for a review ) .
these clusters are observed to disrupt the dense molecular clouds in which they are born ( e.g. , see @xcite ) .
this stellar feedback ( due to strong radiation , stellar winds and supernovae ) disperses cold gas and suppresses further star formation . because of the stellar initial mass function ( imf ) and the main sequence lifetimes , the energy input is roughly constant per unit time over the life time of stars more massive than @xmath6 ( @xmath7 myr ; @xcite ) .
therefore , stellar feedback is sometimes modeled as a constant luminosity driven blast wave ( @xcite ) . a superbubble ( sb )
expands faster than an isolated supernova remnant ( snr ) because of continuous energy injection , and suffers smaller cooling losses because most supernovae ( sne ) go off in a low density bubble . the hot bubble breaks out through the gas disk if the outer shock driven by overlapping sne crosses the scale height with a sufficient mach number within the starburst lifetime ( e.g. , see section 2 of @xcite ) .
after breakout the hot , metal - rich stellar ejecta is spread out into the galactic halo via the rayleigh - taylor instability .
spreading of metals over large scales is required to explain the high metallicity observed in the intergalactic medium ( igm ) , far away from the stellar disk ( @xcite ) .
standard models for feedback through multiple sne assume that a fraction @xmath8 of the total explosion energy is retained in the hot ism ( e.g. , @xcite ) .
this fraction is much larger than the estimates for a single snr ( @xmath9 ) after the radiative phase at @xmath0 myr ( e.g. , @xcite ) .
moreover , there are further radiative and adiabatic losses , such that over the timescale of order tens of myr , the available mechanical energy in the bubble and shell is a negligible ( @xmath10 ) fraction of the initial supernova ( sn ) energy input .
recently @xcite argued that sbs created by multiple sne in a star cluster are more effective as a feedback mechanism , in comparison to incoherent sne .
recent simulations of interacting sne by @xcite show that the fractional energy retained as thermal energy in the hot ism can be as large as @xmath11 only if the explosions are spatially and temporally correlated such that the radiative losses are effectively compensated by new explosions .
sbs produced by compact star clusters are expected to satisfy this condition . in this paper
we assume all sne to be coincident , a good approximation if the snr , when it becomes radiative , encompasses the whole cluster .
this makes sbs a more effective feedback agent which is responsible for magnificent galactic outflows .
in addition to elucidating differences between isolated sne and sbs , we compare various methods of injecting sn energy in numerical simulations of the ism .
different feedback processes , such as radiation pressure , photoionization , and cosmic rays , are important in explaining outflows in galaxies of different masses and redshifts ( e.g. , @xcite and references therein ) . in this paper
we only focus on the most important feedback component , namely , thermal feedback due to sne .
we find that the sn thermal energy must be deposited over a sufficiently small volume for it to create a hot bubble and to have an impact on the surroundings .
for a large energy deposition radius the cooling time is shorter than the thermalization timescale and thermal feedback is artificially suppressed .
most early implementations of sn feedback suffered from this problem ( see @xcite and references therein ) . the formation of a low density bubble is essential for thermal sn feedback to work because the energy of subsequent sne is deposited in the low density bubble and is not radiated instantaneously ; cooling is only restricted to the outer shock .
while this fact has been appreciated ( e.g. , @xcite ) , we present _ quantitative _ conditions for the formation of a strong shock and a hot , dilute bubble for different thermal feedback prescriptions .
the main culprit responsible for the inefficacy of thermal feedback , in both eulerian ( e.g. , @xcite ) and lagrangian ( @xcite ) simulations , is the lack of resolution . in reality
, a sn affects its surroundings starting at small ( @xmath12 1 pc ) scales by launching very fast ejecta ( @xmath13 km s@xmath14 ) into the ism ( e.g. , see @xcite ) .
simulations in which sn energy is not deposited at small enough scales have to resort to artificial measures ( turning off cooling for several energy injection timescales , depositing energy into artificial ` hot ' phase that does not cool , etc . ) in order for feedback to have any impact at all on the ism ( e.g. , @xcite ) .
these measures are necessary for including the effects of supernova heating in large - scale simulations .
the continuous injection of mass and energy by sne deep within the hot sb is expected to launch a steady wind ( as first calculated by @xcite ; hereafter cc85 ) .
cc85 obtained analytic solutions for a superwind assuming a constant thermal energy and mass input rate with an injection radius . by modeling realistic sne as fast moving ejecta within sbs
we show that a steady cc85 wind is obtained only if a large number of sne ( @xmath3 ) go off within the star cluster . for a smaller number of sne , an individual sn s kinetic energy
is not thermalized within a small injection radius and there is an unsteady outflow .
this should have implications for works that simply assume a cc85 wind within the sb .
our paper presents analytic results that can be readily used for the numerical implementation of thermal feedback using various recipes .
these analytic criteria are verified and extended using idealized numerical simulations . the fundamental difference between isolated supernovae and superbubbles that the former is ineffective on galactic scales
is highlighted .
the paper is organized as follows . in section [ sec : fb_pres ] we describe various ways of implementing thermal feedback due to sne .
section [ sec : num_sim ] describes the numerical setup used to study sne and sbs . in section [ sec :
anal ] we present different analytic criteria for feedback to work with a range of feedback prescriptions .
we also derive the conditions for obtaining a thermalized cc85 wind within a sb . in section [ sec : results ] we present 1-d numerical simulations of different feedback recipes with and without cooling , and compare the results with our analytic estimates .
we also briefly discuss the effects of magnetic fields and thermal conduction . in section [ sec : conc ]
we discuss the implications of our results on galaxy formation , feedback simulations , and on galactic wind observations .
although ism is multiphase and extremely complex , for simplicity we consider a uniform , static model with a given density ( typically @xmath15 @xmath16 ) and temperature ( @xmath17 k corresponding to the warm ism ) .
we do not consider stratification because the disk scale height is typically a few 100 pc and the fizzling of sn feedback is essentially a small scale problem .
moreover , the scales of interest ( 100s of pc ; few myr ) are much bigger than the cluster size and the local ism / circumstellar inhomogeneities . for simplicity
we also assume that all sne explode at @xmath18 .
this is a good approximation because the size of a typical ( super ) star - cluster is smaller than the bubble size at the beginning of the radiative phase . the sn prescriptions that we consider in our analytic estimates and numerical simulations cover the full range of methods used in the literature .
these are : 1 . _ kinetic explosion models _ ( ke ) : in these models the sn energy ( @xmath19 , chosen to be 1 bethe @xmath20 erg ) is given to a specified ejecta mass ( @xmath21 , chosen to be 1 @xmath22 ) distributed uniformly with an ejecta density @xmath23 within an ejecta radius @xmath24 .
the ejecta velocity is homologous with @xmath25 within the ejecta ; the normalization is such that the kinetic energy of the ejecta is @xmath19 ; i.e. , @xmath26 .
the ejecta temperature is taken to be small ( @xmath27 k ) .
after every ( fixed ) sn injection time @xmath28 the innermost @xmath24 of the volume is _ overwritten _ by the ejecta density and velocity , thereby pumping sn energy into the ism . after the reverse shock propagates toward the bubble center ,
once the swept - up mass is comparable to the ejecta mass , the bubble density structure is fairly insensitive to the ejecta density distribution ( @xcite ) .
this model most closely resembles a physical snr in early stages at small ( @xmath12 1 pc ) scales when the sn ` piston ' at large speed rams into the ism .
this prescription is not widely used in galaxy formation simulations ( there are some exceptions , e.g. , @xcite ) .
2 . _ thermal explosion models _ ( te ) : in these models the energy is deposited within the ejecta radius in form of thermal energy at an interval of @xmath28 . there are two variants of this model . in one class ,
the mass and internal energy densities are _ overwritten _ within the ejecta radius ( @xmath24 ) such that the _ uniformly distributed _
ejecta thermal energy is @xmath19 ( 1 bethe ) and the _ uniformly distributed _ ejecta mass is @xmath29 .
we abbreviate these models as teo and they behave like ke models .
the second class of models is where we _ add _ ( in contrast to overwrite ) the ejecta mass ( with uniform density ) to the preexisting mass and the ejecta thermal energy ( distributed uniformly ) to the preexisting internal energy within @xmath24 .
we refer to these models as tea .
there are significant differences between the tea and teo / ke models in presence of cooling because tea ejecta can become radiative if thermal energy is added to a dense ism .
most models in the literature are analogous to tea models with some variations ( e.g. , @xcite , @xcite , @xcite ; some works such as @xcite , @xcite , @xcite unphysically turn off cooling for some time for sn feedback to have an impact ) . sometimes all the sn energy is deposited in a single grid cell ( e.g. , @xcite ) or in a single particle ( e.g. , sn particle method in section 3.2.4 of @xcite ) .
luminosity driven models _ ( ld ) : as discussed earlier , for typical imfs , the mechanical energy input due to ob stars per unit time is roughly constant .
this motivates a model in which internal energy and mass within an injection radius ( denoted by @xmath24 ) increase at a constant rate corresponding to internal energy @xmath19 and mass @xmath21 for each sn
. some of the works that use this prescription are @xcite .
all our models are identified with the number of ob stars @xmath30 ( which equals the number of sne in the star cluster ) or luminosity @xmath31 , where @xmath32 is the lifetime of the ob association ( taken to be 30 myr ) and @xmath33 is the time interval between sne .
we note that the overwrite models do not strictly conserve mass and energy .
while the supernova energy is much larger than the overwritten thermal energy , one needs to choose a small enough ejecta radius such that the overwritten mass is subdominant relative to the ejecta mass .
the explosion models are substantially slower compared to the smooth luminosity driven models because of very high temperatures created due to sudden energy injection in the former .
in this section we describe the numerical setup corresponding to our one - dimensional simulations discussed in section [ sec : results ] .
our numerical simulations use the ism setup and the various feedback prescriptions described in section [ sec : fb_pres ] .
the ism density and temperature are chosen to 1 @xmath16 and @xmath17 k , respectively , unless specified otherwise .
the mean mass per particle is @xmath34 and per electron is @xmath35 .
the initial density and temperature are uniform , and the velocity is zero .
we use the grid - based zeus - mp code in spherical , one - dimensional geometry ( @xcite ) to solve the standard euler equations with source terms mimicking sn energy / momentum / mass injection for the chosen feedback model , and a sink term in the internal energy equation representing radiative cooling .
our equations are similar but not identical to eqs .
20 - 22 in @xcite .
we note that the zeus code does not conserve energy to the machine precision .
however , we have confirmed that energy conservation holds to better than 90% in most of our runs .
in fact , the sedov - taylor blast wave in 1-d spherical coordinates is one of the standard test problems presented in @xcite ; the numerical results match analytic solutions very closely .
while most of our runs are hydro , in section [ sec : b_cond ] we briefly discuss runs with simple models of magnetic fields and thermal conduction ; various parameters for these runs are discussed there .
the radial velocity is set to zero at the inner radial boundary and other fluid variables are copied in the ghost cells .
outflow boundary conditions are applied at the outer radial boundary .
note that the outer boundary is out of causal contact .
since the setup produces strong shocks , we use the standard zeus artificial viscosity to prevent unphysical oscillations at shocks ( @xcite ) .
the cfl number of 0.2 is found to be more robust compared to the standard value of 0.5 , and is used in all the simulations .
the cooling function that we use is the collisional - ionization - equilibrium based solar metallicity table of @xcite , with the cooling function set to zero below @xmath17 k. the cooling step is implemented via operator splitting using the semi - implicit method of @xcite . cooling is subcycled and the number of subcycles is limited to be less than 100 .
most of our runs use a 1024 resolution grid extending from 1 pc to 2 kpc .
a logarithmically spaced grid is used to better resolve smaller radii ; there are equal number of grid points covering 1 pc to @xmath36 pc and @xmath36 pc to 2 kpc .
some runs with stronger sn feedback use a larger spatial extent ( c.f .
@xmath37 runs in fig .
[ fig : rho_vs_r_cc85 ] ) , and some uniform very high resolution runs ( 16384 grid points ; c.f . fig .
[ fig : d_t ] ) use a smaller extent ( 1 to 200 pc ) .
all our simulations ( except the very high resolution ones that are run for 3 myr ) are run for 30 myr , typical age of a young star cluster .
in this section we present the analytic criteria that need to be satisfied for various feedback models discussed in section [ sec : fb_pres ] to work .
these analytic estimates help us understand the results of numerical simulations discussed in section [ sec : results ] . while radiative cooling is the most discussed phenomenon in the context of fizzling sn feedback , the feedback prescription should satisfy additional constraints for the energy input to couple realistically with the ism .
a recurring concept in what follows is that of thermalization ; i.e. , in order to be effective the input energy should have time to couple to the ism before it is radiated or is overwritten . in section [ sec : cc85 ]
we show that a steady superwind within a superbubble , as envisaged by @xcite , occurs only if the number of sne is sufficiently large . in the following sections we derive upper limits on the ejecta radius within which the feedback energy must be deposited for it to be effective .
we can easily convert this radius limit into a critical mass resolution needed in smooth - particle - hydrodynamics ( sph ) simulations ; namely , @xmath38 , where @xmath39 is the ism density , @xmath40 is the number of neighbors used in the sph smoothing kernel and @xmath41 is the maximum sph gas particle mass required for feedback to work . in models where energy within the ejecta radius is overwritten ( ke , teo )
the ejecta radius should be smaller than a critical radius ( @xmath42 ) for the input energy to get coupled to the ism .
the critical radius equals the sedov - taylor shock radius at @xmath28 , because the shock quickly transitions from a free - expanding to a sedov - taylor state ; the sedov - taylor radius ( when the swept up ism mass equals the ejecta mass ) is @xmath43 , much smaller than the estimate in eq .
[ eq : r_therm ] , where @xmath44 is the ejecta mass in solar units . ] the time lag between sne , [ eq : r_therm ] r_crit ( ) ^1/5 50 pc n^-1/5 e_ej,51 ^ 1/5 t_sn,0.3 ^ 2/5 , where @xmath39 ( @xmath45 ) is the ism ( number ) density ( assuming @xmath34 ) , @xmath46 is the ejecta energy in units of @xmath47 erg , and @xmath48 is the time between consecutive sne in units of @xmath49 myr . if the ejecta radius is larger than this value the ejecta energy is overwritten before it can push the outer shock .
thus , in such a case , the input sn energy is overwritten without much affecting the ism . for thermal sn feedback to launch a strong shock the energy
should be deposited over a small enough volume , such that the post - shock pressure is much larger than the ism pressure .
this is equivalent to demanding the outer shock velocity to be much larger than the sound speed in the ism .
the shock velocity ( @xmath50 ; @xmath51 is the outer shock radius ) , expressed in terms of the shock radius in sedov - taylor stage , is @xmath52 for an isolated sn and @xmath53 for a luminosity driven sb ( @xcite ) .
the condition for a strong shock for an isolated sn is ( @xmath54 ; @xmath55 is the ism isothermal sound speed ) [ eq : r_sn ] r_ej 174 pc e_ej,51 ^ 1/3 n^-1/3 t_4 ^ -1/3 and for a sb is ( see eq . 3 in @xcite ) [ eq : r_sb ] r_ej 1.5 kpc l_ej,38 ^ 1/2 n^-1/2 t_4 ^ -3/4 , where @xmath56 is the ejecta luminosity ( @xmath57 for explosion models ) in units of @xmath58 erg s@xmath14 ( this corresponds to @xmath59 over @xmath60 myr ) and @xmath61 is the ism temperature in units of @xmath17 k. the sonic constraint ( @xmath54 ) is typically less restrictive than the compactness requirements due to cooling in a dense ism ( see next section ) .
@xcite , who considered supernova feedback in the hot ism ( @xmath62 k ) of galaxy clusters and elliptical galaxies , found that the shock can quickly ( when outer radius is only @xmath63 pc ; see eq .
[ eq : r_sn ] ) decelerate to attain the sound speed in the hot ism .
after this the outer shock propagates as a sound wave .
while the sound wave can spread the sn energy over a larger radius ( @xmath64 for a sound wave , unlike a strong blast wave in which @xmath65 ) , energy dissipation is not as efficient as in shocks . in luminosity driven models ( ld ) ,
sn feedback does not fizzle out ( in fact , the shock can get started ) only if the cooling rate is smaller than the energy deposition rate , i.e. , @xmath66 ( @xmath67 $ ] is the cooling function ) , or [ eq : lum_fizzle ] r_ej 20 pc l_ej,38 ^ 1/3 n^-2/3 _
-22 ^ -1/3 , where @xmath68 is the cooling function in units of @xmath69 erg @xmath70 s@xmath14 .
the above criterion ( eq . [ eq : lum_fizzle ] ) for the luminosity driven models is quite different from the criterion that we now derive for the widely used thermal explosion models with energy and mass addition ( tea model in section [ sec : fb_pres ] ) .
since energy is _ added _ to the ( possibly dense ) pre - existing medium , cooling in this model can be substantial .
in contrast , since the ejecta density is low , cooling losses are smaller in the overwriting models ( ke , teo ) . for tea models to launch a shock , radiative losses over the timescale in which the shock from a point explosion reaches the ejecta radius , [ eq : t_therm ] t_ej = e_ej^-1/2 r_ej^5/2 ^1/2 , should be smaller than the energy deposited ( here @xmath39 is the density of the medium in which energy is injected , not necessarily the ism density ) ; i.e. , [ eq : cool_cond ] n^2 t_ej 3 e_ej/4r_ej^3 . plugging in the expression for @xmath71 ,
we get [ eq : eadd_fizzle ] r_ej 31 pc e_ej , 51 ^ 3/11 _
-22 ^ -2/11 n^-5/11 .
this condition is much more restrictive than the one obtained by replacing @xmath71 in eq .
[ eq : cool_cond ] by the cfl stability timestep .
moreover , this is the appropriate timescale to use because the relevant timescale for the injected energy to couple to the ism is the thermalization time ( @xmath71 ) .
@xcite and @xcite have used similar arguments and derived results not too different from ours for the thermal explosion addition models . a slight difference from our work
is that they consider energy deposition over a resolution element ( a necessity because of a larger range of scales in cosmological galaxy simulations ) , but we allow for energy deposition over a resolved region . @xcite
have expressed their resolution limit in terms of the cooling rate per unit mass and @xcite in terms of the post - shock temperature ; we use the cooling function ( @xmath72 ) to express the critical radius within which the energy needs to be deposited . in models where the energy and mass densities are overwritten within @xmath24 ,
the condition for overcoming cooling losses and launching a shock is @xmath73 where ejecta number density @xmath74 and @xmath75 ; note that this expression is different from eq .
[ eq : cool_cond ] in that the ejecta density is used instead of the ism density .
the overwrite models ( ke , teo ) behave quite differently from addition ( tea , ld ) models because a larger ejecta radius means a smaller density ejecta to which energy is added . replacing the ism density by the ejecta density in eq .
[ eq : t_therm ] gives @xmath76 , and the condition for energy thermalization is r_ej 0.003 pc m_ej,^5/4 _
-22 ^ 1/2 e_ej,51 ^ -3/4 . in order to avoid radiative losses the ejecta radius should be _ larger _ than above .
this early cooling of the mass loaded sn ejecta , responsible for creating cold filaments in young sne ( e.g. , @xcite ) , is physical ( unlike fizzling out of the energy addition models ) and should reduce the energy available to drive the sn .
all our simulations use an ejecta radius much greater than this limit .
@xcite ( hereafter cc85 ) found analytic solutions for a luminosity driven wind with a fixed injection radius .
luminosity injection is expected to drive both an outer shock bounding the bubble and a wind that shocks within the hot bubble at the termination shock ( see fig . 1 in @xcite ; see also ld run in fig .
[ fig : rho_vs_r_cc85 ] ) . in this section
we show that for a small number of sne ( c.f .
[ eq : cc_crit ] ) the sn ejecta does not thermalize within the termination shock . in that case , the density inside the bubble is much lower than the cc85 wind because most sne occur in the dilute bubble created by earlier sne and the thermalization radius is comparable to the outer shock radius .
this has important implications on cooling and luminosity of sn ejecta .
following @xcite , the outer shock radius of a luminosity driven bubble is given by @xmath77 , velocity by @xmath78 , and the post - shock pressure by @xmath79 .
assuming a steady superwind , the ram pressure at the termination shock ( @xmath80 ; the wind is assumed to be supersonic at this radius ) is @xmath81 , where @xmath82 is the wind velocity , @xmath83 is the density upstream of the termination shock , and @xmath84 is the mass injection rate .
the wind termination shock ( @xmath80 ) is located where the wind ram pressure balances the bubble pressure ; i.e. , @xmath85 using @xmath53 and @xmath86 gives , [ eq : r_ts ] ( ) ^1/2 0.08 e_ej , 51 ^ -1/12 m_ej , ^1/4 n^-1/6 r_os , 2 ^ -1/3 t_sn , 0.3 ^ -1/6 , where @xmath87 is the outer shock radius in units of 100 pc , @xmath88 is the time between sne normalized to 0.3 myr ( corresponding to @xmath59 ) ; we have used @xmath57 and @xmath89 .
the ratio @xmath90 depends very weakly on time ( @xmath91 ) ; this comes from the time dependence of @xmath51 in eq .
[ eq : r_ts ] .
the reverse shock for an isolated sn very quickly ( at the beginning of sedov - taylor stage ) collapses to a point but the termination shock for a sb is present at all times . thus the non - radiative termination shock can power a sb long after the outer shock becomes radiative , unlike a sn which dies off shortly after the outer shock becomes radiative ( see section [ sec : energy_budget ] for our results from simulations ) .
the condition for the existence of a smooth cc85 wind is that the ejecta thermalization radius should be smaller than the termination shock radius .
the superwind is mass loaded by previous sne ( the bubble density in the absence of mass loading is quite small because most of the mass is swept up in the outer shell ) . the swept up mass till radius @xmath92 in a cc85 wind is @xmath93 where @xmath94 is the wind density profile ; here we have assumed that the swept up mass is dominated by the supersonic portion of the wind . now
the thermalization radius ( the radius within which the deposited energy is thermalized and which should correspond to cc85 s injection radius ) of the ejecta is where the swept up mass roughly equals the ejecta mass , or [ eq : rth_cc ] r_th v_ts t_sn 3 kpc e_ej , 51 ^ 1/2 m_ej , ^-1/2 t_sn , 0.3 .
since the thermalization radius is quite large , a thermalized cc85 solution will only occur for large clusters ( with shorter @xmath28 , the time lag between sne ) ; for modest star clusters the ejecta will only thermalize beyond the termination shock . of course
, the thermalization radius can not be smaller than the size of the star cluster launching the energetic ejecta .
[ eq : r_ts ] and @xmath77 , the termination shock radius can be expressed as @xmath95 where @xmath96 is time in units of 0.3 myr .
a cc85 solution will appear only if this termination shock radius is larger than the thermalization radius ( eq . [ eq : rth_cc ] ) ; i.e. , if [ eq : cc_crit ] t_sn,0.3
0.007 e_ej,51 ^ -9/26 t_0.3 ^ 4/13 n^-3/13 m_ej,^15/26 .
this means that @xmath97 ( recall that @xmath33 , where @xmath60 myr is the cluster lifetime and @xmath30 is the number of sne ) is required for a cc85 wind to appear by 30 myr .
thus , a thermally driven cc85 wind occurs only for a sufficiently big starburst , with a large mass loading , and at late times .
in this section we present the results from our one - dimensional numerical simulations .
we vary the ism density and sn injection parameters to assess when sn energy can significantly affect the ism , both with and without cooling .
we also numerically verify the various analytic constraints presented in section [ sec : anal ] .
we discuss the structure of a radiative sb and compare the energetics of isolated sne and sbs .
while isolated sne lose most of their mechanical energy by a few myr , sbs can retain up to @xmath2 of the input energy long after the outer shock becomes radiative .
thus , sbs , and not isolated sne , are the viable energy sources for global , galactic - scale feedback . in section [ sec : b_cond ] we briefly discuss the impact of magnetic fields and thermal conduction on sbs .
the sn shock is launched once a protoneutron star forms at the center of a massive evolved star ( with size @xmath98 cm ) . in the ejecta dominated state
( when the swept up ism mass is less than the ejecta mass ) the cold ejecta is dominated by kinetic energy ( e.g. , @xcite ) . in our ` realistic ' simulations ( ke models ;
see section 2 ) we choose the ejecta to have a constant density and a velocity proportional to the radius ( homologous expansion ; this is a similarity solution for the freely expanding ejecta ) within the ejecta .
the sn shock develops a reverse shock after sweeping up its own mass in the ism ; this slows down the ejecta and communicates the presence of the ism to the supersonic ejecta . in this section
we compare the evolution of _ adiabatic _
( cooling is turned off ) ke models with different parameters , highlighting the importance of having a small ejecta radius ( @xmath24 ) even in absence of cooling for overwrite ( ke , teo ) models .
we have verified that kinetic explosion ( ke ) and thermal explosion overwrite ( teo ) models behave in a similar fashion .
[ cc][][1.][0]@xmath99 , title="fig : " ] figure [ fig : rho_vs_r ] shows the density profile as a function of radius ( normalized to the self similar scaling , @xmath100^{1/5}$ ] , where @xmath101 ) for different realistic runs ( results are similar for teo models ) with @xmath102 at 10 myr .
the runs with a large ejecta radius ( 100 pc ) give a smaller outer shock radius because most of the energy is overwritten without being thermalized ( see section [ sec : over ] for a discussion ) .
the problem is worse for larger @xmath30 ( shorter @xmath28 ) , as expected from eq .
[ eq : r_therm ] .
the normalized location of the outer shock falls almost on top of each other for a small ejecta radius ( @xmath103 pc ) .
as expected , the shock is weaker , broader , and with a modest density jump for a smaller number of sne .
while the ke model is most realistic , we expect other models in section [ sec : fb_pres ] to give a similar location for the outer shock after the swept - up ism mass equals the ejecta mass and the shock is in the sedov - taylor regime .
the structure within the bubble depends on sn prescription , as we show in section [ sec : cc85_sims ] .
[ ] [ ] [ 1.][0]@xmath104 [ ] [ ] [ 1.][0]@xmath105 ) is sufficiently small ; otherwise energy is overwritten before getting coupled to the ism .
there is no such problem for energy addition and luminosity driven models . at early times the outer shock radius scales with the sedov - taylor scaling ( @xmath65 ) and later on , after many sne go off , it steepens ( @xmath106 ) .
[ fig : r_vs_t],title="fig : " ] figure [ fig : r_vs_t ] shows the location of the outer shock ( measured by its peak density ) as a function of time for various models ( ke , ld , tea ) and sn parameters in absence of cooling .
the solid line at the bottom shows the transition from a single blast wave ( outer shock radius , @xmath65 ) to a continuously driven bubble ( @xmath106 ; @xcite ) for @xmath59 run .
the runs with more sne show such a transition very early on .
the dot - dashed line shows the outer shock radius for the ke run using a large ejecta radius violating the criterion in eq .
[ eq : r_therm ] ; the outer shock radius is much smaller than expected because energy is overwritten before it energizes the hot bubble ( see section [ sec : over ] ) .
the luminosity driven ( ld ) and kinetic explosion ( ke ) models agree only if the ejecta radius satisfies eq .
[ eq : r_therm ] for ke models ( we have verified that this constraint also applies to the thermal explosion overwrite [ teo ] models ) .
tea ( thermal explosion addition ) runs and ld runs fall on top of each other for both choices of @xmath24 ( 2 , 100 pc ) .
the outer shock radii for the runs with @xmath107 pc increase only after a thermalization time ( eq . [ eq : t_therm ] ; although in this case @xmath39 is not the ism density but the much lower density of the bubble within which energy is added ) .
[ cc][][1.][0]@xmath99 , but not for ke / tea models with @xmath59 ; the smooth cc85 wind is identified by the density profile varying @xmath108 between the ejecta radius and the termination shock ( various regions have been marked for the ld run ) .
the cc85 wind density using @xmath37 is slightly smaller for the ke model compared to the tea model because density is overwritten ( and hence mass is lost ) in ke models .
[ fig : rho_vs_r_cc85],title="fig : " ] in this section we show that a simple steady wind , as predicted by cc85 , exists within the bubble only if the number of sne is sufficiently large ( see section [ sec : cc85 ] ) .
figure [ fig : rho_vs_r_cc85 ] shows the density profile as a function of the scaled radius for various models .
the solid line shows density for a luminosity driven ( ld ) model with @xmath59 and @xmath103 pc ; various regions for the smooth cc85 wind within the bubble are marked .
the superwind has a structure identical to the cc85 wind ; the sonic point is just beyond the energy injection radius ( 2 pc ) .
the wind shocks at the termination shock ( @xmath80 ) where the wind ram pressure balances the bubble pressure .
the ratio of the termination shock and the outer shock ( @xmath51 ) is @xmath109 , in good agreement with eq .
[ eq : r_ts ] . for comparison ,
figure [ fig : rho_vs_r_cc85 ] also shows the density profiles for the kinetic explosion ( ke ) and thermal explosion addition ( tea ) models with the same parameters . while the outer shock radius agree for these runs , the density profiles within the bubble are quite different .
the most blatant difference , for runs with @xmath59 , is the absence of a cc85 wind in ke and tea models . in accordance with the discussion in section
[ sec : cc85 ] , sn shocks do not thermalize within the termination shock for a small number of sne ( see eqs .
[ eq : rth_cc ] & [ eq : cc_crit ] ) ; therefore a smooth cc85 wind is not expected in any model with small @xmath30 except ld . only for a large enough @xmath30 and late enough times
does a cc85 wind start to appear within the hot bubble .
figure [ fig : rho_vs_r_cc85 ] includes the density profiles for kinetic explosion ( ke ) and tea models using @xmath37 ( the inner [ outer ] radius of the computational domain for these runs is 0.5 pc [ 5 kpc ] ; @xmath110 pc is chosen to satisfy the constraint in eq .
[ eq : r_therm ] ) .
clearly , in these cases we see the appearance of the cc85 wind solution within the termination shock because the injected energy is thermalized . for the ke run with @xmath37 one
can still see the internal shocks due to isolated sne interacting with the superwind . the density profile for the ke model using @xmath111
is shown by the dotted line in figure [ fig : rho_vs_r ] . in agreement with eq .
[ eq : r_ts ] , the ratio @xmath90 increases with an increasing @xmath30 . for @xmath111 thermalization
is less complete as compared to @xmath37 , but happens within the termination shock . in comparison ,
a clear termination shock is absent for @xmath59 because the thermalization radius is larger than the termination shock radius ( see eq .
[ eq : cc_crit ] ) . in this section
we study the effects of radiative cooling on sne and sbs .
we focus on a few aspects : the fizzling out of thermal feedback in some models in which energy is not injected over a sufficiently small scale ; comparison of cooling losses and mechanical energy retained by radiative snrs and sbs ; the influence of magnetic fields and thermal conduction . as we mentioned in section
[ sec : cool_cond ] , some models ( tea , ld ) in which we _ add _ sn thermal energy in a dense ism , over a large radius , can suffer unphysical catastrophic radiative cooling . in such cases a hot bubble is not even created and sn feedback has no effect , whatsoever .
early sn feedback simulations suffered from this problem because of low resolution .
, ke model indeed shows a bubble and a forward shock .
also shown is the density profile for tea model with a lower density ( 5 @xmath16 ) ism ; at later times it shows a bubble which pushes the shell outwards .
the outer shock radius is larger for a lower density ism because @xmath112 . [
fig : fizzling ] ] figure [ fig : fizzling ] shows the density profiles at 3 myr for three of our energy injection models ( ke , ld , tea ) with @xmath59 and the ism density of 20 @xmath16 .
the ejecta radius is chosen to be large such that it violates conditions in eqs .
[ eq : lum_fizzle ] & [ eq : eadd_fizzle ] .
the figure shows a comparison of the luminosity driven ( ld ) and thermal explosion addition ( tea ) models that fizzle out , and kinetic explosion ( ke ) model which shows a hot , dilute bubble .
thus , our results are in agreement with the analytical considerations of section [ sec : cool_cond ] .
the outer shock location for the ke model roughly agrees with the self similar scaling of @xcite if the luminosity is reduced by a factor @xmath113 ; this is comparable to the fraction of mechanical energy retained by sbs after the outer shock becomes radiative ( see section [ sec : energy_budget ] and the right panel of fig .
[ fig : rad_loss_time ] ) . for most runs in figure [ fig : fizzling ] we have chosen a rather high density ( @xmath114 @xmath16 ) compared to the critical values in eqs .
[ eq : lum_fizzle ] & [ eq : eadd_fizzle ] .
for lower densities ( e.g. , 5 @xmath16 for tea model in fig .
[ fig : fizzling ] ) we find that the energy does not couple at early times .
energy injection excites large amplitude sound waves and associated density perturbations , such that at late times the lowest density regions no longer violate eqs .
[ eq : lum_fizzle ] & [ eq : eadd_fizzle ] . after this
, a hot bubble starts to grow because of energy injection in a dilute medium ( see dotted line in fig .
[ fig : fizzling ] ) , and eventually the outer shock radius starts to agree with analytic estimates .
this and later sections , which study the influence of radiative cooling on sbs and snrs , use the realistic kinetic explosion ( ke ) model for supernova energy injection with ejecta radius @xmath103 pc .
however , we have verified that other models discussed in section [ sec : fb_pres ] give similar results , as long as the conditions in section [ sec : anal ] are satisfied .
spherical adiabatic blast waves , both snrs and sbs , have shells with finite thickness .
an estimate for the shell thickness is obtained by assuming that all the swept - up ism mass lies in a shell and that the post shock density is 4 times the ism density for a strong shock ; this gives @xmath115 .
of course , the shock transition layer is unresolved in simulations , and in reality is of order the mean free path .
the structure of an adiabatic blast waves is fairly simple .
the density jump at the shock is 4 for a strong shock , and as the shock becomes weaker the density jump decreases and the shell becomes broader .
eventually , the outer shock is so weak that it no longer compresses gas irreversibly , but instead becomes a sound wave with compressions and rarefactions ( see fig . 2 in @xcite ) .
since the evolution of isolated snrs with cooling has been thoroughly studied in past ( e.g. , @xcite ; hereafter t98 ) , we only highlight the differences between isolated snrs and sbs .
the fundamental difference between the two is that snrs suffer catastrophic losses just after they become radiative because , unlike in sbs , there is no energy injection after this stage . in sbs , the cool ( yet dilute ) , fast sn ejecta periodically thermalizes within the bubble and powers it long after the forward shock becomes radiative .
this keeps the radiative forward shock moving ( like a pressure - driven snowplow ) , as long as sne go off within the hot bubble .
the structure of a radiative shell is quite complex .
the shell become radiative when the cooling time of the post - shock gas is shorter than its expansion time ( which is of order the age of the blast wave ) . moving inward from the upstream ism
, the outer shock transition happens over a mean free path , which is followed by a thin radiative relaxation layer of order the cooling length ( see , for example , pp .
226 - 229 of @xcite and the top - right panel of fig .
[ fig : d_t ] ) .
the radiative relaxation layer is followed by a dense shell , which is separated by a contact discontinuity from the dilute hot bubble . in steady state , radiative cooling
is concentrated at two unresolved boundary layers , the outer radiative relaxation layer and the inner contact discontinuity . here
the density is high and the temperature is conducive to radiative cooling .
figure [ fig : d_t ] shows the zoomed - in density and temperature structure of the radiative shell for a sb ( with @xmath111 ; upper panels ) and a snr ( @xmath116 ; lower panels ) using high resolution ( with 16384 grid - points ) runs .
it clearly shows an outer radiative shock and an inner contact discontinuity . within the contact discontinuity of the sb ( @xmath111 )
is the shocked sn ejecta ; figure [ fig : rho_vs_r_cc85 ] shows the full structure of a superwind within the sb .
just when the outer shock becomes radiative the coolest / densest part is compressed by high pressure regions sandwiching it ( left panels of fig .
[ fig : d_t ] ) . after a sound crossing time
the post - shock region is roughly isobaric and in the pressure - driven snowplow phase ( right panels of fig .
[ fig : d_t ] ) . unlike sbs , for isolated snrs
there is no energy injection at later times ; the pressure in the bubble falls precipitously after the outer shock becomes radiative at @xmath1170.05 myr .
by @xmath118 myr the bubble pressure becomes comparable to the ism pressure , the shell density falls and it becomes momentum conserving with a velocity comparable to the sound speed in the ism . at even later times
( @xmath4 few myr ) the hot bubble just oscillates as a weak acoustic wave
. /[total cooling rate ] ) and bubble ( [ bubble cooling rate]/[total cooling rate ] ) for ke models ( @xmath111 ) with and without conduction ( the run with thermal conduction is discussed in section [ sec : b_cond ] ) .
most radiative energy losses happen at the radiative relaxation layer ahead of the dense shell . at late times , as the outer shock weakens , radiative losses in the bubble become more dominant .
bubble is comparatively more radiative ( in fact , bubble losses exceed shell losses after 5 myr ) with conduction because of mass loading of the bubble by evaporation from the dense shell .
results from the high resolution run and the luminosity driven ( ld ) model are similar .
the minimum in fractional radiative losses corresponds to the time when the outer shock becomes radiative .
[ fig : rad_ratio ] ] figure [ fig : rad_ratio ] shows the distribution of radiative losses in the shell and in the bubble ( shell is defined as the outermost region where density is above 1.01 times the ism density ; bubble comprises of all the grid points with radius smaller than the inner shell radius ) for a superbubble with @xmath111 ; results from runs with and without thermal conduction are shown . here
we only discuss the run without conduction ; the run with conduction is highlighted in section [ sec : b_cond ] .
unlike figure [ fig : d_t ] , here we use our standard resolution runs ( 1024 grid points ) because we are running for a much longer time . results from the higher resolution runs match our standard runs , highlighting the fact that the volume integrated cooling is the same even if the radiative relaxation layer and the contact discontinuity are unresolved .
the time and volume integrated losses ( @xmath119 ) in bubble and shell are sampled appropriately and differentiated in time to obtain their respective cooling rates .
as already discussed , cooling is concentrated at the radiative relaxation layer , which is included in the shell , and at the contact discontinuity , part of which is included in the bubble .
consistent with our previous discussion , most radiative losses are concentrated in the shell .
fractional radiative losses in the bubble ( concentrated at the contact discontinuity ) are @xmath120 , which increase with time as the outer shock becomes weaker .
the structure of a radiative shock is well known ( see , e.g. , pp . 226 - 229 in @xcite ) . applying mass and momentum conservation across the radiative relaxation layer in the shock frame , and assuming the same temperature upstream / downstream of it ( see the top - right panel of fig .
[ fig : d_t ] for different regions of the outer shock ) gives @xmath121 ( eq .
16.36 in @xcite ) , where @xmath122 is the density downstream ( upstream ) of the radiative relaxation layer , @xmath123 ( @xmath124 ) is the upstream ( downstream ) velocity in the shock rest frame , and @xmath55 is the isothermal sound speed of upstream ism ( at @xmath125 k below which radiative cooling vanishes ) .
thus , we expect larger density jump across stronger shocks ( @xmath126 ) .
this is evident from the shell density for the two cases ( @xmath127 ) in figure [ fig : d_t ] .
the thickness of the cold , dense shell can be estimated by equating the swept - up ism mass with the mass in the constant density shell ; @xmath128 .
this thickness is quite small , with @xmath129 for a 100 km s@xmath14 shock .
this estimate agrees with our results in figure [ fig : d_t ] , and as predicted , the shell is thicker for a smaller @xmath30 and becomes thicker with time as the shock becomes weaker . the thickness of the radiative relaxation layer can also be estimated .
the size of the radiative relaxation layer is @xmath130 ( @xmath130 is the distance behind the outer shock after which the advection time becomes longer than the cooling time ) , such that [ eq : rl ] _
0^l_cool = _ 0^t_cool dt = t_cool , where @xmath131 is the velocity in the relaxation layer in the shock rest frame .
while this equation can only be solved after numerically solving for the structure of the relaxation layer , we can make an order of magnitude estimate .
the integral on the lhs of eq .
[ eq : rl ] can be estimated as @xmath132 , where @xmath133 is the geometric mean of the velocity at the front of the relaxation layer ( @xmath134 for a strong shock ) and just downstream of it ( @xmath135 ) .
similarly , the cooling time @xmath136 in eq .
[ eq : rl ] can be estimated by using a geometric mean of densities across the relaxation layer ; i.e. , @xmath137 , where @xmath138 and we can use the peak of the cooling function for @xmath139 and @xmath72 . putting this all together gives , l_cool ~a_t ( ) , which is @xmath140 pc for fiducial numbers , far from being resolved even in our highest resolution runs .
while the transition layers ( contact discontinuity and radiative relaxation layer ) where all our cooling is concentrated are unresolved , we find that the volume integrated quantities such as radiative losses , kinetic / thermal energy in shell / bubble are converged even at our modest ( 1024 grid points ; results are similar even for 256 grid points ) resolution . )
have been combined cumulatively ( see eq .
[ eq : cum ] ) , assuming that sne go off independently in the ism .
pre - radiative phase energetics are similar but isolated snrs are extremely deficient in mechanical energy ( after 1 myr ) as compared to a sb with the same energy input .
the arrow on top right shows the bubble thermal energy at the end for an adiabatic sb run .
isolated sn results are only shown till 2 myr because snrs become weak sound waves by then .
[ fig : energy ] ] in this section we focus on the energetics of sb shell / bubble and compare it with the results from isolated snrs .
we define the shell to be the outermost region where the density is larger than 1.01 times the ism density .
all gas at radii smaller than the shell inner radius is included in the bubble ( this definition is convenient but not very precise as it includes small contribution from the unshocked sn ejecta ) .
figure [ fig : energy ] shows a comparison of kinetic and thermal energies in bubble and shell as a function of time for a sb driven by @xmath141 sne . the bubble kinetic energy
is not included because it is much smaller .
also included is a comparison of the same quantities for the same frequency of sne that go off independently .
the results for multiple isolated sne are obtained by combining the single sn run at different times .
we simply use the data at an interval of @xmath28 ( time between individual sne ) and add them to obtain total kinetic / thermal energy in the shell and bubble at a given time .
for instance , the thermal energy in bubbles of all independent sne at time @xmath142 is obtained by summing up the bubble thermal energy from a single sn ( @xmath116 ) run at @xmath143 , @xmath28 , @xmath144 , ... , till @xmath142 .
this is equivalent to a cumulative sum over time for a single sn run , [ eq : cum ] e_cum(t ) = _
i=0^i < n e(i t_sn ) = _ 0^t
e(t^ ) dt^ , where @xmath145 stands for , say , bubble thermal energy and @xmath146 is the number of sne till time @xmath147 .
@xcite have given analytic predictions for energy in different components of sbs : the total energy of the shell is @xmath148 ( @xmath149 of this is kinetic energy and @xmath150 is thermal ) and the thermal energy of the bubble is @xmath151 ( kinetic energy of the bubble is negligible ) .
these analytic predictions agree well with our numerical results in the early adiabatic ( non - radiative ) sb phase in figure [ fig : energy ] .
figure [ fig : energy ] shows that the sb shell loses most of its thermal energy catastrophically at @xmath152 myr ; the trough in shell thermal energy can be estimated by assuming that all the swept - up mass till then cools to the stable temperature ( @xmath17 k ) .
the thermal energy of the cold shell increases after that as it sweeps up mass from the ism ; this is not a real increase in the thermal energy because the newly added material , which was previously part of the ism , simply becomes a part of the dense shell at the same temperature .
the bubble thermal energy and the shell kinetic energy show only a slight decrease in slope after the radiative phase because they are energized by the non - radiative termination ( internal ) shock(s ) driven by sn ejecta . however , there are some losses because of cooling at the contact discontinuity ( see figs .
[ fig : d_t ] & [ fig : rad_ratio ] ) .
the shell kinetic energy and the bubble thermal energy in radiative sb simulations at 20 myr are roughly half of the values obtained in adiabatic simulations ( which agree with analytic predictions ) .
thus , the mechanical energy retained in the sb is @xmath153 .
this should be contrasted with the energy evolution in isolated snrs .
the isolated snr becomes radiative much earlier ( @xmath117 0.05 myr ; when the shell thermal energy shown by dashed line flattens suddenly in fig .
[ fig : energy ] ) because of a weaker shock compared to a sb .
note that the energies for isolated sne in figure [ fig : energy ] are _ cumulative sums _ over time of a single sn run ( see eq .
[ eq : cum ] ) .
the bubble thermal energy and shell kinetic energy also drop for an individual snr after it becomes radiative ( albeit not catastrophically , unlike the shell thermal energy ; see fig . 3 in t98 ; this is the pressure - driven snowplow stage ) because of cooling at the contact discontinuity and adiabatic losses , and because there is no new energy source ( unlike the termination / internal shocks in a sb ) .
the total mechanical energy in the bubble and shell of a single snr at the beginning of the momentum conserving phase ( 1 myr ; when bubble pressure is comparable to the ism pressure ) is @xmath154 erg , which is only 10% of the input energy ( see fig . 3 in t98 ) .
this agrees with the energy fraction available as mechanical energy of the snr , as quoted by t98 .
after a few myr the snr should be considered a non - energetic part of the ism , as the thermal energy of the swept up ism becomes larger than the snr s mechanical energy , and the bubble becomes a weak acoustic wave . in order to compare isolated snrs and superbubbles over the cluster lifetime
we must extrapolate our cumulative sn energies to 30 myr .
this is also the relevant timescale for preventing large - scale galactic inflows from efficiently forming stars ( the free - fall timescale at @xmath155 kpc for galactic halos is few 10s of myr ) . for isolated snrs the shell kinetic energy
+ bubble thermal energy is @xmath47% of the input energy by 2 myr , and only 0.7% when extrapolated to 30 myr .
we should not extrapolate the shell thermal energy because its rise at late times in figure [ fig : energy ] is due to the sweeping up of the ism into the shell , without an increase in the temperature . to conclude , isolated sn feedback is much weaker ( by a factor @xmath7 ) as compared to the feedback due to sbs over the cluster lifetime .
the left panel of figure [ fig : rad_loss_time ] shows the total radiated energy over the whole computational domain as a function of time for an isolated snr ( @xmath116 ; solid line ) and for sbs ( dashed lines ) .
the results are qualitatively different for sbs ( even for @xmath156 ) and isolated sne .
while an isolated sn radiates almost all of the input energy ( @xmath47 erg ) over a myr timescale , sbs radiate a smaller fraction ( @xmath157 ) of their energy even till late times . the runs with smaller @xmath30 and larger
density become radiative at an early time as the shock is weaker , but sb runs are qualitatively similar . unlike in isolated sne , a significant fraction of the input energy in sbs is retained in the bubble thermal energy and the shell kinetic energy ( @xmath158 ; see fig .
[ fig : energy ] ) .
the key reason for the difference between isolated sne and sbs is that in sbs the non - radiative termination / internal shocks keep the bubble overpressured but isolated sne , which do not have further energy input after the initial explosion , simply fizzle out soon after they become radiative .
the right panel of figure [ fig : rad_loss_time ] shows the fraction of energy retained ( 1-the fraction radiated ) as a function of time for several runs .
all sb runs , including a higher density run ( @xmath159 @xmath16 ) and the run with conduction ( see section [ sec : b_cond ] ) , show the asymptotic fraction of energy retained to be @xmath160 .
in contrast , an isolated sn loses 90% of the input energy by few myr ( and almost all of it by 10 myr ; see also fig .
[ fig : energy ] ) .
radiative losses for an isolated snr at late times ( @xmath161 few myr ) are more than the energy input ( @xmath47 erg ) ; these come at the expense of the thermal energy of the swept - up ism .
we can compare our results of coincident sne with the case of multiple sne distributed over space in a random manner .
the two cases presented in figure [ fig : energy ] represent two extreme limits : spatially coincident sne in a sb and totally independent sne . for spatially distributed sne
we expect results somewhere in between these two extremes .
@xcite compare the total explosion energy that remains as the thermal energy of hot gas in the case of spatially distributed sn explosions .
they study the effects of coherent explosions , as defined in @xcite , which implies that sne overlap before they become radiative .
if the shell radius of a snr when it becomes radiative is @xmath162 and the corresponding time scale is @xmath163 , then for a sn rate density of @xmath164 , the coherency condition is that @xmath165 . @xcite compare the cases in which explosions occur coherently with those in which they do not .
they find that a fraction @xmath166 of the explosion energy is retained in the gas with temperature @xmath167 k if the explosions occur coherently , and the fraction is @xmath168 if the explosions are incoherent .
our results here for sbs correspond to the coherent case , since @xmath28 is always shorter than the cooling time of the gas in the bubble .
therefore , our result of a fraction @xmath169 being retained as sb s mechanical energy is consistent with @xcite .
since the ism is magnetized , we try to assess the qualitative effects of magnetic fields using an idealized high resolution ( 16384 grid points ) mhd simulation .
we assume an azimuthal ( @xmath170 ) component of the magnetic field so that only magnetic pressure forces ( and no tension ) are present .
we choose a plasma @xmath171 ( ratio of gas pressure and magnetic pressure ) of unity in the ism , and our sn ejecta is also magnetized with the same value of @xmath171 .
since the ejecta is dominated by kinetic energy and the bubble is expanding , we do not expect magnetic fields to affect the bubble and the ejecta structure .
however , the radiative shell is compressed because of cooling , and due to flux - freezing magnetic pressure is expected to build up in the dense shell .
this is indeed what we find in our simulations with magnetic fields .
the left panel of figure [ fig : dt_b_cond ] shows the zoomed - in density and temperature structure of the radiative outer shock with and without magnetic fields .
the key difference between the hydro and mhd runs is that the dense shell in mhd has a lower density and is much broader .
this is because magnetic pressure prevents the collapse of the dense shell .
the dense shell ( @xmath172 in the left panel of fig .
[ fig : dt_b_cond ] ) is magnetically dominated with plasma @xmath173 .
the mhd run has two contact discontinuities ; one at the boundary of the hot bubble ( @xmath174 pc ) , and another at @xmath175 pc left ( right ) of which the plasma is dominated by thermal ( magnetic ) pressure .
another important physical effect , especially in the hot bubble is thermal conduction .
we carry out a 1024 resolution hydro run with thermal conduction to study its qualitative influence . however , it is difficult to determine the ism conductivity in a magnetized ( presumably turbulent ) plasma
. therefore , we use the spitzer value with a suppression factor of 0.2 ( see eq . 11 in @xcite ) . moreover , since the bubble can become very hot such that the diffusion approximation breaks down , we limit the conductivity to an estimate of the free streaming diffusivity ( chosen to be 2.6 @xmath176 where @xmath177 is the local isothermal sound speed and @xmath92 is the radius ) .
thermal conduction is operator split , and implemented fully implicitly through a tridiagonal solver using the code s hydro time step .
conduction is expected to evaporate matter from the dense shell and deposit it into a conductive layer in the bubble ; in steady state the rate of conductive transport of energy from bubble to the shell is balanced by the rate of heat advection from shell to the bubble @xcite .
the outer and termination shock locations are not affected much by conduction .
however , the density and temperature structure in the hot bubble is affected significantly . without conduction
the bubble is very hot ( @xmath178 k ) , but with conduction the temperature drops into the x - ray range ( @xmath179 k ) and density is higher .
this can enhance the x - ray emissivity of sbs ; a rough estimate of hard x - ray luminosity ( @xmath180 over the hot bubble ) at 10 myr from the right panel of figure [ fig : dt_b_cond ] is few @xmath58 erg s@xmath14 ( which is @xmath181 the energy put in by sne by that time ) .
since galactic superwinds are copious x - ray emitters , we expect thermal conduction to be a very important ingredient for explaining observations .
figure [ fig : rad_ratio ] confirms that the fraction of radiative losses from the bubble is much higher with conduction than without conduction because of a higher density .
however , the right panel of figure [ fig : rad_loss_time ] shows that the total fractional radiative losses with thermal conduction are only slightly higher compared to the non - conductive @xmath111 sb run .
we emphasize that our treatment of magnetic fields and thermal conduction is extremely simplified .
realistic calculations must be done in three dimensions with tangled magnetic fields and with anisotropic thermal conduction along fields lines .
however , we expect the qualitative effects of realistic magnetic fields and thermal conduction to have some semblance with our simplified treatment .
we have obtained several important results in this paper , on both numerical implementation of sn / sb feedback and on differences between isolated sne and sbs .
sbs are a result of spatially and temporally correlated sne .
since most massive stars are expected to be born in star clusters few 10s of pc in extent ( e.g. , @xcite ) , pre - sn stellar winds and first sne are expected to carve out a low density bubble , which by a fraction of myr encloses the whole star cluster ( see fig . [
fig : d_t ] ) .
therefore , subsequent sne happen in the low density bubble and we are in the sb regime of coherent sne ( see @xcite ) .
magnificent galactic outflows , such as m82 , are powered by multiple super - star - clusters and the problem of understanding coalescing sbs is important .
star clusters more massive than @xmath182 ( and hence with @xmath183 sne ) are rare ( e.g. , see @xcite ) ; therefore , the superbubbles in m82 and in our @xmath111 sb model should be considered as giant bubbles driven by 100s of overlapping superbubbles due to individual star clusters . indeed ,
100s of star clusters have been observed in the central few 100 pc of m82 ( @xcite ) .
we note that vertical stratification is important for the acceleration and assimilation of the metal - rich bubble into the halo . in this paper
we consider the idealized smaller - scale problem of the behavior of isolated snrs and multiple coincident sne within a sb in a uniform ism .
some of our most important results are : * our most realistic kinetic explosion ( ke ) models ( and other models in which sn energy is _ overwritten _ ) , in which the sn energy in kinetic form is overwritten in a small volume , give correct results only if the energy is deposited within a small length scale ( see section [ sec : over ] ) ; otherwise , energy is overwritten without coupling to the ism .
this is true even without considering any radiative losses . * with cooling , if feedback energy is deposited within a length scale @xmath24 larger than the critical values mentioned in section [ sec : cool_cond ] ,
such that the input energy is radiated before it is thermalized , a hot bubble is not formed in the widely used luminosity driven and thermal explosion addition ( ld , tea ) models .
thus sn fizzles out at early stages due to artificially large cooling losses . * with insufficient resolution and large
ism densities the bubble fizzles out completely in the luminosity driven ( ld ) and thermal explosion addition ( tea ) models ( in which energy is _ added _ to the ism ; see section [ sec : fb_pres ] ) and can not have any effect on the ism ( see fig .
[ fig : fizzling ] ) . as also pointed out previously ( e.g. , by @xcite ) , early galactic - scale sn feedback simulations failed mainly because of this .
however , for a realistic sn ( as mimicked by our ke model ) a bubble is formed and subsequent sne occuring within the non - radiative bubble power the radiative outer shock .
another , probably more serious , problem faced by numerical simulations is that the sn energy is not typically put in coherently over a small volume in space and within a short interval .
feedback due to sne in young star clusters is expected to be coherent and much more effective than a similar number of isolated sne ( see figs .
[ fig : energy ] , [ fig : rad_loss_time ] ) . for correlated sne
, isolated sn bubbles overlap to form a superbubble which is powered by the non - radiative termination / internal shocks , long after the outer shock becomes radiative .
in contrast , an isolated supernova becomes powerless a bit after ( @xmath184 myr ) the outer shock becomes radiative . *
a smooth cc85 wind within the superbubble is possible only if the number of sne ( @xmath30 ) over the cluster lifetime is large ( i.e. , @xmath185 ) . only in these cases , individual sne going off inside the superbubble are able to thermalize within the termination shock .
this result has implications for modeling the x - ray output , for example , in individual bubbles blown by star clusters and in the inner regions of galactic outflows , since the cc85 wind structure is often assumed where it may not be valid .
* most of the radiative losses come from the unresolved radiative relaxation layer at the outer shock .
the fractional radiative losses from the interior region , concentrated at the contact discontinuity between the shocked ism and the shocked ejecta , varies between @xmath186 , with larger losses occurring at later times .
while these radiative layers are unresolved even in our highest resolution simulations , the volume integrated radiative losses in them converge even for a modest resolution . * as compared to isolated sne , superbubbles can retain a larger fraction of the initial energy of explosions as thermal / kinetic energy of the gas .
isolated sne are mixed with the ism soon after they become radiative ; by few myr they are incapable of affecting the ism at all . while most energy is radiative away ( close to 100% , and not @xmath187% , as is often assumed ) for isolated sne over 10 myr , a sb can retain a fraction @xmath169 ( for @xmath15 @xmath16 ) as the bubble thermal energy + the shell kinetic energy .
this fraction is only weakly affected by a higher ism density and by thermal conduction ( see the right panel of fig . [
fig : rad_loss_time ] ) .
thus , sbs are expected to significantly affect even a dense ism .
substantial radiative losses can partly explain the smaller observed bubble sizes compared to what is expected by modeling the stellar populations ( see @xcite for a summary ) .
* the temperature profile of sbs strongly depend on thermal conduction , whose inclusion can decrease ( increase ) the temperature ( density ) and thereby enhance the x - ray luminosity . thermal conduction ( and other sources of mass loading of the hot bubble , such as turbulent mixing ) plays an important role in explaining the x - ray emission from galactic superbubbles because very little gas is expected to be in the x - ray emitting regime ( @xmath188 k ) in its absence ( see the right panel of fig .
[ fig : dt_b_cond ] ) .
our simple one - dimensional simulations show that isolated supernova remnants , owing to large radiative losses , are much weaker feedback agents compared to superbubbles driven by coherently overlapping supernovae .
however , detailed three - dimensional calculations , particularly with a realistic distribution of stars in a cluster , and magnetic fields and thermal conduction , are required in order to make quantitative comparisons with observations .
this will be done in future .
ps thanks ramesh narayan for useful discussions .
ps is partly supported by dst - india grant no .
sr / s2/hep-048/2012 .
ys is partly supported by rfbr ( project code 12 - 02 - 00917 ) .
agertz , o. , teyssier , r. , & moore , b. 2011 , mnras , 410 , 1391 chevalier , r. 1974 , mnras , 169 , 229 chevalier , r. 1977 , ann .
astr . , 15 , 175 chevalier , r. a. & clegg , a. w. 1985 , nature , 317 , 44 chevalier , r. a. & blondin , j. m. 1995 , apj , 444 , 312 cox , d. p. 1972 ,
apj , 178 , 159 creasey p. , theuns , t. , bower , r. g. , & lacey , c. g. 2011 , mnras , 415 , 3706 creasey , p. , theuns , t. , & bower , r. g. 2013 , mnras , 429 , 1922 cooper , j. l. , bicknell , g. v. , sutherland , r. s. , & bland - hawthorn , j. 2008 , apj , 674 , 157 dalla vecchia , c. & schaye , j. 2012 , mnras , 426 , 140 dubois , y. , teyssier , r. 2008 , a&a , 477 , 79 gerritsen , j. p. e. 1997 , phd thesis , university of groningen gnedin , n. y. 1998 , mnras , 297 , 407 guedes , j. , callegari , s. , madau , p. , & mayer , l. 2011 , 742 , 76 hayes , j. c. et al .
2006 , apjs , 165 , 188 hopkins , p. f. , quataert , e. , & murray , n. 2011 , mnras , 417 , 950 hopkins , p. f. , quataert , e. , & murray , n. 2012 , mnras , 421 , 3522 joung , m. k. r. & mac low , m. 2006 , apj , 653 , 1266 katz , n. 1992 , 391 , 502 kay , s. t. , pearce , f. r. , frenk , c. s. , & jenkins , a. 2002 , mnras , 330 , 113 larsen , s. s. 1999 , a&a supp .
ser . , 139 , 393 leisawitz , d. , bash , f. n. , & thaddeus , p. 1989
, apjs , 70 , 731 mac low , m. , mccray , r. , & norman , m. l. 1989 , apj , 337 , 141 mac low , m. & ferrara , a. 1999 , apj , 513 , 142 mccray , r. & kafatos , m. 1987 , apj , 317 , 190 nath , b. b. & shchekinov , y. 2013 , apjl , 777 , 12 navarro , f. f. & white , s. d. m. 1993 , mnras , 265 , 271 oconnell , r. w. , gallagher , j. s. , hunter , d. a. , & colley , w. n. 1995 , apj , 446 , l1 oey , m. s. 2009 , aip conf .
proc . , 1156 , 295 ( 2009 ) palous , j. , wnsch , r. , martnez - gonzlez , s. , hueyotl - zahuantitla , f. , silich , s. , & tenorio - tagle , g. 2013 , apj , 772 , 128 portegies zwart , s. f. , mcmillan , s. l. , & gieles , m. 2010 , ann .
astr . , 2010 , 48 , 431 recchi , s. & hensler , g. 2013 , a&a , 551 , 41 roy , a. , nath , b. b. , sharma , p. , & shchekinov , y. 2013 , mnras , 434 , 3572 sharma , p. , parrish , i. j. , & quataert , e. 2010 , apj , 720 , 652 shu , f. h. , the physics of astrophysics , vol .
ii , gas dynamics , university science books , 1992 silich , s. , tenorio - tagle , g. , torres - campos , a. , munz - tun , monreal - ibero , a. , & melo , v. 2009 , apj , 700 , 931 springel , v. & hernquist , l. 2003 , mnras , 339 , 289 springel , v. et al .
2005 , nature , 435 , 629 stinson , g. , seth , a. , katz , n. , wadsley , j. , governato , f. , & quinn , t. 2006 , mnras , 373 , 1074 stone , j. m. & norman , m. l. 1992 , apjs , 80 , 753 strickland , d. k. & stevens , i. r. 2000 , mnras , 314 , 511 strickland , d. k. & heckman , t. m. 2009 , apj , 697 , 2030 suchkov , a. a. , balsara , d. s. , heckman , t. m. , & leitherer , c. 1994 , apj , 430 , 511 sutherland , r. s. & dopita , m. a. 1993 , apjs , 88 , 253 tang , s. & wang , q. d. 2005 , apj , 628 , 205 tasker , e. j. & bryan , g. l. 2006 , apj , 641 , 878 thacker , r. j. & couchman , h. m. p. 2000 , apj , 545 , 728 thornton , k. , gaudlitz , m. , janka , h. , & steinmetz , m. 1998 , apj , 500 , 95 truelove , j. k. & mckee , c. f. 1999 , apjs , 120 , 299 tumlinson , j. , thom , c. , werk , j. k , prochaska , j. x. , tripp , t. m. , weinberg , d. h. , peeples , m. s. , omeara , j. m. , oppenheimer , b. d. , meiring , j. d. , katz , n. s. , dav , r. , ford , a. b. , & sembach , k. r. 2011 , science , 2011 , 334 , 948 vasiliev , e. o. , nath , b. b. , bondarev , r. , & shchekinov , y. 2014 , arxiv:1401.5070 weaver , r. , mccray , r. , castor , j. , shapiro , p. , & moore , r. 1977 , apj , 218 , 377 zhang , q. , fall , s. m. , & whitmore , b. c. 2001 , apj , 561 , 727 | using idealized 1-d eulerian hydrodynamic simulations , we contrast the behavior of isolated supernovae with the superbubbles driven by multiple , collocated supernovae . continuous energy injection via successive supernovae going off within the hot / dilute bubble maintains a strong termination shock .
this strong shock keeps the superbubble over - pressured and drives the outer shock well after it becomes radiative .
isolated supernovae , in contrast , with no further energy injection , become radiative quite early ( @xmath0 myr , 10s of pc ) , and stall at scales @xmath1 pc .
we show that isolated supernovae lose almost all of their mechanical energy by a myr , but superbubbles can retain up to @xmath2 of the input energy in form of mechanical energy over the lifetime of the star cluster ( few 10s of myr ) .
these conclusions hold even in the presence of realistic magnetic fields and thermal conduction .
we also compare various recipes for implementing supernova feedback in numerical simulations . for various feedback prescriptions
we derive the spatial scale below which the energy needs to be deposited for it to couple to the interstellar medium ( ism ) .
we show that a steady thermal wind within the superbubble appears only for a large number ( @xmath3 ) of supernovae . for smaller clusters
we expect multiple internal shocks instead of a smooth , dense thermalized wind .
[ firstpage ] galaxies : ism ism : bubbles shock waves
supernova remnants |
the anomalous magnetic moment of the muon has served over many years as an important test of the standard model , see the reviews @xcite .
it is also sensitive to potential contributions from new physics .
the current status of the muon @xmath0 is summarized in table [ tab : status_g-2 ] where we list the different contributions in theory ( qed , weak , hadronic ) from various recent sources and compare with the experimental value .
more references to earlier work can be found in the quoted papers and in refs .
the experimental world average is dominated by the final result of the brookhaven muon @xmath0 experiment @xcite , corrected for a small shift in the ratio of the magnetic moments of the muon and the proton @xcite .
we observe a difference between experiment and theory of more than three standard deviations @xmath2 unfortunately , the theoretical uncertainties @xcite from hadronic vacuum polarization ( hvp ) and hadronic light - by - light scattering ( hlbl ) make it difficult to interpret this discrepancy as a clear sign of new physics .
most recent evaluations @xcite , which differ slightly in the treatment of the hadronic contributions , obtain deviations of @xmath3 . in ref .
@xcite the hadronic cross - section data below @xmath4 was fitted to a ( broken ) hidden local symmetry ( hls ) model and a discrepancy in the muon @xmath0 between @xmath5 was observed , depending on the selected data .
.standard model contributions to @xmath6 and comparison of theory and experiment . [ cols="<,>,>,^",options="header " , ] a selection of estimates for hlbl is presented in table [ tab : hlbl_summary ] .
note that the refs .
@xcite are the only full calculations of hlbl to date , using , as much as possible , one model for all the contributions ( hls model in ref .
@xcite , extended nambu - jona - lasinio ( enjl ) model in ref .
both calculations showed that the exchanges of the lightest pseudoscalar states , @xmath7 , dominate numerically , which can be understood from the large-@xmath8 counting .
the contributions from the ( dressed ) pion - loop and the ( dressed ) quark - loop are subdominant , but not negligible , and they happen to largely cancel each other numerically .
the final results for the total hlbl contribution were rather close in both models . in ref .
@xcite an ansatz for the pion - photon transition form factor with a minimal number of narrow vector resonances in large-@xmath8 qcd ( lowest meson dominance ( lmd , lmd+v ) ) was matched to short - distance constraints from the ope .
the reevaluation of the pion - pole contribution to hlbl in ref .
@xcite with the ansatz from ref .
@xcite then revealed a sign error in the earlier calculations @xcite .
furthermore , for the neutral pion contribution , a 2-dimensional integral representation was derived in ref .
@xcite for a certain class of vector meson dominance ( vmd)-like form factors . in the end , one integrates over the lengths of the two independent loop - momenta @xmath9 in euclidean space .
schematically , @xmath10 with universal weight functions @xmath11 and where the ( model ) dependence on the form factors resides in the functions @xmath12 , see ref .
@xcite for details .
note that the expressions in ref .
@xcite with on - shell form factors are in general not valid as they stand . as pointed out in ref .
@xcite , one needs to set the form factor at the external vertex to a constant to obtain the pion-_pole _ contribution .
but the expressions are valid for the constant wess - zumino - witten ( wzw ) form factor and the off - shell vmd form factor .
the weight functions @xmath11 are plotted in figure [ fig:2d_weightfunctions ] .
one can see that the relevant momentum regions are about @xmath13 , see also the numerical analysis in ref .
as long as the form factors in different models lead to a damping , we expect comparable results for @xmath14 at the level of 20% , which is indeed the case , see the papers quoted in table [ tab : hlbl_summary ] and ref .
@xcite . for the general case with off - shell form factors ,
a 3-dimensional integral representation was derived in ref .
@xcite , where the integrations run over @xmath15 and @xmath16 , where @xmath17 . in eq .
( [ 2d_integral ] ) from ref .
@xcite.,scaledwidth=80.0% ] later , ref .
@xcite derived new short - distance constraints from the four - point function on the pion - pole and axial - vector - pole contributions , which do not allow for any form factors at the external vertex .
reference @xcite also included the mixing of two axial - vector nonets and studied the pion - loop within the hls model in more detail .
all this lead to a substantial enhancement of these contributions to hlbl .
more recently , a new short - distance constraint on the off - shell form factor at the external vertex in pion - exchange was derived in ref .
@xcite , which yielded , again in the framework of the lowest meson dominance approximation to large-@xmath8 qcd , a value for this contribution about half - way between the results in refs .
@xcite and those in ref .
note that the compilations @xcite and ref .
@xcite are largely based on the full calculations @xcite , with revised or newly calculated values for some of the contributions .
more recent estimates , mostly for the pseudoscalar contribution , can be found in ref .
while most of these evaluations agree at the level of 15% , if one takes the extreme values , there is a spread of @xmath18 . until 2010
, a consensus had been reached about the central value @xmath19 , but there was a discussion on how to estimate the error , more progressively , @xmath20 , in ref .
@xcite and more conservatively , @xmath21 , in refs .
@xcite . in view of the precision goal of future @xmath0 experiments at fermilab and j - parc @xcite with @xmath22 and the continued progress in improving the error in hvp
, the hlbl contribution might soon be the main uncertainty in the theory prediction , if it can not be brought under better control @xcite . in the last few years
, several works have appeared which yield much larger ( absolute ) values for some of the contributions , see table [ tab : hlbl_summary ] . in ref .
@xcite the quark - loop was studied using a dyson - schwinger equation approach .
in contrast to refs .
@xcite , no damping compared to the bare constituent quark - loop result was seen , when a dressing was included .
note that this calculation of the quark - loop is not yet complete and that earlier , very large results for the quark - loop seem to have been affected by some errors in the numerics in certain parts .
the large size of the quark - loop contribution in ref .
@xcite was questioned in the papers @xcite , using different quark - models and approaches , see also the ballpark prediction for hlbl in ref .
the pion - loop contribution was analyzed in ref .
the authors stressed the importance of the pion - polarizability effect and the role of the axial - vector resonance @xmath23 , which are not included in the models used in refs .
@xcite . depending on the value of the pion - polarizability and the model for the @xmath23 resonance
used , a large variation was seen .
the issue was taken up in ref .
@xcite where different models for the pion - loop were studied .
the inclusion of the @xmath23 resonance was attempted , but no finite result for @xmath0 could be achieved . with a cutoff of @xmath4 , a result close to the earlier estimate in ref .
@xcite was obtained .
reference @xcite also pointed out that the very small ( absolute ) value for the pion - loop in ref .
@xcite could be due to the fact that the hls model used in ref .
@xcite has a wrong high - energy behavior and that there is some cancellation between positive and negative contributions in the pion - loop in hlbl .
concerning the future , maybe lattice qcd will provide a reliable calculation of hlbl at some point , see ref .
@xcite for some promising recent results . in the meantime
, only a close collaboration between theory and experiment can lead to a better controlled estimate for hlbl . on the theory side ,
the hadronic models can be improved by short - distance constraints from perturbative qcd to have a better matching at high momenta .
one can also use dispersion relations to connect the theory with experimental data , e.g. in @xmath24 @xcite .
also the issue about whether the dressing of the bare constituent quark - loop leads to a suppression or an enhancement needs to be studied further .
this problem is also related to the question whether there is any double counting involved . on the experimental side ,
the information on various processes ( decays , form factors , cross - sections ) of hadrons interacting with photons at low and intermediate momenta , @xmath25 , can help to constrain the models .
important experiments which should be pursued include more precise measurements of the ( transition ) form factors of light pseudoscalars with possibly two off - shell photons in the process @xmath26 and the two - photon decay width and the ( double ) dalitz decays of these mesons .
this could further reduce the error of the dominant pseudoscalar exchange contribution @xcite . concerning the pion - loop contribution , in addition to studying @xmath27 , measurements of the pion - polarizability in various processes , e.g. in radiative pion decay @xmath28 , in radiative pion photoproduction @xmath29 or with the hadronic primakoff effect @xmath30 or @xmath31 ( with some nucleus @xmath32 ) , can help to improve the models @xcite . for the development of models with the axial - vector resonance @xmath23 and estimates of the sizable axial - vector contribution , information about the decays @xmath33 would be useful as well .
finally , to extract the needed quantities from experiment will also require the development of dedicated monte carlo programs for the relevant processes @xcite .
if the recent results for the quark - loop and pion - loop are taken at face value , one obtains the range @xmath34 .
while the new approaches raise some important issues and point to potential shortcomings in the previously used models , these calculations are also still preliminary and further studies are needed .
therefore , the estimate @xmath35 from refs .
@xcite still seems to give a fair description of the current situation .
i would like to thank the organizers of the _ international workshop on @xmath36 collisions from @xmath37 to @xmath38 2013 ( phipsi13 ) _ , of the _ working group on radiative corrections and monte carlo generators for low energies _ and of the _
lc13 workshop : exploring qcd from the infrared regime to heavy flavour scales at b - factories , the lhc and a linear collider _ for the opportunity to present reviews on the hlbl in the muon @xmath0 .
generous support from the organizers , the infn , the ect@xmath39 , trento , italy , and the heinrich - greinacher - stiftung , physics institute , university of bern , switzerland , is gratefully acknowledged .
the aec is supported by the `` innovations- und kooperationsprojekt c-13 '' of the `` schweizerische universittskonferenz suk / crus . ''
m. davier _ et al .
. phys .
j. c * 71 * , 1515 ( 2011 ) [ erratum - ibid .
c * 72 * , 1874 ( 2012 ) ] ; k. hagiwara _ et al .
_ , j. phys .
g * 38 * , 085003 ( 2011 ) ; t. teubner , talks at phipsi13 , rome and at working group on radiative corrections and monte carlo generators for low energies , frascati , september 2013 .
m. hayakawa , t. kinoshita , a.i .
sanda , phys .
lett . * 75 * , 790 ( 1995 ) ; phys .
d * 54 * , 3137 ( 1996 ) ; m. hayakawa , t. kinoshita , phys .
d * 57 * , 465 ( 1998 ) [ erratum - ibid .
d * 66 * , 019902 ( 2002 ) ] .
j. prades , e. de rafael , a. vainshtein , in _
advanced series on directions in high energy physics - vol .
20 `` lepton dipole moments '' _ , edited by b.l roberts , w.j .
marciano ( world scientific , singapore , 2010 ) , p.303 ; and arxiv:0901.0306 .
m. zahiri abyaneh , arxiv:1208.2554 ; j. bijnens , m. zahiri abyaneh , epj web conf . * 37 * , 01007 ( 2012 ) [ arxiv:1208.3548 ] ; j. bijnens , talk at mesonnet 2013 international workshop , prague , june 2013 [ see : arxiv:1308.2575 ] .
dorokhov , w. broniowski , phys . rev .
d * 78 * , 073011 ( 2008 ) ; a.e .
dorokhov , a.e .
radzhabov , a.s .
zhevlakov , eur .
j. c * 71 * , 1702 ( 2011 ) ; * 72 * , 2227 ( 2012 ) ; a.e .
radzhabov , talk at phipsi13 , rome , september 2013 ; d.k .
hong , d. kim , phys .
b * 680 * , 480 ( 2009 ) ; l. cappiello , o. cata , g. dambrosio , phys .
d * 83 * , 093006 ( 2011 ) ; k. kampf , j. novotny , phys . rev .
d * 84 * , 014036 ( 2011 ) ; p. masjuan , phys .
d * 86 * , 094021 ( 2012 ) ; r. escribano , p. masjuan , p. sanchez - puertas , arxiv:1307.2061 . | we give an update on the status of the hadronic light - by - light scattering contribution to the muon @xmath0 .
we review recent work by various groups , list some of the open problems and give an outlook on how to better control the uncertainty of this contribution .
this is necessary in order to fully profit from planned future muon @xmath0 experiments to test the standard model .
despite some recent developments , we think that the estimate @xmath1 still gives a fair description of the current situation .
december 17 , 2013 + andreas nyffeler + _ regional centre for accelerator - based particle physics , harish - chandra research institute , chhatnag road , jhusi , allahabad - 211 019 , india _ + _ albert einstein center for fundamental physics , institute for theoretical physics , university of bern , sidlerstrasse 5 , ch-3012 bern , switzerland _
pacs numbers : 14.60.ef , 13.40.em , 12.38.lg |
a ramanujan graph is a @xmath0-regular finite graph with nontrivial eigenvalues @xmath1 satisfying @xmath2 this condition is equivalent to the `` graph theoretical riemann hypothesis '' for the ihara zeta function associated to the graph , see @xcite for more information .
the above equivalence was first formulated by sunada in @xcite .
infinite families of @xmath0-regular ramanujan graphs of increasing size are examples of the more general families of expander graphs .
in fact , the families of ramanujan graphs are the expander families that are optimal from the spectral point of view , according to the bounds by alon and boppana . furthermore , the known constructions of families of ramanujan graphs are essentially number theoretical ; for instance , the proof of the ramanujan property ( [ eqn : ramprop ] ) for the original families is related to the ramanujan conjecture for modular forms of weight 2 .
the families of expanders have applications to engineering and computer science , among others areas @xcite .
the original construction of families of ramanujan graphs was presented by lubotzky , phillips and sarnak in @xcite _ _ * * _ _ , and independently by margulis in @xcite .
the former construction consists of cayley graphs on projective linear groups over a finite field .
the use of cayley graphs allows to utilize the properties of the underlying group and the generating set to determine structural and spectral properties of the graph .
for example , the irreducible representations of the underlying group are directly related to the eigenvalues of the graph .
recently , there has been work on generalizing the notion of group determinant as a function of a group - subgroup pair using the representation theory of @xmath3-determinants @xcite .
the resulting wreath determinant for group - subgroup pairs shares some of the properties of the group determinant , but there is a nontrivial relation with the chosen ordering of the group and subgroup elements .
some results on factorization for this determinant can be found on @xcite , it turns out to be closely related to the representation of the symmetric groups associated with the given group and subgroup and their wreath product .
it was suggested to the author by masato wakayama that one could employ the same strategy to extend existing constructions , or to conceive new constructions , of ramanujan graphs and expanders based on groups , in particular , by generalizing cayley graphs . in this paper
, we apply these ideas to introduce an extension of cayley graphs for a group - subgroup pair .
the motivation is twofold , one is to have further tools to solve problems , in particular the construction of ramanujan graphs ; and to continue the work on this incipient group - subgroup study philosophy . the main purpose of this paper is to introduce the concept of group - subgroup pair graph based on these considerations and show its elementary properties . in particular , the degree of the vertices of the pair graph is determined by the coset structure of the subgroup and the chosen generating set .
the organization of the paper is as follows . in section 2 ,
the definition , examples and basic properties are presented .
namely , the determination of the degrees of the vertices of the graph , and the conditions for the graph to be regular . in section 3 ,
conditions for connectedness of the graph are completely determined , along with the number and properties of connected components for disconnected graphs .
additionally , sufficient conditions for the graphs to be bipartite and the notion of trivial eigenvalues are introduced for the group - subgroup pair graphs . in section 4
we limit the discussion to the regular graph case and assert a symmetry relation between the spectrum for different choices of generating set for fixed group and subgroup , this result is applied to give a sufficient condition on the size of the generating set that guarantees that the resulting graphs are ramanujan graphs .
we provide two examples of ramanujan graphs obtained by using this condition .
the common theme throughout this paper is to compare and relate the results on group - subgroup pair graphs with results on cayley graphs .
in this section we introduce the group - subgroup pair graph , a cayley - type graph construction for group - subgroup pairs ; relate the definition with the group matrix and establish some elementary properties of the graphs .
first , we introduce the conventions and notation used in this paper .
all groups are assumed to be finite unless otherwise stated and @xmath4 always represents the identity of a given group g. for a subgroup @xmath5 of @xmath6 , all cosets are assumed to be right cosets and we say that @xmath7 is incongruent to @xmath8 modulo @xmath5 when @xmath9 .
a subset @xmath10 is said to be symmetric if @xmath11 . for a given group @xmath6 and symmetric subset @xmath12
we denote the corresponding cayley graph by @xmath13 .
the characteristic function of a subset @xmath14 is denoted @xmath15 and the notation @xmath16 $ ] for @xmath17 is used for the set @xmath18 .
[ def]let @xmath6 be a group , @xmath5 a subgroup and @xmath19 a subset such that @xmath20 is a symmetric subset of @xmath6 .
then the * group - subgroup pair graph * @xmath21 is the graphwith vertex set g and edges @xmath22 equivalently , the group - subgroup pair graph @xmath21 can be defined as @xmath23 where @xmath24 , @xmath25 and @xmath26 is the generalized edge sum operator , as defined in @xcite .
the graph @xmath27 is a cayley graph and the graph @xmath28 is the graph with vertices @xmath6 and edges @xmath29 according to the alternative definition , the group - subgroup pair graph @xmath21 can be thought informally as the undirected graph consisting of an inner cayley graph of @xmath5 determined by @xmath30 and an outer graph that connects vertices of @xmath5 with vertices of @xmath31 .
the notation @xmath32 and @xmath30 is used throughout this paper with the same meaning as in the previous definition . when @xmath33 , the generating set is @xmath34 and the corresponding graphs is a cayley graph , in other words , @xmath35 .
this justifies the claim that the group - subgroup pair graphs are a generalization of the cayley graphs and the seemingly artificial condition of symmetry of @xmath36 .
when @xmath12 is empty , we call the resulting pair graph @xmath21 or cayley graph @xmath13 trivial .
[ example1]let @xmath37 , @xmath38 , and @xmath39 , then @xmath40 and @xmath41 .
the corresponding pair - graph @xmath21 can be seen on figure [ graph1 ] .
[ finitefield]let @xmath42 , and the prime field @xmath43 of @xmath44 considered as a subgroup of the additive group @xmath44 , then @xmath44 is the direct sum of seven copies of @xmath5 .
let @xmath45 be the norm map of @xmath44 as a field extension of @xmath5 , then if @xmath46 , we obtain the following pair - graph @xmath47 this graph consists of vertices of degree 2,4 and 16 .
[ random]matrix groups over finite fields have been used in the construction of families of ramanujan graphs . set @xmath48 , where @xmath49 is the finite field of @xmath50 elements and @xmath51 , then for a random subset @xmath12 of 7 elements taken from the complement of @xmath5 in @xmath6 , we obtain the following pair - graph @xmath21 .
this connected graph has 480 vertices with vertices of degrees @xmath52 and @xmath53 .
note that none of the graphs of the preceding examples can be constructed as cayley graphs .
we briefly mention another generalization of cayley graphs for group @xmath6 and subgroup @xmath5 , the _ schreier coset graph_. for a symmetric subset @xmath12 of @xmath6 , the schreier coset graph is defined as the graph with the set @xmath54 of coset as vertices and where two cosets @xmath55 and @xmath56 are adjacent when there is an @xmath57 such that @xmath58 a schreier coset graph can have multiples edges and loops ( even when @xmath59 ) .
this kind of graphs have been used for coset enumeration techniques .
a detailed exposition can be found in @xcite . for a given group @xmath6 , the schreier coset graph is a cayley graph when @xmath60 , whereas the group - subgroup pair graph is a cayley graph when @xmath61 .
as stated in the introduction , one of the motivation for the group - subgroup pair graph comes from the extension of the group determinant for a group - subgroup pair , called _ wreath determinant for group - subgroup pair_s . in this subsection
we show how one can relate the adjacency matrix of a cayley graph with the group matrix of the corresponding group ; then , by extending the idea for the matrix used for the wreath determinant for group - subgroup pairs we obtain the rows corresponding to the subgroup on the adjacency matrix of a certain group - subgroup pair graph , which is enough to determine the complete adjacency graph . for a group @xmath62 , consider a polynomial ring @xmath63 containing the indeterminates @xmath64 , for @xmath65 , then the _ group matrix _ is a matrix @xmath66 in @xmath67 defined by @xmath68 for @xmath69 $ ] and where @xmath70 $ ] is an enumeration function for @xmath6 , used implicitly .
the determinant @xmath71 of the group matrix is called _ group determinant _ of @xmath6 and does not depend on the chosen enumeration of the elements of @xmath6 .
for @xmath72 we have @xmath73 , therefore for any element @xmath74 , the corresponding transpose element is @xmath75 .
similarly , for a group @xmath6 of order @xmath76 , and subgroup @xmath5 of order @xmath77 one can construct the matrix @xmath78 by @xmath79 for @xmath80 , @xmath81 , @xmath82 , j \in [ n k ] $ ] and where @xmath83 $ ] and @xmath84 $ ] are enumerations functions for @xmath6 and @xmath5 .
considering only the columns corresponding to elements of @xmath5 of the matrix @xmath85 one obtains the group matrix of @xmath5 with respect to the ordering @xmath86 . for a matrix @xmath87
, the wreath determinant of @xmath88 is defined as @xmath89 } ) , \ ] ] where @xmath90}$ ] is the row @xmath0-flexing of the matrix @xmath88 and @xmath91is the @xmath3-determinant , for an extensive exposition of the wreath determinant and its properties the reader is referred to @xcite . in the paper @xcite ,
the authors define the wreath determinant for the pair @xmath6 and @xmath5 by @xmath92 in contrast with the ordinary group determinant , this wreath determinant for @xmath6 and @xmath5 depends on the enumeration functions @xmath93 and @xmath86 . for a given group @xmath6 and symmetric subset @xmath12 , by evaluating the corresponding group matrix @xmath66 by the rule @xmath94 for @xmath57 and @xmath95 for @xmath96 one obtains a symmetric matrix .
since @xmath97 implies @xmath98 , the corresponding matrix is the adjacency matrix of the cayley graph @xmath13 .
consider @xmath99 , the symmetric group on three letters with the ordering @xmath93 given by @xmath100 , the group matrix is @xmath101 where @xmath102 stands for @xmath64 .
considering @xmath103 and evaluating in the way mentioned before we get @xmath104 which can be verified to be the adjacency matrix of the cayley graph @xmath105 .
likewise , for group @xmath6 , subgroup @xmath5 and subset @xmath12 as in definition [ def ] , by evaluating the group - subgroup matrix with the rule @xmath94 for @xmath57 and @xmath95 otherwise , we obtain a matrix with nonzero entries @xmath106 when @xmath107 . in other words ,
there are ones in the matrix exactly when @xmath108 , which is the relation for the edges of the group - subgroup pair graph @xmath21 in definition @xmath109 .
the resulting matrix corresponds to the rows associated with the elements of @xmath5 in the adjacency matrix of the pair - graph @xmath110 and can be completed by symmetry to obtain the complete adjacency matrix .
let @xmath37 , @xmath38 , and @xmath111 as in example [ example1 ] , the corresponding matrix with respect to natural orderings @xmath93 and @xmath86 is @xmath112 then , evaluating as describe before @xmath113 which can be verified to correspond to the rows of the vertices of @xmath5 of the adjacency matrix of the pair - graph @xmath114 of example [ example1 ] .
note that the group matrix can also be defined as @xmath115 . using this definition
the resulting cayley graph is defined by left multiplication , the same is true for the group - subgroup matrix for the wreath determinant and the group - subgroup pair graph .
an isolated vertex is one that is not connected to any other vertices .
in contrast with nontrivial cayley graphs , group - subgroup pair graphs may contain isolated vertices even when the generating subset is non empty .
the following result characterizes the presence of isolated vertices in group - subgroup pair graphs .
[ isolated]i ) the pair - graph @xmath21 contains no isolated vertices if and only if @xmath12 contains a representative for each coset of @xmath5 on @xmath6 different from @xmath116 .
\ii ) the vertices @xmath5 are isolated in @xmath21 if and only if @xmath12 is the empty set .
suppose @xmath12 is not empty and contains a representative for each coset , then take @xmath117 , and @xmath118 the representative of @xmath55 , then there is @xmath119 such that @xmath120 , and therefore @xmath121 is connected to @xmath122 .
conversely , if there are no isolated vertices , by the definition we must have @xmath123 .
the second statement follows directly from the definition .
consider any group @xmath6 of order @xmath77 and @xmath124 , then the graph @xmath21 with @xmath125 has no isolated vertices .
in fact , @xmath21 is a @xmath126 _ star graph_. the second part of the proposition shows the relation with cayley graphs , as a cayley graph is empty if and only if @xmath12 is the empty set .
a graph in which all the vertices have the same degree is called a _ regular graph _ , more precisely , if the vertices have degree @xmath0 , the graph is called a @xmath0-_regular graph . _ an important property of a cayley graph @xmath13 is that it is @xmath127-regular .
example [ example1 ] shows that this is not true in general for @xmath110 , but there is still uniformity on the degree of the vertices in each coset .
[ coset]in a pair - graph @xmath21 , all the vertices in the same coset have the same degree .
[ cor1]namely , the vertices in @xmath5 have degree @xmath127 and for @xmath128 the degree of the vertices in the coset @xmath55 is @xmath129 .
it is clear by the definition of the outer graph @xmath130 that any vertices @xmath131 in the same coset @xmath55 have the same degree @xmath132 .
the vertices in @xmath5 have degree @xmath127 by construction .
returning to example [ example1 ] , @xmath133 , @xmath134 , and the cardinality of these sets corresponds to the degree of the vertex in the respective cosets .
we call those types of graphs , _ multi - regular _ or more precisely @xmath135-regular graphs , where @xmath136 is the degree of the vertices on a given partition .
the preceding discussion suggests that the structure of @xmath54 , the set of cosets of @xmath5 on g , is closely related to the structure of the graph .
let @xmath6 be a group and @xmath5 a subgroup of index @xmath137 = k+1 $ ] .
for a subset @xmath138 with @xmath30 symmetric , consider the pair - graph @xmath21 .
if @xmath139 is a set of representatives of the cosets of h incongruent to @xmath4 , then for @xmath140 , @xmath141 with equality only when @xmath142 .
in particular , a nontrivial @xmath21 is regular if and only if @xmath40 and @xmath137 = 2 $ ] , or @xmath137 = 1 $ ] .
the inequality follows from proposition [ cor1 ] and the fact that @xmath143 , and the equality happens only when @xmath144 .
the _ if _ part of the proof follows directly from the inequality and the definitions . for the _ only if _
part , consider a @xmath145-regular @xmath21 graph , then by the preceding proposition @xmath146 and @xmath147 for @xmath148 .
it follows from the inequality above that @xmath149 and therefore @xmath0 is necessarily @xmath150 or @xmath151 .
if @xmath137 = 2 $ ] , then @xmath152 , so @xmath153 and the case @xmath137 = 1 $ ] gives a cayley graph .
note that in view of proposition [ coset ] , we can consider @xmath154 as the degree of any of the elements of the coset . in that case
, the above identity can be written as @xmath155 with equality happening only when @xmath30 is empty .
in this section we consider the connectedness for pair graphs @xmath156 . recall that a cayley graph @xmath13 is connected if and only if @xmath157 .
we begin by considering the case @xmath158 empty , in other words , none of the vertices of @xmath5 are adjacent in @xmath21 .
[ conectbip]let @xmath159 be a group , h a subgroup and @xmath160 subset with @xmath30 symmetric . if @xmath142 then in the pair - graph @xmath21 the vertices of @xmath5 are in the same connected component if and only if @xmath161 = h. if @xmath162 , then it suffices to prove that the identity @xmath4 is connected to an arbitrary @xmath140 . for @xmath140 , we have @xmath163 with @xmath164 , then if we set @xmath165 , @xmath166 is adjacent to @xmath167 and @xmath122 is adjacent to @xmath168 so @xmath166 is connected to @xmath122 . by repeating this process
we conclude that @xmath4 is connected to @xmath122 . on the other hand ,
if in the graph @xmath21 , all the vertices of @xmath5 are in the same connected component , @xmath140 is connected to @xmath169 .
since there are no direct connections between two elements of @xmath5 or @xmath170 , there must be path from @xmath4 to @xmath122 where every even vertex is is an element of @xmath5 , so we have a sequence @xmath171 of elements of @xmath5 , where @xmath172 and @xmath173 is connected to @xmath102 @xmath174 with @xmath175 .
that is , we have a sequence of edges @xmath176 , as shown in the following diagram . to @xmath177 .
] then , for @xmath178 @xmath179 thus , @xmath180 it follows that @xmath181 and @xmath182 .
note that since a group - subgroup pair graph may contain isolated vertices , the condition of the lemma alone is not sufficient for connectedness .
[ connectbip]with the same notation as before , if @xmath142 then the pair - graph @xmath21 is connected if and only if @xmath183 and s contains representatives of all the cosets of @xmath184different from @xmath5 .
the result follows from part follows from lemma [ conectbip ] , proposition [ isolated ] and the observation that any vertex @xmath185 must be connected to some @xmath140 which is in turn connected to the identity @xmath186 .
[ connectness]a pair - graph @xmath187is connected if and only if @xmath188 and @xmath12 contains representatives of all the cosets of @xmath5 different from the coset @xmath5 .
first we see that the vertices of @xmath5 are in the same connected component if and only if @xmath189=h . the proof of this fact is the same as that of lemma [ conectbip ] while considering that in the path from a @xmath190 to @xmath140 there may be edges connecting elements @xmath191 from @xmath5 , in such case we have @xmath192 , with @xmath193 .
then the result follows like in proposition [ connectbip ] .
[ example2]let @xmath194 , @xmath195 .
set @xmath196 and @xmath197 , the corresponding group - subgroup pair graphs are the following .
note that @xmath198 and @xmath199 , so neither graph is connected , as can be seen in the diagrams . on the other hand ,
as there are no elements of the coset @xmath200 on @xmath201 , all the vertices of that coset are isolated on @xmath202 . if a graph is not connected , the characterization of the connected components of the graph is desirable . for cayley graphs ,
the connected component of the identity is the subgroup @xmath203 , and each of the cosets in @xmath6 are the connected components of the graph .
this identification is not identical for group - subgroup pair graphs as the connected component of the identity may include vertices from @xmath170 . in particular , the subgroup @xmath204 of @xmath5 contains the elements of @xmath5 that are in the connected component of the identity and the cosets of this subgroup are the intersection of @xmath5 with certain connected components of the graph .
[ comps]with the same notation as before , let @xmath205 , then the identity component @xmath206 of @xmath207 consists of the vertices @xmath208 .
the remaining connected components of the pair - graph @xmath21 are either of the type @xmath209 for @xmath140 or the type @xmath210 for @xmath211 .
the first statement follows from the preceding discussion and the definition of the pair - graph @xmath21 .
any path @xmath212 from the identity @xmath4 to @xmath213 corresponds uniquely to a path @xmath214 from @xmath122 to @xmath215 so the connected component of @xmath140 is @xmath209 . for @xmath185
if @xmath121 is an isolated vertex its connected component is @xmath210 , otherwise it is connected to an @xmath140 so its connected component is of the type @xmath216 .
a consequence of the above proposition is that an arbitrary connected component @xmath217 of @xmath21 has cardinality equal to @xmath218 or 1 .
moreover , for first case , we also have @xmath219 . for cayley graphs , the number of connected components of the graph is the index @xmath220 $ ] .
the existence of isolated vertices even for non empty generating sets makes the situation for pair - graphs slightly more complicated .
[ components]the number of connected components of @xmath221 is @xmath222 { } { } + \vert g - h \vert - \left\vert \bigcup_{s \in s_{o } } h s \right\vert .\ ] ] by the preceding proposition , the first term in the formula is the number of connected components @xmath216 that occur on @xmath5 , the second and third terms count the number of isolated points in @xmath170 , by proposition [ isolated ] .
since there are not connections between elements of @xmath170 , this is the number of connected components of the graph . proposition [ comps ] and theorem [ components ]
completely characterize the connected components for the pair - graphs @xmath21 for given group @xmath6 , subgroup h and valid subset @xmath138 . for the pair - graph of example [ example1 ]
we have @xmath39 , since @xmath223 , @xmath5 is generated by @xmath224 and the first term is 1 , the second term is 8 and since all the cosets are represented the last term is 8 , and we get 1 connected component .
for the pair - graphs generated by @xmath201 and @xmath225 of example [ example2 ] , in both cases the first term is 2 , the next term is @xmath226 , and the final term is @xmath227 for the graph generated by @xmath201 and @xmath226 for the graph generated by @xmath225 , therefore @xmath228 has 6 connected components and @xmath229 has 2 connected components as confirmed visually in the diagrams .
a graph is _
vertex transitive _ when for any pair of different vertices @xmath121 and @xmath230 there is graph automorphism @xmath45 such that @xmath231 .
cayley graphs are naturally vertex transitive by means of left translations @xmath232 with @xmath233 .
any vertex transitive graph must be regular , therefore by proposition [ coset ] , we have the following result .
nontrivial pair - graphs @xmath21 are not vertex transitive when @xmath137 \geqslant 3 $ ] . however , we still see restricted transitivity when considering left translations by elements of @xmath5 on @xmath21 . [ leftaction]the left action of h in @xmath21 is a graph automorphism . moreover , for any @xmath234 , the action is transitive .
the same holds for any coset @xmath5x , @xmath235 . for @xmath140 and @xmath57 ,
the edge @xmath236 is in @xmath237 .
now , for @xmath238 , the images of @xmath122 , and @xmath239 under the left action @xmath240 are @xmath241 and @xmath242 , which are adjacent in @xmath243 .
the map is a bijection , and has inverse @xmath244 so it is a graph automorphism . for @xmath234 , consider @xmath245 , then @xmath246 as required . for @xmath247 ,
take @xmath238 such that @xmath248 .
note that the left action of arbitrary @xmath233 is not necessarily a graph automorphism .
for instance , in example [ graph1 ] the action of @xmath249 is not a graph endomorphism , as the image of @xmath150 is @xmath151 , and @xmath250 , but @xmath251 .
let @xmath252 be the set of complex valued functions defined on @xmath6 , and consider @xmath253 the left regular representation of @xmath5 given by the action @xmath254 for h@xmath255 and @xmath235 .
recall that for any graph with vertices on @xmath6 we can associate the adjacency operator @xmath88 , acting on @xmath252 in the following way , @xmath256 [ leftreg]let @xmath88 be the adjacency matrix for a group - subgroup pair graph @xmath21 , then for any @xmath140 @xmath257 note that for the pair - graph @xmath21 and @xmath258 , the adjacency operator is given by @xmath259 the result then follows from direct calculation by considering that for any @xmath235 , @xmath121 and @xmath260 are in the same coset , in other words , @xmath261 for any @xmath140 .
for @xmath140 , @xmath262 is a permutation matrix for the elements of @xmath6 corresponding to the left multiplication by @xmath122 . by theorem 15.2 of @xcite ,
a bijection @xmath45 of the vertices of a graph is a graph automorphism if @xmath263 , where @xmath264 is the permutation matrix associated with @xmath45 ; therefore proposition [ leftreg ] is only a representation theoretic restatement of proposition [ leftaction ] .
the trivial eigenvalue of a @xmath0-regular graph is
@xmath265 and it corresponds to any constant eigenfunction @xmath266 on the vertices of the graph . by extension ,
the trivial eigenvalue of a cayley graph @xmath13 is @xmath265 . in this section
we extend this notion to the group - subgroup pair graphs .
[ trivialeigen]let @xmath6 be a group , @xmath5 a subgroup of @xmath6 of index @xmath267 = k+1 $ ] with @xmath268 , and @xmath138 a nonempty subset with @xmath30 symmetric and @xmath269 .
consider @xmath270 a set of representatives of the cosets of @xmath5 in @xmath6 and set @xmath271 , for @xmath272 $ ]
. then @xmath273 are eigenvalues of the graph @xmath21 .
a corresponding eigenfunction is defined by @xmath274 .
\end{cases}\ ] ] note that the numbers @xmath275 satisfy @xmath276 then , for @xmath140 , we have @xmath277 similarly , for @xmath278 , @xmath279 , @xmath280 note that for the case @xmath281 , @xmath282 is an eigenvalue with corresponding eigenvector @xmath283 as defined in the above theorem , but @xmath284 so it is not an eigenfunction . with the same notation as before ,
the eigenvalue @xmath282 is the largest eigenvalue of the graph @xmath21 with multiplicity @xmath285 $ ] .
first , consider the case of a connected pair - graph @xmath21 , in particular @xmath286 for all @xmath272 $ ] and the eigenfunction @xmath283 takes only positive values . by thm .
8.1.4 and cor .
8.1.5 of @xcite , any eigenfunction that only takes nonzero values of the same sign corresponds to the largest eigenvalue , which has multiplicity 1 .
the proposed eigenfunction satisfies this condition , therefore corresponds to the largest eigenvalue and it is an eigenvalue of multiplicity one , so the statement of the proposition follows . for the remaining case , let @xmath140 and consider the connected component @xmath216 as a subgraph of @xmath21 and note that @xmath287 is an eigenfunction of @xmath216 with eigenvalue @xmath282 .
then by the same argument as before , @xmath282 is the largest eigenvalue of @xmath216 with multiplicity 1 .
now , it is well known that the characteristic polynomial @xmath288 of the graph @xmath289 is the product of the characteristic polynomials of its connected components , @xmath290 where @xmath291 is the number of connected components of @xmath21 containing elements of @xmath5 and @xmath292 the number of isolated vertices . moreover , since @xmath275 is the largest root of @xmath293 for each @xmath80 , then is the largest root of @xmath294 and therefore the largest eigenvalue of @xmath21 .
furthermore , since @xmath282 is a simple eigenvalue for each of the subgraphs @xmath295 , then by theorem [ components ] @xmath282 is an eigenvalue of @xmath21 with multiplicity equal to @xmath285 $ ] . from the decomposition of the characteristic polynomial of the adjacency matrix , we can obtain a bound for the multiplicity of the eigenvalue @xmath296 . with the same notation as before ,
@xmath297 is an eigenvalue of multiplicity at least @xmath298 of the nontrivial pair - graph @xmath21 .
first , suppose that the pair - graph @xmath21 contains isolated vertices , then from the decomposition of the characteristic polynomial of the adjacency matrix we have that @xmath296 is an eigenvalue with multiplicity at least the number of isolated components , by [ components ] this number is @xmath299 .
for a general pair - graph @xmath21 , consider an eigenfunction @xmath266 associated with the eigenvalue @xmath300 with @xmath301 for @xmath140 , then @xmath266 must satisfy the @xmath302 linear equations @xmath303 the matrix @xmath304 corresponding to this system is a @xmath305 matrix , then from elementary linear algebra it holds that @xmath302 is an upper bound for the rank of @xmath304 .
therefore the kernel of @xmath306 has dimension at least @xmath307 , so the multiplicity of the eigenvalue @xmath296 is at least @xmath307 .
the result follows from considering the two cases at the same time .
the pair - graph in example [ finitefield ] , has @xmath308 , with four cosents of degree 2 and two cosets of degree 4 , therefore the trivial eigenvalues are @xmath309 .
the pair - graph in example [ random ] has @xmath310 , with two cosets of degree 2 and one coset of degree 3 , then @xmath311 are the corresponding trivial eigenvalues .
the pair - graph in example [ bipartite ] has @xmath312 , @xmath313 , therefore the trivial eigenvalues are @xmath314 and @xmath315 .
a graph is _ bipartite _ when there is a bipartition @xmath316 of the vertices such that any pair of vertices in the same subset are not adjacent . for a group @xmath6 and symmetric subset @xmath12 , if there is an homomorphism @xmath317 , such that @xmath318 then the cayley graph @xmath13 is bipartite , this condition is also necessary when the graph is connected . if a group homomorphism @xmath317 , such that @xmath319 exists , then the graph @xmath21 is bipartite .
the converse is true if @xmath21 is connected and @xmath12 is symmetric .
the proof of this proposition is the same as the one for cayley graphs , see for example chapter 4 of @xcite .
note that if such a homomorphism exists then @xmath320 , so the elements of @xmath12 and @xmath321 would be on the same partition .
[ bipartite]let @xmath322 , the alternating group of 4 letters , @xmath5 , the klein four group embedded as a subgroup of @xmath6 , and @xmath323 , @xmath324 , @xmath325 , @xmath326 , @xmath327 , @xmath328 , using cycle notation .
observe that @xmath329 , the inverse of @xmath330 , is not contained in @xmath12 .
the resulting @xmath21 is a bipartite graph .
any homomorphism @xmath317 with @xmath331 , would have @xmath332 , but in this case @xmath333 , therefore there are no homomorphisms that satisfy the conditions . for the case @xmath334 , the sets @xmath5 and @xmath170 are a bipartition of @xmath21 , so the nontrivial pair - graphs @xmath110 with with @xmath334 are bipartite .
an example can be seen in figure [ graph1 ] .
from the results of section @xmath335 , a group - subgroup pair graph @xmath156 is regular when @xmath61 , or when @xmath5 is a subgroup or order 2 and @xmath30 is empty . in this section
we restrict to the latter case , where the resulting pair - graph is also bipartite .
[ bipcayley]let @xmath6 , @xmath5 and @xmath12 be as described above . if @xmath12 is a symmetric set , the resulting @xmath21 is a cayley graph .
namely , @xmath336 .
the conditions imply that @xmath41 , then by the alternative definition of the pair - graph the edges are given by @xmath337 since s is symmetric one can simply write @xmath338 , @xmath339 , @xmath340 , which is the definition of cayley graph .
proposition [ bipcayley ] states another way in which the graph @xmath341 can be seen as a generalization of certain class of bipartite cayley graphs . as an example , for @xmath342,@xmath343 prime numbers , when @xmath342 is not a square modulo @xmath343 , the @xmath344 ramanujan graphs from lubotzky , phillips and sarnak @xcite can be identified with pair - graphs @xmath345 . in this section and the next one
we fix a group @xmath6 and subgroup @xmath5 of order @xmath77 and index 2 .
let @xmath346 denote the power set of @xmath170 . in this section
we consider transformations on the generating sets that result in isomorphic group - subgroup pair graphs .
[ rigact]let @xmath347 , if @xmath348 , where @xmath349 is the right translation by @xmath238 , then @xmath350 .
for @xmath238 consider the map @xmath351 , defined as @xmath352 for @xmath117 and @xmath353 for @xmath140 .
the map is clearly bijective , then for @xmath354 , we have @xmath355 with @xmath356 , therefore is an graph homomorphism with inverse @xmath357 and the graphs @xmath228 and @xmath229 are isomorphic .
let @xmath347 , if @xmath358 , where @xmath359 is a group automorphism of g , then the pairs graphs @xmath228 and @xmath360 are isomorphic . since @xmath137 = 2 $ ] , @xmath5 is invariant under @xmath359 , then for any @xmath361 , @xmath362 with @xmath363 , therefore @xmath359 is a graph isomorphism with inverse @xmath364 .
any orbit on @xmath346 of right actions @xmath365 by elements of @xmath5 and group isomorphisms @xmath359 of @xmath6 consists of sets that generate isomorphic graphs . as mentioned in the beginning of the section , under the current assumptions the pair - graphs @xmath21 are regular bipartite .
the spectrum of these graphs is symmetric about 0 and its largest eigenvalue is the trivial eigenvalue @xmath366 .
moreover , any eigenvalue @xmath1 satisfies latexmath:[$| \mu pair - graph @xmath21 is given by @xmath368 if @xmath369 and @xmath370 , then the adjacency operator of @xmath371 is the sum of the adjacency operators of @xmath228 and @xmath229
. follows from the definition @xmath372 for a graph @xmath373 , the complement graph is @xmath374 , where @xmath375 , if and only if @xmath376 for @xmath377 .
the eigenvector and eigenvalues of a @xmath0-regular graph and its complement can be related , see for example lemma 8.5.1 of @xcite .
we present a relation between the of eigenvalues and eigenvectors for pair - graphs @xmath228 and @xmath378 with @xmath369 and @xmath370 .
first , assume that @xmath379 , @xmath380 , then any constant function @xmath266 is an eigenfunction of both of the pair - graphs @xmath381 and @xmath229 corresponding to @xmath382 and @xmath383 respectively .
similarly , for @xmath384 , the function @xmath385 is an eigenfunction of @xmath228 and @xmath229 corresponding to @xmath386 and @xmath387 .
[ symmetry]let @xmath201 and @xmath225 as before with @xmath388 for numbers @xmath389 , then \(i ) if @xmath228 is not connected , then there are independent eigenfunctions @xmath266 and @xmath390 of @xmath228 associated to @xmath391 such that @xmath392 is an eigenfunction of of @xmath229 corresponding to the eigenvalue @xmath393 .
\(ii ) if @xmath266 is an eigenfunction of @xmath228 associated with an eigenvalue @xmath394 , then @xmath266 is an eigenfunction of @xmath229 corresponding to the eigenvalue @xmath393 .
denote the adjacency graph operator of graphs @xmath395 , @xmath229 , @xmath371 as @xmath396 respectively , so that @xmath397 . _
( i ) _ for a connected components @xmath217 of @xmath202 , consider the function @xmath398 or @xmath399 , which can be verified to be eigenfunctions corresponding to @xmath265 and @xmath400 , respectively .
we can similarly define @xmath390 with respect to a different connected component @xmath401 of @xmath228 .
note that there are no isolated vertices on the graph , therefore by proposition [ comps ] , all the connected components have the same cardinality , in particular @xmath402 and @xmath403 .
then , we have @xmath404 it follows that @xmath405 or @xmath406 depending on the whether @xmath265 or @xmath400 . by the preceding discussion
, we have @xmath407 and @xmath408 _ ( iii ) _ for an eigenfunction @xmath266 of @xmath88 corresponding to an eigenvalue @xmath409 , we have @xmath410 therefore it suffices to prove @xmath411 . in other words , it suffices to prove the two equalities @xmath412 @xmath413 first , not that for fixed @xmath414 we have @xmath415 then , by summing over all elements of @xmath201 , @xmath416 since @xmath266 is an eigenfunction associated with @xmath1 , we have @xmath417 therefore we have @xmath418 similarly , by fixing @xmath414 and using the equation @xmath419 we obtain the relation @xmath420 finally , combining ( [ eqn : eqn1 ] ) and ( [ eqn : eqn2 ] ) , we obtain the equations @xmath421 @xmath422 since @xmath423 , we conclude @xmath424 @xmath425 and the proof is complete .
recall that when @xmath228 is not connected , by theorem [ trivialeigen ] the eigenvalue @xmath265 has multiplicity equal to the number @xmath426 of connected components of @xmath228 , then by @xmath427 the graph @xmath229 also has the eigenvalue @xmath428 with multiplicity @xmath429 .
we can reformulate the result as follows .
[ symm]for group @xmath6 and subgroup @xmath5 of index @xmath335 and order @xmath77 .
let @xmath430 with @xmath431 and @xmath432 the spectrum of the pair - graph @xmath21 .
then there is a @xmath433-regular pair - graph @xmath434 with spectrum @xmath435 such that @xmath436 for @xmath437 . for @xmath438 , @xmath439 , and @xmath440 , @xmath441 , @xmath442 , @xmath50 , @xmath443 , @xmath444 , @xmath445 , @xmath446
we have the @xmath228 and @xmath229 pair - graphs shown in figure [ fig : corrgraphs ] .
table [ table1 ] contains the values of the positive eigenvalues for both of the graphs .
since both graphs are bipartite the remaining eigenvalues correspond to the negatives of the ones shown .
.the positive values of the spectrum of the previous graphs . [ cols="^,^",options="header " , ] note that @xmath447 , but @xmath448 is such that @xmath449 by proposition [ rigact ] , and @xmath450 .
recall that a ramanujan graph is a connected @xmath0-regular graph such that for every eigenvalue @xmath1 different from @xmath451 , one has @xmath452 [ eigenbound]with the same assumptions as before .
if the pair graph @xmath21 is connected and @xmath453 then it is a bipartite ramanujan graph . due to theorem [ symmetry ] any @xmath0-regular pair - graph @xmath237 has nontrivial eigenvalues @xmath1 satisfying @xmath454 . also , the pair - graph @xmath21 is a ramanujan graph when said trivial eigenvalues satisfy @xmath455 .
considering the two inequalities , it follows that all @xmath0-regular pair - graphs @xmath21 with @xmath456 or @xmath457 are ramanujan graphs .
let @xmath458 , @xmath459 .
the set @xmath460 , @xmath461 , @xmath462 , @xmath463 , @xmath464 , @xmath465 , @xmath466 , @xmath467 is such that @xmath468 satisfy the bound of the corollary , so the corresponding @xmath21 graph is ramanujan .
its spectrum consist of @xmath469 with multiplicity 1 , @xmath470 with multiplicity 2 and 0 with multiplicity 18 .
note that it contains the eigenvalues @xmath470 and as expected by theorem [ symmetry ] , the corresponding @xmath227-regular graph , generated by @xmath471 , @xmath472 , @xmath473 , @xmath474 has 3 connected components .
both of the pair - graphs are shown in figure [ fig : ramgcor ] .
let @xmath475 and @xmath476
. then @xmath477 and @xmath478 , in this case by corollary [ eigenbound ] , for any subset @xmath12 with @xmath479 , the resulting pair - graph @xmath21 graph is ramanujan . for a random choice of @xmath12 we obtain the @xmath445-regular ramanujan pair graph @xmath21 shown in figure [ fig : ramgff ] . in this case , the pair - graph @xmath480 of figure [ ramgraph2 ] , generated by the complement of @xmath12 in @xmath170 , is a @xmath53-regular ramanujan graph .
this shows that the result of corollary [ eigenbound ] is not a necessary condition , as mentioned in the above remark .
as mentioned in the introduction , ramanujan graphs are graphs with good connectivity ( as measured by the isoperimetric constant of the graph ) .
since the addition of edges to a graph results in a graph with greater or equal connectivity , it is expected for graphs with large amount of vertices to have good connectivity .
this can be confirmed in the above examples , as the ramanujan pair - graphs @xmath21 resulting from using the bound provided by corollary @xmath481 , for group @xmath6 and subgroup @xmath5 of order @xmath482 , have generating set @xmath12 satisfying @xmath483 .
then , corollary @xmath481 can also be stated as follows : the set of graphs @xmath484 consists only of ramanujan graphs when @xmath485 , where @xmath5 is a subgroup of order @xmath77 and @xmath137 = 2 $ ] .
[ diagramsummary ] the general properties of group - subgroup pair graphs according to the choice of @xmath486 and @xmath487 are shown in diagram [ diagramsummary ] . for arbitrary pair - graph @xmath21 a set of trivial eigenvalues can be determined including the largest eigenvalue of the graph . according to the index of the subgroup
the pair - graph can be regular ( index 1 or 2 ) or never be regular . if the generating set is contained in complement of the subgroup , the pair - graph is bipartite or multipartite .
the pair - graphs are cayley graphs in two cases : when the subgroup and the group are equal and when the subgroup is of order 2 and the graphs is regular bipartite with symmetric generating set .
the symmetry on the generating set also makes easier to verify if the given pair - graph is bipartite .
these observations suggest that further research on the matter may be started with graphs with symmetric subsets before considering the general case .
the author was supported by the japanese government ( monbukagakusho : mext ) scholarship .
the author would like to express his gratitude to professor masato wakayama for providing the motivation for the present work and for the invaluable discussions and support .
the author would also like to thank professors yoshinori yamasaki , miki hirano and kazufumi kimoto for their valuable comments and discussion , regarding the definition of the group - subgroup pair graph during the event `` zeta functions in okinawa 2013 '' ; and mr .
hiroto inoue for his valuable suggestions on the statements and proofs of some of the results .
g. a. margulis .
explicit group - theoritical constructions of combinatorial schemes and their applications to the design of expanders and concentrator .
_ journal problems of information transmission _ , 24,1:5160 , 1988 . | in this paper we introduce a cayley - type graph for group - subgroup pairs and present some elementary properties of such graphs , including connectedness , their degree and partition structure , and vertex - transitivity .
we relate these properties to those of the underlying group - subgroup pair . from the properties of the group , subgroup and generating set some of the eigenvalues can be determined , including the largest eigenvalue of the graph .
in particular , when this construction results in a bipartite regular graph we show a sufficient condition on the size of the generating sets that results on ramanujan graphs for a fixed group - subgroup pair
. examples of ramanujan pair - graphs that do not satisfy this condition are also provided , to show that the condition is not necessary . |
charm is predicted to be produced primarily in gluon fusion reactions in the early stages of a collision @xcite .
if charm is indeed produced in initial hard ( high @xmath3 ) processes , one will observe that charm production scales with the number of binary ( nucleon - nucleon ) collisions .
the charm production cross - section can be calculated by summing up feynman diagrams at the `` fixed - order plus next - to - leading - log ( fonll ) '' level .
the total charm cross - section per binary collision as predicted by fonll for @xmath1 = 200 gev is @xmath4 @xcite .
both the star and phenix experiments at rhic observe binary scaling when comparing p+p to au+au ( phenix ) and d+au to au+au ( star ) ; however , previously measured charm cross - sections in star and phenix differ above declared uncertainties @xcite .
also , current fonll calculations under - predict measured star charm cross - section values @xcite .
this may suggest that different physical processes are at play than are assumed by fonll calculations .
this analysis uses data taken with the time projection chamber ( tpc ) at the solenoidal tracker at rhic ( star ) , one of the experiments at the relativistic heavy ion collider ( rhic ) at brookhaven national lab ( bnl ) .
the tpc measures an energy loss per unit distance , de / dx , by particles ionization of the p10 gas in the detector s chamber .
the momentum is measured through the particles curvature in a 0.5 t magnetic field @xcite . by using a bichsel parameterization , a particle identification ( pid )
is acquired .
the @xmath5 invariant mass spectrum is reconstructed via the @xmath2 decay channel . the @xmath6 ( or @xmath7 )
invariant mass is given by @xmath8 where @xmath9 is the angle between @xmath10 and @xmath11 .
the invariant mass spectrum was extracted from 28.7 million cu+cu minimum bias events at @xmath12 gev .
then the uncorrelated background was subtracted via the rotational background subtraction method . to implement this method one of the daughter tracks in a two body decay is rotated in momentum space to every 5 degrees from 150 to 210 degrees in the plane transverse to the beam line .
after the background is subtracted a peak with a statistical significance of @xmath13 ( @xmath14 , s = area of signal peak , b = background under peak ) is seen ( see figure [ d0]a ) . @xmath15 % \cr .
\mbox{\bf ( 1 ) } & \mbox{\bf ( 2 ) } \cr \mbox{\bf } & \mbox{\bf } % \mbox{\bf ( a ) } & \mbox{\bf ( b ) } \end{array}$ ]
figure [ d0]b shows a transverse momentum spectrum for @xmath16 at mid - rapidity . efficiency and acceptance corrections , as calculated from a detailed geant simulation of the star detector , and a branching ratio correction of @xmath17 @xcite were applied .
an @xmath18 exponential fit is done to the spectrum ( @xmath19 ) . by integrating over the entire @xmath20 spectra
, we find that @xmath21 = @xmath22 .
the @xmath21 yield is converted to a cross - section via @xmath23 here , @xmath24 is the charm cross - section per binary collision , @xmath25 mb is the proton+proton inelastic cross - section @xcite , @xmath26 is the average number of binary nucleon collisions in a cu+cu minimum bias collision found using the glauber model , @xmath27 is a normalization factor to extrapolate to full rapidity ( based on pythia simulation ) @xcite , and @xmath28 is the ratio of @xmath6 to @xmath29 , measured in @xmath30 collider events @xcite . the preliminary cu+cu charm cross - section is thus calculated to be @xmath31 mb and is shown in figure [ cucu_cross_section ] .
evaluation of systematic error bars for the cu+cu measurement is in progress .
this new measurement confirms previous star measurements @xcite showing that charm production follows a binary scaling from p+p to central au+au collisions .
star measured @xmath5 in cu+cu collisions with a statistical significance of @xmath13 . the charm cross - section in @xmath1
= 200 gev minimum bias cu+cu collisions was measured to be @xmath31 mb .
this is consistent with binary scaling when compared to other star measurements , verifying charm production via hard processes in the initial stages of the collision .
z. lin and m. gyulassy , phys .
c * 51 * 4 2177 ( 1995 ) .
m. cacciari , p. nason , and r. vogt , phys .
lett . * 95 * 122001 ( 2005 ) .
s.s . adler _
et al . _ ( phenix ) , phys .
lett . * 94 * 082301 ( 2005 ) .
et al . _ ( phenix ) , phys .
lett . * 97 * 252002 ( 2005 ) .
j. adams _ et al . _
( star ) , phys .
lett . * 94 * 062301 ( 2005 ) .
y. zhang ( star ) , e - print arxiv : nucl - ex/0607011 .
m. anderson _
( star tpc group ) , nucl .
instr . and meth . a * 499 * 659 ( 2003 ) .
w - m yao _ et al .
_ ( particle data group ) , j. phys .
g : nucl . part .
phys . * 33 * 1 ( 2006 ) . c. adler _
et al . _ ( star ) , phys .
91 * 172302 ( 2003 ) . | we present the mid - rapidity @xmath0 yield measured in @xmath1 = 200 gev cu+cu collisions at rhic via direct reconstruction through the @xmath2 decay channel .
a charm cross - section is reported and compared to theoretical predictions and to previous rhic experimental results .
open charm , cu+cu , quark - gluon plasma , fonll , star 12.38.mh , 21.65.qr |
until recently , only supernova remnants ( snrs ) in our galaxy and the magellanic clouds were resolved at x - ray wavelength . however , studies of galactic snrs can be limited by the lack of reliable distance estimates , high interstellar absorption at optical and x - ray wavelengths , and the high level of confusion in many galactic fields ( magnier et al .
these limitations are overcome by studies of snrs in nearby extragalactic systems such as the large and small magellanic clouds ( see , e.g. , hughes 2001 ) .
more recently , kong et al . ( 2002 , 2003 ) and williams et al . (
2004 ) discovered five x - ray resolved snrs in the nearest ( 780 kpc ) spiral galaxy , m31 .
more importantly , these snrs were also seen at optical and radio wavelengths .
such observations provide a rare opportunity to perform multi - wavelength morphological studies of snrs beyond the magellanic clouds .
the question naturally arises whether there are x - ray resolved snrs in some nearby dwarf galaxies at a distance between the magellanic clouds and m31 .
ngc 6822 is one of the nearest dwarf irregular galaxies in the local group beyond the magellanic clouds and has a rich history in observations ( see , e.g. , perrine 1922 ; hubble 1925 ) . at a distance of about 500 kpc ( mcalary et al .
1983 ; gallart , aparicio , & vilchez 1996 ) , ngc 6822 is sufficiently close that one can explore its snrs with high resolution imaging .
however , the relatively high absorption of ngc 6822 is a substantial problem ( @xmath4 ; schlegel et al .
1998 ) . despite its proximity
, there is only one known snr in ngc 6822 .
the snr ho 12 was first identified as an emission - line object by hodge ( 1969 , 1977 ) but it was not clear if ho 12 was an h ii region or a snr . ho 12
was then confirmed as a snr based on narrow - band imaging and optical spectroscopy ( smith 1975 ; dodorico , dopita , & benvenuti 1980 ; dodorico , & dopita 1983 ) . in the x - ray band ,
an x - ray source was found associated with ho 12 from the _ einstein _ hri ( markert & donahue 1985 ) , and _ rosat _ pspc ( eskridge & white 1997 , ew97 hereafter ) .
although both _ einstein_and _ rosat _ observations indicated that the x - ray source is likely to be a snr , they could not rule out that it may be an h ii region or an x - ray binary .
the only published radio detection near ho 12 was from the very large array ( vla ) observations ( dickel et al .
an extended source with an integrated flux density of @xmath5 mjy at 1.4 ghz was associated with ho 12 . in this paper
, we report on an archival _ chandra _ observation of the luminous snr , ho 12 , in ngc 6822 .
in addition , we present narrow band optical imagings from the local group surveymassey / lgsurvey ] ( lgs ; massey et al .
2001 ) and a new radio image of ho 12 from combined vla archival data sets . in section 2
, we detail the observations and the results from observations in different wavelengths .
a discussion about the nature of ho 12 is presented in section 3 .
ngc 6822 was observed with the _ chandra _ advanced ccd imaging spectrometer ( acis - i ) on 2002 november 4 for 28 ksec .
the acis data were telemetered in very faint mode ( vfaint ) .
we reprocessed the x - ray data with the newest calibration ( caldb v2.26 ) and at the same time , we screened the raw data in order to utilize the vfaint mode , and applied the charge transfer inefficiency correction to correct the loss of charge in a ccd as it is shifted from one pixel to the next during readout .
we also removed the 0.5-pixel randomization during the pipeline processing to improve the resolution of on - axis _
chandra _ source ( see also garcia et al .
2001 ) . in order to reduce the instrumental background , we screened the data to allow only photon energies in the range of 0.37 kev .
we also searched for periods of high background using source free regions but none was present .
we therefore used all the data for analysis .
data were reduced and analyzed with the ciao v3.0.2 package , and spectral analysis was performed with sherpa v3.0 . in the _ chandra _ image , it is clear that a bright x - ray source is located at about @xmath1 from the aim - point of the observation .
the position of this x - ray source is consistent with previous detections of ho 12 by _
einstein _
( markert & donahue 1985 ) and _ rosat _ ( ew97 ) . to improve the absolute astrometry of the _ chandra _ data , we cross - correlated detected
_ chandra _ sources ( by wavdetect in the ciao ) with the usno - b1.0 catalog ( monet et al .
2003 ) . within @xmath6 of ho 12 ,
two x - ray sources were less than @xmath7 from bright usno stars ( @xmath8 , @xmath9 ) ; they are very likely to be foreground stars .
we matched the _
chandra _
positions with the usno stars and shifted the _
chandra _
image by @xmath10 and @xmath11 in right ascension and declination , respectively . after correcting for this aspect offset , the coordinates of the center of ho 12
are found to be r.a.=19@xmath12 44@xmath13 [email protected] , decl.=@xmath15 48@xmath16 30@xmath17 ( j2000.0 ) .
the source is clearly extended with a diameter of @xmath18 ( @xmath19 pc at 500 kpc ) . to improve the spatial resolution of the image , we applied subpixel event repositioning ( ser ) techniques ( li et al .
2004 ; see also tsunemi et al .
2001 ; li et al .
ser techniques are based on the premise that the impact position of events can be refined , according to the distribution of charge among affected ccd pixels .
it is proven that ser techniques can improve the spatial resolution by as much as @xmath20 .
we here employed an energy and charge - split dependent ser model on the _ chandra _ image .
all the results reported here are based on the reprocessed image .
figure 1 shows the smoothed `` true color '' x - ray image of ho 12 .
the color scheme is defined as 0.30.9 kev ( red ) , 0.91.5 kev ( green ) , and 1.57 kev ( blue ) .
it is clear that most of the x - ray emission is from below 1.5 kev .
we extracted the energy spectrum from a @xmath21 radius and background from an annulus centered on the source ; the total source counts are 211 while the background subtracted counts are 205.75 . in order to allow @xmath22 statistics to be used
, the spectrum was grouped into at least 20 counts per spectral bin .
the fitting statistic was then used by sherpa ( with chi dvar ) to find the model parameters that best fit the x - ray spectral data .
response files were created according to the ccd temperature with standard ciao routines ; quantum efficiency degradation of acis was also corrected by using the caldb v2.26 .
we fitted the data with several single - component spectral models including absorbed power law , blackbody , raymond - smith , and nonequilibrium ionization ( nei ) models .
the raymond - smith model is a simple collisional equilibrium ionization model , while the nei model is appropriate for modeling snrs whose age is smaller than the time required to reach ionization equilibrium .
the nei model consists of an electron temperature ( @xmath23 ) and an ionization timescale ( @xmath24 ) , where @xmath25 and @xmath26 are the mean electron density and the elapsed time after the plasma was shock heated to a constant temperature @xmath23 .
both models are often applied to study x - ray emission from extragalactic snrs ( see , e.g. , wang 1999 ; schlegel , blair , & fesen 2000 ; hughes , hayashi , & koyama 1998 ) . for the raymond - smith and nei models ,
we fixed the chemical abundances in number of atoms relative to hydrogen ( he : @xmath27 , c : @xmath28 , n : @xmath29 , o : @xmath30 , mg : @xmath28 , si : @xmath31 , s : @xmath32 ) which gave the best fit from optical spectroscopy ( dodotico & dopita 1983 )
. the results of the spectral fits are given in table 1 and figure 2 .
it is clear that except for the nei model , all models did not give a good fit ( @xmath33 ) .
the best - fit electron temperature of the nei model is @xmath34 kev with @xmath35 @xmath36 .
the galactic absorption for ngc 6822 is @xmath37 @xmath36 , while the observed intrinsic column associated with this region of ngc 6822 is about @xmath38 @xmath36 ( hodge et al .
therefore , the absorption derived by our spectral fit is consistent with the optical measurement .
the 0.37 kev luminosity is @xmath2 ergs s@xmath39 .
note all quoted uncertainties are 90% confidence . + @xmath40 power - law slope .
+ @xmath41 ionization timescale in units of s @xmath42 .
+ @xmath43 0.37 kev luminosity ( @xmath44 ergs s@xmath39 ) , assuming a distance of 500 kpc .
we obtained the h@xmath45 , [ o iii ] , [ s ii ] , @xmath46-band , and @xmath47-band images of ngc 6822 from the lgs .
these images have been properly flat - fielded and the geometric distortions removed so that the coordinates in the images are good to @xmath48 and the images at the different bandpasses are registered with one another .
we therefore were easily able to subtract the @xmath46-band continuum from the [ o iii ] image and the @xmath47-band continuum from the [ s ii ] and h@xmath45 images in order to make the line - emitting sources stand out .
these images were then compared to the x - ray images of ho 12 , revealing similar x - ray and optical sizes and morphologies , as shown in figures 1 and 3 .
to provide convenient optical flux estimates for ho 12 which compliment those from the x - ray and radio data , we roughly calibrated the lgs narrow - band images .
the h@xmath45 calibration was done using the bright h ii regions _ hubble _
@xmath46 and _ hubble _ @xmath49 ( hubble 1925 ) .
the published h@xmath45 fluxes of these objects are 4.8 and 3.8 @xmath50 ergs @xmath36 s@xmath39 , respectively ( odell , hodge & kennicutt 1999 ) .
we measured the counts of these h ii regions in an aperture with 30@xmath17 radius in the star - subtracted lgs h@xmath45 image .
these measurements provided a rough calibration for the lgs h@xmath45 image of 6@xmath51 ergs @xmath36 s@xmath39 ct@xmath39 .
this rough calibration contains a systematic error because the @xmath47-band contains the h@xmath45 ; however , the ratio of the broadband to narrow - band filter widths ( 1510/80 ) suggests that the nebular contamination in the broadband filters should be @xmath5210% .
we then measured the counts in the star - subtracted lgs narrow - band images of ho 12 using a 4.5@xmath17 radius aperture . applying the calibration from _
@xmath46 and _ hubble _
@xmath49 , the total h@xmath45 flux of ho 12 is 6.9 @xmath53 ergs @xmath36 s@xmath39 . this h@xmath45 flux is in good agreement with the surface brightness measured by killen and dufour ( 1982 ) , which results in a total flux of 8@xmath53 ergs @xmath36 s@xmath39 if a radius of 4@xmath17 is applied .
the emission - line ratios for ho 12 have already been spectroscopically measured by smith ( 1975 ) , who found [ s ii]/h@xmath45 = 0.5 and [ o iii]/h@xmath54 = 0.9 . using his reddening value of @xmath55 , we infer an observed [ o iii]/h@xmath45 = 0.2 .
these ratios provide [ s ii ] and [ o iii ] total flux estimates of 3.5@xmath5610@xmath57 and 1.5@xmath5610@xmath57 erg s@xmath39 @xmath36 , respectively . applying the measured source counts from ho 12 in the lgs [ s ii ] and [ o iii ] images , rough calibrations for the lgs [ s ii ] and [ o iii ] images of ngc 6822 are 6.6 @xmath51 ergs @xmath36 s@xmath39 ct@xmath39 and 1.2 @xmath58 ergs @xmath36 s@xmath39 ct@xmath39 , respectively .
we note that these rough calibrations also contain a systematic error because the @xmath46-band contains [ o iii ] and the @xmath47-band contains [ s ii ] ; however , the ratio of the broadband to narrow - band filter widths for these bandpasses ( 940/55 and 1510/81 for @xmath46/[o iii ] and @xmath47/[s ii ] , respectively ) suggest that the nebular contamination in the broadband filters should be @xmath5210% .
the radio data were obtained from the vla archive .
it includes the original dickel et al .
( 1985 ) data set taken in 1984 , as well as two longer observations taken by others for interests other than the snr in 1985 and 1995 .
all three data sets were observed in vla b - array , i.e. , with a spatial resolution of about 5 at 1.4 ghz .
we did not use the 5 ghz observations included in the data as they had different phase centers and thus much less sensitivity on the position of the snr .
after some initial flagging of bad data points , all data sets were reduced with the new vlarun procedure in nrao s aips package .
the final image was made using the combined data after one iteration of self - cal on strong sources in the field .
the resulting total effective on - source time of the combined 1.4 ghz observations is about 7 hours , yielding an image noise of about 60 @xmath59jy / beam .
however , the archival vla data sets were taken using different calibrators and different epochs .
in addition , some of the calibrator positions have been corrected or updated to more acurate positions , resulting in small uncertainties in the radio positions of each individual archival data set .
compared to the nrao vla sky survey ( nvss , condon et al .
1998 ) , we have measured an offset of 4.4 in r.a . and 0.4 in dec . using seven bright radio sources in our final image . given the measured shift , and the observation that the size and position of the radio source overlap in coordinates with the optical and x - ray source within one beam , we conclude that the radio , optical and x - ray source all are from the same remnant , ho 12 .
we have shifted the radio data 3.9 westward to obtain the overlays shown in fig .
3 . with the correction in radio
coordinates and this shift , this suggests that the _ chandra_/optical alignment ( sect.2.1 ) and radio reference frame coincide within a fraction of 1 , i.e. , about one pixel in the _ chandra _ image . also due to combining different data sets , the spatial resolution ( @xmath60 ) and the @xmath61 deconvolved size of the snr are different from the values measured by dickel et al .
the integrated flux of the snr , [email protected] mjy , is nevertheless consistent .
the radio data show a shell - type snr , with a brighter region west of the center , and some asymmetric extentions toward the north and east rims , possibly related to the regions with the brighter h@xmath45 knots . due to the low resolution
, there is no clear correspondence between the radio emission and x - ray knots .
ho 12 is clearly resolved with _
chandra _ into a shell - shaped object .
the diameter of ho 12 as measured by _ chandra _ is about @xmath1 , corresponding to 24 pc .
the size and the morphology of the x - ray remnant roughly agrees with its optical counterpart .
it is therefore no doubt that the x - ray source is the snr , ho 12 .
the x - ray spectrum of ho 12 can be fit with an absorbed nei model although the uncertainties of the electron temperature are large due to the low count rate . using this model , the 0.37 kev luminosity of the snr is about @xmath2 ergs s@xmath39 , which is similar to some snrs in the magellanic clouds ( e.g. , hughes et al .
1998 ; williams et al . 1999 ) .
the relatively high electron temperature ( 2.8 kev ) is not unusual ; for example , n132d in the large magellanic cloud requires a two - component nei model for which the temperatures are 0.8 and 2.7 kev ( favata et al .
the usual explanation for the hot component is that it comes from the shock - heated swept - up circumstellar medium or it is due to the inhomogeneity of the interstellar medium ( ism ) .
assuming ho 12 is in the adiabatic expansion phase , we can estimate its physical parameters through the sedov solution . using equations ( 1 ) and ( 2 ) from kong et al .
( 2002 ) and assuming an initial explosion energy of @xmath63 ergs ( blair , kirshner , & chevalier 1981 ) , radius of 12 pc , and shock temperature of @xmath64 , we obtain an age estimate of @xmath65 years and a density estimate of @xmath66 @xmath42 . the snr nature of ho 12 was called into question due to the x - ray variability as observed by _ einstein _ and _ rosat _ ( ew97 ) .
however , the _ einstein
_ hri count rate quoted by ew97 was from fabbiano et al .
( 1992 ) in which it was derived from the entire field of view .
this is misleading since there is another bright source in the field as discussed by markert & donahue ( 1985 ) .
the count rate of ho 12 measured by markert & donahue ( 1985 ) is about 33% of the rate used by ew97 .
if we take this into account , there is indeed no significant variation between the _ einstein _ and _ rosat _ observations ( see table 1 of ew97 ) . to compare the fluxes derived from different instruments
, we converted the count rates into fluxes ( 0.22 kev ) by using the best - fit nei model with pimms .
for the _ einstein _
hri data ( taken in 1979 ) , we derived a background subtracted count rate of 0.0014 c / s which is consistent with markert & donahue ( 1985 ) .
we also checked the hricfa database for which the count rate is corrected for vignetting , deadtime , mirror scattering , and quantum efficiency , the count rate of ho 12 is 0.002 c / s and we used this value for calculating the flux . for the _ rosat _ pspc ( taken in 1992 ) ,
our measured count rate is 0.011 c / s , consistent with the pspc wga catalog .
there are two _ rosat _ hri datasets ( taken in 1995 and 1996 ; see zang & meurs 2001 ) of ho 12 . from the brera multi - scale wavelet hri catalog ( bmw - hri ; panzera et al .
2003 ) , ho 12 was clearly detected with count rates of 0.0035 c / s ( 1995 ) and 0.0037 c / s ( 1996 ) .
our own measurements of the archival data yield 0.0031 c / s and 0.0035 c / s ; it was underestimated as we did not correct the count rate for the vignetting and point spread function .
we therefore used the values from the bmw - hri catalog for analysis .
using pimms , we derived the history of fluxes ( in units of ) as follows : @xmath67 ( _ einstein _
hri ) , @xmath68 ( _ rosat _ pspc ) , @xmath69 , and @xmath67 ( _ rosat _ hri ) .
the 0.22 kev _ chandra _ flux is @xmath70 .
the difference among them is less than 20% .
therefore , we conclude that ho 12 has not shown variability in the past two decades . at optical wavelengths , ho 12
is clearly detected as an extended object in the narrow - band lgs images ( see figure 1 ) .
the [ s ii]/h@xmath45 ratio ( @xmath710.4 ; smith 1975 ; dodorico et al . 1980 ) of ho 12 is typical of snrs ( e.g. levenson et al .
the reddening - corrected spectroscopic [ o iii]/h@xmath45 ratio is low ( 0.3 ; smith 1975 ) .
this ratio is about one fourth the ratio predicted for a typical type iia snr shock in a medium with half the solar o abundance ( vancura et al .
the low [ o iii]/h@xmath45 ratio is therefore reasonable considering the known o abundance of the ngc 6822 ism ( @xmath7125% solar ; lequeux et al .
1979 ; pagel et al .
1980 ) .
our analysis of archival vla data provides a tighter constraint on the integrated 1.4 ghz flux density .
dickel et al . ( 1985 ) first detected this source as an extended ( @xmath72 ) faint ( @xmath73 mjy at 1.4 ghz ) object .
however , the noise of their radio image was dominated by confusion due to nearby very bright objects . therefore , the error of the flux is comparable with the observed value and it can not be claimed to be a solid detection .
we here obtained a more sensitive radio image by adding more observations .
the resulting flux is @xmath74 mjy at 1.4 ghz , consistent with dickel et al .
( 1985 ) .
the morphology of ho 12 is very similar across all wavelengths .
there is a hint of brightening in the northwestern region from the narrow - band optical images and the _ chandra _ image .
another `` bright spot '' located at the east side of the remnant in the optical images , however , corresponds to a breakout region in x - rays .
there is no sinificant difference in morphology among the three narrow - bands images . in x - ray , although most of the emission is from below 1.5 kev ( see fig . 1 and 2 ) , it is interesting that the distributions of the soft ( 0.30.9 kev ) and medium ( 0.91.5 kev ) photons are slightly different .
the soft band elongates along the northeast - southwest direction , while the medium band tends to distribute along the northwest - southeast direction .
although the radio image of ho 12 is in low resolution ( @xmath21 ) , the morphology is consistent with the optical and x - ray images .
in particular , the assymetry of the radio emission toward the north and east sides may assoicate with the h@xmath45 knots .
however , because of the large uncertainty in astrometry and low spatial resolution , these spatial coincidences are difficult to determine and interpret . a high resolution radio image and spectral index measurement of ho 12 are therefore required to further investigate the radio emission and multiwavelength morphology of the remnant .
the authors thank k. johnson and a. seth for sharing their 5 ghz ngc6822 source positions prior to publication .
was supported by nasa under go3 - 4049x and an ltsa grant nag5 - 10705 .
b.f.w . acknowledges support from nasa through grant go-3103x from the chandra x - ray center .
the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
blair , w.p . ,
kirshner , r.p .
, & chevalier , r.a .
1981 , apj , 247 , 879 condon , j. j. , cotton , w. d. , greisen , e. w. , yin , q. f. , perley , r. a. , taylor , g. b. , & broderick , j. j. 1998 , aj , 115 , 1693 dickel , j. r. , silverman , a. , & dodorico , s. 1985 , , 90 , 414 dodorico , s. , dopita , m. a. , & benvenuti , p. 1980
, , 40 , 67 dodorico , s. & dopita , m. 1983 , iau symp . 101 : supernova remnants and their x - ray emission , 101 , 517 eskridge , p. b. & white , r. e. 1997 , , 114 , 988 ( ew97 ) fabbiano , g. , kim , d .- w . ,
& trinchieri , g. 1992 , , 80 , 531 favata , f. , vink , j. , parmar , a. n. , kaastra , j. s. , & mineo , t. 1997 , a&a , 324 , l45 garcia , m. r. , kong , a. k. h. , primini , f. a. , mcdowell , j. , & murray , s. s. 2001 , bulletin of the american astronomical society , 33 , 1178 hodge , p.w .
1969 , , 18 , 73 hodge , p.w .
1977 , , 22 , 69 hodge , p.w . , smith , t. , eskridge , p.b . , macgillivray , h. , & beard , s. 1991 , , 379 , 621 hubble , e. p. 1925
, , 62 , 409 hughes , j. 2001 , in aip conf .
565 , young supernova remnants , ed . s. s. holt & u. hwang ( new york : aip ) , 419 hughes , j. p. , hayashi , i. , & koyama , k. 1998 , , 505 , 732 levenson , n. a. , krishner , r. p. , blair , w. p. , & winkler , p. f. 1995 , , 110 , 739 lequeux , j. , peimbert , m. , rayo , j. f. , serrano , a. , & torres - peimbert , s. 1979 , , 80 , 155 li , j. , kastner , j. h. , prigozhin , g. y. , & schulz , n. s.2003 , , 590 , 586 li , j. , kastner , j. h. , prigozhin , g. y. , & schulz , n. s. , feigelson , e. d. , & getman , k. v. 2004 , apj , submitted ( astro - ph/0401592 ) killen , r. m. & dufour , r. j. 1982 , , 94 , 444 kong , a. k. h. , garcia , m. r. , primini , f. a. , & murray , s. s. 2002 , , 580 , l125 kong , a. k. h. , sjouwerman , l. o. , williams , b. f. , garcia , m. r. , & dickel , j. r. 2003 , , 590 , l21 magnier , e. a. , prins , s. , van paradijs , j. , lewin , w. h. g. , supper , r. , hasinger , g. , pietsch , w. & tremper , j. 1995 , a&as , 114 , 215 markert , t .
h. & donahue , m. e. 1985 , , 297 , 564 massey , p. , hodge , p. w. , holmes , s. , jacoby , g. , king , n. l. , olsen , k. , saha , a. , & smith , c. 2001 , bulletin of the american astronomical society , 33 , 1496 monet d. g. , et al .
2003 , , 125 , 984 odell , c. r. , hodge , p. w. , & kennicutt , r. c. 1999 , , 111 , 1382 panzera m.r .
, campana s. , covino s. , lazzati d. , mignani r.p . , moretti a. , & tagliaferri g. 2003 , a&a , 399 , 351 perrine , c. d. 1922 , , 82 , 486 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , apj , 500 , 525 schlegel , e. m. , blair , w. p. , & fesen , r. a. 2000 , aj , 120 , 791 smith , h.e . 1975 , , 199 , 591 tsunemi , h. , mori , k. , miyata , e. , baluta , c. , burrows , d. n. , garmire , g. p. , & chartas , g. 2001 , , 554 , 496 wang , q. d. 1999 , apj , 517 , l27 williams , b. f. , sjouwerman , l. o. , kong , a. k. h. , gelfand , j. d. , garcia , m. r. , murray , s. s. 2004 , , submitted williams , r. , chu , y .- h . , dickel , j. , petre , r. , smith , r.c . , tavarez , m. 1999 , apjs , 123 , 467 zang , z , & meurs , e.j.a . 2001 , , 556 , 24 | the supernova remnant ( snr ) , ho 12 , in the center of the dwarf irregular galaxy ngc 6822 was previously observed at x - ray , optical , and radio wavelengths . by using archival _
chandra _ and ground - based optical data , we found that the snr is spatially resolved in x - rays and optical .
in addition , we obtained a @xmath0 resolution radio image of the snr .
these observations provide the highest spatial resolution imaging of an x - ray / optical / radio snr in that galaxy to date .
the multi - wavelength morphology , x - ray spectrum and variability , and narrow - band optical imagings are consistent with a snr .
the snr is a shell - shaped object with a diameter of about @xmath1 ( 24 pc ) .
the morphology of the snr is consistent across the wavelengths while the _ chandra _ spectrum can be well fitted with a nonequilibrium ionization model with an electron temperature of 2.8 kev and a 0.37 kev luminosity of @xmath2 .
the age of the snr is estimated to be @xmath3 years .
# 1to 0pt#1 [ firstpage ] |
the samples of ho(co@xmath13al@xmath14)@xmath5 alloys were prepared from pure components by melting in an induction furnace under a protective ar atmosphere and were subsequently annealed in vacuum at 1100 k for about one week .
the x - ray analysis showed no traces of impurity phases .
a four probe dc method was used for the electrical resistivity measurements ; for the thermopower measurements a differential method was applied .
typical size of the samples was about 1@xmath15 1@xmath15 10 mm@xmath16 .
the estimated error in the absolute value of the electrical resistivity is @xmath1710% .
this is mainly due to uncertainty in the sample geometry which is closely related to the mechanical quality of the samples .
the thermopower was measured with an accuracy of @xmath170.2 @xmath18vk@xmath19 .
the resistivity temperature dependencies are depicted in figure [ resd ] .
the residual resistivity , shown in the inset , rapidly increases with al content .
simultaneously , the temperature dependent part of the resistivity decreases , so that the high temperature total resistivity shows a comparatively small change with composition .
room temperature values are in the range of 130 to 150 @xmath20 cm . temperature dependent resistivity of ho(co@xmath13al@xmath14)@xmath5 for various al concentrations .
the inset shows the composition dependence of the residual resistivity , measured at 4.2 k. ] the curie temperature is marked by an abrupt change of the resistivity of the samples with @xmath21=0 , 0.005 , 0.025 , and 0.05 , whereas samples with @xmath21=0.075 and 0.1 clearly show a second order - like variation of the resistivity at @xmath11 .
data of fig .
[ resd ] indicate that the boundary between first - order and second - order transitions is between @xmath21=0.05 and @xmath21=0.075 .
these results agree with the data of duc et al .
@xcite .
figure [ thpw ] presents experimental results for the thermopower from 5 k to 300 k. dramatic changes of @xmath9 with the alloy composition are clearly seen in both , the paramagnetic and in the magnetically ordered state of the alloys . at high temperatures the @xmath9 minimum temperature @xmath10 , observable for pure hoco@xmath5 at 250 k , and the absolute @xmath9 values decrease with increasing al content . in the low temperature region , ( @xmath22 ) , @xmath9 is also strongly dependent on the al - content . temperature dependent thermopower of ho(co@xmath13al@xmath14)@xmath5 .
the lower panel shows @xmath9 of the _ x _ = 0 , _ x _ = 0.005 and _ x _ = 0.025 alloys in a vicinity of curie temperature . ] at the lowest temperatures all compounds exhibit a minimum with negative @xmath9 values ; its temperature gradually increases with al concentration .
the @xmath9 minimum , as well at the following maximum , become suppressed for @xmath21=0.05 and they almost disappear for @xmath21=0.075 .
the curie temperatures are associated with the discontinuous changes of @xmath9 for @xmath21=0 , 0.005 , 0.025 , whereas the sample with @xmath21=0.075 clearly shows a second order - like variation of the thermopower at @xmath11 .
the temperature variation of the thermopower in the paramagnetic temperature region can be understood within the model proposed in @xcite .
calculations of the density of states , @xmath23 for yco@xmath5 @xcite revealed the fermi level , @xmath24 , lying near a sharp peak of dos , primarily composed of 3_d _ states of co ( figure [ dos ] ) . dos of _
r_co@xmath5 compounds in a vicinity of the fermi level ( solid line ) @xcite .
the gray line shows @xmath25 at _
t_=300 k ( @xmath26 is fermi distribution function ) to indicate the energy region important for transport properties .
the broken line is the model state density , used for thermopower calculations .
@xmath27 is the width of the model dos peak , @xmath28 - is the distance from fermi level to the center of the peak . ] since co is present in all _ r_co@xmath5 compounds , it is assumed , that @xmath23 has the same general features throughout the _
r_co@xmath5 series .
it has been shown that the high - temperature minimum , observed in @xmath9 of nearly all _ r_co@xmath5 compounds ( with the exception of gdco@xmath5 , dyco@xmath5 , and smco@xmath5 where @xmath10 falls into the temperature range where these compounds order magnetically ) , is associated with the 3_d _ peak of the dos , @xmath10 being the measure of the peak width @xcite .
since the width of an itinerant electronic band depends on the extent of the corresponding atomic orbital overlapping , one should expect that the 3_d _ peak width increases when the lattice parameter decreases .
the lattice parameter within _
r_co@xmath5 compounds reveals considerable variation , being smallest for scco@xmath5 and largest for ndco@xmath5 .
this implies that the 3_d _ dos
peak width is largest in case of scco@xmath5 and smallest for ndco@xmath5 with corresponding variation of @xmath10 .
this assumption has been confirmed by a good correlation between @xmath10 and the lattice constants of the _ r_co@xmath5 compounds @xcite .
figure [ t - dcorr]a reproduces , in part , this correlation . *
a*. dependence of @xmath10 against lattice parameter for _ r_co@xmath5 compounds ( @xmath29 - data from @xcite ) .
the triangles show this work results for ho(co@xmath13al@xmath14)@xmath5 alloys . *
b. * @xmath10 ( triangles ) and curie temperature @xmath11 ( @xmath30 ) against composition for ho(co@xmath13al@xmath14)@xmath5 . * @xmath15 * @xmath11 against composition for ho(co@xmath13si@xmath14)@xmath5 alloys according to ref . @xcite . ] the decrease of @xmath10 with increasing al content in ho(co@xmath13al@xmath14)@xmath5 alloys ( fig .
[ t - dcorr]b ) indicates a narrowing of the 3_d _ dos peak .
partially this narrowing can be connected with the increase of the lattice parameter in the alloys .
however , the expansion of the crystal lattice is insufficient to entirely account for the narrowing of the dos peak and the observed increase of the curie temperature .
if the decrease of @xmath10 was related only to the lattice expansion , then one would expect that @xmath10 in the alloys decreases roughly at the same rate as in pure _ r_co@xmath5 compounds .
however , the results for ho(co@xmath13al@xmath14)@xmath5 reveal a much stronger dependence of @xmath10 on the lattice parameter ( fig.[t - dcorr]a ) , implying that some additional factor is responsible for the narrowing of the peak width and increase of @xmath11 .
literature results on y@xmath31lu@xmath32(co@xmath13al@xmath14)@xmath5 @xcite alloys with invariable lattice parameter show , that at small _ x _ ( up to about _
x_=0.060.1 ) the critical field for metamagnetic transition in these alloys decreases at the same rate as in lu(co@xmath13_m_@xmath14)@xmath5 ( _ m _ = ga , sn , al ) @xcite and y(co@xmath13al@xmath14)@xmath5 @xcite alloys , where lattice parameter varies considerably . similarly , in ho(co@xmath13si@xmath14)@xmath5 diluted alloys @xcite with ivariable lattice parameter and in ho(co@xmath13al@xmath14)@xmath5 alloys @xmath11 increases with @xmath21 at the same rate , see fig
. [ t - dcorr]b .
this indicates , in agreement with our results , that lattice expansion can not be the main cause for the magnetic enhancement of the alloys . on the other hand , the 3_d _ band de - population model
also can not explain these results since the substituting elements have different configurations of the outer electronic shells .
the results show that the mechanism , which causes the decrease of @xmath7 in non - magnetic compounds ( or increase of @xmath11 in the magnetic ones ) does depend at small _ x _ neither on the type of substituting element nor on the lattice expansion . what all above mentioned alloys have in common , is disorder within the co - sublattice due the substitution of co by other elements .
we therefore conclude , that the principal reason for the narrowing of the 3_d _ band in ho(co@xmath13al@xmath14)@xmath5 for low al contents , is the increasing localization of the 3_d _ electrons caused by disorder within co - sublattice .
experimental results on transport in y(co@xmath13al@xmath14)@xmath5 alloys has led us to a similar conclusion @xcite .
a different situation takes place when the impurity element has unfilled 3_d_-shell , like ni and fe , in that case the effects of the change of 3_d _ electron concentration play a dominant role @xcite . of course
, one can not exclude a change of the 3_d _ electron concentration upon the co - substitution in ho(co@xmath13al@xmath14)@xmath5 .
moreover , the decrease of @xmath9 values in the paramagnetic temperature region , as will be demonstrated later , implies such a change .
it also follows from literature that the behavior of y(co@xmath13_m_@xmath14)@xmath5 alloys at large _
( above about 0.1 ) , such as the stabilization of a ferromagnetic ground state , depends on the kind of substituting elements , suggesting a hybridization of _ s p _ and 3_d _ states .
however , our analysis shows that the change of the 3_d _ electron concentration does not play major role in magnetic enhancement of diluted alloys . to summarize this part ,
our experimental results on transport properties of ho(co@xmath13al@xmath14)@xmath5 alloys and analysis of the results , reported in the literature , led us to the conclusion that the principal mechanism , responsible for the increase of @xmath11 in magnetic alloys , or a decrease of @xmath7 in paramagnetic alloys , is the narrowing of the 3_d _ dos peak due to localization of co 3_d _ electron states , induced by disorder in co sublattice of the alloys .
the rapid increase of the residual resistivity and the modest variation of high temperature resistivity with al content are the most characteristic features observed in the resistivity of the alloys .
the temperature dependent part of the resistivity systematically decreases with _ x _ as it is shown in fig .
[ tdepr ] .
we should note that such behavior is common for many _ r_co@xmath5 - based alloys @xcite , which include both , _ r_(co@xmath13_m_@xmath14)@xmath5 and y@xmath13_r_@xmath14co@xmath5 systems .
temperature induced part of resistivity ( @xmath33 ) of ho(co@xmath13al@xmath14)@xmath5 alloys . ]
however , there has been no commonly accepted model , able to account for the observed variation of the temperature dependent resistivity .
the independence of the phenomenon on the constituents of the alloys implies a common mechanism . as in the case of the above
discussed co - localization , disorder and metamagnetism of the magnetic co subsystem are the natural candidates .
baranov et al .
@xcite were probably the first to emphasize the important role , which disorder can play in _ r_co@xmath5 - based alloys . in a certain range of alloy composition
, 3_d _ magnetic subsystem can be in `` partially '' ordered ground state , i.e. in the alloy there are spatial regions with high and low 3_d _ magnetization .
this static magnetic disorder originates from a combination of fluctuating exchange field ( in ( y@xmath34_r_@xmath14)co@xmath6 alloys ) , or fluctuating local susceptibility ( in _ r_(co@xmath13_m_@xmath14)@xmath5 alloys ) , and of the metamagnetism of 3_d _ electron system .
scattering of conduction electrons of this static magnetic disorder gives contribution to the alloy resistivity , which is comparable to the high temperature limit of 3_d _ spin fluctuation ( sf ) resistivity ( the resistivity due to scattering on dynamic , i.e. temperature - induced , fluctuations of local 3_d _ magnetization ) in _ r_co@xmath5 compounds @xcite .
the static magnetic disorder can be viewed as `` frozen '' sf .
when temperature increases this frozen disorder melts and is replaced by the dynamic disorder . since the corresponding contributions to resistivity
are of similar magnitude , the temperature variation of the total resistivity is considerably reduced .
figure [ resrel ] shows the resistivity of ho(co@xmath13al@xmath14)@xmath5 , dyco@xmath5 and tbco@xmath5 @xcite , normalized to its value at _
t_=300 k , plotted against @xmath35 to account for the different width of the 3_d _ band . since for the ho(co@xmath13al@xmath14)@xmath5 with @xmath36 0.075 the high temperature minimum in @xmath9 is not observable due to the high @xmath11 , we use an interpolation of @xmath37 , as it is shown in fig .
[ t - dcorr]b by the broken line , to obtain @xmath10 of these alloys .
the other two interesting features of the resistivity can be seen from figure [ resrel ] : + 1 .
the kind of the phase transition at curie point is changed from first to second order around @xmath11=@xmath10 .
this is another manifestation that @xmath10 is an essential physical parameter , directly reflecting dos features .
the relation between @xmath10 and the change of the transition order follows from the model of the metamagnetic transition of co subsystem @xcite as a consequence of the temperature smearing of dos features at the fermi level @xcite .
normalized resistivity @xmath38 against normalized temperature @xmath35 for ho(co@xmath13al@xmath14)@xmath5 alloys and dyco@xmath5 , tbco@xmath5 compounds @xcite . ]
essentially the same mechanism leads to the development of the minima in @xmath9 @xcite .
there is , however , an alternative theoretical model , in which the order of a magnetic phase transition depends basically on the lattice constant value @xcite . since the width of the dos peak ( and
, therefore , @xmath10 ) is determined in pure _ r_co@xmath5 compounds by the lattice parameter , @xmath39 represents a more general condition for the boundary between second and first order phase transition in these compounds .
the magnitude of the resistivity drop at the curie temperature @xmath40 decreases as @xmath11 approaches @xmath10 .
this last feature is also present in binary _
r_co@xmath5 compounds , as it can be seen comparing the resistivities of hoco@xmath5 and dyco@xmath5 , shown in fig .
[ resrel ] .
the resistivity drop at @xmath11 of heavy _ r_co@xmath5 compounds , which exhibit a first - order transition , has been attributed to a sudden suppression of sf by strong molecular field of the ordered 4_f _
moments @xcite .
this , however , does not explain the decrease of @xmath40 with increasing @xmath11 .
we propose that the origin of this decrease is the same as the origin of resistivity saturation of nearly magnetic compounds at high temperatures @xcite . in the model ,
put forth in @xcite by ueda and moriya , the sf contribution to the resistivity ( @xmath41 ) is determined by the _
d _ band dynamical susceptibility , which is enhanced due to _ d _ band sf .
it has been shown that the saturation tendency , which is characteristic for the temperature dependent resistivity of nearly magnetic materials , originates from the strong decrease of the _ d _ band enhancement owing to interactions among sf at high temperatures .
we use here a very simplified model with a rectangular approximation for the co 3_d _ dos ( fig .
[ mdos ] ) to illustrate the essential physics of the resistivity saturation and of the decrease of @xmath40 .
we consider an enhanced 3_d _ band paramagnetic compound ( such as yco@xmath5 ) . at _
t_=0 k , without external field , the spin - up and spin - down sub - bands are equally occupied , fig .
[ mdos]a , and there are no sf . at a nonzero temperature thermal fluctuations of the 3_d _ sub - band occupation , amplified by exchange interaction ,
result in fluctuations of local magnetization , i.e. sf . according to @xcite the magnitude of the local magnetization fluctuations @xmath42 increases with temperature
, however it saturates at a constant value in a high temperature limit .
both , the saturation value and cross - over temperature , are determined by the 3_d _ band parameters . in magnetic _ r_co@xmath5 compounds the 3_d _ system at _
t_=0 k is in high magnetization state . in the following we consider 3_d _ band in this case as fully polarized by the exchange field of ordered 4_f _ moments , as it schematically shown in fig .
[ mdos]*b * , with saturation magnetic moment @xmath43 . as temperature increases
, there will be also temperature - induced fluctuations of the 3_d _ magnetization around this polarized state . for our qualitative model
we assume that the sf resistivity is determined by the spin
spin correlation function @xmath44 where @xmath45 and @xmath46 are the local 3_d _ magnetization at lattice point with coordinates @xmath47 and @xmath48 , respectively , and @xmath49 is distance of order of conduction electron ( s electron ) mean free path .
the scattering magnitude of conduction electron by a magnetic fluctuation is proportional to its squared magnitude , measured relative to the local background at a distance of order of conduction electron mean free path .
therefore , the most effective will be the scattering on correlated magnetic fluctuations with largest difference of local magnetization at points @xmath47 and @xmath48 .
for the paramagnetic state , this corresponds to fluctuations at @xmath47 and @xmath48 with the opposite magnetization . while for a system in ferromagnetic ground state with fully polarized 3_d _
band , the largest difference is between this fully polarized state moment @xmath50 and local fluctuation with a reduced moment .
basing on the above arguments , we define the effective sf magnitude ( @xmath51 - paramagnetic state , @xmath52 - ferromagnetic state ) , determining the sf resistivity , as : @xmath53 where @xmath54 and @xmath55 with @xmath56 and @xmath57 the definitions of @xmath58 , @xmath59 and @xmath60 are given in fig .
@xmath61 and @xmath62 are local spin - up and spin - down density , @xmath63 is the fermi distribution function , and @xmath64 is the exchange enhancement factor . due to the charge neutrality condition ( @xmath65 ) , the magnitude of the sf has an upper limit .
both @xmath51 and @xmath52 have maximum value of @xmath66 . for the unpolarized 3_d
_ band the sf attain the maximum amplitude at a temperature @xmath67 satisfying condition @xmath68 , whereas for the fully polarized 3_d _ band the corresponding condition is @xmath69 . at higher temperatures
the itinerant sf behave as localized magnetic moments with fixed magnitude .
the corresponding sf resistivity follows from eqns .
[ np ] and [ pol ] : @xmath70 for the non - polarized ( paramagnetic ) state , and @xmath71 for the polarized ( magnetically ordered ) 3_d _ band .
expression for @xmath72 gives a roughly correct representation of the overall temperature dependence of the sf resistivity of paramagnetic _
r_co@xmath5 compounds : it increases as @xmath73 at low temperatures and saturates to a temperature - independent value at high temperatures @xcite .
the model of 3_d _ dos : * a. * non polarized state at _
t_=0 k ; * b. * polarized state at _
t_=0 k. 2@xmath58 is the 3_d _ band width , @xmath59 is the distance from fermi energy to the top of 3_d _ band and @xmath60 dos magnitude . ] both , @xmath72 and @xmath74 are displayed in fig .
[ rmod ] . schematic temperature dependence of normalized @xmath41 for polarized ( @xmath75 ) and non - polarized ( @xmath72 ) states of the 3_d _ band and temperature dependence of resistivity jump at first order transition from non - polarized to polarized ( ferromagnetic ) state @xmath76 .
the circles show the experimental difference between the normalized resistivities of paramagnetic yco@xmath5 and ferrimagnetic tbco@xmath5 @xcite .
the temperature is normalized to @xmath77 of tbco@xmath5 = 214 k @xcite . t@xmath78=@xmath79 . ] in _ r_co@xmath5 compounds with a first - order magnetic phase transition , the 3_d _ band undergoes at @xmath11 a transition from the non - polarized to the polarized state ( shown schematically by the arrows in fig .
[ rmod ] ) , with corresponding change of the sf resistivity : @xmath80 which is also presented in fig .
[ rmod ] . for comparison the experimental difference between the normalized resistivity of paramagnetic yco@xmath5 @xcite ( a representative of a non - polarized state of the 3_d _ band ) and of ferrimagnetic tbco@xmath5 @xcite ( a representative of the polarized state of the 3_d _ band ) is shown on this figure too . in spite of very schematic model , used to calculate @xmath12 ,
there is close correspondence between theoretical and experimental curve .
the experimental resistivity jump observed at @xmath11 of _ r_co@xmath5 and ho(co@xmath13al@xmath14)@xmath5 is presented in fig .
[ delrho ] .
@xmath81 is shown for pure tmco@xmath5 @xcite , erco@xmath5 , hoco@xmath5 , dyco@xmath5 @xcite compounds and for ho(co@xmath13al@xmath82)@xmath5 alloys .
also shown are the results for erco@xmath6 , hoco@xmath5 and dyco@xmath5 under pressure up to 16 kbar @xcite .
the resistivity drop is plotted against @xmath83 to account for the different width of co-3_d _ band . the resistivity drop @xmath12 at curie temperature of ho(co@xmath13al@xmath14)@xmath5 alloys and of some _ r_co@xmath5 compounds @xcite .
broken line is theoretical @xmath84 , scaled with @xmath85=100 @xmath86 cm .
the gray symbols stand for data under pressure for corresponding compounds @xcite . ]
we do not have experimental data on thermopower of all _ r_co@xmath6 under pressure , therefore the same value of @xmath87 k / kbar was used to calculate @xmath10 for compounds under pressure , shown in fig .
[ delrho ] .
@xmath88 was obtained experimentally for a ( gd@xmath89y@xmath90)co@xmath5 alloy @xcite . as it can be seen from the fig .
[ delrho ] there is qualitative agreement between theoretical and experimental values of @xmath91 the model shows two important features in the resistivity behavior : the saturation at high temperatures , and the decrease of @xmath40 with increasing @xmath11 .
the important parameters of the model are the bandwidth 2@xmath58 and the distance from the fermi level to band edge @xmath92 .
they can be estimated in the following way .
the empty portion of the 3_d _ band can be evaluated from the experimentally observed saturation co - moment _
m_=1 @xmath93/co for _ r_co@xmath5 compounds . from this value
it follows that @xmath94 .
using the experimentally observed temperature range in which @xmath95 is non - zero ( fig .
[ delrho ] ) as the measure for @xmath92 ( 200
k ) , and taking @xmath96 @xcite ( which gives @xmath97 ev ) , we get for co3_d _ bandwidth 2_w _ about 2.4 ev , in a very reasonable agreement with the value of 2.5 ev , according to the result of ref .
@xcite .
the model dos used here to demonstrate the mechanism of @xmath98 decrease with increasing @xmath11 , is obviously too oversimplified to account for the very details of @xmath99 and @xmath100 .
the omission of 4_f _
3_d _ exchange interaction is less obvious .
this effect , however , is clearly visible , if one compares @xmath101 of paramagnetic and magnetic _ r_co@xmath5 compounds , depicted in fig .
[ rco2-res ] .
normalized resistivity of _
r_co@xmath5 compounds @xcite and ho(co@xmath102al@xmath103)@xmath5 alloy ( @xmath30 ) .
the solid line is the resistivity of yco@xmath5 .
@xmath104 . * @xmath105 * and * @xmath106 * indicate resistivity at @xmath11 , including corrections for @xmath107 and @xmath108 in paramagnetic and ferrimagnetic phase , respectively . ] for a hypothetical magnetic _ r_co@xmath5 compound ,
whose resistivity includes only sf , phonon , and impurity contributions , its normalized resistivity above @xmath11 would coincide with the normalized resistivity of paramagnetic yco@xmath5 , while below @xmath11 it should be similar to the normalized resistivity of tbco@xmath5 .
however , as we can see from fig .
[ rco2-res ] , the normalized resistivity of the other magnetic _ r_co@xmath5 compounds deviates on approach to @xmath11 from the resistivity of tbco@xmath5 at @xmath22 and from the resistivity of yco@xmath5 at @xmath109 . in part
, this deviation refers to additional contributions owing to scattering on 4_f _ magnetic moments in magnetic _ r_co@xmath5 . at @xmath109 , the corresponding contribution , @xmath107 ,
is independent of temperature .
therefore , the normalized resistivity of a magnetic compound : @xmath110 differs from the normalized resistivity of yco@xmath5 . below @xmath11 , instead of @xmath107 there is a temperature
dependent 4_f _ spin - wave contribution to the resistivity , @xmath108 .
it increases with temperature , reaching at @xmath11 a maximum value . in ferromagnetic metals with second - order phase transition , @xmath111 , however in _
r_co@xmath5 with first - order transition it should be strongly suppressed by the 3_d _ exchange field just below the transition .
the estimated resistivity at @xmath11 , including contributions of @xmath107 or @xmath108 is shown in fig .
[ rco2-res ] . for these estimations
we use @xmath112 , with @xmath107 from ref .
corrections due to @xmath108 are certainly overestimated .
nevertheless , it is clear from fig .
[ rco2-res ] that these corrections can not account for the observed deviations , especially above @xmath11 .
the origin of these deviations is the mutual interaction between the 4_f _ and the 3_d _ magnetic systems .
below @xmath113 , the spin - wave excitations in the 4_f _ system amplify the itinerant sf of the 3_d _ polarized band , leading to strong increase of the resistivity on approaching @xmath11 . in the paramagnetic state
there is a similar mutual amplification of 4_f _ and 3_d _ sf on approach to @xmath11 from higher temperatures
. this amplification increases with decreasing @xmath11 since the susceptibility of the localized 4_f _ moments increases with decreasing temperature .
@xmath9 above @xmath11 is related to dos features in vicinity of the fermi energy .
results for ho(co@xmath13al@xmath82)@xmath5 alloys provide further support to the model outlined in our previous publications @xcite .
we use similar approach and the model dos , depicted in fig .
[ dos ] by the dotted line , to explore the dependence of the thermopower on co 3_d _ band parameters . according to mott s _ s__d _ model , the conductivity is due to high mobility _
s_electron states , from which the charge carriers are scattered into 3_d _ states with low mobility and , effectively , are eliminated from conduction process . the scattering probability of conduction electrons is proportional to the state density value , into which an electron is scattered . therefore , within _
s__d _ model , the _
s_electron scattering probability @xmath114 is expected to depend on the electron energy as : @xmath115 where @xmath116 is due to other than _ s__d _ scattering transitions .
this relation is valid for elastic or quasi - elastic scattering .
thermopower is expressed as @xcite : @xmath117 where @xmath118 and @xmath119 where @xmath120 is the velocity of conduction electrons , @xmath121 is the cross - sectional area of the fermi surface , perpendicular to the electrical field direction .
@xmath122 is the relaxation time . in the _
s__d _ model approximation @xmath123 using equations ( [ s][tau ] ) @xmath9 is calculated with the model dos , shown in fig .
the results of the calculations are shown in fig .
[ s - tcalc ] .
the calculations were performed using different values of two parameters describing the model dos : @xmath27 is the width of the 3_d _ dos peak , and @xmath28 - is the distance from fermi energy to the center of the peak .
= 0.16 ev , the distance from the fermi energy to the center of the dos peak , @xmath124 , varies , simulating the de - population of the 3_d _ band . * b. * the occupation of the 3_d _ band @xmath124 is fixed at -0.03 ev , the width of the 3_d _ peak varies . * c. * both , @xmath27 and the absolute value of @xmath28 decrease .
the arrows indicate position of the thermopower minimum . ]
two important conclusions follow from these results : + 1 .
the position of the thermopower minimum depends mainly on the 3_d _ peak width ; + 2 .
the de - population of the 3_d _ band affects the magnitude of the high temperature thermopower , however has almost no effect on the position of the thermopower minimum .
comparison of the theoretical result of fig .
[ s - tcalc]c and experimental data on thermopower ( fig . [ thpw ] ) indicates that both effects de - population and narrowing of 3_d _ band , play an important role determining @xmath9 in the paramagnetic temperature range . both also lead to an increase of 3_d _ dos at fermi level . from the variation of parameters @xmath124 and
@xmath125 we can roughly estimate that the narrowing of 3_d _ band increases dos at fermi level by more than 50% , while de - population by less than 20% .
therefore , this estimation support our conjecture , that narrowing of 3_d _ band due to disorder plays dominant role in the magnetic enhancement of _
r_(co@xmath13_m_@xmath14)@xmath5 alloys .
the calculations of thermopower were made under assumption of elastic scattering , which is valid at high temperatures , i.e. above about 100 k. at lower temperatures scattering on sf is non elastic .
the distinction between elastic and non - elastic scattering is essential in case of thermopower ( however it is not important for resistivity ) . for elastic scattering
the change of conduction electron energy satisfies the condition : @xmath126 where @xmath127 and @xmath128 denote the final and initial states , respectively . in non - elastic scattering
the change of the electron energy in a scattering event is of order of @xmath129 . on average , an electron from a state above @xmath130
will be scattered into a state below @xmath130 and vice versa .
therefore , while for elastic scattering ( due to negative @xmath131 values at fermi energy ) @xmath132 , for the non - elastic scattering it can be reversed : @xmath133 , leading to a positive thermopower , instead of negative values for elastic scattering .
the effect of these different scattering regimes is clearly visible in @xmath9 of paramagnetic _
@xcite : with decreasing temperature below about 100 k @xmath9 shows a clear tendency to change sign , which , however , is interrupted by another process , leading to large negative thermopower values at low temperatures .
this another process is likely caused by sf drag effect .
@xmath9 of ho(co@xmath13al@xmath14)@xmath5 at @xmath22 exhibits a complicated variation as obvious from fig .
[ lowtr ] .
low temperature thermopower of ho(co@xmath13al@xmath14)@xmath5 .
the solid lines represent the nordheim - gorter expression ( [ stot ] ) with parameters , listed in table [ tab1 ] . ]
data imply that these temperature dependencies result from an interplay between impurity ( elastic ) and spin wave ( non - elastic ) scattering . for diluted alloys at low temperatures ,
the impurity scattering constitutes the main contribution to thermopower .
this gives rise to negative @xmath9 values , exhibiting an almost linear temperature variation .
at higher temperatures , non - elastic scattering on spin waves becomes important and results in the pronounced positive peak , especially at lower al content . according to the nordheim - gorter rule @xcite , thermopower of a conductor with
two dominating scattering processes , can be expressed as : @xmath134 with @xmath135 the subscripts refer to impurity and spin - wave scattering contributions .
the resistivity related to scattering on spin - waves follows from @xmath136 we assume that both , @xmath137 and @xmath138 , are linear at low temperatures : @xmath139 and @xmath140 based on these assumptions we obtain for the total thermopower : @xmath141 , \label{stot}\ ] ] where @xmath142 is temperature , defined from the condition : @xmath143 .fitted parameters of eqn .
[ stot ] [ cols="^,^,^,^",options="header " , ] expression ( [ stot ] ) was fitted to the low temperature experimental thermopower , using @xmath144 , @xmath145 and @xmath142 as free parameters .
the results are shown in fig .
[ lowtr ] by solid lines , and the parameter values are listed in table [ tab1 ] . for diluted alloys ( _ x_=0 , _
x_=0.005 ) the parameters are reasonable and show expected dependencies on @xmath21 : @xmath144 and @xmath142 increase with @xmath21 , while @xmath145 is independent of @xmath21 . at larger values of @xmath21 , however , the dominant low - temperature scattering mechanism is due to static magnetic disorder , related to the formation of a partially - ordered ground state of the 3_d _ electron system .
this scattering is not temperature independent , which leads to a decrease of @xmath144 and @xmath145 in concentrated alloys .
in conclusion , we have studied electrical resistivity and thermoelectric power of ho(co@xmath13al@xmath14)@xmath5 alloys ( @xmath146 ) for temperatures from 4.2 k to 300 k. the experimental temperature dependencies of thermopower indicate that the width of the co 3_d _ band decreases with al content .
this decrease can not be accounted for only by the increasing lattice parameter of the alloys .
the analysis of experimental results on magnetic and transport properties of ho(co@xmath13al@xmath14)@xmath5 and other _
r_(co@xmath34_m_@xmath14)@xmath5 alloys led us to the conclusion that the principal mechanism , responsible for the narrowing of the 3_d _ dos peak is the localization of the co 3_d _ electron states , induced by disorder in co sublattice of the alloys .
the narrowing of the 3_d _ dos gives main contribution to enhancement of dos at fermi energy and to increase of @xmath11 in magnetic alloys , or a decrease of @xmath7 in paramagnetic alloys .
this new mechanism of a magnetic enhancement of the co 3_d _ band in _ r_co@xmath5-based alloys upon substitution of co by non - magnetic elements reconciles two previous , apparently conflicting , models . the magnitude of the resistivity jump at curie temperature , observable in ho(co@xmath13al@xmath14)@xmath5 and _ r_co@xmath5 compounds , exhibiting first order magnetic phase transition ,
is a non - monotonic function of @xmath11 .
our simple model shows that the non - monotonous variation of the resistivity jump with @xmath11 is due to the saturation of the 3_d _ band sf magnitude at high temperatures .
we postulate that fluctuations of the local magnetic susceptibility in ho(co@xmath13al@xmath14)@xmath5 leads to development of partially ordered ground state of the 3_d _ magnetic subsystem , as it was shown previously for y@xmath13_r_@xmath14co@xmath5 alloys ( _ r _ are magnetic rare earth elements ) .
static magnetic disorder , associated with this partially ordered ground state , induces the huge residual resistivities observed . with increasing temperature the static magnetic disorder
is replaced by dynamic , temperature - induced sf .
the corresponding contributions to the resistivity have similar magnitude .
therefore , the overall temperature variation of the resistivity in the alloys is strongly suppressed . our analysis
demonstrate that the temperature behavior of the thermopower above 100 k is determined by a narrow peak in the 3_d _ dos at the fermi energy .
the complex behavior of the thermopower at low temperatures results from an interplay of elastic impurity scattering and non - elastic temperature - induced sf scattering .
we want to thank dr .
a.yu.zyuzin for stimulating discussions .
we measured resistivity of pure _ r_co@xmath5 compounds ( except tmco@xmath5 ) using new samples , prepared according to the same procedure as the samples of ho(al@xmath82co@xmath13 ) alloys .
these new results agree within experimental uncertainty with earlier data from refs . @xcite . | we present experimental results on thermoelectric power ( _ s _ ) and electrical resistivity ( @xmath0 ) of pseudobinary alloys ho(al@xmath1co@xmath2)@xmath3 ( @xmath4 ) , in the temperature range 4.2 k to 300 k. the work focuses on the effects of static ( induced by alloying ) and dynamic ( induced by temperature ) disorder on the magnetic state and electronic transport in a metallic system with itinerant metamagnetic instability .
spatial fluctuations of the local magnetic susceptibility in the alloys lead to the development of a partially ordered magnetic ground state of the itinerant 3_d _ electron system .
this results in a strong increase of the residual resistivity and a suppression of the temperature - dependent resistivity .
thermopower exhibits a complex temperature variation in both the magnetically ordered and in the paramagnetic state .
this complex temperature variation is attributed to the electronic density of states features in vicinity of fermi energy and to the interplay of magnetic and impurity scattering .
our results indicate that the magnetic enhancement of the co 3_d _ band in _ r_co@xmath5based alloys upon a substitution of co by non - magnetic elements is mainly related to a progressive localization of the co
3_d _ electrons caused by disorder .
we show that the magnitude of the resistivity jump at the curie temperature for _ r_co@xmath5 compounds exhibiting a first order phase transition is a non - monotonic function of the curie temperature due to a saturation of the 3_d_band spin fluctuation magnitude at high temperatures .
hoco@xmath5 belongs to the family of intermetallic _
r_co@xmath6 cubic laves phases ( _ r _ stands for rare earth elements ) which are well known for their outstanding magnetic properties associated with itinerant 3_d _ electron subsystem @xcite .
long - range magnetic order of the 3_d _ electron subsystem of paramagnetic yco@xmath5 and luco@xmath3 can be induced by external magnetic field exceeding a certain critical value @xmath7 .
this critical field was found to be about 70 t for yco@xmath5 @xcite and about 77 t for luco@xmath3 @xcite .
partial replacement of cobalt by aluminium leads to a decrease of @xmath7 and to the appearance of weak itinerant ferromagnetism in y(al@xmath1co@xmath2)@xmath3 for @xmath80.12 @xcite . in the case of magnetic _ r_co@xmath5 compounds ,
long - range magnetic order of the 3_d _ electron subsystem is induced by ordering of the localized magnetic moments of the rare earths .
a substitution of al on co - sites of the magnetic _ r_co@xmath5 compounds , results in a considerable increase of the curie temperature @xcite .
this behavior was interpreted within the framework of the itinerant magnetism by an increase of the density of states ( dos ) at the fermi level as co is replaced by al .
however , the mechanism , responsible for the increase of dos at the fermi level , remains obscure .
mainly two scenarios have been proposed : in the first , dos increases due to de - population of 3_d_-co band when co is replaced by al @xcite ; in the second , the dos increase is caused by a narrowing of the 3_d _ band due to an expansion of the crystal lattice in the alloys @xcite . in the present work we use the temperature dependent thermopower @xmath9 to study the mechanism of the dos enhancement in al - substituted _ r_co@xmath6 compounds .
it has been shown that @xmath9 high - temperature minimum , observed in @xmath9 of nearly all _ r_co@xmath5 compounds in a range from 150 k to 400 k , is associated with a sharp peak in the density of states related mainly to the co 3__d - electron density @xcite .
particularly , it was shown that the temperature @xmath10 , at which the thermopower has this minimum , is a measure of the dos peak width @xcite .
another unsolved problem , related to metamagnetism and charge transport in _
r_co@xmath5 compounds is the different order of the phase transition at curie temperature ( @xmath11 ) and a non - monotonic variation of the resistivity jump ( @xmath12 ) at @xmath11 . among the
_ r_co@xmath5 compounds , erco@xmath5 , hoco@xmath6 and dyco@xmath5 undergo a first order phase transition at @xmath11 , whereas it is generally accepted that in tbco@xmath5 , gdco@xmath5 , tmco@xmath5 and smco@xmath5 the magnetic ordering is via second order transition .
the kind of the phase transition at @xmath11 in prco@xmath5 and ndco@xmath5 has been a subject of recent discussion @xcite . in this context
the ho(al@xmath1co@xmath2)@xmath3 alloys are proper materials to study , since this alloy system includes compounds with both , first order , and second order transition at @xmath11 . |
red - sequence galaxies are an important population for understanding galaxy formation and evolution in general .
they represent systems with very little or no on - going star - formation and thus are a unique tracer of the past activity of galaxies .
@xcite found a deficit of faint red - sequence galaxies in clusters at [email protected] relative to local clusters .
this would suggest that star formation ended earlier for the most luminous / massive galaxies at high redshift and would support the `` down - sizing '' picture first proposed by @xcite , in which the termination of star - formation progresses from the most massive to the least massive galaxies as the universe ages . similar results for clusters spanning a range of redshifts have been found by @xcite , @xcite , @xcite , @xcite and others .
similarly , surveys of field galaxies have attempted to characterise the evolution of the red - sequence luminosity function .
the relative contributions of passive evolution / termination of star - formation and number density evolution / dry merging are still open questions ( e.g. , @xcite , @xcite ) .
red - sequence galaxies can thus place constraints on galaxy formation prescriptions in theoretical models .
recent observations in the local universe using sdss data @xcite have found that the fractions of satellite galaxies which are red are too high in current semi - analytic models . with the recent abundance of surveys targeting red - sequence galaxies both in the cluster and the field ,
the time is now ripe to draw together these works to build a picture of the build - up of galaxies on the red - sequence as a function of environment and to confront these with predictions of galaxy formation models .
we aim to construct the most uniform sample possible using colour - selected red - sequence galaxy data from the literature . to this end
, we transform all our data onto the same system .
a simpler quantity than the luminosity function ( lf ) , which has been employed in the past , is the red - sequence dwarf - to - giant ratio , hereafter dgr .
this is essentially just a lf reduced to two bins and avoids the complications of having to fit an analytic function ( usually with degeneracies between the fitted parameters ) to the data .
following @xcite , we define luminous , or giant , galaxies as those brighter than @xmath1 and faint , or dwarfs , as those with @xmath2 ( k- and evolution - corrected ) .
where data are presented on a different photometric system , we use @xcite stellar population models to calculate the transformation using a @xmath3 , solar metallicity , single stellar population with a chabrier imf . where dgrs have been given in the original paper using these limits ( as is the case for much of the cluster data ) we use these numbers and their associated errors directly .
for other works , we take the @xmath4 and @xmath5 values from the schechter function fit , transform to @xmath6 and integrate the transformed lf to calculate dgr .
we estimate conservative errors using a monte carlo method taking the errors in the fitted parameters and allowing for their covariance by employing the error ellipse given in the original reference .
in addition we propagate a 0.05 magnitude uncertainty ( discussed below ) in the magnitude limits to allow for any residual mismatch in the converted photometric system . for the field samples ,
we carefully examine all of the schechter function fits in each redshift bin and only use measurements where the data fully cover the dwarf and giant range .
for example , the ( * ? ? ?
* noaodwfs ) data do a good job of constraining the shape of the lf over the magnitude range we require for [email protected] but at higher redshift are insufficiently deep to constrain the number of dwarfs .
conversely , the ( * ? ? ?
* vvds ) data are sufficiently deep to measure the faint end of the lf at higher redshift , but at [email protected] are subject to considerable uncertainty around @xmath4 , presumably due to the relatively limited area of the survey .
the remainder of our field sample uses points from ( * ? ? ?
* sdss ) , ( * ? ? ?
* sdss ) , ( * ? ? ?
* combo-17 ) , ( * ? ? ?
* millennium galaxy catalogue ) and ( * ? ? ?
* cosmos ) .
1 shows dgr as a function of redshift for these surveys .
where the redshift ranges of different surveys overlap , we find good agreement in our measured dgrs ( e.g. , vvds and cosmos at @xmath8 ) .
clusters show considerable scatter in their individual dgrs ( e.g. , @xcite ) and so we only focus on works measuring dgr from ensemble averages of clusters in each redshift bin ( @xcite , @xcite , @xcite , @xcite , @xcite ) .
two results are immediately apparent from the observational data in fig . 1 : the dgr of field red - sequence galaxies is always lower than that of cluster galaxies at the same redshift , and the dgr of both samples evolves with redshift .
we parameterise the evolution of the dgr with a fit of the form @xmath9 finding @xmath10 and @xmath11 for the cluster and field samples respectively .
the first systematic to be addressed is that of transforming published photometry on different systems to a uniform one .
most of the published cluster studies give rest - frame @xmath12 magnitudes ( e.g. , @xcite ) , and most field studies give rest - frame @xmath13 ( e.g. , @xcite ) .
this is not exclusively the case , however , and we find good agreement where different studies measured in different passbands in the same environments at similar redshifts overlap ( e.g. , * ? ? ?
* ; * ? ? ?
thus we believe the transformation is not a large source of error .
we have propagated a generous 0.05 magnitude error through our analysis to allow for this .
this also allows for the dominant transformation uncertainty which is that of the systematic calibration in the individual surveys ( e.g. , hansen et al .
2007 quote their calibration uncertainty as 0.03 mag ) .
in fact , comparing @xcite and @xcite , the measured dgrs agree within the errors even if this additional uncertainty is neglected .
so , we believe that our error bars are ample .
the important other systematic for field studies is cosmic variance .
the authors of the various works included here discuss the size of this effect for their surveys , and this is not likely to be a major contribution to the error bars in fig . 1 .
cluster studies require more care in the sample selection to minimise systematic errors .
as found by e.g. , @xcite and @xcite , the faint end slope of the cluster rslf depends on cluster richness / mass ( at least for [email protected] ) , in the sense that low mass clusters / rich groups are expected to have systematically lower dgrs . in this compilation ,
we have attempted to utilise the richest / most massive systems at each redshift for the cluster sample , to maximise the difference between the clusters and the field . in
so much as , on average , the most massive systems at one redshift evolve into the most massive systems at lower redshift , such a selection is appropriate for evolutionary studies .
if we consider the three data points nearest to [email protected] , these were drawn from very differently selected samples : a statistical sample of 54 red - sequence selected clusters , having richnesses corresponding to a mass limit of @xmath15m@xmath16 @xcite ; a sample of 11 clusters selected via their unresolved cluster light from shallow optical imaging , having various masses @xcite ; and a sample of ten of the most x - ray luminous clusters @xcite .
the good agreement of the points from these ensemble clusters suggests that the dgr is relatively insensitive to mass or selection method over this mass range .
furthermore , it suggests that the evolution of the cluster dgr is likely not due to a systematically changing mass limit with redshift .
additionally , the virial radius of a cluster depends on its mass roughly as @xmath17 ( e.g. , * ? ? ?
* ) , so using a fixed physical radius within which to measure a property of a cluster probes a different fraction of the cluster s extent depending on the mass of the cluster .
the effect of this on dgr is small as can be seen from fig . 5 of @xcite .
by limiting the mass range of the clusters to the most massive systems , we have minimised the possible impact of such an effect , since the difference between a fixed physical radius and a fixed fraction of the virial radius only depends on the mass to the @xmath18 power .
@xcite recently argued , based on a small number of individual clusters , that the cluster dgr does not appear to evolve with redshift .
we indicate his data points on fig . 1 , omitting the three z@xmath191 clusters for which the data do not bracket the 4000 break .
all of the individual clusters are in reasonable agreement with the ensemble averages , and with our best fit line , except for one at @xmath20 .
furthermore , his sample only contains two @xmath21 clusters , one of which is coma , for which the dgr is in good agreement with the @xcite value .
given the significant cluster - to - cluster scatter , the ensemble samples compiled here provide a better reflection of the global average in clusters , and allow us to measure significant evolution with redshift .
@xcite and @xcite both constructed composite samples of local sdss clusters from the c4 cluster catalogue @xcite .
however , @xcite found a low dgr , comparable with their two higher redshift cluster [email protected] and [email protected] ) and therefore concluded that the cluster dgr might not evolve with redshift .
both of their higher - z individual clusters lie on the best - fit relation of fig . 1 , so the discrepancy arises from the z@xmath00 point .
@xcite found a higher dgr ( for their @xmath23 600 km s@xmath24 sample ) more consistent with the results of @xcite for the richest clusters ; @xcite for the most x - ray luminous clusters ; and also for the regions of highest density in the sdss @xcite ( open square ) .
we suggest that the reason for the low dgr measured by @xcite may be that their sample was dominated by lower mass systems .
we note that their measurement is just consistent with the lowest dgr point at [email protected] in fig . 1 which comes from @xcite and comprised a set of 57 abell clusters over a range of richnesses .
we draw samples from the millennium simulation using the semi - analytic prescription of @xcite , hereafter b06 .
galaxies are selected based on their halo mass ( @xmath1910@xmath25 m@xmath16 for clusters ) and their observed colours and magnitudes in this prescription .
red - sequence galaxies were selected by fitting a gaussian to the red peak of the @xmath26 ) colour distribution and locating the local minimum on the blue side to isolate the extent of the red - sequence . in order to capture the uncertainties in sampling the simulation in the same way as the observations , we vary the slope and normalisation of this red - sequence limit ( by 15% and 3% respectively ) as well as the magnitude limits used to define dgr ( 0.1 mags each limit ) and monte - carlo the results . the points for clusters and field in three redshift bins ( 0.0 , 0.5 , 1.0 ) are shown as asterisks and triangles respectively in fig . 1 .
statistical errors are negligible , so the error bars reflect this systematic variation .
when confronted with the observational data , the semi - analytic model fails to reproduce either the redshift evolution of the dgr or the relative difference between clusters and the field at a given redshift .
the model predicts approximately the correct dgr for clusters in the local universe , but this dgr value then remains constant out to z@xmath01 .
these results are insensitive to the exact halo mass chosen to define the clusters , as can be inferred from the similarity between the cluster and field points at each redshift .
i.e. , the dgr is not a strong function of halo mass in the model .
previous works have suggested that there appears to be little or no evolution in the bright end of the rslf , both in clusters and the field ( e.g. , @xcite ; @xcite ) .
if this is the case , the evolution of the dgr can be attributed to the build - up of faint galaxies on the red - sequence with increasing cosmic time .
this implies that star - formation is terminated in giant galaxies first and later in dwarf galaxies , in agreement with the `` down - sizing '' picture of star - formation .
furthermore , the lower value of dgr in the field compared with clusters implies that star - formation was completed in cluster galaxies before field galaxies . performing a similar analysis on the b06 semi - analytic model predicts a non - evolving dgr out to z@xmath221 , with little difference between clusters and the field .
such a result suggests that faint galaxies move onto the red - sequence too efficiently in the model , with this build - up of the faint end of the red - sequence already having been completed by the earliest epoch considered here ( z@xmath01 ) .
the methods available in the models for the quenching of star - formation include : feedback from supernovae or agn and hot halo gas removal .
this latter process only applies to galaxies which are satellites ( i.e. not the most massive , central object ) in their dark matter halo .
the relative importance of the various quenching mechanisms are summarised in fig . 4 of b06 .
at the bright end of the red - sequence , most of the galaxies are the central object in their dark matter halo and thus feedback from agn or supernova winds are required to turn off star - formation . at the faint end ( in the dwarf regime ) , red - sequence galaxies are predominantly satellites .
we have verified that this is still the case for all the redshifts we consider here . in most galaxy formation models ( e.g. , b06 , @xcite ) ,
when any galaxy enters a more massive dark matter halo and becomes a satellite , it instantly has all of its hot gas removed in a process generally referred to as strangulation
thus , once it has consumed the cold gas already present in its disk , its fuel can no longer be replenished by hot gas from the halo and its star - formation is terminated .
this prescription may be too efficient .
recent simulations by @xcite found that the removal of the initial hot galactic halo gas does not occur instantaneously , but likely takes @xmath0 1 gyr to remove the bulk of it , and that as much as 30% can still remain after 10 gyr .
in addition , the models of @xcite predict the existence of a critical halo mass ( @xmath27 m@xmath16 ) for haloes , below which strangulation may not occur .
thus , we expect additional parameters involving strangulation efficiency / timescale and/or a threshold halo mass for strangulation to be important in tuning the models to reproduce the observations presented in fig . 1 . to illustrate the effect of a threshold halo mass for strangulation , we use the following toy model to predict the evolution of the field dgr .
field surveys do not just probe isolated ( i.e. non - group ) galaxies .
indeed , they are dominated by groups , especially toward lower redshift as cosmic structure grows .
thus it is group - scale physics which is important in determining the properties of galaxies in field surveys .
we assume further that galaxies in haloes below some ( group - size ) threshold mass , @xmath28 , have no red - sequence satellites , i.e. dgr @xmath29 .
we assume that haloes above @xmath28 have a dgr given by that of the observed cluster population at the same redshift ( from the best - fit cluster line in fig . 1 ) .
we consider @xmath30 ) m@xmath16 and measure the fractional contribution of haloes in the appropriate mass range from the millennium simulation .
we overplot the expected evolution of the field dgr for the different threshold masses in fig .
it can be seen that both the @xmath31 and @xmath32 m@xmath16 models give good agreement with the observed field dgr . in order to reproduce the observations so well ,
our model requires both an increasing contribution to the field by groups due to growth of structure _ and _ an intrinsic evolution of the dgr in haloes above @xmath28 .
if either of these factors is removed ( e.g. if a constant group fraction at all redshifts or a constant cluster dgr at all redshifts is assumed ) then the predicted evolution of the field dgr is too flat with respect to the observations .
if groups do not exhibit exactly the same dgr as clusters , but instead are intermediate between clusters and the field ( this is currently an open question ) , then this would effectively shift our predicted lines down and a lower @xmath28 would then reproduce the observations .
of course , a scenario in which the timescale for the removal of hot gas is much longer may fit the observations equally well and this or some alternative combination of threshold mass / timescale can not be ruled out .
indeed , the observed evolution of the cluster dgr we have had to include in our model may reflect the effect of a strangulation timescale . disentangling the effects of a timescale and threshold mass for the quenching of satellite galaxies
is a complex observational problem .
the colour difference between central and satellite red - sequence galaxies may be a useful test of models .
for example , in the b06 prescription , it is predicted that satellite red - sequence galaxies are significantly redder than central red - sequence galaxies .
@xcite recently measured the average colours of satellite and central galaxies in the local universe from an approximately halo mass - selected sample and found that the former were indeed on average redder .
they also showed that measuring halo masses from statistical samples for @xmath010@xmath33 m@xmath16 systems is not possible with high completeness and thus , if there is a threshold mass for strangulation around this limit , such studies will require detailed targeted observations of groups ( e.g. , @xcite ) . an alternative approach to get at the timescale for strangulation is the study of the star formation histories of individual galaxies undergoing transformation ( e.g. , @xcite ) .
it is possible that the observations in fig . 1 , coupled with new strangulation prescriptions in the semi - analytic models may improve constraints on the likely values of the threshold mass and timescale for strangulation
this research was supported by an early researcher award from the province of ontario , and by an nserc discovery grant . | red - sequence galaxies record the history of terminated star - formation in the universe and can thus provide important clues to the mechanisms responsible for this termination .
we construct composite samples of published cluster and field galaxy photometry in order to study the build - up of galaxies on the red - sequence , as parameterised by the dwarf - to - giant ratio ( dgr ) .
we find that the dgr in clusters is higher than that of the field at all redshifts , implying that the faint end of the red - sequence was established first in clusters .
we find that the dgr evolves with redshift for both samples , consistent with the `` down - sizing '' picture of star formation .
we examine the predictions of semi - analytic models for the dgr and find that neither the magnitude of its environmental dependence nor its evolution is correctly predicted in the models .
red - sequence dgrs are consistently too high in the models , the most likely explanation being that the strangulation mechanism used to remove hot gas from satellite galaxies is too efficient . finally we present a simple toy model including a threshold mass , below which galaxies are not strangled , and show that this can predict the observed evolution of the field dgr .
[ firstpage ] galaxies : clusters : general galaxies : evolution galaxies : general |
as discovered by swift , there is a shallow decay phase ( temporal indices @xmath0 $ ] ) from post - burst several tens of seconds to several hours ( even days ) occurring in the x - ray afterglow light curves of a significant fraction of gamma - ray bursts ( grbs ) ( nousek et al .
2006 ; obrien et al . 2006 ; willingale et al . 2007 ; liang et al . 2007 ) .
this shallow decay phase is obviously beyond understanding of the standard afterglow model ( mszros & rees 1997 ; sari et al .
1998 ) , but is generally accepted to be due to continuous energy injection into relativistic blast waves ( e.g. , zhang et al .
2006 ; nousek et al .
2006 ; fan & xu 2006 ; sollerman et al . 2007 ; de pasquale et al . 2006 , 2007 ) .
in particular , liang et al .
( 2007 ) recently compared the closure relations derived from a simple energy - injection form , @xmath1 , with the observed temporal and spectral indices of afterglows of 53 grbs and then argued that a roughly constant injection luminosity may be favored by the observations .
however , although this simple injection form is usually assumed in model fitting , the specific nature of the energy injection should be concerned and studied further .
the physical nature of an injecting energy flow discussed in the literature involves two possible candidates : ( i ) ejecta that consists of a series of shells with a distribution of lorentz factors , which are dominated in energy by baryons ( rees & mszros 1998 ; panaitescu et al .
1998 ; sari & mszros 2000 ; granot & kumar 2006 ; liu et al . 2007 ) , and ( ii ) a post - burst energy flow that results from a long - lasting activity of the central engine ( dai & lu 1998a , 1998b ; zhang & mszros 2001a ; wang & dai 2001 ; dai 2004 ; fan & xu 2006 ; yu & dai 2007 ) . in the former case ,
the total energy carried by the ejecta is released from the central engine during the prompt phase of a grb .
subsequently , during the afterglow phase , lower - velocity shells catch up and collide with foregoing , higher - velocity but decelerated shells continuously because of an assumed power - law distribution of lorentz factors in the ejecta .
this persistent collision leads to a long - lasting newtonian or trans - relativistic reverse shock that propagates into the ejecta .
meanwhile , the shock - heated ejecta pushes the outward - moving relativistic blast wave and thus reduces the deceleration of the blast wave effectively , which accounts for the shallow decay . in the latter case ,
most of the energy that produces an afterglow is argued to be continuously released after ( not during ) the burst , but the grb ejecta may provide only a relatively small amount of energy for the blast wave . initially , the post - burst energy flow may be dominated in energy by the poynting flux . however , as it propagates outward , the flow would evolve to an ultrarelativistic kinetic - energy flow dominated by a component of electron - positron pairs through some mechanisms ( e.g. , magnetic reconnection ) at larger radii ( coroniti 1990 ; michel 1994 ; kirk & skjraasen 2003 ) .
it is this ultrarelativistic leptonic wind ( rather than the electromagnetic flux ) that feeds the blast wave and maintains a long - lasting relativistic reverse shock .
the emission of the more and more energetic blast wave and the emission of the reverse shock together give rise to the shallow decay ( dai 2004 ) . therefore ,
we conclude from the above discussion that the injecting energy flows are likely to be matter - dominated .
furthermore , based on the component that is dominative in energy , the injecting flows perhaps have two types : a baryon - dominated outflow and a lepton - dominated outflow .
the two corresponding representative models as described above are here called the radially structured ejecta ( rse ) model ( rees & mszros 1998 ) and the relativistic wind bubble ( rwb ) model ( dai 2004 ) , respectively . in order to provide an effective test on the models mentioned above ,
a careful investigation of high - energy emission is important and urgent , as the launch and detection of glast are upcoming ( ritz 2007 ) . in this paper , we calculate high - energy afterglow emission during the shallow decay phase of x - ray afterglows based on these two models and give corresponding observational signatures .
recently , wei & fan ( 2007 ) , fan et al .
( 2007 ) , and gou & mszros ( 2007 ) have made some attempts on high - energy afterglows . in their papers
, the authors studied only the effect of the forward - shocked medium with the simple energy - injection form of @xmath1 , but did nt consider the effect of the injecting flow itself . however , as found by dai ( 2004 ) , uhm & beloborodov ( 2007 ) , genet et al .
( 2007 ) , yu & dai ( 2007 ) , and liu et al .
( 2007 ) , the reverse shock that propagates into the injecting flow could play an important role in the emission in x - ray and/or optical bands in some situations .
therefore , high - energy emission features due to the reverse shock are also expected . in the rse and rwb models ,
the high - energy photon emission from shocked materials is mainly produced by inverse compton ( ic ) scattering of shock - accelerated electrons ( for simplicity , electrons and positrons are not differentiated ) off synchrotron seed photons ( sari & esin 2001 ) . besides the synchrotron self - compton ( ssc ) radiation from two shocked regions , we also consider two combined ic ( cic ) processes , i.e. , scattering of reverse shock photons by forward - shocked electrons and scattering of forward shock photons by reverse - shocked electrons , as pointed out by wang , dai & lu ( 2001 ) .
the structure of this paper is organized as follows : in the next section we provide a unified description for the dynamics of fireballs in the rse and rwb models . in 3 , we present the energy distributions of shocked electrons that are determined by the shock - acceleration effect and the synchrotron & ic cooling effect , and formulate calculations of the synchrotron and ic radiation ( including ssc and cic ) . in 4 ,
we show numerical results of the dynamics , spectra and light curves for some typical parameters in the two models .
finally , in 5 , a summary is given and the observability of high - energy emission by the large area telescope ( lat ) instrument of glast is discussed .
as illustrated in figure 1 , in both the rwb and rse models , the system can be divided into four regions by the forward and reverse shocks : ( 1 ) the unshocked ambient medium , ( 2 ) the forward - shocked medium , ( 3 ) the reverse - shocked materials ( i.e. , shocked leptonic wind for the rwb model or shocked grb ejecta for the rse model ) , and ( 4 ) the unshocked cold wind or grb ejecta , where regions 2 and 3 are separated by a contact - discontinuity surface .
figure 1 also shows illustrative @xmath2-distributions in all the regions .
the main differences between the two models are in the structure and composition of injecting flows : \(i ) _ the rwb model_. following the analysis of dai ( 2004 ) for a magnetar - driven wind , we simply assume that the leptonic wind propagates outward with a constant luminosity @xmath3 and a constant bulk lorentz factor @xmath4 during a period of @xmath5 after the burst with a prime is measured in the comoving frame of a certain region , while the other quantities @xmath6s are measured in the local medium s rest frame except for specific declarations .
in addition , a subscript @xmath7 " represents a quantity describing region @xmath8 " . ] .
thus , we can calculate the number density ( @xmath9 ) and the bulk lorentz factor ( @xmath10 ) of the pre - shock materials at the reverse shock front as @xmath11 @xmath12 where @xmath13 is the rest mass of electrons and @xmath14 is the radius of the system in the thin shell approximation .
\(ii ) _ the rse model_. as suggested by rees & mszros ( 1998 ) , the mass distribution in the wide grb ejecta associated with a distribution of bulk lorentz factors reads @xmath15 moreover , we assume that the @xmath16-distribution in the grb ejecta satisfies @xmath17 where @xmath18 is the displacement of the reverse shock propagating into the ejecta .
if the lorentz factor of the end of the ejecta ( i.e. , the minimum bulk lorentz factor ) is denoted by @xmath19 , then the total displacement of the reverse shock when it crosses the ejecta could be estimated as @xmath20 cm . under these assumptions ,
the associated injecting energy distributes with respect to @xmath18 as @xmath21 , and @xmath22 and @xmath23 are given by @xmath24 @xmath25 where @xmath26 is the proton rest mass .
now , we describe the dynamic evolution of the systems under the effect of the two types of injecting flow .
a resultant long - lasting reverse shock transforms the injecting energy into the internal energy of the reverse - shocked materials continuously .
meanwhile , a part of fresh energy is further transformed to the kinetic energy ( @xmath27 ) of region 2 through the work done by region 3 : @xmath28 , where @xmath29 is the pressure of region 3 .
then , we have @xmath30,\ ] ] where @xmath31 is the rest mass of the swept - up medium and @xmath32 is the average bulk lorentz factor of the shocked medium .
thus , the dynamic evolution of region 2 is described by ( dai 2004 ) @xmath33\over2\gamma_2m_{\rm sw}},\label{g2}\ ] ] where @xmath34 is applied and @xmath35 is the proton number density of the ambient medium . to integrate the above equation ,
the pressure of region 3 should be calculated by @xmath36where @xmath37 is the lorentz factor of region 3 measured in region 4 and @xmath38 is the velocity in units of @xmath39 , and @xmath40 represents the electron and proton rest masses for the rwb and rse models , respectively .
the evolution of @xmath41 can be obtained from the equality of lorentz factors of the two sides of the contact discontinuity surface , assuming that the shocked medium in region 2 satisfies the adiabatic self - similar solution of blandford & mckee ( 1976 ) @xmath42 to fix the self - similar variable @xmath43 , we use the following relationship , @xmath44 combining eqs . [ p31 ] , [ g3 ] , and [ p32 ] to eliminate @xmath43 , we can solve @xmath41 and @xmath29 as functions of @xmath32 and thus integrate eq . [ g2 ] .
furthermore , the rest mass of region 3 is obtained from @xmath45where @xmath46 is the velocity of the reverse shock . when the reverse shock crosses the wind or grb ejecta , the dynamic evolution of region 3 is described by kobayashi & sari ( 2000 ) and kobayashi ( 2000 ) .
then , the pressure of region 3 in eq . [ g2 ] decreases significantly and thus is negligible . as a result ,
the dynamic equation of region 2 returns to the form of the standard afterglow model in huang et al .
after the dynamic evolution equations are given , the internal physics of the shocked regions in the rwb and rse models can be considered as follows .
as the forward and reverse shocks propagate , the bulk kinetic energies of the shells are gradually transformed into the internal energy of the shocked materials , the density of which is denoted by @xmath47 .
the internal energy will be partially carried by the accelerated electrons and magnetic fields , the energy densities of which are factions @xmath48 and @xmath49 of @xmath47 , respectively . through the shock acceleration , the electrons behind the forward / reverse
shock will obtain an initial energy distribution @xmath50 with the minimum lorentz factor @xmath51\epsilon_{{\rm e},i}(\tilde{\gamma}_i-1)$ ] and the maximum lorentz factor @xmath52^{-1/2}$ ] , where @xmath53 , @xmath54 , @xmath55 , @xmath56 , and @xmath57 is the magnetic field strength .
the occurrence of the compton parameter @xmath58 is induced by the ic cooling effect . by considering the cooling effect of the synchrotron and ic radiation , we can obtain the actual electron energy distributions
( huang et al . 2000 ) : + for @xmath59 , @xmath60 and for @xmath61 , @xmath62 the cooling lorentz factor @xmath63 is defined by equaling the cooling timescale and dynamic expansion timescale of the system , and it reads @xmath64 with @xmath65 being the thomson scattering cross section . here , the dynamic timescale measured in the observer s frame in calculating @xmath66 is assumed to be approximately equal to the observer s time @xmath67 since the trigger .
in addition , the self - absorption of the synchrotron photons is ignored in eqs .
[ ng1 ] and [ ng2 ] for the high - energy emission that we are interested in . following sari & esin ( 2001 ) , the compton parameter @xmath58 in eq .
[ gc ] , defined as the ratio of the ic ( including ssc and cic ) luminosity to the synchrotron luminosity of electrons , is estimated by @xmath68 @xmath69 where @xmath70 and @xmath71 are the energy densities of synchrotron seed photons and magnetic fields , respectively .
the radiation efficiency @xmath72 reads @xmath73 a factor of @xmath74 in eqs .
[ y2 ] and [ y3 ] occurs , because only about one - half of seed photons from one shocked region will diffuse into the other one for the cic process . in two extreme situations , equations [ y2 ] and [ y3 ]
can be simplified as + i ) for @xmath75 , @xmath76 ii ) for @xmath77 , @xmath78 once the electron distribution is known , the synchrotron emissivity of electrons in region @xmath8 at frequency @xmath79 is calculated directly by ( rybicki & lightman 1979 ) @xmath80 where @xmath81 is the electron charge , @xmath82 , @xmath83 , and @xmath84 is the bessel function . accompanying with the synchrotron radiation ,
the electrons also lose their energy through upscattering the synchrotron seed photons . as usual , the first - order ic scattering is considered and the higher - order processes are neglected .
according to blumenthal & gould ( 1970 ) , when the klein - nishina suppression is considered , the ic emissivity ( at a frequency @xmath79 ) of electrons in region @xmath8 upscattering seed photons from region @xmath85 is calculated by ( @xmath86 for ssc and @xmath87 for cic ) @xmath88 where @xmath89 with @xmath90 , @xmath91}$ ] .
@xmath92 and @xmath93 are the frequency and the corresponding flux density of the incident photons from region @xmath85 , respectively , which are measured in the comoving frame of region @xmath8 .
the observed synchrotron and ic flux densities at a frequency @xmath94 ( measured in the observer s frame ) from region @xmath8 are given respectively by ( huang et al .
2000 ) @xmath95 @xmath96 where @xmath97 is the luminosity distance and @xmath98 is the doppler factor . here
, the viewing angle is assumed to be zero and @xmath99 is the angle between the moving direction of the emitting material and the line of sight .
the volumes @xmath100 of the shocked regions in eqs .
[ fsyn ] and [ fic ] are functions of @xmath99 because the integrations are performed over the equal - arrival surface .
in addition , the jet correction should be considered if the grb ejecta is collimated .
for the rwb and rse models , we first calculate the dynamic evolution with radius .
the initial value of @xmath14 is taken to be @xmath101 cm as the deceleration radius .
for the rwb model , as in yu & dai ( 2007 ) , the isotropic wind luminosity is @xmath102 , the wind duration @xmath103 s , and the bulk lorentz factor of the wind @xmath104 , where @xmath105 is the magnetic field strengthen of the central magnetar in unit of @xmath106 g. here , we take @xmath107 for a typical magnetar .
thus the total energy of the wind is @xmath108 erg .
however , the isotropic kinetic energy of the previous grb ejecta is relatively small , @xmath109 erg . for consistency , we assume that , in the rse model , the isotropic kinetic energy carried by the wide grb ejecta is @xmath110 erg , and the distribution - related parameters are @xmath111 , @xmath112 , and @xmath113 . in addition , the number density of the ambient interstellar medium is @xmath114 in both cases .
the dotted lines in figure 2 represent the dynamic evolution of blast waves with a certain energy ( i.e. , without energy injection ) , and the solid lines show the energy injection case .
we can see clearly from this figure that the energy ( @xmath115 ) of the shocked medium increases gradually until the energy injection is over . as analyzed in yu & dai ( 2007 ) , the final energy carried by the shocked medium
is a fraction @xmath116 of the total injecting energy for the rwb model versus @xmath117 for the rse model .
meanwhile , the other fraction of the injecting energy ( @xmath118 and @xmath119 for the rwb and rse models , respectively ) should be shared by the reverse - shocked material .
in addition , we can know that the reverse shock is relativistic in the rwb model , but newtonian or trans - relativistic in the rse model , according to an estimation of the lorentz factor of region 3 relative to region 4 from @xmath120 with the values of @xmath41 as shown in figure 2 .
this difference between the reverse shocks in the two models is just the reason why the reverse shock in the rwb model can share more injecting energy than the one in the rse model . to calculate the emission of shocked materials ,
we take the microphysical parameters @xmath121 and @xmath122 for all shocked regions .
then , @xmath123 is calculated by @xmath124 for the baryon - dominated regions ( i.e. , regions 2 & 3 in the rse model and region 2 in the rwb model ) as argued by medvedev ( 2006 ) and @xmath125 for the lepton - dominated region ( i.e. , region 3 in the rwb model ) .
the luminosity distance of a grb is taken to be @xmath126 gpc , corresponding to a redshift of 0.2 .
figure 3 shows the light curves of an rwb . from this figure
, we obtain the following results .
first , the shallow decay phase of an x - ray ( kev ) afterglow is produced during the early hours ( left upper panel ) , which is mildly dominated by the reverse shock emission ( for a detailed discussion see yu & dai 2007 ) .
the synchrotron radiation is the dominative mechanism in this band .
second , an obvious hump accompanying with the x - ray plateau occurs in a high - energy ( mev & gev ) afterglow , which is dominated by the ic ( especially ssc ) emission from the reverse - shocked wind .
finally , a comparison of light curves in different bands ( right lower panel ) shows that the high - energy ( especially gev ) emission flux is mildly or even significantly higher than the one in x - ray band .
this feature is also implied obviously by the spectra as shown in figure 4 .
the peak flux of the ic emission at @xmath127 gev is about one order of magnitude higher than the synchrotron peak at @xmath128 ev .
moreover , the ic component is dominative above sub - mev and , in all bands , the emission from the reverse - shocked wind is more important than the one from the forward - shocked medium at 1000 s. the results for the rse model are shown in figures 5 and 6 . from these two figures , we find that , first , the x - ray shallow decay phase during the early hours is also produced in this model as shown in left upper panel of figure 5 , which is dominated by the forward shock .
second , the high - energy afterglow light curves also have a flattening segment .
however , differing from the rwb model , the contribution to the high - energy plateau from the ic emission plays a role only in gev band , whereas the mev emission is totally contributed by the synchrotron mechanism during the total afterglow phase .
this result arises from the relative weakness of the ic component in the rse model as shown in figure 6 .
therefore , the flux in mev or gev band is lower than or approximately equal to the x - ray flux as shown in right lower panel of figure 5 .
in addition , figure 6 shows that the contribution of the emission from the reverse - shocked ejecta is negligible in all bands for the parameters that we adopt . in conclusion
, we obtain very different high - energy afterglows in the rwb and rse models with a same amount of injecting energy that gives rise to similar x - ray afterglows .
the rwb model predicts more significant ( about one order of magnitude stronger ) high - energy emission than the rse model .
the reasons for this difference are that the reverse shock in the rwb model is relativistic and that the energy of the reverse shock is almost totally carried by electrons .
the former reason enables the reverse shock to share more ( i.e. , 33% vs 10% ) injecting energy as discussed in sect .
4.1 , and the latter reason dramatically increases the emission efficiency of the reverse - shocked material through the enhanced ic emission .
therefore , a higher fraction of injecting energy can be radiated from the shocked regions in the rwb model . in the baryon - dominated injection model , however ,
this significant enhanced ic component does nt happen even if the emission flux from the reverse shock exceeds the one from the forward shock under some extreme conditions , because most of the injecting energy in both shocked regions is locked in the baryons whose emission is weak
. therefore , we argue that the difference in high - energy emission between the rwb and rse models is essentially due to the physical distinction between the two types of energy injection .
thus , reasonable variations of the parameters for the models should not change our results significantly .
the discovered shallow decay phase of grb x - ray afterglows suggests long - lasting energy injection into relativistic blast waves .
the injecting flow is likely to be dominated in energy by the component of either baryons or leptons . in this paper , we first provide a unified description for dynamics and radiation in two representative models , i.e. , the rse and rwb models . through maintaining a long - lasting reverse shock and doing work to forward - shocked medium , both types of energy injection can produce the flattening segment in x - ray afterglow light curves easily .
second , we pay attention to calculations of the simultaneous high - energy emission that is due to synchrotron and , especially , ic ( including ssc and cic processes ) radiation from the shocked materials .
our results show that , during the shallow decay phase of x - ray afterglows , there is a plateau ( even a hump ) in high - energy light curves in both the rse and rwb models .
as argued by wei & fan ( 2007 ) , we suggest that the plateau / hump might account for the delayed high - energy emission of some bursts such as grb 940217 .
we also find that the high - energy emission derived from the two models has different observable features , e.g. , morphologies of the light curves and spectra .
in particular , more significant high - energy emission is predicted by the rwb model , because more injecting energy in this model is carried by electrons and thus the ic scattering is enhanced significantly .
this difference in high - energy emission between the two models could be tested in the upcoming glast era . as proposed by zhang & mszros ( 2001b ) and gou & mszros ( 2007 )
, the fluence threshold of the lat instrument aboard glast for an effective observation time @xmath129 in an observation energy band centered around a photon energy @xmath130 is estimated by @xmath131et_{\rm eff}$ ] , where the flux sensitivity , @xmath132 $ ] , scales as @xmath133 due to a limitation by the background for a long - time observation and @xmath134 is the sensitivity over an exposure time @xmath135 . specifically , over a one - year exposure period
, the sensitivity above 100 mev is @xmath136 for point - source observations of grb afterglows ( gou & mszros 2007 ) . on the other hand , for a short - time observation
, the fluence threshold is instead calculated by @xmath137 with the assumption that at least 5 photons are collected on an effective area @xmath138 of the instrument .
taking @xmath139 mev and @xmath140 , we obtain the fluence threshold of glast lat , @xmath141 where an observation efficiency @xmath142 due to the occultation by the earth is considered . with this fluence threshold ,
the observability of high - energy emission in the rwb and rse models by glast lat is shown in figure 7 .
it can be seen that both the high - energy emissions predicted by the two models can be detected by glast lat for the given parameters .
the partial fluence of the high - energy emission in the rwb model is about an order of magnitude higher than that in the rse model for a same amount of injecting energy . here
, the partial fluence is defined as an integration of the flux density ( @xmath143 ) over the glast lat energy band [ 20 mev , 300 gev ] and the time intervals [ @xmath144 as done in gou & mszros ( 2007 ) .
although the detection of the rwb model by glast lat is obviously easier , joint observations of glast and swift are necessary to distinguish between the rwb and rse models
. simultaneous observations of a high - energy afterglow by glast lat and an x - ray afterglow by swift xrt for a same grb could provide a ratio of fluxes ( @xmath145 ) in the high - energy band to the x - ray . as implied by the right lower panels of figures 3 and 5 , for example , the injecting flow could be dominated in energy by baryons if the ratio @xmath146 is close to 1 . on the other hand ,
if the ratio is significantly higher than 1 , there could be lepton - dominated energy injection .
we would like to thank lijun gou and an anonymous referee for valuable comments and suggestions that have helped us to improve an earlier version of this manuscript .
this work is supported by the national natural science foundation of china ( grant no .
10221001 and 10640420144 ) .
ywy is also supported by the visiting phd candidate foundation of nanjing university and the national natural science foundation of china ( grant no.10603002 ) .
blandford , r. d. , & mckee , c. f. 1976 , phys . fluids . , 19 , 1130 blumenthal , g. r. , & gould , r. j. 1970 , rev . mod
. phys . , 42 , 237 coroniti , f. v. 1990 , apj , 349 , 538 dai , z. g. 2004 , apj , 606 , 1000 dai , z. g. , & lu , t. 1998a , a&a , 333 , l87 dai , z. g. , & lu , t. 1998b , phys .
lett . , 81 , 4301 de pasquale , m. , et al .
2006 , mnras , 365 , 1031 de pasquale , m. , et al . 2007 , accepted by mnras , arxiv : astro - ph/0703447 fan , y. z. , & xu , d. 2006 , mnras , 372 , l19 fan , y. z. , piran , t. , narayan , r. , & wei , d. m. , 2007 , submitted to mnras , arxiv:0704.2063 genet , f. , daigne , f. , & mochkovitch , r. 2007 , arxiv : astro - ph/0701204 gou , l. j. , & @xmath147 , p. 2007
, accepted by apj , arxiv:0705.1545 granot , j. , & kumar , p. , 2006 ,
mnras , 366 , l13 huang , y. f. , dai , z. g. , & lu , t. 1999 , mnras , 309 , 513 huang , y. f. , gou , l. j. , dai , z. g. & lu , t. 2000 , apj , 543 , 90 kirk , j. g. , & skjraasen , o. 2003 , apj , 591 , 366 kobayashi , s. 2000 , apj , 545 , 807 kobayashi , s. , & sari , r. 2000 , apj , 542 , 819 liang , e. w. , zhang , b. b. , & zhang , b. 2007 , accepted by apj , arxiv:0705.1373 liu , x. w. , wu , w. f. , zou , y. c. , & lu , t. 2007 , to be submitted medvedev , m. v. 2006 , apj , 651 , l9 @xmath147 , p. , & rees , m. j. 1997 , apj , 476 , 232 michel , f. c. 1994 , apj , 431 , 397 nousek , j. a. , et al .
2006 , apj , 642 , 389 obrien , p. t. , et al .
2006 , apj , 647 , 1213 panaitescu , a. , @xmath147 , p. , & rees , m. j. 1998 , apj , 503 , 314 rees , m. j. , & @xmath147 , p. 1998 , apj , 496 , l1 ritz , s. 2007 , a review talk in the first glast symposium ( 5 february 2007 ) rybicki , g. b. , & lightman , a. p. 1979
, radiative processes in astrophysics ( new york : wiley ) sari , r. , & esin , a. a. 2001 , apj , 548 , 787 sari , r. , piran , t. , & narayan , r. 1998 , apj , 497 , l17 sari , r. , & @xmath147 , p. 2000 , apj , 535 , l33 sollerman , j. , et al .
2007 , a&a , 466 , 839 uhm , z. l. , & beloborodov , a. m. 2007 , apj , 665 , l93 wang , w. , & dai , z. g. 2001 , chin .
lett . , 18 , 1153 wang , x. y. , dai , z. g. , & lu , t. , 2001 , apj , 556 , 1010 wei , d. m. , & fan , y. z. 2007 , chjaa , 7 , 509 willingale , r. , et al .
2007 , apj , 662 , 1093 yu , y. w. , & dai , z. g. 2007 , a&a , 470 , 119 zhang , b. , et al . 2006 , apj , 642 , 354 zhang , b. , & @xmath147 , p. 2001a ,
apj , 552 , l35 zhang , b. , & @xmath147 , p. 2001b
, apj , 559 , 110 | the widely existing shallow decay phase of the x - ray afterglows of gamma - ray bursts ( grbs ) is generally accepted to be due to long - lasting energy injection .
the outflows carrying the injecting energy , based on the component that is dominative in energy , fall into two possible types : baryon - dominated and lepton - dominated ones .
the former type of outflow could be ejecta that is ejected during the prompt phase of a grb and consists of a series of baryonic shells with a distribution of lorentz factors , and the latter type could be an electron - positron - pair wind that is driven by the post - burst central engine .
we here provide a unified description for the dynamics of fireballs based on these two types of energy injection , and calculate the corresponding high - energy photon emission by considering synchrotron radiation and inverse compton scattering ( including synchrotron self - compton and combined inverse - compton ) of electrons .
we find that , in the two energy - injection models , there is a plateau ( even a hump ) in high - energy light curves during the x - ray shallow decay phase .
in particular , a considerable fraction of the injecting energy in the lepton - dominated model can be shared by the long - lasting reverse shock since it is relativistic .
furthermore , almost all of the energy of the reverse shock is carried by leptons , and thus the inverse - compton emission is enhanced dramatically .
therefore , this model predicts more significant high - energy afterglow emission than the baryon - dominated model .
we argue that these observational signatures would be used to discriminate between different energy - injection models in the upcoming _ gamma - ray large area space telescope _ ( glast ) era . |
a finite vector configuration @xmath6 defines the _ toric ideal _ @xmath7 in the polynomial ring @xmath8 = { \mathbb k } [ { \bf x}]$ ] where @xmath9 is a field .
let @xmath10 and @xmath11 denote the cone , lattice and semigroup spanned by the @xmath12 and @xmath13-linear combinations of @xmath0 where @xmath13 is the set of non - negative integers .
let @xmath14 be the krull dimension of @xmath15 which equals the rank of @xmath16 .
assume @xmath17 .
the configuration @xmath0 is _ normal _ if @xmath18 and _ graded _ if @xmath0 spans an affine hyperplane in @xmath19 . a finite set @xmath20 such that @xmath21 is called a _ hilbert basis _ of the semigroup @xmath22 .
if @xmath0 is normal , the zero set of @xmath3 is a normal toric variety in @xmath23 of dimension @xmath24 , and when @xmath0 is also graded , it is a projectively normal toric variety in @xmath25 of dimension @xmath26 .
see @xcite for details on toric ideals .
a survey of recent results and open questions on normal configurations can be found in @xcite .
it is well known that initial ideals of a polynomial ideal inherit important invariants of the ideal such as dimension , degree and hilbert function .
thus it is natural to ask if certain initial ideals inherit further properties of the ideal such as cohen - macaulayness , betti numbers or reducedness ( of the corresponding scheme ) . a result of hochster @xcite shows that when @xmath0 is normal , @xmath27 is cohen - macaulay .
if @xmath0 is also graded , then @xmath27 is generated by homogeneous binomials of degree at most @xmath24 ( * ? ? ?
* thm . 13.14 ) . motivated by this , sturmfels asked and conjectured the following .
[ cmconjecture ] if @xmath0 is normal ( more generally , if @xmath3 is cohen - macaulay ) , is there a monomial initial ideal @xmath28 of the toric ideal @xmath3 that is cohen - macaulay ?
[ bigconjecture ] ( * ? ? ?
* conjecture 2.8 ) if @xmath0 is a graded , normal configuration then @xmath27 has a grbner basis whose elements have degree at most @xmath29 . in this paper , we show that question [ cmconjecture ] has a positive answer and conjecture [ bigconjecture ] is true when @xmath0 is @xmath5-_normal_. these configurations were defined by hoc sten and thomas in @xcite .
we recall the definition .
let @xmath5 be a pure @xmath30-dimensional simplicial complex with vertex set @xmath31 : = \{1 , \ldots , n\}$ ] .
we denote the set of _ facets _
( @xmath24-element faces ) of @xmath5 by @xmath32 . for a set
@xmath33 $ ] , let @xmath34 .
we say that @xmath5 is a _ triangulation _ of @xmath0 if @xmath35 and @xmath36 for all @xmath37 .
stanley - reisner _ ideal of @xmath5 is the squarefree monomial ideal @xmath38 .
a cornerstone in the combinatorics of toric initial ideals is the result by sturmfels that the radical of a monomial initial ideal @xmath39 of the toric ideal @xmath3 is the stanley - reisner ideal of a triangulation @xmath40 of @xmath0 ( * ? ? ?
the ideal @xmath39 is said to be _ supported on _ @xmath40 .
triangulations supporting initial ideals of @xmath3 are the _ regular triangulations _ of @xmath0 .
a triangulation @xmath41 of @xmath0 is _ unimodular _ if for each @xmath42 ,
@xmath43 .
[ def_dnormal ] a configuration @xmath0 is * @xmath5-normal * if it has a regular triangulation @xmath5 such that for each @xmath44 , @xmath45 is a hilbert basis of @xmath46 .
note that @xmath16 is used in the semigroups of definition [ def_dnormal ] .
all @xmath5-normal configurations are normal . in sections [ cmsection ] and [ dnormalboundsection ]
, we prove our main results .
* theorem [ cm ] .
* let @xmath0 be a @xmath5-normal configuration .
then there exists a term order @xmath47 such that @xmath48 and @xmath49 is cohen - macaulay .
* theorem [ dnormalbound ] .
* let @xmath0 be a graded @xmath5-normal configuration .
then there exists a term order @xmath47 such that @xmath48 and the grbner basis of @xmath3 with respect to @xmath47 consists of binomials of degree at most @xmath29 .
it was shown in @xcite that if @xmath0 is @xmath5-normal then @xmath3 has a monomial initial ideal that is free of embedded primes . in section [ setup ]
we recall the main features of this initial ideal .
theorems [ cm ] and [ dnormalbound ] are proved by showing that this same initial ideal is cohen - macaulay and , when @xmath0 is graded , generated in degree at most @xmath24 .
our proofs are combinatorial and rely heavily on the structure of this initial ideal .
the set of @xmath5-normal configurations is a proper subset of the set of normal configurations .
they occur naturally in two ways .
if @xmath0 has a regular unimodular triangulation @xmath5 , then @xmath0 is @xmath5-normal .
if @xmath50 is simplicial and we assume that its extreme rays are generated by @xmath51 , then @xmath0 is @xmath5-normal with respect to the coarsest regular triangulation @xmath5 consisting of the unique facet @xmath52 .
these were the only examples known so far .
specific instances of normal configurations that are not @xmath5-normal for any @xmath5 are also known @xcite . in section [ families ]
we construct non - trivial families of both @xmath5-normal and non-@xmath5-normal configurations
. * theorem [ babysfirstdeltanormalfamily ] .
* there are families of @xmath5-normal configurations @xmath53 where @xmath54 is non - simplicial and @xmath55 has no regular unimodular triangulations .
* theorem [ babysfirstnormalfamily ] .
* there is a family of normal , graded configurations @xmath56 , that are not @xmath5-normal for any regular triangulation @xmath5 .
we now recall from @xcite the initial ideal of @xmath3 without embedded primes when @xmath0 is @xmath5-normal
. the construction uses the _
standard pair decomposition _ of a monomial ideal @xmath57 @xcite which carries detailed information about @xmath58 , the set of associated primes of @xmath57 .
recall that every monomial prime ideal of @xmath59 is of the form @xmath60 for some @xmath33 $ ] .
the monomials of @xmath59 that do not lie in @xmath57 are the _ standard monomials _ of @xmath57 .
the _ support _ of a monomial @xmath61 is defined to be the support of its exponent vector @xmath62 i.e. , @xmath63 \ , : \ , v_i \neq 0\}$ ] .
@xcite let @xmath64 be a monomial ideal .
for a standard monomial @xmath65 of @xmath57 and a set @xmath33 $ ] , @xmath66 is a * pair * of @xmath57 if all monomials in @xmath67 $ ] are standard monomials of @xmath57 .
we call @xmath68 a * standard pair * of @xmath57 if : 1 .
@xmath68 is a pair of @xmath57 , 2 .
@xmath69 , and 3 .
the set of monomials in @xmath70 $ ] is not properly contained in the set of monomials in @xmath71 $ ] for any @xmath72 satisfying ( 1 ) and ( 2 ) .
the set of standard pairs of @xmath57 is unique and is called the standard pair decomposition of @xmath57 since this set provides a decomposition of the standard monomials of @xmath57 .
if @xmath61 is a standard monomial of @xmath57 then there is a standard pair @xmath73 of @xmath57 such that @xmath65 divides @xmath61 and @xmath74 . in this case
we say that @xmath61 is _ covered _ by @xmath68 .
we also use @xmath68 to denote the set of all standard monomials covered by it .
[ ap56 ] ( a ) @xcite let @xmath57 be a monomial ideal in @xmath59 .
then , 1
. @xmath75 if and only if @xmath57 has a standard pair of the form @xmath76 .
2 . @xmath77 is a minimal prime of @xmath57 if and only if @xmath78 is a standard pair of @xmath57 .
\(b ) @xcite let @xmath79 be a monomial initial ideal of the toric ideal @xmath3 .
then , 1 . if @xmath75 then @xmath80 is a face of the regular triangulation @xmath40 of @xmath0 , 2 .
@xmath81 is a minimal prime of @xmath57 if and only if @xmath82 , and 3 . for @xmath83 , the number of standard pairs of @xmath57 of the form @xmath84 is @xmath85 , the normalized volume of @xmath86 in @xmath40 .
we call @xmath65 and @xmath80 the _ root _ and _ face _ of the standard pair @xmath68 .
let @xmath87 ( respectively , @xmath88 ) be the matrix whose set of columns is @xmath0 ( respectively , @xmath89 ) .
the normalized volume of @xmath90 is the absolute value of the determinant of @xmath91 divided by the g.c.d . of the non - zero maximal minors of @xmath87 .
[ dnormal ] ( * ? ? ?
4.7 ) let @xmath0 be a @xmath5-normal configuration .
then there exists a term order @xmath47 such that @xmath48 and @xmath92 is free of embedded primes .
the term order @xmath47 needed in theorem [ dnormal ] is described in @xcite and is not directly used in this paper .
the ideal @xmath92 is shown to be free of embedded primes via an explicit description of its standard pairs .
this structure is crucial for this paper and hence we recall it now
. assume without loss of generality that @xmath93 .
for @xmath94 , let @xmath95 be the half open _ fundamental parallelopiped _ of @xmath96 .
then @xmath97 has @xmath85 elements including the origin . for @xmath98 ,
let @xmath99 be the cheapest monomial with respect to @xmath47 among all @xmath100 with @xmath101 .
it was shown in @xcite that @xmath102 .
the standard pairs of the initial ideal @xmath103 in theorem [ dnormal ] are precisely the pairs @xmath104 as @xmath105 varies in @xmath106 for each @xmath86 in @xmath107 . by theorem [ ap56 ] , @xmath103 is thus free of embedded primes . for the remainder of this paper we will denote the toric initial ideal @xmath49 of theorem [ dnormal ] by @xmath108 and its set of standard pairs by @xmath109 .
[ example1 ] let @xmath0 be the vector configuration consisting of the @xmath110 columns of @xmath111 then @xmath0 is a graded _ supernormal _ configuration @xcite which means that it is @xmath5-normal with respect to every regular triangulation . consider the regular triangulation @xmath112 the configuration @xmath0 and its regular triangulation @xmath5 are shown in figure [ ex1 ] .
the toric ideal @xmath3 lives in @xmath113 $ ] . in figure
[ ex1 ] , we have labeled the points of @xmath0 by the variables @xmath114 corresponding to the columns of @xmath87 , instead of by @xmath115 .
the term order @xmath47 in theorem [ dnormal ] can be induced via the weight vector @xmath116 refined by the reverse lexicographic order with @xmath117 .
the initial ideal ( computed using macaulay 2 @xcite ) is @xmath118 its standard pairs , grouped by the facets of @xmath5 are : + @xmath119 + for @xmath120 , @xmath121 consists of nine lattice points @xmath122 for each exponent vector @xmath123 of the roots @xmath124 .
the last of these is @xmath125 .
the monomials @xmath65 of @xmath59 such that @xmath126 in increasing order with respect to @xmath47 are : @xmath127 .
thus , @xmath128 .
in this section we prove that the initial ideal @xmath108 of theorem [ dnormal ] is cohen - macaulay .
this is done by showing that @xmath108 has a particular _ stanley filtration _
@xcite which implies cohen - macaulayness @xcite .
stanley filtrations are special _
stanley decompositions_. let @xmath64 be a monomial ideal .
a * stanley decomposition * of @xmath57 is a set of pairs of @xmath57 , @xmath129 , that _ partition _ the standard monomials of @xmath57 .
every monomial ideal @xmath57 has the trivial stanley decomposition @xmath130 .
there can be many stanley decompositions of a monomial ideal .
the standard pair decomposition of @xmath57 is usually not a partition of the standard monomials of @xmath57 .
we will show that the standard pair decomposition @xmath109 of @xmath108 can be modified first to a stanley decomposition and then to a stanley filtration of the needed form . for @xmath33 $ ]
let @xmath131 $ ] be the projection map where @xmath132 if @xmath133 and @xmath134 if @xmath135 .
the following is a well known fact .
@xcite[artproj ] if @xmath136 for a regular triangulation @xmath40 of @xmath0 , then @xmath137 is an artinian monomial ideal in @xmath138 with @xmath85-many standard monomials which are precisely the roots of the standard pairs @xmath139 of @xmath28 .
[ rootsdivisors ] if @xmath140 is a standard pair in @xmath109 , then every divisor of @xmath65 is also the root of a standard pair in @xmath141 .
[ rootslemma ] let @xmath140 and @xmath142 be two standard pairs in @xmath109 .
if @xmath143 then @xmath144 .
suppose @xmath145 .
then @xmath146 with @xmath147 , @xmath148 and @xmath149 outside the vertices of @xmath150 and thus in particular , outside @xmath151 .
hence , @xmath152 .
[ simpcase ] if @xmath0 is normal , @xmath50 is simplicial ( generated without loss of generality by @xmath51 ) , and @xmath108 is the special initial ideal of theorem [ dnormal ] supported on @xmath153 , then @xmath109 is a stanley decomposition .
here @xmath0 is @xmath150-normal .
all standard pairs in @xmath109 have face @xmath154 $ ] and roots the standard monomials of @xmath155}(j)$ ] .
thus no two standard pairs of @xmath109 have the same root and by lemma [ rootslemma ] , no two standard pairs intersect .
however , if @xmath156 , then it is precisely the standard pairs in @xmath109 with a common root that stop @xmath109 from being a partition .
such pairs always exist when @xmath156 for instance , @xmath157 is in @xmath109 for all @xmath94 .
we will use the combinatorial notion of _ shellings _ to create new pairs that partition the standard monomials covered by each set of standard pairs with a common root . in section [ setup ] we defined @xmath158 .
the following lemma identifies the faces in all standard pairs that share a root .
[ sigmasets ] if @xmath65 is a root of a standard pair in @xmath109 , then @xmath159 .
recall from section [ setup ] that if @xmath160 , then @xmath161 .
conversely , suppose @xmath68 is a standard pair in @xmath109 and @xmath162 for some @xmath163 in @xmath164
. then @xmath165 . then @xmath166 . since @xmath167 , it is also in @xmath168 .
further , @xmath65 is the cheapest monomial with respect to @xmath47 among all monomials @xmath61 in @xmath59 with @xmath169 .
this implies that @xmath140 is also a standard pair of @xmath108 .
[ shelling ] ( * ? ? ?
* chapter 3 , definition 2.1 ) let @xmath170 be a pure simplicial complex .
a * shelling * of @xmath170 is a linear ordering @xmath171 of the facets of @xmath170 such that for each @xmath172 , @xmath173 , the subcomplex supported in @xmath174 has a unique minimal face .
a simplicial complex @xmath170 is * shellable * if it has a shelling .
let @xmath175 be a face of a simplicial complex @xmath170 .
then @xmath176 is the simplicial complex generated by all @xmath177 containing @xmath175 .
we sometimes write @xmath178 for @xmath179 when @xmath170 is obvious . the following is a mild generalization of lemma 8.7 in @xcite .
[ starshell ] let @xmath170 be a pure shellable simplicial complex with shelling order @xmath171 . if @xmath175 is any face of @xmath170 then the restriction of the global shelling order to @xmath180 yields a shelling of @xmath180 . for a root @xmath65 in @xmath109 ,
let @xmath181 . in the following arguments we fix a root @xmath65 of a standard pair in @xmath109 .
by lemma [ sigmasets ] , @xmath182 and the set of facets @xmath183 if @xmath61 divides @xmath65 , then @xmath184 .
the regular triangulation @xmath150 is shellable @xcite . in the rest of this section
, we fix a shelling of @xmath150 . by lemma [ starshell ] , this induces a shelling of @xmath185 .
let us assume without loss of generality that @xmath186 is the induced shelling order of the facets of @xmath187 .
for @xmath188 , let @xmath189 be the unique minimal face described in definition [ shelling ] .
it is known that @xmath190 * example [ example1 ] continued . *
consider the root @xmath191 of @xmath192 and the shelling order @xmath193 and @xmath194 on @xmath150 .
then @xmath195 is the induced shelling order of @xmath196 . from this
we obtain the sets @xmath197 and @xmath198 .
consider the interval @xmath199
@xcite [ localgeompart ] the simplicial complex @xmath200 is the disjoint union of the intervals @xmath201 , @xmath202 .
[ subshellingremark ] note that by construction , the partial union @xmath203 is a partition of the subcomplex with maximal faces @xmath204 and that @xmath189 is not contained in this partial union for any @xmath205 . for a root @xmath65 in @xmath109 and a facet
@xmath206 define the monomial @xmath207 recall that we have fixed a shelling of @xmath150 and thus , by lemma [ starshell ] , a shelling of @xmath185 for each root @xmath208 . therefore ,
if @xmath209 , @xmath210 is uniquely defined .
we return to the fixed root @xmath65 and the shelling @xmath211 of @xmath185 .
[ partition ] the standard monomials of @xmath108 in @xmath212 are partitioned by the pairs @xmath213 , @xmath202 . by lemma [ localgeompart ]
, @xmath214 is a partition of @xmath215 .
hence , if @xmath216 then @xmath217 for a unique @xmath201 . by construction , @xmath218 and so @xmath219 .
this implies that @xmath220 where the inclusion @xmath221 follows from the previous line and @xmath222 from the fact that @xmath223 for each @xmath202 . to see that @xmath224 is a partition , suppose @xmath225 where @xmath226 .
then @xmath227 has support in @xmath228 which implies that @xmath229 .
however , for @xmath230 , @xmath231 as @xmath232 which means that @xmath65 lies only in @xmath233 . * example [ example1 ] continued . * as before , the monomial @xmath191 is a root of @xmath192 with the shelling order induced on @xmath234 as above .
we obtain the monomials @xmath235 and @xmath236 from @xmath237 and @xmath238 .
then @xmath239 with the latter union being disjoint .
[ stanley ] let @xmath240 be the fixed shelling of @xmath150 .
then @xmath241 is a stanley decomposition of @xmath108 .
lemma [ partition ] showed how to make the union of the standard pairs of @xmath109 with a common root a disjoint union of pairs of @xmath108 . by lemma [ rootslemma ] , ( [ standecomp ] ) ,
the union of these disjoint unions is a stanley decomposition of @xmath108 .
the above stanley decomposition can be organized to have more structure .
[ filtration ] @xcite let @xmath64 be a monomial ideal .
a * stanley filtration * of @xmath57 is a stanley decomposition of @xmath57 with an ordering of the pairs @xmath242 such that for all @xmath243 the set @xmath244 is a stanley decomposition of @xmath245 .
equivalently , the ordered set @xmath246 is a stanley filtration provided the modules @xmath247 form a filtration @xmath248 with @xmath249 $ ] .
( from @xcite ) let @xmath250 $ ] .
then @xmath251 is a stanley decomposition of @xmath57 but no ordering of these pairs is a stanley filtration of @xmath57 .
alternatively , the ordered pairs @xmath252 form a stanley filtration of @xmath57 .
we now show that the pairs in ( [ standecomp ] ) can be ordered to yield a stanley filtration of @xmath108 .
the significance of this for us comes from a result of simon @xcite , interpreted as follows by maclagan and smith @xcite .
[ filtgivescm ] if @xmath64 is a monomial ideal with a stanley filtration such that for each face @xmath80 of a pair in the filtration , the prime ideal @xmath77 is a minimal prime of @xmath57 , then @xmath57 is cohen - macaulay .
the faces of pairs in ( [ standecomp ] ) already index minimal primes of @xmath108 .
thus to show that @xmath108 is cohen - macaulay all we need to do is to order the pairs in ( [ standecomp ] ) so that the ordered decomposition is a stanley filtration .
we do this using the following algorithm .
[ list ] * input : * the stanley decomposition ( [ standecomp ] ) of @xmath108 . + * output : * a stanley filtration of @xmath108 with the same faces as those in ( [ standecomp ] ) . 1 : : ( local lists ) for each @xmath253 , @xmath254 , order the pairs in ( [ standecomp ] ) with face @xmath253 in any way such that if @xmath255 precedes @xmath256 then @xmath61 does not divide @xmath65 .
call this list @xmath257 .
2 : : ( global list ) the global list @xmath258 is obtained by appending @xmath257 to the end of @xmath259 for @xmath260 .
let @xmath261 for @xmath262 .
then @xmath263 is the total number of pairs in ( [ standecomp ] ) .
write @xmath258 as @xmath264 $ ] where @xmath265 when @xmath266 ( @xmath267 ) and @xmath268 is the root of the @xmath269-th pair in the local list @xmath257 constructed in step 1 of the algorithm . for
@xmath243 define the partial list @xmath270 $ ] and the ideal @xmath271 .
we need to prove that @xmath272 is a stanley decomposition of @xmath273 .
since @xmath272 is already a partition , it suffices to show that the set of monomials in the pairs in @xmath272 is the set of standard monomials of @xmath273 .
\(i ) _ the standard monomials of @xmath273 are contained in the pairs in @xmath272 _ : a standard monomial @xmath65 of @xmath273 is a standard monomial of @xmath108 and hence is covered by a unique pair @xmath274 in @xmath258 . also , @xmath275 which implies that @xmath276 for any @xmath277 . hence @xmath278 and @xmath279 .
\(ii ) _ the monomials in the pairs in @xmath272 are standard monomials of @xmath273 _ : suppose @xmath65 lies in the ( unique ) pair @xmath280 .
since @xmath281 , it suffices to show that @xmath282 .
suppose @xmath283 .
then there exists @xmath284 , @xmath285 such that @xmath286 where @xmath287 is a monomial with support in @xmath288 .
since @xmath289 and @xmath290 are both in @xmath31 \backslash ( \tau_p \cup \tau_l)$ ] , it follows that @xmath291 .
since @xmath292 , by step 1 of the algorithm , @xmath293 . recall that @xmath294 and @xmath295 are standard pairs of @xmath108 .
since @xmath291 , by corollary [ rootsdivisors ] , @xmath296 is also in @xmath109 .
this implies that @xmath297 and @xmath288 are two distinct facets in @xmath298 . since @xmath299 , @xmath300
however , this is a contradiction since @xmath288 precedes @xmath297 in the shelling order on @xmath150 and hence @xmath301 can not be in @xmath288 . thus @xmath302 and @xmath303 and thus not in @xmath273 .
* example [ example1 ] continued . * as before ,
@xmath193 and @xmath194 is a shelling order on @xmath150 .
the ( ordered ) local lists in the stanley filtration produced by algorithm [ list ] are : + @xmath304 $ ] , + @xmath305 $ ] , + @xmath306 $ ] , + @xmath307 $ ] .
[ cm ] let @xmath0 be a @xmath5-normal configuration .
then there exists a term order @xmath47 such that @xmath308 and @xmath49 is cohen - macaulay .
algorithm [ list ] shows that the initial ideal @xmath108 of theorem [ dnormal ] has a stanley filtration that satisfies the conditions of theorem [ filtgivescm ] .
this theorem guarantees that @xmath108 is cohen - macaulay .
we remark that even when @xmath0 is @xmath5-normal it is not true that all initial ideals of @xmath3 without embedded primes are cohen - macaulay .
take @xmath0 to be the columns of @xmath309 then @xmath0 admits a unimodular regular triangulation and is hence @xmath5-normal . the toric ideal @xmath310 $ ] has codimension four and has 46 initial ideals without embedded primes . among them ,
the following two have projective dimension five .
1 . @xmath311 2 .
@xmath312 the initial ideals of @xmath27 were computed using the software package cats @xcite and then checked for embedded primes and cohen - macaulayness using macaulay 2 .
we remark that the first example of a monomial toric initial ideal without embedded primes that is not cohen - macaulay was found by laura matusevich @xcite . in that example
, @xmath3 is not cohen - macaulay and thus @xmath0 is not normal .
[ dnormalbound ] if @xmath0 is a graded @xmath5-normal configuration , then there exists a term order @xmath47 such that @xmath48 and the grbner basis of @xmath3 with respect to @xmath47 consists of binomials of degree at most @xmath29 .
theorem [ dnormalbound ] settles conjecture [ bigconjecture ] for the subset of normal configurations that are @xmath5-normal .
since @xmath0 is graded , @xmath3 is homogeneous with respect to the usual grading of @xmath59 where @xmath313 for @xmath314 .
hence it suffices to show that @xmath3 has an initial ideal of degree at most @xmath24 .
we will show that the initial ideal @xmath108 from theorem [ dnormal ] satisfies this degree bound when @xmath0 is graded .
proposition 13.15 in @xcite shows that conjecture [ bigconjecture ] is true whenever @xmath0 admits a regular unimodular triangulation .
( see also proposition 13.18 in @xcite . )
such configurations form a proper subset of the set of @xmath5-normal configurations .
if we are allowed to replace a graded normal @xmath0 by all the lattice points in a `` big enough '' multiple of the convex hull of @xmath0 , then it is known that this new configuration admits regular unimodular triangulations and thus has a grbner basis of degree at most @xmath24 .
see @xcite and @xcite for many such results .
conjecture [ bigconjecture ] requires that @xmath0 be both graded and normal .
[ needgradedandnormal ] _ graded , but not normal : _ when @xmath315 with @xmath316 , @xmath317 and @xmath318 , then @xmath319 .
its two initial ideals are therefore generated in degree @xmath320 .
+ _ normal , but not graded : _ the normal configuration @xmath321 where @xmath322 has the toric ideal @xmath323 .
hence @xmath324 is the unique element in both its reduced grbner bases .
[ sharp ] ( @xcite , chapter 13 ) the bound in conjecture [ bigconjecture ] is best possible .
consider the graded @xmath5-normal configuration @xmath325 where @xmath326 .
( note that @xmath50 simplicial ) . then @xmath327 .
consider the initial ideal @xmath108 from theorem [ dnormal ] for a graded @xmath150-normal @xmath0 .
since @xmath0 is graded , we may assume without loss of generality that @xmath328 for @xmath314 .
we will show that @xmath108 is generated in degree at most @xmath24 . for a @xmath329 , recall that @xmath330 .
define @xmath331 .
then @xmath332 is a partition of @xmath31 $ ] .
let @xmath333 be the artinian ideal in @xmath334 $ ] from theorem [ artproj ] .
recall that the standard monomials of @xmath335 are the roots of standard pairs in @xmath109 with face @xmath86 . since the supports of these roots lie in @xmath336 ,
@xmath337 $ ] is a monomial ideal @xmath338 with @xmath339 , and @xmath340 [ lemma1ofdnormalbound ] each minimal generator @xmath341 of @xmath342 is a minimal generator of @xmath108 of degree at most @xmath24 .
a minimal generator @xmath341 of @xmath342 is the projection via @xmath343 of a minimal generator @xmath344 of @xmath108 where @xmath345 .
suppose @xmath346 .
then @xmath341 is a standard monomial of @xmath108 with @xmath339 .
hence @xmath341 is covered by a standard pair @xmath347 of @xmath108 .
this implies that all monomials of the form @xmath348 as @xmath349 varies are standard monomials of @xmath108 which contradicts that @xmath350 is in @xmath108 .
thus @xmath351 which implies that @xmath352 is a minimal generator of @xmath108 . since @xmath328 for @xmath353 $
] , each lattice point in the half open fundamental parallelopiped @xmath121 of @xmath96 lies on one of the @xmath24 hyperplanes @xmath354 in @xmath19 . therefore , if @xmath355 , then the @xmath356-norm of @xmath357 which equals the first co - ordinate of @xmath358 which equals @xmath359 is at most @xmath26 .
this implies that @xmath360 .
thus all standard monomials of the artinian ideal @xmath335 have degree at most @xmath26 which implies that the minimal generators of @xmath335 ( and @xmath342 ) have degree at most @xmath24 . * example [ example1 ] continued . * for @xmath120 , @xmath361 .
note that all minimal generators of @xmath342 are minimal generators of @xmath108 of degree at most three .
[ simplexbound ] if @xmath0 is a graded normal configuration with @xmath50 simplicial then @xmath27 has a grbner basis consisting of binomials of degree at most @xmath24 . assuming that @xmath50 is generated by @xmath362
, @xmath0 is @xmath150-normal where @xmath150 is the regular triangulation of @xmath0 with the unique facet @xmath363 $ ] . here
@xmath364 .
we argue that all minimal generators of @xmath108 have support in @xmath365 .
suppose @xmath366 is a minimal generator of @xmath108 with @xmath367 = f \neq \emptyset$ ] .
let @xmath368 $ ] .
then @xmath369 since otherwise @xmath366 would lie on the standard pair @xmath370)$ ] of @xmath108 which is a contradiction . write @xmath371 where @xmath372 and @xmath373 . since @xmath374 , @xmath375 is a standard monomial of @xmath108 which implies that @xmath376 is also a standard monomial of @xmath108 as @xmath375 lies on some standard pair with face @xmath154 $ ] .
this is a contradiction and so @xmath377 .
the above argument shows that @xmath108 and @xmath342 have the same minimal generators .
the degree bound then follows from the proof of lemma [ lemma1ofdnormalbound ] .
theorem [ simplexbound ] proves theorem [ dnormalbound ] in the case where @xmath50 is simplicial .
when @xmath50 is not simplicial , @xmath108 may have minimal generators that are not pre - images under @xmath343 of the minimal generators of @xmath342 ( or even @xmath335 ) as @xmath86 varies in @xmath378 .
our next step is to show that for a @xmath206 , the minimal generators of @xmath108 that project under @xmath343 to the minimal generators @xmath379 of @xmath335 have degree at most @xmath24 .
we need a preliminary lemma .
let @xmath380 be a @xmath30-polytope in @xmath381 and let @xmath382 be the cone over @xmath380 .
then there exists a matrix @xmath383 such that @xmath384 where each row of @xmath385 is the normal to a facet of @xmath382 . hence @xmath386 .
let @xmath387 be the system obtained by reversing all the inequalities in @xmath380 : @xmath388 [ halfspaces ] the polyhedron defined by @xmath387 is empty .
we may assume that @xmath380 has been translated so that the unit vector @xmath389 lies in the relative interior of @xmath380 .
if @xmath390 then by our assumption , @xmath391 which implies that @xmath392 .
this implies that @xmath393 where @xmath394 is the dual cone to @xmath382 .
( recall @xmath395 . )
thus there exists some @xmath396 such that @xmath397 .
therefore , if we choose @xmath398 such that @xmath399 then @xmath400 , @xmath401 and @xmath402 let @xmath403 be the right hand side vector in the description of @xmath387
. then @xmath404 and by farkas lemma ( * ? ? ?
* prop . 1.7 . ) , @xmath405 .
[ lemma2ofdnormalbound ] let @xmath86 be a facet of @xmath150 .
then for a @xmath406 , the minimal generators of @xmath108 that are preimages of the minimal generator @xmath407 of @xmath335 under the map @xmath343 are squarefree monomials of degree at most @xmath24 .
let @xmath329 , @xmath406 and @xmath408 be a minimal generator of @xmath108 with @xmath409 .
all minimal generators of @xmath108 that project to @xmath407 under @xmath343 look like @xmath410 .
if @xmath411 , then @xmath407 is the only minimal generator of @xmath108 that projects to @xmath407 and we are done . therefore , we consider the case where @xmath412 .
suppose @xmath410 is not squarefree .
then there exists an @xmath413 such that @xmath414 where @xmath415 is the @xmath416-th co - ordinate of @xmath417 . since @xmath410 is a minimal generator of @xmath108
, @xmath418 is a standard monomial of @xmath108 with @xmath419 .
hence there exists @xmath420 such that a standard pair with face @xmath80 covers @xmath418 .
this implies that @xmath421 .
since @xmath422 , @xmath423 and thus , @xmath424 .
if @xmath425 , then @xmath410 is covered by the standard pair @xmath78 which contradicts that @xmath410 is in @xmath108 . if @xmath426 , then @xmath427 lies in @xmath428 .
since @xmath0 lies on the hyperplane @xmath429 , @xmath427 is in fact in the minimal hilbert basis of both @xmath430 and @xmath50 and hence @xmath431 is the unique vector in @xmath432 that satisfies @xmath433 .
consequently @xmath434 is a standard pair of @xmath108 .
but this implies that @xmath410 lies on this standard pair which is again a contradiction .
therefore , @xmath410 is squarefree . to argue that
@xmath435 , it therefore suffices to prove that @xmath436 .
suppose @xmath437 $ ] , @xmath438 } : = \prod_{i \in \sigma } x_i$ ] and @xmath439}$ ] .
then for each @xmath440 $ ] , @xmath418 is a standard monomial of @xmath108 and is therefore covered by a standard pair @xmath441 of @xmath108 .
the face @xmath442 does not contain @xmath416 since otherwise @xmath410 would be a standard monomial of @xmath108 .
in particular , @xmath443 $ ] for any @xmath440 $ ] .
also , @xmath444 for each @xmath445 $ ] .
we now show that we may assume @xmath446 = [ d]\backslash \ { i \}$ ] for all @xmath440 $ ] .
clearly , @xmath446 \subseteq [ d ] \backslash \ { i \}$ ] since @xmath447 .
suppose a monomial in @xmath448 \backslash \ { i \}]$ ] lies in @xmath108 .
then it is divisible by a minimal generator of @xmath108 that projects to @xmath407 under @xmath343 , all of which are squarefree .
such a minimal generator would properly divide @xmath410 which contradicts that @xmath410 is a minimal generator of @xmath108 . hence @xmath449
\backslash \{i\})$ ] is a pair of @xmath108 and therefore , contained in a standard pair of @xmath108 .
we may assume that @xmath442 is the face of this standard pair . thus @xmath154
\backslash \{i\ } \subset \tau^i$ ] and @xmath446 = [ d]\backslash \ { i \}$ ] as claimed
. since @xmath0 is graded , @xmath450 index @xmath30-simplices in a regular triangulation of @xmath451 , the convex hull of @xmath0 .
the simplex indexed by @xmath154 $ ] is geometrically @xmath452 = \{{\bf x } \in { { \mathbb r}}^d \ , : \ , { \bf s}_i \cdot { \bf x } \geq 0 , \ , x_1 = 1 , \
, i = 1,2,\ldots , d \ } $ ] where @xmath453 for all @xmath454 \backslash \{i\}$ ] and @xmath455 .
now @xmath456 for each @xmath440 $ ] implies that @xmath457_{rev}$ ] where @xmath458_{rev } = \ { { \bf x } \in { { \mathbb r}}^d \ , : \ , { \bf s}_i \cdot { \bf x } \leq 0 , \ , x_1 = 1 , \ , i = 1,2,\ldots , d \ } $ ] .
but by lemma [ halfspaces ] , @xmath452_{rev } = \emptyset$ ] which creates a contradiction .
therefore , @xmath459}$ ] is not a minimal generator of @xmath108 and all preimages @xmath410 of @xmath407 have degree at most @xmath24 .
* example [ example1 ] continued . * for @xmath120 , @xmath460 which index the variables @xmath461 .
the minimal generators of @xmath108 that map to these variables under @xmath343 are @xmath462 .
finally we consider the minimal generators of @xmath108 that do not project under @xmath343 to minimal generators of @xmath335 for any @xmath329 .
such generators may exist .
* example [ example1 ] continued .
* consider the minimal generator @xmath463 of @xmath108 .
then @xmath464 is not a minimal generator of @xmath335 for any of the four facets @xmath86 of @xmath150 .
[ lemma3ofdnormalbound ] let @xmath465 be a minimal generator of @xmath108 whose image under @xmath343 is not a minimal generator of @xmath335 for any facet @xmath86 of @xmath150 .
then @xmath465 is a quadratic squarefree monomial .
let @xmath80 and @xmath466 be facets of @xmath150 and let @xmath467 , @xmath468 with @xmath469 .
then @xmath470 is not covered by any standard pair of @xmath108 and hence lies in @xmath108 .
since @xmath0 is graded , @xmath471 and @xmath472 are standard pairs of @xmath108 which implies that @xmath473 is a minimal generator of @xmath108 .
further , @xmath474 is not a minimal generator of @xmath335 for any @xmath475 .
we will prove that @xmath476 is precisely the set of minimal generators of @xmath108 that do not project under @xmath343 to a minimal generator of @xmath335 for a @xmath477 .
this will prove the lemma .
suppose @xmath465 is a minimal generator of @xmath108 such that @xmath478 is not a minimal generator of @xmath335 for any @xmath329 .
let @xmath479 . _ case ( i )
@xmath480 for some @xmath481 _ : then @xmath482 $ ] . since @xmath483 is not a minimal generator of @xmath484 ( and hence @xmath342 ) , some minimal generator of @xmath342 properly divides @xmath483 . by lemma [ lemma1ofdnormalbound ] , every minimal generator of @xmath342 is a minimal generator of @xmath108 which contradicts that @xmath483 is a minimal generator of @xmath108 .
_ case ( ii ) @xmath485 for some @xmath481 _ : then @xmath483 is covered by the standard pair @xmath486 which contradicts that @xmath487 . _ case ( iii ) @xmath488 for some @xmath329 , with @xmath489 and @xmath490 _ : write @xmath491 where @xmath492 and @xmath493
. then @xmath494 all of whose minimal generators are minimal generators of @xmath108 .
this implies that a divisor of @xmath495 is a minimal generator of @xmath108 .
therefore , @xmath483 is not a minimal generator of @xmath108 , a contradiction .
the above cases have shown that there is no single @xmath481 such that @xmath496 .
therefore , there exists two distinct @xmath497 and two indices @xmath498 such that @xmath499 and @xmath500 .
_ case ( a ) @xmath413 _ : since @xmath501 , @xmath502 is not covered by any standard pair of @xmath108 and so lies in @xmath108 . since @xmath503 divides @xmath483 and @xmath483 is a minimal generator of @xmath108 it must be that @xmath504 .
but then @xmath505 , @xmath501 is a minimal generator of @xmath335 which contradicts our choice of @xmath483 .
therefore this case can not arise .
_ case ( b ) @xmath425 _ : by a symmetric argument to the previous , this can not happen .
therefore , the only possibility is that @xmath506 and @xmath507 . since @xmath508 and @xmath501 , @xmath509 . by the argument in the beginning of the proof , @xmath503 is a minimal generator of @xmath108 and so @xmath504 .
thus @xmath483 lies in the set @xmath510 as claimed .
_ proof of theorem [ dnormalbound ] .
_ lemmas [ lemma1ofdnormalbound ] , [ lemma2ofdnormalbound ] and [ lemma3ofdnormalbound ] account for all minimal generators of the initial ideal @xmath108 and show that they all have degree at most @xmath24 . since @xmath0 is graded , the reduced grbner basis of @xmath3 with initial ideal @xmath108 consists of homogeneous binomials . hence these binomials have degree at most @xmath24 .
in this last section we construct non - trivial families of both @xmath5-normal and non-@xmath5-normal configurations . recall that any configuration @xmath2 is always @xmath5-normal with respect to all its regular unimodular triangulations @xmath5 .
also , a configuration @xmath512 for which @xmath513 is simplicial is @xmath5-normal with respect to its coarsest ( regular ) triangulation @xmath514 if we assume that @xmath515 .
call @xmath2 _ simplicial _ if @xmath513 is simplicial .
we construct families of @xmath5-normal configurations that are not simplicial and do not admit regular unimodular triangulations . by computer search ,
firla and ziegler @xcite found hundreds of normal simplicial configurations @xmath2 in @xmath516 and @xmath517 ( in the course of writing @xcite ) that admit no unimodular triangulations .
our first result in this section is a construction that extends a firla - ziegler configuration to a family of _ non - simplicial _ @xmath5-normal configurations one in @xmath518 for each @xmath519 without unimodular triangulations . in the second part of this section
we construct a family of normal configurations ( an @xmath520 for each @xmath521 ) that are not @xmath5-normal for any regular triangulation @xmath5 .
each firla - ziegler normal simplicial @xmath522 without unimodular triangulations is the hilbert basis of the cone generated by @xmath523 , the first three unit vectors in @xmath524 , and a vector @xmath525 of the form @xmath526 with @xmath527 . in this subsection
we let @xmath2 denote such a firla - ziegler configuration and let @xmath528 be the extreme rays of @xmath513 . by definition
, @xmath2 is the unique minimal generating set of the semigroup @xmath529 and @xmath530 .
the vector @xmath531 is contained in @xmath532 . since @xmath533 and @xmath534 , @xmath535 .
for every @xmath536 with @xmath537 , @xmath538 for @xmath539 since the @xmath540-linear combination of elements in @xmath541 that expresses @xmath349 as an element of @xmath542 must involve a positive multiple of @xmath62 . on the other hand , the @xmath13-linear combination of elements in @xmath0 that expresses @xmath543 as an element of @xmath544 is the sum of distinct vectors in @xmath545 .
at least one of these @xmath546 vectors say @xmath547 has a positive last co - ordinate which implies that @xmath548 .
therefore , @xmath549 is in @xmath0 , the minimal hilbert basis of @xmath550 .
[ firstfz ] the first firla - ziegler @xmath2 in @xmath551 has @xmath552 and @xmath2 consists of the columns of the matrix @xmath553 the hilbert basis of any rational polyhedral cone can be computed using the software package normaliz @xcite . from a firla - ziegler @xmath2
we will now recursively construct configurations @xmath55 for each @xmath519 such that @xmath554 is @xmath5-normal , @xmath54 is not simplicial and @xmath55 has no unimodular triangulations . for each @xmath519
let @xmath555 , @xmath556 and @xmath557 . here
@xmath558 is the @xmath24-th unit vector in @xmath19 . letting @xmath559 ( a firla - ziegler configuration in @xmath551 ) and @xmath560 , recursively define @xmath561 , @xmath562 and @xmath563 .
we assume that @xmath564 and @xmath565 are always the first and second elements of @xmath566 and that @xmath86 is the index set of @xmath567 in @xmath55 .
let @xmath568 and @xmath569 .
consider the triangulation @xmath570 of @xmath571 consisting of the maximal subcones @xmath572 and @xmath573 .
[ a5 ] the configuration @xmath574 has the following properties : 1 .
@xmath575 , 2 .
@xmath574 is non - simplicial , 3 .
@xmath574 is @xmath576-normal , and 4 .
@xmath574 admits no unimodular triangulations . 1 .
since @xmath577 , @xmath578 and the first three unit vectors of @xmath579 belong to @xmath580 , it follows that all unit vectors of @xmath579 lie in @xmath581 which gives the result .
similarly , @xmath582 .
2 . since @xmath583 lies in the interior of @xmath584 , the vectors @xmath585 and @xmath586 do not lie on a common facet of the cone .
hence @xmath584 is a bipyramidal cone over @xmath587 with six extreme rays and is hence non - simplicial .
the triangulation @xmath576 is the regular triangulation of @xmath574 with respect to the weight vector @xmath588 .
we first argue that @xmath589 is a minimal generating set of the semigroup @xmath590 .
suppose @xmath591 . since @xmath585 is the unique generator of @xmath592 with a positive fifth co - ordinate , @xmath593 where @xmath594 is the unique expression of @xmath595 as an @xmath596-combination of @xmath585 and the other extreme rays of @xmath592 . since @xmath597 , in fact , @xmath598 where the equality @xmath599 follows from the normality of @xmath600 .
this in turn implies that @xmath601 .
( note that @xmath602 . )
similarly , @xmath603 is its own hilbert basis .
thus , @xmath574 is @xmath576-normal .
4 . suppose @xmath41 is a unimodular triangulation of @xmath574 and @xmath80 is a facet of @xmath41 .
then by ( 1 ) , @xmath604 and @xmath605 .
if @xmath606 , then @xmath607 which shows that @xmath608 , a contradiction . hence each maximal simplex in @xmath41 contains exactly one of @xmath585 or @xmath586 and @xmath41 induces a triangulation @xmath609 of @xmath610 . since @xmath41 is unimodular , @xmath609 gives a unimodular triangulation of @xmath600 which is a contradiction as @xmath600 has no unimodular triangulations .
therefore , we conclude that @xmath574 has no unimodular triangulations .
[ babysfirstdeltanormalfamily ] for each @xmath519 , the configuration @xmath55 has the following properties : 1 .
@xmath611 , 2 .
@xmath55 is non - simplicial , 3 .
@xmath55 is @xmath570-normal , and 4 .
@xmath55 admits no unimodular triangulations .
this theorem is proved by induction using lemma [ a5 ] as the base step . 1 .
suppose the result is true for @xmath612 .
then it follows that @xmath613 contains the first @xmath26 unit vectors of @xmath614 which are therefore also in @xmath615 and @xmath616 .
since @xmath617 ( and @xmath618 ) , we also get that @xmath619 ( and @xmath620 ) .
hence @xmath621 .
2 . assume by induction that @xmath622 is non - simplicial and that @xmath623 lies in the interior of @xmath624 .
then , @xmath625 lies in the interior of @xmath54 and hence @xmath626 and @xmath565 do not lie on a common facet of @xmath54 .
this implies that @xmath627 has exactly two more extreme rays than @xmath628 and hence is non - simplicial .
3 . as in lemma
[ a5 ] , @xmath570 can be induced as the regular triangulation of @xmath55 with respect to the weight vector @xmath629 for each @xmath519 .
we assume by induction that @xmath622 is @xmath630-normal and hence normal .
the arguments that @xmath631 and @xmath632 are minimal generating sets for @xmath633 and @xmath634 respectively follow from a straight generalization of the arguments in lemma [ a5 ] .
4 . again
we assume by induction that @xmath622 admits no unimodular triangulations .
the rest of the argument is also a straight generalization of the arguments in lemma [ a5 ] ( 4 ) .
we have thus produced non - simplicial @xmath5-normal configurations without unimodular triangulations in every dimension beyond four , starting with a firla - ziegler @xmath2 in @xmath551 .
the construction applies to all such firla - ziegler configurations .
consider the graph @xmath635 shown in figure [ ho ] .
in @xcite hibi and ohsugi showed that the graded normal configuration @xmath636 admits no regular unimodular triangulations , although it does have non - regular unimodular triangulations .
further , the @xmath637 points in @xmath638 are all extreme points of the convex hull of @xmath638 , denoted as @xmath639 .
the @xmath640-polytope @xmath641 is _ empty _ which means that it has no lattice points other than its vertices .
thus @xmath642 which is a cone over @xmath639 has @xmath637 extreme rays and is therefore non - simplicial .
[ emptynotdeltanormal ] let @xmath643 be a normal graded non - simplicial configuration in @xmath644 such that @xmath451 is empty .
if @xmath0 does not have a regular unimodular triangulation then @xmath0 is not @xmath5-normal for any regular triangulation @xmath5 . without loss of generality
, we can assume that @xmath645 .
by the hypothesis , every regular triangulation @xmath5 of @xmath0 has a maximal face @xmath86 such that @xmath646 .
thus the hilbert basis of @xmath96 contains at least one vector @xmath647 not in @xmath89 .
since all vectors in @xmath0 are extreme rays of @xmath50 , none of them lie in @xmath96 unless they are in @xmath89 .
this implies that @xmath648 is not normal and hence @xmath0 is not @xmath5-normal .
the hibi - ohsugi configuration @xmath638 is not @xmath5-normal for any regular triangulation @xmath5 .
from @xmath638 we now recursively construct configurations @xmath55 for each @xmath521 such that @xmath55 is normal and graded but not @xmath5-normal for any regular triangulation @xmath5 .
for each @xmath521 let @xmath649 .
letting @xmath650 , recursively define @xmath651 [ babysfirstnormalfamily ] for each @xmath521 , the configuration @xmath55 is normal and graded but not @xmath5-normal for any regular triangulation @xmath5 .
it suffices to show that @xmath55 satisfies the conditions of lemma [ emptynotdeltanormal ] for each @xmath24 . for a given @xmath24 ,
@xmath55 is graded since it lies in @xmath652 and @xmath653 is a @xmath640-polytope and hence empty .
further , @xmath54 is non - simplicial as @xmath624 is non - simplicial for all @xmath521 .
the configuration @xmath654 is normal . to see this
let @xmath655 .
since the only extreme ray of @xmath656 with non - zero eleventh co - ordinate is @xmath657 , @xmath658 .
further , @xmath659 where @xmath660 is the unique expression of @xmath595 as an @xmath596-combination of the extreme rays of @xmath656 .
the integral vector @xmath661 lies in @xmath662 since @xmath663 is normal and hence it lies in @xmath664 .
thus @xmath665 .
by induction , it follows that @xmath55 is normal for all @xmath521 .
suppose @xmath654 had a regular unimodular triangulation
. then @xmath666 would be a vertex in every maximal face of this regular unimodular triangulation of @xmath451 .
this in turn induces a regular unimodular triangulation in @xmath667 and hence in @xmath663 , a contradiction .
again , a straightforward inductive argument shows that @xmath55 has no regular unimodular triangulation for all @xmath521 .
we thank diane maclagan for alerting us to the interpretation of simon s condition for a monomial ideal to be cohen - macaulay in her paper @xcite with greg smith .
we also thank winfried bruns , joseph gubeladze , serkan hoten , francisco santos and gnter ziegler for discussions on constructing @xmath5-normal configurations .
, computing the integral closure of an affine semigroup , effective methods in algebraic and analytic geometry , 2000 ( krakow ) .
_ acta math .
_ * 39 * , ( 2001 ) , 5970 .
+ software : _ normaliz _ , ftp.mathematik.uni-osnabrueck.de/pub/osm/kommalg/software . | a normal ( respectively , graded normal ) vector configuration @xmath0 defines the toric ideal @xmath1 of a normal ( respectively , projectively normal ) toric variety .
these ideals are cohen - macaulay , and when @xmath2 is normal and graded , @xmath1 is generated in degree at most the dimension of @xmath3 . based on this , sturmfels asked if these properties extend to initial ideals when @xmath2 is normal , is there an initial ideal of @xmath1 that is cohen - macaulay , and when @xmath2 is normal and graded , does @xmath1 have a grbner basis generated in degree at most @xmath4 ?
in this paper , we answer both questions positively for @xmath5-normal configurations .
these are normal configurations that admit a regular triangulation @xmath5 with the property that the subconfiguration in each cell of the triangulation is again normal .
such configurations properly contain among them all vector configurations that admit a regular unimodular triangulation .
we construct non - trivial families of both @xmath5-normal and non-@xmath5-normal configurations . |
internet of things ( iot ) is a novel concept that allows a number of devices to be connected through the internet .
such devices can be sensors / actuators which are able to operate and transmit data without or with minimal human intervention .
iot has brought a great influence to many areas , and there have been many iot applications implemented such as healthcare , transportation , logistics , and manufacturing @xcite .
however , the development of iot is facing many challenges especially for data management @xcite . due to the special characteristics of iot systems and services ,
e.g. , heterogeneous large - scale architecture , diverse and enormous data , traditional data management approaches may become intractable such that new solutions are required .
recently , the concept of _ smart data pricing ( sdp ) _ @xcite has been introduced as an alterative to address network resource management issues .
its major benefit is the ability to provide effective solutions from both system and economic aspects . in particular , with sdp , prices are used not only to gain profits for providers , but also to provide tools to improve system and data management . in this article
, we propose to apply sdp to iot so that prices are used to incentivize sensor owners to contribute their data to iot services , improving the service quality and generating higher revenue from selling iot services to users .
we first present an overview of iot including its intrinsic features , system architecture , benefits , data management , and business model .
we then introduce the sdp and review some related work .
moreover , we demonstrate the applications of economic models in iot by introducing a pricing scheme to optimize the sensing data buying price and iot service subscription fee for sensor owners and service users , respectively . we adopt the bundling strategy for multiple iot providers to form a coalition and offer their services as a bundle . with bundling
, the profit of the providers can be improved by encouraging many users to subscribe more services .
finally , we highlight a few important future research directions .
though the technology development and applications of iot are tremendously growing , its definition can be diverse and fuzzy .
@xcite provides one of the formal , concrete , and standardized definitions of iot : `` _ _ internet of things envisions a self - configuring , adaptive , complex network that interconnects ` things ' to the internet through the use of standard communication protocols .
the interconnected things have physical or virtual representation in the digital world , sensing / actuation capability , a programmability feature and are uniquely identifiable . _ _ '' in the iot context , things can be objects that have internet capability . the objects with unique identification can offer services in terms of data capture , communication , and actuation .
thus , derived from the definition , the fundamental features of iot are as follows @xcite : * _ connected to the internet : _
things must be collected to the internet using wired or wireless connections . * _ uniquely : _ things are uniquely identifiable via ip addresses . * _ sensing / actuation capability : _ things are able to perform sensing / actuation tasks . * _ embedded intelligence : _ things are embedded with intelligent functions , e.g. , self - configurability . * _ interoperable communication capability : _ the iot system has a communication capability based on standard technologies . a general architecture for iot systems can be illustrated as in figure [ iot_architecture ] .
the architecture consists of four layers . *
_ physical layer _ is composed of smart devices , e.g. , sensors and actuators , that interact and/or gather data from the physical entities , e.g. , environment and human . * _ network and communication layer _ provides data communications and networking infrastructure to transfer data collected from devices at the physical layer to the data center . *
_ data center layer _ provides infrastructures to support data storage and processing to meet requirements of iot applications . *
_ service layer _ is a set of software that provides services to iot users .
as shown in figure [ iot_architecture ] , the interactions among layers in the iot architecture is through data .
data is first gathered at the physical layer by sensors and then transmitted to the data center through the network and communication layer . at the data center layer ,
the data is stored and processed to extract useful information for users in the service layer .
based on the received information , the applications or users can make appropriate decisions , e.g. , sending commands to control sensors / actuators at the physical layer .
@xmath0 \end{array}$ ] with an idea to connect everything to the internet , iot has brought a variety of benefits .
the major benefits are to improve system efficiency and user satisfaction , enhance flexibility , safety , and security , and finally open new business opportunity and revenue stream .
because of many benefits , there are numerous applications of iot in various sectors such as education , economics , transportations , and healthcare @xcite .
we discuss one example in the following .
tomtom ( www.tomtom.com ) , a well known gps manufacturer , introduces an iot service for a congestion index .
the aim is to capture anonymous travel time information particularly in urban areas .
data obtained from smart devices , e.g. , vehicle journey recorders , gps , and traffic cameras , will be sent to the tomtom servers through communication channels such as 3g / wifi .
the data is used to extract meaningful traffic information . based on the information ,
the tomtom service provides travel time information and real - life driving patterns presented to the general public , industry , and policy makers showing global congestion level .
thus , using the tomtom service , drivers can have more efficient and safe journeys .
government agencies and authorities can make appropriate policies , rules , and regulations in controlling road traffic , reducing accidents , and constructing road infrastructure .
finally , businesses can have useful information for their operations , e.g. , to open retail stores , gas stations , and repair shops at the best locations . as shown in the iot architecture ( figure [ iot_architecture ] ) , data is the key component , and hence data management is a main concern in iot .
data management involves the following aspects . * _ data collection : _ sensing is the first step in the data flow to obtain iot data . in the sensing process , to collect a large amount of data with high quality , a number of sensors have to be deployed which will result in high costs .
sensor deployment , data gathering and preprocessing have to be optimized .
alternatively , participatory sensing can be adopted . * _ data communications : _ together with a large number of devices , data communication and networking become important issues that need further analysis and optimization to meet specific requirements of iot .
machine - type - communications ( mtc ) has been introduced as the solutions for cellular networks . alternatively , bluetooth low energy ( ble ) , wifi , and 6lowpan can be used for local and personal area networks . *
_ data storage and processing : _ the amount of iot data is enormous which needs to be stored and processed efficiently and securely . cloud computing becomes an effective solution .
* _ information trading : _
after data is processed , the useful information will be extracted to provide ( sell ) services to iot users .
this is an important step in business models in terms of profitability and sustainability .
market structures , incentive and pricing mechanisms have to be newly defined because of specific characteristics of iot systems , businesses , and users .
the core economic benefit of iot is to generate revenues for businesses , and thus an iot business model is important .
generally , a business model _ describes the rationale of how a company creates , delivers , and captures value _
there are four major components in the business model as shown in figure [ business_model_framework ] .
the _ value proposition _ of data is the core component in the business model .
the value proposition specifies what to be actually delivered and the price that will be charged to the customers .
the main purpose of the value proposition is for the company to demonstrate to the customers that they will gain more benefits than what they pay .
accordingly , it is important in the business model to specify pricing mechanisms . to determine the price
, the company needs to analyze and know the total cost incurred to operate the business and to offer iot services .
the major cost is from investing in and operating _ infrastructure _ components .
moreover , the company has to quantify the willingness - to - pay value of the customers .
the economic and marketing techniques can be employed to understand the utility structure based on the types and segments of the customers . with the information about cost and willingness - to - pay
, the company can derive an optimal price that maximizes the provider profit and user satisfaction .
@xmath1 \end{array}$ ] in the next section , we introduce and review pricing models developed for iot data management .
sdp @xcite is a new concept to enhance network performance and to support data management through using pricing incentives .
for example , during a congestion period , dynamic pricing is used to defer some non - urgent users from accessing networks , improving qos performance .
there are two important attributes that make sdp a suitable solution for the data management in iot .
* _ flexibility : _ instead of using static pricing , e.g. , the usage - based method ( byte - counting ) that requires users to pay the same price per unit of data accessed , sdp provides the flexible and dynamic pricing mechanisms based on the demand and requirements of the users .
* _ incentive : _ devices may belong to different owners who may have no interest in the joint data management . thus , the use of sdp can attract such independent device owners to participate in iot , improving the service quality . in this section ,
we review related applications of sdp in iot data management .
the summary of related applications is given in table [ table_sec3_sum ] .
auction - based pricing @xcite & multiple sellers ; a buyer ; and an auctioneer & the auctioneer conducts a reverse auction in which sellers submit their asking prices .
the auctioneer selects the seller with the lowest asking price as the winner & * widely applied to theoretical researches * economics : buyer s revenue maximization * system : data aggregation ; resource allocation ; and routing protocol & nash equilibrium sealed - bid auction - based pricing @xcite & multiple buyers ; a seller ; and an auctioneer & the auctioneer conducts an auction in which buyers submit their bidding prices .
the auctioneer selects the buyer with the highest bidding price as the winner & * widely applied to theoretical researches * economics : seller s revenue maximization * system : data aggregation ; resource allocation ; and routing protocol & competitive equilibrium cost - based pricing @xcite&a seller ( or multiple sellers ) ; and a buyer ( or multiple buyers ) & the seller calculates selling price of an item based on costs charged for the item & * widely applied to real world market * economics : seller s profit maximization * system : data aggregation ; and packet forwarding & optimal solution stackelberg game - based pricing @xcite & multiple sellers , i.e. , leaders ; multiple buyers , i.e. , followers ; and multiple brokers & the sellers and brokers determine their own selling prices based on profit functions & * applied to real world market and security domain * economics : sellers and brokers profit maximization * system : relay selection & stackelberg equilibrium crowdsensing or participatory sensing is a concept to use independent mobile devices to collect and deliver sensing data with the aim to reduce the cost of deploying and maintaining sensor devices .
for example , gigwalk ( http://www.gigwalk.com/ ) provides a marketplace for sensing tasks performed through smartphones .
mobile users , who have installed the gigwalk client application , can submit their sensing data and receive rewards .
however , an important issue is how to set the price that is profitable for the provider and attractive for mobile users . in @xcite
, the authors introduced a reverse auction based dynamic price ( radp ) incentive mechanism .
the objective is to minimize and stabilize the incentive cost , while achieving sufficient number of participants . in this model
, mobile users can sell their sensing data to the provider through their claimed bidding prices . then
, the provider will select users who have the lowest bidding prices .
the authors demonstrate by simulation that the incentive cost can be reduced by more than 60% compared with random selection with a fixed price .
the authors in @xcite proposed a cost - based pricing model which can minimize the cost for transmitting packets to the provider through using short - range communications , e.g. , bluetooth and wifi , with their neighbors instead of transmitting data directly to the provider . after receiving the data
, the neighbors can then use the sensing data for themselves or resell it to the provider to gain revenue . to determine the selling price ,
each device builds a one - hop neighbor table .
the device then defines the total cost of sending packets to its neighbors according to remaining energy , resource usage , and costs .
the selling prices and corresponding profits are calculated as a function of the total cost , and the device will select a neighbor that has the minimal total cost .
bitmesh ( https://www.bitmesh.network/ ) is one of such data forwarding services that can adopt the aforementioned pricing model .
bitmess allows users to share their internet connection with peers in a local `` marketplace '' . due to the flexibility and efficiency
, cloud computing becomes a typical infrastructure to store and process a large amount of iot data collected from devices and sensors . in cloud computing , computing and storage resources can be used in an on - demand basis , reducing total cost of operation .
a variety of economic models have been developed for cloud computing services to achieve the highest profits and to meet the user requirements @xcite .
for example , multiple cloud providers can cooperate and share resources to increase scalability and reliability through the federated cloud model .
the strategic - proof dynamic pricing ( spdp ) scheme @xcite was proposed to improve resource utilization .
the spdp scheme is based on an auction mechanism that the payment for resource allocation is a function of the demand of users and the supply of cloud providers .
such pricing scheme can be applied to practical cloud services such as zuora ( https://www.zuora.com/ ) that offer cloud resource sharing services .
after data is processed , the useful information will be extracted for services to iot users .
information can be treated as goods that can be sold and bought in a market . in @xcite
, the authors studied the information service pricing by formulating a hierarchical game , i.e. , a stackelberg game , among information providers , brokers , and customers .
the brokers acquire information from the providers , i.e. , leaders , and sell the services to the customers , i.e. , followers .
the authors also adopted the bundling strategy for selling the information services .
the relationship between the information collection cost and willingness - to - pay value of the customers . although there are some sdp schemes applied to iot systems , few works considered bundled services of multiple providers . moreover ,
the service quality due to variable sensors selling data to providers was ignored .
thus , in the next section , we introduce a new pricing scheme for sensing data buying and iot service subscription with bundling .
in this section , we introduce a smart data pricing scheme for iot services that incorporate sensing data buying and service subscription with bundling @xcite . we first discuss the motivations and scenarios .
then , we present the detail of the pricing scheme . the numerical results are presented afterwards .
some works considered the bundling strategy in the smart data pricing , e.g. , @xcite @xcite . additionally , multi - tier market models were proposed @xcite and @xcite .
however , the consideration of the bundling strategy in the multi - tier market model taking the unique requirements of iot services into account was not done before .
we consider iot service providers who act as brokers .
each provider buys sensing data from a set of sensors belonged to other owners .
the provider then transfers the sensing data and processes it with the purpose to offer a value - added iot service to a set of consumers or users .
the data processing and service delivery can be performed in the facility , i.e. , private cloud , of the provider or in the public cloud that the provider rents from other cloud services .
figure [ fig : servicemodel ] illustrates the iot service offered by the provider .
there can be multiple providers collecting sensing data from different sets of sensors , and the providers offer the same or different services to the same set of users , i.e. , unimodal and multimodal sensing , respectively .
we assume that the users need the services from different providers separately with different preferences and values .
for example , the sensing data of each provider is from a different geographical area .
thus , the providers do not compete each other .
@xmath2 \end{array}$ ] this iot service is inspired and is applicable to many applications and businesses .
the following are some pertinent examples . * _ placemeter : _ placemeter ( https://www.placemeter.com/ ) is a startup that allows a user to set up a video camera to capture a view from streets and cities and to stream the video data to the placemeter .
the placemeter then processes it with video analytics .
the users will be paid by the placemeter depending on the view and quality of the video .
the video analytics can be used to extract meaningful and useful information to trace crowds and road traffic .
many businesses and government agencies can benefit from the information , e.g. , for retailers to open shops and for city agencies to plan public area usage and expansion . * _ road transportation services : _ there are a few services that allow drivers to share and obtain road traffic information .
some examples are google maps for mobile ( http://www.google.com/maps/about/ ) and waze ( https://www.waze.com/ ) .
drivers simply allow mobile apps on their mobile phones to send related information , e.g. , moving speed , to the service provider .
then , the provider translates and processes the information , and informs other drivers of the current road conditions in different areas . * _ iot search engine : _ while aforementioned examples are the services for specific purposes , recently iot search engine services have been introduced .
the services provide a generic capability for users to acquire sensing data with various types .
sensor owners can share , exchange , or sell their sensing data to other users .
one example is thingful ( https://thingful.net/ ) that lets sensors be connected and users can browse and obtain desired sensing data .
the iot search engine is able to locate , index , and make sensing data searchable . to develop and operate the iot services economically and profitably , in the following , we introduce an effective pricing scheme . as shown in figure [ fig : servicemodel ] , there are three major entities in the iot service under consideration .
* _ sensors _ are suppliers of sensing data .
sensor owners sell sensing data to one of the providers .
each sensor has a _ reservation wage _ which is a value of sensing data .
in particular , the sensor ( owner ) will sell its sensing data to a provider if the provider offers to buy with a _ buying price _
greater than the reservation wage .
otherwise , the sensor will not sell the data .
the reservation wage can be based on the cost that incurs because of collecting and transmitting the data to the provider . *
_ users _ are the consumers of iot services offered by the providers .
the users can buy any iot services .
each user has a _ reservation price _ which is a willingness - to - pay value of the iot service .
specifically , the user will subscribe , i.e. , buy , the iot service from the provider if the provider sells the service with a _ subscription fee _
lower than the reservation price .
otherwise , the user will not subscribe the service .
each user has different reservation prices for different iot services .
the user can also subscribe to multiple iot services simultaneously . * _ providers _ buy sensing data from a particular set of sensors , perform value - added data processing , and deliver an iot service to users . to buy sensing data from the sensors , the provider sets a buying price .
likewise , to sell the iot service to the users , the provider sets a subscription fee .
in addition to setting the price and fee , the providers have an option to cooperate with each other to offer iot services as a bundle . the objective of the providers is to maximize their profits .
note that the reservation wage and reservation price of the sensors and users , respectively , can be determined from cost - benefit analysis and performance requirements .
alternatively , the conjoint analysis that employs various marketing techniques to quantify such values can also be employed .
moreover , a sensor and user can be physically co - located .
for example , vehicle drivers can be data suppliers that transmit their location , traveling speed , and other information to the google map or waze .
likewise , the same drivers , as consumers , can access the google map or waze services to obtain global road network conditions to determine their best driving routes .
the iot service providers will maximize their profits through the pricing scheme .
firstly , they have to determine the buying price offered to the sensors and the subscription fee proposed to the users .
secondly , the providers decide whether the providers cooperate to offer their iot services as a bundle , the bundled subscription fee needs to be optimized .
we first consider the case without bundling .
the provider determines the buying price and subscription fee considering the reservation wage and reservation price of the sensor and user , respectively . *
firstly , the utility of sensor @xmath3 is determined by @xmath4 , where @xmath5 represents the buying price offered by provider @xmath6 , and @xmath7 is the reservation wage .
the sensor will sell its sensing data if the utility is positive . given the buying price @xmath5 , the number of sensors @xmath8 that sell its data can be determined from @xmath9 , where @xmath10 is an indicator function that returns one if the utility of sensor @xmath3 is greater than zero , and zero otherwise . here , @xmath11 is the number of potential sensors .
* secondly , the utility of user @xmath12 subscribing to provider @xmath6 is expressed as @xmath13 , where @xmath14 is the service quality , which is defined as an increasing function of the number of sensors @xmath15 participating in the iot service of provider @xmath6 , @xmath16 is the maximum reservation price , and @xmath17 is the subscription fee .
we assume that the service quality varies between 0 and 1 . naturally , when there are more sensing data from more sensors , the quality of iot service tends to be better .
thus , the users will appreciate more from the service .
for example , if more drivers supply their driving status , it is likely that the estimated road traffic information will be more accurate . here
, @xmath18 is basically a reservation price of the user .
similar to the sensor , the user will subscribe the service if the utility is positive .
the profit of provider @xmath6 is @xmath19 where @xmath20 is an indicator function that returns one if the utility of user @xmath12 is greater than zero , and zero otherwise .
@xmath21 is the total number of potential users .
thus , the optimal buying price and subscription fee are obtained from @xmath22 . a set or a coalition of providers , denoted by @xmath23 ,
can cooperate to offer their iot services as a bundle with the fee @xmath24 .
they can self - organize the coalitions or the third party can facilitate the formation .
for example , an iot search engine can let subscribers , who pay the fee , access all sensing data such as the weather and road traffic conditions .
the cooperative providers will jointly optimize the buying prices and subscription fee of the bundle . while the utility of a sensor is the same as that in the case without bundling , the utility of a user to a bundle
is defined as @xmath25 .
basically , the user will buy the bundle if the total reservation price of all services in the bundle is higher than the bundle subscription fee .
the profit of the coalition of providers is @xmath26 similarly , the buying prices and bundle subscription fee are optimized to maximize the profit .
notably , @xmath27 is the profit of every provider in coalition @xmath23 .
thus , a fair profit sharing scheme is needed .
the scheme must ensure that all providers gain their profit higher than that without joining a coalition , i.e. , making a bundle . from the cooperative game theory literature , the solution concepts such as a shapley value and nash bargaining solution @xcite can be applied . to simplify the presentation of the numerical analysis , we use the following simple setting .
there are two sets of sensors , and thus two providers offering services 1 and 2 .
the reservation wage of the sensors and the reservation price of users are uniformly distributed between 0 and 1 .
the number of sensors in each set is 50 , and the number of users is 200 . the service quality function of a set of sensors is logarithmic , i.e. , @xmath28 , where @xmath15 is the average number of participated sensors .
@xmath11 is the total number of sensors in a set .
@xmath29 is a sensing quality factor to be varied .
a concave quality function is reasonable in the sense that the improvement of the service quality becomes diminishing when there are more sensors .
@xmath30 \end{array}$ ] we first show the impact of the sensing data buying price and service subscription fee to a profit of one service provider .
the provider buys sensing data from one set of sensors and sells a service from processing the sensing data to the users .
figure [ fig : result_profit ] shows the profit of the provider .
apparently , there are an optimal sensing data buying price and service subscription fee that maximize the profit .
the profit is a unimodal function .
thus , numerical methods , e.g. , a simplex method , can be applied to obtain the optimal solutions . in this case , when the buying price is low , the service quality is low due to few sensors selling data , i.e. , low supply .
thus , the provider can not charge a high subscription fee to the users and can not gain a high profit .
if the buying price is too high , the service quality improves , but the cost increases , and the profit plunges .
similarly , when the subscription fee is low , the revenue of the provider is small .
however , when the subscription fee is too high , few users will purchase the service , i.e. , low demand .
consequently , the revenue and profit tumble .
@xmath31 \end{array}$ ] we then consider the impact of the bundle . here
, we consider a symmetric setting of both providers for simplicity .
figures [ fig : result_bundle](a ) and ( b ) show the optimal subscription fees when two providers offer their services separately and as a bundle , respectively .
the locations of the markers in the figures correspond to the reservation prices of users .
for selling service separately ( figure [ fig : result_bundle](a ) ) , the users will subscribe a service if their reservation price is higher than each of subscription fee . here
, the optimal sensing data buying price from both sets of sensors is 0.486 which leads to the maximum reservation price of 0.571 .
the optimal service subscription fees for both providers are 0.286 .
since the services are sold separately , the optimal fees are shown as vertical and horizontal straight lines for providers 1 and 2 , respectively . thus , there are four regions .
users subscribe no service if none of their reservation price is higher than the fees .
users subscribe one of the services if one of their reservation prices is greater than the fee .
finally , users subscribe both the services if all their reservation prices are higher than the fees . for selling service as a bundle ( figure [ fig : result_bundle](b ) )
, the users subscribe both services if the sum of their reservation prices is higher than the subscription fee of the bundled services . here
, the optimal sensing data buying price from both sets of sensors is 0.517 which results in the maximum reservation prices of 0.601 .
the optimal bundled service subscription fee is 0.491 .
thus , there are two regions that correspond to the users subscribing and not subscribing the bundled services . from figures [ fig : result_bundle](a ) and ( b )
, the profit of selling the bundled services is higher than that of selling services separately , i.e. , 38.723 versus 33.524 .
this can be observed from that more users subscribe both services in figure [ fig : result_bundle](b ) than that in figure [ fig : result_bundle](a ) .
additionally , selling services as a bundle allows providers to charge a higher subscription fee .
@xmath32 \end{array}$ ] next , we consider an asymmetric case that the sensing quality factor for the set of sensors for service 2 is varied , while that of provider 1 is fixed at one .
we apply the shapley value for profit sharing between two service providers .
figure [ fig : result_sq ] shows the profits obtained by the two providers when they cooperate and do not cooperate .
clearly , when they cooperate to sell their services as a bundle , the individual profits are higher than that without cooperation , i.e. , selling services separately .
we can observe that the gains from the cooperation for both providers are different .
this implies that the providers can tolerate the integration cost differently .
here , the integration cost incurs because of the cooperation and bundling .
consider the sensing quality factor at 0.7 for service 2 .
the gain for provider 2 is smaller than that of provider 1 , i.e. , 0.97 versus 1.79 .
thus , if the integration cost to provider 2 is more than 0.97 , then provider 2 will not want to sell its service as a bundle .
likewise , if the integration cost to provider 1 is more than 1.79 , then provider 1 will not cooperate . based on the proposed sensing data buying and service subscription with bundling , the following points can be considered for the future work . *
_ strategic sensors and users : _ sensors and users can adjust their reservation wage and reservation price , respectively , based on market conditions . in this case
, an auction can be one of the suitable tools to determine the equilibrium reservation wage and reservation price .
the discriminatory pricing scheme can be developed . *
_ cooperation and collusion : _ to buy sensing data from sensors and to sell services to users , providers can compete or cooperate with each other . in a competitive environment
, the provider will set the buying price and the subscription fee to maximize its individual profit given the strategies of other providers .
a nash equilibrium solution can be adopted .
alternatively , the providers can collude to maximize their profits collectively . the collusion formation and prevention in the market can be studied . * _ quality of data : _ the providers can adjust the buying price for each sensor individually to encourage it to supply high quality sensing data . in particular , the sensors can optimize data quality based on the buying price and their resource usage for collecting , processing , and transmitting data to the provider .
for example , a camera as a sensor can supply higher video quality , but at the cost of more energy consumption and bandwidth usage .
hence , the camera will do so only when the buying price is sufficiently high .
joint pricing and performance optimization models can be developed .
in this article , we have considered smart data pricing ( sdp ) for iot systems and services .
we have first introduced an overview of iot including its architecture , benefits , and business models .
then , we have reviewed some related work of applying sdp to iot .
we have proposed a new pricing scheme for iot service providers taking into account sensing data buying and subscription with bundling .
the numerical results have clearly shown that with bundling , multiple providers can form a coalition to achieve higher profit .
j. gubbi , r. buyya , s. marusic , and m. palaniswami , `` internet of things ( iot ) : a vision , architectural elements , and future directions , '' _ future generation computer systems _ , vol .
1645 - 1660 , september 2013 .
j. s. lee and b. hoh , `` sell your experiences : a market mechanism based incentive for participatory sensing , '' in _ ieee international conference on pervasive computing and communications _ , pp . 60 - 68 , 2010 .
l. mei , w. li , and k. nie , `` pricing decision analysis for information services of the internet of things based on stackelberg game , '' in _ proceedings of international conference on logistics , informatics and service science ( liss ) _ , springer , pp .
1097 - 1104 , 2013 . t. le , m. beluri , m. freda , j .-
gauvreau , s. laughlin , and p. ojanen , `` on a new incentive and market based framework for multi - tier shared spectrum access systems , '' in _ proceedings of ieee international symposium on dynamic spectrum access networks ( dyspan ) _ , pp .
477 - 488 , 1 - 4 april 2014 .
x. zhang and b. li , `` on the market power of network coding in p2p content distribution systems , '' _ ieee transactions on parallel and distributed systems _ , vol .
2063 - 2070 , december 2011 . | internet of things ( iot ) has emerged as a new paradigm for the future internet . in iot , enormous devices are connected to the internet and thereby being a huge data source for numerous applications . in this article , we focus on addressing data management in iot through using a smart data pricing ( sdp ) approach . with sdp ,
data can be managed flexibly and efficiently through intelligent and adaptive incentive mechanisms .
moreover , it is a major source of revenue for providers and partners .
we propose a new pricing scheme for iot service providers to determine the sensing data buying price and iot service subscription fee offered to sensor owners and service users , respectively .
additionally , we adopt the bundling strategy that allows multiple providers to form a coalition and bid their services as a bundle , attracting more users and achieving higher revenue .
finally , we outline some important open research issues for sdp and iot .
iot , bundling strategy , smart data pricing , pricing mechanism , incentive . |
in close analogy with the case of the spins ( eqns .
( [ soodd ] ) and ( [ soeven ] ) in the text ) , we can construct string order parameters also in the charge sector .
we define @xmath91 and @xmath92 as @xmath93 where @xmath94 . as shown in fig .
[ chargestring ] , both @xmath92 and @xmath91 are finite for a generic band insulator and - up to a trivial factor @xmath95 - very similar ( but not identical ) to the spin string order parameters of fig .
[ stringorder ] . in the purely diagonal ( @xmath96 ) and vertical ( @xmath97 ) limits only one of the two string orders is different from 0 .
the presence of charge string order comes as a natural consequence of the symmetry existing in the one - dimensional band insulator between the spin and the charge sectors . in bosonization language ,
both the spin and charge fields are pinned due to the presence of a relevant cosine term in the low energy bosonic hamiltonian @xcite , leading therefore to finite string order parameters @xcite . as already discussed in the main text , this scenario is stable also in presence of interactions ( at least for truly one - dimensional systems ) , as long as both the bosonic fields are massive . in the limit of large @xmath18 , the charge string orders will be be strongly suppressed by powers of @xmath98 due to the smallness of the charge fluctuations ( becoming strictly zero for a pure spin model , i.e. @xmath99 ) but will remain finite for any finite @xmath18 . in the case of the ionic hubbard model ( corresponding to the higher symmetry manifold @xmath40 and @xmath70 in the parameter space of eq .
( [ hamiltonian ] ) ) the charge field becomes massless at @xmath100 due to the competition between two relevant cosine perturbations @xcite ( the on - site energy and the coulomb interaction ) . as both for @xmath101 and @xmath75 the charge field
is locked , the charge string order will be finite but it will vanish at the gapless point @xmath100 ( the spin string order stays finite at @xmath74 ) .
starykh , l. balents , phys .
lett . * 93 * , 127202 ( 1997 ) .
kolezhuk , h .- j mikeska , phys .
b * 56 * , 11380 ( 1997 ) .
m. fabrizio , a.o .
gogolin and a.a .
nersesyan , phys .
* 83 * , 2014 ( 1999 ) .
m. khl _ et al .
_ , phys .
lett . * 94 * , 080403 ( 2005 ) .
diener and t.l . ho , phys . rev .
lett . * 96 * , 010402 ( 2006 ) .
gogolin , a.a .
nersesyan , a.m. tsvelik , _ bosonization of strongly correlated systems _ , cambridge ( 1998 ) . | the ground state of spin-1 haldane chains is characterized by the so - called string order .
we show that the same hidden order is also present in ordinary one - dimensional band insulators .
we construct a family of hamiltonians which connects adiabatically band insulators to two topologically non - equivalent spin models , the haldane chain and the antiferromagnetic spin-1/2 ladder .
we observe that the localized spin-1/2 edge - state characteristic of spin-1 chains is smoothly connected to a surface - bound state of band insulators and its emergence is not related to any bulk phase transition .
furthermore , we show that the string order is absent in any dimensions higher than one .
the antiferromagnetic spin-1 chain is a prominent example of a spin - liquid , an exotic state of matter where interactions play a dual role : on the one hand a strong hund s rule coupling locally leads to the formation of magnetic spin-1 degrees of freedom , on the other hand spin correlations screen magnetism completely . as conjectured by haldane in the early eighties
@xcite , spin-1 chains are characterized by a finite gap in their spectrum and by exponentially decaying spin - spin correlations . since this seminal work a lot of progress have been made , theoretically and experimentally , towards a full description of the so - called haldane phase . in particular , affleck ,
kennedy , lieb and tasaki @xcite provided a microscopic understanding of this correlated ground state , devicing an exactly solvable model ( aklt ) adiabatically connected to the spin-1 chain .
the aklt ground state is made up solely of nearest - neighbor singlets ( or valence bonds ) and the gap in the system ( and therefore in the spin-1 chain ) corresponds to the energy needed to break a bond . in this paper
we want to address the question to what extent the haldane chain is different from an ordinary band insulator .
the latter has also a gap and an even number of electrons per unit cell and , recently , one of the authors has shown @xcite that luttinger surfaces can not be used to distinguish band- from correlated insulators .
two properties , though , are considered peculiar of the haldane phase : first , when a spin-1 chain is cut , a localized fractionalized excitation , namely a spin-1/2 , is formed at the boundary of the chain .
these boundary spins have for example been observed in the nmr profile close to the chain ends of mg - doped @xmath0 @xcite .
second , the ground state possesses a hidden long - range order characterizing the entanglement of the spins : den nijs and rommelse @xcite noticed that , even though true nel order is absent in the ground state , any site with @xmath1 is followed by another with @xmath2 , separated from the first by a string of @xmath3 of arbitrary length .
this implies that the so - called string order ( so ) parameter @xmath4 is always finite in the haldane phase .
a spin-1 chain can be deviced on a spin-1/2 ladder using a strong ferromagnetic rung coupling @xmath5 which binds two spin 1/2 into a single spin 1 . in this case
the string order parameter takes the form @xcite @xmath6 and a typical ground state wave function in the haldane phase is shown at the bottom of the first column of fig .
[ string ] . as emphasized by bonesteel @xcite and kim _ et .
al . _
@xcite , this wave function has a topological feature , namely a vertical line ( dashed in the figure ) crosses always an _ odd _ number of singlets . as can be easily seen from the figure , this property
is directly related to the existence of edge states at the boundaries . on the other hand
, an antiferromagnetic rung coupling leads to another gapful phase , with a ground state in a different topological sector of the space of singlet wave functions . here
, always an _ even _ number of singlets is crossed by a vertical and edge states are absent ( see fig .
[ string ] ) .
correspondingly @xcite the odd string order @xmath7 vanishes and instead one gets a finite even string order @xmath8 @xmath9 in spin - ladder models these two orders never coexist as they characterize two different topological sectors . in the following
we will first show that the ground state of an ordinary band insulator can be adiabatically deformed into the ground state of either the haldane chain or the antiferromagnetic ladder .
then we will investigate the role both of the string order and of the boundary states in the band - insulating phase .
finally , we consider the fate of the string order in dimensions larger than one . the starting point of our discussion is the following family of ladder hamiltonians ( see top of fig . [ string ] ) : @xmath10 where @xmath11 indicates the two legs of the ladder ( or two orbitals of an atom ) .
varying the set of parameters @xmath12 $ ] , it is possible to span a rich phase diagram including a mott - insulating haldane phase ( @xmath13 , @xmath14 ) , an antiferromagnetic ladder ( @xmath15 , @xmath16 ) and band - insulators ( @xmath17 ) .
we now describe how these states can be connected adiabatically following the route depicted in fig .
[ string ] .
first , it has been shown both numerically @xcite and using analytic arguments @xcite that the haldane chain can be adiabatically deformed into a spin-1/2 model with only diagonal antiferromagnetic couplings , and , similarly , the antiferromagnetic ladder is connected to a pure rung singlet model ( second last row of fig .
[ string ] ) .
these local antiferromagnets can be realized as the large @xmath18 limit of the two - sites hubbard hamiltonian @xmath19 for the diagonal or the rung , respectively . for arbitrary @xmath20 ,
the three lowest eigenvalues of @xmath21 are a singlet with energy @xmath22 , a triplet with @xmath23 and a charge excitation with @xmath24 .
the gap @xmath25\end{aligned}\ ] ] is always finite and , therefore , one can switch off the interaction @xmath18 completely , connecting adiabatically the mott- and the band - insulating ground states .
note that the nature of the lowest excitation changes completely when @xmath26 while the ground state rotates smoothly @xcite from a state describing two localized spins to one where two electrons occupy the bonding band .
( @xmath27 , @xmath28 and @xmath29 ) . in the inset
the localization length of the bound state is shown . ] as a final step , we have to show that these two local insulators can be adiabatically connected in the non - interacting limit , @xmath30 where the band - structure is given by @xmath31^{1/2}.\end{aligned}\ ] ] the states in the second row of fig .
[ string ] are described by @xmath32 and @xmath33 , respectively . for @xmath34 and @xmath35
, the state is always a band insulator with a finite gap .
therefore by reducing @xmath36 and increasing @xmath37 for finite @xmath38 we can complete the adiabatic path of fig .
[ string ] . for @xmath39 the hamiltonian of eq .
( [ hamiltonian ] ) has a quantum critical point at @xmath40 ( related to an extra particle - hole symmetry of this model ) . in fig .
[ bondstate ] we show that for small @xmath38 the gap remains finite upon tuning from the diagonal to the rung insulator . ) .
the dashed lines denote the adiabatic path of fig .
[ string ] .
for sufficiently large interactions , either a second or first order transition or new intervening phases are expected . ]
after having shown the adiabatic continuity , we consider now the evolution of the string order along the adiabatic path .
in eqns .
( [ soodd ] ) and ( [ soeven ] ) the @xmath41 were operators living in the spin hilbert space while for the fermionic model we have to replace those by @xmath42 .
for the two local singlet states of the third row of fig .
[ string ] the string order is maximal @xcite , with @xmath43 , @xmath44 and @xmath45 , @xmath46 , respectively . remarkably , the string order is also finite for the two local band insulators of the second row : @xmath47 , @xmath44 and @xmath48 , @xmath46 , respectively .
the factor @xmath49 reduction arises because empty and doubly occupied states appear with a probability of @xmath50 , affecting the spin operators at the beginning and at the end of the string .
the string , however , still contributes with a factor @xmath51 . how can these different order parameters be connected ? to answer this question we calculate the two string orders along the adiabatic path connecting the two local insulators ( i.e. on segment b of fig .
[ adipath ] ) . with the help of the identities @xmath52 and @xmath53 where @xmath54 denotes normal ordering
, we can reformulate eq .
( [ soodd ] ) as @xmath55 and a similar expression is valid for eq .
( [ soeven ] ) . in analogy to the string order in the spin sector
one can also introduce a similar quantity in the charge sector , see appendix [ chargeorder ] .
for a non - interacting system , one can easily evaluate expectation values of the form @xmath56 $ ] using functional integration , where @xmath57 is the equal - time greens function matrix .
we therefore obtain @xmath58 , \label{determinant}\ ] ] where the matrices @xmath59 can be read off from eq .
( [ norder ] ) and the @xmath57 matrix can easily be obtained numerically for arbitrary @xmath60 .
also the determinant in eq .
( [ determinant ] ) can be easily computed for large systems and the so parameters are obtained from the limit @xmath61 , where @xmath62 is the correlation length . and
@xmath63 , for the non - interacting ladder as a function of @xmath64 ( @xmath27 , @xmath28 , @xmath29 ) .
inset : in the presence of charge fluctuations a vertical line can cross both even and odd number of singlets . accordingly
the two string orders can coexist . ]
we find that in one - dimensional band insulators string order is always present even in the absence of interactions !
in contrast to the pure spin models generically both string order parameters are finite ( see fig .
[ stringorder ] ) and they are smooth functions of @xmath64 for @xmath65 . in our model , for @xmath40 the two string orders are equal by symmetry , @xmath66 ( this follows from the more general relation @xmath67 that can be understood in terms of the following transformation on the ladder : shift of the lower leg by one lattice site and mirror - reflection of the ladder respect the x - axis ) . only for @xmath39 and @xmath40 , when the gap vanishes , both @xmath68 and @xmath69 deplete to 0 because at the critical point singlets of arbitrary length are generated .
the two orders , @xmath69 and @xmath68 , coexist in gapped insulators also in the presence of strong interactions as long as ( virtual ) charge fluctuations are allowed . this can be seen from the inset of fig .
[ stringorder ] : a doubly occupied or empty site allows to switch from diagonal to rung singlets .
we now outline the consequences of our discussion for the phase diagram of generalized hubbard ladders , sketched in fig .
[ adipath ] .
first , gapped systems are stable against small perturbations and therefore there is a finite range of parameters which allows to connect adiabatically haldane states and antiferromagnetic ladders .
second , in a pure spin model it is _ not _ possible to connect these two phases as their wave functions live in different topological sectors of the spin hilbert space .
therefore , we expect that beyond some critical value of @xmath18 a phase transition line ( first or second order ) or an intervening phase should separate the even and odd sectors@xcite . for pure spin models , starykh and balents@xcite analyzed the possible intervening phases and concluded that without extra fine - tuning there is never a direct second order transition from the haldane to the antiferromagnetic ladder . either the transition is first order or intervening phases with broken translational invariance appear .
the absence of a direct second - order transition is consistent with our finding that the two high - symmetry phases are adiabatically connected .
to gain further insight in nature of the phase diagram sketched in fig .
[ adipath ] at finite @xmath18 , we observe that for @xmath40 and @xmath70 , our ladder gains an extra mirror symmetry ( this is evident unfolding the ladder on a chain , as shown in the lower part of fig .
[ gaps ] ) and we expect that its properties are similar to the ones of the well - studied@xcite one - dimensional _ ionic _ hubbard model ( i.e. a hubbard model with a staggered potential ) which has the same symmetries . for this system upon increasing @xmath18 ( see upper part of fig . [ gaps ] )
first the charge gap of the band insulator closes and one obtains a dimerized phase for @xmath71 . for @xmath72 the system becomes a uniform , gappless mott insulator ( effectively a uniform heisenberg _ chain_@xcite ) .
we therefore expect that @xmath73 in the schematic phase diagram of fig .
[ adipath ] has a similar role as @xmath74 , i.e. the charge gap vanishes and a dimerized phase ( not shown in fig .
[ adipath ] ) appears for @xmath75 , @xmath76 , where @xmath68 is bigger or smaller then @xmath69 , dependingly on which of the two degenerate ground states one considers . according to the results of starykh and balents@xcite discussed above , we do , however
, not expect a gapless phase for generic parameters ( e.g. @xmath77 ) .
the role of string order in the charge sector of the ionic hubbard model is briefly discussed in appendix [ chargeorder ] . .
@xmath78 and @xmath79 are the charge and spin gaps and @xmath80 is the dimerization order parameter @xcite .
bottom : the ladder of eq .
( [ hamiltonian ] ) unfolded on a chain .
an extra mirror symmetry is present for @xmath40 and @xmath70 . ] at this point we would like to investigate the emergence of the edge - states following again the path depicted in fig .
[ string ] .
the localized spin 1/2 at the boundary of the haldane phase evolves smoothly in a surface bound state of the diagonal band insulator .
for finite interactions and a proper choice of the chemical potential , this surface bound state will be singly occupied giving rise to a localized spin . in fig .
[ bondstate ] we show how this edge state merges with one of the bands when @xmath37 is increased for @xmath81 . at this point , the localization length of this state diverges with @xmath82 ( inset of fig .
[ bondstate ] ) .
nevertheless , this is not related to any bulk phase transition . up to now
we have only discussed one - dimensional ladders and it is interesting to study the role of string order in higher dimensions . especially in gapped systems one would naively expect that any type of long - range order is stable against small perturbations like a coupling to neighboring ladders . using the methods described above , we can easily calculate the string order for two- and three - dimensional band insulators numerically .
surprisingly we find that for arbitrarily weak inter - ladders coupling , @xmath83 , the string order decays exponentially , @xmath84 with @xmath85 ( see fig . [ 2dstring ] ) .
a similar observation has been made previously by todo _
et al . _
@xcite , who studied spin-1 ladders numerically .
the decay of the string order is a consequence of ( rare ) inter - ladders singlets which introduce a random phase @xmath86 in the string .
we note that obvious generalizations @xcite of the string order parameter to higher dimensions ( replacing strings by squares or cubes ) are ineffective again due to dangling singlets at the infinite surface of such structures . ) for a two - dimensional lattice decays as @xmath87 ( with @xmath27 , @xmath28 , @xmath29 , @xmath88 and @xmath89 from up to down ) .
inset : @xmath11 as a function of @xmath90 . ] in conclusions , we have shown that three phases with both spin- and charge gaps and two electrons per unit cell , namely haldane chains , band - insulators and antiferromagnetic ladders , are actually all the same in the sense that their unique ground states can be adiabatically connected ( the excitation spectra , however , may change ) .
this cousinship is reflected in the prevalence of string order in all these one - dimensional phases , even in non - interacting band insulators . however , string order is not robust in higher dimensions , where it is destroyed by arbitrarily weak coupling@xcite .
an interesting open question is the precise role of the schrieffer - wolff transformation , a central tool to derive effective models .
such a unitary transformation eliminates charge degrees of freedom completely and therefore maps the non - magnetic ground state of hamiltonians with charge gap into one of the topological sectors of the spin models . as we have connected two _
different _ sectors adiabatically along our path , this seems to imply that the schrieffer - wolff transformation has to break down at some point .
this is consistent with a scenario where at @xmath73 in fig .
[ adipath ] the charge gap closes .
however , one also has to take into account that a schrieffer - wolff transformation is not uniquely defined as it depends e.g. on the choice of a single - particle basis and therefore the topological sector may not be uniquely defined in the presence of charge fluctuations .
the rapid progress in the control of fermionic atoms in optical lattices will make possible to study the adiabatic evolution of correlated ground states also experimentally in the near future @xcite .
we acknowledge useful discussions with r. moessner , e. mller - hartmann , a.a .
nersesyan , a. schadschneider , a.m. tsvelik and j. zaanen , j. zittartz and , especially , g.i .
japaridze .
we thank for financial support of the dfg under sfb 608 . |
the wlm galaxy ( ddo 221 ) is a local group dwarf irregular ( class irr iv - v ) , in an isolated part of the local group approximately 900 kpc from the milky way .
it was discovered by wolf ( 1909 ) and rediscovered by lundmark and melotte ( melotte 1926 ) , thus earning its name of wolf - lundmark - melotte , abbreviated wlm .
its distance is approximately 900 kpc , making it close enough to allow ground - based studies of its red giant branch ( rgb ) tip ( @xmath1 ) and upper main sequence , but distant enough that such studies will not reach the horizontal branch ( @xmath2 ) .
ables & ables ( 1977 ) used electrographic techniques to make photometric measurements of supergiants in wlm , getting @xmath3 and @xmath4 photometry for 50 stars , the faintest having a magnitude @xmath3 of 21.6 . with this photometry , they were able to estimate a distance modulus of @xmath5 .
they also noted a `` slightly diffuse starlike object about 2 west of the center '' of the galaxy , which they determined was most likely a globular cluster with @xmath6 and @xmath7 .
finally , they made contour maps in @xmath3 and @xmath4 of the galaxy , finding a color gradient with @xmath8 in the core and @xmath9 in the outer regions .
sandage & carlson ( 1985 ) studied the brightest supergiants in wlm using a photometer and determined periods for 15 cepheids . with the cepheid results ,
the distance modulus was heavily revised to @xmath10 , a drop of 1.3 magnitudes .
the first ccd study of this galaxy was made by ferraro et al .
( 1989 ) , with later work made by minniti & zijlstra ( 1997 ) .
both works reached @xmath11 , deep enough to accurately measure the rgb tip but not the hb .
the ferraro et al .
( 1989 ) data consisted of two fields , both partially in the bar and partially in the halo .
they were able to conclude that the two fields had different star formation rates in the very recent past ( and therefore that wlm s star formation was localized rather than global ) , and that a salpeter imf was adequate for their data .
minniti & zijlstra ( 1997 ) made a @xmath4 and @xmath12 study of wlm , which allowed them to study the older stars in greater detail .
they found that star formation began at least 10 gyr ago in the galaxy , with a metallicity [ fe / h ] of -1.45 @xmath13 0.2 .
a distance modulus of @xmath14 was calculated from the rgb tip .
importantly , they observed that there was a lack of bright main sequence stars in the galaxy s halo , and therefore the color gradient observed by ables & ables ( 1977 ) was due to a stellar population gradient . with hst observations of wlm , we are now able to peer much deeper into the stellar content of this galaxy , with quality @xmath4 photometry to magnitude 27 ( @xmath15 ) .
a distance modulus of @xmath16 was calculated for the globular cluster from color - magnitude diagram fitting ( hodge et al .
1999 ) , which was centered in the pc camera of wfpc2 .
the data from the wf cameras also provide the opportunity to study the star formation history in a quantitative manner , using the method detailed in dolphin ( 1997 ) .
the metallicity enrichment history has not been studied extensively for this galaxy either , with only two solid estimates of the metallicity .
hodge & miller ( 1995 ) used spectra of two hii regions in wlm to deduce a current metallicity of approximately -1.15 @xmath13 0.2 , while minniti & zijlstra ( 1997 ) estimated that the old stars have a metallicity of -1.45 @xmath13 0.2 through photometry .
these two figures provide rough endpoints of the enrichment history , but are very uncertain ( to the point that both are consistent with -1.3 )
. a better understanding of this galaxy s enrichment history is possible through the hst data presented here .
goals for this study will be : * determining the distance from cmd analysis , and comparing it with that obtained by hodge et al .
( 1999 ) and an rgb tip analysis . * dating the oldest stars in the galaxy , and determination of star formation rates since that time .
* determining the chemical enrichment history and comparing with estimates of the old star metallicity and measurements of the hii region metallicities . *
determining the nature of the stellar population differences . has the star formation rate in the bar
increased recently , that in the halo decreased , or a combination of the two ? * quantifying the recent star formation history differences : how much different are the recent star formation rates in different regions , and how long have they been different ?
as part of a cycle 6 hst go program , four images of part of wlm were taken on 28 september , 1998 .
two images were taken through the f555w filter with exposure times of 2600 and 2700 seconds , and two were taken through the f814w filter with exposure times of 2700 seconds each .
the pc chip was centered on wlm s globular cluster , the data for which was analyzed separately ( hodge et al . 1999 ) , and the orientation of the camera was such that the wf chips lay approximately along the galaxy s minor axis , providing a sample of wlm bar and halo stars .
the alignment of the chips put wf2 the closest to the bar of the galaxy , wf3 further out , and wf4 well outside the bar .
cosmic ray cleaning was made with a routine similar to that of the iraf task crrej .
the cleaned @xmath4 image is shown in figure [ fig - image ] .
stellar photometry was carried out using the hstphot package ( dolphin 1999 ) used in hodge et al .
this program is optimized for undersampled wfpc2 images , and uses a hybrid psf fitting and aperture photometry routine to determine the position and brightness of each star located .
cmds are given for all three wf chips combined , and for each chip separately , and are shown in figure [ fig - cmds ] .
for all cmd analyses in this paper , the padova isochrones ( girardi et al .
1996 , fagotto et al .
1994 ) were used for the generation of synthetic cmds .
these models include later ( post - helium flash ) phases of stellar evolution , despite the large uncertainties in those phases , providing completeness at the cost of accuracy .
for the purpose of matching the entire observed cmd with synthetic data , however , it is far better to use models with slightly incorrect agb , red clump , and horizontal branch phases than models that omit them entirely . the star formation history solution method used here is based on that described in dolphin ( 1997 ) , and involves modeling the observed cmd with synthetic cmds .
this process involves three steps .
first is the generation of synthetic cmds , which requires theoretical isochrones , an artificial star library , and parameters of star formation : star formation rate , metallicity ( both functions of time ) , distance , imf , and extinction . for this data set , which is not deep enough to observe below the ancient main sequence turnoff ,
a salpeter imf was assumed , with endpoints of 0.1 and 120 @xmath17 . for a single solution
, a distance , extinction , and metallicity history would be assumed , while the star formation rates would be the free parameters . since any complex population
( multiple ages and abundances ) can be modeled as the sum of its component populations , its cmd is thus the sum of the cmds of the component populations .
thus all that is required to generate any number of cmds with various past star formation rates is a set of cmds of each of the individual component populations .
this was accomplished by dividing the time from the beginning of the galaxy s star formation to the present into segments , each with a constant star formation rate of 1 @xmath17/yr . by multiplying each cmd by the average star formation rate over its period
, one will then have the total synthetic cmd .
the second step is the conversion of the data into a format that can be fit . to accomplish this ,
each cmd was divided into a large number of square bins , with the number of stars falling in each counted .
the synthetic cmds , generated from the theoretical isochrones and the artificial star library , were given the same treatment .
two sizes of bins were used for fits . a lower - resolution fit , with bin size 0.1 mag in @xmath18 and 0.3 in @xmath4 , would be more sensitive to larger features , such as luminosity functions along the main sequence and rgb . a higher - resolution fit , with bin sizes half those of the lower - resolution fit , would be more sensitive to details . finally , a fit must be made for the star formation rates of each period of time , attempting to find the star formation rates for the single population cmds that , when the cmds are combined , will create the `` best '' fit to the observed data .
a standard least-@xmath19 fit can be easily thrown off when dealing with common observational errors , most notably the presence of observed `` stars '' ( cosmic rays , noise , binary stars , galactic foreground contamination , etc ) in regions of the cmd where the models do not predict any stars .
such a case will give a number of expected stars of @xmath20 in that region of the cmd , which will clearly make it the most important part of the fit . to compensate
, a small constant value was added to the @xmath21 denominators of the @xmath19 calculations .
additionally , the red clump , horizontal branch , and agb regions of the cmd were given less weight in the fits , again so that unfittable parts of the cmd would be less important in the fitting .
finally , @xmath19 value for any bin not allowed to exceed 10 , so that no one piece of the cmd would determine the fit .
this modified @xmath19 goodness - of - fit parameter was thus used in by an amoeba algorithm to solve for the star formation rates , given the assumed enrichment , distance , and extinction . in order to make a search for those values as well , a large number of fits was made , each with a different combination of initial metallicity , current metallicity , metallicity enrichment function , distance , and extinction . all solutions with a modified @xmath19 near that of the best fit
were averaged together for the determination of best values and uncertainties of the metallicities , distances , extinctions , and star formation rates .
thus a parameter that strongly affected the solution would give a small uncertainty , while one ( such as initial metallicity ) with only a weak effect would give a large uncertainty .
additionally , any relations between parameters ( such as the age - metallicity - distance `` degeneracy '' ) are accounted for in the uncertainties .
in addition to the distance and extinction , the time of initial star formation in wlm is an essential parameter to calculate before determining the galaxy s star formation history .
this is because the evolutionary models reproduce neither the horizontal branch nor the red clump structures well , and thus a blind solution using the models can easily be thrown off in dealing with these old populations . in the case of wlm ,
doing so would cause the models to attempt recreating the tight red clump as a red horizontal branch .
but clearly the field has little or no horizontal branch , which sets the upper limit on large amounts of star formation to @xmath2212 gyr ago . to determine the age of initial star formation , a set of history calculations
was made that were sensitive to the age of initial star formation , with the distance and extinction adopted from above and the enrichment history allowed to vary .
of the possible starting points , the oldest ( 12 gyr ) gave the best fits and is used below .
the question of whether the galaxy actually had _ no _ star formation from the formation of the globular cluster ( 14.8 @xmath13 0.7 gyr ago ) to 12 gyr ago , or whether the galaxy had a non - zero but very small star formation rate can not be solved from this data , as there are stars where the horizontal branch would be expected , but no horizontal feature is seen in the cmd , placing an upper limit of @xmath23/yr on the star formation rate between 12 and 15 gyr ago .
regardless of whether or not the galaxy waited until 12 gyr to form any stars , it clearly waited until that point to form a measurable number of stars .
the star formation history study was broken into two parts .
first was the study of the global star formation history and enrichment , which used the combined cmd from all three wf chips .
the distance and extinction were also determined here , permitting a self - consistent solution to be found . for the size of the region studied ( @xmath22500 pc ) , any stars formed more than a gyr ago should be mixed by the present time ( with a random velocity of 1 km / sec , a 500 pc region would be mixed in 500 myr ) , and thus the combination of the cmds will simply improve the signal - to - noise of the results . additionally , the best star formation history information is only available to 1.5 gyr , beyond which point the main sequence drops below the limit of good photometry , so the signal - to - noise improvement obtained by combining the fields is essential . to study the global star formation history , a set of star formation history calculations
was made allowing metallicity enrichment history ( initial metallicity , current metallicity , and function of metallicity increase ) , distance , and extinction to vary , with the best - fitting solutions averaged with a weighted average to determine the best results .
uncertainties were calculated by adding the standard deviations from the fitting procedure to the standard deviations of monte carlo simulations in quadrature , so that both uncertainties resulting from the parameters and those from the fit itself are included in the final results .
this solution was made using 9335 stars with @xmath24 and @xmath25 .
the measured distance modulus was @xmath0 , with an extinction of @xmath26 .
this distance is somewhat larger than that measured for the globular cluster ( @xmath16 in hodge et al .
1999 , later revised to @xmath27 using improved photometry ) and the rgb tip distance measured by minniti & zijlstra ( 1997 ) of @xmath28 , but is within uncertainties of both .
an rgb tip analysis , using the method given by lee et al .
( 1993 ) , gives a distance of @xmath29 , which is consistent with the above measurement .
this distance is also consistent with the cepheid distances given in section 1 .
star formation rates are given assuming a salpeter imf with cutoffs at 120 and 0.1 @xmath17 , which was used out of necessity because this data does not extend beyond the old main sequence turnoff and therefore the imf can not be measured .
an imf error for the old stars would introduce a gradient into the data . for example , using too shallow of an imf would generate fewer stars in the older synthetic cmds , and thus return too large of a star formation rate in the very old times .
the lower cutoff is also a simplification , again made because the data contains no information about low - mass stars .
most likely the true mass of stars formed is lower than the numbers given here , scaled by some constant factor .
results are shown in table [ tab - oldsfh ] and figure [ fig - oldsfh ] .
a synthetic cmd for the region reconstructed from the models , artificial star results , and determined star formation history is shown in figure [ fig - synth]a .
a glance shows that the fit appears to be good , with the primary differences between the two diagrams ( red clump position , strongly populated blue loop , and brighter agb ) located in the regions where theoretical models are highly uncertain .
the lower part of the visible main sequence is more populated in the reconstructed cmd , which is most likely a result of the assumption of constant star formation rate across the time of each bin . since the youngest bin lasts from 200 myr ago to the present , any change in the star formation rate over that time will alter the ms luminosity function and not be reproduced in the synthetic cmd . in this case , the recent star formation rate is likely higher than the 200 myr average , thus producing a flatter ms luminosity function in the observed data . at any rate ,
the recent star formation rates are studied in the next section , and will not be dealt with here .
it is encouraging that the cmd features that are expected to be well - reconstructed , namely the positions of the main sequence and the red giant branch , are indeed correct .
the reconstructed rgb is somewhat too broad , a result of the uncertainties in modeling old star formation rates and metallicity enrichment .
it should be stressed that there were no built - in assumptions of the metallicity , distance , or extinction in this star formation history solution ; the star formation history program was able to determine all of these values and produce a synthetic cmd that closely resembles the observed cmd .
a more detailed comparison is shown in figures [ fig - comp ] and [ fig - comp2 ] , which show the binned cmds that were used for the solution ( 0.1 magnitude resolution in @xmath18 and 0.3 magnitude resolution in @xmath4 for figure [ fig - comp ] ; 0.05 @xmath18 and 0.15 @xmath4 for figure [ fig - comp2 ] ) , as well as the residuals and the fit parameter values .
magnitude limits in these plots are @xmath30 and @xmath31 .
the subtracted cmds are shown at 2.5 times the contrast level of the two cmds , while the modified @xmath32 plots are shown on a scale from -9 to + 9 @xmath33 . for both ,
a point darker than average is one with more synthetic points than observed .
the largest errors in fitting the cmd come in the red clump and horizontal branch regions , which along with the agb are the regions with the highest uncertainty in the models .
thus errors here are not surprising , and do not necessarily mean that the fit quality is poor .
most notably , the models still show a slight red rgb , despite the 12 gyr age limit placed on the fit , while the observed cmd shows a red clump slightly brighter and bluer .
errors from the red clump region are up to 9 @xmath33 ( in the undersubtracted region bluewards of the red clump and blue loop base ) .
additionally , the theoretical agb has a much different slope than that observed here , thus making it impossible to adequately model .
however , the erroneous agb modeling provides very little to the fit ( no more than a 1 @xmath33 error at any point ) , as it contains few stars .
the red giant branch is another source of error , with the red edge fit very well , but the blue edge problematic .
the blue edge is drawn out by the program s attempt to fit the bluest red giant branch stars as the old population . a solution which breaks the old bin ( 9 - 12 gyr old ) into multiple bins would solve this problem , allowing a more flexible enrichment and star formation history , although this would imply better resolution than is really possible in this type of work .
undersubtraction of the observed blue edge of the rgb produces errors of up to 6 @xmath33 , while oversubtraction in the modeled blue edge produces errors of up to 3 @xmath33 .
as noted above , the main sequence is oversubtracted at the faint end ( @xmath34 ) .
this contributes a 3 @xmath33 error in the solution , and is the result of changes in the star formation rate within the past 200 myr .
looking at the calculated star formation history , it seems that the star formation of wlm began with a strong initial star formation episode , beginning approximately 12 gyr ago .
this event contained about half ( by mass ) of the total star formation that wlm has had over its lifetime .
the star formation rate fell considerably after that episode , although uncertainties are sufficiently large that nothing can be said for certain about the period between 2.5 and 9 gyr ago except that there was some star formation , and likely with an average rate of 1 or 2 @xmath35/yr .
the recent star formation history will be covered in the next section .
the metallicity enrichment of wlm shows an initial abundance of -2.18 @xmath13 0.28 , although a glance at the cmd tells that few stars formed with such a low metallicity .
the metallicity at the tail end of the 3 gyr initial star - forming episode was -1.34 @xmath13 0.14 .
this result is consistent with the result of minniti & zijlstra ( 1997 ) , that the average metallicity of old rgb stars is -1.45 @xmath13 0.2 , as well as the hodge et al .
( 1999 ) measurement of the globular cluster metallicity of -1.51 @xmath13 0.09 ( revised to -1.63 @xmath13 0.14 with improved photometry ) . the metallicity has continually climbed since that time , reaching a current value of -1.08 @xmath13 0.18 .
this value for the current metallicity agrees well with the [email protected] found by hodge & miller ( 1995 ) for two hii regions in wlm .
in contrast to the well - mixed old stars , minniti & zijlstra ( 1997 ) and ferraro et al .
( 1989 ) found a stellar population gradient in wlm , with the bar region contributing more recent star formation .
as their data extended to @xmath36 , it was impossible to quantify this result , but with the hst data here a star formation history analysis can be done separately on different parts of the galaxy . for the study of recent star formation ,
the three fields were kept separate , with 6381 stars in wf2 , 2165 in wf3 , and 789 in wf4 .
the analysis was carried out using the distance , extinction , and enrichment calculated above , with results shown in table [ tab - newsfh ] and figure [ fig - newsfh ] . reconstructed cmds of the three fields
are shown in figures [ fig - synth]b - d .
as with the combined cmd , the chief differences between observed and reconstructed cmds are in the blue edge of the rgb , the red clump sharpness and position , and the agb position , resulting from the models uncertainties in late phases of evolution .
the reconstructed wf2 field , in particular , shows a very wide red clump , apparently due to overestimated incompleteness and crowding errors in that field
. otherwise , the fits are excellent , as with the combined cmd .
as shown by the global star formation history , wlm has had a burst of star formation activity during the past gyr and possibly longer .
determining star formation rates beyond 2.5 gyr is much more difficult , as the main sequence is below the limit of the good photometry , so only star formation rates to 2.5 gyr are shown for the fields . the wf2 field , which is the closest to the galaxy s center ( and includes the outer regions of hii complexes c1 and c2 as well as including all of hii region 1 listed in hodge & miller 1995 ) , contains most of the recent star formation ( 85% of the star formation in the past 200 myr ) , while the 1000 to 2500 myr star formation rate is more evenly distributed between the regions , which can be explained through mixing .
the hodge & miller ( 1995 ) data on wlm provide an additional constraint on the current star formation , as they find a formation timescale ( defined as total star formation amount divided by current star formation rate ) of 150 gyr for the galaxy , significantly longer than the value of 14 gyr that would be calculated from the current star formation rate shown in table [ tab - oldsfh ] .
( interestingly , the 14 gyr calculated from the data presented here implies that the galaxy s current star formation rate is roughly equal to the average rate for its lifetime . ) two possible causes of this difference can be pinpointed .
first , our sample consists of only a small fraction of wlm ( one including a small hii region and bordering two large hii complexes ) , and the global star formation timescale could be significantly different from that found in this region .
second , it is possible that the estimate of the lifetime - averaged star formation rate presented here is more accurate , as this is the first star - formation history work for wlm using hst wfpc2 data .
however , without data of hst quality elsewhere in the galaxy , it is likely not possible to determine which ( or perhaps a combination of the two ) is the source of the difference .
overall , the results confirm the previous findings of minniti & zijlstra ( 1997 ) and ferraro et al .
. the wf2 field , which is mostly in the wlm bar , has seen the most star formation activity in the recent burst of activity , accounting for @xmath2285% of the total recent star formation .
( for comparison , the wf2 field contains just 62% of the stars with @xmath37 . ) the wf3 field , which is located on the edge between the bar and the halo , is intermediate , containing 12% of the recent star formation and 26% of the red stars . finally , the wf4 field , located entirely in the halo , appears to have almost no recent star formation ( 3% of the total ) despite containing 11% of the red stars .
hst f555w and f814w photometry of a portion of the wlm galaxy has been presented .
a distance modulus of 24.88 @xmath13 0.09 , is calculated using cmd analysis and shown to be consistent with a distance determination from the rgb tip and previous measurements by hodge et al .
( 1999 ) , minniti & zijlstra ( 1997 ) , sandage & carlson ( 1985 ) , and later cepheid recalibrations . the galaxy s oldest stars appear to be @xmath2212 gyr old , with an abundance [ fe / h ] of -2.18 @xmath13 0.28 for the initial star - forming episode .
it is calculated that roughly half of the galaxy s star formation occurred before 9 gyr ago .
star formation rates in the intermediate ages are more difficult to calculate , because of no main sequence stars and poor modeling of agb stars by the evolutionary tracks used .
the metallicity has risen to a current value of [ fe / h ] = -1.08 @xmath13 0.18 , consistent with the hodge & miller ( 1995 ) measurement of hii region abundances in wlm .
increased activity has taken place recently , for approximately the past 1 gyr .
the activity for stars older than this can not be pinpointed because of mixing within the galaxy , but the current star formation is limited to the bar of the galaxy .
this activity has continued to the present , with a measured formation timescale of 14 gyr given the current star formation rate .
ables , h. d. & ables p. g. , 1977 , apjs , 34 , 245 dolphin , a. e. 1997 , new astronomy 2 , 397 dolphin , a. e. 1999 , in prep fagotto , f. , bressan , a. , bertelli , g. , & chiosi , c. 1994 , a&as 105 , 29 ferraro , f. r. , pecci , f. f. , tosi , m. , & buonanno , r. 1989 , mnras 241 , 433 girardi , l. , bressan , a. , chiosi , c. , bertelli , g. , & nasi , e. 1996 , a&as 117 , 113 hodge , p. & miller , b. w. 1995 , apj 451 , 176 hodge , p. w. , dolphin , a. e. , smith , t. r. , & mateo , m. , 1999 , apj , in press lee , m. g. , freedman , w. l. , & madore , b. f. , 1993 , apj , 417 , 92 melotte , p. j. , 1926 , mnras , 86 , 636 minniti , d. & zijlstra , a. a. 1997 , aj 114 , 147 sandage , a. & carlson , g. 1985 , aj 90 , 1464 wolf , m. , 1909 , astron .
nachr . , 183 , 187 crr 0@xmath381 & 27 @xmath13 6 & -1.08 @xmath13 0.18 [email protected] & 37 @xmath13 8 & -1.13 @xmath13 0.16 2.5@xmath385 & 8 @xmath13 8 & -1.20 @xmath13 0.14 5@xmath387 & 13 @xmath13 15 & -1.25 @xmath13 0.13 7@xmath389 & 27 @xmath13 25 & -1.34 @xmath13 0.14 9@xmath3812 & 67 @xmath13 15 & -2.18 @xmath13 0.28 crrr [email protected] & 30 @xmath13 3 & 4 @xmath13 1 & 1 @xmath13 1 [email protected] & 10 @xmath13 1 & 1 @xmath13 1 & 2 @xmath13 1 [email protected] & 34 @xmath13 4 & 8 @xmath13 2 & 6 @xmath13 13 [email protected] & 8 @xmath13 7 & 3 @xmath13 3 & 23 @xmath13 36 [email protected] & 8 @xmath13 12 & 7 @xmath13 5 & 4 @xmath13 4 | hst f555w and f814w photometry of a portion of the wlm galaxy are presented .
the distance modulus is determined via fitting of the entire color - magnitude diagram to be @xmath0 , which is consistent with the rgb tip distance .
the galaxy s measurable star formation history appears to have begun no more than 12 gyr ago , with about half of wlm s total star formation ( by mass ) formed before 9 gyr ago .
the star formation rate gradually decreased , until a recent increase in activity starting between 1 and 2.5 gyr ago .
this is still continuing to the present time , and is concentrated in the bar of the galaxy , as shown by the difference in recent star formation rates in the three wf chips . |
the development of self - consistent many - body theories aiming at the description of low - energy nuclear phenomena provides the necessary theoretical tools for an exploration of the nuclear chart into known and unknown regions .
theoretical methods ( both relativistic and non - relativistic ) formulated within the framework of density functional theory ( dft ) and effective field theory ( eft ) are the most promising tools for the global investigation of the properties of atomic nuclei .
the dft and eft concepts in nuclear structure models have been extensively discussed in a number of recent articles @xcite .
the power of the models based on these concepts is essentially unchallenged in medium and heavy mass nuclei where ab - initio type few - body calculations are computationally impossible and the applicability of the spherical shell model is restricted to a few regions in the vicinity of doubly shell closures .
the self - consistent mean - field approach to nuclear structure represents an approximate implementation of kohn - sham density functional theory ( dft ) @xcite , which is successfully employed in the treatment of the quantum many - body problem in atomic , molecular and condensed matter physics .
the dft enables a description of the nuclear many - body problem in terms of energy density functionals ( edf ) , and self - consistent mean - field models approximate these functionals , which include all higher - order correlations , with powers and gradients of ground - state nucleon densities ( see refs .
@xcite and references therein ) .
edf functionals are universal in the sense that they can be applied to nuclei all over the periodic table .
although they model the effective interaction between nucleons , edf are not necessarily related to any nucleon - nucleon ( nn ) potential . by employing these energy functionals ,
adjusted to reproduce the empirical properties of symmetric and asymmetric nuclear matter , and bulk properties of some spherical nuclei , the current generation of self - consistent mean - field methods has achieved a high level of accuracy in the description of the ground states and the properties of excited states in arbitrarily heavy nuclei , exotic nuclei far from @xmath0-stability , and in nuclear systems at the nucleon drip - lines ( see refs .
@xcite and references therein ) .
the self - consistent methods ( such as hartree - fock ( hf ) or hartree - fock - bogoliubov ( hfb ) ) based on zero range skyrme forces or finite range gogny forces are frequently used in nuclear structure calculations @xcite .
these approaches represent non - relativistic energy density functionals based on the schrodinger equation for many - body nuclear problem @xcite . on the other hand
, one can formulate the class of relativistic models based on the dirac formalism , which can generally be defined as _ covariant density functionals _ ( cdf ) @xcite .
these models , such as quantum hadrodynamics ( qhd ) @xcite , are based on concepts of non - renormalizable effective relativistic field theories and dft , and they provide a very interesting relativistic framework for the studies of nuclear structure phenomena at and far from the valley of @xmath1stability @xcite . relativistic mean field ( rmf ) models @xcite are analogs of the kohn - sham formalism of dft @xcite , with local scalar and vector fields appearing in the role of local relativistic kohn - sham potentials @xcite .
the energy density functional is approximated with the powers and gradients of auxiliary meson fields or nucleon densities .
the eft building of the energy density functional allows error estimates to be made , provides a power counting scheme which separates long- and short - distance dynamics and , therefore , removes model dependences from the self - consistent mean field approach @xcite . in the description of nuclear ground states and the properties of excited states the self - consistent mean - field implementations of quantum hadrodynamics , the relativistic hartree - bogoliubov model ( rhb ) and the relativistic ( quasiparticle ) random phase approximation ( rqrpa ) and their subversions ,
are employed @xcite .
the _ mean field _ is a basic concept of every dft .
one can specify _ time - even _ and _ time - odd _ mean fields @xcite dependent on the response of these fields to the action of time - reversal operator .
the properties of time - even mean fields in nuclear density functionals are reasonably well understood and defined @xcite .
this is due to the facts that ( i ) many physical observables such as binding energies , radii etc .
are sensitive only to these fields , and ( ii ) the model parameters are fitted to such physical observables .
on the other hand , the properties of time - odd mean fields , which appear only in nuclear systems with broken time - reversal symmetry , are still poorly understood .
however , it is already known that these fields are important for proper description of rotating nuclei @xcite , band terminations @xcite , magnetic moments @xcite , isoscalar monopole vibrations @xcite , electric giant resonances @xcite , large amplitude collective dynamics @xcite , fussion process @xcite , the strengths and energies of gamow - teller resonances @xcite , the binding energies of odd - mass nuclei @xcite and the additivity of angular momentum alignments @xcite .
they also may play a role in the @xmath2 nuclei @xcite and affect the definition of the strength of pairing correlations @xcite .
there was a dedicated effort to better understand time - odd mean fields in the framework of the skyrme energy density functional ( edf ) theory ( see refs .
@xcite and references therein ) . on the contrary ,
much less attention has been paid to these fields in covariant density functional theory ( cdft ) @xcite .
this is due to the fact that time - odd mean fields are defined through the lorentz invariance in the cdft @xcite , and thus they do not require additional coupling constants . on the other hand , time - odd mean fields are not well defined in non - relativistic density functional theories @xcite and , as a consequence , there are a number of open questions related to these fields .
the current manuscript aims at better and systematic understanding of time - odd mean fields and their impact on physical observables in non - rotating nuclei in the framework of the rmf realization of the cdft .
the results of the study of these fields in rotating nuclei will be presented in a forthcoming article @xcite which represents a continuation of the current investigation .
the manuscript is organized as follows .
the cranked relativistic mean field theory and its details related to time - odd mean fields are discussed in sect .
[ theory ] .
section [ sect - odd ] is devoted to the analysis of the impact of time - odd mean fields on binding energies of odd - mass nuclei .
the mass and particle number dependences of this impact and their connections with odd - even mass staggerings are also considered .
the microscopic mechanism of additional binding in odd - mass nuclei induced by time - odd mean fields is analyzed in sect .
[ mech - bind - to ] .
the impact of time - odd mean fields on the properties of proton - unstable nuclei is studied in sect .
[ sect - proton ] . section [ sect - odd - odd ] considers how time - odd mean fields modify the properties of odd - odd nuclei .
finally , sect .
[ sect - concl ] contains the main conclusions of our work .
the results presented in the current manuscript have been obtained using cranked relativistic mean field ( crmf ) theory @xcite .
this theory has been successfully employed for the description of rotating nuclei ( see ref .
@xcite and references therein ) in which time - odd mean fields play an important role , but it is also able to describe the nuclear systems with broken time - reversal symmetry in intrinsic frame at no rotation . in this theory
the pairing correlations are neglected which allows to better isolate the effects induced by time - odd mean fields .
the crmf computer code is formulated in the signature basis . as a result ,
the breaking of kramer s degeneracy of the single - particle states is taken into account in a fully self - consistent way .
this is important for an accurate description of time - odd mean fields in fermionic channel ( see sect . [ mech - bind - to ] ) .
the most important features of the crmf formalism related to time - odd mean fields are outlined below ( for more details see refs .
@xcite ) for the case of no rotation ( rotational frequency @xmath3 . in the hartree approximation , the stationary dirac equation for the nucleons in the intrinsic frame
is given by @xmath4 where @xmath5 is the dirac hamiltonian for the nucleon with mass @xmath6 @xmath7 it contains the average fields determined by the mesons , i.e. the attractive scalar field @xmath8 @xmath9 and the repulsive time - like component of the vector field @xmath10 @xmath11 a magnetic potential @xmath12 @xmath13 originates from the space - like components of the vector mesons .
note that in these equations , the four - vector components of the vector fields @xmath14 , @xmath15 , and @xmath16 are separated into the time - like ( @xmath17 , @xmath18 and @xmath19 ) and space - like [ @xmath20 , @xmath21 , and @xmath22 components . in the dirac equation
the magnetic potential has the structure of a magnetic field
. therefore the effect produced by it is called _ nuclear magnetism _ ( nm ) @xcite .
the corresponding meson fields and the electromagnetic potential are determined by the klein - gordon equations @xmath23 \nonumber \\ & & -g_{2}\sigma^{2}({{{\mbox{\boldmath $ r$}}}})-g_{3}\sigma^{3}({{{\mbox{\boldmath $ r$ } } } } ) , \label{kgsigma } \\
\left\ { -\delta + m_{\omega}^{2}\right\ } \omega_{0}({{{\mbox{\boldmath $ r$ } } } } ) & = & g_{\omega } [ \rho_v^n({{{\mbox{\boldmath $ r$ } } } } ) + \rho_v^p({{{\mbox{\boldmath $ r$ } } } } ) ] , \label{kgomega0 } \\
\left\ { -\delta + m_{\omega } ^{2}\right\ } ~{{{\mbox{\boldmath $ \omega$}}}}({{{\mbox{\boldmath $ r$ } } } } ) & = & g_{\omega } [ { { { \mbox{\boldmath $ j$}}}}^n({{{\mbox{\boldmath $ r$ } } } } ) + { { { \mbox{\boldmath $ j$}}}}^p({{{\mbox{\boldmath $ r$ } } } } ) ] \label{kgomegav } \\ \left\ { -\delta + m_{\rho}^{2}\right\ } \rho_{0}({{{\mbox{\boldmath $ r$ } } } } ) & = & g_{\rho } [ \rho_v^n({{{\mbox{\boldmath $ r$ } } } } ) - \rho_v^p({{{\mbox{\boldmath $ r$ } } } } ) ] , \label{kgrho0 }
\\ \left\ { -\delta + m_{\rho}^{2}\right\ } ~{{{\mbox{\boldmath $ \rho$}}}}({{{\mbox{\boldmath $ r$ } } } } ) & = & g_{\rho } [ { { { \mbox{\boldmath $ j$}}}}^n({{{\mbox{\boldmath $ r$ } } } } ) - { { { \mbox{\boldmath $ j$}}}}^p({{{\mbox{\boldmath $ r$ } } } } ) ] , \label{kgrhov } \\
-\delta a_0({{{\mbox{\boldmath $ r$}}}})=e\rho^p_v({{{\mbox{\boldmath $ r$ } } } } ) , & & -\delta { { { \mbox{\boldmath $ a$}}}}({{{\mbox{\boldmath $ r$}}}})=e{{{\mbox{\boldmath $ j$}}}}^p({{{\mbox{\boldmath $ r$}}}}),\end{aligned}\ ] ] with source terms involving the various nucleonic densities and currents @xmath24 where the labels @xmath25 and @xmath26 are used for neutrons and protons , respectively . in the equations above
, the sums run over the occupied positive - energy shell model states only ( _ no - sea approximation _ ) @xcite .
note that the spatial components of the vector potential @xmath27 are neglected in the calculations since the coupling constant of the electromagnetic interaction is small compared with the coupling constants of the meson fields .
and @xmath28 parameters of different modern parametrizations of the rmf lagrangian .
they are combined into four groups dependent on how self- and mixed - couplings are introduced .
group a represents the parametrizations which include non - linear self - couplings only for the @xmath29-meson .
group b contains the parametrizations which include self - couplings for the @xmath29- and @xmath30-mesons ( and @xmath31mesons in the case of pk1r ) .
group c represents the parametrizations which include density - dependent meson - nucleon couplings for the @xmath29- , @xmath30- , and @xmath32-mesons .
the other parametrizations are included into the group d. the svi and svi-2 parametrizations neglect non - linear self - couplings for the @xmath29-meson but include isoscalar - isovector couplings .
the other parametrizations of the group d include mixed interaction terms ( such as isoscalar - isovector couplings @xcite ) in addition to non - linear self - couplings for the @xmath29-meson .
the parameters are taken from refs .
@xcite ( nl1 ) , @xcite ( nl3 ) , @xcite ( nl3 * ) , @xcite ( nlz ) , @xcite ( nlz2 ) , @xcite ( nlsh ) , @xcite ( nlra-1 ) , @xcite ( pk1,pk1r ) , @xcite ( tma ) , @xcite ( tm1 ) , @xcite ( svi1,svi-2 ) , @xcite ( fsugold ) , @xcite ( fsugz00,fsugz03,fsugz06 ) , @xcite ( tw99 ) , @xcite ( ddme1 ) , @xcite ( ddme2 ) , @xcite ( pkdd ) .
note that we omitted mass - dependent terms for @xmath28 in the tma parametrization which is a good approximation for heavy nuclei since @xmath33 @xcite.,width=302 ] the magnetic potential @xmath12 in the dirac equation as well as the currents @xmath34 in the klein - gordon equations do not appear in the rmf equations for time - reversal systems @xcite .
similar to the nonrelativistic case , their presence leads to the appearance of time - odd mean fields .
thus , we will use the terms _ nuclear magnetism _ and _ time - odd mean fields _ interchangeably throughout this manuscript .
the magnetic potential is the contribution to the mean field that breaks time - reversal symmetry in the intrinsic frame and induces non - vanishing currents @xmath35 ( eq . ( [ current ] ) ) in the klein - gordon equations ( eqs .
( [ kgomegav ] ) , ( [ kgrhov ] ) ) , which are related to the space - like components of the vector mesons . in turn
, the space - like components of the vector @xmath36 and @xmath37 fields form the magnetic potential ( [ magnetic ] ) in the dirac equation .
note that the current @xmath38 change the sign upon the action of time - reversal operator @xcite .
together with densities it forms covariant four - vector @xmath39 . as a consequence ,
these two quantities ( @xmath32 and @xmath40 ) do not transform independently under lorentz transformation .
this explains why the structure of the klein - gordon equations for time - like and space - like components of vector mesons is the same ( compare , for example , eqs .
( [ kgomega0 ] ) and ( [ kgomegav ] ) for @xmath30-meson ) and why the same coupling constant stands in front of the densities and currents on the right hand side of these equations .
the spatial components of the vector @xmath30 and @xmath32 mesons lead to the interactions between possible currents . for the @xmath30-meson
this interaction is attractive for all combinations ( @xmath41 , @xmath42 and @xmath43-currents ) , and for the @xmath32-meson it is attractive for @xmath41 and @xmath42-currents but repulsive for @xmath43-currents . within mean field theory
such currents occur only in the situations of broken time - reversal symmetry .
note that time - odd mean fields related to nm are defined through the lorentz invariance @xcite and thus they do not require additional coupling constants : the coupling constants of time - even mean fields are used also for time - odd mean fields .
the currents are isoscalar and isovector in nature for the @xmath30 and @xmath32 mesons ( eqs .
( [ kgomegav ] , [ kgrhov ] ) ) , respectively . as a consequence ,
the contribution of the @xmath32-meson to magnetic potential and total energy is marginal in the majority of the cases even at the neutron - drip line ( see sect . [ pol - odd ] for details ) .
thus , time - odd mean fields in the rmf framework depend predominantly on the spatial components of the @xmath30 meson . neglecting the contribution of the @xmath32 meson
, one can see that only two parameters , namely , the mass @xmath44 and coupling constant @xmath28 of the @xmath30 meson define the properties of time - odd mean fields ( eqs .
( [ magnetic ] ) , ( [ kgomegav ] ) , and ( [ kgrhov ] ) ) . fig .
[ paramet ] clearly indicates that these parameters are well localized in the parameter space for the parametrizations of the rmf lagrangian in the groups a , b , and c ( see caption of fig .
[ paramet ] for group definitions ) .
this suggests that the parameter dependence of the impact of time - odd mean fields on the physical observables should be quite weak for these types of the parametrizations .
indeed , the analysis of terminating states in ref .
@xcite showed that time - odd mean fields are defined with an accuracy around 15% for the parametrizations of the rmf lagrangian containing only non - linear self - couplings for the @xmath29-meson ( group a ) . on the other hand , fig .
[ paramet ] suggests that the time - odd mean fields maybe less accurately defined in the parametrizations of group d which include mixed interaction terms such as isoscalar - isovector couplings .
however , it is premature to make such conclusion because these parametrizations have not been tested extensively even on nuclear structure data sensitive to time - even mean fields .
this is contrary to the parametrizations of groups a , b , and c which have passed successfully such test .
the investigation of all these parametrizations is definitely beyond the scope of this study .
thus , the present investigation has been focused on the study of time - odd mean fields in the cdft with the parametrizations of the rmf lagrangian including only non - linear self - couplings of the @xmath29-meson ( the group a of the parametrizations ) .
the results of the study of time - odd mean fields in the groups b , c , and d of the parametrizations of meson - coupling models as well as within the point - coupling models will be presented in a forthcoming manuscript @xcite .
the total energy of the system is given in refs .
@xcite . in order to facilitate the discussion we split it into different terms as @xmath45 where @xmath46 and @xmath47 represent the contributions from fermionic degrees of freedom , while the other terms are related to mesonic ( bosonic ) degrees of freedom . in eq .
( [ etot ] ) @xmath48 is the energy of the particles moving in the field created by the mesons ( @xmath49 is the energy of @xmath50-th particle and the sum runs over all occupied proton and neutron states ) @xmath51 , \end{aligned}\ ] ] is the linear contribution to the energy of isoscalar - scalar @xmath29-field @xmath52 , \end{aligned}\ ] ] is the non - linear contribution to the energy of isoscalar - scalar @xmath29-field @xmath53 , \end{aligned}\ ] ] is the energy of the time - like component of isoscalar - vector @xmath30-field @xmath54 , \end{aligned}\ ] ] is the energy of the time - like component of isovector - vector @xmath32-field @xmath55 , \label{e - s - omega}\end{aligned}\ ] ] is the energy of the space - like component of isoscalar - vector @xmath30-field @xmath56 , \label{rho - space}\end{aligned}\ ] ] is the energy of the space - like component of isovector - vector @xmath32-field @xmath57 is the coulomb energy @xmath58 is the correction for the spurious center - of - mass motion approximated by its value in a non - relativistic harmonic oscillator potential .
the total energy of the system can alternatively be written as ( similar to refs .
@xcite ) @xmath59 where the kinetic energy @xmath60 is given by @xmath61 and the interaction energy between the nucleons @xmath62 by @xmath63 however , this representation of total energy has a disadvantage as compared with eq .
( [ etot ] ) since it does not provide a direct access to the particle energy @xmath46 .
the latter plays an important role in the understanding of the breaking of kramer s degeneracy of time - reversal orbitals in the presence of time - odd mean fields ( see sect .
[ e - split ] for details ) .
thus , further discussion of total energy will be based mostly on eq .
( [ etot ] ) .
however , we will also provide the results of calculations for kinetic energy @xmath60 .
the crmf equations are solved in the basis of an anisotropic three - dimensional harmonic oscillator in cartesian coordinates characterized by the deformation parameters @xmath64 ( @xmath65 in the case of superdeformed states ) and @xmath66 as well as the oscillator frequency @xmath67 mev .
the truncation of basis is performed in such a way that all states belonging to the shells up to fermionic @xmath68=12 and bosonic @xmath69=16 are taken into account in the calculations of light and medium - mass nuclei .
the fermionic basis is increased up to @xmath68=14 in the calculations of actinides .
numerical analysis indicates that this truncation scheme provides sufficient numerical accuracy for the physical quantities of interest .
single - particle orbitals are labeled by @xmath70\omega^{sign}$ ] .
@xmath70\omega$ ] are the asymptotic quantum numbers ( nilsson quantum numbers ) of the dominant component of the wave function .
the superscripts _ sign _ to the orbital labels are used sometimes to indicate the sign of the signature @xmath71 for that orbital @xmath72 .
the majority of the calculations are performed with the nl3 parametrization @xcite of the rmf lagrangian .
many - particle configurations ( further nuclear configurations or configurations ) are specified by the occupation of available single - particle orbitals . in the calculations without pairing , the occupation numbers @xmath25 are integer ( @xmath73 or 1 ) . in odd nuclei , all single - particle states with exception of one are pairwise occupied .
we will call this occupied single - particle state of fixed signature for which its time - reversal ( signature ) counterpart state is empty as _ blocked state _ in order to simplify the discussion . the total signature and the parity of the configuration
are the same as the ones of the blocked state . in the crmf code , it is possible to specify the occupation of either @xmath74 or @xmath75 signature of the single - particle state .
the specification of nuclear configuration by means of listing all occupied single - particle states is unpractical .
thus , we label the nuclear configuration in odd mass nuclei by the nilsson label and the signature of the blocked state .
note that many physical observables , such as additional binding due to nm , do not depend on the signature of the blocked state in odd - mass nuclei . in these cases
, we will omit the signature from the configuration label . in odd - odd nuclei ,
the nilsson labels of the blocked proton and neutron states and their signatures are used for configuration labelling . note that the labelling by means of nilsson labels is performed only when the calculated shape of nuclear configuration is prolate or near - prolate . in order to investigate the impact of nm ( time - odd mean fields ) on physical observables ,
the crmf calculations are performed in three calculational schemes for the fixed configurations : * fully self - consistent calculations with nm included ( further denoted as nm calculations ) which take into account space - like components of the vector mesons ( eqs .
( [ kgomegav ] ) , ( [ kgrhov ] ) and ( [ magnetic ] ) ) , currents ( eqs .
( [ kgomegav ] ) , ( [ kgrhov ] ) , and ( [ current ] ) ) , and magnetic potential v(r ) ( eq .
( [ magnetic ] ) ) ; * fully self - consistent calculations without nm ( further denoted as wnm calculations ) which omit space - like components of the vector mesons ( eqs .
( [ kgomegav ] ) , ( [ kgrhov ] ) and ( [ magnetic ] ) ) , currents ( eqs .
( [ kgomegav ] ) , ( [ kgrhov ] ) , and ( [ current ] ) ) , and magnetic potential v(r ) ( eq .
( [ magnetic ] ) ) .
note that the results of the nm and wnm calculations are always compared for the same nuclear configuration ; * perturbative calculations ( the physical quantities of interest are indicated by superscript _
fully self - consistent calculations with nm provide a starting point . using their fields as input fields ,
only one iteration is performed in the calculations without nm : this provides perturbative results .
time - even mean fields are the same in both ( fully self - consistent and perturbative ) calculations .
then , the impact of time - odd mean fields on calculated quantities ( for example , different terms in the total energy ( see eq .
[ etot ] ) ) is defined as the difference between the values of this quantity obtained in these two calculations . in this way
, the pure effects of time - odd mean fields in fermionic and mesonic channels of the model are isolated because no polarization effects are introduced into time - even mean fields .
these are the ways in which the effects of time - odd mean fields can be studied , and as such they are frequently used in the dft studies , both in relativistic and non - relativistic frameworks @xcite .
one should , however , keep in mind that if time - odd fields are neglected , the local lorentz invariance ( galilean invariance in non - relativistic framework @xcite ) is violated .
the inclusion of time - odd mean fields restores the lorentz invariance .
the time - reversal invariance is conserved in the ground states of even - even nuclei .
the nucleon states are then pairwise degenerated , and the contribution of the state to the currents cancels with the contribution of its time - reversed partner .
time - odd mean fields reveal themselves in odd- and odd - odd mass nuclei and in two-(multi-)quasiparticle states of even - even nuclei .
this is because an unpaired ( odd ) nucleon breaks the time - reversal invariance in intrinsic frame and produces the contribution to the currents and spin . in this case , the kramer s degeneracy of time - reversal partner orbitals is also broken .
the modifications of the binding energies and quasiparticle spectra are the most important issues when considering time - odd mean fields in non - rotating systems .
the binding energies are important in nuclear astrophysics applications @xcite , and their modifications due to time - odd mean fields may have considerable consequences for the @xmath71- and @xmath76-process abundances .
thus , it is important to understand the influence of time - odd mean fields on binding energies of odd- and odd - odd mass nuclei , especially in the context of mass table fits @xcite . with the current focus on the spectroscopic quality dft @xcite , the knowledge on how time - odd mean fields influence the relative energies of different ( quasi)particle states in model calculations
is also needed .
but for the results obtained with the indicated parametrizations of the rmf lagrangian in the ar isotopes .
the structure of the ground states is shown by the nilsson labels .
only when the structure of the ground state is the same as in the nl3 parametrization , the results obtained with other parametrizations are shown.,width=302 ] while there was a considerable interest in the study of time - odd mean fields in odd- and odd - odd mass nuclei at no rotation within the skyrme edf @xcite , relatively little is known about their role in the framework of the cdft .
so far , the impact of time - odd mean fields on binding energies has been studied in the cdft framework only in odd - mass nuclei around doubly magic spherical nuclei in ref .
@xcite , and in few deformed nuclei around @xmath77s @xcite and @xmath78no @xcite .
the impact of nm on the binding energies of light odd - mass nuclei is shown in fig .
[ light - a ] .
one can see that in all cases the presence of nm leads to additional binding the magnitude of which is nucleus and state dependent .
the absolute value of this additional binding is typically below 200 kev and only reaches 300 kev in some lower mass nuclei . on the average , the magnitude of additional binding due to nm is inversely correlated with the mass of the nucleus ; it is the largest in the lightest nuclei and the smallest in the heaviest nuclei . for each isotope chain , it is the largest in the vicinity of the proton - drip line and the smallest in the vicinity of the neutron - drip line .
the polarization effects induced by nm and the energy splitting between blocked state and its unoccupied signature partner induced by nm decrease with the increase of mass ( compare tables [ nm - odd ] and [ nm - odd-3 ] below ) .
this explains the observed trends in additional binding due to nm .
[ light - par ] shows that additional binding due to nm only weakly depends on the rmf parametrization ; this is also seen in the analysis of terminating states in ref .
@xcite . in both cases , the largest deviation from the nl3 results
is observed in the case of the nlsh parametrization .
it is interesting to compare these results with the ones obtained in the skyrme edf ( see fig . 4 in ref .
the modifications of total binding energy due to time - odd mean fields are given by the @xmath79 quantity in ref .
@xcite , which is an analog of the @xmath80 quantity .
the general dependence of both quantities on @xmath81 is similar in odd - mass nuclei apart from a few cases such as @xmath82ti and @xmath82sc in sly4 skyrme edf ( fig . 4 in ref .
neither rmf nor skyrme edf calculations in odd - mass nuclei indicate the enhancement of time - odd mean fields in the vicinity of the @xmath2 line .
this is contrary to ref .
@xcite which suggested that the effects of time - odd mean fields are enhanced at the @xmath2 line .
the absolute values of @xmath79 and @xmath80 quantities are similar being below 300 kev in the majority of the cases .
the principal difference between the rmf and the skyrme edf lies in the fact that time - odd mean fields are always attractive and show very small dependence on the parametrization in the rmf calculations ( this is also supported by the analysis of terminating states , see ref .
@xcite ) , while they can be both attractive ( sly4 force ) or repulsive ( siii force ) and show considerable dependence on the parametrization in skyrme edf ( ref .
@xcite ) . )
the upper panel shows additional binding @xmath80 due to nm and its configuration dependence .
the configurations are labelled by the nilsson labels of the blocked states ; the configurations at and to the right of the nilsson label up to the next nilsson label have the same blocked state .
the bottom panel shows the corresponding deformations of the configurations .
the calculations have been performed with the nl3 parametrization of the rmf lagrangian .
they cover the nuclei from the proton - drip line up to the neutron - drip line.,width=302 ] but for negative parity configurations.,width=302 ] the role of time - odd mean fields is studied here in medium mass ce isotopes in order to facilitate the comparison with the results obtained within the skyrme edf with the sly4 force in ref .
this reference represents the most detailed study of time - odd mean fields in odd - mass nuclei within the skyrme edf .
we consider the lowest configurations of positive and negative parities , while ref .
@xcite studies only the lowest configurations in each nucleus
. figs .
[ ce - pos ] and [ ce - neg ] show the additional binding due to nm .
the comparison with the skyrme edf results of ref .
@xcite reveals a number of important differences .
first , similar to the results in light nuclei ( sect .
[ bind - light ] ) and in actinide region ( sect .
vi h in ref .
@xcite ) , time - odd mean fields are attractive in the rmf calculations for the ce isotopes . on the contrary , they are repulsive in the sly4 parametrization of the skyrme edf @xcite .
note that the sly4 force produces attractive time - odd mean fields in light nuclei ( ref .
this mass dependence of the effects of time - odd mean fields in the skyrme edf may be due to the competition between isovector and isoscalar effects @xcite .
the average absolute magnitude of the change of binding due to time - odd mean fields in the rmf calculations is only half of the one seen in the skyrme calculations with the sly4 parametrization .
it was also checked on some examples that additional binding due to nm only weakly depends on the parametrization of the rmf lagrangian .
second , the results of the calculations do not reveal a strong dependence of additional binding due to nm on deformation .
for example , the deformation of the @xmath8411/2 $ ] configuration in the @xmath85ce chain changes drastically from @xmath86 down to @xmath87 ( fig .
[ ce - pos ] , bottom panel ) , but the additional binding due to nm remains almost the same ( fig .
[ ce - pos ] , top panel ) .
the @xmath887/2 $ ] and @xmath8911/2 $ ] configurations are another examples of this feature ( fig .
[ ce - neg ] ) .
third , the binding energy modifications due time - odd mean fields are completely different in the rmf and skyrme edf calculations . in the skyrme edf calculations ,
the magnitude of these binding energy modifications is related with three properties of the blocked orbital . in
decreasing order of importance they are @xcite : a small @xmath90 quantum number , a down - sloping behavior of the energy of the single - particle state with mass number @xmath91 , and a large total angular momentum @xmath92 for the spherical shell from which the single - particle state originates .
for example , the binding energy modifications due to time - odd mean fields will be larger for the configuration based on blocked single - particle state with small @xmath90 than for the configuration with large @xmath90 of the blocked state if both blocked states belong to the same @xmath92-shell . on the contrary , the rmf calculations do not reveal this type of correlations between additional binding due to nm and the structure of the blocked state .
indeed , the configurations which have the largest changes in binding energies due to nm ( @xmath93 mev ) are [ 413]5/2 , [ 404]7/2 , [ 640]1/2 , [ 631]3/2 , [ 505]9/2 and [ 501]1/2 . when discussing current distributions , it is important to remember that the calculations are performed in one - dimensional cranking approximation .
although the rotational frequency is equal to zero in the calculations , the results for the currents still obey the symmetries imposed by the cranking approximation .
this is clearly seen when considering the signature quantum number in the limit of vanishing rotational frequency @xmath94 ( see ref .
@xcite ) . in this case
definite relations exist between the states @xmath95 of good signature @xmath96 ( @xmath97 denotes the set of additional quantum numbers ) and the single - particle states employed usually in the low spin limit .
for the latter states in axially symmetric nuclei , we obtain doubly degenerate single - particle states @xmath98 and @xmath99 , where @xmath100 denotes the projection of angular momentum on the symmetry axis . here , @xmath98 is an eigenstate with definite angular momentum projection @xmath100 , while @xmath99 denotes the time - reversed state ( with angular momentum projection @xmath101 ) . in the limit of vanishing rotational frequency @xmath102 , the states @xmath95 with definite signature
@xmath96 become linear combinations of the states @xmath98 and @xmath99 @xmath103 these relations may be considered as a transformation between two representations of the single - particle states : the one with good projection @xmath100 ( the @xmath98 representation ) and the other with good signature @xmath71 ( cranking representation ) . in the @xmath98-representation the alignment of angular momentum vector of a particle
is specified along the axis of symmetry . as a result ,
the axial symmetry is conserved and only azimuthal currents with respect of the symmetry axis are present . in the cranking formalism ( which allows also triaxial shapes )
, the alignment of angular momentum vector of a particle is specified along the @xmath104axis perpendicular to the axis of symmetry . as a result ,
the currents follow the symmetries of cranking approximation and have the distributions discussed below .
total neutron current distributions in the configurations of selected nuclei having similar quadrupole deformations are shown in figs .
[ curr - distr ] and [ curr - distr-2 ] .
they are predominantly defined by the currents generated by the blocked orbitals .
this is clearly visible from the comparison of figs .
[ curr - distr ] and [ curr - distr - single ] : the latter figure shows the currents produced by single neutron in different nilsson states of the @xmath1051/2 $ ] configuration in @xmath106ce .
neutron currents are characterized by the complicated patterns in different cross - sections of the nucleus .
[ curr - distr - single ] clearly shows that these patterns are defined by the density distributions of the blocked states . moreover , there are clear correlations between the patterns of the currents in the @xmath107 plane ( @xmath108 is the symmetry axis and @xmath109 is the rotation axis in the crmf theory ) and the @xmath90 value of the nillson label of the blocked orbital ( figs .
[ curr - distr ] and [ curr - distr - single ] ) . at @xmath110 ,
the single - particle densities are concentrated in the vicinity of the axis of symmetry , and , as a consequence , the currents show circulations ( vortices ) which are concentrated in the central region of the nucleus . however , with increasing @xmath90 , the densities ( and , as a consequence , the currents ) are pushed away from the axis of symmetry of the nucleus towards the surface area .
in addition , the strength of the currents correlates with @xmath90 .
as follows from the values of factor @xmath111 the strongest currents appear for the @xmath110 states .
these orbitals are aligned with the axis of rotation ( @xmath109-axis ) already at no rotation . as a result
, the single - particle angular momentum vector of the @xmath110 orbitals performs the precession around the @xmath109-axis , thus orienting the currents predominantly in the @xmath107 plane .
this extra mechanism is not active in other configurations .
the strength of the currents decreases with the increase of @xmath90 .
for example , the currents generated by the blocked @xmath112 orbitals are weaker by a factor of almost 200 than the currents generated by the blocked @xmath110 orbitals ( compare scaling factors f in fig .
[ curr - distr ] for the blocked states with different @xmath90 values ) . in the @xmath113 plane ,
the majority of the configurations show the current pattern ( although with different strength of the currents and their localization in space ) visible on fig .
[ curr - distr-2]b , while the typical pattern of the currents in the @xmath114 plane is shown in fig . [ curr - distr-2]a . figs .
[ curr - distr]d , e show that the change of the signature of the blocked orbital leads only to a change in the direction of the currents .
neutron and proton number dependences of additional binding due to nm ( the @xmath115 quantity ) are presented in fig .
[ part - dep ] .
these figures are based on the results obtained in sects .
[ bind - light ] and [ ce - isotopes ] and on some extra calculations .
these extra calculations include odd-@xmath116 nuclei with @xmath117 , odd-@xmath118 nuclei with @xmath119 , and odd-@xmath116 nuclei with @xmath120 and cover these isotope and isotone chains from proton- to neutron - drip lines .
the calculations of nuclei around @xmath121cf were also performed in order to check the impact of pairing on the @xmath122 quantity .
for the same blocked states , the @xmath122 values obtained in the calculations without pairing in the present manuscript were compared with the ones obtained in the relativistic hartree - bogoliubov calculations of ref .
@xcite ( see table iv of ref .
although the pairing decreases additional binding due to nm in the most of the cases , there are still one-(quasi)particle configurations in which the @xmath115 quantity is smaller in the calculations without pairing .
this is a consequence of the complicated nature of the @xmath80 quantity defined by ( i ) the interplay of time - odd mean fields and the polarization effects ( sect .
[ mech - bind - to ] ) and by ( ii ) the differences in the impact of pairing on different terms of total energy .
the calculated @xmath115 quantities were fitted by simple parametrization @xmath123 where @xmath124 is equal either to proton @xmath116 or neutron @xmath118 numbers .
note that the @xmath115 values from odd - proton ( odd - neutron ) nuclei were used in the fit of @xmath125dependence of @xmath126 .
the results of the fits are shown by solid lines in fig .
[ part - dep ] .
one can see that the powers @xmath127 are similar for different fits ( proton or neutron ) . on the other hand ,
the magnitudes @xmath128 differ considerably between proton and neutron quantities , indicating weaker additional binding due to nm for odd - proton nuclei .
this result is consistent with the analysis obtained in sect .
[ sect - odd - odd ] .
the three - point indicator @xcite @xmath129 \label{neut - oes}\end{aligned}\ ] ] is frequently used to quantify the odd - even staggering ( oes ) of binding energies . here
@xmath130 is the number parity and @xmath131 is the ( negative ) binding energy of a system with @xmath118 particles . in eq .
( [ neut - oes ] ) , the number of protons @xmath116 is fixed , and @xmath118 denotes the number of neutrons , i.e. this indicator gives the neutron oes .
the factor depending on the number parity @xmath132 is chosen so that the oes centered on even and odd neutron number @xmath118 will both be positive .
an analogous proton oes indicator @xmath133 is obtained by fixing the neutron number @xmath118 and replacing @xmath118 by @xmath116 in eq .
( [ neut - oes ] ) .
the @xmath134 ( and similarly @xmath133 ) quantity will be modified in the presence of time - odd mean fields as @xmath135 where the subscripts to and wto indicate the values obtained in the calculations with and without time - odd mean fields , and @xmath136 is the contribution coming from time - odd mean fields .
if the @xmath134 quantity is centered at odd-@xmath118 nucleus , the @xmath136 quantity represents the change of binging energy of this odd - mass nucleus induced by time - odd mean fields .
this is because the time - odd mean fields have no effect on the binding energies of the ground states of even - even nuclei .
note that with such selection @xmath136 is negative if the time - odd mean fields provide additional binding in odd - mass nucleus . in the cdf theory ,
the @xmath136 quantity is equal to @xmath80 and thus it is always negative : this result does not depend on the rmf parametrization ( see sect .
[ bind - light ] for the dependence of the @xmath80 quantity on the rmf parametrization ) .
in addition , the magnitude of the @xmath136 quantity depends only weakly on the rmf parametrization . on the contrary , the sign and the magnitude of @xmath136 depends strongly on the parametrization in the skyrme edf calculations .
for example , in the calculations with the sly4 force the @xmath136 quantity is positive for medium mass nuclei ( refs .
@xcite ) but negative in light nuclei ( ref . @xcite ) . on the other hand
, the @xmath136 quantity will be positive in light nuclei in the calculations with the siii parametrizations @xcite .
it is interesting to compare the averaged effects of time - odd mean fields as given by the @xmath126 quantity with the experimental global trends for oes as shown by dashed lines in fig . 2 in ref .
the latter trends were obtained using phenomenological parametrization with the same functional dependence as in eq .
( [ delta ] ) with @xmath137 mev ( 4.31 mev ) and @xmath138 for neutron ( proton ) data sets .
the comparison of theory and experiment suggests that time - odd mean field contributions into oes can be as large as 10% in light systems and around 5 - 6% in heavy systems .
these are non - negligible contributions which have to be taken into account when the strength of pairing interaction is defined from the fits to experimental oes . the analysis of the sn isotopes in ref .
@xcite showed that time - even and time - odd polarization effects induced by odd nucleon produce oes reduced by about 30% as compared to the ones obtained in standard spherical calculations .
as a consequence , an enhancement of pairing strength by about 20% is required to compensate for that effect .
our calculations show much smaller reduction of oes in part because the polarization effects in time - even channel are already taken into account in the calculations without nm .
thus , the current calculations suggest that a much smaller increase of the strength of pairing ( by approximately 5% ) would be required to compensate for the reduction of oes due to time - odd mean fields .
in the current section , a detailed analysis of the impact of nm on the energies of the single - particle states and on different terms in the total energy expression ( eq .
( [ etot ] ) ) is performed in order to better understand the microscopic mechanism of additional binding due to nm .
we use the @xmath1395/2 $ ] configuration of @xmath140ce as an example in this analysis .
[ esp - splitting ] shows that the presence of time - odd mean fields leads to the energy splitting @xmath141 of the single - particle states which are time - reversal counterparts .
this corresponds to the removal of the kramer s degeneracy of these states .
one of these states moves up by @xmath142 as compared with its position in the absence of nm , while another moves down by @xmath142 .
detailed analysis of the single - particle spectra in @xmath140ce and @xmath143xe reveals general features which are also found in other nuclei .
the @xmath140ce nucleus is axially symmetric ( @xmath66 ) while @xmath143xe is triaxial with @xmath144 .
this difference in the symmetry of nucleus results in important consequences : the energy splittings appear in all single - particle states in triaxial nuclei , while only the states with @xmath145 ( the subscript _
bl _ indicates the blocked state ) experience such splittings in axially symmetric nuclei .
the former feature is due to the fact that @xmath90 is not a good quantum number in triaxial nuclei and each single - particle state represents a mixture of the basic states with different values of @xmath90 .
it is important to mention that the occupied and unoccupied states as well as the proton and neutron states show energy splittings ( fig .
[ esp - splitting ] ) .
the splittings of the proton and neutron states of the same structure are similar .
this is because the largest contribution to magnetic potential ( eq . ( [ magnetic ] ) ) is due to space - like components of the @xmath30-meson fields which do not depend on the isospin .
in addition , the occupied state is always more bound than its unoccupied time - reversal counterpart . between different signatures of the single - particle states in the presence of nm .
the results of the calculations are shown for the configurations of @xmath140ce in which either the @xmath1395/2 ^ -$ ] ( column ( 1 ) ) or @xmath1395/2^+$ ] ( column ( 3 ) ) states are blocked .
these signatures are degenerated in energy in the calculations without nm ( column ( 2 ) ) .
note that the single - particle states of interest are shown at arbitrary absolute energy in the column ( 2 ) .
solid ( open ) circles indicate occupied ( unoccupied ) states .
solid ( dotted ) lines are used for the @xmath74 ( @xmath75 ) states .
, width=264 ] the change of the signature of the blocked state leads to the inversion of the signatures in all pairs of time - reversal orbitals ( compare columns ( 1 ) and ( 3 ) in fig . [ esp - splitting ] ) .
the explanation for this process is the following .
the change of the signature of the blocked state results in the change of the direction of the currents to the opposite one ( compare fig .
[ curr - distr]d with fig . [ curr - distr]e ) .
this leads to the change of the direction of the vector potential @xmath146 in the dirac equation to the opposite one , which in turn causes the inversion of the signatures in all pairs of time - reversal orbitals .
however , the additional binding due to nm ( the @xmath80 quantity ) does not depend on the signature of the blocked state in odd - mass nuclei .
the polarization effects induced by nm are investigated by considering its impact on different terms of the total energy ( eq . ( [ etot ] ) ) .
the results of this study are shown in table [ nm - odd ] .
one can see that the total energy terms can be split into two groups dependent on how they are affected by nm .
the first group includes the @xmath147 , @xmath148 , @xmath149 and @xmath150 terms which are only weekly influenced by nm , and thus , they will not be discussed in detail .
.the impact of nm on different terms of the total energy ( eq . ( [ etot ] ) ) in the @xmath1515/2^+$ ] configuration of @xmath140ce .
second column shows the absolute energies [ in mev ] of different energy terms in the case when nm is neglected .
third and forth columns show the changes @xmath152 [ in mev ] in the energies of these terms induced by nm in self - consistent ( column 3 ) and perturbative ( column 4 ) calculations .
note that we show only nonzero quantities in column 4 .
the relevant quantities are also shown for kinetic energy @xmath60 ( eq . ( [ ekin - f ] ) ) in the last line . [ cols="^,^,^,^,^",options="header " , ] the situation is more complicated in odd - odd nuclei ( @xmath153p and @xmath154cl ) in which considerable energy splitting between the @xmath155 and @xmath156 configurations is obtained in the calculations .
the microscopic mechanism of binding modifications is illustrated in table [ table - cl34 ] on the example of configurations a and b in @xmath154cl .
nm provides additional binding of around 0.4 mev in the configuration a which has signature @xmath156 . in this configuration , proton and neutron currents due to the occupation of proton and neutron 5/2[202]@xmath157 states are in the same direction which results in appreciable total baryonic current
this baryonic current leads to sizable modifications in the @xmath46 , @xmath158 , @xmath159 , @xmath160 terms ( table [ table - cl34 ] ) .
these are precisely the same terms which are strongly affected by nm in odd - mass nuclei , see sect .
[ pol - odd ] . the fermionic contribution into @xmath80 ( the @xmath161 term )
is defined by more or less equal contributions from time - odd mean fields and the polarization effects in time - even mean fields .
on the contrary , time - odd mean fields define only @xmath162 ( @xmath163 mev ) of the mesonic contribution into @xmath80 , while the rest is due to the polarization effects in time - even mean fields ( the @xmath164 , @xmath165 terms ) .
nm leads to the loss of binding in the configuration b which has @xmath155 . in this configuration
, the proton and neutron currents due to the @xmath1665/2^+$ ] and @xmath1675/2 ^ -$ ] states are in opposite directions , so the total baryonic current is very close to zero . as a result , the impact of nm is close to zero for the majority of the terms in eq .
( [ etot ] ) ( see table [ table - cl34 ] ) .
the only exception is the @xmath149 term which represents space - like component of the isovector - vector @xmath168-field .
this term depends on the difference of proton and neutron currents ( eq . ( [ rho - space ] ) ) , which for the present case of opposite currents gives a non - zero result . as follows from table [ table - cl34 ] , this term is predominantly responsible for the loss of binding due to nm in the configuration b. it is well known that many physical quantities are additive in the calculations without pairing ( see ref .
@xcite and references therein ) .
the additivity principle states that the average value of a one - body operator @xmath169 in a given many - body configuration @xmath170 , @xmath171 , relative to the average value in the core configuration , @xmath172 , is equal to the sum of effective contributions @xmath173 of particle and hole states by which the @xmath170-th configuration differs from that of the core @xcite @xmath174 coefficients @xmath175 ( @xmath176 ) define the s.p .
content of the configuration @xmath170 with respect to the core configuration ( see ref .
@xcite for details ) .
let us check whether additional binding due to nm ( the @xmath177 quantity ) is additive .
the doubly magic sd configuration @xmath178 in even - even @xmath77s nucleus is used as a core for this analysis : the effective contributions @xmath179 of particle state(s ) to @xmath180 are given by @xmath181^{nm}- [ e_{i } ( { \rm nucleus\,\ , a } ) - e_{i } ( { \rm core})]^{wnm } = e_{i}^{nm } ( { \rm nucleus\,\ , a})- e_{i}^{wnm}({\rm nucleus\,\ , a } ) = \delta e_{i}({\rm nucleus\,\ , a}))$ ] because the core configuration is not affected by nm .
thus , the additivity implies that @xmath182 ( @xmath183 ) for the situation when the proton and neutron currents in @xmath154cl are in the same ( opposite ) directions .
[ multi ] clearly shows that additivity conditions are not fulfilled and that additional binding due to nm is not additive in self - consistent calculations .
the analysis involving odd - odd @xmath153p and odd @xmath184p , @xmath184s nuclei leads to the same conclusion ( see fig .
[ multi ] ) .
the additivity is also violated in perturbative calculations :
the comparison of tables [ table - pert ] ( columns 4 and 5 ) and [ table - cl34 ] ( columns 4 and 6 ) reveals that the conditions @xmath185 are violated both for the total energy ( @xmath186 ) and individual components of the total energy @xmath187 .
the analysis of @xmath188 ( this terms provides the largest contribution to @xmath189 ( see tables [ table - pert ] and [ table - cl34 ] ) ) allows to understand the origin of the violation of additivity for the @xmath189 quantity . in odd - proton @xmath184cl nucleus ,
@xmath190 } \approx - 1/2 \delta e_{split}^{p}$ ] ( @xmath191 is the energy splitting between blocked proton state and its signature counterpart ) and @xmath191 depends predominantly on the proton current induced by odd proton . the same is true in odd - neutron @xmath192s nucleus where @xmath193 depends predominantly on the neutron current induced by odd neutron .
additivity principle implies @xmath194 } \approx - ( 1/2 \delta e_{split}^{p } + 1/2 \delta e_{split}^{n})$ ] for the @xmath195 configuration , in which the proton and neutron currents are in the same direction .
however , proton @xmath196}$ ] ( neutron @xmath197}$ ] ) energy splitting between blocked proton ( neutron ) state and its time - reversal counterpart in odd - odd nuclei depend on total baryonic ( proton+neutron ) current in this configuration . on the contrary , the additivity principle implies that these proton and neutron quantities depend on individual proton and neutron currents in odd - odd nucleus , respectively .
this total current is approximately two times stronger than individual ( proton or neutron ) currents in odd mass nuclei . as a consequence , @xmath196}$ ] and @xmath197}$ ]
values in odd - odd mass nucleus are larger than the same quantities ( @xmath191 , @xmath193 ) in odd - mass nuclei by a factor close to 2 . as a result ,
@xmath198 ( see tables [ table - pert ] and [ table - cl34 ] ) which clearly indicates the violation of additivity for the @xmath188 quantity ( and for the @xmath189 quantity ) . quantity for different signatures .
the structure of the blocked states are shown by the nilsson labels only in the cases when the configurations are near - prolate .
the same state is blocked in the proton subsystem of all nuclei .
the bottom panel shows the @xmath199 and @xmath200-deformations of the configurations under study.,width=302 ] but for the lowest configurations in odd - odd cl nuclei.,width=302 ] line as a function of proton number @xmath116 .
panel ( a ) shows the results for the configurations with different blocked proton and neutron states , while panel ( b ) shows the results for the configurations with the same blocked proton and neutron states .
the structure of the blocked states are shown by the nilsson labels only in the cases when the configurations are near - prolate .
note that in panel ( b ) only one nilsson label is shown since blocked proton and neutron states have the same structure .
the results are shown only in the cases when the convergence has been achieved for both @xmath2 and @xmath201 ( or @xmath202 ) nuclei.,width=302 ] fig .
[ multi ] also shows the results for the 4-particle excited sd states @xmath203 in @xmath77s , for which the calculated rotational structures display the signature separation induced by time - odd mean fields @xcite .
the configurations are formed by exciting proton and neutron from the @xmath2041/2^{-}$ ] orbitals below the @xmath205 and @xmath206 sd shell gaps into the @xmath2075/2^{\pm}$ ] orbitals located above these gaps .
they have the @xmath208 structure in terms of intruder orbitals . when nm is neglected these four configurations are degenerated in energy .
this degeneracy is broken and additional binding , which depends on the total signature of the configuration ( 0.907 mev for the @xmath155 configurations and 0.468 mev for the @xmath156 configurations in the calculations with the nl3 parametrization ) , is obtained when nm is taken into account .
the nl1 and nlsh parametrizations of the rmf lagrangian give very similar values of additional binding due to nm . the essential difference between the relativistic and non - relativistic calculations
lies in ( i ) the size of the energy gap between the @xmath155 and @xmath156 configurations and ( ii ) the impact of time - odd mean fields on the energy of the @xmath156 states .
this energy gap is about 2 mev in the skyrme edf calculations with the sly4 force @xcite , while it is much smaller being around 0.45 mev in the crmf calculations with the nl1 , nl3 and nlsh parametrizations .
the energies of the @xmath156 states are not affected by time - odd mean fields in the skyrme edf calculations @xcite , while appreciable additional binding is generated by nm for these states in the crmf calculations ( fig .
[ multi ] ) . figs .
[ odd - odd - al ] and [ odd - odd - cl ] show the results of the calculations for ground state configurations in odd - odd al and cl nuclei .
the calculations suggest that signature separation due to time - odd mean fields is also expected in the configurations of odd - odd nuclei located at zero or low excitation energies .
the signature separation is especially pronounced in the @xmath2 @xmath209al ( the @xmath1665/2 \otimes \nu [ 202]5/2 $ ] configuration ) and @xmath154cl nuclei .
this is because proton and neutron currents in these configurations are almost the same both in strength and in spatial distribution . as a result ,
their contribution to the total energy is large when these currents are in the same direction ( the @xmath156 configurations ) and close to zero when these currents are in opposite directions ( the @xmath155 configurations ) .
note that @xmath209al is axially deformed while @xmath154cl is triaxially deformed with @xmath210 .
however , both of them show the enhancement of the signature separation at @xmath2 .
the signature separation is rather small for the majority of nuclei away from the @xmath2 line .
this is a consequence of the fact that the strength of the currents in one subsystem ( and thus the impact of nm on binding energies ) is much stronger than in another subsystem . as a result
, there is no big difference ( large signature separation ) between the cases in which proton and neutron currents are in the same and opposite directions .
however , some nuclei away from the @xmath2 line also show appreciable signature separation . these are @xmath211al and @xmath212cl nuclei ( figs .
[ odd - odd - al ] , [ odd - odd - cl ] ) for which the strengths of proton and neutron currents ( but not necessarily the spatial distribution of the currents ) are of the same order of magnitude .
it was suggested in ref .
@xcite that the effects of time - odd mean fields are enhanced at the @xmath2 line
. however , fig .
[ nznz ] clearly shows that the enhancement of signature separation is not restricted to the @xmath2 line . indeed ,
signature separation of the configurations based on the same combination of blocked proton and neutron states are very similar in the @xmath2 and @xmath213 nuclei despite the fact that the deformations of compared nuclei differ sometimes appreciably .
there are considerable signature separation in the configurations based on the same blocked proton and neutron states in the @xmath2 and @xmath201 nuclei ( fig .
[ nznz]b ) . on the other hand ,
almost no signature splitting is observed in the @xmath2 and @xmath202 nuclei when the configurations are based on different blocked proton and neutron states ( fig .
[ nznz]a ) .
this suggests that the enhancement of signature splitting is due to similar proton and neutron current distributions ( see discussion in previous paragraph ) .
when considering odd - odd nuclei one has to keep in mind that the present approach takes into account only the part of correlations between blocked proton and neutron and neglects the pairing .
in particular , the residual interaction of unpaired proton and neutron leading to the gallagher - moshkowski doublets of two - quasiparticle states with @xmath214 and @xmath215 @xcite is not taken into account .
thus , future development of the model is required in order to compare directly the experimental data on odd - odd nuclei with calculations .
this question will be discussed in more details in a forthcoming manuscript @xcite .
time - odd mean fields ( nuclear magnetism ) have been studied at no rotation in a systematic way within the framework of covariant density functional theory by performing the blocking of the single - particle states with fixed signature .
the main results can be summarized as follows : * in odd - mass nuclei , nuclear magnetism always leads to an additional binding indicating its attractive nature in the cdft .
this additional binding only weakly depends on the parametrization of the rmf lagrangian . on the contrary
, time - odd mean fields in skyrme edf can be attractive and repulsive and show considerable dependence on the parametrization of density functional .
this additional binding is larger in odd - neutron states than in odd - proton ones in the cdft framework .
the underlying microscopic mechanism of additional binding due to nm has been studied in detail .
the perturbative results clearly indicate that additional binding due nm is defined mainly by time - odd fields and that the polarization effects in fermionic and mesonic sectors of the model cancel each other to a large degree . *
additional binding due to nm can have a profound effect on the properties of odd - proton nuclei in the ground and excited states in the vicinity of the proton - drip line .
in some cases it can transform the nucleus which is proton unbound ( in the calculations without nm ) into the nucleus which is proton bound .
this additional binding can significantly affect the decay properties of proton unbound nuclei by ( i ) increasing the half - lives of proton emitters ( by many orders of magnitude in light nuclei ) or ( ii ) moving the @xmath216 value inside or outside the @xmath216 window favorable for experimental observation of proton emission .
* relative energies of different ( quasi)particle states in medium and heavy mass nuclei are only weakly affected by time - odd mean fields .
this is because additional bindings due to nm show little dependence on blocked single - particle state . as a result
, the present investigation suggests that time - odd mean fields can be neglected in the fits of covariant density functionals aimed at accurate description of the energies of the single - particle states .
* the phenomenon of signature separation @xcite and its microscopic mechanism have been investigated in detail .
it was shown that this phenomenon is active also in the configurations of odd - odd nuclei .
it is enhanced for configurations having the same blocked proton and neutron states ; this takes place either at ground state or at low excitation energy in the nuclei at or close to the @xmath2 line . some configurations away from the @xmath2 line
also show this effect but signature separation is appreciably smaller .
the present investigation has been focused on the study of time - odd mean fields in the cdft with non - linear parametrizations of the lagrangian .
point coupling @xcite and density dependent meson - nucleon coupling @xcite models are other classes of the cdf theories .
it is important to compare them in order to make significant progress toward a better understanding of time - odd mean fields .
the work in this direction is in progress , and the results will be presented in a forthcoming manuscript @xcite .
the material is based upon work supported by the department of energy under grant number de - fg02 - 07er41459 .
private communication of p. ring @xcite is greatly appreciated .
r. j. furnstahl , proc . of the ect
* school on `` renormalization group and effective field theory approaches to many - body systems '' , to be published in springer lecture notes in physics ; arxiv:0708.1602 at xxx.lanl.gov .
y. m. engel , d. m. brink , k. geoke , s. j. krieger , and d. vauterin , nucl . phys . *
a249 * , 215 ( 1975 ) . v.
o. nesterenko , w. kleinig , j. kvasil , p. vesely , and p .-
reinhard , int .
j. mod .
e * 17 * , 89 ( 2008 ) . | time - odd mean fields ( nuclear magnetism ) are analyzed in the framework of covariant density functional theory ( cdft ) by performing the blocking of the single - particle states with fixed signature .
it is shown that they always provide additional binding to the binding energies of odd - mass nuclei .
this additional binding only weakly depends on the rmf parametrization reflecting good localization of the properties of time - odd mean fields in cdft .
the underlying microscopic mechanism is discussed in detail .
time - odd mean fields affect odd - even mass differences .
however , our analysis suggests that the modifications of the strength of pairing correlations required to compensate for their effects are modest .
in contrast , time - odd mean fields have a profound effect on the properties of odd - proton nuclei in the vicinity of the proton - drip line . their presence can modify the half - lives of proton - emitters ( by many orders of magnitude in light nuclei ) and considerably affect the possibilities of their experimental observation . |
in several recent papers [ 1 - 4 ] we have developed a highly predictive model of quark and lepton masses based on the grand unified group @xmath0 .
this model grew out of our attempt @xcite to construct a realistic grand unified theory ( gut ) in which @xmath0 was broken down to the standard model in the simplest possible , or minimal " way @xcite . in this model there emerged a new mechanism based on certain well - known features of @xmath2 for explaining the large mixing between the mu and tau neutrinos that is seen at superkamiokande@xcite .
in @xcite we gave the structure of the quark and lepton mass matrices for the second and third families , treating the first family as massless .
in @xcite , it was shown how to extend the model to include the first family , which leads to several interesting predictions . in @xcite , it was observed that the mixing of the electron neutrino very naturally falls either within the range @xmath3 , corresponding to the small angle msw solution @xcite of the solar neutrino problem , or very near to the value @xmath4 , corresponding to what is called bimaximal mixing " .
in this paper we present the model in fuller detail , especially in regard to neutrino phenomenology , and to the structure of the higgs sector , yukawa interactions , and flavor symmetries that underlie the quark and lepton mass matrices .
the paper is organized as follows . in section
ii we discuss in general terms , that is apart from a particular model , our mechanism for explaining the large mixing of @xmath5 and @xmath6 . in section iii , we explain what we mean by a minimal " @xmath0 breaking scheme , and show how such minimal breaking and the requirements of simplicity lead one naturally to a certain form for the mass terms of the heavier two families of quarks and leptons .
we then observe that this form realizes the general mechanism for large @xmath7 mixing described in the previous section .
it is important to emphasize that this mechanism emerged not from an attempt to explain neutrino phenomenology , but from other considerations entirely , in particular the attempt to simplify the higgs structure of @xmath0 .
it is most interesting that the same mechanism has also independently been found by other groups attempting to make sense of neutrino phenomenology . in section iv
, it will be explained how this model is best extended to the first family of quarks and leptons , and how this gives rise to several distinctive predictions .
accurate analytic expressions for the predictions at the gut scale will be presented . in section
v , the neutrino sector will be examined in detail .
it will be seen how either the small - angle msw solution of the solar neutrino problem or bimaximal mixing can result with equal simplicity .
finally , in section vi , a concrete model , including all the details of flavor symmetries and of the higgs and yukawa superpotentials , will be presented , showing that the basic scheme is technically natural .
before explaining our mechanism , it will be helpful to explain why the observed large mixing of @xmath8 , presumably with @xmath6 , has been a theoretical puzzle .
the basic reason is simple : the mixing that is seen between the quarks of the second and third families is described by a small mixing angle , namely @xmath9 , and therefore it was expected that the mixing between the second and third family of leptons would also be small . the grounds for this expectation were twofold
first , there is the empirical fact that the masses of the quarks and leptons exhibit roughly similar hierarchical " patterns , and therefore it was natural to assume that their mixing angles would be similar also .
second , the most promising theoretical approaches to understanding the pattern of quark and lepton masses , namely grand unification and flavor symmetry , tend to treat quarks and leptons in similar ways .
for instance , small quark mixing angles might suggest an underlying fundamental family symmetry " that is weakly broken , in which case the lepton mixings might be expected also to be small .
and in grand unification based on @xmath0 there is a close connection between the quark and lepton mass matrices .
there are actually two puzzles associated with the mixing of the second and third families : first , why is the lepton mixing @xmath10 so large
? and , second , why is the quark mixing @xmath9 so small ?
what we mean by saying that these are distinct puzzles is that they are both unexpected within the most commonly assumed framework for explaining quark and lepton masses , the weinberg - wilczek - zee - fritzsch ( wwzf ) idea @xcite .
the wwzf idea was that the cabibbo angle could be understood if the mass matrices of the first and second families of quarks had the following form : @xmath11 this gives @xmath12 , @xmath13 , and @xmath14 , and thus the famous relation @xmath15 since @xmath16 , @xmath17 , and @xmath18 , this relation is satisfied for @xmath19 .
if we apply the same idea to the leptons of the first two families we get @xmath20 the second term on the right is not known , but if it is assumed to be small one has the rough prediction that @xmath21 .
this could be consistent with the small angle msw solution of the solar neutrino problem , which requires that @xmath22 .
thus the wwzf idea appears to work well where it was originally applied , namely to the first and second families .
fritzsch @xcite later extended this idea to explain the mixing of the third family . if a wwzf form is assumed to hold for the second and third family , i.e. , if one takes @xmath23 and @xmath24 in eq .
( 1 ) , one obtains @xmath25 and @xmath26 since @xmath27 , and @xmath28 , one sees that the observed value of @xmath9 is too small by a factor of three or so . assuming that the neutrino mass ratio in eq .
( 5 ) is small , and given that @xmath29 , one sees that the nearly maximal value of @xmath30 that is observed is too large by a factor of three or so .
these eqs .
( 2 - 5 ) are based on the assumption of a hierarchical and symmetric form for the mass matrices .
a key feature in our mechanism for understanding the large mixing of the tau neutrino is that it involves highly asymmetric mass matrices .
as we shall see , the assumption of asymmetric mass matrices naturally explains why @xmath31 is larger than the fritzsch value and @xmath32 is smaller than the fritzsch value by approximately the same factor .
consider , a toy model with @xmath2 symmetry , which has a set of yukawa terms of the following form : @xmath33 , with @xmath34 and the subscript @xmath35 denoting a higgs representation .
these terms yield the following mass matrices for the second and third families of down quarks and charged leptons : @xmath36 with @xmath37 .
here we have labelled the fermions with a family index , instead of the names @xmath38 , @xmath39 , @xmath40 , and @xmath41 , since the mass matrices in this case are far from diagonal . a crucial point to notice
is that the matrix for the leptons , which we will denote by @xmath42 , is the transpose of the matrix for the down quarks , which we will denote by @xmath43 .
this is a feature of minimal @xmath2 .
it arises from the fact that the @xmath44 representation of fermions contains the left - handed leptons , @xmath45 , and the charge conjugate of the right - handed down - quarks , @xmath46 , while the @xmath47 representation of fermions contains the charge conjugate of the right - handed leptons , @xmath48 , and the left - handed down - quarks , @xmath49 .
thus , @xmath2 relates @xmath43 to @xmath42 , but only up to a left - right transposition : @xmath50 . the transposition feature of @xmath2 unification appearing in eq . ( 6 )
results in the large element , @xmath51 , of @xmath42 producing an o(1 ) mixing of @xmath52 with @xmath53 for the leptons , while in @xmath43 for the quarks it produces a large mixing of the right - handed fields @xmath54 and @xmath55 .
the mismatch between the large @xmath56 mixing and the @xmath57 mixing , which is small ( as will soon be seen ) , leads to a large @xmath31 mixing element .
but the right - handed mixings of the quarks are not observable through standard model physics .
what matters is the left - handed mixing of @xmath58 with @xmath59 , which contributes to @xmath32 , and is controlled by the small parameter @xmath60 .
the common statement that grand unification relates quark and lepton mixing angles , and thus @xmath32 to @xmath31 , is very misleading .
what is really true in general is that grand unification relates the mixing of quarks of one handedness to the mixing of leptons of the other handedness .
thus @xmath32 and @xmath31 need not be directly related to each other .
of course , if the mass matrices are symmetric , as has almost always been assumed , the left - handed and right - handed mixings are the same , and hence @xmath32 is directly related to @xmath31 . the most natural interpretation , then , of the experimental discovery that @xmath61 is that the mass matrices are highly asymmetric .
this is the essential point first made in @xcite .
not only does a highly asymmetric , or , as we will call it , lopsided , " form of the mass matrices explain the difference between the size of @xmath31 and @xmath32 , but it also explains the fact , noted above , that @xmath31 is larger than the fritzsch value and @xmath32 is smaller than the fritzsch value by about the same factor .
the point is that the product of the two off - diagonal elements , @xmath60 and @xmath51 , is controlled by the fermion mass ratio . as is evident from eq .
( 6 ) , @xmath62 .
that means that the fritzsch prediction for the mixing of @xmath49 and @xmath63 , which is @xmath64 , goes approximately as @xmath65 .
that shows that the fritzsch prediction for the mixing angles is roughly the geometric mean between the true value of @xmath66 and the true value of @xmath32 @xmath67 . in other words , in our hypothesis of lopsided mass matrices ,
the surprising largeness of @xmath31 and the surprising smallness of @xmath32 are two sides of the same coin .
another important feature of this mechanism should be emphasized .
almost all published explanations of the largeness of the @xmath7 mixing trace it to some special feature or form of the neutrino mass matrix .
perhaps this is due to the purely linguistic fact that we talk about neutrino mixing angles " .
but they could just as well be called the charged - lepton mixing angles " .
they are really the angles expressing the mismatch between the neutrino mass eigenstates and the charged lepton mass eigenstates , just as the ckm angles are the mismatch between the up and down quark eigenstates . in our mechanism
, the large value of @xmath31 is traceable to a peculiarity of the charged lepton mass matrix @xmath42 , namely , having a large off - diagonal entry @xmath51 .
as we shall see in the next section , having such a large entry helps to explain several other features of the quark and lepton mass spectrum . to sum up
, the mechanism for explaining large @xmath7 mixing proposed in @xcite has three salient features : ( 1 ) the largeness of this mixing is due to the charged lepton mass matrix , which is ( 2 ) highly asymmetric , and which is ( 3 ) related to the transpose of the down quark mass matrix by @xmath2 . in the next section
we will see how a model with precisely these features arises very naturally in @xmath0 from very different considerations .
the model that we shall examine in this paper emerged originally from our attempt to construct a realistic model based on @xmath0 @xcite in which @xmath0 is broken to the standard model group , @xmath68 in the simplest possible way .
we shall therefore start by explaining what we mean by minimal @xmath0 breaking . since @xmath0 is a rank 5 group , it requires for its breakdown to @xmath69 at least two higgs fields .
one higgs field is needed to break the rank of the group to 4 , but this generally leaves an unbroken @xmath2 .
the second higgs field is needed to break @xmath2 down to @xmath69 .
the two breakings can occur in either sequence depending upon which higgs field has the larger vev and effects the first breaking .
whatever higgs field gives superlarge mass to the right - handed neutrinos , as required for the standard seesaw explanation of the lightness of the left - handed neutrinos , will also break @xmath0 down to @xmath2 , and thus the rank to 4 .
there are two simple choices for this higgs field : either an antisymmetric five - index tensor @xmath70 or a spinor @xmath71 . in either case , one also expects a higgs field in the conjugate representation , @xmath72 or @xmath73 , to go along with it .
a nice feature of the @xmath70 is that this tensorial representation leaves unbroken a @xmath74 subgroup of the center of @xmath0 that acts as an automatic matter parity , whereas if a spinor higgs is introduced , then matter parity is not automatic . on the other hand , to introduce @xmath75 is to introduce quite large representations that tend to make the unified gauge coupling go non - perturbative below the planck scale , and that may be hard to obtain from superstring theory . in any event , it would seem that the use of a spinor - antispinor pair , @xmath76 , is more economical .
thus we assume that the rank of @xmath0 is broken at the unification scale and the right - handed neutrinos get mass from one such spinor - antispinor pair of higgs fields . to break the group the rest of the way to @xmath69 requires the existence of higgs fields in the adjoint representation @xmath77 and/or in the symmetric two - index tensor representation @xmath78 .
most published realistic @xmath0 models have several of both kinds of multiplets .
however , it has been shown that it is possible to break @xmath0 to @xmath69 with only a _ single _
adjoint higgs and no larger representations @xcite .
this , then , is what we call the
minimal breaking scheme for @xmath0 " : _ the breaking of @xmath0 to @xmath69 is accomplished by the expectation values of a set of higgs fields consisting of _ @xmath79 , _ with the model containing no multiplets larger than the single _ @xmath77 . there , of course , have to be other higgs fields to break the @xmath80 group of the electroweak interactions .
this minimality assumption is restrictive enough that it is possible to say in which direction the expectation values of these fields point .
this can be done by considering the problem of doublet - triplet splitting , whereby the colored partners of the weak - doublet higgs fields of the standard model become superheavy while the weak - scale masses of the doublets themselves are preserved . in @xmath0 the only known way of doing this in a technically natural manner is the dimopoulos - wilczek or
missing vev " mechanism @xcite .
the idea is that if an adjoint higgs field that has an expectation value proportional to the @xmath0 generator @xmath81 couples to higgs fields in the vector representation , it will make their color - triplet components heavy ( since they have @xmath82 ) while leaving their weak - doublet components massless ( since they have @xmath83 ) .
the needed coupling is simply of the form @xmath84 .
of course , the expectation value of the adjoint , by virtue of the definition of the adjoint representation , is necessarily a linear combination of generators of the group .
@xmath81 is one of the @xmath0 generators that is picked out by simple forms of the higgs superpotential for the adjoint multiplet .
it should be noted that there is another version of the missing vev mechanism that works in @xmath0 , in which the vev of the adjoint is proportional to the generator @xmath85 of the @xmath86 subgroup of @xmath0 @xcite .
however , that version is significantly more complicated .
therefore , simplicity dictates the choice that @xmath87 since the assumption of a minimal @xmath0 breaking scheme included the supposition that only _
one _ adjoint exists in the model , no adjoint exists except the one that points in the @xmath81 direction .
as we shall see , this puts an important limitation on the possibilities for constructing realistic mass matrices for the quarks and leptons .
the assumption of a minimal @xmath0 breaking scheme thus acts as an important guide in searching for good models .
the simplest possible terms that would give mass to quarks and leptons in @xmath0 would be @xmath88 , where the subscripts @xmath89 and @xmath90 are family indices .
this would lead to four proportional dirac mass matrices for the up - quarks ( @xmath91 ) , down quarks ( @xmath43 ) , charged leptons ( @xmath42 ) , and neutrinos ( @xmath92 ) .
in fact one would have @xmath93 .
moreover , all these matrices would be symmetric , which is why one can write @xmath94 instead of @xmath50 as in minimal @xmath2 . some of the predictions that follow from these relations are good , notably the famous prediction @xmath95 , where the superscript zero stands for quantities evaluated at the unification scale @xmath96 .
however , @xmath97 also predicts that @xmath98 and @xmath99 .
empirically , one finds instead that @xmath100 and @xmath101 . these factors of three are called the georgi - jarlskog factors @xcite . the simplest possible @xmath0 yukawa terms also predict that all the ckm angles vanish , since @xmath102 . while not exactly true , this is at least a good zeroth order relation , since the ckm angles are all small compared to unity . by contrast , in @xmath2 the matrices @xmath43 and @xmath91 are not related by the unified symmetry and so the ckm angles are unconstrained . the smallness of the ckm angles can be regarded , therefore , as evidence for @xmath0 . on the other hand , the proportionality of @xmath43 and @xmath91 in @xmath0 also predicts that @xmath103 , which fails badly by over an order of magnitude .
what one can conclude is that a way of going beyond the simplest possible @xmath0 yukawa scheme must be found which preserves some of its predictions while breaking others .
one way to do this involves using larger representations to break the electroweak interactions .
for instance , in the original georgi - jarlskog model , a @xmath104 multiplet of @xmath2 ( not to be confused with the adjoint of @xmath0 ) participates in breaking @xmath105 . in the context of @xmath0 , this @xmath104 is contained in a @xmath70 , which is inconsistent with our minimality assumptions .
more economical is to assume that the higgs fields that break @xmath0 at the unification scale , i.e. , the @xmath106 , couple to quarks and leptons and thus introduce the effects of that @xmath0 breaking into the quark and lepton mass relations .
this is the assumption we make . to describe the third family
it is simplest to assume the minimal yukawa term @xmath107 as pictured in fig .
1(a ) . by itself
, this would make all the mass matrices have the form @xmath108 that would give the following predictions , all of which are at least good zeroth approximations to reality : @xmath95 , @xmath109 , and @xmath110 , where @xmath111 is a mass of a fermion of the @xmath112 family .
note that these are just the good " @xmath0 predictions mentioned above .
the second family presents more of a challenge .
the main issue is how to get the georgi - jarlskog factor of 3 between @xmath113 and @xmath114 .
breaking of @xmath2 must be involved , since the bad relation @xmath98 arises already at the @xmath2 level .
the only field that breaks @xmath2 in the framework of minimal @xmath0 breaking is the adjoint , @xmath115 . since @xmath116 , and
the @xmath81 of leptons is @xmath117 times that of quarks , this field has the possibility of giving the needed georgi - jarlskog factor .
thus one must seek an effective yukawa term that involves the @xmath115 .
the simplest such term @xcite , in the sense of the term of lowest dimension , is of the form @xmath118 .
moreover , this term can arise in a simple way by the integration out of a @xmath119 family plus antifamily at the unification scale , as shown in fig .
there are actually two ways to contract the @xmath0 indices of such a term : the product @xmath120 can be contracted symmetrically or antisymmetrically .
it is easy to show that if @xmath121 , only the antisymmetric contraction contributes to the quark and lepton mass matrices .
( the reason is simple .
if the vev of the adjoint is proportional to a generator @xmath122 , then the symmetric / antisymmetric contractions give contributions to fermion masses that go as @xmath123 .
since @xmath81 of an antifermion is minus that of the fermion , the contribution cancels for the symmetric contraction . )
thus , one need only consider the flavor - antisymmetric term , which means only @xmath124 and not @xmath125 , or @xmath126 .
consequently , the only operator of interest is @xmath127 which , together with the operator @xmath107 , gives @xmath128 the desired factor of 3 has been achieved between leptons and quarks , due to the generator @xmath81 to which the adjoint vev is proportional .
one also can see that the @xmath60 entries are flavor antisymmetric for reasons already explained .
as they stand , these forms of the matrices are inadequate to explain even the features of the second and third families of fermions .
there are three inadequacies .
( 1 ) the factor of 3 comes in squared between the mass of the leptons and quarks of the second family .
the reason is that , for @xmath60 small due to the mass hierarchy between families , the second eigenvalue of @xmath42 is given by the seesaw formula @xmath129 , while the second eigenvalue of @xmath43 is given by @xmath130 .
( 2 ) the matrices @xmath43 and @xmath91 are still exactly proportional .
this is a consequence of the fact that the generator @xmath81 does not distinguish up and down quarks .
therefore , the ckm angle @xmath32 still exactly vanishes .
( 3 ) because @xmath43 and @xmath91 are exactly proportional , one still has the bad prediction @xmath103 .
it is clear that the breaking of @xmath0 due to the adjoint can not cure all of these problems , since @xmath81 does not distinguish @xmath43 from @xmath91 .
thus the breaking of @xmath0 done by @xmath131 must come into play . as we shall now show , a single simple operator exists , which involves one of these spinor higgs and cures at one stroke all three of the problems we have identified . the lowest dimension effective yukawa operators that involve the spinor higgs fields are quartic in spinors .
consider , therefore , operators of the form @xmath132 .
the @xmath133 is the spinor higgs field that breaks @xmath0 at @xmath96 down to @xmath2 .
the @xmath134 is a spinor higgs that has a weak - scale vev that breaks @xmath80 . in principle
, these two spinors could be the same field .
however , if they were , it would mean that they had to be contracted symmetrically by bose statistics , which in turn would mean that @xmath135 and @xmath136 would also have to be contracted symmetrically .
a careful examination shows that the resulting flavor - symmetric contributions to the mass matrices do not lead to realistic forms , though it is possible to achieve realistic mass matrices by adding yet another yukawa operator , as in the interesting model of babu , pati , and wilczek @xcite
. therefore @xmath137 must be a distinct field .
as will be seen later , introducing this @xmath138 involves no loss of economy , since it allows a very elegant explanation of the largeness of the ratio @xmath139 without making @xmath140 large .
there are still several operators of this type to be considered : the family indices can take the values @xmath141 , or @xmath142 , and there are three ways to contract the four spinors to make an @xmath0 singlet . here
again , one must examine the various cases to see which gives the most realistic mass matrices . as it turns out ,
there is one operator that is much superior to the others , in the sense that it much more cleanly and simply fits the data .
it is of the form @xmath143 [ { \bf 16}_3 { \bf 16'}_h]$ ] , where @xmath144 $ ] means that the spinors inside are contracted into a @xmath47 .
this can arise very simply by integrating out a @xmath47 of fermions , as shown in fig .
let us write the resulting mass operator in @xmath2 language .
denote by @xmath145 a @xmath146 multiplet of @xmath2 that is contained in a @xmath147 multiplet of @xmath0 .
the vev of @xmath133 lies , of course , in the @xmath148 direction , while the vev of @xmath138 that breaks the weak interactions lies in the @xmath149 direction .
therefore , the resulting mass term is of the form @xmath150 [ { \bf 10}({\bf 16}_3 ) \overline{{\bf 5 } } ( { \bf 16}'_h)]$ ] , which in @xmath2 terms gives effectively the operator @xmath151 .
note that this has the same form as the @xmath2 operator discussed in the last section , which gave the @xmath51 entries in eq .
the result , then , of including this operator @xcite is to make the mass matrices take the form : @xmath152 the new term has given the entries we call @xmath51 .
note that these lopsided entries appear only in @xmath43 and @xmath42 .
the reason is simply that the @xmath138 contains a @xmath44 of @xmath2 but no @xmath153 .
it is easy to see that the new term with @xmath154 cures at once all three of the problems we identified with the forms given in eq .
( 9 ) : ( 1 ) instead of @xmath129 and @xmath155 , one has approximately that @xmath156 and @xmath157 .
more exact expressions will be given later .
thus the desired georgi - jarlskog factor of 1/3 is obtained , instead of 1/9 .
the @xmath51 entry has dominated over one of the factors of @xmath158 and thus prevented the factor of 1/3 from coming in squared .
\(2 ) the @xmath51 entry comes into @xmath43 but not @xmath91 , and thus breaks the proportionality of the two matrices . as a result , @xmath32 no longer vanishes , but is given approximately by @xmath159 .
note that this is of the same order in @xmath60 as @xmath160 , rather than @xmath64 as is the case with fritzsch forms , and accords much better with the actual experimental values .
\(3 ) the fact that @xmath51 breaks the proportionality of @xmath91 and @xmath43 also means that the bad relation @xmath103 is broken .
specifically , @xmath161 is of order @xmath60 , while @xmath162 is still of order @xmath163 and therefore much smaller .
this also accords well with the experimental numbers .
in fact , as we shall see , if one uses @xmath32 and @xmath164 to fix the parameters @xmath51 and @xmath60 , one finds that @xmath165(1 gev ) is predicted to be in agreement with the experimentally determined value of @xmath166 gev .
it should also be noted that the prediction @xmath95 is only very slightly affected by the addition of the @xmath51 term , both @xmath167 and @xmath168 being given to leading order in @xmath60 by @xmath169 .
the economy of the above mass matrix forms is seen in the fact that five quantities ( @xmath32 , @xmath164 , @xmath170 , @xmath171 , and @xmath172 ) are successfully fit with only the two parameters @xmath51 and @xmath60 .
no other published form succeeds in accurately reproducing the masses and mixing of the the heavier two families with so few parameters .
the predictions and fits will be discussed in detail in section iv .
we see that the matrices in eq .
( 10 ) were arrived at by a process of reasoning that had nothing to do with the question of neutrino mass but rather with an attempt to get realistic masses and mixings for the quarks and charged leptons using as simple a higgs sector as possible in @xmath0 .
but what has emerged is a structure with precisely the three critical features identified in the last section as giving a simple explanation of the large mixing of @xmath5 and @xmath6 . in fact
, a fit of @xmath32 and @xmath164 gives @xmath173 and @xmath174 .
consequently , as can be seen directly from eq .
( 10 ) , the angle @xmath175 .
this is quite consistent with what is observed .
we will look more carefully at these predictions later . to summarize , with two parameters , @xmath60 and @xmath51 , four mass ratios and two mixing angles are satisfactorily accounted for , if we include @xmath31 .
no greater economy could be hoped for in explaining the spectrum of the heavy two families . moreover , as we shall see in the next section , the forms in eq .
( 10 ) can be extended to include the first family with equal economy : the introduction of two new parameters ( one of which is complex ) will nicely account for seven quantities pertaining to the first family . before explaining how the model is extended to the first family we will expand on a couple of points made earlier .
first , we said that the introduction of the @xmath138 allows a simple explanation @xcite of why @xmath176 that does not require a @xmath177 .
the point is that the higgs doublet of the mssm that is often called @xmath178 is purely contained in the @xmath179 that couples to @xmath180 and gives rise to the 1 " entry in the mass matrices of eq .
however , the higgs doublet of the mssm that is called @xmath181 does not come purely from @xmath179 .
rather it is a mixture of doublets in @xmath179 and @xmath138 , since they both contain @xmath44 s of @xmath2 .
thus we may write @xmath182 where @xmath183 is some mixing angle that depends on the parameters of the higgs sector . since the 33 elements of the mass matrices all arise purely from the coupling of the @xmath179 , the parameters we called @xmath184 and @xmath185 in eq . (
10 ) are given by @xmath186 leading to the ratio @xmath187 , where @xmath140 is defined to be the ratio of the higgs vev @xmath188 giving mass to the top quark to the higgs vev @xmath189 giving mass to the bottom quark .
hence @xmath190 the point is simply that the large ratio of the top to bottom masses could be the result of @xmath191 being small rather than @xmath140 being large .
in fact , since we do not know anything _ a priori _ about the angle @xmath183 , we can not say whether @xmath192 is large or small .
it should be noted that if one assumes that @xmath181 lies mostly in the @xmath138 ( so that @xmath193 ) , it would explain why the parameter @xmath51 is large since it comes from a coupling to @xmath138 , and also explain why @xmath192 might be small .
a second point we wish to underline here has to do with the reasonableness of asymmetric mass matrices .
in many models it is assumed that all the mass matrices are symmetric .
however , this is not something that is called for by the group theory of grand unification .
it is true that with the minimal yukawa terms @xmath2 gives a symmetric @xmath91 .
but even with minimal yukawa terms @xmath2 does not predict any symmetry of the @xmath43 and @xmath42 matrices . and in @xmath0 , as we have seen , once one introduces the effects of @xmath0 breaking into the yukawa sector , as one virtually must , one easily obtains effective yukawa terms that are asymmetric .
1(c ) shows that very simple diagrams can give terms that are lopsided , in the sense that they contribute only above or below the diagonal . from the point of view of the fundamental grand unified theory ,
then , lopsided terms are as natural as symmetric ones .
the preference for symmetric terms has been the result not of examining what kinds of terms are obtained in a simple way in unification , but rather from the desire to reduce the number of parameters at the level of mass matrices with the aim of making models which are highly predictive . however , putting oneself in the straightjacket of symmetric matrices makes it hard to get a good fit to all the quark and lepton masses and mixings .
it turns out , as we have seen , and will see further below , that allowing asymmetric matrices makes possible a model which gives both a very good fit to the data and is actually much more predictive than most models which assume symmetric matrices .
in arriving at the form of the mass matrices for the heavy two families we were limited in the choices that were possible by the assumption we made about the simplicity of the @xmath0-breaking sector . in extending to the first family
we are not quite so limited . nevertheless , the number of simple possibilities is not very large .
there are several discrete choices : should the contributions to the first row and column of the mass matrices be flavor symmetric like the 1 s in eq .
10 , antisymmetric like the @xmath60 s , or lopsided like the @xmath51 s ?
should they contribute to all the matrices equally like the 1 s , to all the matrices but with non - trivial clebsch factors like the @xmath60 s or only to @xmath43 and @xmath42 like the @xmath51 s ?
it is fairly easy to run through the various cases and see what kinds of relations among masses and mixings result .
as it turns out , one of the simplest possibilities gives a remarkably good fit to the data .
this uniquely simple choice @xcite is the following : @xmath194 we have already mentioned that fits give @xmath195 and @xmath174 . the new parameters @xmath196 and @xmath197 both have magnitude of about @xmath198 .
the parameter @xmath199 is by far the smallest , being about @xmath200 .
the only role that @xmath199 plays in the sector of quarks and charged leptons is in giving the up quark a mass , for it makes negligible contributions to the down quark and electron masses as determined from @xmath43 and @xmath42 , respectively . in fig .
2(a ) we have illustrated a higher - order diagram that can contribute to the parameter @xmath199 . since it is not excluded that the up quark is exactly massless , it is possible to set @xmath199 to zero . in any event , one can see that @xmath201 , which is by orders of magnitude smaller than any other interfamily ratio of masses in the standard model .
it will , however , be of some significance for neutrino masses .
if @xmath199 vanishes , this model gives only the small - angle msw solution to the solar neutrino problem .
but even if @xmath199 is as small as @xmath200 , it allows either the small - angle msw solution or bimaximal neutrino mixing to arise in a simple way . turning to the parameters @xmath196 and @xmath197
, we see that they appear symmetrically and only in @xmath43 and @xmath42 .
such terms are easily obtained in @xmath0 from simple diagrams such as that shown in figs .
2(b ) and 2(c ) .
the effective operators arising from these diagrams are of the form @xmath202[{\bf 16}_h { \bf 16}'_h]$ ] with @xmath203 , where again the spinors in brackets are contracted symmetrically into a @xmath47 of @xmath0 which is integrated out . note , however , that the symmetric contributions @xmath196 and @xmath197 from the two higgs contraction @xmath204 $ ] contributes only to @xmath43 and @xmath42 by virtue of their @xmath2 structure .
contrast these effective operators for @xmath203 with that occurring previously for the term @xmath51 arising from the diagram shown in fig .
the three new parameters we have introduced are , as we shall see , sufficient to account for everything about the first family . before proceeding , however
, we must be careful about complex phases .
it is easy to show that if we allow all parameters of the model to be complex all but two phase angles can be rotated away from the mass matrices @xmath91 , @xmath43 , @xmath42 and @xmath92 , provided we now neglect the negligible @xmath199 contributions to @xmath43 and @xmath42 . we will call these physical phases @xmath205 and @xmath206 which appear as follows , @xmath207 where in these matrices and henceforth @xmath60 , @xmath51 , @xmath196 , @xmath197 and @xmath199 denote the magnitudes of these parameters and the phases are written explicitly .
the phase @xmath205 only comes into the fits of masses at higher order in the small quantity @xmath208 .
numerically , its effect is only a few percent ; moreover , the fits ( especially to @xmath165 ) prefer a value near zero .
therefore , we can ignore @xmath205 and will do so from now on . that leaves only the phase @xmath206 .
its only significant effect , but a very important one , is to give the cp - violating phase angle @xmath209 . instead of using the parameters @xmath196 , @xmath197 and @xmath210
, it will be somewhat more convenient to use the parameters @xmath211 , @xmath212 , and @xmath213 , which are defined in terms of them as follows : @xmath214 and @xmath215 the significance of these parameters is that they are essentially the left - handed and right - handed cabbibo angles .
this can be seen by taking the forms for @xmath43 and @xmath42 given in eq .
( 15 ) and diagonalizing the 2 - 3 block .
when this is done the 1 - 2 blocks of these matrices take the form @xmath216 } \propto \left ( \begin{array}{cc } 0 & t_r \\ t_l & 1 \end{array } \right ) , \;\;\;\ ; l^{[12 ] } \propto \left ( \begin{array}{cc } 0 & t_l \\ t_r & 3 \end{array } \right).\ ] ] in terms of the five dimensionless parameters @xmath60 , @xmath51 , @xmath211 , @xmath212 , and @xmath213 with @xmath199 set equal to zero , we now write down expressions for fourteen observable quantities : seven ratios of quark and lepton masses , three ckm angles and one phase , and three lepton mixing angles .
@xmath217 m_c^0/m_t^0 & \cong & \frac{1}{9 } \epsilon^2 \cdot [ 1 - \frac{2}{9 } \epsilon^2 ] , \\[0.2 in ] m_{\mu}^0/m_{\tau}^0 & \cong & \epsilon \frac{\sigma}{\sigma^2 + 1 } \cdot [ 1 + \epsilon \frac{1 - \sigma^2 - \sigma \epsilon}{\sigma(\sigma^2 + 1 ) } + \frac{1}{18}(t_l^2 + t_r^2 ) ] , \\[0.2 in ] m_s^0/m_b^0 & \cong & \frac{1}{3 } \epsilon \frac{\sigma}{\sigma^2 + 1 } \cdot [ 1 + \frac{1}{3 } \epsilon \frac{1 - \sigma^2 - \sigma \epsilon/3 } { \sigma ( \sigma^2 + 1 ) } + \frac{1}{2 } ( t_l^2 + t_r^2 ) ] , \\[0.2 in ] m_u^0/m_t^0 & = & 0 , \\[0.2 in ] m_e^0/m_{\mu}^0 & \cong & \frac{1}{9 } t_l t_r \cdot [ 1 - \epsilon \frac{\sigma^2 + 2}{\sigma(\sigma^2 + 1 ) } + \epsilon^2 \frac{\sigma^4 + 9 \sigma^2/2 + 3}{\sigma^2(\sigma^2 + 1)^2 } - \frac{1}{9 } ( t_l^2 + t_r^2 ) ] , \\[0.2 in ] m_d^0/m_s^0 & \cong & t_l t_r \cdot [ 1 - \frac{1}{3 } \epsilon \frac{\sigma^2 + 2}{\sigma ( \sigma^2 + 1 ) } - ( t_l^2 + t_r^2 ) + ( t_l^4 + t_l^2 t_r^2 + t_r^4 ) ] , \\[0.2 in ] v_{cb } & \cong & \frac{1}{3 } \epsilon \frac{\sigma^2}{\sigma^2 + 1 } \cdot [ 1 + \frac{2}{3 } \epsilon \frac{1}{\sigma(\sigma^2 + 1 ) } ] , \\[0.2 in ] v_{us } & \cong & t_l [ 1 -\frac{1}{2 } t_l^2 - t_r^2 + t_r^4 + \frac{5}{2 } t_l^2 t_r^2 + \frac{3}{8 } t_l^4 - \frac{\epsilon}{3 \sigma \sqrt{\sigma^2 + 1 } } \frac{t_r}{t_l } e^{- i \theta } ] , \\[0.2 in ] v_{ub } & \cong & \frac{1}{3 } t_l \epsilon \frac{1}{\sigma^2 + 1 } [ \sqrt{\sigma^2 + 1 } \frac{t_r}{t_l } e^{-i \theta } ( 1 - \frac{1}{3 } \epsilon \frac{\sigma}{\sigma^2 + 1 } ) - ( 1 - \frac{2}{3 } \epsilon \frac{\sigma}{\sigma^2 + 1 } ) ] , \\[0.2 in ] u_{\mu 3}^0 & \equiv & \sin \theta_{\mu \tau } \cong \frac{\sigma}{\sqrt{\sigma^2 + 1 } } + o(\epsilon ) , \\[0.2 in ] u_{e 2}^0 & \cong & \cos \theta_{\mu \tau } \left ( \frac{1}{3 } t_r \right ) \cdot [ 1 + \epsilon ( \frac{\tan \theta_{\mu \tau}}{\sigma^2 + 1 } - \frac{t_l}{t_r } e^{i \theta } \frac{(1 + \sigma \tan \theta_{\mu \tau } ) } { \sigma \sqrt{\sigma^2 + 1 } } ) - \frac{1}{18 } t_r^2 - \frac{1}{9 } t_l^2 ] , \\[0.2 in ] u_{e 3 } & \cong & \tan \theta_{\mu \tau } u_{e 2 } \cdot [ 1 + \epsilon \frac{2}{\sin 2 \theta_{\mu \tau } } \left ( \frac{t_l}{t_r } e^{i \theta } \frac{1}{\sqrt{\sigma^2 + 1 } } - \frac{1}{\sigma^2 + 1 } \right ) ] , \\ \end{array}\ ] ] these expansions have been carried to sufficiently high order in small quantities to be accurate to within 0.2% and are useful in doing the fits to the data .
however , the leading terms in these expansions have much simpler forms and thus allow one to see more readily the relationships among various quantities in this model .
we therefore write these simpler expressions for purposes of discussion .
@xmath218 m_c^0/m_t^0 & \cong & \frac{1}{9 } \epsilon^2 , \\[0.2 in ] m_{\mu}^0/m_{\tau}^0 & \cong & \epsilon \frac{\sigma}{\sigma^2 + 1 } , \\[0.2 in ] m_s^0/m_b^0 & \cong & \frac{1}{3 } \epsilon \frac{\sigma}{\sigma^2 + 1 } , \\[0.2 in ] m_u^0/m_t^0 & = & 0 , \\[0.2 in ] m_e^0/m_{\mu}^0 & \cong & \frac{1}{9 } t_l t_r , \\[0.2 in ] m_d^0/m_s^0 & \cong & t_l t_r , \\[0.2 in ] v_{cb}^0 & \cong & \frac{1}{3 } \epsilon \frac{\sigma^2}{\sigma^2 + 1 } , \\[0.2 in ] v_{us}^0 & \cong & t_l , \\[0.2 in ] v_{ub}^0 & \cong & \frac{1}{3 } t_l \epsilon \frac{1}{\sigma^2 + 1 } ( \sqrt{\sigma^2 + 1 } \frac{t_r}{t_l } e^{-i \theta } - 1 ) , \\[0.2 in ] u_{\mu 3}^0 & \equiv & \sin \theta_{\mu \tau } \cong \frac{\sigma}{\sqrt{\sigma^2 + 1 } } + o(\epsilon ) \sim 0.7 , \\[0.2 in ] u_{e2}^0 & \cong & \cos \theta_{\mu \tau } \left ( \frac{1}{3 } t_r \right ) , \\[0.2 in ] u_{e3 } & \cong & \sin \theta_{\mu \tau } \left ( \frac{1}{3 } t_r \right ) .
\\ & & \end{array}\ ] ] it might at first seem surprising that without any information about the majorana mass matrix @xmath219 of the right - handed neutrinos we are able to write down predictions for the three neutrino mixing angles .
however , if @xmath220 , as we are assuming at present , then the dirac mass matrix of the neutrinos ( @xmath92 ) has vanishing first row and column , and therefore , obviously , the same will be true of the mass matrix of the light neutrinos , which is given by the well - known see - saw " formula @xmath221 .
this means that the two mixing elements of the electron neutrino , @xmath222 and @xmath223 , get no contribution from diagonalizing @xmath224 , but come entirely from diagonalizing @xmath42 . since @xmath42 is a known matrix in our model ,
these two mixing elements are predicted . in the case of the mixing of the mu and tau neutrinos ,
@xmath31 does receive a contribution from diagonalizing @xmath224 .
however , as can be seen from the form of @xmath92 this is an effect of @xmath225 .
the contribution to @xmath31 coming from diagonalizing @xmath42 , on the other hand , is of order unity , since it arises from the large parameter @xmath51 .
thus @xmath31 is predicted , although not precisely .
since we have written fourteen quantities in terms of five parameters , there are altogether nine predictions of the model . which quantities one takes as predicted " depends on which quantities are used to determine the values of the parameters .
we will use the lepton mass ratios and the angles @xmath32 and @xmath226 for this purpose as they are the best measured .
as one can see from the third and eighth of eqs .
( 20 ) , one can get the value @xmath51 from the ratio @xmath227 .
one finds ( of course , taking the renormalization effects into account as was done in @xcite ) that numerically @xmath228 . substituting this into the expression for @xmath229
, one obtains that @xmath230 .
this is the small parameter of the model that is responsible for the hierarchy between the second and third families , and is small enough that the expressions in eqs .
( 20 ) are fairly accurate .
one can use @xmath231 and @xmath226 to determine @xmath211 and @xmath212 .
a careful fit , described later , gives @xmath232 and @xmath233 . that @xmath234 is easily understood from eq .
( 18 ) and the well - known weinberg - wilczek - zee - fritzsch result @xcite that the cabbibo angle is well acounted for by symmetric mass matrices for the first two families ; cf .
. the near equality of @xmath211 and @xmath212 is also apparant from the seventh and ninth relations of eqs .
( 20 ) and the fact that numerically @xmath235 .
the phase factor @xmath213 will be determined from the cp - violating phase @xmath209 .
the nine predictions , then , are the following . to begin with , there are the three famous predictions , * ( 1 ) * @xmath236 , * ( 2 ) * @xmath237 , and * ( 3 ) * @xmath238 .
the first is the good " prediction of minimal @xmath2 unification , and the latter two are the georgi - jarlskog relations .
these predictions are manifest from the first , third , fourth , sixth , and seventh of eqs .
it is hardly surprising that the model gives these relations , since we were guided by them in constructing the model .
the fourth prediction is * ( 4 ) * @xmath239 .
even if the @xmath240 quark is not exactly massless this relation is a very good approximation to reality . if one takes the favored value of @xmath241 mev , then , with reasonable assumptions about thresholds in doing the running up to the gut scale , one obtains @xmath242 .
this is far smaller than any other interfamily ratio of masses .
for instance , the comparable ratio for down quarks is @xmath243 , and for leptons is @xmath244 . like the previous three relations , @xmath245 is a reflection of basic group - theoretical aspects of the model .
it comes from the fact , explained above , that the @xmath196 and @xmath197 entries only appear in @xmath43 and @xmath42 .
the remaining five predictions are not simple group - theoretical relations like the foregoing , but are non - trivial quantitative predictions .
they are predictions for * ( 5 ) * @xmath165 , * ( 6 ) * @xmath246 , * ( 7 ) * @xmath31 , * ( 8) * @xmath222 , and * ( 9 ) * @xmath223 .
the prediction for @xmath165 is particularly interesting .
we see immediately that , for reasons having to do with the group - theoretic structure of the model , the ratio @xmath162 is much less than the corresponding ratio @xmath161 for the down quarks because it is of higher order in the small parameter @xmath60 .
this is a highly significant success , because the minimal yukawa terms of @xmath0 notoriously give these ratios to be equal .
moreover , the success is not merely a qualitative one .
when @xmath60 and @xmath51 are fit ( using @xmath32 and @xmath164 ) and the renormalization effects are later taken into account , it is found that @xmath165 comes out within about 5% of the experimentally preferred value , which is quite remarkable given the various experimental and theoretical uncertainties .
this success is non - trivial , because the reasoning that led to the forms of the mass matrices did not depend upon the value of @xmath165 , and hence it could have been expected that @xmath165 would come out wrong by a large factor .
another non - trivial quantitative hurdle for the model is the prediction for @xmath246 .
the eighth , ninth , and tenth relations of eqs .
( 20 ) give @xmath247 .
if we use the facts that @xmath248 and @xmath234 , this gives @xmath249 .
a careful fit gives @xmath250 in other words the model predicts that @xmath246 should lie on a certain circle in the complex plane . as can be seen from fig .
3 , the circle for @xmath251 slices neatly through the middle of the presently allowed region . again , this is a very significant success , since the reasoning that led to the forms in eq . (
15 ) was not based on the value of @xmath246 .
the prediction for the mixing of @xmath5 and @xmath6 has already been discussed .
it is one of the key successes of this model that this mixing turns out to be nearly maximal .
the fact that @xmath248 tells us that the first term in the expression for @xmath252 in eq .
( 20 ) corresponds to an angle near @xmath253 .
as we shall see in the next section , the @xmath225 corrections easily bring this down close to the maximal mixing value of @xmath254 .
the prediction of this model for the mixing of @xmath255 and @xmath5 with @xmath220 is quite interesting . from the sixth relation of eqs .
( 20 ) and the fact that @xmath234 , one sees that @xmath256 .
thus the model predicts that @xmath257 .
the factor of @xmath258 is crucial @xcite since without it one would have @xmath259 , which is about twice the value needed for the small - angle msw solution to the solar neutrino problem . since atmospheric neutrino data tells us that @xmath260 , the model gives just the correct value for the small - angle msw solution . in the future
both @xmath246 and @xmath261 will be known better and will provide a sharp test of the model .
the theoretical uncertainties in the predictions for @xmath246 and @xmath262 are estimated to be only a few percent . in discussing the @xmath263 mixing above
, we have assumed that @xmath220 . if @xmath199 does not vanish , but is around @xmath264 , corresponding to @xmath265 mev , then it turns out that both the small - angle msw solution we just discussed and bimaximal mixing are possible .
this will be discussed in detail in the next section .
finally , there is the prediction of @xmath266 mixing .
one sees from eq .
( 20 ) that there is a prediction that @xmath267 .
it is interesting that even for the bimaximal mixing case that will be discussed in the next section , the numerical value of @xmath223 is virtually unaffected .
thus this prediction is a robust " one of this model .
in analyzing the predictions of this model for @xmath7 mixing , we may make the approximation that @xmath268 .
this means that @xmath92 has vanishing first row and column .
therefore , in computing @xmath221 the first row and column of @xmath269 are irrelevant .
thus we may write @xmath269 as @xmath270 where @xmath271 , @xmath272 , and @xmath273 are in general complex .
there are consequently five real parameters ( the over all phase does not matter ) that come into the masses and mixing of @xmath5 and @xmath6 from @xmath219 .
as observed earlier , this does not prevent us from making a qualitative prediction for the mixing parameter @xmath31 , since the contribution to it from diagonalizing the mass matrix @xmath224 is only of order @xmath60 and @xmath31 comes predominantly from diagonalizing the known matrix @xmath42 .
however , in order to see if more precise predictions can be obtained , we shall look at two simple special cases : @xmath274 and @xmath275 in these cases only three parameters in @xmath219 , namely @xmath276 , @xmath277 and @xmath278 , contribute to the neutrino observables of the second and third families , since @xmath60 has appeared previously and is used as a natural scaling parameter . in the first case , @xmath279 the neutrino mixing matrix @xmath91 , now known as the mns mixing matrix @xcite , is given by @xmath280 , where @xmath281 is the unitary matrix that diagonalizes @xmath282 , and @xmath283 is the unitary matrix that diagonalizes @xmath284 . for case i , @xmath283 is given by @xmath285 where @xmath286 , and @xmath287 .
the ratio of eigenvalues of @xmath224 gives @xmath288 .
one can choose @xmath289 to be real , and one can write @xmath290.\ ] ] where @xmath291 is the phase of @xmath292 .
one readily sees from the form of the charged - lepton mass matrix @xmath42 in eq .
( 15 ) that @xmath293 is given by @xmath294 .
since @xmath248 it is evident that @xmath295 . using the best fit values of @xmath51 and @xmath60 one finds , more precisely , that @xmath296 .
altogether , then , the mixing parameter of @xmath5 and @xmath6 is given by @xmath297 if neutrino masses are hierarchical , and atmospheric neutrino oscillations are @xmath7 oscillations , then @xmath298 ev . and if one further assumes the small - angle msw solution to the solar neutrino problem , then @xmath299 ev .
thus , @xmath300 , within a factor of two or so . taking it to have the value 0.05 , and the phase @xmath301 to vanish , eq .
( 28 ) gives @xmath302 , and @xmath303 . with the same value of the neutrino mass ratio and @xmath301 taken to be @xmath254 , @xmath304 .
we see that there is excellent agreement with the experimental limits from superkamiokande if the complex phase is not too large .
but if @xmath305 , with the same mass ratio , @xmath306 .
the value of @xmath307 corresponds to @xmath308 .
since @xmath309 , for @xmath310 this gives @xmath311 , or @xmath312 .
in other words , no very great hierarchy is required in @xmath219 .
turning now to case ii , we have that @xmath313 consequently , for this case @xmath314 where @xmath315 .
the ratio of eigenvalues of @xmath224 gives @xmath316 .
if we take , @xmath307 , as before , and assume that @xmath317 is real , we have that @xmath318 .
this gives @xmath319 , @xmath320 , and @xmath321 .
again , there is good agreement with the superkamiokande results .
moreover , since this value of @xmath317 corresponds to @xmath322 , or @xmath323 , we see that in this case also no great hierarchy is needed in @xmath219 .
the foregoing discussion is all based on the assumption that the mixing with the first family is small , so that one has the small - angle msw solution to the solar neutrino problem
. this will certainly be the case if @xmath220 . as we will now see , if instead @xmath324 , as needed to have @xmath325 mev , either the small mixing of @xmath255 that we have been considering or large mixing of @xmath255 is possible , depending on the form of @xmath219 .
in the previous discussion , we set @xmath220 in which case , no matter what the form of @xmath219 , the matrix @xmath326 has vanishing first row and column , and the matrix @xmath283 that diagonalizes @xmath284 has the form of eq .
it is easy to show that the matrix @xmath281 which diagonalizes @xmath282 has the form @xmath327 where @xmath328 , @xmath212 is defined in eq .
( 17 ) , and @xmath329 is given after eq .
( 27 ) . putting these together
, one has that the total mixing matrix of the neutrinos , @xmath280 , is @xmath330 where @xmath331 .
this yields the results , already given in eq .
( 20 ) , for @xmath262 and @xmath223 .
now we will consider what happens under what is presumably the more realistic assumption that @xmath332 .
with @xmath333 , there are two basic possibilities to consider .
one possibility is that @xmath219 has a form in which its 12 , 21 , 13 , and 31 elements all vanish or are negligibly small .
if such is the case , then the previous analysis applies , and the mixing of @xmath255 is due entirely to the matrix @xmath42 .
the only effect of the parameter @xmath199 in the lepton sector is then to give @xmath334 a mass of about @xmath335 ev .
the second possibility is that @xmath219 does have significant 12 , 21 and/or 13 , 31 elements .
if this is the case then a strikingly different situation can arise @xcite , namely
bimaximal " mixing @xcite , @xcite .
we will first illustrate what happens with a simple example .
consider the following form for @xmath219 : @xmath336 we normalize @xmath337 and @xmath277 by powers of @xmath60 simply for later convenience .
the mass matrix of light neutrinos resulting from this form is @xmath338 one sees that the 2 - 3 block has vanishing determinant , so that a rotation in the 2 - 3 plane by an angle @xmath339 brings @xmath224 to the form @xmath340 this can be put in a more transparent form by a rotation in the 1 - 3 plane by an angle @xmath341 .
this angle is less than or of order @xmath342 and thus negligible , practically speaking , so @xmath343 it is clear that the 11 element , being higher order in @xmath199 , is likely to be much smaller than the 12 and 21 elements .
the condition for this to be the case is that @xmath344 .
if this very weak condition is satisfied , then the form of the matrix manifestly corresponds to the situation in which the @xmath255 and @xmath5 together form a pseudo - dirac pair .
that in turn would mean that the mixing of these two neutrinos is very close to maximal .
one sees from eq .
( 36 ) that @xmath345 , and that the splitting between @xmath346 and @xmath347 is given by @xmath348 . for the vacuum solution to the solar neutrino problem
, one has @xmath349 ev@xmath350 , so that @xmath351 .
this gives @xmath352 .
thus no great hierarchy is required in @xmath219 to get the vacuum oscillation solution .
the reason for this is that in this scheme the smallness of @xmath353 is due to the extreme smallness of the parameter @xmath199 , which is equal to the ratio @xmath354 .
it is easy to see from what has already been said that the matrix @xmath283 needed to diagonalize @xmath284 is of the form @xmath355 where we have neglected the tiny rotation @xmath356 .
the matrix @xmath281 is already given in eq .
( 31 ) , so that the full neutrino mixing matrix can be written @xmath357 where @xmath358 is given in eq .
( 32 ) , and is just the form that results in the small mixing angle ( sma ) msw case of this model . in other words , the net result of the large mxing of the first family produced by the @xmath337 entries in eq .
( 34 ) is simply to multiply the ( sma ) msw form of @xmath91 on the right by a rotation of @xmath254 in the 1 - 2 plane .
consequently , the predictions for @xmath31 and @xmath223 are essentially unaffected .
however , @xmath262 becomes @xmath359 instead of the value given in eq .
( 20 ) .
the interesting lesson is that bimaximal mixing " is easy to achieve if the large mixing of @xmath5 and @xmath6 comes from the charged lepton sector , i.e. , from diagonalizing @xmath42 , while the large mixing of @xmath255 comes from diagonalizing @xmath224 .
the simple form given in eq .
( 33 ) gives @xmath360 , and thus @xmath361 , corresponding to @xmath362 .
somewhat larger values of @xmath363 can arise from a more general form @xmath219 .
consider , for example , @xmath364 then @xmath365 here , as in eq .
( 34 ) , the 2 - 3 block has vanishing determinant .
the crucial difference is that the diagonalization of this matrix involves a rotation in the 2 - 3 plane by an angle @xmath366 . with @xmath367 , for instance , @xmath368 comes out very close to @xmath369 .
otherwise , this case is quite similar to that of eq .
in the previous sections we have presented the construction of our @xmath0 minimal higgs model in the framework of effective @xmath0 and @xmath2 operators . we now show that one can construct a technically - natural realization of this scheme by introducing sets of higgs and matter superfields with a well - defined family symmetry .
we first address the higgs sector .
the doublet - triplet splitting problem in @xmath2 , and therefore @xmath0 , arises because the colored higgs in the @xmath371 pairs of each @xmath179 must be made superheavy at the gut scale , while just one pair of higgs doublets should remain massless there and be free to develop vev s at the electroweak scale .
this problem has been addressed and solved in @xcite by the introduction of just one @xmath77 adjoint higgs field with its vev pointing in the @xmath372 direction , together with two pairs of @xmath373 spinor higgs fields , two @xmath47 higgs in the vector representation plus several higgs singlets .
we shall briefly summarize the solution , but first we note that it is necessary to introduce several more higgs fields in the vector and singlet representations in order to generate the yukawa structure for the fermion mass matrices presented earlier .
the authors of @xcite found that the higgs superpotential required to solve the doublet - triplet splitting problem could be neatly obtained from their list of higgs fields by introducing a global family symmetry group of the type @xmath370 , which can arise in a natural fashion from string theory . with this in mind
, we now list in table i all the higgs fields to be considered together with their family charge assignments .
@xmath374 ll + @xmath375 : & @xmath376 + @xmath73 : & @xmath377 + @xmath378 : & @xmath379 + @xmath47 : & @xmath380 + @xmath381 : & @xmath382 + + @xmath47 : & @xmath383 , + & @xmath384 + @xmath381 : & @xmath385 , + & @xmath386 + @xmath374 as noted in the table , in order to complete the construction of the dirac mass matrices , four more vector higgs fields and four additional higgs singlets are needed , while one higgs singlet is introduced to generate the right - handed majorana neutrino mass matrix .
it is then possible to write down explicitly the full higgs superpotential from the higgs @xmath0 and family assignments , where we have written it as the sum of five terms : @xmath387 w_a & = & tr a^4/m + m_a tr a^2\\[0.1 in ] w_{ca } & = & x(\overline{c}c)^2/m^2_c + f(x )
\\ & & + \overline{c}'(pa / m_1 + z_1)c + \overline{c}(pa / m_2 + z_2)c'\\[0.1 in ] w_{2/3 } & = & t_1 a t_2 + y ' t^2_2\\[0.1 in ] w_{h_d } & = & t_1 \overline{c}\overline{c } y'/m + \overline{t}_0 c c ' + \overline{t}_0(t_0 s + t'_0 s')\\[0.1 in ] w_r & = & \overline{t}_0 \overline{t}'_0 v_m\\ \end{array}$$\ ] ] the higgs singlets are all assumed to develop vev s at the gut scale
. we can then determine the fate of the other higgs fields from the f - flat and d - flat conditions .
in particular , @xmath388 fixes @xmath389 through the @xmath390 condition where one solution is @xmath391 , the dimopoulos - wilczek solution @xcite .
@xmath392 gives a gut - scale vev to @xmath393 and @xmath394 by the @xmath395 condition and also couples the adjoint @xmath337 to the spinors @xmath396 and @xmath397 without destabilizing the dimopoulos - wilczek solution or giving goldstone modes , as shown in @xcite .
@xmath398 gives the doublet - triplet splitting by the dimopoulos - wilczek mechanism @xcite , @xcite .
@xmath399 mixes the @xmath400 doublet in @xmath401 with those in @xmath402 ( by @xmath403 ) , and in @xmath404 and @xmath405 ( by @xmath406 ) .
we now turn to the yukawa sector and specify the matter fields and their @xmath370 charge assignments which will complete the realization of the specific model in question . for this purpose , we require three spinor fields @xmath135 , one for each light family , two vector - like pairs of @xmath407 spinors which can get superheavy , a pair of superheavy @xmath47 fields in the vector representation , and three pairs of superheavy @xmath408 fermion singlets . the complete listing is given in table ii . @xmath374 [ cols= " < , < , < " , ] @xmath374 in terms of these fermion fields and the higgs fields previously introduced , one can then spell out all the terms in the yukawa superpotential which follow from their @xmath0 and @xmath370 assignments : @xmath409 where the coupling parameters have been suppressed . to obtain the gut scale structure for the fermion mass matrix elements , all but the three chiral spinor fields in the first line of table ii
will be integrated out to yield froggatt - nielsen diagrams @xcite of the type pictured earlier .
note that the right - handed majorana matrix elements will all be generated through the majorana couplings of the @xmath410 higgs field with conjugate singlet fermions given in the last two terms of eq .
( 42 ) . in order to present a clearer description of how the gut scale mass matrices are determined from the yukawa and higgs superpotentials
, we shall illustrate the procedure for the up quark mass matrix @xmath91 .
the three massless color - triplet quark states each with charge @xmath411 are obtained as linear combinations of all such color and charge states within the fermion supermultiplets given in ( 41 ) .
in particular , the basis for the left - handed states @xmath412 and left - handed conjugate states @xmath413 can be ordered as follows : @xmath414 & \left.|3,2,\frac{1}{6}>_{10(16')},\ |3,1,\frac{2}{3}>_{\overline{10}(\overline{16})},\ |3,1,\frac{2}{3}>_{\overline{10}(\overline{16}')}\right\ } \end{array}\\[-0.2in]$$\ ] ] @xmath415 & \left.|\bar{3},1,-\frac{2}{3}>_{10(16')},\ |\bar{3},2,-\frac{1}{6}>_{\overline{10}(\overline{16})},\ |\bar{3},2,-\frac{1}{6}>_{\overline{10}(\overline{16}')}\right\ } \end{array}$$\ ] ] where the states are labeled by their representations and hypercharge according to @xmath416 .
we then form the yukawa contribution @xmath417 by using the above bases and the superpotentials to obtain the matrix @xmath418 where we have introduced the following shorthand notation : @xmath419 we can then determine from this matrix the three pairs of zero - mass eigenstates at the gut scale where the electroweak vev of @xmath401 vanishes : @xmath420/\sqrt{1 + y'^2/s''^2}\\ |u_{2l}\rangle = & |10(16_2)\rangle \\ |u_{3l}\rangle = & \left [ |10(16_3)\rangle - \frac{a}{p}|10(16)\rangle\right ] /\sqrt{1
+ a^2/p^2}\\ |u^c_{1l}\rangle = & \left [ |10(16_1)\rangle - \frac{y'}{s''}|10(16 ' ) \rangle\right]/\sqrt{1 + y'^2/s''^2}\\ |u^c_{2l}\rangle = & |10(16_2)\rangle \\ |u^c_{3l}\rangle = & \left [ |10(16_3)\rangle - \frac{a}{p}|10(16)\rangle \right]/\sqrt{1 + a^2/p^2}\\ \end{array}$$\ ] ] and where the states are now simply labeled by their @xmath2 and @xmath0 representations . finally , the dirac matrix @xmath91 for the three light quark states @xmath421 is obtained by bracketing the electroweak contributions by the appropriate @xmath422 state on the left and the @xmath423 state on the right .
the result obtained for @xmath91 has exactly the form found earlier in eq .
( 15 ) from the previous effective operator approach , with the identifications : @xmath424 here the subscript on @xmath425 signifies a factor of @xmath426 arising from the @xmath81 vev of the @xmath337 in the adjoint representation , while the subscripts on the @xmath427 terms specify the appropriate doublet vev in the @xmath47 for @xmath401 .
we have neglected the state normalization factors in ( 47 ) but will later argue that they can all be taken to be approximately unity .
the dirac matrices , @xmath428 are constructed in a similar fashion . in the case of @xmath43 and @xmath42 , the bases corresponding to eqs .
( 43 ) and ( 44 ) are enlarged by two states lying in the @xmath429 and @xmath430 representations of @xmath0 . for @xmath92 ,
on the other hand , in addition to the two above states , one must add the singlet fermion contributions from the representations @xmath431 for @xmath432 and @xmath433 for @xmath434 .
we list the zero - mass state vectors for @xmath43 , @xmath92 and @xmath42 in analogy to eqs .
( 46 ) : @xmath435/\sqrt{1 + y'^2/s''^2}\\ |d_{2l}\rangle = & |10(16_2)\rangle \\ |d_{3l}\rangle = & \left [ |10(16_3)\rangle - \frac{a}{p}|10(16)\rangle\right ] /\sqrt{1 + a^2/p^2}\\ |d^c_{1l}\rangle = & \left [ |\bar{5}(16_1)\rangle - \frac{y'}{s '' } |\bar{5}(16')\rangle\right]/\sqrt{1 + y'^2/s''^2}\\ |d^c_{2l}\rangle = & \left[|\bar{5}(16_2)\rangle - \frac{c}{y } |\bar{5}(10_2)\rangle\right]/\sqrt{1 + c^2/y^2}\\ |d^c_{3l}\rangle = & \left [ |\bar{5}(16_3)\rangle - \frac{a}{p } |\bar{5}(16)\rangle\right]/\sqrt{1 + a^2/p^2}\\ \end{array}$$\ ] ] @xmath436/\sqrt{1 + y'^2/s''^2}\\ |n_{2l}\rangle = & \left [ |\bar{5}(16_2)\rangle - \frac{c}{y } |\bar{5}(10_2)\rangle\right]/\sqrt{1 + c^2/y^2}\\ |n_{3l}\rangle = & \left [ |\bar{5}(16_3)\rangle - \frac{a}{p } |\bar{5}(16)\rangle\right]/\sqrt{1 + a^2/p^2}\\ |n^c_{1l}\rangle = & \left [ |1(16_1)\rangle - \frac{y'}{s''}|1(16 ' ) \rangle - \frac{\bar{c}_1}{x}|1^c_1\rangle\right ] /\sqrt{1 + y'^2/s''^2 + \bar{c}^2_1/x^2}\\ |n^c_{2l}\rangle = & \left [ |1(16_2)\rangle - \frac{\bar{c}_2}{p_{22 } } |1^c_2\rangle\right]/\sqrt{1 + \bar{c}^2_2/p^2_{22}}\\ |n^c_{3l}\rangle = & \left [ |1(16_3)\rangle - \frac{a}{p}|1(16)\rangle - \frac{\bar{c}_3}{z}|1^c_3\rangle\right]/\sqrt{1 + a^2/p^2 + \bar{c}^2_3/z^2}\\ \end{array}$$\ ] ] @xmath437/\sqrt{1 + y'^2/s''^2}\\ |\ell_{2l}\rangle = & \left[|\bar{5}(16_2)\rangle - \frac{c}{y } |\bar{5}(10_2)\rangle\right]/\sqrt{1 + c^2/y^2}\\ |\ell_{3l}\rangle = & \left [ |\bar{5}(16_3)\rangle - \frac{a}{p } |\bar{5}(16)\rangle\right]/\sqrt{1 + a^2/p^2}\\ |\ell^c_{1l}\rangle = & \left [ |10(16_1)\rangle - \frac{y'}{s''}|10(16 ' ) \rangle\right]/\sqrt{1 + y'^2/s''^2}\\ |\ell^c_{2l}\rangle = & |10(16_2)\rangle \\
|\ell^c_{3l}\rangle = & \left [ |10(16_3)\rangle - \frac{a}{p } |10(16)\rangle\right]/\sqrt{1 + a^2/p^2}\\ \end{array}$$\ ] ] in the above we have introduced , in analogy with eqs . ( 45 ) , the additional shorthand notation : @xmath438 the dirac matrices @xmath439 and @xmath42 are found by forming matrix elements of the electroweak symmetry breaking vev s with the appropriate basis vectors .
again these matrices have exactly the structures given in eqs .
( 15 ) , provided the state normalization factors are approximated by unity , i.e. , we assume that the zero - mass states have their large components in the chiral representations @xmath440 and @xmath441 and that all the other components are small .
we shall return to this point in the next section . in the meantime
we make the identifications @xmath442 in terms of the notation given in eqs .
( 45 ) and ( 51 ) and the following : @xmath443 the phase @xmath206 appearing in the @xmath197 term can be understood to arise from a phase in the vev for @xmath444 .
the structures of the dirac matrix elements given in eqs . (
15 ) , ( 47 ) and ( 52 ) can be understood in terms of the simple froggatt - nielsen diagrams of fig . 1 and 2
, with the higgs fields labeled as in table i. turning to the right - handed majorana mass matrix , we use the zero mass left - handed conjugate states that were obtained implicitly above for the dirac matrix @xmath92 to form the basis for @xmath219 . the right - handed majorana matrix
is then obtained by bracketing the majorana higgs @xmath410 with the appropriate zero mass conjugate neutrino states in ( 49 ) .
we obtain @xmath445 where @xmath446 the lighter two right - handed majorana masses are degenerate and have opposite cp - parity .
note that the whole right - handed majorana matrix has been generated in this simple model by one majorana vev coupling superheavy conjugate fermion singlets as shown in fig .
we conclude this section with a summary of the gut scale predictions derived from the dirac and majorana mass matrices with the particular parameters appropriate for the model in question . for convenience we give the whole set equations which are the counterpart of eqs .
( 19 ) .
@xmath447,\quad m_b^0/m_{\tau}^0 \cong 1 - \frac{2}{3 } \frac{\sigma}{\sigma^2 + 1 } \epsilon , \\
\cong & \frac{1}{3 } \epsilon \frac{\sigma}{\sigma^2 + 1}\cdot [ 1 + \frac{1}{3 } \epsilon \frac{1 - \sigma^2 - \sigma \epsilon/3 } { \sigma ( \sigma^2 + 1 ) } + \frac{1}{2 } ( t_l^2 + t_r^2 ) ] , \\
\cong & t_l t_r \cdot [ 1 - \frac{1}{3 } \epsilon\frac{\sigma^2 + 2}{\sigma ( \sigma^2 + 1 ) } - ( t_l^2 + t_r^2)\\ & \quad + ( t_l^4 + t_l^2 t_r^2 + t_r^4 ) ] , \\
\cong & \epsilon \frac{\sigma}{\sigma^2 + 1 } \cdot[1 + \epsilon \frac{1 - \sigma^2 - \sigma \epsilon}{\sigma(\sigma^2 + 1 ) } + \frac{1}{18}(t_l^2 + t_r^2 ) ] , \\
\cong & \frac{1}{9 } t_l t_r \cdot [ 1 - \epsilon \frac{\sigma^2 + 2}{\sigma(\sigma^2 + 1 ) } + \epsilon^2 \frac{\sigma^4 + 9 \sigma^2/2 + 3}{\sigma^2(\sigma^2 + 1)^2}\\ & \quad - \frac{1}{9 } ( t_l^2 + t_r^2 ) ] , \\
v_{cb}^0 \cong & \frac{1}{3 } \epsilon \frac{\sigma^2}{\sigma^2 + 1 } \cdot [ 1 + \frac{2}{3 } \epsilon \frac{1}{\sigma(\sigma^2 + 1 ) } ] , \\
v_{us}^0 \cong & t_l [ 1 -\frac{1}{2 } t_l^2 - t_r^2 + t_r^4 + \frac{5}{2 } t_l^2
t_r^2 + \frac{3}{8 } t_l^4\\ & \quad - \frac{\epsilon}{3 \sigma \sqrt{\sigma^2 + 1 } } \frac{t_r}{t_l } e^{- i \theta } ] , \\
v_{ub}^0 \cong & \frac{1}{3 } t_l \epsilon \frac{1}{\sigma^2 + 1 } [ \sqrt{\sigma^2 + 1 } \frac{t_r}{t_l } e^{-i \theta } ( 1 - \frac{1}{3 } \epsilon \frac{\sigma}{\sigma^2 + 1})\\ & \quad - ( 1 - \frac{2}{3}\epsilon \frac{\sigma}{\sigma^2 + 1 } ) ] , \\[0.1 in ] m_2 ^ 0/m_3 ^ 0 \cong & \left(\frac{\eta}{a\epsilon\sqrt{1 + \epsilon^2}}\right ) \left [ 1 + \frac{\eta}{a\epsilon^3\sqrt{1 + \epsilon^2 } } \right ] , \\[0.1 in ] m_1 ^ 0/m_3 ^ 0 \cong & \left(\frac{\eta}{a\epsilon\sqrt{1+\epsilon^2 } } \right)\left [ 1 - \frac{\eta}{2a\epsilon^3\sqrt{1+\epsilon^2 } } \right],\\[0.1 in ] u_{\mu 3}^0 \cong & - \frac{1}{\sqrt{\sigma^2 + 1}}(\sigma - \epsilon \frac{\sigma^2}{\sigma^2 + 1}),\\[0.1 in ] u_{e 2}^0 \cong & -\frac{1}{\sqrt{2}}\left [ 1 - \frac{\epsilon}{3\sigma}t_l e^{i\theta}\right.\\ & \quad + \left . \frac{1}{3\sqrt{\sigma^2 + 1}}(1 + \epsilon\sigma ) t_r\right],\\[0.1 in ] u_{e 3}^0 \cong & \frac{1}{3\sqrt{\sigma^2 + 1}}(\sigma - \epsilon ) t_r - \frac{\eta}{a\epsilon^2}\\[0.1 in ] \end{array}\ ] ] to the quark equations we have added the ratio @xmath448 which involves the ratio of @xmath449 to @xmath450 , i.e. , @xmath451 , as well as giving the leptonic mass ratios and mixings specific to the model in question .
we have elaborated above how the simple explicit model proposed gives precisely the structure for the dirac mass matrices that was obtained from the effective operator approach .
we now show that the entries are also numerically in the range to fit the quark and lepton mass and mixing data . in order to compare the gut scale predictions in eq .
( 56 ) with the low scale data , the gut scale values are first evolved from @xmath452 gev down to the susy scale which is taken to be @xmath453 gev by use of the 2-loop mssm beta functions . for this purpose ,
the mass ratios at the two scales are related by the @xmath454 running factors , while the quark mixing elements are scaled by the @xmath455 factor according to @xmath456 ( v_{ij})_{\rm susy } = & v^0_{ij}/\eta_{ij},\quad ( ij ) = ( ub),\ ( cb ) , \
( td),\ ( ts)\\ \end{array}\ ] ] the remaining evolutions to the bottom and charm quark or tau lepton running mass scales , or to the 1 gev scale in the case of the light quarks and leptons , is carried out with the 3-loop qcd and 1-loop qed renormalization group equations . here
the running factors are @xmath457 with the mass ratios scaled according to @xmath458 or similarly , with the running mass scale @xmath111 replaced by 1 gev . with @xmath459
used for the numerical evaluations for reasons that will become apparently shortly , @xmath460 and @xmath461 , the running factors are given by @xmath462 \eta_u(m_t ) = 0.4235,\quad & \eta_c(m_t ) = 0.4733,\quad & \eta_t(m_t ) = 1.0000\\ \eta_d(m_t ) = 0.4262,\quad & \eta_s(m_t ) = 0.4262,\quad & \eta_t(m_t ) = 0.6540\\ \eta_e(m_t ) = 0.9816,\quad & \eta_\mu(m_t ) = 0.4816,\quad & \eta_\tau(m_t ) = 0.9836\\ \end{array}\ ] ] finally , finite corrections must be applied to @xmath463 , @xmath464 and the evolved quark mixing matrix elements which arise from gluino and chargino loops .
the correction factors are conventionally written as @xmath465 and @xmath466 where we have used @xmath467 as explained below . using the quantities @xcite @xmath468 , and @xmath469 to determine the input parameters , one obtains for them @xmath470 , and @xmath471 . with these inputs
the remaining quark masses and mixings are obtained , to be compared with the experimental values @xcite in parentheses : @xmath472 where the finite susy loop corrections for @xmath473 and @xmath32 have been rescaled to give @xmath474 gev for @xmath475 had we chosen @xmath476 as input , on the other hand , we would find instead @xmath477 . with the numerical values in ( 61 )
we find for @xmath478 and the @xmath479 and @xmath183 angles of the unitarity triangle pictured in fig .
3 @xmath480 the upper vertex of the triangle appears to be circled precisely in the allowed experimental region .
additional predictions follow for the neutrino sector .
the effective light neutrino mass matrix of eq . ( 34 ) or ( 36 ) with @xmath481 leads to bimaximal mixing with a large angle solution for atmospheric neutrino oscillations @xcite and the `` just - so '' vacuum solution @xcite involving two pseudo - dirac neutrinos , if we set @xmath482 gev and @xmath483 .
we then find @xmath484 \multicolumn{2}{c}{m_3 = 2.4 \times 10^{14}\ { \rm gev},\ m_2 = m_1 = 3.66 \times 10^{10}\ { \rm gev}}\\[0.1 in ] \multicolumn{2}{c}{u_{e2 } = 0.733,\ u_{e3 } = 0.047,\ u_{\mu 3 } = -0.818 , \
\delta'_{cp } = -0.2^o}\\[0.1 in ] \delta m^2_{23 } = 3.0 \times 10^{-3}\ { \rm ev^2},\quad & \sin^2 2\theta_{\rm atm } = 4|u_{\mu 3}|^2|u_{\tau 3}|^2 = 0.89 \\[0.1 in ] \delta m^2_{12 } = 3.6 \times 10^{-10}\ { \rm ev^2},\quad & \sin^2 2\theta_{\rm solar } = 4|u_{e1}|^2|u_{e2}|^2 = 0.99 \\[0.1 in ] \delta m^2_{13 } = 3.0 \times 10^{-3}\ { \rm ev^2},\quad & \sin^2 2\theta_{\rm reac } = 4|u_{e3}|^2(1 - |u_{e3}|)^2 = 0.009\\ \end{array}\ ] ] the effective scale of the right - handed majorana mass contribution occurs two orders of magnitude lower than the susy gut scale of @xmath485 gev . the effective two - component reactor mixing angle given above should be observable at a future neutrino factory , whereas the present limit from the chooz experiment @xcite is approximately 0.1 for the above @xmath486 . in principle , the parameter @xmath337 appearing in @xmath219 can also be complex , but we find that in no case does the leptonic cp - violating phase , @xmath487 exceed @xmath488 in magnitude .
hence the model predicts leptonic cp - violation will be unobservable .
the vacuum solar solution depends critically on the appearance of the parameter @xmath199 in the matrix @xmath92 , corresponding to the non - zero @xmath199 entry in @xmath91 which gives the up quark a mass at the gut scale .
should we set @xmath220 , only the small - angle msw solution @xcite would be obtained for the solar neutrino oscillations .
the large angle msw solution is disfavored by the larger hierarchy , i.e. , very small @xmath337 value , required in @xmath219 .
finally we address the issue that the state normalization factors were all replaced by unity in eqs .
( 47 ) and ( 52 ) for the various matrix parameters .
this is a good approximation provided the three fermion spinor states @xmath489 provide the dominant contributions to the zero - mass quark and lepton states at the gut scale . in particular , the following ratios must be much less than unity : @xmath490 let us assume for simplicity that the electroweak couplings of @xmath491 in @xmath492 and @xmath493 in eq .
( 45 ) and of @xmath494 in @xmath495 of ( 53 ) are identical . then with @xmath496 , we find the condition @xmath497 holds . to obtain an up quark mass @xmath498 mev
, we need @xmath499 at the gut scale , which easily satisfies ( 64 ) . requiring that @xmath500 and with the result from eqs .
( 52 ) that @xmath501 leads us to the results that @xmath502 \tan \beta \simeq & \sqrt{\sigma^2 + 1}m^0_t ( cos \gamma)/m^0_b \ll m^0_t / m^0_b\\ \end{array}\ ] ] in terms of the @xmath503 mixing angle , @xmath183 , in eq .
( 11 ) . with @xmath504 , for example , @xmath505 which implies @xmath506 , a very reasonable mid - range value allowed by experiment .
for this reason , we have chosen to illustrate the numerical results above with @xmath459 .
the remaining ratios in eq .
( 64 ) can also be satisfied . for comparable @xmath507
s , @xmath508 obtained from eq .
( 55 ) requires that @xmath509 .
this ratio is consistent with the vev s needed in the higgs superpotential of eq .
( 41 ) in order to solve the doublet - triplet splitting problem .
turning now to the parameters @xmath196 and @xmath197 , we note that the near equality of their magnitudes leads to the ratio @xmath510 . moreover ,
if we assume @xmath511 , we obtain the estimate @xmath512 with the numbers obtained earlier , whereas the actual value needed is @xmath513 .
thus we have found that not only are the desired forms of the dirac ( and majorana ) matrices generated by the model of this section , but that the numerical values required for the matrix parameters are also quite reasonable .
both the largeness of the atmospheric neutrino mixing @xmath31 and the smallness of the quark mixing @xmath32 can be elegantly accounted for by the idea that the charged lepton mass matrix @xmath42 is highly asymmetric or lopsided " and that the down - quark mass matrix @xmath43 is related to the transpose of @xmath42 by an @xmath2 symmetry .
this idea was discovered independently by several groups and has since been used in numerous models of fermion masses .
remarkably , exactly such mass matrices emerged in our work from quite other considerations than neutrino masses and mixings , specifically from an attempt to construct the simplest possible realistic @xmath0 model .
advances have been made in recent years in simplifying the higgs structure of susy @xmath0 models . if one assumes the minimal set of higgs fields that can break @xmath0 down to the standard model group , the possibilities for yukawa terms for the quarks and leptons become significantly restricted .
it turns out that there is what seems to be a uniquely simple set of @xmath0 yukawa terms that gives realistic masses and mixings .
this set consists of only six effective yukawa terms ( five if @xmath514 ) which satisfactorily fits all nine masses of the quarks and charged leptons as well as the four ckm parameters .
in addition , large @xmath7 mixing emerges automatically .
moreover , in this uniquely simple model , the simplest possibilities for the majorana mass matrix @xmath219 of the right - handed neutrinos lead either to small angle msw values for the solar neutrino mixing or to vacuum oscillation values . in this paper
we have studied in detail the consequences of different forms of @xmath219 for the neutrino mixing angles and mass ratios . in the published literature
no more predictive and economical a model for quark and lepton masses than the one discussed here exists that is also consistent with present knowledge .
it is striking that in this model a single term and a single parameter ( which we call @xmath51 ) accounts for no less than four puzzling aspects of the light fermion spectrum : the largeness of @xmath31 , the smallness of @xmath32 , the smallness of @xmath171 compared to @xmath170 , and the georgi - jarlskog factor of three between @xmath515 and @xmath463 at the gut scale .
it should be emphasized that , while many satisfactory neutrino mixing ideas and also many interesting ideas for explaining the pattern of quark and charged lepton masses have been proposed , very few models exist which not only give a satisfactory account of neutrino phenomenology but are at the same time highly predictive .
we have shown that the model defined by the existence of these five ( or six ) effective yukawa terms can be realized in a complete and specific renormalizable susy @xmath0 model that is technically natural . we have presented the details of such a model , including all the higgs and quark and lepton superfields , the abelian flavor symmetries , and the transformation properties of the fields under these symmetries .
finally , we have done a quantitative comparison of the predictions of the model to experiment . in the future
this model will be rigorously testable in several ways .
the most important are ( 1 ) a relation between the real and imaginary parts of @xmath246 including a precise test of the angles of the unitarity triangle ; ( 2 ) a prediction for @xmath262 , which in the small angle msw case gives a sharp relation between the solar and atmospheric angles ; and ( 3 ) a definite prediction for @xmath223 .
the research of smb was supported in part by department of energy grant number de fg02 91 er 40626 a007 .
one of us ( cha ) thanks the fermilab theoretical physics department for its kind hospitality where much of his work was carried out .
fermilab is operated by universities research association inc . under contract with the department of energy .
albright , k.s .
babu , and s.m .
barr , phys .
lett . * 81 * , 1167 ( 1998 ) ; j. sato and t. yanagida , phys .
b * 430 * , 127 ( 1998 ) ; n. irges , s. lavignac , and p. ramond , phys .
d * 58 * , 035003 ( 1998 ) .
see also k.s .
babu and s.m .
barr , phys .
b * 381 * , 202 ( 1996 ) .
h. georgi , _ particles and fields 1974 _ , ed .
carlson , ( aip , ny , 1975 ) , p.575 ; h. fritzsch and p. minkowski , ann . phys .
* 93 * , 193 ( 1975 ) ; r. barbieri , d.v .
nanopoulos , g. morchio , and f. strocchi , phys .
b * 90 * , 91 ( 1980 ) ; j.a .
harvey , p. ramond , and d.b .
reiss , nucl .
b * 199 * , 223 ( 1982 ) .
s. dimopoulos and f. wilczek , report no . nsf - itp-82 - 07 ( 1981 ) , in _ the unity of fundamental interactions _ , proceedings of the 19th course of the international school of subnuclear physics , erice , italy , 1981 , ed .
a. zichichi ( plenum press , new york , 1983 ) ; k.s . babu and s.m .
barr , phys .
d * 48 * , 5354 ( 1993 ) .
h. fritzsch and z.z .
xing , phys .
b * 372 * , 265 ( 1996 ) ; e. torrente - lujan , phys .
b * 389 * , 557 ( 1996 ) ; v. barger , p. pakvasa , t.j .
weiler , and k. whisnant , phys .
b * 437 * , 107 ( 1998 ) ; a. baltz , a.s .
goldhaber and m. goldhaber , phys .
lett . * 58 * , 5730 ( 1998 ) ; h. georgi and s.l .
glashow , hep - ph/9808293 ; c. giunti , phys .
d * 59 * , 077301 ( 1999 ) .
y. nomura and t. yanagida , phys .
d * 59 * , 017303 ( 1998 ) ; g. altarelli and f. feruglio , phys .
b * 439 * , 112 ( 1998 ) ; m. jezabek and y. sumino , phys .
b * 440 * , 327 ( 1998 ) ; h. fritzsch and z. xing , phys . lett .
b * 440 * , 313 ( 1998 ) ; r. mohapatra and s. nussinov , phys .
b * 441 * , 299 ( 1998 ) ; m. tanimoto , phys .
d * 59 * , 017304 ( 1999 ) ; s.k .
kang and c.s .
kim , phys .
d * 59 * , 091302 ( 1999 ) ; c. jarlskog , m. matsuda , s. skadhauge , and m. tanimoto , phys .
b * 449 * , 240 ( 1999 ) ; y .- l .
wu , hep - ph 9901245 ; e. ma , hep - ph 9902392 ; a. ghosal , hep - ph 9905470 . | a full account is given of the procedure used by the authors to construct an @xmath0 supersymmetric grand unified model of the fermion mass matrices .
various features of the model which gives remarkably accurate results for the quark and lepton masses and mixings were presented earlier in separate publications . the construction of the matrices is first discussed in the framework of effective operators , from which one naturally obtains the maximal @xmath1 mixing , while the small angle or maximal mixing solutions for the solar neutrinos depend upon the nature of the majorana matrix . a set of higgs and fermion superfields is then introduced from which the higgs and yukawa superpotentials uniquely give the structure of the mass matrices previously obtained .
the right - handed majorana matrix arises from one higgs field coupling to several pairs of superheavy conjugate neutrino singlets . for the simple version considered
, 10 input parameters accurately yield the 20 masses and mixings of the quarks and leptons , and the 3 masses of the right - handed neutrinos . |
topological phase transitions in two dimensions , which are induced by the proliferation of topological defects , are naturally described via coulomb gas formulations . in the description of the kosterlitz - thouless transition of the xy model ,
the corresponding coulomb gas ( cg ) with integer charges is obtained as an effective theory for the topological defects of the model : the vortices @xcite .
similarly , a collection of dislocations whose proliferation induce the melting transition in a two dimensional elastic lattice can be described by a coulomb gas with vector charges which belong to the reciprocal lattice@xcite .
finally , most of two dimensional statistical models , such as the ising , potts and askin - teller models , can be transformed into coulomb gases @xcite ( see also the review @xcite ) . in all these cases , the transitions can be studied by renormalization in considering the screened interaction between two test coulomb charges . this interaction is logarithmic at large distance @xmath0 , and the renormalization procedure consists in neglecting the higher terms in an expansion in @xmath1 as being irrelevant @xcite .
this renormalization is thus valid for a dilute gas of charges , and is usually implemented by an expansion of the coulomb gas partition function in powers of the _ charge fugacity _ @xmath2 .
this fugacity @xmath3 is related to the _ core energy _ @xmath4 of the charges ( i.e the local energy to create a defect or their chemical potential ) .
thus this coulomb gas renormalization procedure is perturbatively controlled in the limit of large @xmath4 , or equivalently at finite temperature in the limit of small fugacity .
several topological transitions in two dimensions have been successfully described using this cg technique @xcite . soon afterwards , several authors attempted to extend these techniques to models with quenched disorder @xcite .
the randomness in the original statistical models translates into random fields in the coulomb gas formulation .
the averaged free energy ( instead of the partition function ) of this coulomb gas is then expanded , as in the pure case , in powers of the fugacity @xmath2 of the charges .
usual scaling techniques then allow to study the topological transitions in these disordered systems
. it does not seem to have been realized at that time that these approaches rely on the crucial assumption of a _ uniform fugacity _ for the charges in the sample , or equivalently of a core energy _
spatially uniform over the sample_. although this assumption is natural for pure models , it is at least questionable in the presence of a random potential @xcite .
the randomness may favor the appearance of topological defects ( the charges of the cg ) in some sites where the core energy will be effectively lower than on other sites . at high temperature ,
the randomness of the core energy @xmath4 is irrelevant as it is averaged out by thermal phase fluctuations . in that case
the large scale behaviour of the model can indeed be described by simply considering the averaged fugacity over the sample @xmath5 , as in @xcite . however , as we show here , when the temperature is lowered , the spatial inhomogeneities in the local core energy become more and more relevant . as a result
, approaches based on a single uniform fugacity are doomed to fail .
a correct detailed description of the scaling behaviour of the site - dependent fugacities ( or core energies ) becomes then necessary to determine the phase diagram and describe quantitatively the transitions in the random model .
it is the purpose of this work to define a novel renormalization method which allows this description in a consistent and perturbatively controlled manner .
we present here a detailed analysis , a shorter account of the method and results has appeared in @xcite .
although we focus here on the random phase xy model , these techniques can be applied to a much wider class of systems which can be formulated as disordered coulomb gases ( e.g. see section [ part : sinegordon ] for a random sine - gordon formulation ) . in particular , extensions to the melting of two dimensional crystals in presence of disorder are studied in @xcite and applications to the statistics of localized wave functions of electrons in random magnetic fields as well as entropic transitions in liouville theory and other models are studied in @xcite . in this paper , we focus on the 2d xy model with randomness in the phase ( see eq .
( [ xy ] ) ) , originally studied in @xcite in the context of xy magnets with random dzyaloshinskii - moriya interactions . the topological defects ( vortices ) of this model are represented by integer charges @xmath6 at sites @xmath0 .
two charges @xmath7 interact for a large separation via the usual coulomb interaction of strength @xmath8 , with the corresponding energy @xmath9 where @xmath10 is the typical size of the core region of the vortices , _
i.e _ the short distance cut - off ( as discussed in section [ part : model ] ) .
these charges also couple to a random potential @xmath11 , which arises from the randomness of the xy model , via a term @xmath12 which depends on the sign of the charge .
the crucial property of this random potential is that it has long range logarithmic correlations : @xmath13 the on - site variance being @xmath14 where @xmath15 is the system size .
rubinstein , schraiman and nelson @xcite identified the most relevant vortices as the @xmath16 charges .
they derived scaling equations for the stiffness @xmath17 , the disorder strength @xmath18 and a _ single fugacity _ for these charges @xmath19 .
as in the pure case , the nature of the phase is determined by the behaviour of the fugacity : in a quasi - ordered phase , @xmath20 decreases with the scale while in a disordered phase it increases . in this last case , vortices appear at the scale @xmath21 at which their renormalized core energy @xmath22 is about zero , which corresponds to a fugacity of order @xmath23 : @xmath24 ( see _ e.g _ @xcite ) .
interestingly the divergence of the correlation length at criticality found in @xcite , @xmath25 , is identical to the result of kosterlitz - thouless for the pure system .
the phase diagram was found to be reentrant at low temperature , and is shown as a dashed line in fig .
[ fig : phasediag - intro ] as a function of the renormalized value of @xmath26 and @xmath8 .
recently , several authors @xcite have proposed a modified phase diagram for the same model , where the reentrance of the phase transition of fig .
[ fig : phasediag - intro ] disappears .
the main point , further developed in @xcite , is an energy argument for a single vortex at zero temperature in a finite size sample , in the spirit of the kosterlitz - thouless argument for the transition in the pure model @xcite .
the energy to create a single defect has two main components ( neglecting the bare core energy ) : the elastic energy @xmath27 and the disorder energy .
indeed the defect can take advantage of the spatial variations of @xmath11 and choose the minimal value of @xmath11 in the @xmath28 sites of a sample of size @xmath15 . to estimate this minimum @xmath29 of @xmath11 over @xmath30 sites
, the above authors neglected the spatial correlations of @xmath11 .
under that hypothesis , the problem of this single particle in the random potential @xmath11 becomes _ identical _ to the random energy model ( rem ) defined and solved in the seminal work of derrida @xcite .
in particular , the averaged @xmath31 of the minimum behaves for large @xmath30 as @xmath32 , where @xmath33 is the onsite variance ( see above ) . adding the elastic contribution @xmath34 yields a creation energy @xmath35 .
hence this simple single vortex argument points towards the existence of a topological phase transition at @xmath36 where vortices proliferate .
the modified phase diagram is shown as a solid line on figure [ fig : phasediag - intro ] .
although appealing , this single vortex argument is not sufficient by itself to prove the existence of a phase transition in the system of many interacting charges , and even less so to describe the critical behaviour .
screening by mutual interactions usually plays an important role in coulomb systems , and screening of the disorder could also in principle modify the results .
the single vortex argument can only become valid asymptotically ( at infinite scaling length ) if there indeed exists a phase with finite renormalized values for @xmath8 and @xmath26 and vanishing fugacity at zero temperature .
the very existence of the xy phase in the present problem has been questionned in recent works @xcite and a further analysis is thus necessary .
thus , we can not avoid the use of a renormalization approach to determine the true phase diagram .
such a controlled rg description should correctly take into account the correlations of the random potential , neglected in the above argument .
it should allow to precisely caracterize the universal features of the critical behaviour at the disorder driven transition .
several attempts to construct a rg method were proposed @xcite prior to this work , but no agreement was found between the results of these different approaches . as we discuss below , this is largely because none of these approaches was internally consistent .
in particular , they were all based on expansions in _ dipole _ fugacities while considering single charge fugacities not only appears more natural but is essential in constructing a consistent renormalization procedure . as a result
these approaches missed the very important contribution of the fusion of environments ( see below ) .
korshunov and nattermann started by approximating the free energy of a given disordered coulomb gas by the sum of the free energies of _ independent dipoles _ ] of charges @xcite .
they found no renormalization of the disorder strength @xmath26 and a modified phase diagram in agreement with the figure [ fig : phasediag - intro ] .
an interesting replica approach was developed by scheidl @xcite , with the use of a coulomb gas of @xmath37-component charges @xmath38 , where @xmath37 is the number of replica .
scheidl noticed that , in the renormalization procedure , one must a priori take into account charges with a number @xmath39 of non zero components , contrarily to @xcite where only single component charges were considered .
his rg equations are not compatible with the one of @xcite : since the disorder strength @xmath26 is renormalized , it leads to a different phase diagram when expressed in the bare constants . upon closer examination in section [ part : other - approaches ] and interpretation within our formalism , the assumption implicitly underlying scheidl s work @xcite appears to be the gaussian nature of the renormalized local disorder , while we are able to prove that the local disorder does not remain gaussian upon coarse - graining .
although we are also able to show a posteriori that in the xy phase some results are compatible between the two approaches , the non gaussian nature of the local disorder becomes crucial to determine the critical behaviour at the transition .
finally tang @xcite developed an appealing physical picture for the single vortex freezing phenomena in this model beyond the rem approximation mentionned above , and correctly foresaw the existence of a connection @xcite with the problem of the directed polymer on the cayley tree ( dpct ) .
however , no precise coulomb gas renormalization procedure was developed following these ideas .
finding a controlled rg procedure which allows to describe this physics of freezing in a collection of interacting coulomb charges in a random environments is thus a non trivial and challenging problem . as we will see ( and as can be expected from the above single charge argument ) , it requires a quantitative description of the scale dependence of the probability distribution of the fugacities associated with the rare sites were the charges are frozen . before embarking in the ( sometimes technical ) remainder of the paper , it is useful to depict the spirit of the method and summarize the main results . to capture the physics of freezing in a gas of coulomb charges we find it crucial to start by realizing that upon increasing the size of the charges ( cutoff @xmath10 ) , their core energy aquires a random component from the potential @xmath11 . increasing @xmath10 to @xmath40 , this potential @xmath11 with logarithmic correlation splits into the rescaled logarithmically correlated potential @xmath41 at scale @xmath42 , and a random part @xmath43 uncorrelated at scales larger than @xmath42 .
this local potential is naturally incorporated in the ( renormalized ) core energy @xmath4 , which thus becomes random and site dependent : @xmath44 .
thus , upon coarse graining , the fugacities of the @xmath45 charges of the coulomb gas become random ( and dependent on the sign of the charge ) : @xmath46 .
let us stress that the definition of @xmath43 ( and thus of @xmath47 ) depends on details of the cutoff procedure .
however the rg procedure developed here depends only on the logarithmic behaviour of the correlator of @xmath41 at large distances , which is cut - off independent . as a result
, a remarkable universality will emerge at the end of the procedure , which happens to be related to universality of front solutions in non - linear equations ( see below ) .
we are now faced with the description of the scale - dependence of a coulomb gas with random fugacities , and thus _ a priori _ of the full corresponding fugacity distribution .
since we find that , at low temperature , the distribution of the local core energy does not remain gaussian under coarse graining we can not restrict a priori the renormalization study to follow a small number of variables ( such as the mean and the variance ) as illustrated in figure [ fig : distribution ] .
instead , we must study the scale dependence of the full distribution @xmath48 and especially the precise form of its tails around @xmath49 ( i.e @xmath50 ) which control the low temperature physics ( at @xmath51 defects appear where the local core energy is negative ) .
this requires new techniques in two dimensions . before developing a new rg procedure
, we have to find a correct perturbative parameter to study the disorder driven transitions at low temperature . around the pure topological transition , this parameter corresponds to the charge fugacity @xmath2 ( for the most relevant charges @xmath16 ) .
however , we expect that , upon lowering the temperature , a freezing of the defects occurs . in that case
, most sites have a large core energy ( @xmath52 ) while only a few sites with @xmath49 are favorable to the defects .
a natural choice for a perturbative parameter at low temperature in this highly non - homogeneous coulomb gas is thus the _ density of favorable sites _
depends on the sign of the charge : @xmath53 , and the perturbative parameter is the density @xmath54 of sites favorable either to @xmath55 charges or to @xmath56 charges .
one must thus follow the rg flow of the full probability distribution @xmath57 . ] which we note @xmath58 .
@xmath59 plays an analogous role here than @xmath20 in the pure case : the quasi - ordered phase is caracterized by a density @xmath59 decreasing with the scale @xmath60 , while a disordered phase , where disorder induced topological defects proliferate , corresponds to an increasing @xmath59 .
this new small parameter allows us to study in a controlled perturbative expansion the physics of the vortices at low temperature and around the transitions .
an essential contribution to the renormalization of @xmath57 , absent in previous approaches , originates from what we call the _ fusion of random environments_. the main idea is the followoing .
each random fugacities @xmath61 is associated with a region of size @xmath10 around @xmath0 . upon increasing the cut - off , the size of these regions increases , and we thus have to merge two regions distant from less than the new cutoff @xmath42 ( see fig .
[ fig : fusion - intro ] ) .
the probability distribution of the fugacities in the new region can be deduced from the one in the two merged regions by a _
fusion rule _
( see fig .
[ fig : fusion - intro ] ) .
the final renormalization of @xmath62 can be formulated into a single differential equation for the distribution function @xmath57 ( given in section [ part : replica ] ) .
most interestingly we find that the universal features at the transition of the model ( @xmath63 ) do not depend on the precise definition of these regions .
the fusion of environments corresponds to a non - linear term in the differential equation , which can thus be affected by a change of cut - off procedure .
quite remarkably , the universality class at the new transition is determined by the properties of the front solutions of this differential equation ( i.e their velocity ) , that _ do not depend _ on the precise form of this non - linear term .
hence universality in the usual rg sense finally appears from non trivial results of front propagation in non - linear equations .
this differential rg equation for @xmath62 , together with the screening equations for @xmath8 and @xmath64 , can be obtained by a systematic and perturbatively controlled rg procedure .
one formulation consists in using the replica trick to average the free energy over disorder , leading to a coulomb gas with @xmath37-component charges . in this formulation ,
the infinite set of fugacities , associated with the vector charges @xmath65 $ ] , encode the distribution function @xmath62 of the random fugacities @xmath47 .
hence , since this last distribution can not be _ a priori _ described by a finite number of moments at low temperature , we have to consider the scaling behaviour of the fugacities of all the replica charges with components @xmath66 .
the @xmath67 limit of this infinite set of rg equations is then taken by using an appropriate parametrization in terms of @xmath68 which yields the non linear rg equation for @xmath68 . the second , equivalent formulation ,
does not rely on replica .
it is constructed using a new expansion of physical quantities in the `` number of independent regions '' , introduced in this paper .
after some transformations , we reduce the non - linear rg equation that governs the scale dependence of the distribution @xmath62 to the celebrated kolmogorov - petrovskii - piscounov ( kpp ) equation ( also named the fisher equation ) .
the known results on this equation allow us to derive the form of the tails of the distribution and thus to obtain the phase diagram .
in particular , the velocity of the front solutions of the kpp equation directly determines whether the small parameter @xmath69 increases or decreases . hence using known results on the selection of this velocity ,
we obtain the phase diagram of the figure [ fig : phasediag - intro ] expressed in renormalized variables .
the critical behaviour at the transition @xmath70 , follows from the finite size corrections to the velocity corrections of the kpp front and defines a new universality class .
in particular , the correlation length @xmath71 diverges at the @xmath72 transition as @xmath73 in contrast with the kt behaviour found in @xcite .
note that recent numerical simulations @xcite of this model seem to agree with the phase diagram of fig .
[ fig : phasediag - intro ] and it would be interesting to also determine numerically the precise critical behaviour at low temperature . besides the caracterization of the new critical behaviour our work also enables to study the freezing of vortices , which occurs below a temperature @xmath74
( see figure [ fig : phasediag - intro ] ) .
this freezing corresponds to a transition for the single charge problem , whose study with our new rg technique is presented in @xcite . in particular , this renormalization approach draws a precise connection between the physics of freezing of the xy defects and the problem of branched processes ( or of random directed polymers on cayley trees ( dpct ) for a discrete version ) studied by derrida and spohn @xcite .
this connection naturally emerges from our coulomb gas rg equations via the kpp equation which has also appeared in the exact solution of the dpct problem @xcite .
it does not rely on any ad - hoc construction .
the paper is organized as follows : in section [ part : model ] , the random xy model is defined and its cg formulation is carefully derived . in particular , the relation between the continuum limit and the decomposition of the random potential @xmath11 in to a local part ( random core energy ) and a long - range potential is discussed in section [ part : disdec ] .
part [ part : replica ] describes the renormalization method of the replicated coulomb gas , and while a direct method ( without replica ) , which consists in expanding the free energy into the number of independent regions ( random environments ) is presented in section [ part : direct ] .
the rg equations are analyzed in section [ part : analysis ] using results on the propagation of kpp - like fronts .
the consequences for the determination of the phase diagram and the critical behaviour is detailed in section [ part : xyphase ] .
a formulation in terms of a sine - gordon model is given in section [ part : sinegordon ] .
the comparison with previous approaches is postponed to section [ part : other - approaches ] , and most of the technical details can be found in the appendices .
in this section and in the following we study the xy model with random phases @xcite .
we start with the model defined on the 2d square lattice by its partition function : @xmath75=\prod_{i}\int_{-\pi}^{\pi } d \theta_{i}~ e^{- \beta h[\theta , a ] } \quad \text { with } \quad h[\theta , a]=\sum_{\langle i , j \rangle } v(\theta_{i}-\theta_{j}-a_{ij})\ ] ] where the sum is over pairs of nearest neighbors ( i.e over bonds ) on the lattice and @xmath76 is the inverse temperature .
the @xmath77 are random gauge fields , independent from bond to bond , each with gaussian distribution of variance @xmath78 .
the periodic potential @xmath79 is defined for the xy model by @xmath80 where @xmath8 is the stiffness . in the limit
@xmath81 this model corresponds to the 2d `` gauge glass model '' @xcite .
for finite @xmath26 it was studied in ref .
@xcite .
the standard way to study this model is to decompose it into spin waves and vortex degrees of freedom @xmath82 = z_{sw } z_{cg}$ ] .
this decomposition can be performed exactly ( see appendix [ part : cg ] ) for the corresponding villain model defined by the potential @xmath83 technically this is the model which we study here .
it is reasonable however to expect that this villain model and the xy model should be in the same universality class ( as is the case without disorder ) .
the rg analysis contained in the following sections is consistent with this assumption .
the vortex part is described in terms of a coulomb gas with integer charges @xmath84 defined on the sites @xmath0 of the dual infinite ( square ) lattice : [ square ] @xmath85 where @xmath86 is the 2d coulomb potential with @xmath87 at large distance .
$ ] is the lattice laplacian and we denote @xmath89 where @xmath90 is the lattice spacing .
note that the energy associated with a dipole of unit charge of size @xmath91 is @xmath92 . as discussed in the appendix [ part : cg ] we can consider a neutral cg ( since @xmath93 ) with @xmath94 , and neutrality has already been used to arrive at ( [ square ] ) . in the vortex representation the random gauge fields @xmath95
translate into _ random dipoles _ ( along @xmath96 ) @xmath97 which couple to the vortex charges via the coulomb potential .
as detailed in appendix [ part : cg ] this results in a gaussian bare disorder potential @xmath98 .
an important feature of this problem is that the disorder potential seen by the charges ( the vortices ) has logarithmic _ long range correlations _
@xmath99 for @xmath100 , since @xmath101 .
note that at a given point @xmath102 where @xmath15 is the system size and that up to now , these are exact transformations . before defining the continuum limit of this coulomb gas , we note that an alternative definition to this lattice coulomb gas consists in labelling the _ non zero charges _ in ( [ square ] ) by their positions @xmath103 and their corresponding charge @xmath104 . instead of the integer field @xmath84 of ( [ square ] ) defined at each site of the lattice ,
a configuration is represented by the set @xmath105 .
we obviously have @xmath106 . with this representation
, the partition function of ( [ square ] ) reads @xmath107 where @xmath108 and @xmath109 have been defined after eq .
( [ square ] ) and the sum over the charge configurations ( primed sum ) counts each distinct neutral configuration of non zero charges only once in the definition of the configuration sum of the cg , where @xmath110 is the total number of particles of charge @xmath111 . in order to implement a renormalization procedure
one first needs to introduce a continuum version of this model . in the usual approach to 2d coulomb gas
the continuum limit is obtained by replacing the lattice coulomb interaction by the approximation @xcite @xmath112 where @xmath113 for @xmath114 and @xmath115 otherwise , and @xmath116 with @xmath117 is the euler constant .
this approximation is excellent @xcite on the lattice for @xmath118 . in the standard method the continuum cg model
is then defined by considering a gas of integer hard core charges @xmath119 at point @xmath120 of diameter @xmath90 which interact with @xmath121 . using neutrality @xmath122 and ( [ approx ] ) ,
allows to rewrite @xcite the hamiltonian ( [ square - bis ] ) as a sum of a simple logarithmic interaction @xmath123 between the hard core charges and a _ fugacity term _
@xmath124 of bare value @xmath125 where @xmath126 can be interpreted as the bare core energy for the defects of the model ( [ square ] ) . in presence of disorder
, special care has to be taken to define properly the continuum limit for the random potential , since its correlator is logarithmic . since on the lattice the correlator of the disorder @xmath127 is the same as the coulomb interaction , it is consistent to use the same @xmath128 for the disorder in the continuum model .
this immediately leads to the fact that the disorder @xmath11 must be separated in two parts , using ( [ approx ] ) : a _ long range correlated gaussian _ part @xmath41 and a _ local _
part @xmath129 : [ decomposition ] @xmath130 with no cross correlation . using this decomposition
we can now write the partition function of the continuum model : @xmath131 } \\ & & h[n , r ] = - j \sum_{i \neq j } n_i \ln \left(\frac{|{\bf r}_i-{\bf
r}_j|}{a_{o}}\right ) n_j - \sum_{i } n_i v^{>}_{{\bf r}_i } - \sum_{i } \ln y[n_i,{\bf r}_i]\end{aligned}\ ] ] where the primed configuration sum counts only once each distinct neutral charge configuration .
we have introduced the _ spatially dependent fugacity _ , of bare value , from ( [ approx],[decomposition ] ) : @xmath132 = - \gamma \beta j n^2 + \beta n v_{\bf r } \end{aligned}\ ] ] thus we find that disorder favors some regions resulting in a local fugacity for @xmath45 charges @xmath133 with a core energy @xmath134 which now varies from point to point .
thus one anticipates that problems will arise in the conventional fugacity expansion if @xmath135 varies _ strongly _ from point to point .
in addition , note that the local fugacities are different for @xmath136 charges and @xmath137 charges ( although there is still a statistical @xmath138 symmetry ) .
we have thus defined a continuum model with a particular cutoff procedure .
this is apparent , e.g. in the decomposition of the disorder that we have used .
this decomposition is cutoff dependent and we have taken care to chose the same cutoff procedure for the disorder term and the interaction .
we have chosen here a real space hard cutoff procedure , which is often used in cg studies .
other cutoff procedures can be used .
it is natural to expect , and we will partially verify that the large scale results will not depend on the particular procedure chosen .
the alert reader will have already noticed that the disorders @xmath41 and @xmath43 as defined above in ( [ decomposition ] ) are not strictly speaking physical , since the fourier transform of their respective correlators is not positive .
this is an artefact of this particular choice of a real space hard cutoff .
it is not a serious problem , and is easily cured by choosing instead a cutoff in fourier space .
the resulting decomposition of disorder is then completely legitimate .
this is further explained in appendix [ part : cutoff ] . for simplicity
, we will however proceed using the above cutoff choice , which has illustrative value , keeping in mind that the more legitimate choice detailed in appendix [ part : cutoff ] can be used instead , completely equivalently at all stages , for technical rigor .
we now turn to the renormalization of the model .
in this section we study , using replica , the renormalization group properties of the disordered coulomb gas defined by ( [ square ] ) . in the present case
the replica method is particularly convenient in order to perform the combinatorics necessary to renormalize consistently the model .
the strategy is first in ( [ part : replicacg ] ) to transform the model ( [ square ] ) into a vector coulomb gas with @xmath37-replica charges .
the fugacities of these @xmath37-vector charges will then naturally encode the distribution of the spatially dependent fugacities defined above ( see ( [ locfug ] ) ) .
the renormalization group ( rg ) equations for these fugacities are derived in ( [ part : replicarg ] ) for fixed @xmath37 . by a suitable parametrization , in ( [ part : replicalimit ] ) we then extract in the @xmath139 limit the rg equations which describe the scale dependence of the full distribution @xmath140 of the local fugacities of the topological defects in the original disordered model ( [ square ] ) , as well as the scale dependence of the stiffness @xmath8 and long wavelength disorder @xmath26 .
we start again from the model ( [ square ] ) on a square lattice .
as is well known , disordered averaged correlation functions and free energy can be obtained by studying the replicated partition function ( generating function ) @xmath141 in the limit @xmath139 . for integer @xmath37 ,
@xmath141 can then be written exactly as a cg with m - component _ vector _ charges @xmath142 , @xmath143 ( each @xmath142 is integer ) , living on the sites of the lattice ( note that @xmath10 denotes a replica index while
@xmath90 denotes the cutoff ) .
averaging over the bare disorder one obtains the partition sum of a fully coupled , translationally invariant , vector coulomb gas on a square lattice : @xmath144 with @xmath145 where @xmath146 and summation over repeated replica indices is assumed unless otherwise specified .
the next step is to approximate the lattice replica model ( [ latticereplica ] ) by a continuum coulomb gas with @xmath37-component vector charges . in the following we consider a hard core cutoff in real space .
the problem of the choice and consequences of the cutoff procedure is rather subtle here and will be discussed below . using the approximate propagator ( [ approx ] ) we obtain the continuum hamiltonian [ replicah ] @xmath147 & = & - \sum_{i \neq j } k_{ab } n^a_i g_{{\bf r}_{i}-{\bf r}_{j}}^{\text{app } } n^b_j\\ & = & - \sum_{i \neq j } k_{ab } n^a_i \ln\left(\frac{|{\bf r}_i-{\bf r}_j|}{a_{o } } \right ) n^b_j - \sum_{i } \ln y[{\bf n}_i]\end{aligned}\ ] ] where the charge @xmath148 is located in @xmath103 . in the second equality
we have used the neutrality of the coulomb gas , i.e. @xmath149 for each @xmath143 , to introduce the local fugacity @xmath65 $ ] which is a function of the whole set of components of the _ vector _ charge @xmath150 .
its bare value from ( [ approx ] ) is a simple quadratic function @xcite : @xmath151_{\text{bare } } = e^{- n_a \gamma k^{ab } n_b}\end{aligned}\ ] ] this quadratic form results from the _ gaussian nature of the bare local disorder _ and corresponds to ( [ locfug ] ) in the unreplicated version .
if this form was preserved by the rg , as was implicitly _ assumed _ in @xcite , one would be able to study the model using only two coupling constants .
however this is not the case .
as shown below , the vector charge fugacity @xmath65 $ ] has a non trivial flow under rg and does not remain purely quadratic
. the local disorder does not remain gaussian and we will have to follow its full probability distribution .
we now study the scale dependent properties of the @xmath37-component vector coulomb gas using the _ expansion in the vector charge fugacity _ @xmath65 $ ] .
although it is the natural way to study the renormalization of a vector coulomb gas , it may seem at this stage somewhat formal .
this is not so however since , as will become clear below , it turns out to correspond exactly to the expansion in the number of rare favorable regions of local disorder , which is the physically relevant ( and novel ) expansion for this model . for the replicated partition function in the continuum model
this vector fugacity expansion reads : @xmath152}\end{aligned}\ ] ] which contains fugacities via ( [ replicah ] ) and the primed sum over charge configurations counts each distinct neutral configuration only once . in presence of disorder ,
infrared divergences appear everywhere in the low temperature xy phase @xcite . to treat these divergences we now turn to the rg method .
we perform the rg analysis of the present @xmath37-component vector coulomb gas on the partition function @xmath153 .
it is a simple extension of the analysis for the scalar coulomb gas @xcite .
details are presented in the appendix [ part : rgreplica ] and we only sketch the method here . for any fixed @xmath37 it is possible to leave the form of the expansion ( [ expansion ] ) unchanged under the increase of the hard core cut - off
@xmath154 provided one defines scale dependent coupling constants @xmath155 and fugacities @xmath156 $ ] .
this corresponds to the ( one loop ) renormalizability of the @xmath37-component vector model , which we checked here to order @xmath65 ^ 2 $ ] .
the rg flow equations which determine these couplings are found as ( see appendix [ part : rgreplica ] ) : [ rgrep ] @xmath157 y[-{\bf n } ] \\
\label{rgrep2 } & & \partial_l y[{\bf n } ] = ( 2 - n^a k_{ab } n^b ) y[{\bf n } ] + c_2 \sum _ { \genfrac{}{}{0pt } { } { { \bf n}'+ { \bf n } '' = { \bf n } } { { \bf n}',{\bf n } '' \neq 0 } } y[{\bf n } ' ] y[{\bf n } '' ] \end{aligned}\ ] ] with @xmath158 , @xmath159 for our hard cut - off procedure , and the second equation ( [ rgrep2 ] ) is defined only for @xmath160 . the first equation ( [ rgrep1 ] )
comes from the annihilation of dipoles of opposite replica vector charges separated by @xmath161 .
it gives the renormalization of the interaction ( screening by small dipoles ) and of the disorder .
simple rescaling gives the first term of the second equation ( [ rgrep2 ] ) , i.e. the naive scaling dimension of @xmath65 $ ] .
the second contribution in ( [ rgrep2 ] ) comes from the possibility of _ fusion of two replica vector charges _ upon coarse graining ( see fig .
[ fusioncg2 ] ) . while such term can be neglected in the ( pure ) scalar coulomb gas ( as it yields less relevant operators ) it is usually crucial when studying most vector cg models , as e.g. in the analysis of two dimensional melting transition @xcite . indeed following too closely
the analysis for the pure xy model has led previous studies @xcite to miss the possibility of fusion and thus such a contribution . as we see in the following it has important and non trivial consequences for the physics of the low t phase and the transition . for the present disordered coulomb gas , contrarily to the conventional analysis @xcite
, one can not hope to capture the most relevant operators by restricting to single component charges ( e.g. @xmath162 ) .
this was recently emphasized by scheidl @xcite .
however , since this leads to considering multicomponent vector charges , it is thus crucial to treat properly this fusion term , which was not done previously ( e.g. in @xcite ) .
moreover , discarding this term in ( [ rgrep2 ] ) leads to a set of rg equations which is _ not consistent _ to their lowest order @xmath163 ^ 2)$ ] .
this term may _ a priori _ modify the scaling dimensions in a non trivial way in the @xmath139 limit , and it is thus crucial to study carefully its effect . before doing so in the next section ,
let us give for completeness the renormalization of the free energy density per replica , defined as @xmath164 .
it reads : @xmath165 y[-{\bf n}']\end{aligned}\ ] ] from which the flow of the free energy density can be obtained as @xmath166 .
we now have to find an analytical continuation to @xmath139 of the whole set of rg equations ( [ rgrep ] ) .
this is a priori a formidable task because ( [ rgrep ] ) are in fact , for arbitrary @xmath37 , an infinite set of coupled equations .
remarkably , in the process of performing this analytical continuation , an appealing physical interpretation in terms of probability distributions of local fugacities ( local disorder : see section [ part : disdec ] ) will emerge naturally and be our guide in the following . as a first step
, we will consider only charges with components @xmath167 in each replica .
how to incorporate higher charges ( e.g. @xmath168 ) is discussed in appendix [ highercharges ] , where it is shown that they are less relevant in the region of the phase diagram studied here .
we first remark that the possible forms of the @xmath65 $ ] are severely constrained . replica permutation symmetry , which we will assume here and is preserved by the rg , together with @xmath167 implies that @xmath65 $ ] depends only on the number @xmath169 and @xmath170 of @xmath171 components of @xmath38 .
a natural possible parametrisation of @xmath65 \equiv y[n_{+},n_{-}]$ ] consists in introducing a function of two arguments @xmath172 such that : @xmath173 = \langle z_{+}^{n_{+ } } z_{-}^{n_{- } } \rangle_\phi = \langle \prod_{a}\left[\delta_{n^{a},0}+z_{+}\delta_{n^{a},+1}+z_{-}\delta_{n^{a},-1 } \right ] \rangle_\phi \end{aligned}\ ] ] where we denote @xmath174 .
our strategy is to establish an rg equation for @xmath175 ( in the limit @xmath139 ) whose solutions @xmath176 will parametrize solutions @xmath177 $ ] of ( [ rgrep1 ] , [ rgrep2 ] ) .
let us now examine how ( [ rgrep2 ] ) can be transformed in an integro - differential equation for @xmath175 .
the technical details are given in appendix [ part : replimit ] .
the first terms in the r.h.s . of ( [ rgrep2 ] ) translate into a differential operator @xmath178 where : @xmath179 using new `` core energy '' variables @xmath180,@xmath181 such that @xmath182 , and the corresponding function @xmath183 such that @xmath184 , it can be interpreted as a _ diffusion process _ since the first term of ( [ rgrep2 ] ) now translates into @xmath185 with @xmath186 . to deal with the second term we first extend the rg equation ( [ rgrep2 ] ) so as to allow for zero charge @xmath187 , since it is easier to continue analytically unrestricted sums .
after some combinatorics ( see appendix [ part : replimit ] ) , we find that using the representation ( [ eq : ydef ] ) , ( [ rgrep2 ] ) can be rewritten completely equivalently in terms of @xmath188 as : @xmath189 where @xmath190 .
this equation describes the scale dependence ( @xmath60 ) of the function @xmath176 which parametrizes the whole set of scale dependent fugacities @xmath177 $ ] in the limit @xmath139 . up to now
the function @xmath191 has been introduced as a generating function to parametrize the fugacities @xmath65 $ ] .
it is a priori an arbitrary function and in particular @xmath192 is still undetermined . in this paragraph
we will exchange @xmath172 for a _ physical _ function @xmath193 of norm unity , which will be interpreted in the following as the probability distribution for the local fugacities @xmath194 of @xmath45 charges .
we start from the above equation ( [ loose ] ) for @xmath175 which can be simply interpreted as describing the sum of two processes . defining from the random fugacities
@xmath195 the random core energy variables @xmath196 , the first process in ( [ loose ] ) corresponds to a brownian diffusion for @xmath181 ( i.e the local disorder potential as in ( [ locfug ] ) ) together with a convection for @xmath180 .
the second process involves a fusion , with a rate @xmath197 upon increase of the cutoff , of two sets of random variables @xmath198 , @xmath199 into a single one @xmath200 according to the transformation law : @xmath201 as in a @xmath202 reaction .
the term @xmath203 in ( [ loose ] ) corresponds to a loss of two charges , while the last term corresponds to a creation of the fused one .
the term @xmath204 keeps track of the `` dead charges '' which disappear by setting them to @xmath115 ( since they decouple from the system ) .
it is in a sense only a counting device , since by construction @xmath192 is unchanged upon fusion .
we thus introduce @xmath205 restricted to @xmath206 , such that @xmath207 where @xmath208 is the total weight of non zero charges . integrating ( [ loose ] ) over @xmath209
we obtain that : @xmath210 thus in the presence of fusion it converges quickly towards @xmath211 .
since @xmath212 converges to a constant , this suggests to introduce a normalized function @xmath213 . as shown below it
is natural to interpret @xmath140 as a _ probability distribution_. from ( [ loose ] ) we find that it obeys the following rg equation : @xmath214 where @xmath215 denotes @xmath216 and the probability conserving diffusion operator @xmath217 has been defined in ( [ oo ] ) .
the limit @xmath139 of the other rg equations ( [ rgrep1 ] ) which give the renormalization of the stiffness @xmath8 and disorder strength @xmath26 is performed in appendix [ part : replimit ] using @xmath188 .
reexpressed in terms of @xmath218 they read : [ screening ] @xmath219 where we have chosen ( arbitrarily ) to keep @xmath220 fixed and renormalize @xmath8 ( only the combination @xmath221 flows ) . the above formulae ( [ rgeqp],[screening ] ) forms our complete set of rg equations .
as will become clear in the following sections , @xmath140 represents the distribution of the fugacities @xmath194 of local environments and the last term in ( [ rgeqp ] ) corresponds to fusion of environments upon coarse graining .
remarkably , once expressed in terms of @xmath218 the coefficients of the above rg equations exhibit some _
universality_. the factor of 2 in the last term in ( [ rgeqp ] ) arises from the fraction of environments @xmath222 which are fused when increasing the cutoff .
note also that the coefficient @xmath223 which naturally appears in ( [ screening ] ) is not affected by a uniform rescaling of the fugacities .
note that some features of these equations are cutoff dependent , as will be discussed in a following section .
finally , the flow of the free energy density is found to be : @xmath224 these rg equations will be studied in section [ part : analysis ] .
first we will present another renormalisation procedure , without replicas .
although it is technically more difficult to implement , it allows for a more direct physical interpretation , which in turns sheds some light on the more systematic replica method presented above .
in addition it may be more appealing to the replicaphobic section of the community .
the above method which relies on an expansion in the vector fugacities @xmath65 $ ] can be justified provided there are _ few vector charges _ in the system i.e in the dilute limit . even though the @xmath65 $ ]
may appear as formal fugacities , the above rg equations can be justified in an expansion of the exact renormalised potential @xmath225 seen by two test charges distant of @xmath91 in @xmath226 , as was emphasized by nienhuis @xcite .
it is even claimed to be exact @xcite in that limit \sum_{{\bf n } ' \neq 0 } y[{\bf n } ' ] y[-{\bf n}']$ ] with @xmath227 in the second equation .
we omitted this term as we found that it vanishes in the limit @xmath139 ] .
as will become apparent in the following section the physical meaning of this diluted limit of _ vector charges _ exactly corresponds to the limit of a small density of regions favorable to the creation of frozen defects which is the physically relevant limit in the regimes studied in this paper . to understand how cutoff dependence comes in the method used here ,
it is instructive to study the limit of zero disorder .
one can indeed check that one recover the usual results in the limit of the pure case @xmath229 .
this however , requires some careful consideration of the cutoff procedure for the vector cg representation . as discussed in the appendix
f , even in the pure case the distribution @xmath191 which parametrizes the vector fugacities solution of the cg rg equations can be non trivial . it does satisfy @xmath230 but the fugacity @xmath231 still has a non trivial distribution
@xmath232 for a generic choice of cutoff .
there is no paradox there and it is compatible with the standard korsterlitz - thouless rg equation of the pure case , as is explained in appendix f , universality being recovered at small fugacity .
in this section we introduce a method to study the model ( [ square ] ) and more generally coulomb gas with disorder without using replicas .
we start in section [ part : physics ] from the general motivation and rederive the rg equation for @xmath140 in ( [ rgeqp ] ) in a more physical way .
we also identify the small parameter which allows to study perturbatively the present problem . in section [ part : exp ]
we introduce a quantitative and systematic method to expand in this small parameter . the direct renormalization approach using this expansion
is performed in section [ part : directrg ] .
the connection with the replica method of section [ part : replica ] is presented in section [ part : connect ] .
let us first explain the spirit of the direct method and illustrate how one is led to the rg equation ( [ rgeqp ] ) , derived more quantitatively in the next sections .
we have seen in section [ part : disdec ] that the local disorder @xmath129 defines the site dependent fugacities .
we concentrate on @xmath45 charges for which these fugacity variables read ( see ( [ locfug ] ) ) @xmath233 and are quenched random variables with only short range spatial correlations .
one now studies the system under a change of cutoff @xmath234 ( coarse graining ) which includes an integration over the corresponding degrees of freedom .
we find that the coarse grained model remains of the same form as the original one , with a renormalized stiffness @xmath235 , a renormalized gaussian long range disorder strength @xmath236 and a local disorder distribution @xmath68 .
note that , although the bare local disorder @xmath129 is gaussian , it becomes _ non gaussian _ under coarse graining .
this is a novel feature of the present approach , at variance with previous attempts at renormalizing the model @xcite .
it complicates the analysis but is necessary to capture correctly the physics of the model which is driven by the rare events . the rg equations ( [ rgeqp ] , [ screening ] ) for the fugacity distribution @xmath68 of local environments ( higher charges are less important and considered later ) , for the stifness @xmath235 and for the correlated disorder strength @xmath237 can be understood from the following considerations . the correction to the fugacity distribution @xmath68 is the sum of two contributions : \(i ) _ rescaling _ : the first observation is that upon changing the cutoff , as can be seen from its correlator ( [ decomposition ] ) and is detailed below , the long range disorder @xmath238 produces an additional local disorder contribution which can be written as a renormalisation of the local charge fugacity : @xmath239 where @xmath240 is a gaussian random variable , uncorrelated from site to site and with @xmath241 .
this contribution leads to an effective diffusion and drift in the random core energy variables @xmath242 as @xmath243 and thus produces the first terms in ( [ rgeqp ] ) .
\(ii ) _ fusion of environments_. the second contribution comes from the fusion of environments . upon a change of cutoff ,
any two regions located around @xmath244 and @xmath245 with @xmath246 have to be considered as a single region in the system with the rescaled cutoff . as a consequence the two corresponding pairs of fugacities @xmath247 and @xmath248 must be combined and replaced by a single pair of effective fugacity variables @xmath249 associated with the new region at @xmath250 , as illustrated in fig . [
fig : fusion ] .
@xmath251 can be determined by estimating the relative boltzman weight @xmath252 to have a configuration with charge 1 ( which lies either in @xmath244 or @xmath245 ) versus a neutral one ( either no charges or a dipole in @xmath244,@xmath245 ) , and similarly for @xmath253 .
this gives the fusion rule : @xmath254 the corresponding correction to the distribution @xmath68 produces the last two terms of ( [ rgeqp ] ) .
finally , the rg equation for @xmath235 and @xmath237 can be obtained from the screening by small dipoles of the effective interaction and disorder between two infinitesimal test charges as described in section [ part : directrg2 ] .
several comments are in order concerning this rg procedure .
first we note that in defining local fugacity variables ( [ locz ] ) we have added an explicit spatial dependence to the part @xmath255 of the fugacity which does not distinguish between a @xmath55 and a @xmath56 charge .
this dependence is not explicitly present in the bare model formula ( [ locfug ] ) ( although it is present if an additional small disorder in the local stiffness @xmath256 is included ) but , as we can see from ( [ fusion1 ] ) it appears as soon as fusion takes place ( the fusion rule is not compatible with a uniform @xmath257 ) .
second , there are some assumptions underlying the rg procedure : technically we treat the local regions as independent from point to point , we restrict @xmath41 to be strictly gaussian , together with the usual assumptions ( e.g. short distance expansions ) of the cg renormalization .
these assumptions are consistent and amount to discard less relevant operators .
these irrelevant operators can be identified within the method using replica of section [ part : replica ] where the above assumptions appear as standard in the rg of the @xmath37-component vector cg .
for instance , the separation of the disorder into the two components @xmath258 and @xmath129 corresponds in the replica method to the natural splitting in ( [ replicah ] ) between the vector fugacity local operator @xmath65 $ ] ( originating from @xmath129 ) and the off - diagonal replica coulomb interaction @xmath259 ( from @xmath258 )
. this will be further apparent on the equivalent sine gordon representation of the problem presented in section [ part : sinegordon ] .
accordingly , the definition of the independent local regions ( and thus of the local disorder environments and of the detailed form of the distribution @xmath260 ) is clearly cutoff dependent .
so is the detailed form of the fusion rule ( [ fusion1 ] ) .
however , universality of the physical results will be recovered in a remarkable way in section [ part : analysis ] , independently of the details of the cutoff procedure .
as we will see , this is because , in the low temperature limit , the above definitions and fusion rules capture correctly ( to order @xmath261 ) the rare events which dominates the physics .
they correctly evaluate the universal part of @xmath260 ( its tails in the low temperature region where they dominate the physics ) while they also correctly describe the weak disorder regime at higher temperatures .
finally , we note that usual charge fusion between certain types of replica charges , represented on figure [ fusioncg2 ] corresponds , in the method without replica , to fusion of environments . to renormalize consistently the present model we need a method which can handle in a systematic way broad distributions of local fugacities .
we have found such a method , which we now introduce .
it is based on a _ systematic expansion of physical quantities in the number of independent points_. it generalises the conventional fugacity expansion in @xmath2 of the pure case , but is more powerful . in effect
, it amounts to a partial resummation of the conventional expansion .
it is versatile since , as we will see , it yields back the conventional expansion in the pure case or at high temperature , but is also able to handle the broad distributions which arise at low temperature .
the idea there is that only few rare regions ( favorable to the charges ) in each environment will dominate the observables and thus it becomes possible again to expand in the density of such rare regions .
the idea underlying this expansion in the number of `` independent points '' is that the site - dependent fugacities associated with distinct `` points '' can be considered as statistically independent . on a lattice
, these `` independent points '' naturally correspond to the sites of the lattice , while in the continuum model , their definition is necessarily cut - off dependent . upon coarse graining
, we will be able to use this systematic expansion to renormalize consistently the fugacity distribution associated with each point . for definiteness
, we will show how to construct the expansion in number of points on the free energy @xmath262=-\beta^{-1 } \ln z[v]$ ] .
it can be performed in a given environment , keeping the full functional dependence in the set of potentials @xmath263 .
the construction can then be easily generalized to any physical quantity , such as arbitrary powers of the free energy @xmath264^p}$ ] ( which yield an expansion of all moments averaging term by term over disorder ) or all correlation functions of the field @xmath6 which can be obtained from products of free energies @xmath265\dots f^{a_{k}}[v_{k}]}$ ] by differentiation with respect to the potentials @xmath266 . in a second stage ( next section )
we will use this expansion to justify the rg equation for the disorder distribution .
we recall that the coulomb gas model we consider is defined by its partition function @xmath267 = 1 + \sum_{p>0 } \sum'_{\{n_{1},\dots n_{p } \ } } \sum_{{\bf r}_{1}\neq \dots \neq { \bf r}_{p } } e^{\beta j \sum_{{\bf r}_{i}\neq { \bf r}_{j } } n_{i}g_{{\bf r}_{i}-{\bf r}_{j } } n_{j } + \beta \sum_{i } n_{i}v_{{\bf r}_{i}}}\ ] ] here and below , as in ( [ zcont ] ) , all formulae can be extended to the continuum model by replacing discrete sums over distincts sites @xmath268 by integrals with , _
e.g _ hard core conditions @xmath269 and introducing the uniform fugacity @xmath2 as was done in section [ part : model ] .
note that in the above expression ( [ free ] ) we do not make use of the decomposition ( [ decomposition ] ) and @xmath270 denotes the original disorder .
the expansion in the number of points of the free energy has the form : @xmath271&=&f^{(0 ) } + \sum_{{\bf r}_{1}\neq { \bf r}_{2 } } f^{(2)}_{{\bf r}_{1},{\bf r}_{2}}[v ] + \sum_{{\bf r}_{1}\neq { \bf r}_{2}\neq { \bf r}_{3 } } f^{(3)}_{{\bf r}_{1},{\bf r}_{2},{\bf r}_{3}}[v ] + \dots\end{aligned}\ ] ] where , in the disordered case , each term @xmath272 depends on the values taken by the disorder potential @xmath11 exactly and only at points @xmath273 .
the explicit construction of these terms is given both for the disordered and pure case ( @xmath274 ) in appendix [ part : appexp ] . from a practical point of view
, the explicit expression of @xmath272 is in all cases : @xmath275 } f_{{\bf r}_{i_{1}},\dots { \bf r}_{i_{l}}}[v]\end{aligned}\ ] ] where @xmath276 $ ] is the free energy of the coulomb gas defined only on the set of @xmath60 points @xmath277 ( instead of the full lattice ) and the summation is over all distinct subsets of the set @xmath278 .
the definition ( [ free ] ) is unambiguous , though subtle .
looking at the explicit expression ( [ fk ] ) as a sum over smaller subset of points one could imagine adding other terms to the @xmath272 depending on less than @xmath279 points .
this is not possible in a global way , as the whole series must add up to the free energy , and the formula ( [ fk ] ) enforces it order by order .
we refer to the appendix [ part : appexp ] for further details about the precise definition and construction .
note that the term @xmath280 vanishes here because due to neutrality one can not define a cg on a single site and that the expansion in the number of points of powers @xmath281 $ ] involves a rearrangment of the expansion of @xmath262 $ ] .
it is important to stress that each term @xmath282 of the above expansion corresponds to an infinite sum over terms of _ arbitrary high order in the conventional fugacity @xmath2 _ ( as defined in ( [ locfug ] ) ) .
this can also be seen when setting disorder to zero , where we find that each @xmath283 ( @xmath284 ) and @xmath285 ( @xmath286 ) starts as @xmath287 plus an infinite number of additional higher order terms in @xmath2 ( see below ) .
indeed the expansion ( [ free ] ) corresponds to a complete resummation of the conventional fugacity expansion usually performed in coulomb gas studies @xcite except that it is usually performed on the partition function @xmath288 while here we perform it on the free energy .
the expansion ( [ free ] ) using the free energy is the appropriate expansion when the fugacities ( or the core energies ) are random and strongly site dependent with a broad distribution ( e.g. most of the sites have @xmath289 except for a few which have @xmath290 ) .
indeed , in this case the only small parameter is the probability @xmath291 that a given point is favorable for a charge .
thus the @xmath279-th term of the expansion ( [ free ] ) is of order @xmath292 , since it is associated with @xmath279 independent points . we can thus consider ( [ free ] ) as a perturbative expansion in the small parameter @xmath291 , valid in the @xmath293 phase , which replaces the conventional expansion in @xmath2 . to compute the @xmath272 we thus have to consider a coulomb gas defined by the partition function ( [ square - recall ] ) ( in the continuum limit ) , restricted to the system of points @xmath273 .
we will in addition restrict ourselves to charges @xmath294 , as higher charges , examined later will be less relevant .
let us closely examine the lowest order terms @xmath295 and @xmath296 . for @xmath295
we need only two points and the partition function ( [ square - recall ] ) reads simply @xmath297 the three terms of the partition function ( [ z-2 ] ) corresponds respectively to no charges or a dipole in @xmath298 , and the two possible positions for the dipole .
this results in the boltzmann weigth for a dipole @xmath299 .
the first terms of ( [ free ] ) thus reads @xmath300 restriction of the expansion ( [ free ] ) to its first nonvanishing term , _
i.e _ @xmath301 corresponds to the so - called _ independent dipole approximation_. this is the approximation on which the analysis of @xcite was based .
this approximation , which neglects all interactions between dipoles , may seem at first sight to be a good enough approximation in the xy phase and at the transition .
although this is reasonable in the pure xy model , this turns out to be incorrect here : by discarding the next term @xmath302 one throws away crucial statistical correlations . indeed , when renormalizing the distribution of local fugacities , we have to take into account correlations between dipoles induced by the disorder , which arise as follows .
suppose that the site @xmath303 is favorable to the creation of a @xmath136 defect , _
i.e _ if @xmath304 and @xmath305 and @xmath306 are both favorable to creation of a @xmath137 defect ( while other neigbouring sites are unfavorable ) . within the independent dipole approximation
the dominant configuration would be to put both a dipole in @xmath307 and one on @xmath308 to take advantage of these three favorable sites .
these are the configurations ( 4 - 6 ) in fig .
[ fig : expansion ] .
however this configuration is forbidden because of the hard core constraint ( and we have restricted to @xmath45 charges , higher ones being less favorable energetically ) .
thus we need to take into account the effective correlations between dipoles which arise because of rare favorable sites .
this is done by considering the second term of the expansion @xmath302 .
furthermore , as we will see below , consistent one loop renormalization requires to examine all terms in the expansion ( [ free ] ) and how they change under coarse graining , and thus to go well beyond the independent dipole approximation .
let us derive the explicit formula for @xmath302 .
the partition function with three sites reads : @xmath309= -\beta^{-1}\ln \left ( 1+w_{{\bf r}_{1}{\bf r}_{2 } } + w_{{\bf r}_{1}{\bf r}_{3}}+w_{{\bf r}_{2}{\bf r}_{3}}\right)\ ] ] however , since _ all terms of the expansion _ of @xmath310 in terms of @xmath311 depend exactly and only on @xmath312 , we have to substract to this free energy the terms depending on less than three sites , which can be identified as @xmath313 . the final expression for the second term of the expansion of the free energy
is thus @xmath314\\ \nonumber & = & -\beta^{-1}\ln \left ( \frac{1+w_{{\bf r}_{1}{\bf r}_{2 } } + w_{{\bf r}_{1}{\bf r}_{3}}+w_{{\bf r}_{2}{\bf r}_{3 } } } { ( 1+w_{{\bf r}_{1}{\bf r}_{2}})(1+w_{{\bf r}_{1}{\bf r}_{3 } } ) ( 1+w_{{\bf r}_{2}{\bf r}_{3 } } ) } \right)\end{aligned}\ ] ] let us first notice that now the three sites component of the expansion ( [ free ] ) restricted to the first two terms is exactly @xmath315 . thus adding this second term has cured the problem coming from the configurations ( 4,5,6 ) of figure ( [ fig : expansion ] ) which , as discussed above are not allowed .
it is interesting to note that the term @xmath310 is present ( and important ) in the expansion of @xmath316 , even though there are no actual configuration with three charges in a given environment ( from neutrality ) .
indeed each term @xmath282 in the expansion ( [ free ] ) contains contributions from every even number ( less or equal to @xmath279 ) of charges .
as can be seen from above ( fig [ fig : expansion ] ) it takes into account effective _ correlations _ between the distributions of fugacities of three differents sites @xmath312 induced by the hard core constraints . by contrast , there would be no such term involving three sites ( because of neutrality ) in the similar expansion carried out on @xmath288 ( conventional fugacity expansion ) .
it was thus missed in previous studies , while in fact we show in the next section that it gives rise to a crucial contribution to the renormalization of the distribution of disordered fugacities : the _ fusion of environments_. the above defined expansion in number of independent points ( a type of cluster - virial expansion ) which we have illustrated on the first few terms , can be performed systematically to all orders . by coarse graining all terms of the expansion self - consistently , we will now obtain the renormalisation of the disorder distribution .
we now propose a renormalisation scheme based on the above expansion ( [ free ] ) . from ( [ part : physics ] ) one captures the relevant physics by defining random local charge fugacities @xmath317 at each point and splitting the disorder distribution as @xmath318=\prod_{\bf r } p(z^{+}_{\bf r } , z^{-}_{\bf r } ) p[v^>_{\bf r}]$ ] as in ( [ decomposition ] ) .
we know from the results of section [ part : replica ] that one can renormalize the model by defining only three scale dependent quantities : the full local disorder distribution @xmath319 , the stiffness @xmath235 and the second moment @xmath320 of the long range disorder @xmath321 .
we now separately obtain the renormalization of each of these quantities , in a systematic perturbative expansion in the small parameter @xmath291 deduced from the expansion in the number of points .
there are three types of contributions as follows .
as can be seen from ( [ zcont ] , [ decomposition ] ) there is an explicit dependence in the cutoff @xmath322 in the expression of the interaction energy between charges and of the correlator of the long range part of the disorder @xmath321 .
this dependence will appear in each term @xmath282 of the expansion ( [ free ] ) .
upon changing the cutoff this will results in contributions of order @xmath323 which can be absorbed by appropriate redefinitions as follows .
the interaction term changes as : @xmath324 where we have used the neutrality condition @xmath325 .
the first term is the interaction term with the rescaled cutoff while the additional term produces an additive contribution to the @xmath45 charge fugacity @xmath326 .
similarly from ( [ decomposition ] ) the correlator of @xmath238 can be rewritten as @xmath327 explicitly as the sum of a new long range disorder correlator with cutoff @xmath328 and a short range disorder correlator ( we have discarded terms of order @xmath329 ) .
thus the original problem with cutoff @xmath322 can be rewritten as one with cutoff @xmath330 with ( i ) a new gaussian long range disorder with identical form of the correlator ( [ decomposition ] ) with @xmath322 replaced by @xmath330 ( ii ) a new local ( short range ) disorder @xmath331 with @xmath332 since it is clear from ( [ change ] ) that when @xmath333 the lr disorder produces an additive _ gaussian _ contribution @xmath334 to the sr disorder .
adding the two contributions we find that the change of cutoff produces the total rescaling contribution ( [ rgrescaling ] ) .
having introduced the expansion ( [ free ] ) one can now coarse grain its continuum version by increasing the cutoff @xmath335 and integrating over the corresponding degrees of freedom . upon this increase two points @xmath336 from the @xmath279 points integral ( @xmath282 ) will be fused if they satisfy @xmath337 .
this will produce a contribution which is an integral at scale @xmath338 depending only on @xmath339 _ independent points _ and thus , by definition of the @xmath282 , produces a correction of order @xmath323 to the @xmath339 terms @xmath340 .
all these corrections can be reabsorbed into a correction to the fugacity distribution ( together with an additive change to @xmath341 the free energy contribution of all degrees of freedom which have been eliminated in the change of cutoff ) .
we now illustrate how this works on the case @xmath296 , and indicate in the appendix [ app : kpoint ] how it works for arbitrary @xmath279 .
to lowest order in @xmath342 , this correction is independent of @xmath279 and can be easily performed by considering the three points integral , using ( [ free3 ] ) with _
@xmath343 : @xmath344\end{gathered}\ ] ] where we assume latexmath:[$|{\bf r}_{1}-{\bf r}_{3}|\geq \tilde{a}_{0 } , two points @xmath307 are fused to a single point @xmath346 and one obtains a correction to @xmath347 . using the decomposition ( [ decomposition ] ) of the disorder potential into a correlated component and a local part : @xmath348 , we can rewrite the boltzmann weigth for a dipole @xmath299 , defined in ( [ z-2 ] ) , as : @xmath349 where we have used the definition @xmath350 for the local charge fugacities .
several simplifications now occur in evaluating ( [ correct1 ] ) .
we first note that in ( [ correct1 ] ) the integral is of order @xmath342 .
thus to find the leading correction of @xmath351 to order @xmath323 , we only have to consider the integrand expanded in order @xmath115 in @xmath342 . to this order , from the correlator ( [ decomposition ] ) of the disorder potential @xmath41 and using @xmath352 , we find that @xmath353 thus the weight of the fused pair can be simplified as @xmath354 .
similarly , @xmath355 and @xmath356 simplify to order @xmath115 in @xmath342 using ( [ correct2 ] ) : @xmath357 using these simplifications , we can now rewrite the first term of ( [ correct1 ] ) : we obtain for the corresponding dipole weigth : @xmath358 it is now simple to verify that the change in the disorder averaged @xmath359 > _ v$ ] originating from the @xmath360 term is correctly accounted for by following rules for the random fugacity variables .
the last two terms of ( [ correct1 ] ) produce , with the simplification ( [ correct1 - 3],[correct1 - 2 ] ) , the rules @xmath361 from ( [ correct1 - 1 ] ) , we find the rule corresponding to the first term of ( [ correct1 ] ) : @xmath362 this is true for any point @xmath306 _ different _ from @xmath303 and @xmath305 .
one can thus rewrite these fusion corrections as a correction to the ( unnormalised ) distribution of local fugacities @xmath175 : @xmath363\end{aligned}\ ] ] the coefficient @xmath364 comes from the integration over the relative position @xmath365 .
thus from ( [ fusion - direct ] ) and ( [ rgrescaling ] ) , we recover exactly the rg equation ( [ loose ] ) for the function @xmath188 .
one can similarly check that the above rule correctly account for the corrections to @xmath339 term from the @xmath279 term upon change of cutoff . to be exhaustive
, we must also consider a constant term in the expansion ( [ free ] ) of the free energy : @xmath366 which satisfies @xmath367 .
this term corresponds to the free energy sum of all degrees of freedom which have been eliminated up to scale @xmath60 .
indeed , @xmath140 ( or @xmath368 ) contains an average over all disorder configurations at smaller scales ( all environments which have been eliminated ) . a full and systematic proof can in principle be made by considering all averages of powers @xmath281 $ ] expanded in number of points and all fusions to order @xmath279 giving corrections to order @xmath339 .
we will not attempt it here , as the systematic study to all orders is much easy to perform within the ( equivalent ) replica formalism . to derive the scaling behaviour of both the stiffness @xmath8 and the strength of the correlated disorder @xmath41
, we consider the screening of the interaction and the correlation of the disorder between two infinitesimal test charges @xmath369 and @xmath370 in the sample . to implement the study of this screening within our expansion in number of independentpoints ,
we first define @xmath371 $ ] as the free energy of the disordered coulomb gas defined in ( [ zcont ] ) with the two additional test charges @xmath372 fixed in @xmath307 .
these test charges interact with the other integer charges of the coulomb gas , which screen the interaction between them , and the correlator of the disorder @xmath41 . from @xmath371
$ ] , one can define the screened interaction and disorder by [ def - screened ] @xmath373 } { \partial e_{1 } \partial e_{2 } } \right|_{e_{1}=e_{2}=0}\\ -\beta v^{r}({\bf r}_{1 } ) & = & \left .
\frac { \partial f[v , e_{1},e_{2 } ] } { \partial e_{1 } } \right|_{e_{1}=e_{2}=0}\end{aligned}\ ] ] at large distance we expect from renormalisability @xmath374 , which imply the definitions for the renormalized coupling constant and disorder strength @xmath375 where the fourier transform of the 2d laplacian @xmath376 has been defined after equation ( [ square ] ) .
these definitions can be transformed exactly using ( [ zcont ] ) into the relation @xmath377 up to now , these are only standard definitions of the renormalised stiffness @xmath378 and disorder strength @xmath379 . from this one
deduces the rg equations for @xmath8 and @xmath26 since by coarse graining ( [ jr],[sr ] ) we obtain @xcite @xmath380 where the @xmath381 factor arises because we are dealing with correlations of the charge _
density_. the novel and most tricky part is how to evaluate the right hand side in a systematic way .
our method is to expand these correlation functions in the number of independent points .
this can be done in a systematic way using ( [ free ] ) and the definitions @xmath382 } { \partial v_{{\bf 0 } } \partial v_{{\bf r } } } \\
\label{def - correl2 } \langle n_{\bf 0}\rangle \langle n_{\bf r } \rangle & = & \frac { \partial f[v ] } { \partial v_{{\bf 0 } } } \frac { \partial f[v ] } { \partial v_{{\bf r}}}\end{aligned}\ ] ] note that while the expansion of the first term ( connected correlation ) in the number of points is readily obtained from the expansion ( [ free ] ) , the expansion of the second term ( disconnected correlation ) is more subtle since we need to square the derivatives of ( [ free ] ) and rearrange it again in an expansion in the number of points .
fortunately , to the order we are working , we need only the two point term in the expansion .
thus we can restrict ourselves to terms with two points on both expansions .
the expansion of the first term ( [ def - correl1 ] ) follows straightforwardly from ( [ free ] ) : @xmath383 } { \partial v_{{\bf 0 } } \partial v_{{\bf r } } } = \sum_{{\bf r}_{1}\neq { \bf r}_{2 } } \frac { \partial^2 f^{(2)}_{{\bf r}_{1},{\bf r}_{2 } } } { \partial v_{{\bf 0 } } \partial v_{{\bf r } } } + \sum_{{\bf r}_{1}\neq { \bf r}_{2}\neq { \bf r}_{3 } } \frac { \partial^2 f^{(2)}_{{\bf r}_{1},{\bf r}_{2},{\bf r}_{3 } } } { \partial v_{{\bf 0 } } \partial v_{{\bf r}}}+ \dots \end{aligned}\ ] ] and thus from ( [ f-2 ] ) one finds : @xmath384 } { \partial v_{{\bf 0 } } \partial v_{{\bf r}}}|_{2 points } & & = \frac { \partial^2 f^{(2)}_{{\bf 0},{\bf r } } } { \partial v_{{\bf 0 } } \partial v_{{\bf r } } } \\ & & = \frac{-\beta^{-1}}{1+w_{{\bf 0},{\bf r } } } \frac { \partial^2 w_{{\bf 0},{\bf r } } } { \partial v_{{\bf 0 } } \partial v_{{\bf r } } } + \frac{\beta^{-1}}{(1+w_{{\bf 0},{\bf r}})^{2 } } \frac { \partial w_{{\bf 0},{\bf r } } } { \partial v_{{\bf 0 } } } \frac { \partial w_{{\bf 0},{\bf r } } } { \partial v_{{\bf r}}}\end{aligned}\ ] ] since we can keep only terms depending of the two points ( @xmath385 and @xmath386 ) in the last equality ( the other contributions vanish ) .
similarly one finds that : @xmath387 } { \partial v_{{\bf 0 } } } \frac { \partial f[v ] } { \partial v_{{\bf r}}}|_{2 points } & & = \left(\sum_{{\bf r}_{1}\neq { \bf r}_{2 } } \frac { \partial f^{(2)}_{{\bf r}_{1},{\bf r}_{2 } } } { \partial v_{{\bf 0 } } } + \dots \right ) \left . \left(\sum_{{\bf r}_{3}\neq { \bf r}_{4 } } \frac { \partial f^{(2)}_{{\bf r}_{3},{\bf r}_{4 } } } { \partial v_{{\bf r } } } + \dots \right)\right|_{2 points } \\ \nonumber & & = \frac { \partial f^{(2)}_{{\bf 0},{\bf r } } } { \partial v_{{\bf 0 } } } \frac { \partial f^{(2)}_{{\bf 0},{\bf r } } } { \partial v_{{\bf r } } } = \frac{\beta^2}{(1+w_{{\bf 0},{\bf r}})^{2 } } \frac { \partial w_{{\bf 0},{\bf r } } } { \partial v_{{\bf 0 } } } \frac { \partial w_{{\bf 0},{\bf r } } } { \partial v_{{\bf r } } } \label{expcorr}\end{aligned}\ ] ] where the dipole weights @xmath388 have been defined in ( [ z-2 ] ) these expressions further simplify since it is sufficient to work to lowest order in @xmath342 , i.e to order @xmath115 in @xmath342 .
we can use that @xmath389 since we need only to evaluate the correlation at @xmath390 .
this yields that @xmath391 .
one can also replace the derivatives with respect to the full disorder @xmath392 by derivatives with respect to the local disorder @xmath393 ( see ( [ decomposition ] ) ) and thus we find : @xmath394 by using this expansion into ( [ def - correl1],[def - correl2 ] ) we find the screening rg equations ( [ screening ] ) . thus the expansion in the number of points has allowed us to derive consistent rg equations _ without using replicas_. note the difference with the previous unsuccesful attempt in @xcite to derive rg equations for the stiffness and disorder without replicas .
the authors of @xcite were working in the independent dipole approximation .
thus they computed the correlation functions as in the first line of ( [ expcorr ] ) except that they kept the full expression @xmath395 .
this led to intractable expressions involving up to _ four _ independent points .
the beauty of the present method is that we know that we can keep , in a consistent way to this order , only the terms in this expression which involve two points .
we have thus developed two rg procedures , one using replicas based on an expansion in the _ vector fugacity _
@xmath65 $ ] which may have seemed somewhat formal , the other one , direct without replicas , as an expansion in the number of favorable regions @xmath291 ( at low temperature , while at higher temperature it smoothly crosses over into the usual expansion in the uniform fugacity @xmath396 ) .
in fact these two approaches are _
equivalent_. indeed , we have checked using some combinatorics detailed in appendix [ part : connection - replica ] , that the fugacity expansion in @xmath65 $ ] ( [ expansion ] , [ replicah ] ) is _ identical _ term by term , in the limit @xmath139 to the expansion in number of points of the free energy ( [ free ] ) .
thus , working with one or the other is equivalent .. ] it also shows that the limit of diluted _ vector charges _ physically corresponds to the limit of rare favorable regions .
[ part : kpp ] in this section we analyse the rg equations derived in the previous sections and we obtain the phase diagram and the critical behaviour of the xy model with random phases .
our task now is to study the closed set of rg equations ( [ rgeqp],[screening ] ) for the scale dependent distribution of fugacities @xmath397 , the long range disorder strength @xmath237 and the stiffness @xmath235 .
since we study below the zero temperature limit , we have chosen to keep @xmath220 fixed and renormalize @xmath8 ( only the combination @xmath221 flows ) .
we started by studying the set ( [ rgeqp],[screening ] ) numerically . although we will not reproduce here the details we describe the main results .
they confirmed the overall physical picture and led us to introduce an approximation , described below , which allows for an analytical solution .
we found that at low @xmath220 and for @xmath26 smaller than some critical value @xmath398 an xy phase exists ( as in fig .
[ fig : phasediag - intro ] ) , where both the stiffness @xmath8 and @xmath26 converge to finite non zero values at large @xmath60 : @xmath399 in general , at low @xmath220 we found that , starting at scale @xmath400 from ( [ locfug ] ) with a small initial averaged fugacity ( and small local disorder ) the distribution @xmath397 becomes broad and develops power law tails in the variables @xmath401 , @xmath402 which quickly extend up to fugacities @xmath403 . however , in the xy phase ( [ xydef ] ) we found that the typical @xmath404 goes to zero and that the concentration of rare favorable regions , after an initial increase , ends up decreasing towards zero at large @xmath60 .
it is useful to define the _ single fugacity distribution _ : @xmath405 which does not satisfy a closed rg equation .
then we find in the xy phase that @xmath406 decreases at large @xmath60 ( or equivalently @xmath407 where @xmath408 is some arbitrary threshold ) .
thus the probability that either @xmath401 or @xmath402 is of order @xmath23 decreases .
on the other hand , for @xmath409 we find that @xmath406 ends up increasing at large scale and the whole distribution @xmath410 ends up drifting towards increasing @xmath96 values .
this corresponds to the disordered phase where @xmath235 vanishes at large scale ( @xmath411 ) .
the most interesting flow occurs in the critical regime near the transition .
there , if one interprets the fraction of favorable regions @xmath406 as the perturbative parameter , the structure of the flow near @xmath412 is reminiscent of the rg flow a la kosterliz - thouless with a separatrix , in the plane @xmath413 between the xy and the disordered phase .
accordingly , exactly at the transition , i.e on the separatrix ( which , to be accurate , is a critical manifold in the space of distributions and couplings @xmath414 ) the distribution @xmath140 becomes very broad and develops a fixed shape , with @xmath406 slowly decreasing to zero ( which makes the critical regime perturbative in @xmath415 ) . a schematic representation of the distribution is given in fig .
[ fig : distrib-2d ] .
let us close this paragraph by noting that at higher @xmath220 , by contrast , we found that the distribution remains peaked and that the overall picture becomes more consistent with considering a uniform average fugacity ( as in @xcite ) .
the xy transition then occurs when this uniform average fugacity ceases to flow to zero at large scale . to quantify these ( mostly numerical ) observations
, we now turn to the single fugacity approximation .
we now argue that , given the structure of the rg flow observed numerically , we can , with no loss of accuracy in all the regimes of interest ( i.e within and near the boundaries of the xy phase ) approximate the full rg equations ( [ rgeqp ] ) for @xmath68 by a simpler equation for the single fugacity distribution @xmath410 , and similarly for ( [ screening ] ) .
let us first focus on the low @xmath220 regime , where the distribution @xmath397 is broad and the physics is dominated by rare favorable regions .
for such a broad distribution the parameter which allows to organize perturbation theory is : @xmath416 where we distinguish symbolically the rare configurations @xmath417 from the typical ones where @xmath418 .
the first observation is that @xmath401 and @xmath402 are even more rarely simultaneously @xmath419 .
we note symbolically : @xmath420 indeed such configurations , absent from the start , are generated from the fusion term in ( [ rgeqp ] ) . to generate from the fusion rule @xmath421 a configuration where both @xmath422 one clearly needs at least that either @xmath423 or @xmath424 , which is of probability of order @xmath425 ( symbolically one can write @xmath426 ) . thus , and this can be further checked , the estimate ( [ p11 ] ) is valid ( the diffusion term does not change it ) .
one can now inspect the corrections to the stiffness @xmath8 and to the long range disorder @xmath26 in ( [ screening ] ) which come mainly from configurations where either @xmath423 or @xmath424 .
this leading contribution is thus of order @xmath427 .
other fugacity configurations , e.g. with both @xmath428 contribute in ( [ screening ] ) but , from the above estimate ( [ p11 ] ) , their probability is @xmath429 and are thus subleading in an expansion in @xmath415 . to summarize ,
one can see schematically the rg equations ( [ screening ] ) as a correction of leading order @xmath430 to @xmath235 and @xmath237 and the rg equation ( [ rgeqp ] ) as a correction to @xmath415 of order @xmath415 ( diffusion term ) and @xmath430 ( fusion term ) .
guided by these observations , we now approximate the full equation ( [ rgeqp ] ) by a rg equation for a function of a single variable which should correctly describe the behaviour of @xmath431 around @xmath432 . this approximation amounts to neglect the denominators in the fusion rule of ( [ rgeqp ] ) .
indeed , in the spirit of our previous estimates , this denominator contributes only when @xmath423 ( or @xmath424 ) and then yields a contribution of order @xmath430 but which corrects only @xmath433 since this fusion produces both a @xmath434 .
this corresponds to a correction of order @xmath435 to @xmath415 ( which we neglect , since the leading order of the correction to @xmath415 coming from the fusion term is @xmath425 ) .
this can be seen e.g. since @xmath433 itself in turns contributes to @xmath415 to order @xmath425 ( through integration over one of the two variables ) , or by more detailed arguments , not reproduced here .
neglecting the denominators gives the equation for the probability distribution @xmath436 this equation can now be explicitly integrated over one variable , say @xmath253 and yields a closed equation ) admits solutions of the form @xmath437 .
] for the single fugacity distribution @xmath438 defined in ( [ singledef ] ) : @xmath439p(z ) - 2 p(z ) \\ \nonumber & & + 2 \int_{z',z''>0}dz'dz '' \delta ( z - ( z ' + z '' ) ) p ( z ' ) p ( z'')\end{aligned}\ ] ] we can now obtain the corresponding equations for the corrections to @xmath8 and @xmath26 . in ( [ screening ] )
one keeps only the configurations corresponding to either @xmath440 or @xmath441 but discard the much less probable configurations where all four fugacities are @xmath419 .
the expressions then nicely factor out in terms of @xmath410 and one obtains : [ screening2 ] @xmath442 where the additional factor of @xmath443 counts the equivalent integral with @xmath136 and @xmath137 exchanged . we now summarize the obtained closed set of rg equations in terms of the variable @xmath180 defined as : @xmath444 of distribution @xmath445 .
physically the random variable @xmath180 can be interpreted as the random local core energy @xmath446 .
one has : [ rg - eq-1d - u ] @xmath447 and thus the corrections to @xmath8 and @xmath26 involve only the convolution @xmath448 .
we have thus justified , in the low temperature regime , that these approximate rg equations should describe the tails of the fugacity distribution exactly to the order @xmath430 in the expansion in @xmath415 to which we are working . indeed , since we have studied the rg in the previous sections only to order @xmath65 ^ 2 $ ] ( in the replica formulation , corresponding to @xmath430 in the rare events formulation ) it probably does not make sense at this stage to try to be more accurate .. are correct , only an exhaustive analysis of the two dimensional front solutions of ( [ rgeqp ] ) , clearly a complex task which goes beyond this paper , could confirm the exactness of the approximation and the validity of the arguments based on organizing the perturbation theory using a single small parameter @xmath415 . ] finally , let us note that although we have focused until now on the low @xmath220 regime it is rather obvious that the approximation of discarding the denominators will be even more valid at higher temperature since in the conventional fugacity expansion ( in a uniform averaged fugacity ) these terms corresponds to @xmath449 terms while we work to @xmath450 .
so these new rg equations interpolate all limits correctly .
one can now easily see on the form ( [ rg - eq-1d - u ] ) that the rg equations admit a well defined @xmath451 limit . in the equation ( [ toy ] ) , the fusion rule @xmath452 now becomes a max rule @xmath453 and ( [ toy ] ) transforms as @xmath454 similarly , the function of @xmath455 in ( [ toy2],[toy3 ] ) become respectively a delta and a theta function when @xmath451 , so that in that limit one has : [ rg1d-2 ] @xmath456 note that the last integral exactly evaluates the probability to find two local regions with a total negative core energy , energetically favorable for a dipole , in agreement with the physical picture valid at low @xmath220 .
we thus obtain a close set of equations ( [ rg1d-1],[rg1d-2 ] ) which describes the scaling behaviour of the system at zero temperature .
the equation for the distribution ( [ rg1d-1 ] ) can be conveniently simplified using the parametrization @xmath457 with @xmath458 .
the function @xmath459 then satisfies @xmath460 in the case where @xmath26 and @xmath8 are @xmath60-independent , this equation is known as the kolmogorov - fisher equation and we will recall some results on this equation in the next section . before turning to its study , let us notice that the screening equations ( [ rg1d-2 ] ) can be rewritten using this new parametrization simply substituting @xmath461 .
although ( [ rg - eq-1d - u ] ) at finite temperature @xmath463 could be in principle studied directly , it is much more convenient to introduce the generalization of the @xmath51 parametrization ( [ g - t0 ] ) as @xmath464 with @xmath465 . in the limit @xmath466 , this function reduces to the previous one ( [ g - t0 ] ) . using this new function and
variable the integral equation ( [ toy ] ) can again be transformed exactly into a simpler differential equation which , interestingly , remains exactly the same as in the @xmath51 case : @xmath467
only the initial condition explicitly depends on temperature ( see below ) .
thus solving the kpp equation ( [ kolmogorov ] ) allows us in principle to obtain the scale dependent fugacity distribution @xmath468 at any @xmath220 .
however the relation between this function @xmath459 and the distribution @xmath469 becomes much more involved at @xmath463 . as a result , and contrarily to the @xmath51 case
, the screening equations do not admit a simple expression in term of @xmath459 . for any fixed @xmath60 one can reconstruct the integer moments @xmath470 of the distribution @xmath469 by simply expanding the generating function in powers of @xmath471 since : @xmath472 taking back this expansion in the kpp equation ( [ kolmogorov ] ) and identifying the coefficient of the exponentials @xmath473 , we find the exact rg equations for the moments @xmath474 of the distribution @xmath469 : @xmath475p_{l } ( z')p_{l } ( z'')\end{aligned}\ ] ] starting from a reasonable initial distribution with finite moments , the moments remain finite for finite @xmath60 , but , as we now discuss , increase quickly as @xmath476 .
let us examine the scaling dimension of the moments , neglecting for now the bilinear ( fusion ) term .
we find that for fixed @xmath220 and @xmath8 , each successive moment @xmath474 diverges with the scale @xmath60 when the long range disorder @xmath26 becomes larger than a critical disorder @xmath477 ( see fig .
[ fig : y - moments ] ) ( these values will be slightly renormalized by the screening equations but the conclusion will be similar ) . putting back
the fusion term simply implies that the moments with @xmath478 also diverge when the first one does , as indicated in the fig .
[ fig : y - moments ] . the divergence of the first moment @xmath479 which can be identified with the uniform fugacity @xmath2 of rubinstein et al .
yields the ( incorrect ) reentrant phase diagram of @xcite where the xy phase is destroyed above the line @xmath480 .
indeed it is already clear that the uniform fugacity approximation can not work since we now see that higher moments diverge for even smaller disorder strengths . thus even within the xy phase this result for the moments
show that the distribution @xmath469 rapidly becomes broader and broader .
atypical sites where @xmath96 is large appear and dominate the behaviour of the higher moments .
thus it becomes meaningless to study the scale dependence of the integer moments , but rather we must now consider the whole probability distribution @xmath469 and in particular understand its tail .
this can be achieved with the help of the known solutions of the kpp equation which we briefly review in the next section , before coming back to describing the scaling behaviour of @xmath468 .
it is interesting to note at this stage that the above kpp equation also arises in the problem of the directed polymer with quenched disorder on the cayley tree ( dpct ) @xcite .
there the variable @xmath60 corresponds to the number of generations and @xmath481 to the distribution of free energy @xmath482 .
thus we have demonstrated in an explicit and non trivial way , i.e at the level of the rg equations , that there are close connections between the two problems , which both exhibit a similar freezing transition .
there are also notable differences between the two problems .
for instance , while the diffusion coefficient @xmath483 is constant in the dpct studied in @xcite , in the disordered cg , @xmath484 depends on the scale and , in a self consistent manner , on the solution of the kpp equation itself , via the equations ( [ screening2 ] ) which describes the physics of screening , absent in the dpct . as a result ,
additional phase transitions exist here as will be detailed in section [ part : xyphase ] .
we recall in this section some known facts on the kolmogorov - petrovskii - piscounov ( kpp ) equation ( also known as the kolmogorov - fisher equation ) which we will need in the following sections .
the equation reads , in a general form : @xmath485 where the diffusion coefficient is constant , the function @xmath486 satisfies @xmath487 , @xmath488 positive between @xmath115 and @xmath23 and @xmath489 between @xmath115 and @xmath23 .
the usual case corresponds to @xmath490 .
this equation has been applied to a wide range of problems , from chemistry to hydrodynamic instabilities or to the propagation of the meissner phase into the normal phase in a superconductor @xcite .
it is the prototype of equations describing the diffusive invasion of an unstable state by a stable one .
this can be seen by writing it as a landau equation @xmath491 whose free energy @xmath492 takes by construction its local maximum in @xmath493 ( the unstable state ) and its minimum in @xmath494 ( stable state ) .
one usually chooses an initial condition @xmath495 monotonously decreasing from @xmath496 to @xmath497 . for a large class of initial conditions ,
the solutions of the kpp equations are known to converge uniformly towards traveling waves solutions of the form : @xmath498 with @xmath499 and @xmath500 .
however the question of the determination , given an initial condition , of the asymptotic traveling wave @xmath501 and its velocity @xmath502 has been largely debated for kpp equations or for similar more complex non linear equations , and is still of current interest .
it is known as the front _ velocity selection problem_. it can be illustrated as follows .
a family of possible front solutions exists , parametrized by the velocity @xmath503 and noted @xmath504 , as can be seen by substituting ( [ front ] ) in ( [ kpp - general ] ) .
constraints exist for the velocity .
indeed , one can linearize the kpp equation in the region ahead of the front for large positive @xmath505 where @xmath501 is very small . as discussed below it
is in fact this region which determines the velocity and is universal . in this region
one has : @xmath506 and thus @xmath501 is a superposition of two exponentials @xmath507 with ( see figure [ fig : vitesse ] ) @xmath508 with @xmath509 .
the large @xmath510 behaviour is dominated by the smaller @xmath511 and thus correspond to the left branch of the curve @xmath512 in figure [ fig : vitesse ] when @xmath513 . in the marginal case where @xmath514 and @xmath515 , the two eigenvalues are degenerate and the front thus has the asymptotic behaviour for @xmath516 :
@xmath517 let us now give the known results for the selection of the asymptotic front among the family of possible @xmath518 .
\(i ) _ velocity selection .
_ although for more complex equations one relies on stability analysis and a marginal stability criterion
@xcite , in the case of the kpp equation a rigorous result is available .
a theorem due to bramson @xcite shows that the asymptotic traveling wave is determined by the behaviour at @xmath519 of the initial condition @xmath495 in the following manner .
if @xmath495 decays fast enough , as @xmath520 for @xmath516 with @xmath521 ( or faster ) theorem b of @xcite states that @xmath522 uniformly converges towards the traveling wave solution @xmath523 of velocity @xmath524 .
if @xmath495 decays slower , as @xmath520 with @xmath525 then @xmath522 uniformly converges towards @xmath526 of velocity @xmath512 continuously depending on @xmath511 and given by ( [ velocity ] ) . the asymptotic velocity is thus given by @xmath527 for steep enough initial condition @xmath528 and @xmath529 otherwise .
moreover , the leading corrections to the velocity are also given by the theorem of bramson and are _ independent of the function _
@xmath530 in ( [ kpp - general ] ) .
the corresponding position of the front @xmath531 is given by @xcite [ velocity - corrections ] @xmath532 note that cases ( [ 3demi ] ) and ( [ 1demi ] ) differ only by the velocity corrections but not by the asymptotic front shape which is @xmath533 in both case .
let us emphasize again the remarkable universality which arises in this problem .
clearly the detailed shape of the asymptotic front depends on the detailed form of the non linear term @xmath530 in the kpp equation .
however , the selection itself , the selected velocity , its corrections , and the tail of the selected front function @xmath501 are all independent of @xmath530 .
it is also natural physically that it is the region in which the front penetrates ( region ahead of the front @xmath516 ) which determines the selection .
this universality has been explored further @xcite and it has even been shown that the next leading corrections to @xmath534 are also universal .
finally the marginal case which interpolates between ( [ 3demi ] ) and ( [ 1demi ] ) has been also explored with @xmath535 one has : @xmath536 .
the asymptotic front shape is the same for all these cases , but not the detailed finite @xmath60 shape as we now discuss .
\(ii ) _ shape of the front for finite @xmath60 _ the problem has also been studied for finite but large @xmath60 @xcite .
we start with the case where @xmath495 decays fast for @xmath516 as @xmath520 with @xmath537 , most relevant for the following sections
. then one must distinguish two regions in the traveling wave solution as illustrated in figure [ fig : front - shape ] .
the central region is the `` bulk '' or `` interior front '' for @xmath538 fixed and finite .
there , the shape of the front corresponds to its asymptotic form @xmath539 and the center moves as @xmath540 with @xmath541 .
far ahead of this `` interior front '' , @xmath542 , @xmath522 still decays faster that in the asymptotic solution @xmath543 . thus there
exist an intermediate region ( which we call the `` intermediate front region '' ) which matches between the interior region and the region at infinity . in this intermediate region which corresponds to @xmath544
one must take into account diffusion , and one can solve the linearized kpp equation without assuming a front solution , but assuming a scaling form .
it is found @xcite that in this scaling `` far front '' region @xmath544 , the solution @xmath522 behaves as with @xmath535 with a scaling function noted @xmath545 in @xcite ( for @xmath546 one gets the above result , but the form for general @xmath10 is more complicated ) . ]
: @xmath547 note that later on we will need to distinguish a region even further ahead of the front for @xmath548 .
let us close this section by returning to the question of the dependence of the rg equation of the random xy model in the cutoff procedure .
we can show that the cutoff procedure will determine the function @xmath530 .
let us consider for instance again the limit of zero disorder , discussed in appendix f. since in that case @xmath549 one should have : @xmath550 with in that case @xmath551 and @xmath218 characterizes the `` disorder '' associated with the choice of cutoff procedure .
an interesting choice corresponds to @xmath552 where @xmath553 satisfies the kosterlitz thouless rg equation @xmath554 .
it is easy to see that in that case @xmath522 satisfies ( [ zerodis ] ) with @xmath555 .
this is further discussed in @xcite
we now use the solutions of the kpp equation discussed in the previous section to determine the phase diagram of the xy model with random phase shift .
let us first look for the xy phase where we expect that @xmath235 and @xmath556 reach limits , respectively @xmath557 and @xmath558 at large @xmath60 .
thus in the xy phase at large @xmath60 the kpp equation with constant @xmath559 can be used .
precise behaviour near the phase transition away from the xy phase , as well as intermediate scale dependence inside the xy phase _ a priori _ requires taking into account the @xmath60 dependence of @xmath484 which will be done in the following sections . note that @xmath60 dependence of @xmath235 itself results only in a shift that can always be trivially taken into account .
first we note that the phase diagram will be entirely determined by the velocity selected in the kpp equation .
indeed , we know from the previous section that @xmath522 converges to a traveling front solution @xmath560 of velocity @xmath503 .
the parametrization ( [ g - tfinie ] ) then implies that the distribution @xmath410 of vortex fugacity for the random xy model , itself converges to a traveling front solution , more conveniently expressed in the random core energy variable @xmath561 as : @xmath562 where @xmath531 is given in ( [ velocity - corrections ] ) and @xmath563 . the center of the front of @xmath564 , located in @xmath565 corresponds to the maximum of the distribution @xmath566 ( as can be easily seen on ( [ g - t0 ] ) ) and to the typical values of the random variable @xmath180 .
the front shape of @xmath564 is simply related to the kpp front solution @xmath518 through @xmath567 .
the asymptotic velocity of the front of @xmath564 is thus : @xmath568 the total velocity in the @xmath180 variable is thus the kpp velocity minus the stiffness .
the former comes from the spread of the distribution due to disorder , while the latter from the effect of interactions . in previous sections
we have explained that the xy phase corresponds to a decrease of the density of favorable regions @xmath569 ( i.e @xmath570 ) and to the absence of topological defects at large scale . in particular , for @xmath8 and @xmath483 to reach finite asymptotic values @xmath557 and @xmath571 , it is necessary that @xmath572 decreases fast enough .
@xmath415 can be estimated crudely as @xmath573 .
thus , as shown in fig .
[ fig : relative - velocity ] , when the total velocity @xmath574 is negative ( xy phase ) , the front moves to the left ( large @xmath96 or large @xmath180 ) and thus the probability of events @xmath575 decreases while if @xmath574 is positive , @xmath576 increases asymptotically ( disordered phase ) . the xy phase thus corresponds to the region where the total velocity is negative and thus to : @xmath577 in which case the whole probability distribution of the core energy @xmath578 , its typical and average values drift to @xmath579 at larger scale .
the transition line between the disordered and the xy phase can be located , in the plane @xmath557 , @xmath580 , by finding the line where this relative velocity @xmath581 vanishes .
the phase diagram can now be obtained by determining the velocity @xmath503 as a function of @xmath220 and @xmath26 .
the crucial observation is that by construction ( as is the case in @xcite ) the initial condition @xmath495 decays for large @xmath510 as : @xmath582 thus , since @xmath583 is finite , it decays exponentially and one can apply bramson s results ( [ velocity - corrections ] ) detailed in the previous sections with the identification : @xmath584 thus in effect it is the temperature which selects the velocity @xmath503 .
we find that , as depicted in fig [ fig : phase - diag ] : \(i ) for @xmath585 , i.e at high enough temperature @xmath586 , the velocity continuously depends on temperature : @xmath587 and thus , in that regime , the xy phase exists for : @xmath588 which is exactly the condition which would be obtained from the averaged fugacity ( at least when expressed in the renormalized parameters ) and leads to the transition line of rubinstein et al . as we have discussed earlier
( see eq.([rg - moments ] ) ) .
\(ii ) the velocity of the front _ freezes _ at @xmath589 and for @xmath590 , i.e at low temperature @xmath591 , where it becomes temperature independent : @xmath592 and the total velocity of the front @xmath481 is now @xmath593 thus we obtain that below this freezing temperature at : @xmath594 the transition between the xy phase and the disordered phases occurs at : @xmath595 . for @xmath596
the system is unstable to the proliferation of topological defects induced by disorder . to describe the xy phase it is important to understand the various regions of the scale dependent fugacity distribution @xmath410 .
indeed the fugacity distribution also gives the distribution of the charge correlation functions in the coulomb gas from the relations ( [ exp : correl1 ] ) . in the xy phase
one can describe the large scale behaviour of the fugacity distribution by neglecting the slow scale dependence of @xmath26 and @xmath8 ( adiabatic approximation ) .
the behaviour of @xmath481 can be obtained by studying @xmath597 since one has @xmath598 with @xmath599 . at @xmath51
this simplifies into @xmath600 .
it is also useful to note that the laplace transform of @xmath410 is @xmath601 . at low @xmath220 in the xy phase @xmath410
becomes very broad and one must distinguish for large but finite @xmath60 several different regions represented in fig .
[ fig : front - shape ] . using the results of the preceding sections on the kpp equation
we now describe these regions in details : _ interior front _ : the first region corresponds to the bulk of the probability distribution centered around the typical value of the fugacity @xmath602 . from the results for the front velocity ( [ xlowt ] , [ xhight ] ) one thus obtains that in the xy phase at high temperature @xmath603
the _ typical _ renormalized fugacity decays with the scale as : @xmath604 which in that case is also the scaling of the average fugacity @xmath605 .
on the other hand , below the freezing temperature @xmath606 it decays as : @xmath607 the bulk of the distribution is thus well described by the asymptotic front of the kpp equation as @xmath608 and @xmath609 .
its precise shape is of course non universal as it depends on the details of the definition of the fugacity at the level of the cutoff ( e.g. the function @xmath530 in ( [ kpp - general ] ) ) .
however , since the physically interesting region @xmath610 corresponds to the region ahead of the kpp front ( even though the _ total _ velocity @xmath611 is negative , see fig . [ fig : front - shape ] ) universal results about the kpp equation can be used .
in particular the _ near tail _ ( i.e the scaling region @xmath612 fixed but large ) is universal and can be obtained since @xmath613 . through laplace inversion one
gets that the fugacity distribution in that near tail region behaves as a power law : @xmath614 which is valid for @xmath615 and in the regime @xmath616 fixed ( as @xmath60 grows ) but large .
the freezing which occurs at @xmath617 thus concerns _ typical regions _ which , deep in the xy phase have a very small fugacity @xmath618 .
it corresponds to the temperature
at which the first moment of the distribution on the r.h.s . of ( [ typical ] ) becomes infinite .
thus for @xmath606 the true average fugacity @xmath619 and becomes dominated by rare local environments corresponding to values of @xmath96 outside of the bulk of @xmath410 ( where ( [ typical ] ) is invalid ) .
_ intermediate front _ : the second region is the `` intermediate front region '' ( see fig . [
fig : front - shape ] ) and corresponds to @xmath620 .
there we know from ( [ front - intermediate ] ) that @xmath621 with @xmath622 .
this region will be crucial to describe the critical behaviour in the following section .
_ far tail region _ : finally the most important region to describe the xy phase is the `` far tail region '' far ahead of the front with @xmath623 ( see fig . [
fig : front - shape ] ) . indeed , to obtain the renormalization of @xmath8 and @xmath26 ( via the screening equations ( [ screening2 ] ) ) we are interested in the rare events @xmath624 of small probability @xmath415 , but which dominate the _
average _ correlations .
these events correspond to the region of fixed @xmath180 and thus to @xmath625 .
fortunately , this region is so far ahead of the front that @xmath522 can be obtained with excellent accuracy by solving the linearized kpp equation : @xmath626 by straightforward integration from from @xmath400 to @xmath60 .
this leads to : @xmath627 we also notice that in this regime @xmath481 can also be obtained explicitly . indeed ,
since the relation between @xmath481 and @xmath522 is a simple linear ( @xmath60-dependent ) convolution it is straighforward to show ( e.g. via @xmath510 fourier space ) that @xmath522 obeying the linearized kpp equation ( [ linkpp ] ) is equivalent to @xmath481 satisfying : @xmath628 this linearized equation was not entirely obvious to guess directly from ( [ rg - eq-1d - u ] ) and justifies , even in this simplest regime , the detour through the rigorous results for the kpp equation .
using @xmath599 it yields : @xmath629 it is interesting to note that for _ fixed _ @xmath510 and @xmath180 , @xmath630 and @xmath481 can be estimated in the large @xmath60 limit as simple exponentials : @xmath631 with @xmath632 and an identical expression for @xmath633 with a different prefactor @xmath634 .
these forms , which are valid in the region of fixed fixed @xmath635 ( and fixed @xmath636 respectively ) will be useful to estimate the renormalization of @xmath8 and @xmath26 below ) and ( [ kpplinearu ] ) by approximating @xmath637 , and similarly for @xmath180 ( to be precise the @xmath483 and @xmath8 which appear here are @xmath638 and @xmath639 respectively ) .
they are valid only as long as the integrals which defines @xmath640 and @xmath641 are convergent .
since @xmath642 for large positive @xmath643 the estimate for @xmath644 is valid only as long as @xmath645 . for @xmath646 one must instead perform saddle point estimate , see below ] .
thus at low temperature we find the following decay of the probability of rare favorable local environments : @xmath647 with @xmath648 .
we also find that the distribution @xmath469 has an algebraic tail at low temperature in the region @xmath624
( i.e. @xmath96 fixed as @xmath649 ) as : @xmath650 with is a different power law behaviour the averaged fugacity @xmath651 is controlled by @xmath652 , while for @xmath653 is controlled by @xmath654 than the one which characterizes typical fugacities ( [ typical ] ) .
these two different power law behaviours are represented in figure [ fig : front - shape ] . finally , one can check that the three regions match properly and thus we have a fairly complete description of @xmath655 .
for instance , using the expression valid in the region ( ii ) for _ fixed _ @xmath510 at large @xmath60 one gets : @xmath656 which always gives the result ( [ papp ] ) up to the logarithmic corrections @xmath60 prefactors factors , and reproduces even the @xmath60 prefactors correctly for @xmath657 , which is when we expect the matching to become exact coefficient in the logarithmic correction to the velocity @xcite . ] . to study the screening equations ( [ screening2 ] ) in the xy phase we need the distribution @xmath658 of dipole core energy @xmath659 .
fortunately in this regime it can be computed simply from ( [ kpplinearu ] ) and is given simply by : @xmath660 the last equality being valid for fixed @xmath180 and @xmath476 , and @xmath661 in terms of the initial dipole core energy distribution . substituting this last form in ( [ screening2 ] ) we obtain , in the large @xmath60 limit : [ screeningc ] @xmath662 with @xmath663 a constant roughly independent of the temperature .
the two above integrals @xmath664 are convergent respectively only for @xmath665 and @xmath666 , in which cases it is indeed legitimate to replace @xmath667 by its above asymptotic form .
note that the @xmath51 limit is well defined since in that case @xmath668 and @xmath669 .
at higher temperature @xmath670 the full front solution controls the renormalization of @xmath8 . using the simple above gaussian form yields that the @xmath180 integral is dominated by the saddle point @xmath671 which gives : @xmath672 which corresponds for @xmath670 , to the behaviour of @xmath673 ( and for @xmath674 of @xmath675 ) , as expected .
a similar , though more involved , analysis can be performed for @xmath26 .
thus we have obtained the equations ( [ screeningc ] ) for the renormalization of @xmath8 and @xmath26 in the xy phase .
since we have shown that @xmath415 decreases exponentially as ( [ pl1 ] ) we conclude that @xmath8 and @xmath26 reach their finite limits @xmath676 and @xmath677 as power laws of the systems size .
this analysis also yields the full distribution of the correlation function of the charges in the xy phase .
indeed , let us recall that : @xmath678 thus by following the rg up to the scale @xmath679 we can obtain from @xmath680 the distribution of the charge correlations at large @xmath681 .
for instance , from the above and ( [ typ2 ] ) one finds the following decay of the _ typical _ thermal and disorder correlations : @xmath682 for @xmath683 and @xmath606 . for @xmath674 one
has instead @xmath684 .
the averaged moments can be obtained as above by substituting the exponential form of the distribution @xmath685 for @xmath686 . performing the corresponding integrals ,
one gets : @xmath687 with @xmath688 \gamma\left[p -\frac{t}{t^*}\right]}{\gamma[2 p ] } \\ & & b_p(t ) = \frac{c''}{\sqrt{8 \pi \sigma j^2 } } t ~\frac{\gamma\left[\frac{t}{t^*}\right ] \gamma\left[2 p - \frac{t}{t^*}\right]}{\gamma[2 p ] } \end{aligned}\ ] ] these formulae are valid only at low enough temperature @xmath689 . finally ,
although we have not attempted a precise rg calculation of the xy order correlation functions , the following behaviour should hold in the quasi ordered xy phase : @xmath690 with @xmath691 and @xmath692 . at the zero temperature transition point @xmath693 , @xmath51 , the value of these exponents are universal @xmath694 .
we now study the transition from the xy to the disordered phases at low temperature at @xmath695 . from the previous sections we know
that it occurs when the total velocity of the front of the distribution @xmath481 vanishes , i.e the critical region is defined by @xmath696 in a large @xmath60 regime . while in the xy phase the physics was dominated by the _
far tail _ of the traveling front @xmath623 in the critical regime it is the _ interior front _ @xmath697 , as well as the intermediate front region @xmath620 ( see fig .
[ fig : front - shape ] ) which controls the transition .
thus the correct description of the transition _ requires _ the knowledge of the kpp physics in an essential way , and is thus entirely novel . for simplicity
, we will only present the analysis at @xmath51 .
we will also work to first order in @xmath698 .
we first assume that the coefficient @xmath699 varies sufficiently slowly near the transition so that the results from the kpp equation with a constant @xmath483 can be used .
this assumption will be self consistently verified at the end .
near the transition the center of the front is located at @xmath700 with @xmath701 as indicated by ( [ velocity - corrections ] ) .
thus in the critical regime one still has @xmath702 , although logarithmically in @xmath60 exactly at the transition .
thus this critical regime can still be studied perturbatively , as @xmath415 remains very small .
the front velocity has the following scale dependence : @xmath703 where the h.o.t . contains the universal @xcite @xmath704 subdominant corrections to velocity in the kpp equation as well as additional subdominant corrections originating from the slowly varying @xmath484 . in the critical region at @xmath51
we can use the kpp front solution @xmath705 with , from ( [ front - intermediate ] ) : @xmath706 an expression valid as long as @xmath707 .
the rg equations for @xmath8 and @xmath26 in the critical region thus read : [ sct0 ] @xmath708 we need to evaluate these expressions for @xmath709 large and negative ( typically either as @xmath710 on the critical manifold or as @xmath711 very close to it ) . at criticality
they behave approximately as @xmath712 , which can be guessed by setting @xmath686 in the first expression . a more accurate estimate of the above integrals
is performed in the appendix [ part : integrals ] .
the end result is that , in the regime of interest ( where we can discard terms of order @xmath713 ) one has : @xmath714 where @xmath640 is a constant . introducing the small parameter : @xmath715
the density of favorable regions reads : @xmath716 since we can discard terms of order @xmath713 . from ( [ velcritical ] )
one finds that @xmath717 satifies precisely the rg flow equation : @xmath718 to lowest order in @xmath719 .
note the @xmath720 _ universal _ factor which arises from the universal velocity corrections in the kpp equation ( [ velocity - corrections ] ) .
a natural choice of parameters to describe the universal behaviour around the transition is thus @xmath721 ( which up to logarithmic corrections is equal to the density of favorable sites @xmath722 ) and @xmath26 which satisfies : @xmath723 these two equations ( [ eq - g ] , [ eq - sigma - g ] ) form our complete set of rg equations projected on the plane @xmath724 .
they are somewhat analogous to the one describing a kosterlitz - thouless type transition , with an important difference .
here one readily finds that the separatrix is _ vertical_. introducing the deviation from criticality , @xmath725 ( where @xmath726 ) one has for @xmath727 the flow : @xmath728 and thus satisfy @xmath729 . thus our rg equations yields @xmath730 at criticality @xmath731 at criticality ] and a correlation length : @xmath732 since starting away from criticality @xmath733 one finds read @xmath734 $ ] with @xmath735 = \int_x^{+\infty } dx x^{-3 } e^{-32 x}$ ] up to logarithmic corrections . ]
this critical behaviour correspond to a new universality class which is different from kosterlitz - thouless and from the prediction of @xcite .
note the crucial role of the @xmath720 universal factor . replacing it by any number less than @xmath23 would have led to usual kt behaviour .
we can now check that the variations of @xmath483 should be unimportant at the transition .
indeed one finds .
writing @xmath737 $ ] as a function of @xmath709 , we get that @xmath738 $ ] since @xmath709 appears only explicitly in the bound of the integral in ( [ sct0 ] ) .
thus one finds @xmath739 ( 1 - j \sigma \frac{d}{d x_l } \ln a[x_l])$ ] . using that using that @xmath740 = 2/\sqrt{d } + o(1)$ ] one gets that the leading variations cancel at the transition point @xmath741 . ] that @xmath742 at most . on the other hand we can estimate that if @xmath743 varies faster than @xmath744 the kpp results should not be affected ( see the scaling with @xmath60 of all terms in e.g. eq ( a7 ) in @xcite )
let us close by noting again how universality appears in the derivation of the critical behaviour .
although most details of the fugacity distribution @xmath745 ( e.g. its bulk , the fusion rule .. ) depend on the cutoff procedure the universality in the xy transition appears in a remarkable way .
it arises from the independence @xcite of the velocity , the velocity corrections , and front tail on the precise form @xmath746 $ ] of the non linear term in ( [ kpp - general ] ) .
this gives us confidence in the method developped here .
as is well known the coulomb gas can also be equivalently formulated as a sine gordon model ( see _ e.g _ @xcite ) . in this section
we identify the random version of the sine gordon model to which our analysis applies .
the scalar sine gordon model , of partition function @xmath747 is defined , in the absence of disorder by the action : @xmath748 where @xmath749 $ ] .
as is well known it is equivalent to a coulomb gas since , expanding in @xmath721 one has : @xmath750 where @xmath751 denotes averages with respect only to the quadratic gradient part . the interaction @xmath752 has been defined in ( [ part : model ] ) .
the above partition sums involve only @xmath753 neutral pairs of dipoles since , in the large size limit @xmath754 .
finally @xmath755 in the last line involves a summation over all distinct charges with @xmath756 for @xmath757 and @xmath758 for @xmath759 .
let us turn to the disordered version of the model . to reproduce the bare version ( [ zcont ] ) of the model
, one must first add a short range correlated random _
imaginary _ field as follows : @xmath760 with correlations @xmath761 .
since it imposes now that @xmath762 , each factor @xmath763 in the @xmath721 expansion in ( [ sgexp ] ) yields an additional @xmath764 where @xmath765 and thus reproduces the bare cg version ( [ zcont ] ) of our model .
note that the above model still contains a uniform `` fugacity '' @xmath721 .
it thus corresponds to the version studied in @xcite by rubinstein et al .
however we know that this can not be the correct form under renormalization .
the first obvious idea is to generalize @xmath766 , i.e a disordered fugacity ( the above expansion can be immediately generalized to this case )
. this term will be generated , but is not the end of the story .
indeed let us consider the symmetries of the above action @xmath767 ( even in the presence of a @xmath768 ) .
when @xmath769 the action is real and invariant through @xmath770 . in the presence of the random field
this symmetry is broken ( as @xmath771 acquires a non zero , disorder environment dependent average ) and the action is complex .
thus nothing prevents that under coarse graining the action will become : @xmath772 and this is indeed precisely what we have found in the cg formulation , i.e each sign of charge acquires a _
different _ local random fugacity .
there is however a symmetry constraint on the distribution of the local random fugacities . indeed in the above bare model ( [ sgaction ] ) the full partition sum is real , as there is still a _ statistical symmetry _ : @xmath773 for any configuration @xmath771 and environment @xmath774 . since the probabilities of @xmath774 and @xmath775 are the same , all physical averages will be real .
in the coarse grained model ( [ sgcoarse ] ) , since we have that @xmath776 the probability ( over environments ) of the random fugacity disorder configuration @xmath777 should be equal to the probability of @xmath778 .
we can now check that this random sine gordon model ( [ sgcoarse ] ) is indeed equivalent to the coarse grained disordered cg considered in this paper by expanding its the partition sum in a given environment , which yields : @xmath779 and thus the @xmath95 random field disorder is associated to the long range part @xmath321 of the disorder in the cg ( see ( [ decomposition ] ) ) while the random couplings constants @xmath780 in ( [ sgcoarse ] ) are associated to the local random fugacities in the cg .
similarly one can establish a correspondence directly on the replicated versions of both models . replicating the above sg model and averaging over disorder yields ( the limit @xmath139 is implicit ) : @xmath781 expanded to first order and treating the @xmath782 and @xmath783 as uncorrelated in space and reexponentiating yields the replicated sg model defined as : @xmath784 e^{i { \bf n } \cdot { \bf \phi } } \label{replicatedsg}\end{aligned}\ ] ] where @xmath785 is the scalar product in replica space and @xmath65 $ ] are analogous to the vector fugacities introduced previously in the replicated vector cg model .
although the bare model obtained from ( [ averagesg ] ) contains only single component replicated charges , all multicomponent charges will be generated upon coarse graining , as in the replicated vector cg . as we have seen all these charges should be taken into account and thus the generic replicated sg model should contain all possible @xmath65 $ ] with @xmath786 .
this sg model can also be studied using rg , either in its replicated form ( [ replicatedsg ] ) , or directly ( [ sgcoarse ] ) , very similarly to the vector cg studied in this paper .
although variations in definitions of the fugacities @xmath65 $ ] and different cutoff procedures can induce some irrelevant differences in the details of the renormalization ( and different rg equations ) , the two models have the same physics .
the sine gordon formulation has several advantages , such as exhibiting by construction the decomposition of the disorder in two physically very different components ( see [ decomposition ] ) .
it is also more amenable to replica variational methods than the cg , left for future study .
let us compare our method and results with the work of scheidl @xcite
. in this work the multicomponent charges ( restricted to @xmath66 in each replica ) were considered , but the fusion was not taken into account . also only dipole fugacities were introduced .
we can recover the rg equations and results of scheidl within our approach by ( i ) artificially setting the fusion coefficient @xmath197 to @xmath115 in ( [ loose ] ) ( ii ) assuming a log - normal distribution @xmath787 for the fugacities ( which is consistent only when fusion is neglected ) ( iii ) defining random `` dipole fugacities '' as @xmath788 and @xmath789 ( or equivalently @xmath790 ) identical to the `` dummy gaussian variables '' introduced by scheidl . within our diffusion formalism ( see section([part : replicalimit ] ) ) we find that the norm @xmath791 diverges exponentially , which allows to recovers the extra factors @xmath792 appearing in scheidl s screening equations . going from our diffusion formalism back to the replica formulation yields back the rg equations of @xcite . in presence of a fusion term @xmath793 ,
the equations become a priori very different
. the fugacity distribution does not remain log - normal as we do not assume a priori the form of the distribution .
interestingly , although the log - normal does not reproduce correctly the true distribution of fugacities ( and misses connections such as the one with the freezing of the dpct via the kpp equation ) , some of our results deep in the xy phase , such as the renormalization of @xmath794 and @xmath26 ( see ( [ screeningc ] ) ) , agree with the one of scheidl .
it was not obvious a priori that the approach without fusion in the xy phase did not miss extra relevant physics in this model , and indeed near the transition fusion appears to be crucial and must be taken into account .
to conclude , we have constructed in this paper a novel renormalization group method which allows to study perturbatively , and in a consistent way , a large class of disordered models which can be formulated as two dimensional coulomb gases with quenched disorder .
we have applied it specifically to the xy model with quenched random phases .
we have obtained the phase diagram for this model , confirmed the existence of a low temperature xy phase , and elucidated the critical behaviour at the transition where topological defects proliferate .
it would be interesting to check our predictions in numerical simulations the present rg method is not based on the conventional perturbative expansion in a vortex fugacity @xmath2 spatially uniform over the system , which , as we have shown , is only justified for pure models or for disordered models at high enough temperature .
instead , it is constructed by first defining the local random vortex fugacity ( or core energy ) and then following its full probability distribution under coarse graining . below a freezing temperature
this distribution becomes very broad and can not be followed by conventional cg methods .
our renormalization procedure allows to follow this broad distribution in a controlled way , by defining the new perturbative parameter as the concentration @xmath406 of rare regions favorable to vortices .
this running parameter flows to zero in the xy phase and at the transition ( marginal flow ) .
hence both can be studied perturbatively .
the underlying physical picture obtained here is that the transition is controlled , as the scale increases , by the proliferation of vortices in less and less rare favorable regions .
we find that it has some features reminiscent of a kosterlitz - thouless transition , with important differences , such as the scaling of the correlation length . to derive the rg equation for the distribution of vortex fugacities , we have introduced two equivalent methods .
one is based on the replicated vector coulomb gas version of the model , and in an expansion in the vector fugacities .
the second one is direct , with no use of replicas , and is based on a systematic expansion of all physical quantities in the number of points , i.e in independent local regions and thus , in the end , in powers of the concentration @xmath406 of rare favorable regions . as we have shown these two methods
are fully equivalent : the first one being more systematic and the second one allowing for a clear physical understanding of the problem in terms of probability distributions .
these two expansion methods are highly non perturbative in the original uniform fugacity variable @xmath2 .
since they are constructed from charge fugacities they can be made fully consistent ( by contrast with previous approaches based on dipole fugacities ) .
our method sheds light on the broader issue of universality in random systems .
the spirit of the rg method is that at large scale most information about the system is irrelevant and can be discarded . in constructing our rg procedure
we have first shown , using the very special properties of logarithmic interactions , that it is enough to follow , in addition to the two parameters @xmath794 and @xmath26 , the distribution of only one or two local ( i.e uncorrelated ) random variables @xmath795 . at this stage
it is clear that we still keep too much information .
indeed we have found that the precise form the non linear rg equation obeyed by @xmath795 as well as the detailed shape of this distribution are largely cutoff dependent .
however , this complicated looking rg equation can be generally recast , up to irrelevant terms , as a well known non linear front propagation of the kpp type . using the known remarkable universality property of this type of non linear equations
, we found that all the information needed to describe the universal properties of the xy phase and of the transition is indeed independent of the detailed shape of the fugacity distribution , or of the precise form of its rg equation .
only its tails , and the finite size corrections to the front velocity seem to be needed to determine the physical quantities and the critical behaviour .
since we are following a full distribution of local disorder , the present method could be termed a functional rg .
we can indeed draw some parallel between this functional rg procedure for the disordered cg and two other known examples where the universal behaviour of a disordered finite - range system is extracted from a functional rg equation .
the first one is the asymptotically exact real space rg in @xmath796 @xcite well suited to `` infinite disorder '' fixed points .
as emphasized in @xcite there are indeed similarities when treating the single vortex , @xmath796 version of the present model .
the method of @xcite can be applied for the sinai potential which has correlations growing as a power of distance . for the present case of weaker , logarithmic correlations
, these methods can also be applied in principle ( at least around zero temperature where disorder is still very broad ) but become very hard to work out analytically .
the other example of functional rg appears in a dimensional expansion for the problem of an interface in a random potential @xcite , which has infinite number of marginal directions at the upper critical dimension @xmath797 .
an important question is whether in the problem studied in this paper there is also an infinite number of marginal directions . as discussed above
, the results extracted from the kpp equation seem to suggest that a smaller amount of information than the exact full distribution may be needed here .
thus one can speculate that the critical theory studied here could be equivalently formulated as a more conventional field theory , yet to be identified , with a small number of marginal or relevant operators . in any case
the rg method developed here should provide a physically transparent guide to study the system . given the wide applications of coulomb gases in two dimensions
it is likely that other two dimensional disordered models can be studied using methods similar to the one introduced here .
finally , another outcome of the present work has been to unveil some interesting connections between the renormalization of a disordered system and the universal features of the propagation of invading fronts in non linear systems .
the existence of intriguing relations between freezing transitions and velocity selection in non linear fronts was noticed previously by derrida and spohn @xcite in their study of the dpct . here
, we found an even deeper and puzzling connection , betwen the `` universality '' in these selection mecanisms and the universality in the critical phenomena captured by the renormalization group .
since attempts have been made to construct renormalization methods in order to extract the universal features of such non linear equations @xcite , this suggests that a common framework could be developed in connexion with two dimensional disordered models . * acknowledgements * it is a pleasure to thank b. derrida , v. hakim and w. van saarloos for useful discussions .
in this appendix we derive , by exact transformations , a lattice disordered coulomb gas formulation of the villain form of ( [ xy ] ) , extending to this disordered case the approach of kadanoff @xcite .
an alternative route to @xcite in the pure case in the method used in @xcite . for simplicity
we firt turn to the villain version of ( [ xy ] ) before making the duality transformations , although the inverse procedure could be used ( see @xcite ) .
this villain model corresponding to ( [ square ] ) is @xcite : @xmath798\in \mathbb{z } } \prod_{i}\int_{-\pi}^{+\pi } \frac{d \theta_{i}}{2 \pi } \exp \left [ - \frac{k}{2\pi } \sum_{\langle i , j \rangle}(\theta_{i}-\theta_{j}-2\pi l_{ij}-a_{ij } ) ^{2}\right]\end{aligned}\ ] ] where @xmath221 and we used the notation @xmath799 ( see fig .
[ reseau ] ) .
this villain partition function indeed corresponds to a @xmath800-gauge theory , since the action is invariant under @xmath801 with @xmath802 . choosing the _ gauge field _
@xmath803 such that for each horizontal link @xmath804 ( _ i.e _ @xmath805 ) ( free boundary conditions along @xmath510 ) we obtain a partition function of a gaussian field @xmath806 ( @xmath807 ) : @xmath808 } \prod_{i}\int_{-\infty}^{+\infty } d \phi_{i } e^{- 2\pi\beta j \sum_{i } \left ( \left ( \nabla_{i , y}\phi + l_{i , y } -\frac{1}{2\pi } { \bf a}_{i , y } \right)^{2 } + \left ( \nabla_{i , x}\phi -\frac{1}{2\pi } { \bf a}_{i , x } \right)^{2 } \right ) } \end{aligned}\ ] ] with the notation for the discrete gradient @xmath809 . integrating over this gaussian field
yields the coulomb gas action @xmath810 } e^ { - \beta j \sum_{i , j } \left[(l_{i-\hat{e}_{y},y}-l_{i , y})- \frac{1}{2\pi}{\bf \nabla}_i.{\bf a}\right ] g_{ij } \left[(l_{j-\hat{e}_{y},y}-l_{j , y})- \frac{1}{2\pi}{\bf \nabla}_j.{\bf a}\right ] - \sum_i 2 \pi k~ l_{i , y}^2}\nonumber \\ & = & \label{latticecg } \sum_{n_{\alpha } } e^ { - \beta j \sum_{\alpha,\beta } ( n_{\alpha}-q_{\alpha } ) \gamma_{\alpha \beta } ( n_{\beta}-q_{\beta } ) } \end{aligned}\ ] ] where we have discarded the @xmath811 self interaction and we defined the dual lattice charges @xmath812 with @xmath813 ( see fig . [ reseau ] ) : @xmath814 and the quenched random dipoles @xmath815 note that the neutrality of disorder charges @xmath816 follows directly from the definition of @xmath817 ( e.g. taking @xmath818 to vanish at the boundary ) and implies neutrality for the integer charges @xmath819 ( from the divergence of @xmath820 ) . the _ coulomb potential _
@xmath821 is defined on the lattice by @xmath822 which gives the expression @xmath823 where the @xmath824 is the euler constant . using the neutrality of the charges and the definition @xmath825 , and the definition of the disorder potential @xmath826 , the lattice disordered cg action reads @xmath827
the coulomb gas , or its equivalent sine gordon version , can also be renormalized using smooth cutoff procedures .
we will not give details here but refer the reader to @xcite ( see also @xcite ) .
let us simply point out how , in that case , the full disorder @xmath270 can indeed be decomposed , as in ( [ decomposition ] ) , into two bona fide disorders . in the case of a soft cutoff ,
the continuum approximation of the lattice coulomb interaction reads ( instead of ( [ approx ] ) ) : @xmath828 where @xmath829 and we will choose @xmath830 a positive monotonously decreasing function of @xmath510 .
one has the asymptotic large @xmath91 behaviour @xmath831 ( see e.g. @xcite ) .
one also has : @xmath832 where @xmath833 .
thus one can write the cutoff dependent decomposition @xmath834 and one gets upon increase of cutoff : @xmath835 and @xmath836 with : @xmath837 thus both @xmath41 and @xmath838 are well defined physical gaussian disorders since their correlators have positive fourier transform .
in addition @xmath838 is short range correlated . for instance , taking @xmath839 one finds : @xmath840
in this appendix , for completeness , we explicitly renormalize the vector coulomb gas defined by ( [ expansion ] ) .
this amounts to extend to @xmath37-component vector charges the renormalisation of scalar cg @xcite .
the partition function reads : @xmath841 e^{\mathcal{a}_{a_{o } } [ { \bf n}_{1}\dots { \bf n}_{n}]}\end{aligned}\ ] ] where as usual the primed sum is over all distinct neutral charge configurations and the notation h.c .
stand for all the hard core constraints @xmath842 for all pairs @xmath843 , implicit in the following .
the action @xmath844 is defined by @xmath845=\frac{1}{2 } \sum_{i\neq j } 2k_{ab}~ n_{i}^{a}\ln \left(\frac { |{\bf r}_{i}-{\bf r}_{j}|}{a_{o } } \right ) n_{j}^{b}\ ] ] a common way to renormalise usual coulomb gas consists in coarse graining the partition function , leaving the expansion in number of fugacity ( here @xmath846 ) unchanged .
this amounts to define scale dependent replica stiffness @xmath847 and fugacities @xmath848 $ ] .
we will follow this scheme in this appendix . note that another equivalent way would be to renormalise the correlation function directly , following
_ e.g _ @xcite . as in the scalar case @xcite , renormalisation of the generalised vector coulomb gas @xcite proceeds in the same three steps : rescaling , fusion and annihilation ( screening ) of small dipoles .
we now turn to the description of these three contributions : first we increase the hard - core cut - off @xmath849 with @xmath850 .
this increase of cut - off produces a naive rescaling : @xmath851 e^{\mathcal{a}_{a_{o}}[{\bf n}_{1}\dots { \bf n}_{n } ] } = \prod_{i=1}^{n}\int \frac{d^{2}{\bf r}_{i}}{\tilde{a}_{o}^2 } y[{\bf n}_{i } ] e^{\mathcal{a}_{\tilde{a}_{o}}[{\bf n}_{1}\dots { \bf n}_{n } ] } e^{dl ( 2-k_{ab}n_{i}^{a } n_{i}^{b})}\ ] ] where we used the neutrality @xmath852 to express the correction coming from the action .
we can absorb the extra factor in ( [ app : rescaling ] ) to all order in @xmath846 by the change of fugacities corresponding to the equation @xmath853= ( 2-k_{ab}n^{a}n^{b } ) y[{\bf n}]\ ] ] upon the increase of cut - off , two charges @xmath854 and @xmath855 have to be coarse grained if they are located in @xmath856 and @xmath857 with @xmath858 . within the small charge density hypothesis , we consider only one such pair . for a dipole
, these two charges have to be integrated out at scale @xmath859 : this corresponds to the annihilation , while for a non neutral pair , the coarse grained charge is simply the sum of the two charges at scale @xmath90 ( fusion ) . in both cases the partition function splits into @xmath860 where @xmath861 involves configurations with one pair of charges @xmath862 distant of less than @xmath330 while @xmath863 does nt .
@xmath864 can be written as @xmath865 \\
\times \sum_{{\bf n}_{p},{\bf n}_{q } } \int_{a_{o}\leq |{\bf r}_{p}-{\bf r}_{q}|\leq a_{o}e^{dl } } \frac{d^{2}{\bf r}_{p}}{\tilde{a}_{o}^2 } \frac{d^{2}{\bf r}_{q}}{\tilde{a}_{o}^2 } y[{\bf n}_{p}]y[{\bf n}_{q } ] e^{\mathcal{a}_{p , q}[{\bf n}_{1}\dots { \bf n}_{n}]}\end{gathered}\ ] ] where , with the notation @xmath866 , the action reads @xmath867 we must now distinguish between a neutral pair and a non neutral one . in this case
the small pair of charges @xmath868 at scale @xmath90 gives an effective charge @xmath869 at scale @xmath859 , located in @xmath870 .
thus we must integrate over the relative position of the two charges @xmath871 when coarse graining the coulomb gas . to obtain the corresponding correction to @xmath863 ( of order one in @xmath342 ) ,
it is enough to expand ( [ expans ] ) to order 0 in @xmath872
. this expansion reads @xmath873 \sum_{{\bf n}_{p},{\bf n}_{q}}\int \frac{d^{2}{\bf r}}{a_{o}^2}~ y[{\bf n}_{p}]y[{\bf n}_{q } ] \left(\int_{a_{o}\leq \rho\leq a_{o } e^{dl } } \frac{d^{2}{\bf \rho}}{a_{o}^2 } \right ) \\ \times \prod_{i \neq p , q } \left ( \frac{|{\bf r}_{i}-{\bf r}|}{a_{o}}\right)^{\alpha_{ip}+\alpha_{iq } } \prod_{i < j \neq p , q } \left ( \frac{|{\bf r}_{i}-{\bf r}_{j}|}{a_{o}}\right)^{\alpha_{ij}}\end{gathered}\ ] ] using @xmath874 , we can rewrite this correction to @xmath863 as a single contribution to the fugacity @xmath875 $ ] of the _ non zero _ charge @xmath869 : @xmath876= 2 \pi \sum_{{\bf n}_{p},{\bf n}_{q } } y[{\bf n}_{p}]y[{\bf n}_{q}]\ ] ] note that in this expression @xmath854 and @xmath855 are distinguishable charges : this explains the factor 2 between ( [ rg - fusion ] ) and ( [ rgrep2 ] ) . in this case
the small dipole of size @xmath877 is integrated out at scale @xmath859 .
when coarse graining , we sum over both @xmath872 and @xmath386 , yielding a factor @xmath878 from the integration measure .
thus diverging contributions correspond to the expansion of ( [ expans ] ) to order 2 in @xmath872 . using @xmath879
, this expansion of @xmath864 can be expressed as @xmath880 \prod_{i < j \neq p , q } \left ( \frac{|{\bf r}_{i}-{\bf r}_{j}|}{a_{o } } \right)^{\alpha_{ij } } \sum_{{\bf n}_{p}}y[{\bf n}_{p}]y[-{\bf n}_{p}]\\ \times
\int_{a_{o}\leq \rho\leq a_{o } e^{dl } } \frac{d^{2}{\bf \rho}}{a_{o}^2 } \int \frac{d^{2}{\bf r } } { a_{o}^2 } \left ( 1+\sum_{i , j \neq p ,
q}^{n } \alpha_{i , p } \alpha_{j , p } \frac{1}{4 } { \bf \rho}.{\bf \nabla}\ln ( { \bf r}_{i}-{\bf r } ) { \bf \rho}.{\bf \nabla}\ln ( { \bf r}_{j}-{\bf r } ) \right)\end{gathered}\ ] ] performing the integral and reexponentiating the last term using the neutrality of the configuration @xmath881 , we get the correction to @xmath863 coming from @xmath882 : @xmath883 \\ \times \prod_{i < j \neq p , q } \left ( \frac{|{\bf r}_{i}-{\bf r}_{j}|}{a_{o}}\right)^{\alpha_{ij}-\pi^{2}dl \sum_{{\bf n}_{p } } \alpha_{i , p}\alpha_{j , p}~y[n_{p}]y[-n_{p } ] } \end{gathered}\ ] ] the constant @xmath884 corrects the intensive free energy while the second term can be absorbed in a correction to the coupling constant @xmath885 } n^c n^d y[{\bf n } ] y[- { \bf n}]\ ] ] the three above contributions can be summarized into the set of rg equations given in the text ( [ rgrep ] ) valid for _ all non zero vector charge _ , and for all @xmath37 . the coefficient @xmath886 and @xmath364 depends on the ir regularisation . for our
hard cut - off , we find within our procedure that @xmath887 and @xmath888 .
note that the ratio @xmath889 is independent of a uniform rescaling of the fugacities and is known , in the case of single component charges to be universal at a transition @xcite . for
a discussion of regularisation of replicated coulomb gas , see discussion in appendix [ app - cutoff ] .
in this appendix we explicitly perfom the @xmath67 limit of the whole set of rg equations ( [ rgrep1],[rgrep2 ] ) with the restriction that the vector charges have only @xmath890 components .
this limit is taken using the parametrisation ( [ eq : ydef ] ) of the fugacities @xmath891 $ ] in terms of the function @xmath175 . in the three different terms of ( [ rgrep1],[rgrep2 ] ) , corresponding to rescaling , annihilation and fusion contributions ( see appendix [ part : rgreplica ] ) , we first modify sums to include fugacities for null charge , translate the expression in terms of @xmath188 and naturally take the @xmath139 limit .
we will use the notation @xmath892 .
the term corresponding to rescaling in ( [ rgrep2 ] ) is @xmath893 = ( 2 - n_a k^{ab } n_b ) y[{\bf n } ] \end{aligned}\ ] ] which holds for any vector charge @xmath38 .
the second term can be expressed in terms of @xmath188 using ( [ eq : ydef ] ) and @xmath894 with @xmath221 .
however the expression is much simpler if one uses , instead of ( [ eq : ydef ] ) , the equivalent parametrisation @xmath65=\int_{u , v } e^{\beta pu}e^{\beta qv } \tilde{\phi } ( u , v)$ ] with @xmath895 , @xmath896 and @xmath897 . with @xmath898 ,
this yields @xmath899&= & ( \beta j p - \sigma \beta^2 j^2 q^2 ) \int_{u , v } \tilde{\phi}(u , v ) e^{\beta pu } e^{\beta qv } \\ & = & \int_{u , v } \tilde{\phi}(u , v ) \left ( j \frac{\partial}{\partial u } - \sigma j^2 \frac{\partial^2}{\partial v^2 } \right ) e^{\beta pu } e^{\beta qv } \\ & = & - \int_{u , v } e^{\beta pu } e^{\beta qv } \left ( j \frac{\partial}{\partial u } + \sigma j^2 \frac{\partial^2 } { \partial v^2 } \right)\tilde{\phi}(u , v ) \label{lim - repar}\end{aligned}\ ] ] as this is true for any vector charge @xmath900 satisfying @xmath901 , _
i.e _ for any @xmath902 or equivalently @xmath903 , to satisfy ( [ dim - y ] ) we can search for a function @xmath898 such that : @xmath904 in terms of @xmath175 , this corresponds to the equation : @xmath905 where we have used that @xmath906 . note that for this process , the integral of @xmath188
, @xmath192 satisfies simply @xmath907 . in ( [ rgrep2 ] )
, the fusion term of two charges @xmath908 is restricted to @xmath909 and @xmath910 .
furthermore since @xmath911 corresponds to the annihilation which is treated separately in ( [ rgrep1 ] ) ( see below ) and must not be counted twice , the equation ( [ rgrep2 ] ) can be used only for @xmath912 .
it is convenient to extend the equation ( [ rgrep2 ] ) to include @xmath913 = \int \phi = \mathcal{n}$ ] for which the fusion contribution should be absent .
thus the extended equation corresponding to fusion reads : @xmath914= c_{2 } \left ( \sum _ { \genfrac{}{}{0pt}{}{{\bf n}'+ { \bf n } '' = { \bf n}}{{\bf n}',{\bf n}''\neq 0 } } y[{\bf n } ' ] y[{\bf n } '' ] - \delta_{{\bf n},0 } \sum_{{\bf n } ' \neq 0 } y[{\bf n } ' ] y[-{\bf n } ' ] \right ) \\ & & = c_{2 } \left ( \sum _ { { \bf n}'+ { \bf n } '' = { \bf n } } y[{\bf n } ' ] y[{\bf n } '' ] - 2 y[{\bf 0 } ] y[{\bf n } ] - \delta_{{\bf n},0 } \left ( \sum_{{\bf n } ' } y[{\bf n } ' ] y[-{\bf n } ' ] - 2 y[0]^2 \right ) \right ) \nonumber\end{aligned}\ ] ] where in the second equality we allow for @xmath915 or @xmath916 . turning to the representation in terms of @xmath903 we have : @xmath917 y[{\bf n}'']\\ & & = \left < \prod_a \bigl [ ( 1 + z'_- z''_+ + z'_+ z''_- ) \delta_{n_a,0 } + ( z'_+ + z''_+ ) \delta_{n_a,+1 } + ( z'_- + z''_- ) \delta_{n_a,-1 } \bigr ] \right>_{\phi,\phi}\end{aligned}\ ] ] using permutation symmetry we find that ( [ eqphi1 ] ) can be written as : @xmath918 = \\ & & c_2
\left < ( 1 + z'_- z''_+ + z'_+ z''_- ) ^{m } \left(\frac{z'_+ + z''_+}{1 + z'_- z''_+ + z'_+ z''_-}\right)^{n_{+ } } \left(\frac{z'_- + z''_-}{1 + z'_- z''_+ + z'_+ z''_-}\right)^{n_{- } } \right>_{\phi \phi } \\ & & - 2 c_2 \mathcal{n } \left < z_+^{n_+ } z_-^{n_- } \right>_{\phi } - c_2 \delta_{n_+,0 } \delta_{n_-,0 } \left ( \left < ( 1 + z'_- z''_+ + z'_+ z''_- ) ^{m } -1 \right>_{\phi \phi } - \mathcal{n}^2 \right)\end{aligned}\ ] ] the following choice for @xmath919 allows to satisfy the above equation for all @xmath38 ( up to now @xmath37 is still arbitrary ) : @xmath920 we can now take the limit @xmath139 _ explicitly _ on this equation , which yields : @xmath921 putting all these contributions together , we obtain the scaling equations for the coupling constant @xmath8 , the correlated disorder strength @xmath26 and the distribution of local disorder @xmath175 : @xmath929 where the diffusion operator has been defined in ( [ def - o ] ) .
in this appendix we explore the connections between the expansion in number of sites of the free energy , and the expansion of the replicated partition function in power of composite charge fugacities @xmath65 $ ] .
as we will see , both expansions coincide exactly , and we can thus consider the expansion of the replicated partition function as a generating functional of the site - expansions .
fist we consider the first term @xmath930 given by ( [ z-2 ] ) of the expansion in number of independent sites ( [ free ] ) .
this term corresponds exactly to the approximation of independent dipoles considered in @xcite : dipoles do not interact with each other , and the free energy is thus the sum over all positions of the pairs of the free energy of a pair : @xmath931 , and the free energy of a dipole is simply @xmath932 using the decomposition of the disorder ( section [ part : disdec ] ) @xmath348 , and averaging over the correlated disorder @xmath41 using replica , we obtain @xmath933 where we used the non local correlation of the disorder @xmath934 and the definition @xmath935 . with the bare definition of the replica charge fugacities ,
this finally gives after average over @xmath43 , the expected first term of the expansion of @xmath936 : @xmath937 } y[{\bf n } ] y[-{\bf n } ] \left(\frac{|{\bf r}-{\bf r}'|}{a_{o}}\right)^{-2 n^{a}k^{ab}n^b}\end{aligned}\ ] ] the methods is the same as in previous section , but the calculations are slightly more tedious .
we consider the second term of the free energy expansion : @xmath938 with the help of the previous section , it is enough to consider only @xmath939 using the decomposition of figure ( [ fig : secondorder ] ) , we obtain the sum @xmath940 by averaging the second term over @xmath41 , and using @xmath935 , we get @xmath941 inserting this result in the above expression averaged over @xmath41 , the sum over the partitions of the interval @xmath942 $ ] in @xmath943 can be exactly rewritten as a sum over all _ neutral _ triplets @xmath944 of @xmath37 component vector charges with components @xmath66 ( see figure [ fig : secondorder ] ) .
thus one recovers the expression @xmath945(r ) y[n'](r ' ) y[n''](r'')e^{-s_{n , n',n '' } } -1\ ] ] we now note that the last three logarithmic terms in ( [ logs ] ) just give additional restrictions on this sum ( as they correspond respectively to @xmath946 ) .
thus , averaging over the local fugacities , we end up with the expected second term ( see [ replicah ] ) of the expansion of @xmath153 in power of the vector fugacities @xmath65 $ ] : @xmath947 y[{\bf n ' } ] y[{\bf n '' } ] e^{- \beta h_{{\bf n , n',n''}}}\end{aligned}\ ] ] these simple combinatorics can be done on higher order terms : it gives the equivalence term by term between the @xmath67 limit of the expansion in @xmath65 $ ] and the expansion in the number of independent sites of the moments of the free energy .
in this appendix we discuss the consequences on the rg equations of the choice of cutoff made in this paper for the replicated coulomb gas .
we illustrate for simplicity only the case of zero disorder .
although we will be mainly concerned with coulomb gas , most of this discussion can be applied to other general replica field theory .
let us first recall the results for a single component cg ( @xmath948 ) .
we restrict to the most relevant charges @xmath45 for simplicity .
it is defined by the action @xmath949 the corresponding rg equation was derived by kosterlitz : @xmath950 we now consider @xmath37 copies of this model in the absence of disorder , as illustrated in fig .
( [ fig : cutoff ] ) .
they are a priori physically completely uncoupled .
the most natural cutoff procedure in that case would be independent cutoffs ( e.g. hard core for each ) ( left figure ) .
another procedure , which becomes much more convenient in the presence of disorder , is to reformulate the @xmath37 copies as a single coulomb gas of vector charges with @xmath37 components .
however in that case the cutoff is by definition columnar ( right figure ) ( e.g hard core vector charge are a hard columnar disk ) and in a sense the copies are coupled , via the cutoff .
we now check that in the pure case the ensuing vector cg rg equations are still perfectly compatible with ( [ cutoff : rg1 ] ) as they should .
they read : @xmath951 = \left ( 2 - k { \bf n } \cdot { \bf n } \right ) y[{\bf n } ] + c_{2 } \sum_{{\bf n}'\neq 0,{\bf n } } y[{\bf n } ' ] y[{\bf n}-{\bf n } ' ] + \text{higher order terms } \nonumber\ ] ] let us first illustrate the case of two copies @xmath952 .
we can choose @xmath953=y[-1,0]=y[0,1]=y[0,-1 ] \equiv y_1 $ ] and @xmath954=y[-1,1]=y[1,-1]=y[-1,-1 ] \equiv y_2 $ ] . then considering all the possible fusions within this set the rg equations read : @xmath955 a solution of these equations , to the order @xmath956 at which we are working , is : @xmath957 = \lambda_1 y_l \quad y_l[1,1 ] = \lambda_2 y_l^2\end{aligned}\ ] ] with @xmath958 , where @xmath553 satisfies the single copy equation ( [ cutoff : rg1 ] ) .
this can be generalized to higher charges so that in general one can find solutions of the type : @xmath959 } = \lambda_{i } y_l^{i}\ ] ] with coefficients @xmath960 which can be determined for a given regularisation procedure .
they depend on both the initial ( @xmath948 ) and the replica regularisation ( see fig [ fig : cutoff ] ) .
this example illustrates how a generic cutoff procedure will produce non trivial coefficients @xmath961 . thus if one reinterprets formally @xmath65
= < y^{n_+ + n_- } > _ { \phi(y)}$ ] in terms of a `` disorder '' @xmath232 as done in this paper , ] one must keep in mind that even the pure system corresponds to a non trivial `` bare disorder '' in the fugacities , which solely originates from the ( convenient ) choice of a _ columnar _ cutoff for a particular choice which does not introduce additional disorder ] .
it is thus clear that the definition of the local fugacity distribution is strongly cutoff dependent .
in this appendix we derive formula ( [ fk ] ) for the expansion of the free energy in the number of points .
the same method can be applied to other physical observables . to organize this expansion
we start by introducing _
fictitious site dependent fugacities _ for the charges : @xmath962 .
these are introduced only as a trick in this appendix and should not be confused with the real disordered fugacities of ( [ decomposition ] ) . indeed , to recover the original model
, we will set them back at the end either to @xmath963 ( for the lattice model ( [ square - bis ] ) ) or @xmath964 for the continuum model , where @xmath2 is the corresponding fugacity for the charge in the pure case .
these fugacities are introduced by writing a more general form of the the partition function ( [ square - bis ] ) as @xmath965 = \sum_{p } \sum'_{\{n_{1},\dots n_{p } \ } } \sum_{{\bf r}_{1}\neq \dots \neq { \bf r}_{p } } \left(\prod_{i } \zeta_{{\bf r}_{i}}^{n_i^2 } \right ) e^{\beta j \sum_{{\bf r}_{i}\neq { \bf r}_{j } } n_{i}g_{{\bf r}_{i}-{\bf r}_{j } } n_{j } + \beta \sum_{i } n_{i}v_{{\bf r}_{i}}}\ ] ] here and below , as in ( [ zcont ] ) , all formulaes can be extended to the continuum model by replacing discrete sums over distincts sites @xmath268 by integrals with , _
e.g _ hard core conditions @xmath269 .
note that in the above expression ( [ square - fict ] ) we do not make use of the decomposition ( [ decomposition ] ) and @xmath966 denotes the original disorder .
let us consider for simplicity a system of @xmath30 distinct points @xmath967 .
the free energy functional @xmath968 = - t \ln z[v , \zeta ] $ ] is a function of the @xmath30 variables @xmath969 , .. , @xmath970 .
let us write the conventional taylor expansion of the free energy around @xmath971 : @xmath972 & = & \sum_{p_1,p_2, .. p_n = 0}^{+\infty } \frac{1}{p_1 ! \dots p_n ! } \frac{\partial^{p_1+ .. p_n } f[v,\zeta ] } { \partial \zeta_{{\bf r}_1}^{p_1 } \dots \partial \zeta_{{\bf r}_n}^{p_n}}|_{\zeta=0 } \times \zeta_{{\bf r}_1}^{p_1 } ...
\zeta_{{\bf r}_n}^{p_n}\end{aligned}\ ] ] we now separate in this sum the terms which have only one non zero @xmath973 , then only two non zero @xmath973 , etc .. thus we can rewrite : @xmath974 & = & \sum_{k=1}^{+\infty } \sum_{\ { r_{i_1}, .. ,r_{i_k } \ } } \sum_{q_1, ..
q_k = 1}^{+\infty } \frac{1}{q_1 ! \dots q_k ! } \frac{\partial^{q_1+ .. q_k } f[v,\zeta ] } { \partial \zeta_{{\bf r}_{i_1}}^{q_1 } \dots \partial \zeta_{{\bf r}_{i_k}}^{q_k}}|_{\zeta=0 } \times \zeta_{{\bf r}_{i_1}}^{q_1 } ...
\zeta_{{\bf r}_{i_k}}^{q_k } \\ & = & \nonumber \sum_{k=1}^{+\infty } \sum_{\ { r_{i_1}, .. ,r_{i_k } \ } } f^{(k)}_{r_{i_1}, .. ,r_{i_k}}[v,\zeta]\end{aligned}\ ] ] where the sum is over all distincts _ sets _ @xmath975 of @xmath279 distincts points ( among the @xmath30 points ) and the sum over each @xmath803 goes from @xmath23 to @xmath579 . we have thus obtained an expansion of @xmath976 $ ] as a sum of terms of the form @xmath977 which depends exactly and only on the variables @xmath978 ( and thus also only on the variables @xmath966 ) evaluated at the @xmath279 _ distinct points _ @xmath979 . for a neutral coulomb gas
it starts with @xmath295 and reads : @xmath980&= & \sum_{\ { { \bf r}_{i_1 } \neq { \bf r}_{i_2 } \ } } f^{(2)}_{{\bf r}_{i_1},{\bf r}_{i_2}}[v,\zeta ] + \sum_{\ { { \bf r}_{i_1 } \neq { \bf r}_{i_2 } \neq { \bf r}_{i_3 } \ } } f^{(3)}_{{\bf r}_{i_1},{\bf r}_{i_2},{\bf r}_{i_3}}[v,\zeta ] + \dots\end{aligned}\ ] ] with the definition @xmath981 & = & \sum_{q_1, ..
q_k = 1}^{+\infty } \frac{1}{q_1 ! \dots q_k ! } \frac{\partial^{q_1+ .. q_k } f[v,\zeta ] } { \partial \zeta_{{\bf r}_{i_1}}^{q_1 } \dots \partial \zeta_{{\bf r}_{i_k}}^{q_k}}|_{\zeta=0 } \times \zeta_{{\bf r}_{i_1}}^{q_1 } ... \zeta_{{\bf r}_{i_k}}^{q_k}\end{aligned}\ ] ] note that in this last expression we can drop out the dependence of @xmath976 $ ] on the fugacities @xmath982 and the potential @xmath11 at points different from @xmath983 .
a more explicit expression can be obtained by summing over the @xmath984 in ( [ def - fk ] ) : @xmath985 & = & \sum_{l=0}^{k } ( -1)^{k - l}\sum_{i_{1},\dots i_{l}\in [ 1,\dots k ] } f_{{\bf r}_{i_{1}},\dots { \bf r}_{i_{l}}}[v,\zeta]\end{aligned}\ ] ] where @xmath986 $ ] is the free energy associated with the system of sites @xmath277 ( instead of the full lattice ) .
equivalently , it does depend only on the fugacities @xmath982 ( or potential @xmath11 ) at points @xmath277 . after setting @xmath987
( lattice model ) or @xmath988 ( continuum model ) this gives the definition of the @xmath989 $ ] that we use throughout this paper , namely the equation ( [ fk ] ) .
this last expression ( [ fk - expl ] ) allows to explicitly compute the coefficient @xmath989 $ ] of the expansion ( [ free - zeta ] ) for arbitrary order @xmath279 . in the section ( [ part : exp ] ) we use it to give explicit expressions for the first few terms of the expansion of the free energy .
as shown in the previous section the free energy expansion involves the partition function @xmath990 of a system of finite number of sites .
we illustrate here how the rule of fusion of environments described in section [ part : directrg ] works more generally .
let us consider a system of @xmath991 sites , with charges restricted to @xmath992 and of energy and fugacities given by ( [ zcont ] ) .
upon increase of the cutoff one must take into account fusion of the sites @xmath244 and @xmath245 into the site @xmath993 , one finds that : @xmath994 with : @xmath995 } ( w^{(q)}_{1,{\bf r}_{i_2}, .. {\bf r}_{i_q } } + w^{(q)}_{2,{\bf r}_{i_2}, .. {\bf r}_{i_q } } ) \end{gathered}\ ] ] where @xmath996 and @xmath997 } \end{aligned}\ ] ] where @xmath998 has been defined in ( [ zcont ] ) and the charge @xmath999 is located in @xmath1000 .
note that @xmath1001 the dipole weight defined in section [ part : directrg1 ] . in ( [ sump ] )
the first term corresponds to the total weight of configurations with no charge in @xmath244 and @xmath245 .
the second term correspond to configurations with a dipole in @xmath1002 .
its original expression is complicated but simplifies when @xmath244 and @xmath245 are fused into @xmath1000 ( the interaction energy of any given other charge @xmath1003 in @xmath1004 with @xmath1005 becomes opposite to the one with @xmath1006 ( up to higher order terms ) and @xmath1007 simplifies into @xmath1008 to lowest order in @xmath342 as explained in section [ part : directrg ] ) .
the last two terms correspond to all configurations with one charge either in @xmath244 or in @xmath245 .
the important property with respect to the free energy expansion is that one can factor the term @xmath1009 and rewrite : @xmath1010 } \tilde{w}^{(q)}_{\tilde{{\bf r}},{\bf r}_{i_2}, .. {\bf r}_{i_q } } ) \nonumber\end{aligned}\ ] ] where @xmath1011 as the same definition as @xmath1012 except that the fugacity at @xmath1000 has been modified according to the fusion rule for fugacities ( [ rule2 ] ) .
the total factor inside the last logarithm is exactly the partition function @xmath1013 of a system of @xmath1014 sites with the new fugacity given by ( [ rule2 ] ) on the site @xmath1000 .
in this appendix we briefly indicate how higher charges can be included in the same ( fusion - diffusion ) formalism and why including them does not affect any of our results .
let us consider for instance the charges @xmath1015 and define @xmath1016 and @xmath1017 respectively as the number of component charges @xmath1018 and @xmath1019 in the vector charge @xmath38 .
the parametrization ( [ eq : ydef ] ) can be readily extended to encode also for the charges @xmath1015 simply by writing the vector fugacity @xmath65 $ ] as an average over a function @xmath1020 as @xmath1021 = \int_{z_{+},z_{-},z_{++},z_{-- } } \phi(z_{+},z_{-},z_{++},z_{-- } ) z_{+}^{n_{+ } } z_{-}^{n_{- } } z_{++}^{n_{++ } } z_{--}^{n_{-- } } \nonumber\end{aligned}\ ] ] where the random variables @xmath1022 and @xmath1023 represent the random local fugacities for the charges @xmath1015 .
similar manipulations as in appendix ( [ part : replimit ] ) lead to a rg equation for a normalized @xmath1024 which we will not write explicitly .
it does contain a diffusion operator as well as a fusion term . for illustration we simply give the diffusion operator , expressed using more convenient variables @xmath1025 and @xmath1026 and @xmath1027 .
it reads , for the corresponding probability distribution @xmath1028 : @xmath1029 the detailed study of the corresponding rg equation , together with the fusion term will not be reported here .
instead let us indicate how the irrelevance of the higher charges can be justified .
since one has : @xmath1030 it is clear that the reduced probability @xmath1031 satisfies the same diffusion equation as @xmath193 in ( [ rgeqp ] ) but with the change @xmath1032 and @xmath1033 .
it becomes relevant by power counting only at @xmath1034 .
thus it is less relevant than @xmath193 in the region of the phase diagram of interest and we can rightly neglect the new fugacities @xmath1022 and @xmath1023 .
one can also check that the fusion terms imply that @xmath1035 is of order @xmath1036 when this parameter is small ( since at lowest order the rg equation contains a term proportional to @xmath1037 ) .
the rare events which involve the charges @xmath1015 are thus subdominant .
to evaluate the integral ( [ sct0.1 ] ) in the screening equation for @xmath8 we consider separately three intervals @xmath1038 , @xmath1039 and @xmath1040 , where @xmath1041 is a number such that @xmath1042 can be replaced by its asymptotic expression for @xmath1043 with good accuracy . only in the middle interval
can we use the universal tail expressions for both factors @xmath1044 .
thus we have , using symetries : since , remembering that the integral @xmath1047 , the above integral is at most a finite number . the second integral in ( [ cutintwo ] ) is estimated as : @xmath1048 and gives the leading contribution .
for @xmath657 it behaves as @xmath1049 and gives the estimate in the text .
it is encouraging to note that the leading behaviour precisely originates from the interval where @xmath1050 can be replaced everywhere by its universal asymptotic form .
the integral ( [ sct0.2 ] ) in the screening equation for @xmath26 can also be estimated by considering for each variable @xmath180 , @xmath1051 two intervals @xmath1052 and @xmath1053 .
the end result is that we find to leading order : | we study the two dimensional xy model with quenched random phases and its coulomb gas formulation . a novel renormalization group ( rg ) method is developed which allows to study perturbatively the glassy low temperature xy phase and the transition at which frozen topological defects ( vortices ) proliferate .
this rg approach is constructed both from the replicated coulomb gas and , equivalently without the use of replicas , using the probability distribution of the local disorder ( random defect core energy ) . by taking into account the fusion of environments ( i.e charge fusion in the replicated coulomb gas )
this distribution is shown to obey a kolmogorov s type ( kpp ) non linear rg equation which admits travelling wave solutions and exhibits a freezing phenomenon analogous to glassy freezing in derrida s random energy models .
the resulting physical picture is that the distribution of local disorder becomes broad below a freezing temperature and that the transition is controlled by rare favorable regions for the defects , the density of which can be used as the new perturbative parameter .
the determination of marginal directions at the disorder induced transition is shown to be related to the well studied front velocity selection problem in the kpp equation and the universality of the novel critical behaviour obtained here to the known universality of the corrections to the front velocity .
applications to other two dimensional problems are mentionned at the end . |
the bulk of the naturally occurring nuclei heavier than iron observed in the solar system were synthesized via neutron capture processes . in the case of the so - called astrophysical
_ s _ process , the neutron flux is moderate , elements are synthesized through slow neutron captures and @xmath5 decays up to @xmath6bi @xcite . on the contrary , in an astrophysical environment where the neutron densities are orders of magnitude higher than the ones available for the _ s _ process , the neutron captures become faster than the @xmath5-decays and extremely neutron - rich nuclides close to the neutron drip line can be synthesized .
these nuclei decay back into the valley of stability when the neutron flux ceases . this process the so - called astrophysical
_ r _ process
can only take place in explosive environments @xcite .
furthermore , there are about 30 - 35 proton - rich nuclides between @xmath7se and @xmath8hg , which can not be formed by neutron capture processes @xcite .
they are the so - called _ p _ isotopes .
the number of _ p _ isotopes depends on the one hand on the state of art _ s _ process models : a recent investigation @xcite showed that there are large _ s _
process contributions to the abundance of @xmath9er , @xmath10gd and @xmath11ta . on the other hand ,
the precise estimation of the _ r _ process contribution to the abundance of certain _ p _ isotopes ( e.g. @xmath12 in or @xmath13sn , see in @xcite ) is essential , too .
consequently the modification of the list of _ p _ isotopes could be necessary .
the natural isotopic abundance of the _ p _ isotopes is typically 10 to 100 times less than that of the more neutron - rich isotopes of the same element that were created in the _ s _ or _ r _ processes . in their production ,
photon - induced reactions at temperatures around a few gk play a crucial role .
it is generally accepted that the main stellar mechanism synthesizing these nuclei the so - called @xmath4 process @xcite is initiated by photodisintegration reactions on pre - existing neutron - rich _ s _ and _ r _ seed nuclei .
photons with high energy and high flux , necessary for the photodisintegrations , are available in explosive nucleosynthesis scenarios , as in the ne / o burning layer in type ii- @xcite or in the type ia supernovae @xcite
. however , based solely on this model , the calculations are not able to reproduce the solar system _ p _ abundances mainly in two mass regions : the light _
isotopes near the mo - ru region are largely underproduced by the photodisintegration reactions . in recent years
, several other processes have been discussed that could contribute to the production of light _
nuclei through proton captures in a highly proton - rich material : the so - called _
rp_-(rapid proton capture ) process @xcite , @xmath14p - process @xcite , and proton captures on highly enriched _
process seeds during type ia supernova explosion @xcite .
however , the role of these processes in nucleosynthesis remains unclear , e.g. , for the _
rp_-process it is uncertain whether the produced nuclei can be expelled into the interstellar medium and plays any role in the galactic chemical evolution . the _ p _ nuclei in the intermediate region between 150 @xmath15 a @xmath15 165 are underproduced , too @xcite .
it remains unclear whether this deficiency is due to nuclear cross sections , stellar physics , or if alternative / additional process has to be invoked .
although the basics of the @xmath4 process were laid down several decades ago , many details of the process are still unknown . on the one hand , this concerns the ambiguities in the astrophysical conditions under which the process takes place ( seed isotope abundances , peak temperatures , time scale , etc ) . on the other hand ,
large uncertainties are introduced into the calculations by the nuclear physics input , most importantly by the reaction rates ( determined from cross sections ) .
all @xmath4 process models require the use of huge reaction networks including tens of thousands of nuclear reactions , and the rates of these reactions at a given stellar temperature are necessary inputs to the network calculations .
the reaction rates are generally taken from calculations using the hauser - feshbach ( h - f ) statistical model @xcite .
the accuracy of the h - f predictions mainly depends on the adopted nuclear models for optical - model potentials , @xmath4-ray strength functions , and nuclear - level densities .
previous @xmath4 process related experiments showed that above the a @xmath16 100 region the precise knowledge on the @xmath2 + nucleus optical potential would be highly desirable @xcite , in order to increase the predictive power of the model calculations .
the subject of the present work is the study of the angular distributions of elastically scattered alpha particles on the @xmath0cd _ p _ nucleus .
on the one hand experimental angular distributions measured at higher energies are available in the literature @xcite and therefore , the opportunity to study the energy dependence of the imaginary part of the @xmath2 + nucleus optical potential in a wide energy region ranging from e@xmath3 = 27 mev down to e@xmath3 = 16.1 mev is given .
furthermore , recently the cross sections of the @xmath2-induced reactions on @xmath0cd were measured at very low energies , close to the astrophysically relevant ones @xcite .
it was found that there is about a factor of 3 difference between the experimental cross sections and the h - f predictions in the case of the ( @xmath17 ) reaction .
the inaccurate knowledge on the @xmath2 + nucleus optical potential was identified as the source of this disagreement .
finally , the elastic scattering data obtained extends the already existing data set of experiments performed on several proton - rich nuclei @xcite and provides additional data to test the global parametrizations used in astrophysical network calculations .
the paper is organized as follows .
experimental details are described in sec .
[ sec : exp_proc ] .
the measured angular distributions are presented in sec .
[ sec : res ] .
here , the new data are analyzed using a locally adjusted potential , and predictions from global @xmath2nucleus potentials are shown . in sec .
[ sec : capture_reac ] , we compare the results of the h - f calculations calculated with the use of the local and global potentials discussed in the present work with the available @xmath2-induced cross section data taken from the literature @xcite .
finally , conclusions are drawn in sec .
[ sec : summary ] .
the experiments were performed at the cyclotron laboratory of atomki , in debrecen , hungary . the alpha elastic scattering angular distributions between @xmath18 = 20@xmath19 and @xmath18
= 170@xmath19 were measured on the _ p _ nucleus @xmath0cd at three energies around the coulomb barrier .
highly enriched @xmath0cd targets ( @xmath20 ) of 250 @xmath21g/@xmath22 thickness were prepared using vacuum evaporation technique .
the experimental setup was similar to that used in previous experiments @xcite ( and references therein ) . in the following paragraphs , a short outline of the most relevant characteristics of the experimental setup
a detailed description of the experimental setup can be found in e.g. @xcite .
the angular distributions of the elastically scattered alpha particles were measured with the help of two pairs of silicon detectors , mounted on two independently rotatable turntables .
the solid angles of the detectors were between @xmath23 sr and @xmath24 sr .
in addition , two silicon detectors under @xmath18 = 15@xmath25 , with a solid angle of @xmath26 sr were used for absolute normalization , to monitor the target thickness and to take into account the beam position on the target .
the energies of the alpha particle beam were e@xmath3 = 16.1 mev , 17.7 mev and 19.6 mev with typical beam intensities in the order of 250 na .
two different sets of spectra registered during the measurement of the elastic scattering cross section are shown in fig [ fig : spectra ] .
on the left side of the figure , the spectra measured at position @xmath18 = 25@xmath25 are shown for the three measured energies .
the right side of the figure presents the spectra recorded at e@xmath3 = 19.6 mev , for three different angular positions .
as it can be seen , the elastic peak can be easily separated from the additional components , both at forward ( carbon and oxygen peaks ) and backward ( inelastic scattering on @xmath0cd ) angles .
in addition , pulser signals were fed into the pre - amplifiers in order to have a control on the dead - time of the detection system .
the pulser signal amplitude was set so that the pulser peak always falls above the @xmath0cd elastic peak ( see fig .
[ fig : spectra ] ) .
knowledge of the exact angular position of the detectors is of crucial importance for the precision of a scattering experiment because the rutherford cross section depends sensitively on the angle .
the uncertainty in the rutherford normalized cross section is dominated by the error of the scattering angles in the forward region . in a similar way to previous experiments @xcite ( and references therein ) we took advantage of the kinematics of the reaction @xmath27c(@xmath1)@xmath27c and measured the elastically scattered alpha particles in coincidence with the recoil @xmath27c .
we fixed one of the detectors on one side of the chamber at @xmath28 = 80@xmath25 to measure the alpha particles , and measured in coincidence the recoil @xmath27c nuclei around the expected angle of @xmath29 = 40.38@xmath25 .
the result can be seen in fig .
[ fig : ang_calib ] .
the uncertainty of the scattering angle was determined to be @xmath30 = 0.07@xmath25 .
several measurements were performed at other angular positions , and confirmed this result , ensuring the high precision of the experimental data .
complete angular distributions between 20@xmath19 and 170@xmath19 were measured at energies of e@xmath3 = 16.1 mev , 17.7 mev and 19.6 mev in 1@xmath19 ( 20@xmath19 @xmath15 @xmath31 @xmath15 100@xmath19 ) , 1.5@xmath19 ( 100@xmath19 @xmath15 @xmath31 @xmath15 140@xmath19 ) , and 2@xmath19 ( 140@xmath19 @xmath15 @xmath31 @xmath15 170@xmath19 ) steps .
the statistical uncertainties varied between 0.3% ( forward angles ) and 1 to 2% ( backward angles ) @xcite .
the count rates @xmath32 have been normalized to the yield of the monitor detectors @xmath33(@xmath31 = 15@xmath19 ) : @xmath34 the solid angle of the detectors is represented by @xmath35 .
the rutherford - normalized cross sections cover around 2 orders of magnitude , taking into account all the angular distributions measured ( @xmath36 mev to @xmath37 mev ) . over this range of cross sections ,
our measured angular distributions have an accuracy of around 4 to 5% , while the data in @xcite claim around 1 to 2% at forward angles and the authors add @xmath38% for backward angles due to the relative low enrichment . in the following
, the analysis of the new experimental data in the framework of the optical model will be presented .
our analysis is extended up to e@xmath3 = 27 mev by taking into account the elastic scattering angular distributions from @xcite .
the optical model potential combines a coulomb term with the complex form of the nuclear potential , which consists of a real and an imaginary part . usually , the parameters of the optical potential are derived from the analysis of the angular distributions of elastically scattered alpha particles and are adjusted to experimental @xmath2-induced cross sections if they are known .
the @xmath2 + nucleus optical potential is given by : @xmath39 where @xmath40 is the coulomb potential ( calculated from a homogeneously charged sphere ) , @xmath41 and @xmath42 are the real and imaginary parts of the nuclear potential , respectively . in the following , we will first determine a locally adjusted potential ( sec .
[ subsec : opt_mod ] ) and then compare the new experimental data to predictions of global @xmath2 + nucleus optical potentials ( sec .
[ subsec : ang_theo ] ) .
a local optical potential is tailored to describe the experimental data of a small region of the mass range , i.e. , it is a potential with parameters derived from the data of a few alpha elastic angular distributions or even from just a single experiment .
this type of potential is used as an analysis tool , since it is possible to achieve @xmath43 very close to unity , and therefore it is possible to study the impact / sensitivity of its different parameters in the description of the experimental data .
max width= llllllllllll & * @xmath44 * & * @xmath45 * & * @xmath46 * & * @xmath47 * & * @xmath48 * & * @xmath49 * & * @xmath50 * & * @xmath51 * & * @xmath52 * & * @xmath53 * & * @xmath43 * + ( mev ) & ( - ) & ( - ) & ( mev @xmath54 ) & ( fm ) & ( mev ) & ( fm ) & ( fm ) & ( mev @xmath54 ) & ( fm ) & ( mb ) & ( - ) + 15.6 - atomki & 1.218 & 1.029 & 351.84 & 5.3602 & 193.01 & 1.563 & 0.291 & 91.66 & 7.490 & 341.9 & 0.8 + 16.4 - nd & 1.211 & 1.020 & 341.06 & 5.3152 & 169.47 & 1.468 & 0.413 & 101.19 & 7.1421 & 491.3 & 1.0 + 17.0 - atomki & 1.373 & 1.004 & 368.91 & 5.2324 & 166.30 & 1.464 & 0.403 & 96.51 & 7.1158 & 544.4 & 0.9 + 17.0 - nd & 1.423 & 1.007 & 385.38 & 5.2462 & 171.17 & 1.446 & 0.436 & 105.02 & 7.0679 & 581.2 & 0.7 + 18.3 - nd & 1.371 & 1.005 & 369.37 & 5.2370 & 149.35 & 1.428 & 0.464 & 95.30 & 7.0147 & 733.4 & 1.1 + 18.9 - atomki & 1.378 & 1.000 & 365.29 & 5.2076 & 138.73 & 1.402 & 0.505 & 93.26 & 6.9422 & 803.0 & 0.8 + 21.2 - nd & 1.380 & 1.008 & 374.43 & 5.2500 & 134.37 & 1.419 & 0.500 & 91.41 & 7.0108 &
1039.5 & 2.0 + 21.2 - nd@xmath55 & 1.393 & 1.004 & 373.74 & 5.2304 & 137.40 & 1.414 & 0.492 & 91.26 & 6.9812 & 1020.9 & 1.7 + average & 1.336 & 1.010 & 365.18 & 5.2641 & 160.34 & 1.456 & 0.430 & 96.34 & 7.1119 & - & - + av .
& 1.338 & 1.010 & 365.08 & 5.2613 & 160.78 & 1.455 & 0.429 & 96.31 & 7.1077 & - & - + 23.6 - nd & 1.401 & 0.994 & 364.39 & 5.1760 & 159.16 & 1.372 & 0.492 & 99.72 & 6.7903 & 1150.6 & 5.4 + 23.6 - nd@xmath55 & 1.354 & 1.007 & 366.90 & 5.2475 & 152.59 & 1.388 & 0.513 & 102.18 & 6.8891 & 1209.2 & 2.6 + 26.0 - nd & 1.446 & 0.989 & 370.83 & 5.1515 & 132.06 & 1.372 & 0.512 & 86.21 & 6.8157 & 1287.2 & 13.2 + 26.0 - nd@xmath55 & 1.392 & 1.004 & 373.17 & 5.2288 & 128.98 & 1.392 & 0.525 & 88.96 & 6.9209 & 1346.5 & 7.4 + + 23.6 - s+v @xmath56 & 1.433 & 0.990 & 368.14 & 5.1547 & 7.65 & 1.686 & 0.423 & 54.10 & 4.8021 & 1153.2 & 1.4 + & & + 26.0 - s+v @xmath56 & 1.413 & 0.996 & 370.10 & 5.1880 & 10.65 & 1.648 & 0.464 & 54.82 & 5.0561 & 1309.1 & 2.5 + & & + & & & & & & & & & 1006.4 & 1.2 + & & & & & & & & & 1154.7 & 1.1 + & & & & & & & & & 1309.1 & 1.2 + + + the real part of the local potential is based on the double folding procedure of @xcite calculated using charge densities based on electron scattering @xcite ; this @xmath41 potential is modified by a strength parameter @xmath44 and a width parameter @xmath45 , as in the case of the global potential atomki - v1 ( see sec .
[ subsec : ang_theo ] ) .
the nucleon - nucleon interaction is parametrized using the well known ddm3y interaction @xcite .
the fitted local potential has a woods - saxon form imaginary part @xmath42 , which consists of a surface @xmath57 term only at very low energies @xcite . by fitting the data from each of the angular distributions , the @xmath48 , @xmath49 and @xmath50 parameters of the imaginary part
were left as free parameters for the fitting . for the real part
the bare double folding potential was calculated by the dfold code @xcite , using the aforementioned ddm3y interaction , and was fixed for all the @xmath58 .
the width @xmath45 and strength @xmath44 parameters of the real part were left free , in order to modify the bare folding potential ( with @xmath59 ) .
the results of this local potential study are shown in table [ tab : local_pot_parameters ] .
the table lists all the resulting parameters , as well as the calculations for the volume integrals @xmath46 and @xmath51 , root mean - square radii @xmath47 and @xmath52 , the reaction total cross section @xmath53 and the obtained @xmath43 .
it is obvious from table [ tab : local_pot_parameters ] that the angular distributions above 20mev measured at notre dame @xcite can only be described with relatively poor @xmath60 .
there are two reasons for this unexpected large @xmath61 .
first , the absolute values of the notre dame data @xcite have been normalized to rutherford scattering ( @xmath62 ) at the most forward experimental angle .
however , this procedure is not justified at the higher energies .
we have allowed to vary the absolute normalization of these angular distributions , leading to @xmath60 which are about a factor of two smaller than the ones using the original normalization .
the normalization factors are 1.016 , 0.946 , and 0.929 at 21.2mev , 23.6mev , and 26.0mev , respectively .
second , as already mentioned above , a surface imaginary potential is not sufficient at higher energies , and consequently a volume woods - saxon potential has been added ( labelled `` s+v '' in table [ tab : local_pot_parameters ] ) .
this improves the obtained @xmath60 by another factor of two . as a further test , in addition to the above mentioned local potential fits , phase shift fits
were also performed for the notre dame data @xcite using the method of @xcite .
here again @xmath63 can only be obtained with modified absolute normalization of the data , and the best - fit values for the absolute normalization agree within about 1% with the local potential fits .
this confirms that the absolute normalization of these data has to be revised . for the prediction of
low - energy @xmath2-induced cross sections we have chosen an average potential from all angular distributions below 21.2mev which can be described using a pure surface woods - saxon imaginary potential .
these average values are practically insensitive to the modified normalization of the notre dame data @xcite because the strongly affected data at 23.6 mev and 26.0mev do not contribute here ( the additional volume woods - saxon imaginary potential complicates the extrapolation to low energies ) .
the minor renormalization of the 21.2mev data affects the average local potential parameters by the order of 0.1% .
unfortunately , our analysis is not sufficiently sensitive to determine the energy dependence of the parameters of the imaginary part .
figure [ fig : cd106_all_pot_nd ] presents the predicted angular distributions calculated using this average local potential for each of the recently measured elastic scattering cross sections , together with the experimental data taken at university of notre dame @xcite , to illustrate the overall good description . for completeness ,
[ fig : cd106_all_pot_nd ] shows in the case of e@xmath64 = 23.6 mev and 26.0 mev the local potential with additional volume contribution . the local fits using the parameters from table [ tab : local_pot_parameters ] and
the phase shift fits are not shown in fig .
[ fig : cd106_all_pot_nd ] because they follow exactly the experimental data points ( @xmath63 ) . in the framework of the @xmath4 process network studies more than 2 @xmath65 10@xmath66 reactions on about 2000 mostly unstable nuclei
are taken into account . in many of those reactions alpha particles
are present either in the entrance or in the exit channel .
experimentally , most of these reactions are inaccessible for cross section measurements , since short - lived unstable nuclei are involved in the entrance channel .
therefore , global @xmath2 + nucleus optical potentials are used in the reaction network to predict the needed reaction rates . the variation of the potential parameters of the real part as a function of mass and energy is smooth and relatively well understood . on the contrary ,
the imaginary part of the optical potential is strongly energy - dependent especially at energies around the coulomb barrier . as a result of these different energy dependencies of the parameterizations ,
different predictions for reaction cross sections will occur , in particular at very low energies far below the coulomb barrier . in the following , we will compare our new experimental data to the predictions of well - known or recent open - access global potentials .
( i. ) in the middle of the 1960s , alpha elastic scattering experiments at e@xmath67 = 24.7 mev on nuclei ranging from o to u were performed and a consistent optical potential study was carried out . based on this study , a global @xmath2 + nucleus potential was formed .
it is a very simple , four - parameter woods - saxon potential with mass and energy - independent parameters referred to as the potential of mcfadden and satchler @xcite .
this potential is the default for the h - f calculations of astrophysical reaction rates in the non - smoker code @xcite . despite its simplicity
it provides a reasonable description of alpha scattering data and cross sections of @xmath2-induced reactions , at energies slightly above the coulomb barrier .
the h - f calculations based on this mcfadden potential have a tendency to overestimate reaction cross sections at energies below the coulomb barrier , e.g. , @xcite .
recently the depth of this potential was modified by adding a fermi - type function and this modified potential provided excellent predictions for @xmath2-induced cross sections on @xmath68pr and @xmath69er @xcite .
the `` standard '' potential of m. avrigeanu _
et al_. @xcite is an @xmath2 + nucleus optical potential with woods - saxon parameterizations for the real and imaginary parts , with a total of nine independent parameters , three for the real potential , depth @xmath70 , radius @xmath71 and diffuseness @xmath72 , three to describe the imaginary volume potential @xmath73 , @xmath74 , @xmath75 and the remaining three to describe the imaginary surface potential @xmath48 , @xmath49 , @xmath50 .
all of these parameters are mass and energy dependent and the @xmath2 + nucleus optical potential was obtained by fitting alpha particle elastic scattering and reaction cross sections on nuclei in the @xmath76 mass range , and for @xmath77 mev .
consequently , this potential provides a reasonable description for both the scattering and reaction data .
however , the extrapolation of such a many - parameter potential to unstable nuclei with extreme neutron - to - proton ratio @xmath78 may lead to additional uncertainties in the calculation of astrophysical reaction rates . in recent years
, several @xmath2-induced cross sections on nuclei with mass number above a @xmath79 140 @xcite were measured and , furthermore , new experimental alpha elastic scattering angular distributions on even - odd nuclei became available @xcite , thus improvements of this parametrization became possible .
it was shown @xcite that the accuracy of the cross section predictions can be increased using an updated surface imaginary - potential depth @xmath80 . however , for the @xmath0cd target under study in this work , the modifications in the latest version @xcite are very minor and restricted to energies below about 10mev , i.e. , the modifications do not affect the analysis of the elastic scattering data . therefore , we show the results using the potential from @xcite .
( iii . ) finally , the most recent @xmath2 + nucleus optical potential used in this work is the atomki - v1 @xcite optical potential parameter set .
it is based entirely on alpha elastic scattering angular distributions measured in the @xmath81 mass region at energies above and below the coulomb barrier .
the real part @xmath41 consists of a double folding parametrization @xmath82 , based on the widely used ddm3y interaction @xcite and modified by a strength parameter @xmath44 derived from the energy - independent volume integral @xmath46 .
the @xmath48 imaginary part is parametrized by a surface woods - saxon function with fixed @xmath49 and @xmath50 geometry and energy - dependent strength @xmath48 and volume integral @xmath51 .
the atomki - v1 potential has been constructed for low energies where a pure surface woods - saxon potential in the imaginary part is sufficient .
thus , it is not surprising that atomki - v1 is not able to reproduce the angular distributions at the highest energies studied in this work .
figure [ fig : cd106_all_pot_nd ] shows the available experimental angular distributions in comparison with the predictions calculated using the above mentioned global optical potential parameter sets .
it can be seen that each of the different @xmath2 + nucleus optical potentials used in this work provides a slightly different description of the experimental data , however , none of them describes the entire experimental data set . for a strict comparison between the potentials , the @xmath43 values and total reaction cross sections @xmath53
can be found in table [ tab : local_pot_parameters ] .
max width= .
[ tab : table_chi_global ] @xmath43 analysis of the global optical potentials models considered in this work .
please note that the @xmath83 indicates that the @xmath84 100 and nd refers to the notre dame measurements .
[ cols= " < , < , < , < , < , < , < , < , < , < " , ] from the results of table [ tab : table_chi_global ] , on the one hand , we found that the considered global @xmath2 + nucleus optical potentials roughly describe the experimental data in the energy range from e@xmath85 = 15.6 mev to e@xmath85 = 18.9 mev . on the other hand , above e@xmath85 = 18.9 mev ,
all @xmath2 + nucleus optical potentials fail to accurately describe the data , providing relatively large @xmath60 values .
this asks for further improvement of global @xmath2nucleus potentials but does not affect the following study of low - energy @xmath2-induced reaction cross sections .
reaction cross sections of @xmath2-induced reactions for heavy nuclei can be calculated using the statistical model ( stm ) @xcite .
in particular , this model has been widely applied for the calculation of reaction cross sections and stellar reaction rates in nuclear astrophysics @xcite . in a schematic notation ,
the reaction cross section in the stm is proportional to @xmath86 with the transmission coefficients @xmath87 into the @xmath88-th open channel .
the @xmath87 are calculated from global optical potentials ( particle channels ) and from the @xmath4-ray strength function for the photon channel . for details of the definition of @xmath87 ,
see @xcite .
@xmath89 refers to the entrance channel where the target nucleus is in its ground state .
note that this holds only for laboratory experiments ; under stellar conditions , i.e. , for the calculation of stellar reaction rates @xmath90 , thermal excitations of the target have to be considered .
it is typical for @xmath2-induced reactions on heavy nuclei that @xmath91 is much smaller than the other @xmath87 .
a simple qualitative explanation is the high coulomb barrier in the @xmath2 channel . in the neutron channel ,
a coulomb barrier is completely missing and , in the proton channel , the barrier is much lower . as a consequence , the cross section in the stm in eq .
( [ eq : stm ] ) factorizes into a production cross section of the compound nucleus which is proportional to @xmath89 , and a decay branching ratio @xmath92 practically independent of @xmath93 .
the production cross section is thus entirely defined by the chosen @xmath2 + nucleus optical potential whereas the branching ratio @xmath94 depends practically not on the @xmath2 + nucleus optical potential but on all the other ingredients of the stm ( optical potentials for the other channels , @xmath4-ray strength functions , level densities ) .
consequently , all cross sections of @xmath2-induced reactions are sensitive to the @xmath2 + nucleus optical potential , but each individual ( @xmath95 ) , ( @xmath96 ) , or ( @xmath17 ) reaction has further and complicated sensitivities to the other ingredients . as the @xmath2 + nucleus optical potential affects directly the production cross section ,
the sensitivity ( as defined e.g. in @xcite ) is close to 1 for all ( @xmath2,@xmath97 ) reactions at energies around or below the coulomb barrier .
the reaction @xmath98-values for @xmath2-induced reactions on @xmath0cd are @xmath99mev for the ( @xmath96 ) , @xmath100mev for the ( @xmath95 ) , and @xmath101mev for the ( @xmath17 ) reactions .
experimental data for these reactions are available at low energies from about 7mev to 13mev @xcite . following the above argumentation
, it can be expected that the ( @xmath96 ) cross section is dominating at higher energies down to close above the ( @xmath96 ) threshold .
below 10mev , only the ( @xmath95 ) and ( @xmath17 ) channels are open , and it turns out that the ( @xmath95 ) channel becomes very weak towards lower energies .
the focus of the present study is the @xmath2 + nucleus optical potential for @xmath0cd which determines the compound formation cross section . in fig .
[ fig : total ] , we show the formation cross section taken as the sum over the dominating ( @xmath17 ) , ( @xmath96 ) , and ( @xmath95 ) channels for the potentials under study in this work . obviously , the calculated compound formation cross section should be an upper envelope of the experimental ( @xmath17 ) , ( @xmath96 ) , and ( @xmath95 ) data . at higher energies above about 13mev , all potentials lead to very similar cross sections .
this is a general finding which has been explained from the properties of reflexion coefficients @xmath102 , see @xcite .
however , at lower energies , significant differences can be seen .
the local potential which has been adjusted to the elastic scattering angular distibutions overestimates the experimental data at energies below about 10mev .
unfortunately , it was not possible to determine the energy dependence of the imaginary part from elastic scattering and , therefore , the local potential behaves similar to the widely used potential by mcfadden and satchler @xcite which has also been determined from elastic scattering at higher energies and also has an energy - independent imaginary part .
the reaction cross sections from the mcfadden / satcher potential have already been shown in @xcite .
the new atomki - v1 potential uses a very similar parametrization as the local potential , but it has an energy - dependent imaginary part in addition .
the decreasing imaginary strength towards lower energies leads to smaller cross sections , but the calculation still overestimates the experimental data . on the contrary , the potential by avrigeanu _
et al . _
@xcite underestimates the data .
the latest revision of this potential @xcite has a slightly stronger imaginary part at very low energies which leads to slightly larger cross sections by about 10% below 10mev .
let us now turn to the individual ( @xmath17 ) , ( @xmath96 ) , and ( @xmath95 ) reactions .
as already stated above , these cross sections depend not only on the @xmath2 + nucleus optical potential with sensitivity @xmath103 as defined in @xcite , but also on the further ingredients of the stm with varying energy - dependent sensitivities @xmath104 .
this complicates the comparison of the results of the stm with the experimental data .
the following calculations are the results from smaragd code @xcite with its default parameters . following the scope of this study , only the @xmath2 + nucleus optical potential has been varied .
[ fig:106cd_ag_others ] shows the astrophysical s factors of the @xmath0cd(@xmath17)@xmath105sn reaction which dominates below 10mev . here
the ( @xmath17 ) cross section is essentially defined by the @xmath2 + nucleus optical potential .
similar to the compound formation cross section in fig .
[ fig : total ] , the local potential overestimates the low - energy data significantly , the atomki - v1 potential slightly overestimates , and the potential by avrigeanu _
et al . _ slightly underestimates . at higher energies above the neutron threshold ,
the cross sections become @xmath106 with @xmath107 and , thus , no direct conclusion on the @xmath2 + nucleus optical potential is possible here . as a word of caution , it must be mentioned that the behavior of the ( @xmath17 ) cross section at low energies may also be affected by direct reactions , in particular , coulomb excitation of the first @xmath108 state in @xmath0cd @xcite .
[ fig:106cd_an_others ] shows the same comparison for the @xmath0cd(@xmath96)@xmath109sn reaction . at energies sufficiently above the threshold ,
cross sections become the @xmath110 for @xmath107 . in principle , it would be possible to derive properties of the @xmath2 + nucleus optical potential from these data ; however , the experimental data cover only a narrow energy interval of about 2mev from 10mev to 12mev , and the error bars are not sufficiently small to allow clear conclusions .
[ fig:106cd_ap_others ] shows the astrophysical s factors of the @xmath0cd(@xmath95)@xmath109 in reaction .
similar to the ( @xmath96 ) data , the energy range is also very limited ( 9mev - 10mev ) .
as the cross section becomes @xmath111 in the energy range of the experimental data , it is not possible to directly conclude on the @xmath2 + nucleus optical potential from these data . for completeness
it should be mentioned here that the study of cross section ratios like , e.g. @xmath112 allows to constrain the other ingredients of the stm except the @xmath2 + nucleus optical potential . for an application of this idea , see e.g. @xcite . as we have seen ( and has been stated in our earlier study of the reaction data @xcite ) ,
it is not straightforward to draw conclusions on the @xmath2 + nucleus optical potential from the cross sections of the ( @xmath17 ) , ( @xmath96 ) , and ( @xmath95 ) reactions .
we have measured three angular distributions of @xmath0cd(@xmath1)@xmath0cd elastic scattering at energies around the coulomb barrier ; and we have analyzed these data and data from the literature @xcite within the optical model .
reasonable descriptions can be obtained for all data using a real folding potential with two adjustable parameters ( strength @xmath44 and width @xmath45 ) and an imaginary part of woods - saxon type .
similar to previous results @xcite , at low energies , a surface woods - saxon type parametrization is sufficient , whereas at higher energies a combination of volume and surface imaginary part improves the obtained @xmath43 significantly .
a local potential is derived from the average parameters .
unfortunately , it was not possible to determine the energy dependence of the imaginary part at low energies .
the new experimental data are also described very reasonably by the recent global potentials @xcite , whereas the old mcfadden / satchler potential @xcite has a trend to overestimate the elastic data . the locally adjusted potential and the global potentials by avrigeanu _ et al . _
@xcite and atomki - v1 @xcite were used to calculate the cross sections of @xmath2-induced reactions on @xmath0cd .
it is shown that these cross sections factorize into a compound production cross section which depends only on the chosen @xmath2 + nucleus optical potential and into a decay branching which is practically independent of the @xmath2 + nucleus optical potential but shows a complicated dependence on the other ingredients of the stm . therefore ,
clear conclusions on the @xmath2 + nucleus optical potential from @xmath2-induced reaction data are difficult in the case of @xmath0cd .
nevertheless , some conclusions can be drawn .
it turns out that the locally adjusted potential overestimates the reaction data at low energies .
this result is similar to the one obtained with the widely used mcfadden / satchler potential and can be attributed to the missing energy dependence of the imaginary part .
the predictions from the new global potentials @xcite are much closer to the experimental data @xcite but still far from an excellent description with @xmath63 .
the atomki - v1 potential @xcite slightly overestimates the reaction data at low energies whereas the potential by avrigeanu _ et al . _
underestimates the data .
as similar results have been found for @xmath113ce and @xmath68pr , the best prediction of @xmath2-induced reaction cross sections in the mass range @xmath114 could be taken from the geometric mean of these recent global potentials .
this work was supported by otka ( k101328 , k108459 ) , by the hungarian & portuguese intergovernmental s&t cooperation programme no .
tt_10 - 1 - 2011 - 0458 and by the ensar / thexo european fp7 programme .
g. g. kiss acknowledges support from the jnos bolyai research scholarship of the hungarian academy of sciences .
t. rauscher acknowledges support from the swiss nsf , the uk stfc grant bridgce ( st / m000958/1 ) and the european research council .
k. sonnabend acknowledges support from dfg ( so907/2 - 1 ) .
f. kppeler _
et al . , _ rev .
83 ( 2011 ) 157 .
r. reifarth _
et al . , _ j. phys .
g 41 ( 2014 ) 053101 .
m. arnould , s. goriely , the r - process of nucleosynthesis : the puzzle is still with us : astrophysics , ed . prof .
ibrahim kucuk , isbn : 978 - 953 - 51 - 0473 - 5 , intech ( 2014 ) , available from : http://cdn.intechopen.com/pdfs-wm/34260.pdf d. argast _ et al . , _ astronomy and astrophysics 416 ( 2004 ) 997 s. e. woosley and w. m. howard , astrophys .
j. suppl 36 ( 1978 ) 285 .
m. arnould and s. goriely , phys .
384 ( 2003 ) 1 .
m. kusakabe _
et al . , _ astrophys .
j. 726 ( 2011 ) 25 .
c. arlandini _
et al . , _ astrophys .
j. 525 ( 1999 ) 886 .
et al . , _ astrophys .
j. 426 ( 1994 ) 357 .
t. rauscher _ et al .
76 ( 2013 ) 066201 c. travaglio _
et al . , _ the astrophysical journal 739 ( 2011 ) 93 .
c. travaglio _ et al . , _ the astrophysical journal 795 ( 2014 ) 141
. h. schatz _
et al . , _ physics reports 294 ( 1998 ) 167 .
c. frhlich _
et al . , _ phys .
96 ( 2006 ) 142502 .
w. m. howard , b. s. meyer and s. e. woosley , the astrophysical journal letters 373 ( 1991 ) l5 . t. rauscher , a. heger , r. d. hoffman and s. e. woosley , the astrophysical journal 576 ( 2002 ) 323 . m. rayet _
et al . , _ astronomy and astrophysics 298 ( 1995 ) 517 .
t. rauscher , f. k. thielemann and k. l. kratz , phys .
c 56 ( 1997 ) 1613 .
t. rauscher and f. k. thielemann , at .
data nucl .
data tables 75 ( 2000 ) 1 .
g. g. kiss _
et al . , _ physics letters b 695 ( 2011 ) 419 .
t. rauscher _ et al .
c 86 ( 2012 ) 015804 .
a. sauerwein _
et al . , _ phys .
c 84 ( 2011 ) 045808 .
l. netterdon _
_ nuclear physics a 916 ( 2013 ) 149 .
g. g. kiss _
et al . , _ physics letters b 735 ( 2014 ) 40
. j. glorius _
et al . , _ phys .
c 89 ( 2014 ) 065808 .
a. palumbo _
et al . , _ phys .
c 85 ( 2012 ) 035808 .
et al . , _ phys .
c 74 ( 2006 ) 025805 .
et al . , _ at .
data nucl .
data tables 99 ( 2013 ) 651 .
et al . , _ phys .
c 55 ( 1997 ) 1523 .
et al . , _ phys .
c 64 ( 2011 ) 065805 .
d. galaviz _ et al .
c 71 ( 2005 ) 065802 . g. g. kiss _
et al . , _
european physical journal 27 ( 2006 ) 197 .
g. g. kiss _
et al . , _ journal of physics g 35 ( 2008 ) 014037
. g. g. kiss _
et al . , _ phys . rev .
c 80 ( 2009 ) 045807 . g. g. kiss _
et al . , _ phys . rev .
c 83 ( 2011 ) 065807 . g. g. kiss _
et al . , _ phys .
c 88 ( 2013 ) 045804 .
d. galaviz , phd thesis , 2006 technische universitt darmstadt h. abele and g. staudt , phys .
c 47 ( 1993 ) 742 . h. de vries , c. w. de jager and c. de vries , at .
data nucl .
data tables 36 ( 1987 ) 495 .
r. g. satchler and g. w. love , physics reports 55 ( 1979 ) 183 .
u. atzrott _
et al . , _ phys .
c 53 ( 1996 ) 1336 .
m. avrigeanu _
et al . , _ at .
data nucl .
data tables 95 ( 2009 ) 501 .
h. abele , university of tbingen , computer code dfold ( unpublished ) . v. chist _
et al . , _ phys .
c 54 ( 1996 ) 784 . l. mcfadden and g. r. satchler , nucl .
84 ( 1966 ) 177 .
t. rauscher and f. k. thielemann , at .
data nucl .
data tables 79 ( 2001 ) 47 .
g. g. kiss _
et al . , _ phys .
c 86 ( 2012 ) 035801 . m. avrigeanu and v. avrigeanu , phys .
c 82 ( 2010 ) 014606 .
v. avrigeanu _
et al . , _ phys .
c 90 ( 2014 ) 044612 . c. iliadis , _ nuclear physics of stars _ ( wiley , weinheim 2007 ) .
w. hauser and h. feshbach , phys .
87 ( 1952 ) 366 .
t. rauscher , int .
e 20 ( 2011 ) 1071 .
t. rauscher , the astrophysical journal supplement series 201 ( 2012 ) 26 .
p. mohr , phys .
c 84 , ( 2011 ) 055803 .
p. mohr , phys .
c 87 ( 2013 ) 035802 .
t. rauscher computer code smaragd , ( 2009 ) http://nucastro.org/smaragd.html t. rauscher , phys .
111 ( 2013 ) 061104 . | alpha elastic scattering angular distributions of the @xmath0cd(@xmath1)@xmath0cd reaction were measured at three energies around the coulomb barrier to provide a sensitive test for the @xmath2 + nucleus optical potential parameter sets .
furthermore , the new high precision angular distributions , together with the data available from the literature were used to study the energy dependence of the locally optimized @xmath2 + nucleus optical potential in a wide energy region ranging from e@xmath3 = 27.0 mev down to 16.1 mev .
the potentials under study are a basic prerequisite for the prediction of @xmath2-induced reaction cross sections and thus , for the calculation of stellar reaction rates used for the astrophysical @xmath4 process .
therefore , statistical model predictions using as input the optical potentials discussed in the present work are compared to the available @xmath0cd + alpha cross section data .
astrophysical @xmath4 process ; elastic alpha scattering experiments , @xmath2 + nucleus optical potential |
the invention of scanning tunneling microscopy ( stm ) @xcite was a milestone in experimental surface physics .
moreover , it became possible to tailor and analyze small nanostructures on various conducting surfaces @xcite .
perhaps the most spectacular and pioneering examples are the quantum corral experiments , in which closed atomic structures were assembled with help of atomic manipulations @xcite .
as the stm is a real space technique , and is very sensitive to the local atomic and electronic structures , it allows to study the properties of various , not necessarily periodic , structures with atomic resolution .
those include various surface reconstructions @xcite and low dimensional structures , like single adatoms @xcite , islands @xcite or one - dimensional monoatomic chains @xcite-@xcite .
in particular , the one - dimensional structures are very interesting from a scientific point of view , as they exhibit extremely rich phenomena , very often different from those in two and three dimensions @xcite . however , in reality all the one - dimensional chains always stay in contact with their neighborhood ( substrate , external electrodes , etc . ) , thus usually preventing the observations of the exotic physics . moreover , very often they contain various imperfections , like impurities , dislocations or lacks of atoms .
such a situation likely takes place when those monoatomic chains are fabricated in self - assembly processes .
the typical examples are all one - dimensional structures on vicinal si surfaces @xcite .
the electron transport through a chain in two terminal geometry , in which the end atoms of the chain were coupled to external electrodes has been extensively studied , both experimentally and theoretically ( see ref .
@xcite for a review ) .
a number of experiments has revealed many interesting phenomena , like conductance quantization in units of @xmath0 @xcite , deviations from that ( @xmath1 ( @xmath2 ) anomaly ) @xcite , spin - charge separation ( luttinger liquid ) @xcite , oscillations of the conductance as a function of the chain length @xcite or spontaneous spin polarization @xcite .
the problem of impurities or disorder in one - dimensional wires has also been studied both experimentally @xcite and theoretically @xcite-@xcite .
the theoretical studies revealed that even a single impurity can lead to a dramatic modifications of the low energy physics . in particular , the conductance of a wire with interacting impurity shows a power law behavior with the scaling exponent depending on the strength of the impurity @xcite .
the problem of impurities ( single atoms or clusters of atoms ) , but coupled sideways to a wire , has also been extensively studied recently @xcite-@xcite . for a single impurity with a strong coulomb interaction many authors predicted a suppression of the wire conductance due to the fano interference between ballistic ( wire ) channel and the impurity channel @xcite-@xcite . to be more precise
, the fano interference with the impurity reverses the gate voltage dependence of the conductance compared to the case when the impurity is embedded in a wire .
the fano effect in this case can have a classical nature , if the interference comes from the impurity single particle channel , or a many - body nature , when the resonant channel is formed by the kondo effect at the impurity .
the fano effect in a wire with a side coupled quantum dot has been recently confirmed experimentally @xcite .
it is the purpose of the present paper to see what modifications of the transport properties of a chain introduce a side coupled impurity . here ,
however , we shall study transport properties in different geometry , in which all the atoms in the chain as well as the impurity are coupled to one electrode and the second electrode is attached to one particular chain atom or the impurity via additional probing atom .
this geometry simply corresponds to stm tunneling through monoatomic chain with a side attached impurity , and can be related to previously mentioned stm study of self - assembled monoatomic chains on vicinal surfaces @xcite . to our knowledge
, the problem in this geometry has not been studied experimentally so far , except the effects of impurities on the length distribution of atomic chains @xcite .
our system is described by tight binding model with no electron correlations , as we are not interested in many body effects like kondo effect , metal - insulator transition , spin - charge separation , etc .
so our model can be applied to the wires where the interaction energy is smaller than the kinetic energy associated with the hopping along the wire .
for example , this could describe situation on various vicinal surfaces ( like si(335 ) , si(557 ) , etc . ) with monoatomic au chains grown on them , where the correlation effects are negligible due to small carrier concentration @xcite .
moreover , the problem can be solved exactly in this case . in order to calculate the tunneling current we have adapted a non - equilibrium keldysh green s function technique .
the organization of the rest of the paper is as follows . in sec .
[ model ] we introduce our model , and the results of the calculations and discussion are presented in sec .
[ results ] .
we end up with summary and conclusions .
our model system is shown schematically in fig . [ fig1 ] and is composed of a wire ( @xmath3 ) with atomic energies @xmath4 and hopping integrals @xmath5 between nearest neighbor atoms , impurity ( @xmath6 ) with single energy level @xmath7 , which is side - coupled to the wire via hopping @xmath8 .
both the wire and the impurity interact with the surface ( @xmath9 ) via @xmath10 and @xmath11 , respectively .
the surface is treated as a reservoir for electrons with the wave vector @xmath12 , the spin @xmath13 and single particle energies @xmath14 . above the wire
there is a stm tip ( @xmath15 ) modeled by a single atom with the energy level @xmath16 attached to another reservoir ( @xmath17 ) ( with electron energies @xmath18 ) via parameter @xmath19 .
tunneling of the electrons between stm tip and one of the atoms in a wire is described by the parameter @xmath20 .
the whole system is described by the following model hamiltonian @xmath21 where , as usually @xmath22 ( @xmath23 ) stands for the creation ( annihilation ) electron operator in the stm lead ( @xmath24 ) , tip atom ( @xmath25 ) , wire ( @xmath26 ) , impurity ( @xmath27 ) and the surface ( @xmath28 ) . in order to calculate the tunneling current from the stm electrode to the surface
we follow the standard derivations @xcite and get @xmath29 \ ; , \label{current}\end{aligned}\ ] ] where @xmath30 is the fermi distribution function , @xmath31 ( @xmath32 ) is the bias voltage , i.e. the difference between chemical potentials in the stm ( @xmath33 ) and the surface ( @xmath34 ) reservoirs , and the transmittance @xmath35 is given in the form @xmath36 with the coupling parameter latexmath:[$\gamma_{t(s)}(\omega ) = 2 \pi \sum_{{\bf k } \in s(t ) } the stm electrode ( @xmath17 ) and the tip atom ( @xmath15 ) and between the surface ( @xmath9 ) and the wire ( @xmath3 ) .
note that , to get the above expression for the transmittance we have assumed the same values of the coupling of the wire and the impurity with the surface , i.e. @xmath10 = @xmath38 .
@xmath39 ( @xmath40 ) is the fourier transform of the retarded green s function ( gf ) @xmath41_+ \rangle$ ] , i.e. the matrix element ( connecting the tip atom @xmath15 with @xmath42th atom in the chain or with the impurity @xmath6 ) of full gf , obtained from the solution of the equation @xmath43 the full gf @xmath44 is a @xmath45 matrix ( @xmath46 atoms in a chain , the impurity and the tip atom ) , which is obtained by inverting the matrix @xmath47 .
before the presentation of the numerical results , it is worthwhile to comment on choice of the model parameters used in the present work . in numerical calculations
we have assumed equal and energy independent coupling parameters ( @xmath48 ) , which reflects constant energy bands in both electrodes , and chosen @xmath49 as an energy unit .
the other parameters have been chosen in order to satisfy the realistic situation in many experiments .
the hopping integral within a wire is @xmath50 , @xmath51 , @xmath52 , and @xmath53 .
for example , taking @xmath54 ev , we get @xmath55 ev , @xmath56 ev , and @xmath57 ev . such a value of the parameter @xmath20 together with the typical value of the work function @xmath58 ev , gives a tip - surface distance @xmath59 @xcite , and the stm current stays in the range of a few na .
these are typical conditions in real stm experiments .
note that , with such a value of @xmath20 , the modifications of the wire density of states due to the stm tip are negligible . in the following
we will discuss the properties of wires containing even and odd number of atoms , as we expect different behavior for them , showing the results of the calculations for two representative examples , namely , for @xmath60 and @xmath61 atom wires .
however the conclusions drawn from consideration of @xmath60 ( @xmath61 ) atom wire remain valid for any even ( odd ) @xmath46 atom wire , provided the system is in ballistic regime .
moreover , we use the convention , in which the index @xmath42 labels the wire atoms , while @xmath62 indicates the position of the tip with respect to the wire atoms .
let us first discuss the modifications of the wire density of states ( dos ) due to the side - coupled impurity .
the local dos of @xmath42th atom in a wire is related to the corresponding diagonal element of the retarded gf , i.e. @xmath63 .
similarly , for the impurity @xmath64 .
figure [ fig2 ] shows a local dos of a wire consisted of @xmath60 ( left panels ) and @xmath61 atoms ( right panels ) in various impurity - wire configurations , indicated in the insets to the panels .
the top panels show the wire local densities of states in the case when there is no impurity attached , i.e. for clean wire . in this case , for @xmath46 atom wire the local dos on @xmath42th atom is the same as on @xmath65 atom ( @xmath66 ) , i.e. there is a symmetry with respect to the middle of the wire , thus the only non - equivalent @xmath67 are plotted . for @xmath60 atom wire
the @xmath67 has similar structure ( @xmath68 - thin solid line and @xmath69 - thicker solid line ) featuring small dos at the fermi energy @xmath70 and resonances corresponding to the wire atomic energies @xmath53 split by the hopping @xmath50 .
on the other hand , for @xmath61 atom wire there are large dos at the @xmath71 , and the resonances are eventually split by @xmath5 .
such a behavior is similar to the case of two terminal geometry in which the end wire atoms are connected to electrodes @xcite ,
i.e. a local minimum at the fermi energy @xmath70 for even number of atoms in a wire and local maximum at @xmath71 for odd atom wire and can be explained in terms of bonding , antibonding and nonbonding states @xcite . in case of even atom wire there
are always bonding ( @xmath72 ) and antibonding ( @xmath73 ) states .
when @xmath46 is odd , there is additional nonbonding state , which is situated at exactly the same position as the original atomic level ( @xmath74 ) , thus giving large dos at @xmath71 , as @xmath75 in our case .
interestingly , for the second ( or second from the end ) atom in an odd atom wire ( thicker solid line in the top right panel ) , the situation is quite different , namely , the local dos has very small value at the @xmath71 .
it turns out that this is a general tendency in odd atom wires , i.e. every second atom in an odd atom wire shows small dos at the fermi energy .
this could be understood in the following way .
let us forget for a moment about the stm tip , as we mentioned previously , that the stm tip does not influence the wire dos , and assume that the wire is coupled to the surface only .
if we calculated number of non - equivalent electron paths starting and ending in the surface and passing through a given atom , it turns out that if the number of odd paths ( passing through odd number of atoms ) @xmath76 is larger than the number of even paths ( enclosing even number of atoms ) @xmath77 , then the local dos has a resonance at @xmath71 .
thus it behaves as in the case of odd atom wire in two terminal geometry with the end atoms coupled to the leads .
for example , for the first atom in @xmath61 atom wire ( thin solid line in the right top panel of fig
. [ fig2 ] ) we get @xmath61 non - equivalent odd paths and @xmath60 even paths , thus @xmath78 and the resonance at the @xmath71 is produced . similarly , for the third atom , @xmath79 and @xmath80 , and again @xmath78 , thus we get the resonance at the fermi energy ( see thick solid line in the top left panel of fig .
[ fig2 ] ) .
opposite is also true , namely , if @xmath77 is larger than @xmath76 , the local dos has similar behavior as in the case of even atom wire , featuring small @xmath81 .
for example , @xmath82 and @xmath80 for the second atom in @xmath61 atom wire ( thicker solid line on the same picture as previously ) .
this seems to be true for any number of atoms in a wire in the present geometry .
of course , this is only intuitive picture , well working in the present case , namely , when the system is in ballistic regime , so the phase coherence length is longer than the wire length . in general , in the presence of interactions ( electron - electron , electron - phonon or scattering on magnetic impurities ) the phase coherence length will be suppressed , and this simple picture can be no longer valid . in this case one
has to calculate the contributions from different electron paths .
the middle and the bottom panels of fig .
[ fig2 ] show the local densities of states @xmath67 at different wire atoms in the case when the impurity with a single atomic level @xmath83 is attached to the first and second wire atom , respectively .
the presence of the impurity introduces asymmetry in the dos with respect to the middle of the wire , and the condition @xmath66 does not hold anymore , as a result @xmath67 will be different on each wire atom . in both wires
the main modification of the dos features additional resonance and thus splitting by @xmath84 of the resonance around @xmath85 ( @xmath86 ) in the case of @xmath60 ( @xmath61 ) atom wire .
the low energy behavior of dos is little affected , in particular for @xmath60 atom wire .
accidentally , it may a little bit shift the zero energy resonance , leaving small dos at @xmath71 , as it is seen in the middle right panel of fig .
this will be also reflected in the linear conductance , as we will see later .
in two terminal geometry , when the end wire atoms are connected to two different leads , the useful quantity is the total density of states , as in this case the conductance of the system is correlated with it @xcite . on the other hand , in the stm geometry we do not expect that the behavior of the conductance will be governed by this quantity . to see this , we plotted the total ( wire plus impurity ) density of states @xmath87 of a chain consisted of @xmath60 atoms ( left panel ) and @xmath61 atoms ( right panel of fig .
[ fig3 ] ) .
the total dos is defined as a @xmath88 for the wire with no impurity attached , and as a @xmath89 in the case when the impurity is present .
we have distinguished two cases , namely , when the impurity has the same value of the atomic energy @xmath90 as for the wire atoms , and when it has different energy @xmath83 .
the former case would correspond to the situation when the wire atoms and the impurity are composed of the same material , while in the later one , the impurity is a different atom .
first of all , if there is no impurity ( dashed lines ) the total dos shows similar behavior as in the case of the end wire atoms coupled to the leads only .
however one can notice different heights of particular resonances .
this reflects an effect , known from the studies of two impurities on a surface , and associated with different ( even - odd ) symmetries of electron states in resulting system @xcite .
when the impurity is introduced to the system , the total dos is strongly modified due to the parameter @xmath8 , which is responsible for the hoping between those subsystems . in this case
@xmath8 is only two times smaller than the wire hoping @xmath5 , thus one should expect modifications of the wire dos .
the impurity usually introduces additional resonance to the total dos . in both cases
the position of the resonance is around its atomic energy ( compare thin and thick solid lines in in the left panels of fig .
[ fig3 ] ) , and slightly modifies the resonances coming from the wire atoms .
this behavior only slightly depends on which wire atom is in close connection with the impurity .
let us now turn to the transmittance @xmath35 , defined by eq .
( [ transmit ] ) .
figure [ fig4 ] shows the transmittance of the system in various stm tip - wire - impurity configurations for @xmath60 atom ( left panels ) and @xmath61 atom wire ( right panels ) . while the transmittance depends on the stm tip position now , it is possible to identify the impurity induced contribution to @xmath35 .
the impurity introduces additional resonance to @xmath35 around its atomic energy , i.e. at @xmath91 ( compare the left panels of fig . [ fig4 ] ) .
in both cases , i.e. for @xmath60 and @xmath61 atom wires , the transmittance is correlated with the local density of states , and the presence of the impurity leads to similar modifications , especially for @xmath92 ( compare fig . [ fig2 ] ) , although the transmittance is now more asymmetrical .
this is in contradiction with the transport along the wire , as in that case the transmittance depends on the total dos . here
it depends on the local dos .
explanation is simple and intuitive . in the transport along the wire , all the local dos equally contribute to the transmittance @xcite . in the present case
, the transport takes place mainly through the wire atom , just below the stm tip , thus @xmath35 depends mainly on the local dos of a particular atom .
the presence of the impurity usually leads to higher values of @xmath35 for the resonances of the wire origin .
such a behavior is in contradiction with the behavior of @xmath35 when the transport takes place along the wire . in the former geometry
this leads to the reduction of the conductance @xcite-@xcite . here however , the situation is different , as the impurity is also connected to the same lead as the wire is , thus the impurity provides additional tunneling channel .
nevertheless , the zero energy transmittance is influenced by the impurity , similarly as the local dos ( compare fig .
[ fig2 ] ) . at this point
we would like to comment on the fano effect , as there are many channels for electron tunneling from stm tip to the surface .
even the tip is coupled to a single wire atom , the electron can leave the wire entering the surface electrode through different wire atoms .
this is particularly well visible for stm tip placed above second atom in odd atom wire even without impurity attached ( thicker solid line in top right panel of fig .
[ fig4 ] ) . in this case
odd number atoms in the wire have large density of states at the fermi energy , while even atoms have small dos at @xmath71 ( see fig .
[ fig2 ] ) .
when the stm tip is placed above first ( thin solid line ) or third ( thickest solid line ) an electron can tunnel through undelaying atom directly to the surface because neighboring atoms have small dos at @xmath71 .
the fano effect is negligible in this case . on the countrary , when the tip is placed above second atom ,
which has small dos at @xmath71 ( thicker solid line ) , the electron has three tunneling channels to the surface , due to large values of dos at neighboring atoms .
it can directly tunnel to the surface or go through first or third wire atom . in this case
the fano effect is enhanced and is visible as a dip at the fermi energy .
the presense of impurity slightly modifies the above picture , usually enhancing the fano effect when the tip is above the wire atom with attached impurity .
compare thin solid line in midle right panel of fig .
[ fig4 ] and thicker solid line in the bottom right panel .
we do not observe fano effect for even atom wire because dos at @xmath71 is always small ( see fig .
[ fig2 ] ) .
similar effects are reflected in differential conductance @xmath93 , shown in fig .
[ fig5 ] .
again , the presence of the impurity leads to a increase of @xmath94 around @xmath95 ( note that @xmath96 and @xmath32 ) , thus can help us in studying of the properties of impurities attached to wire in real stm experiments .
it also modifies the structures coming from the wire atoms , also leading to larger values of @xmath94 .
note however , that the values of @xmath94 are much smaller ( @xmath97 for @xmath60 atom and @xmath98 for @xmath61 atom wire ) than the conductance unit , i.e. @xmath99 per tunneling channel .
this is due to small value of the stm tunneling parameter @xmath20 . for @xmath100 ,
the conductance reaches the unitary limit , i.e. @xmath101 in our case ( two spin channels ) . on the other hand ,
the linear conductance , i.e. @xmath102 strictly follows the local density of states , similarly as the zero energy transmittance ( compare fig .
[ fig5 ] and figs .
[ fig2 ] and [ fig4 ] ) . in a special case
when the impurity is coupled to the first atom in @xmath61 atom wire , the zero bias maximum is split and the linear transport is strongly reduced for any wire atom ( compare the middle right panel of fig . [ fig2 ] ) .
so far we have discussed the various configurations , in which impurity was coupled to a single atom wire .
this would correspond to a situation in which the impurity is placed close to one particular wire atom .
now , the question arises what will happen if the impurity is side - coupled to two wire atoms , i.e. is placed aside between two wire atoms .
figure [ fig6 ] shows the conductance changes due to different couplings of the impurity to one atom ( top panel ) and two atoms in a wire ( bottom panel ) .
both panels show the conductance maps vs. bias voltage @xmath103 and corresponding couplings @xmath104 ( @xmath105 ) ( see the insets to the figure ) . in first configuration ( top panel )
the impurity is coupled to one wire atom , and the differential conductance has a maximum in the plane ( @xmath106 ) .
the position of the maximum changes linearly with @xmath103 and @xmath104 . on the other hand , in second configuration ,
when the strength of the impurity coupling changes ( bottom panel ) , i.e. for various positions of the impurity between second and third wire atom , we see that the largest values of @xmath94 are obtained when the impurity is close to the second atom ( @xmath107 , @xmath108 ) and slightly smaller close to the third wire atom ( @xmath109 , @xmath110 ) . for intermediate values of @xmath104 ( @xmath105 ) the conductance
is strongly reduced , in particular , when @xmath111 .
this is due to the fano effect , which is further enhanced in this case , and leads to similar reduction of @xmath94 in the case of two terminal geometry with the end wire atoms coupled to the leads @xcite-@xcite .
the fact , that the maximal values of the differential conductance show asymmetry with respect to @xmath111 point , i.e. they are different for the impurity near the second wire atom and near the third wire atom steams from the fact that stm tip is placed above the second wire atom .
finally we would like to comment on the coulomb interactions , as the present model completely neglects them .
we expect qualitative modifications of the results ( like even - odd oscillations of density of states ) , especially at low temperatures where the coulomb blockade and the kondo effect take place .
the picture can change drastically @xcite .
the tunneling can be enhanced when the wire energy levels are placed below the fermi energy and the wire is strongly coupled to the surface ( kondo effect ) or suppressed when it is weakly coupled , leading to the coulomb blockade .
the presense of additional tunneling channel ( due to the kondo effect ) can also enhance the fano effect .
on the other hand , we do not expect qualitative modifications in the mixed valence and the empty regimes , i.e. when the wire energy levels are above the fermi energy .
the present model can be succesfully applied to study monoatomic ( au , ag , pb ) chains grown on various vicinal surfaces @xcite-@xcite .
in conclusion we have studied stm tunneling through a quantum wire with a side - coupled impurity in various coupling configurations , i.e. the impurity was coupled to various atoms in a wire , and moreover , it was also connected to more than single wire atom .
we have found that the impurity strongly modifies the transport properties . in particular
it always produces a resonance around its atomic energy , which can be seen in differential conductance vs. bias voltage .
moreover , we have also shown that if the impurity is side - coupled to two wire atoms with equal strength , it leads to the suppression of the conductance , which is a hallmark of the fano effect in such a system .
those studies could be potentially useful in stm studying of the impurity induced modifications of the wire properties .
this work has been supported by the grant no n n202 1468 33 of the polish ministry of education and science .
in this appendix we give analytical solutions / relations for the local density of states ( ldos ) of a wire disturbed by an adatom .
ldos is connected with the retarded green s function by the following relation @xmath112 where @xmath113 can be obtained from the equation of motion for green functions .
using eq . 2 one can write @xmath114 where @xmath115 is the algebraic complement of the matrix @xmath116 ( cofactor ) . in our calculations
we assume the same coupling strengths between atoms , @xmath5 , and the same single particle energies @xmath4 for all atoms in the wire .
it is worth noting that the stm is weakly coupled with the wire and thus it does not affect the wire density of states ( in hamiltonian eq .
( [ hamilt ] ) we put @xmath117 ) .
first we consider the case of a linear wire without an adatom i.e. @xmath118 . in this case
ldos can be obtained from the relation @xmath119 where @xmath120 and @xmath121 , @xmath122 . to obtain
ldos one needs to know the determinant of @xmath123 which can be expressed as follows @xmath124 the matrix @xmath125 corresponds to an isolated wire ( non - coupled with the surface ) and is a tridiagonal one , @xmath126 , and can be expressed analytically in terms of chebyshev polynomials of the second kind @xcite .
the ldos for a wire coupled with an adatom can be obtained from the relation @xmath127 where @xmath128 and @xmath129 is a vector describing the couplings adatom - qw and adatom - surface , @xmath130 ( @xmath131-th atom of a wire is connected with the adatom .
after some algebra the determinant of the above matrix can be written in the form @xmath132 similar equation one can write for @xmath133 and @xmath134 which also can be expressed in terms of @xmath135 but have more complicated structure .
for the case @xmath136 ( there is no surface under the wire or the coupling wire - surface is very weak ) one can easily find @xmath137 and the ldos can be obtained fully analytically .
it is worth noting that the determinants @xmath138 , @xmath139 and @xmath140 are obtained for arbitrary @xmath131 ( connection adatom - qw atom ) and @xmath46 ( even or odd ) .
the minima of @xmath141 determine high value of ldos and the conductance through the system .
g. binnig , h. rohrer , ch .
gerber , e. weibel , appl .
. lett . * 40 * , 178 ( 1982 ) ; phys .
lett . * 49 * , 57 ( 1982 ) .
g. a. d. briggs , a. j. fisher , surf .
rep . * 33 * , 1 ( 1999 ) .
w. a. hofer , a. s. foster , a. l. shluger , rev .
phys . * 75 * , 1287 ( 2003 ) .
m. f. crommie , c. p. lutz , d. m. eigler , science * 262 * , 218 ( 1993 ) ; e. j. heller , m. f. crommie , c. p. lutz , d. m. eigler , nature * 369 * , 464 ( 1994 ) .
m. jaochowski , prog .
sci . * 74 * , 97 ( 2003 ) .
m. krawiec , m. jaochowski , m. kisiel , surf .
* 600 * , 1641 ( 2006 ) .
f. j. himpsel , k. n. altmann , r. bennewitz , j. n. crain , a. kirakosian , j. -l .
lin , j. l. mcchesney , j. phys . : condens .
matter * 13 * , 11097 ( 2001 ) .
j. n. crain , j. l. mcchesney , fan zheng , m. c. gallagher , p. c. snijders , m. bissen , c. gundelach , s. c. erwin , f. j. himpsel , phys . rev .
* b69 * , 125401 ( 2004 ) .
m. jaochowski , m. strak , r. zdyb , appl .
sci . * 211 * , 209 ( 2003 ) .
m. krawiec , t. kwapiski , m. jaochowski , phys .
( b)242 * , 332 ( 2005 ) ; phys . rev . * b73 * , 075415 ( 2006 ) .
j. m. luttinger , j. math . phys . * 4 * 1154 ( 1963 ) ; f. d. m. haldane , j. phys .
c : solid state phys .
* 14 * 2585 ( 1981 ) ; j. voit , rep .
. phys . * 58 * , 977 ( 1995 ) .
n. agrait , a. levy yeyati , j. m. van ruitebeek , phys .
* 377 * , 81 ( 2003 ) .
b. j. van wees , h. van houten , c. w. j. beenakker , j. g. williamson , l. p. kouwenhoven , d. van der marel , c. t. foxon , phys .
lett . * 60 * , 848 ( 1988 ) .
k. j. thomas , j. t. nicholls , m. y. simmons , m. pepper , d. r. mace , d. a. ritchie , phys .
lett . * 77 * , 135 ( 1996 ) . o. m. auslaender , a. yacoby , r. de picciotto , k. w. baldwin , l. n. pfeiffer , k. w. west , phys .
84 * , 1764 ( 2000 ) .
o. m. auslaender , h. steinberg , a. yacoby , y. tserkovnyak , b. i. halperin , k. w. baldwin , l. n. pfeiffer , k. w. west , science * 308 * , 88 ( 2005 ) . c. j. muller , j. m. van ruitenbeek , l. j. de jongh , phys .
lett . * 69 * , 140 ( 1992 ) .
r. h. m. smit , c. untiedt , g. rubio - bollinger , r. c. segers , j. m. van ruitenbeek , phys .
lett . * 91 * , 076805 ( 2003 ) .
b. e. kane , g. r. facer , a. s. dzurak , n. e. lumpkin , r. g. clark , l. n. pfeiffer , k. w. west , appl .
* 72 * , 3506 ( 1998 ) .
w. h. a. thijssen , d. marjenburgh , r. h. bremmer , j. m. van ruitenbeek , phys .
* 96 * , 026806 ( 2006 ) .
a. luther , i. peschel , phys . rev .
* b9 * , 2911 ( 1974 ) . t. giamarchi , h. j. schulz , phys . rev .
* b37 * , 325 ( 1988 ) . f. dolcini , h. grabert , i. safi , b. trauzettel , phys .
lett . * 91 * , 266402 ( 2003 ) ; f. dolcini , b. trauzettel , i. safi , h. grabert , phys . rev .
* b71 * , 165309 ( 2005 ) .
t. enss , v. meden , s. andergassen , x. barnab - thriault , w. metzner , k. schnhammer , phys . rev . * b71 * , 155401 ( 2005 ) .
a. agarwal , d. sen , phys . rev . *
b73 * , 045332 ( 2006 ) .
m. m. fogler , s. v. malinin , t. nattermann , cond - mat/0602008 .
k. kang , s. y. cho , j .- j.kim , s .- c .
shin , phys . rev .
* b63 * , 113304 ( 2001 ) .
m. e. torio , k. hallberg , a. h. ceccatto , c. r. proetto , phys .
b65 * , 085302 ( 2002 ) ; m. e. torio , k. hallberg , s. flach , a. e. miroshnichenko , m. titov , eur .
j. * b37 * , 399 ( 2004 ) ; m. e. torio , k. hallberg , c. r. proetto , cond - mat/0404146 .
a. a. aligia , c. r. proetto , phys . rev . *
b65 * , 165305 ( 2002 ) .
r. franco , m. s. figueira , e. anda , phys .
rev . * b67 * , 155301 ( 2003 ) .
p. stefaski , solid state commun . * 128 * , 29 ( 2003 ) .
k. kobayashi , h. aikawa , a. sano , s. katsumoto , y. iye , phys .
rev . * b70 * , 035319 ( 2004 ) .
t. kwapiski , vacuum * 74 * , 201 ( 2004 ) .
t. kwapiski , j. phys . : condens .
matter * 17 * , 5849 ( 2005 ) ; _ ibid _ * 18 * , 7313 ( 2006 ) .
v. puthier , c. girarded , surf .
sci . * 511 * , 203 ( 2002 ) .
a. orellana , f. domingues - adame , i. gomez , m. l. ladron de guevara , phys
* b67 * , 085321 ( 2003 ) .
k. s. thygesen , m. v. bollinger , k. w. jacobsen , phys . rev . *
b67 * , 115404 ( 2003 ) .
j. n. crain , m. d. stiles , j. a. stroscio , d. t. pierce , phys .
lett . * 96 * , 156801 ( 2006 ) .
j. n. crain , a. kirakosian , k. n. altmann , c. bromberger , s. c. erwin , j. l. mcchesney , j. -l .
lin , f. j. himpsel , phys .
* 90 * , 176805 ( 2003 ) .
h. haug , a. p. jauho , _ quantum kinetics in transport and optics of semiconductors _
( springer , berlin , 1996 ) .
m. krawiec , k. i. wysokiski , solid state commun .
* 115 * , 141 ( 2000 ) ; semicond .
* 17 * , 103 ( 2004 ) .
y. calev , h. cohen , g. cunibereti , a. nitzan , d. porath , isr . j. chem . * 44 * , 133 ( 2004 ) . f. liu , s. n. khanna , p. jena , phys . rev . *
b42 * , 976 ( 1990 ) .
z. y. zeng , f. claro , phys . rev . *
b65 * , 193405 ( 2002 ) .
m. krawiec , t. kwapiski , surf .
* 600 * , 1697 ( 2006 ) .
d. m. newns , n. read , adv .
36 * , 799 ( 1987 ) . | the stm tunneling through a quantum wire ( qw ) with a side - attached impurity ( atom , island ) is investigated using a tight - binding model and the nonequilibrium keldysh green function method .
the impurity can be coupled to one or more qw atoms .
the presence of the impurity strongly modifies the local density of states of the wire atoms , thus influences the stm tunneling through all the wire atoms .
the transport properties of the impurity itself are also investigated mainly as a function of the wire length and the way it is coupled to the wire .
it is shown that the properties of the impurity itself and the way it is coupled to the wire strongly influence the stm tunneling which is reflected in the density of states and differential conductance . quantum wire , tunneling , stm 68.27.ef , 81.07.vb , 73.40.gk |
limited device battery life has always been a key consideration in the design of modern mobile wireless technologies .
frequent battery replacement / recharging is often costly due to the large number of wireless devices in use , and even infeasible in many critical applications , e.g. , sensors embedded in structures and implanted medical devices .
rf - enabled wireless energy transfer ( wet ) technology provides an attractive solution by powering wireless devices with continuous and stable energy over the air . by leveraging the far - field radiative properties of electromagnetic ( em ) wave , wireless receivers could harvest energy remotely from the rf signals radiated by the energy transmitter .
rf - enabled wet enjoys many practical advantages , such as long operating range , low production cost , small receiver form factor , and efficient energy multicasting thanks to the broadcast nature of em wave .
one important application of rf - enabled wet is _ wireless powered communication _ ( wpc ) , where the wireless devices use the harvested rf energy to transmit / decode information to / from other devices . without being interrupted from energy depletion due to communication usage
, wpc is expected to improve user experience and convenience , with a higher and more sustainable throughput performance than conventional battery - powered communication .
wpc can also be applied in sensors with much lower maintenance cost and enhanced flexibility in practical deployment . due to the high attenuation of microwave energy over distance , rf - enabled wet
is commonly used for supporting low - power devices , such as rfid ( radio - frequency identification ) tags and sensors .
however , the recent advances of antenna technologies and rf energy harvesting circuits have enabled much higher microwave power to be efficiently transferred and harvested by wireless devices @xcite .
therefore , we envision in fig .
@xmath0 that wpc will be an important building block of many popular commercial and industrial systems in the future including the upcoming internet of things / everything ( iot / ioe ) systems consisting of billions of sensing / rfid devices as well as large - scale wireless sensor networks ( wsns ) .
we also envision that rf - enabled wet is key component of the last - mile " power delivery system , with the smart electrical power gird forming the backbone " or core power network . before proceeding to the discussion of rf - enabled wet / wpc
, it is worth pointing out its relation to another green communication technique , namely , _ energy harvesting _ ( eh ) , where wireless devices harness energy from energy sources in environment not dedicated to power the wireless devices , such as solar power , wind power , and ambient em radiations . unlike rf - based eh from ambient transmitters , the energy source of wet is stable , and more importantly , fully controllable in its transmit power , waveforms , and occupied time / frequency dimensions , to power the energy receivers . with a controllable energy source , a _ wireless powered communication network _ ( wpcn ) could be efficiently built to power multiple communication devices of different physical conditions and service requirements .
besides , with rf - enabled wet , information could also be jointly transmitted with energy using the same waveform .
such a design paradigm is referred to as _ simultaneous wireless information and power transfer _ ( swipt ) , which is proved to be more efficient in spectrum usage than transmitting information and energy in orthogonal time or frequency channels @xcite . in this article , we first provide a brief overview of state - of - the - art rf - enabled wet technologies . then , we focus on introducing rf - enabled wpc in the following three topics : * the circuit model and advanced signal processing techniques used for wet ; * the design tradeoffs in joint energy and information transmission for swipt ; * the design challenges and opportunities in wpcn . at last ,
we discuss the future research directions of wpc and conclude the article .
although wet has gained popularity in the recent years , it is in fact a technology under development for more than a century ( see @xcite for the detailed historical development of wet ) .
the existing wet technologies could be categorized into three classes based on the key physical mechanisms employed : inductive coupling , magnetic resonant coupling and em radiation . among them , the first two types exploit the non - radiative near - field em properties associated with an antenna for short - range high power transfer .
currently , inductive coupling wet is well - standardized , with applications such as charging mobile phone and medical implanted devices .
however , due to the drastic drop of magnetic induction effect over distance , inductive coupling typically operates within only several centimeters range
. the operating range of magnetic resonant coupling wet could be as large as few meters .
however , to maintain the resonant coupling , the receiver could not be flexibly positioned as it is optimized for some fixed distance and circuit alignment settings . besides , transmitting energy to multiple receivers is challenging as it requires careful tuning to avoid interference due to mutual coupling effect . on the other hand ,
rf - enabled wet exploits the far - field radiative properties of em wave to power wireless devices over moderate to long distance .
a typical rf - powered rfid tag could be powered at @xmath1 meters away ( with around @xmath2 mw received rf power ) , and some rf energy harvesting chips have a maximum @xmath3-@xmath4 meters line - of - sight operating radius ( with around @xmath5 mw received rf power ) . in practice , rf - enabled wet uses inexpensive rf energy receivers , which could be flexibly positioned and made very tiny to fit into commercial devices . besides
, transmitting energy to multiple receivers is easily achieved with the broadcasting property of microwaves .
the major constraint of the application of rf - enabled wet is the high attenuation of microwave energy over distance . nonetheless , with the continuing decrease of device operating power ( as low as a few microwatts for some rfid tags ) , and the recent application of mimo technology that significantly enhances the wireless energy transfer efficiency
, we could expect more and more important applications of rf - enabled wet in the future .
@xmath6 , we present a network model to illustrate the basic concepts of wpc . in the downlink ( dl ) , wet - enabled energy access points ( aps ) with stable power supply ( e.g. , ap2 in fig .
@xmath6 ) transmit energy to a set of distributed wireless devices ( wds ) .
meanwhile , the wds could use the harvested energy to transmit / receive information to / from the information aps ( e.g. , ap3 in fig .
@xmath6 ) in the uplink ( ul ) and downlink , respectively . besides , energy and information aps could be integrated into a co - located energy / information ap ( e.g. , ap1 in fig .
@xmath6 ) , which both transmits energy and provides data access to the wds . in particular , three canonical operating modes are specified as follows : * wet : energy transfer in the dl only , e.g. , ap1 to wd1 and ap2 to wd5 ; * swipt : energy and information transfer in the dl , e.g. , ap1 to wd4 ; * wpcn : energy transfer in the dl and information transfer in the ul , e.g. , ap1 to wd3 . accordingly ,
the wds all perform energy harvesting ( eh ) in the wet mode , with applications such as charging sensors for sensing operations .
additionally , the wds perform eh in the dl transmission of wpcn mode while sending data in the ul with harvested energy , with applications such as sensor battery charging and data collection in a wsn @xcite . for the swipt mode ,
the wds perform both eh and information decoding ( i d ) in the dl with the same received signals , each using harvested energy to power its information decoder , e.g. , in an energy - self - sustainable information broadcast network @xcite . in practice
, a wpc network could also include other general network models consisting of multiple co - located or separated information / energy transmitters , and receivers with heterogeneous operating modes .
for instance , wd6 harvests energy from an energy ap ( ap2 ) and transmits data to a different information ap ( ap3 ) ; ap1 transmits energy and information respectively to two separated energy and information receivers at the same time , i.e. , wd1 and wd2 . besides
, energy transmission could potentially generate interference to the information receivers operating in the same frequency band ( see the red dashed lines in fig .
@xmath6 ) , for which a joint design of energy / information transmissions is highly desired .
in the following sections , we will focus our discussions on the three canonical operating modes , i.e. , wet , swipt and wpcn , respectively .
in this section , we introduce the rf energy receiver structure and advanced signal processing methods to enhance the energy transfer efficiency of rf - enabled wet .
we consider in fig .
@xmath7 an energy transmitter transferring rf energy to multiple energy receivers ( ers ) , where each transmitter / receiver is equipped with multiple antennas in general .
the transmitted energy signals are in general modulated signals , e.g. , pseudo - random signals , instead of an unmodulated sinusoid tone that is used widely in practice .
the signals could be designed to avoid a spike in the power spectral density ( psd ) , and to satisfy the psd requirement for safety and interference considerations .
therefore , wireless energy transmission occupies a certain bandwidth similar to information transmission , determined by the modulated baseband signal . , the received rf signal is converted to a dc to charge a battery.,scaledwidth=60.0% ] the rf energy harvesting ( eh ) circuit model is also depicted for er @xmath8 in fig .
@xmath7 . the eh receiver
is based on a _ rectifying circuit _ , consisting of a diode and a passive low - pass filter ( lpf ) , which converts the received rf signal to a dc ( direct current ) signal to charge the built - in battery to store the energy . by the law of energy conservation , the harvested energy per unit symbol time at an
er , denoted by @xmath9 , is proportional to the received rf power @xmath10 , i.e. , @xmath11 here , @xmath12 denotes the overall receiver energy conversion efficiency , @xmath13 denotes the transmit power , @xmath14 denotes the distance ( normalized with respect to a given reference distance ) from the er to the transmitter , @xmath15 denotes the path loss factor and @xmath16 denotes the combined antenna gain of the transmit and receive antennas . for instance , using two antennas at both the energy transmitter and the receiver , we could achieve a beamforming gain to increase the harvested energy by about @xmath1 times ( @xmath17 db ) compared to the case with single - antenna transmitter and receiver with the same transmitter power .
this could be more cost - effective in practice than the alternative approach of improving the energy conversion efficiency @xmath18 ( say , from @xmath19 to @xmath20 ) at the receiver with more sophistically designed rectifying circuits . using multiple antennas not only provides antenna power gain as stated above , but also enables advanced _ energy beamforming _ ( eb ) techniques to focus the transmit power in a smaller regions of space to bring significant improvement to the energy transfer efficiency @xcite . by carefully shaping the transmit waveform at each antenna
, eb could control the collective behavior of the radiated waveforms , such that they are combined coherently at a specific receiver , but destructively at the others .
in general , the larger number of antennas installed at the energy transmitter , the sharper the energy beam that could be generated towards a particular spatial direction . with only one er , the transmitter could steer a single sharp beam to maximize the harvested energy . when there are multiple er as in fig .
@xmath7 , however , generating a single beam may result in severe unfairness among the receivers , also known as the ( energy ) near - far problem , where users near the transmitter harvest much more energy than the far users . in this case , the transmitter may need to generate multiple energy beams towards different directions to balance the energy harvesting performance among the receivers @xcite .
an efficient eb design requires availability of accurate knowledge of the channel state information at the transmitter ( csit ) .
however , this is often difficult to achieve in practice . on one hand
, many simple energy receivers have no baseband signal processing capability to perform channel estimation . on the other hand ,
accurate channel estimation consumes significant amount of time and energy , which may offset the energy gain obtained from a refined eb . besides
, receiver mobility could cause time - varying channels , making channel tracking difficult .
various efficient channel estimation methods have been proposed to perform eb under imperfect csit by exploiting the received energy levels over time and balancing the tradeoff between energy consumption and eb gain ( see , e.g. , @xcite and the references therein ) . besides , robust eb design could be pursued to generate energy beams based on the statistical knowledge of csit .
another promising method is to perform eb using distributed antennas .
this effectively reduces the amount of feedback signals for channel estimation , as a receiver harvests energy only from a small subset of nearby transmitting antennas .
besides , the deployment of distributed antennas also reduces the range between the energy transmitters and receivers , thus is effective to solve the near - far problem caused by using a single energy transmitter . in this case , however , efficient coordination among the distributed antennas is needed .
when wet is applied to power communication devices , it will inevitably occupy part of the spectrum used for communication purpose . to avoid the interference to communication ,
a simple but spectrally inefficient method is to transmit energy and information in orthogonal frequency channels . alternatively ,
swipt designs seek to save the spectrum by transmitting information and energy jointly using the same waveform .
this is intuitively achievable , as any waveform for information transmission also carries energy to be harvested by the same or different receivers .
however , an efficient swipt scheme involves a rate - energy tradeoff in both the transmitter and receiver designs to balance the information decoding ( i d ) and energy harvesting ( eh ) performance . in fig .
@xmath21 , we illustrate a swipt network with a multi - antenna hybrid access point ( hap ) transmitting energy and information jointly to multiple receivers ( rxs ) . some of the receivers receive only information ( rx @xmath17 ) or only harvest energy ( rx @xmath22 ) , while some harvest both energy and receive information simultaneously ( rx @xmath23 ) . it is worth pointing out that a typical i d receiver and eh receiver operate with rather different power sensitivities ( e.g. , @xmath24 dbm for eh receivers versus @xmath25 dbm for i d receivers )
. therefore , eh receivers are in general closer to the transmitter than i d receivers for effective energy reception . at the transmitter side ,
the waveforms generated by the hap directly determine the performance of information and energy transfer .
in the extreme case , the hap could ignore the energy ( information ) receivers and optimize the waveforms only to maximize the information ( energy ) transmission efficiency .
however , due to the fundamental difference in the optimal waveforms for information and energy transmissions , such an off - balance design may lead to poor performance of either information or energy transmission .
in general , the waveform design needs to follow a _ rate - energy tradeoff _ to achieve the best possible balance between the two objectives @xcite .
meanwhile , the characterization of rate - energy tradeoff is closely related to the receiver structure and the corresponding signal processing strategies @xcite .
an ideal swipt receiver is assumed to be able to decode information and harvest energy from the same signal @xcite ; however , this could not be realized by practical circuits .
some practical receiver structures are plotted in fig .
@xmath21 , namely , time switching ( ts ) , power splitting ( ps ) , integrated id / eh receiver ( intrx ) , and antenna switching ( as ) @xcite , and will be specified in the sequel along with the respective rate - energy tradeoff characterization . for the simplicity of illustration , we consider each pair of transmitter and receiver separately to discuss the rate - energy tradeoff in a point - to - point channel . in this case , the rate - energy tradeoff is often characterized by the boundary of the _ rate - energy region _ , defined as the union of all the achievable rate - energy pairs by the receiver .
then , any point on the boundary specifies the maximum achievable data rate given a harvested energy requirement .
this corresponds to rx @xmath26 in fig .
a ts receiver consists of co - located i d and eh receivers , where the i d receiver is a conventional information decoder and the eh receiver s structure follows that in fig . @xmath7 . in this case , the transmitter divides the transmission block into two orthogonal time slots , one for transferring power and the other for transmitting data . at each time slot
, the transmitter could optimize its transmit waveforms for either energy or information transmission .
accordingly , the receiver switches its operations periodically between harvesting energy and decoding information between the two time slots .
then , different r - e tradeoffs could be achieved by varying the length of energy transfer slot .
this corresponds to rx @xmath27 in fig .
@xmath1 . the eh and i d receiver components of a ps receiver are the same as those of a ts receiver .
however , the hap can not optimize transmitted signals only for information or energy . instead , the ps receiver splits the received signal into two streams , where one stream with power ratio @xmath28 is used for eh and the other with power ratio @xmath29 is used for i d .
different r - e tradeoffs are achieved by adjusting the value of @xmath30 .
this corresponds to rx @xmath8 in fig .
unlike the ts and ps receivers that split the signal at the rf band , an intrx combines the rf front - ends of i d and eh receivers and splits the signal after converting it into dc current .
then , the dc current is divided into two streams for battery charging and information decoding , respectively .
intrx uses a ( passive ) rectifier for rf - to - baseband conversion , which saves the circuit power consumed by the active mixer used in the information decoder of ts / ps receivers .
however , the i d receiver of intrx needs to perform noncoherent detection from the baseband signal ( dc current ) . in this case , conventional phase - amplitude modulation ( pam ) must be replaced by _ energy modulation _
, where information is only encoded in the power of the input signal resulting a reduction of capacity @xcite . however , intrx is superior than the ps / ts receivers when more harvested energy is required , because active frequency down conversion is not performed . in fig .
@xmath31 , we give an example to illustrate the key characteristics of the rate - energy regions of the three practical receivers and the ideal receiver in a point - to - point additive white gaussian noise ( awgn ) channel .
we could see that , the rate - energy region of the ideal receiver is a box , thus no design tradeoff is involved in this case .
a similar rate - energy region is also observed for intrx .
this is because the optimal strategy is to use infinitesimally small amount of dc current for i d and the remaining dc current for eh .
the rate - energy region of ts receiver is a straight line connecting the two optimal operating points for eh and i d , respectively .
compared to ts receiver , ps receiver has a strictly larger rate - energy region .
so far , the optimal eh - id receiver is not known .
it is unclear if the non - trivial rate - energy region between the ideal and the introduced practical receivers could be achieved or not , which is left for future exploration likely involving different domains , such as physics , circuit theory and information theory .
channel fading and interference are two major challenges to the transceiver design of wireless communication .
however , they result in quite different performance degradation for i d and eh in swipt .
while deep channel fading degrades the performance of both i d and eh , strong interference is only harmful to i d but in fact could be helpful to increase the harvested energy for eh receiver . to optimize the eh and i d performance ,
the receiver could adapt its strategy to the channel conditions and interference level @xcite .
for instance , the ts rx @xmath26 in fig .
@xmath1 should switch to perform i d when the received signal ( information and interference ) is relatively weak and the signal - to - noise ratio ( snr ) is sufficiently high , and eh otherwise .
similarly , the ps rx @xmath27 should allocate more received power to i d receiver when the channel is in poor condition , and more power to eh receiver otherwise @xcite .
intuitively , this is because the gain achieved by i d receiver is less than the gain achieved by eh receiver when the interference is strong ( harmful for i d but helpful for eh ) , and when the channel is in good condition ( logarithmic increase in rate for i d but linear increase in energy for eh ) .
when csit is available , the transmitter could also adapt its transmit power to the channel state to achieve the maximum information and energy transfer efficiency , e.g. , it does not waste energy to transmit in the case of deep channel fading @xcite . the application of mimo technology could significantly mitigate the effect of channel fading for both energy transmission ( energy beamforming ) and information transmission ( spatial diversity and/or multiplexing ) @xcite . in a broadcast channel such as in fig .
@xmath21 , a multi - antenna hap could utilize the spatial degrees of freedom to focus the antenna radiation at specific locations , which not only enhances the harvested energy but also mitigates the interference to unintended information receivers . at the receiver side , the use of multiple antennas enables a low - complexity implementation for ps , named _ antenna switching _
( see rx @xmath1 in fig .
@xmath21 ) , which uses a subset of antennas for eh ( @xmath32 ) and the rest for i d ( @xmath33 ) . while ps requires a power splitter for each antenna , as reduces the hardware complexity by simply connecting an antenna to either an i d receiver or an eh receiver with an inexpensive switch .
the rate - energy region of as approaches that of a ps receiver when the number of receive antennas is large enough @xcite .
swipt could also be extended to other useful application scenarios . in fig .
@xmath21 , for instance , the hap could broadcast energy to the nearby receivers ( rx @xmath22 ) and transmit information to the faraway receivers ( rx @xmath17 ) simultaneously , while meeting the different sensitivities of eh and i d receivers .
besides , information secrecy could be achieved between the hap and information receivers using physical layer secrecy coding techniques @xcite . in addition , a relay node rx @xmath8 in fig . @xmath21 could harvest energy and receive the message dedicated to rx @xmath34 that is located further away from the hap , and then forward the message to rx @xmath34 in another time slot to extend the coverage of the hap .
in swipt , the wireless devices use the harvested energy to decode the information sent to them . here , we consider another scenario that the wireless devices use the harvested energy to transmit information .
this communication architecture is referred to as wireless powered communication network ( wpcn ) @xcite . in this section ,
we introduce the basic operations of a wpcn , the key design challenges and solutions , and interesting extensions to many practical network models .
we consider a single - cell wpcn in fig .
@xmath35 , where a hap broadcasts energy to the multiple wireless devices in the dl , while the wireless devices communicate to the hap in the ul using the energy harvested . due to the half - duplex hardware constraint of hap
, the network operates under a two - phase _ harvest - then - transmit _ protocol within a transmission block of duration @xmath36 . specifically , the wireless devices harvest energy from dl wet in the first phase for @xmath37 ( @xmath38 ) amount of time , and then transmit data in the second phase for the rest of the transmission block .
this could be easily achieved by the time - switching circuit model shown for u@xmath8 in fig .
intuitively , with a larger @xmath39 , the ul data rate could be improved as the devices could harvest more energy in the first phase to transmit data .
however , a larger @xmath39 also decreases the data rate as it leaves a shorter data transmission time . in general , the optimal value @xmath39 that results in the highest ul throughput is related to the users wireless channel conditions .
if the users are all close to the hap , the optimal @xmath39 is small , as each user could still harvest sufficient amount of wireless energy within the short duration of dl wet .
otherwise , a larger @xmath39 is required for the far users to harvest sufficient energy before commencing reliable data transmission .
in fact , it is shown in @xcite that the optimal value of @xmath39 that maximizes the sum - rate decreases as the sum of channel power gains of all users increases . besides setting the optimal duration for wet in the first phase , another important issue in the harvest
- then - transmit protocol is to design efficient multiple access scheme in the second phase for coordinating ul information transmission of users . in conventional wireless system , users
that are far away from the base station in general achieve lower data rates than those in the vicinity .
this fairness issue is even more critical and challenging in a wpcn . due to significant signal attenuation ,
a user far away from the hap ( u@xmath27 ) harvests much lower wireless energy in the dl but consumes more to transmit data in the ul than a user near the hap ( u@xmath26 ) .
this coupled effect is referred to as the _ doubly - near - far problem _
@xcite , which could result in very low throughput for far users , e.g. , @xmath40 times less data rate than a nearby user , if the multiple access scheme is not properly designed .
when tdma is used in the second phase , the hap could allocate a longer data transmission time to the far users in the second phase to tackle the doubly - near - far problem .
on the other hand , when the hap has multiple antennas , _ spatial division multiple access _ ( sdma ) could be applied in the ul . in this case , all the users transmit simultaneously to the hap during the second phase and the hap jointly decodes the user messages using multi - user detection ( mud ) techniques .
sdma in general achieves higher spectrum efficiency than the tdma - based method . besides
, the doubly - near - far problem could be mitigated by user transmit power control in the ul and energy beamforming design in the dl .
specifically , the hap uses eb to steer stronger energy beams towards the far users , and allows them to transmit with a higher power than the near users to balance the throughput performance among all the users @xcite . another effective method to tackle
the doubly - near - far - problem is through user cooperation . in fig .
@xmath35 , after harvesting energy in the first phase , a nearby user u@xmath1 uses part of its resource ( transmit energy and time ) in the second phase to forward a far - away user s ( u@xmath8 ) messages to the hap @xcite .
the more resource u@xmath1 consumes for helping u@xmath8 , the more throughput improvement could be achieved for the far - away user .
due to the transmit time and energy constraints , the relay node u@xmath1 needs to carefully allocate the resource on relaying the other s message and transmitting its own message .
interestingly , it is shown in @xcite that both users could benefit from the cooperation .
for the far - away user , the reason is obvious as the cooperation essentially increases the time and energy used for its message transmission .
for the nearby user , its data rate loss due to cooperation could be made up by an overall longer data transmission time , because the gain from user cooperation allows the hap to allocate more time for data transmission , instead of wet .
the efficient operation of a wpcn is highly dependent on the accurate knowledge of channel state information ( csi ) at the hap , where both information decoding and resource allocation require accurate csi estimate . similar to a conventional wireless network
, the throughput performance of wpcn would benefit from a longer channel estimation period with more accurate csi estimate , but also suffer from a shorter energy / information transmission time .
however , a wpcn - specific design tradeoff arises due to the energy constraints at the wireless devices .
this is because the wireless devices consume energy for channel estimation on decoding the pilot signals sent by the hap , transmitting the csi feedback , or sending pilot signals to the hap in some channel estimation schemes that exploit the ul / dl channel reciprocity @xcite .
evidently , more energy consumption on channel estimation would compromise the transmission rate ( or reliability ) because less energy is left for communication .
however , this results in more accurate csi estimate and hence more precise beamforming that both improves its transmission rate ( or reliability ) and increases the harvested energy in the following transmission blocks . the energy of wireless devices for channel estimation and information transmission should thus be carefully allocated to achieve optimal performance .
another performance enhancing technique to wpcn is massive mimo , which employs a large number of antennas ( tens to more than a hundred ) at the hap to exploit the high antenna array gain .
on one hand , the large degree of freedom provided by massive mimo enables spatial multiplexing to serve more mobile users for ul information transmission at the same time . on the other hand , the large antenna array also improves the eb performance in the dl , by the generation of very sharp beams to enhance the received signal power ( e.g. , more than @xmath41 db power gain ) .
the application of massive mimo to wpcn could result in multiple folds of throughput improvement and also much longer operating range .
interestingly , the high number of antennas at hap does not necessarily translate to high processing complexity and cost .
the property of asymptotic orthogonality of large user - to - hap channels leads to largely simplified beamforming design , multiple access control and power control solutions @xcite .
more generally , there could be a dedicated wireless energy network consisting of multiple power nodes that broadcast energy by means of wet .
moreover , multiple information receivers may decode the data transmitted by the wireless devices in the ul .
in such a multi - cell wpcn ( i.e. , multiple energy transmitters and information receivers ) , it is complicated and often intractable to derive the network capacity using the resource allocation methods for single - cell capacity analysis . instead , the method of _ stochastic geometry _ is widely used to study the scaling laws of network capacity as a function of system parameters . in the context of a cellular wpcn
, @xcite proposes to install wireless energy transferring nodes , named power beacons ( pbs ) , to provide energy for the mobile users to transmit data to some base stations ( bss ) . based on a stochastic geometry model ,
it derives the functional relationships between the densities of bss and pbs as well as their transmission power to achieve a prescribed communication outage probability .
the application of wpcn is also exploited in cognitive radio network in @xcite , where a secondary transmitter ( e.g. , sensor ) could harvest energy from a primary transmitter ( e.g. , mobile phone ) if they are close enough , and transmit to its intended secondary receiver if it is sufficiently far from any primary transmitter , to avoid potential interference to the primary network .
other than the extensions discussed above , wpcn could also be applied to many wireless systems with energy constrained wireless devices .
for instance , multi - hop communications with energy harvesting relays , systems with densely deployed haps using millimeter - wave technologies , and distributed antenna systems with coordinating energy / information beamforming , etc .
wpc contains rich research problems of important applications yet to be studied . in this section .
we highlight several interesting research topics we deem particularly worth investigating . due to the critical power constraints of wireless devices ,
the future wireless system is expected to be a mixture of wireless energy and communication networks . under spectrum scarcity
, it is likely that the two networks operate on overlapped spectrum .
this raises the problem for wireless energy and communication networks to coexist . unlike the two - way interference in conventional multi - cell communication systems ,
the interference is one - way from energy network to communication network .
besides , the highly different sensitivity of information and energy receivers indicates that the interference due to wet is in general much stronger than information signals .
there are many promising solutions to mitigate the interference , such as information / energy transfer scheduling , energy beamforming design , and opportunistic wet based on spectrum sensing , etc . in particular ,
cognitive radio technology could be used to carry out effective spectrum sensing to minimize the interference from wet to communication network .
so far we mainly focus on the physical ( phy ) layer techniques to optimize the performance of wpc . in a practical system ,
medium access control ( mac ) plays the key role in determining the fairness and efficiency of the system .
an efficient wireless system design often takes a cross - layer approach , especially for the closely related phy and mac layers . in the context of wpc ,
an example of cross phy - mac design is for the hap to steer the energy beam towards a user with relatively strong wireless channel and many data packets backlogged in the queue , rather than considering the physical channel condition alone . besides
, an efficient energy scheduling should also consider the residual battery life , the wake up / sleep schedule , and the expected energy consumptions of all wireless devices .
the current studies on wpc are mainly theoretical in nature .
the achievable throughput performance using off - the - shelf energy harvesting and communication modules and under practical wireless environment is not known .
hardware prototyping is urgently needed to evaluate the feasibility of wpc , and to test the applications of various technologies in joint energy and information transmissions , such as massive mimo , millimeter wave , and distributed antenna system .
an extensive testbed could also help identify the most suitable technology , and the proper application scenarios for wpc . with the potential use of massive mimo and advanced beamforming technologies ,
the intensity of microwave at a particular area could be strong enough to harm the human health and cause safety issues .
in practice , the radiation power of any wireless device must satisfy the equivalent isotropically radiated power ( eirp ) requirement on its operating frequency band , e.g. , the fcc ( federal communications commission ) permits a maximum @xmath42 dbm eirp on @xmath43 ghz band . one promising method to solve
the safety problem is to use distributed antenna system , such that for each antenna the radiation is omnidirectional and relatively weak ( thus satisfying the given eirp constraint ) , while the combined effect is constructive only at the destined location but destructive at almost everywhere else .
this will reduce the risk of radiation burn " due to human blockage from a random direction . besides
, the distributed antenna system could be combined with advanced sensing technology to detect the presence of human in real - time , and cease energy transmission if it deems the transmission to be harmful .
in this article , we have provided an overview of state - of - the - art rf - enabled wet technologies and their applications to wireless communications .
promisingly , wireless powered communications could significantly improve upon its battery - powered counterpart , and be practically achieved using simple and inexpensive transceiver structures .
the opportunities and challenges in the design of wireless powered communications were demonstrated by studying two new paradigms : swipt and wpcn .
we hope that the design of wireless powered communications will spur research innovations in wireless technologies , as the future wireless systems is expected to be a mixture of wireless information and energy transfer , with rf - enabled wet , swipt and wpcn being important building blocks .
y. h. suh and k. chang , a high - efficiency dual - frequency rectenna for 2.45-and 5.8-ghz wireless power transmission , " _ ieee trans . microwave theory and techniques _ , vol .
50 , no . 7 , pp . 1784 - 1789 , jul .
2002 .
g. yang , c. k. ho , r. zhang , and y. l. guan ,
throughput optimization for massive mimo systems powered by wireless energy transfer , " to appear in _
ieee j. sel .
areas commun .
_ , available online at arxiv:1403.3991 suzhi bi ( [email protected] ) received the b.eng .
degree in communications engineering from zhejiang university , hangzhou , china , in 2009 .
he received the ph.d degree from the chinese university of hong kong , hong kong , in 2013 .
he is currently a research fellow in the department of electrical and computer engineering at national university of singapore , singapore .
his current research interests include wireless information and power transfer , medium access control in wireless networks , and smart power grid communications .
chin keong ho ( [email protected] ) received the b.eng . and m. eng degrees from national university of singapore .
he received the ph.d .
degree from eindhoven university of technology , the netherlands , where he concurrently conducted research work in philips research .
he is lab head of energy - aware communications lab in institute for infocomm research , a@xmath44star .
his research interest includes green wireless communications with focus on energy - efficient solutions and with energy harvesting constraints , and implementation aspects of multi - carrier and multi - antenna communications .
rui zhang ( [email protected] ) received the b.eng . and m.eng .
degrees from national university of singapore and the ph.d .
degree from stanford university , all in electrical engineering .
he is now an assistant professor with the department of electrical and computer engineering at national university of singapore .
his current research interests include multiuser mimo , cognitive radio , energy efficient and energy harvesting wireless communication , and wireless information and power transfer . | the performance of wireless communication is fundamentally constrained by the limited battery life of wireless devices , whose operations are frequently disrupted due to the need of manual battery replacement / recharging .
the recent advance in radio frequency ( rf ) enabled wireless energy transfer ( wet ) technology provides an attractive solution named wireless powered communication ( wpc ) , where the wireless devices are powered by dedicated wireless power transmitters to provide continuous and stable microwave energy over the air . as a key enabling technology for truly perpetual communications , wpc opens up the potential to build a network with larger throughput , higher robustness , and increased flexibility compared to its battery - powered counterpart .
however , the combination of wireless energy and information transmissions also raises many new research problems and implementation issues to be addressed . in this article
, we provide an overview of state - of - the - art rf - enabled wet technologies and their applications to wireless communications , with highlights on the key design challenges , solutions , and opportunities ahead . |
many experiments have shown that the electrons in cuprate high-@xmath0 superconductors ( hts ) are significantly renormalized by the interaction with collective modes .
this renormalization appears in photoemission measurements as velocity changes in the quasi - particle dispersion ( the `` kinks '' ) accompanied by a drop of the quasi - particle life - time @xcite . in tunneling ,
the renormalization shows up as a depression , or `` dip '' , in the @xmath4 curve with the associated nearby accumulation of spectral weight ( the `` hump '' ) @xcite .
similar signatures observed by tunneling spectroscopy in classical superconductors were successfully explained by the strong - coupling theory of superconductivity @xcite .
there are , however , two striking differences between the structures observed in the cuprates and in low-@xmath0 metals such as pb or hg .
the dip in the cuprates is electron - hole asymmetric , being strongest at negative bias , while no such asymmetry is seen in lead .
the electron - hole asymmetry of the dip is due to the electron - hole asymmetry of the underlying electronic density of states ( dos ) @xcite .
second , while the structures are subtle in low-@xmath0 materials they induce a change smaller than 5% in the tunneling spectrum the cuprate dip is generally a strong effect which , for instance , can reach @xmath5 in optimally doped bi@xmath6sr@xmath6ca@xmath6cu@xmath7o@xmath8 ( bi-2223 )
@xcite . it is tempting to attribute this difference of intensities to a difference in the overall coupling strength , as suggested by the largely different @xmath0 values .
however , a comparison of the effective masses indicates that the couplings are not very different in pb where @xcite @xmath9 and in the bi - based cuprates where @xmath10 varies between @xmath11 and @xmath12 as a function of doping @xcite . here we show that the large magnitude of the dip feature results from the low dimensionality of the materials and the associated singularities in the electronic dos .
tunneling experiments in strongly coupled classical superconductors have been interpreted using a formalism @xcite that neglects the momentum dependence of the eliashberg functions and of the tunneling matrix element , and further assumes that the normal - state dos @xmath13 is constant over the energy range of interest .
the tunneling conductance , then , only depends on the gap function @xmath14 , whose energy variation reflects the singularities of the phonon spectrum @xcite .
the dimensionality of the materials does not enter in this formalism .
the effect of a non - constant @xmath13 on the gap function has been discussed in the context of the a15 compounds @xcite .
however the direct effect of a rapidly varying @xmath13 on the tunneling conductance became apparent only recently in the high-@xmath0 compounds @xcite , and requires to go beyond the formalisms of refs and .
in particular , one can no longer assume that the tunneling conductance is proportional to the product of the normal - state dos by the `` effective superconducting dos '' @xcite @xmath15 $ ] , so that nothing justifies _ a priori _ to normalize the low - temperature tunneling conductance by the normal - state conductance as was done with low-@xmath0 superconductors . among the new approaches introduced to study strong - coupling effects in hts , some have focused on generalizing the classical formalism to the case of @xmath2-wave pairing @xcite , still overlooking the dimensionality .
other models are strictly two dimensional ( 2d ) and pay attention to the full electron dispersion @xcite , taking into account the singularities of @xmath13 .
most of these studies assume that the collective mode responsible for the strong - coupling signatures is the sharp @xmath16 spin resonance common to all cuprates near 3050 mev ( ref . ) , but a phonon scenario was also put forward @xcite . in the present work , we extend these approaches to three dimensions ( 3d ) by means of an additional hopping @xmath17 describing the dispersion along the @xmath18 axis , and we study the evolution of the strong - coupling features in the tunneling spectrum along the 2d to 3d transition on increasing @xmath17 . for simplicity
we restrict to the spin - resonance scenario ; the electron - phonon model can be treated along the same lines , and both models lead to the same main conclusions .
the model we use is described in sec .
[ sec : model ] , results are presented and discussed in sec . [
sec : results ] , and sec .
[ sec : nt ] is devoted to investigating the validity and usefulness of the effective superconducting dos concept .
following previous works @xcite we assume that the differential conductance measured by a scanning tunneling microscope ( stm ) is proportional to the thermally - broadened local density of states ( ldos ) at the tip apex @xcite , and that furthermore the energy dependence of the ldos just outside the sample follows the energy dependence of the bulk dos @xmath19 : @xmath20n(\omega),\ ] ] where @xmath21 is the derivative of the fermi function . in stm experiments , various sources of noise may contribute to broaden @xmath19 further ; when comparing theory and experiment we shall take these into account by a phenomenological gaussian broadening . in the superconducting state ,
the interaction with longitudinal spin fluctuations is described by the @xmath22 nambu matrix self - energy , @xmath23 with @xmath24 the @xmath25 spin susceptibility , @xmath26 the coupling strength with @xmath27 the spin - spin interaction energy , and @xmath28 the @xmath22 bcs matrix green s function .
the sums extend over the @xmath29 vectors @xmath30 in the three - dimensional brillouin zone and the even matsubara frequencies @xmath31 with @xmath32 .
equation ( [ eq : sigma ] ) gives the lowest - order term of an expansion in @xmath27 @xcite .
we follow ref . and use for @xmath24 a model inspired by neutron - scattering experiments on the high-@xmath0 compounds . in this model
, @xmath24 has no @xmath33 dependence and is the product of a lorentzian peak centered at @xmath16 in the 2d brillouin zone and another lorentzian peak centered at the resonance energy @xmath34 .
the widths of the peaks are @xmath35 in momentum space and @xmath36 in energy .
this simple separable form of @xmath24 allows to evaluate analytically the frequency sum in eq .
( [ eq : sigma ] ) , and to perform analytically the continuation from the odd frequencies @xmath37 to the real - frequency axis .
the remaining momentum integral is a convolution which can be efficiently performed using fast fourier transforms .
also , the absence of @xmath33 dependence in @xmath24 implies that the self - energy does not depend on @xmath38 .
we use a bcs green s function broadened by a small phenomenological scattering rate @xmath39 , @xmath40\hat{\tau}_0 + \xi_{\vec{k}}\hat{\tau}_3+\delta_{\vec{k}_{\parallel}}\hat{\tau}_1 } { [ i\omega_n+i\gamma(i\omega_n)]^2-\xi_{\vec{k}}^2-\delta_{\vec{k}_{\parallel}}^2},\ ] ] where @xmath41 , @xmath42 are the pauli matrices with @xmath43 the identity , and @xmath44 is the @xmath2-wave gap which we assume @xmath38 independent for simplicity .
the additional effects resulting from a possible weak modulation @xcite of the gap along @xmath38 will be discussed toward the end of sec .
[ sec : results ] .
we do not address here the origin of the pairing leading to the bcs gap @xmath45 . with the high-@xmath0 compounds in mind ,
we consider a one - band model of quasi-2d electrons with a normal - state dispersion , @xmath46 where @xmath47 , @xmath48 being the chemical potential and @xmath49 a five - neighbor tight - binding model on the square lattice ( @xmath50 ) , @xmath51 . the momentum dependence of the self - energy in eq .
( [ eq : sigma ] ) is not small ( see , e.g. , fig . 1 of ref . and fig .
[ fig : fig2 ] below ) .
this is a major difference with respect to the electron - phonon models describing low-@xmath0 three - dimensional metals , where the momentum dependence of the self - energy can be neglected .
the calculation of the dos is therefore much more demanding since two three - dimensional momentum integrations must be performed for every energy @xmath52 .
the dos is given by @xmath53 where @xmath54 is the first component of the matrix @xmath55^{-1}$ ] . in eq .
( [ eq : dos ] ) , the @xmath38 integration can be performed analytically ( see appendix ) but not in eq .
( [ eq : sigma ] ) . in order to achieve a good accuracy when computing the dos @xmath19 , we evaluate the self - energy using a @xmath56 mesh in momentum space and a value @xmath57 mev . for the evaluation of the tunneling conductance @xmath39
is increased to 2 mev , which allows to decrease the mesh size to @xmath58 . the model in eqs .
( [ eq : didv])([eq : dos ] ) has several parameters but our focus is on the @xmath18-axis hopping energy @xmath17 . in hts ,
@xmath17 is not larger than a few mev , and setting it to zero seems appropriate to discuss tunneling data .
indeed , in the 2d limit the model was found to fit the experimental data for optimally doped bi-2223 very well @xcite .
here we take the parameters determined from one such fit as a starting point , and we vary @xmath17 to demonstrate the role of the dimensionality on the tunneling spectrum .
the band parameters @xmath59 are @xmath60 , @xmath61 , @xmath62 , @xmath63 , and @xmath64 mev , and the chemical potential is @xmath65 mev .
the gap magnitude is @xmath66 mev .
the spin - resonance energy is @xmath67 mev , its energy width @xmath68 mev , and its momentum width @xmath69 .
finally the coupling strength is @xmath70 mev , which implies a quasi - particle residue @xmath71 and a mass renormalization @xmath72 at the nodal point of the fermi surface .
the temperature is set to @xmath73 k unless stated otherwise .
the resulting theoretical tunneling conductance for @xmath74 is compared with experimental data in fig .
[ fig : fig1]c ( topmost curve ) .
the evolution of @xmath19 with increasing @xmath17 is displayed in fig . [
fig : fig1]a . in the 2d limit ,
the dos shows sharp and particle - hole asymmetric coherence peaks , strong and asymmetric dips , as well as humps and shoulders where the spectral weight expelled from the dips is accumulated .
this produces , in particular , a characteristic widening of the coherence peaks basis , which becomes triangular .
the particle - hole asymmetries reflect the particle - hole asymmetry of the corresponding bare dos @xmath13 shown in fig . [
fig : fig1]b , whose van hove singularity ( vhs ) lies slightly below @xmath75 at @xmath62 mev : on the one hand , the spectral weight of the vhs goes to a larger degree into the negative - energy coherence peak , and on the other hand the enhancement of the scattering rate due to the vhs is stronger at negative energy , explaining the stronger dip at @xmath76 @xcite .
this can also be seen in fig .
[ fig : fig2 ] where the electron - scattering rate @xmath77 is displayed for the nodal and anti nodal points of the fermi surface .
the scattering rate vanishes for @xmath78 and has a pronounced , particle - hole asymmetric maximum near @xmath79 ( more precisely between @xmath80 and @xmath81^{1/2}$ ] ) .
it is also clear from the figure that the energy of the scattering - rate peak shows no dispersion with momentum @xcite but its intensity is strongly momentum dependent and larger by a factor @xmath82 in the anti nodal region as compared to the nodal region .
( a ) density of states @xmath19 for superconducting electrons coupled to a @xmath16 spin resonance , as a function of the @xmath18-axis hopping @xmath17 . for @xmath74 ,
the system is two dimensional while for @xmath83 mev , it is three dimensional .
( b ) normal - state bare dos @xmath13 for the same @xmath17 values . ( c )
tunneling conductance for the same @xmath17 values .
the temperature is @xmath73 k and a gaussian broadening of 4 mev has been applied .
the circles show the experimental data of ref . compared to the @xmath74 spectrum .
the curves in ( a ) , ( b ) , and ( c ) are offset vertically for clarity . ] for @xmath84 , the logarithmic divergence in @xmath13 is cut on the scale of @xmath85 due to dispersion along the @xmath18 axis ( fig . [
fig : fig1]b ) .
no significant change in either @xmath19 or @xmath4 is observed for @xmath86 mev .
this value is an upper bound for the @xmath18-axis hopping energy in the cuprates , and the relative insensitivity of the dos to a small @xmath18-axis dispersion justifies the use of two - dimensional models for these systems . at larger @xmath17 values , however , the suppression of the divergence in @xmath13 induces a drop of the coherence peaks in @xmath19 and @xmath4 .
this is a _
direct _ effect of dimensionality on the tunneling spectrum , which was overlooked in the conventional strong - coupling approaches of refs and .
simultaneously the peak in the scattering rate is also suppressed with increasing @xmath17 ( fig .
[ fig : fig2 ] ) , leading to a weakening of the dip feature in @xmath19 and @xmath4 .
this is an indirect effect of dimensionality , that is only revealed in the strong - coupling signatures .
scattering rate @xmath87 as a function of energy for several values of @xmath17 at ( a ) the nodal point @xmath88 and ( b ) the anti nodal point @xmath89 of the fermi surface shown in the insets .
curves are offset vertically for clarity .
the dashed vertical lines delimit the energy range @xmath78 where inelastic scattering by the spin resonance is forbidden .
the dotted vertical lines indicate @xmath90 . ]
partial bcs density of states in two ( 2d ) and three ( 3d ) dimensions .
the thin solid lines show the contribution coming from the @xmath91 region of the two - dimensional brillouin zone , shaded in black in the inset while the dashed lines show the contribution of the remainder of the zone ( shaded in gray ) .
the thick line is the total bcs dos . ] as fig . [
fig : fig2 ] shows , increasing the dimension not only suppresses the peak at @xmath80 in the scattering rate but it also reduces its momentum dependence . in 2d , this peak arises because the @xmath92 sum in eq .
( [ eq : sigma ] ) is dominated by the saddle - point region near @xmath93 and @xmath94 , where the spectral weight of the bcs green s function is largest_i.e .
_ , @xmath95 .
hence the peak energy is determined chiefly by the bcs excitation energy at @xmath96 , shifted by @xmath34 due to the convolution with the spin susceptibility , and the peak intensity is controlled by the momentum dependence of @xmath97 , which is at maximum for @xmath98 .
the situation changes in 3d because the anti - nodal regions no longer dominate the spectral weight , as illustrated in fig .
[ fig : fig3 ] .
this figure displays the partial bcs density of states , _
i.e. _ , the part of the bcs dos originating from states close to the @xmath91 and equivalent points .
while in the 2d limit , a region covering just 14% of the zone around @xmath91 provides 56% of the spectral weight for energies between @xmath34 and @xmath80 , its contribution is reduced to 21% in the 3d limit .
hence the scattering rate in 3d is nearly momentum independent and almost constant above @xmath80 .
finally , the 2d to 3d transition also suppresses the particle - hole asymmetry of the scattering rate .
this again results from the disappearance of particle - hole asymmetry in the underlying bare dos ( fig .
[ fig : fig1]b ) and in the corresponding bcs dos ( fig .
[ fig : fig3 ] ) .
thus the @xmath38 dispersion simultaneously defeats four players who contribute to make the strong - coupling signatures in the 2d high-@xmath0 superconductors distinctly different from those in 3d metals : the van hove singularity , the particle - hole asymmetry , the momentum dependence , and the strong scattering enhancement at @xmath99 , especially near @xmath91 . in the curves of figs .
[ fig : fig1]a and [ fig : fig1]c corresponding to the 3d limit , the strong - coupling signatures are barely visible . their magnitude is @xmath100% , smaller than the @xmath101% value observed in pb .
the origin of this difference lies in the gap symmetry . in @xmath2-wave superconductors ,
the coherence peaks in the bcs dos are weak logarithmic singularities @xcite while in @xmath3-wave superconductors , they are strong square - root divergences .
the strength of the scattering - rate peak at @xmath80 , and consequently the strength of the dip in the dos and tunneling spectrum , are determined by the strength of the coherence peaks in the bcs dos , as is clear from eq .
( [ eq : sigma ] ) . in the case of a @xmath2-wave superconductor ,
the coherence peaks are cut in 3d as compared to 2d ( see fig .
[ fig : fig3 ] ) in the same way as the logarithmic vhs in fig .
[ fig : fig1]b , resulting in the suppression of the scattering - rate enhancement at @xmath80 in fig .
[ fig : fig2 ] .
( note that , roughly speaking , the scattering rate is proportional to the bcs dos shifted in energy by @xmath102 . ) the suppression of the bcs coherence peaks with increasing dimension also occurs in @xmath3-wave superconductors , but with one difference : if , on the one hand , the part of the coherence - peak spectral weight coming from the vhs gets suppressed , on the other hand , the square - root gap - edge singularities persist in any dimension .
therefore , in @xmath3-wave superconductors the strong - coupling signatures remain clearly visible in 3d .
this is illustrated in fig .
[ fig : fig4]a .
the 2d and 3d dos curves of fig .
[ fig : fig1]a are compared to the curves obtained for the corresponding @xmath3-wave model , _
i.e. _ , with all parameters unchanged except the gap which is replaced by @xmath103 mev .
the changes are quite dramatic . the first effect to
notice is a drastic reduction in the peak - to - peak gap @xmath104 in the @xmath3-wave case : a consequence of the pair - breaking nature of the coupling eq .
( [ eq : sigma ] ) in the @xmath3-wave channel @xcite . still , the strong - coupling signatures appear at the same energy @xmath105 mev in both @xmath2 and @xmath3 wave , due to our choice of the lowest - order model @xmath106 in eq .
( [ eq : sigma ] ) .
the second observation is that the strong - coupling signatures look like steps in the @xmath3-wave dos , like in the classical superconductors @xcite , reflecting the asymmetric shape of the bcs @xmath3-wave coherence peaks .
in contrast , the signatures appear as local minima in the @xmath2-wave dos , because the coherence peaks of the @xmath2-wave bcs dos are nearly symmetric about their maximum .
in short , the strong - coupling features give an `` inverted image '' of the bcs coherence peaks @xcite .
an interesting consequence follows : while in @xmath3-wave superconductors , the strong - coupling structures correspond to _ peaks _ in the second - derivative @xmath1 spectrum , for a @xmath2-wave gap they correspond to _ zeros _ in the @xmath1 spectrum , as demonstrated in fig .
[ fig : fig4]b .
this conclusion applies equally to phonon models and calls for a reinterpretation of cuprate @xmath1 data in which peaks were assigned to phonon modes @xcite .
finally , one sees from fig .
[ fig : fig4 ] that in 3d the signatures remain strong for an @xmath3-wave gap , for the reason explained above , while they have almost disappeared in the @xmath2-wave case .
( a ) comparison of the @xmath2-wave and @xmath3-wave dos for @xmath74 ( 2d ) and @xmath83 mev ( 3d ) .
the thick lines show @xmath19 as in fig .
[ fig : fig1]a .
the thin lines show @xmath19 computed with an @xmath3-wave gap of the same magnitude @xmath66 mev , and all other parameters unchanged .
( b ) second - derivative tunneling spectrum , @xmath1 , in the region of the strong - coupling signature at negative bias . in the @xmath3-wave case
, there is a peak in @xmath1 close to the energy @xmath80 while in the @xmath2-wave case , there is a sign change in 2d and no clear signature in 3d . ]
the previous discussion underlines the role of the bcs coherence peaks in the formation , strength , and shape of the strong - coupling signatures . more generally , for such signatures to occur there must be divergences ( or at least pronounced maxima ) in the non - interacting dos .
peaks in the `` bosonic '' spectrum are not sufficient , although they are necessary . indeed
, phonon structures are absent from the normal - state spectra of classical superconductors @xcite because the normal - state dos is flat , in spite of the facts that the phonon spectrum and the electron - phonon coupling do not change significantly at @xmath0 .
in contrast , the normal - state dos of 2d high-@xmath0 superconductors exhibits structures , either the pseudogap @xcite or the bare vhs @xcite .
one can therefore expect to see strong - coupling features in the normal - state spectra of hts , provided that the peaks in the bosonic spectrum subsist above @xmath0 .
figure [ fig : fig5 ] ( thin lines ) shows the normal - state dos implied by setting @xmath107 in our model , keeping the other parameters fixed ( including temperature ) . as expected sharp strong - coupling features remain in 2d at energies @xmath102 and @xmath108 mev while nothing but very weak structures subsist in 3d , signaling the onset of scattering at @xmath102 .
unfortunately it turns out that in the hts the spin resonance is absent above @xmath0or at least below the background level of neutron - scattering experiments @xcite .
the normal state of bi-2223 has not been investigated by neutron scattering so far but we may borrow information from the much studied yba@xmath6cu@xmath7o@xmath109 system ( y-123 ) . in y-123 ,
the normal - state spin susceptibility preserves its separable form with independent momentum and energy variations @xcite .
it is still centered at @xmath16 with a broad maximum at a characteristic temperature - dependent frequency @xmath110 .
for the purpose of illustrating the effect of a broad spin - fluctuations continuum on the normal - state tunneling spectrum , it is sufficient to use the same model as in the superconducting state but with the new parameter @xmath111 mev @xcite .
the resulting dos calculated at @xmath112 k is shown by the thick lines in fig .
[ fig : fig5 ] .
the strong - coupling signatures are almost washed out in 2d and completely in 3d .
this is not due to the thermal broadening of eq .
( [ eq : didv ] ) , not included in the dos @xmath19 , but mostly to the _ intrinsic _ temperature dependence of the self - energy in eq .
( [ eq : sigma ] ) , and , to a lesser extent , to the broader spin response . hence ,
if structures due to interaction with spin fluctuations are unlikely to show up in the normal state of hts , those associated with the interaction with phonons may well be observable if the coupling is strong enough since this coupling will not change appreciably at @xmath0 .
normal - state dos .
the thin lines show the @xmath73 k dos for @xmath107 ; the thick lines show the @xmath112 k dos for @xmath107 and @xmath111 mev ( see text ) .
all other parameters are as in fig .
[ fig : fig1]a . ] in the present study , we have overlooked a possible @xmath38 dependence of the bcs gap , retaining only the @xmath38 dependence of the bare dispersion .
a weak modulation of the bcs gap along @xmath38 is expected in 3d systems @xcite . as shown in ref .
, such a modulation has the effect of cutting the logarithmic coherence peaks on the scale of @xmath113 , with @xmath114 the amplitude of the gap modulation .
this is similar to the effect of @xmath17 on the bcs coherence peaks , which are cut on a scale corresponding to the gap variation along the warped 3d fermi surface , namely , @xmath115 , as seen in fig .
[ fig : fig3 ] .
the expected effect of @xmath114 on the scattering rate is also an additional broadening on top of the one produced by @xmath17 , @xmath28 being replaced by its @xmath38 average in eq .
( [ eq : sigma ] ) .
therefore , we expect that the gap modulation along @xmath38 will contribute to suppress the coherence peaks and the strong - coupling features even further with increasing @xmath17 , as compared to the results in fig .
[ fig : fig1 ] . our results can be summarized as follows .
the formation of clear strong - coupling structures in the tunneling conductance requires two ingredients : ( a ) at least one peak in the spectrum of collective excitations and ( b ) at least one peak in the non - interacting or superconducting dos . in classical superconductors , ( a )
is provided by optical phonons and ( b ) is the asymmetric square - root singularity at the edge of the @xmath3-wave gap : strong - coupling features are asymmetric steps peaks in the @xmath1 curve and dimensionality plays no big role because ( a ) and ( b ) are present in any dimension . in the normal state , there is no signature because ( b ) is absent . in high-@xmath0 layered superconductors , ( a )
is provided by the spin resonance and ( b ) has two sources : ( b1 ) the logarithmic van hove singularity in the bare dos ; ( b2 ) the symmetric logarithmic singularities at the edge of the @xmath2-wave gap .
strong - coupling signatures appear as local minima zeros in the @xmath1 curve but they vanish with increasing dimensionality from 2d to 3d because ( b1 ) and ( b2 ) both get suppressed by the @xmath18-axis dispersion . in the normal state of two - dimensional hts , ( b2 ) is absent , leaving aside the question of the pseudogap but ( b1 ) remains and strong - coupling signatures are thus expected unless ( a ) disappears at @xmath0 .
this is the case for the spin resonance but certainly not for phonons , leaving open the possibility that phonon structures might be observable in the normal - state tunneling spectra .
the conventional theory of electron tunneling into superconductors @xcite leads to an equation identical to eq .
( [ eq : didv ] ) for the tunneling conductance , except that the dos @xmath19 is replaced by an `` effective tunneling dos '' @xmath116 $ ] .
@xmath117 is the normal - state dos at zero energy@xmath118 is assumed and @xmath14 is the gap function .
the latter must be understood as a fermi - surface average of weakly momentum - dependent quantities , @xmath119 with @xmath120 and @xmath121 the eliashberg pairing and renormalization functions . in the notation of eq .
( [ eq : sigma ] ) , they read @xmath122/(2\omega)$ ] and , for an @xmath3-wave gap @xmath123 , @xmath124 . in a @xmath2-wave superconductor , the fermi - surface average of the gap @xmath125 vanishes , and so does the average of the off - diagonal self - energy since @xmath126 . the effective tunneling dos concept is logically generalized @xcite by writing @xmath127
^ 2}\big\rangle_{\text{fs}}$ ] with @xmath128/(2\omega ) } \right\rangle_{\text{fs}}.\ ] ] this form of @xmath129 is an even function of @xmath52 , and can not fit the particle - hole asymmetric spectra in hts .
therefore , a further generalization of the effective tunneling dos has been necessary , namely , @xmath130 ^ 2}}\right\rangle_{\text{fs}}\ ] ] which suggests that the `` true '' superconducting dos can be obtained by dividing the tunneling spectrum in the superconducting state by the spectrum in the normal state @xcite .
equation ( [ eq : nt ] ) is very convenient , but lacks a formal justification .
our model offers the opportunity to investigate the usefulness of eq .
( [ eq : nt ] ) , by comparing numerically the actual tunneling dos @xmath19 of eq . (
[ eq : dos ] ) with the effective tunneling dos @xmath129 . for the practical evaluation of @xmath129
, we define the fermi - surface average as @xmath131 with @xmath132 the zero - energy spectral function in the absence of pairing : @xmath133 . with this definition
, the average is performed on the _ renormalized _ fermi surface , defined by @xmath134 , rather than the bare fermi surface @xmath135 .
furthermore , each state gets correctly weighted if the spectral weight is unevenly distributed along the fermi surface .
( left panels ) real part ( solid lines ) and imaginary part ( dashed lines ) of the pairing function defined in eq .
( [ eq : psi ] ) for @xmath74 ( 2d ) and @xmath83 mev ( 3d ) .
( right panels ) comparison of the effective tunneling dos @xmath129 of eq .
( [ eq : nt ] ) with the actual dos @xmath19 of eq . (
[ eq : dos ] ) .
the dashed lines show @xmath136 . in all graphs ,
the dotted vertical lines mark the energy @xmath90 . ]
a comparison of @xmath19 and @xmath129 is displayed in fig .
[ fig : fig6 ] , where the pairing function @xmath137 is also shown . in two dimensions ,
the real part of @xmath137 has a maximum at @xmath138 , where its imaginary part shows a rapid variation .
this is analogous to the behavior reported in ref . .
the resulting @xmath129 also shows a behavior similar to the one found in ref .
: @xmath129 is _ larger _ than the bcs density of states at energies smaller than @xmath80 and drops below the bcs dos at @xmath80 .
the actual dos @xmath19 , however , behaves differently : it is _ smaller _ than the bcs dos between the coherence peak and some energy above the dip minimum ( see also fig . 1 of ref . ) .
thus , although the positions of the strong - coupling features are identical in @xmath129 and @xmath19 , their shape is markedly different in 2d @xmath2-wave superconductors . in 3d ,
the difference between @xmath129 and @xmath19 is less severe than in 2d , and both curves show very weak signatures , although those in @xmath19 are slightly stronger .
finally , the @xmath129 curves show structures which are absent in the @xmath19 curves . in 2d , a peak at @xmath139 appears due to the vhs in @xmath13 ; this peak is unphysical because in the actual energy spectrum , the vhs is pushed to @xmath140^{1/2}$ ] . in 3d
, @xmath129 has a structure near @xmath141 mev , which also comes from the bare dos @xmath13 as can be seen in fig .
[ fig : fig1]b . in the actual spectrum , this structure
is suppressed due to the persistence of a large scattering rate at energies much higher than the threshold @xmath34 ( see fig . [
fig : fig2 ] ) .
these problems illustrate the limitations of the simple product ansatz eq .
( [ eq : nt ] ) for analyzing the tunneling spectrum of @xmath2-wave superconductors .
i thank john zasadzinski for useful discussions .
this work was supported by the swiss national science foundation through division ii and manep .
if the nambu self - energy has no @xmath38 dependence , the @xmath38 sum in eq .
( [ eq : dos ] ) can be performed analytically .
this is the case in our model defined in eq .
( [ eq : sigma ] ) . solving dyson s equation @xmath55^{-1}$ ] with @xmath28 given by eq .
( [ eq : gbcs ] ) , we find since @xmath143 does not depend on @xmath38 ( although it _ does _ depend on @xmath17 ) , the @xmath38 dependence only comes from @xmath144 in eq .
( [ eq : dispersion ] ) and we can make it explicit by rewriting @xmath145}\\ = \frac{1}{2}\left(1+\frac{\zeta}{\lambda}\right)\frac{1}{\eta+\lambda-2t_z\cos(k_zc)}+ \frac{1}{2}\left(1-\frac{\zeta}{\lambda}\right)\frac{1}{\eta-\lambda-2t_z\cos(k_zc)}. \end{gathered}\ ] ] in eq .
( [ eq : g11 ] ) , the quantities @xmath146 , @xmath147 , @xmath148 , @xmath149 , @xmath150 , and @xmath151 are all functions of @xmath152 and @xmath52 but not of @xmath38 .
explicitly , @xmath153 , @xmath154 , @xmath155 , @xmath156 , @xmath157 , and @xmath158 . in electron - phonon models , the migdal argument allows to neglect vertex corrections and obtain an accurate self - consistent theory by replacing @xmath28 in the self - energy by the full @xmath162 ( refs and ) . as there is no such theorem for spin fluctuations ,
the self - consistent model @xmath163 is not necessarily better ( _ i.e. _ , closer to the exact result including all vertex corrections ) than the lowest - order model @xmath164 .
this issue has been pointed out in the case of the hubbard model ( ref . ) , and is also well known in the context of the @xmath165 approximation for the long - range coulomb interaction , where the self - consistent version often gives poorer results than the non self - consistent one ( ref . ) .
the situation is further complicated if one uses a phenomenological spin susceptibility in eq .
( [ eq : sigma ] ) , rather than the spin susceptibility calculated using a consistent approximation .
no systematic inclusion of vertex corrections is possible in this latter case .
hence the justification of eq .
( [ eq : sigma ] ) and of the corresponding self - consistent model ( ref . ) resides in their ability to properly describe experiments .
just the opposite happens in electron - phonon models , which are pair breaking in the @xmath2-wave channel . in electron - phonon models , the self - energy eq .
( [ eq : sigma ] ) is replaced by , schematically , @xmath166 with @xmath167 the electron - phonon coupling and @xmath168 the phonon propagator ( ref . ) .
the @xmath169 matrices appear because phonons couple to the charge density while spin fluctuations couple to the spin density , and they change the sign of the @xmath170 component controlling the gap renormalization . the 200 k data in fig .
. can be well fitted to the lorentzian form used for the spin susceptibility in eq .
( [ eq : sigma ] ) , namely , @xmath171-(\gamma_{\text{sf}}/\pi)/[(\omega+\omega _ { \text{sf}})^2+\gamma_{\text{sf}}^2]\}$ ] , with the parameters @xmath172 mev , @xmath173 mev , and @xmath174 . on the other hand ,
the momentum width of the spin response was found to be temperature independent . to model the normal state of bi-2223
, we may therefore keep @xmath175 fixed and increase @xmath36 to the value of @xmath173 mev . in the absence of more detailed information
, we do not change the value of @xmath34 , which now plays the role of the broad maximum in the magnetic response .
these new parameters lead to a reduced effective mass @xmath176 in the normal state as compared to the superconducting state . | we investigate numerically the signatures of collective modes in the tunneling spectra of superconductors
. the larger strength of the signatures observed in the high-@xmath0 superconductors , as compared to classical low-@xmath0 materials , is explained by the low dimensionality of these layered compounds .
we also show that the strong - coupling structures are dips ( zeros in the @xmath1 spectrum ) in @xmath2-wave superconductors , rather than the steps ( peaks in @xmath1 ) observed in classical @xmath3-wave superconductors .
finally we question the usefulness of effective density of states models for the analysis of tunneling data in @xmath2-wave superconductors . |
in order to understand the formation processes of low - mass stars , it is important to investigate the properties of dense ( @xmath12 ) cores in molecular clouds .
such compact ( @xmath13 ) cores supply material to newly forming stars through dynamical gravitational collapse , however , detailed physical processes are still uncertain
. one of the investigative approaches is to derive the detailed density and velocity structures from observations of ( pre-)protostellar cores which are expected to retain more information than class i / ii objects for the initial conditions of gravitational collapse @xcite .
dust continuum emission imaging at millimeter and submillimeter wavelengths using single - dish radio telescopes has revealed the radial density profiles , @xmath14 , of ( pre-)protostellar cores .
recent investigations have demonstrated that the profiles of class 0/i sources can be fitted by single power - law profiles over a wide range of radii ( e.g. , * ? ? ?
@xcite modeled class 0 source maps using a single power - law density distribution @xmath15 and found that most of them can be well fitted with a power - law index of @xmath16 .
gas kinematics in dense cores have been investigated by molecular line observations .
it has been shown that asymmetric double - peaked profiles of optically thick lines detected toward star - forming cores are considered to be a signature of collapse motion .
extensive surveys of such blue - skewed spectra in starless cores have been carried out by @xcite .
furthermore , @xcite conducted the profile fitting using a simple two - layer radiative transfer model and suggested an inward motion of subsonic speed ( @xmath17@xmath18 ) that extended to @xmath19 in the pre - protostellar core l1544 .
the velocity structures of dense cores , including not only infalling motion but also rotation , have also been investigated using first and second moment images of line emission taken with single - dish telescopes and interferometers ( e.g. , * ? ? ?
statistical studies for the rotation of dense cores were conducted using @xmath20 and @xmath21 line emissions by @xcite and @xcite , in which the typical velocity gradients are found to be @xmath22@xmath23 . on the other hand , @xcite observed @xmath24@xmath1 line emission toward nine low - mass protostellar envelopes down to @xmath25 scales .
the mean velocity gradient estimated in their samples is @xmath26 , which is much larger than the velocity gradients of dense cores .
@xcite analyzed the kinematics of 17 protostellar systems .
they found that the velocity gradients obtained with interferometric data ( @xmath27 ) are considerably larger than those that also have single - dish data ( @xmath28 ) , which indicates accelerating infall and spinning - up rotational velocities toward the core center .
the environments surrounding forming stars are composed of structures with different scales : circumstellar disks ( @xmath29 ) , infalling envelopes ( @xmath30 ) , and dense cores ( @xmath31 ) .
recently , an approach that combines data obtained with single - dish telescopes and interferometers has been widely used to investigate the physics in protostellar systems ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
@xcite performed @xmath32@xmath33 and @xmath34@xmath35 observations toward the protostellar envelope of b335 with the submillimeter array and single - dish telescopes , and imaged by combining those data .
they derived the specific angular momentum of the envelope and found that specific angular momenta tend to be larger as evolution progresses by comparing with other class 0 , i , and ii sources .
theoretically , two extreme models for the core evolution have been proposed for isolated low - mass star formation .
the similarity solution of larson
penston describes the density evolution of isothermal gas spheres @xcite . when a central object is formed ( @xmath36 )
, the gas sphere reaches the density profile of @xmath37 and velocity field of @xmath38 , where @xmath39 is the isothermal sound speed and @xmath40 is the gravitational constant .
this model of the larson
penston solution is referred to as `` runaway '' collapse .
on the other hand , the isothermal similarity solution proposed by @xcite describes a core that is slowly increasing its central density through ambipolar diffusion while maintaining kinematic balance ( i.e. , @xmath41 ) , and moving toward dynamical collapse .
the density profile achieves a singular isothermal sphere , @xmath42 , at @xmath36 , which is the initial condition of dynamical collapse after protostar formation .
this model is the most static and is referred to as `` inside - out '' collapse .
moreover , extensions for @xmath43 of the larson penston solution and generalization were developed by @xcite and @xcite . at later times ( @xmath43 ) , in both of the solutions , the density and velocity structures attain a free - fall profile , @xmath44 and @xmath45 , respectively , from the center to the outside , with a sound speed for the shu solution that accompanies the rarefaction wave and with a supersonic velocity for the larson penston solution .
the mass infall rate is predicted to be @xmath46 for the shu solution and 48 times higher than this for the larson penston solution . in this paper
, we present an observational study of the dense core associated with a class 0 protostar within b335 .
the bok globule b335 , otherwise known in the literature as cb199 in the catalog of @xcite or l663 , is a typical low - mass star - forming region .
b335 appears as an opaque dark cloud on optical images and is one of the best candidates for studying the initial conditions of star formation , because it is isolated from other star - forming regions and is near the sun . in this paper
, we adopt a distance of @xmath47 @xcite .
b335 contains a far - infrared ( fir ) source , iras 19347 + 0727 , which is bright in submillimeter wavelengths @xcite and shows a combination of characteristics that indicates one of the clearest examples of very young stars .
this fir source is associated with the dense molecular gas envelope , and is classified as a class 0 protostellar object based on its spectral energy distribution @xcite .
this source is considered to be the driving source of the bipolar outflow @xcite , extending from east to west with an inclination angle from the sky plane of @xmath48 and an opening angle of @xmath49 . toward the iras source in b335 ,
@xcite have obtained blue - skewed line profiles in the @xmath50@xmath51@xmath52@xmath53 and @xmath54@xmath55@xmath56 line emissions , which can be explained by a spherically symmetric inside - out collapse model @xcite .
moreover , @xcite computed radiative transfer using a monte carlo method with a model of inside - out collapse .
they confirmed that the double - peaked line profiles indicate an infall motion and derived the physical parameters of the collapsing envelope .
@xcite made interferometric observations in the @xmath57 line and also interpreted the kinematics as an infalling motion , whereas the interferometric images of @xmath58@xmath53 by @xcite were found to be dominated by clumps associated with the outflow cavity .
as mentioned above , the detailed kinematics of infall motion provide crucial information about the properties of the gravitational collapse of the core , which we can compare with the theoretical pictures and distinguish between them .
few observational studies until now , however , have been capable of unveiling the velocity structures within dense cores in the early phases of star formation over a wide range of spatial scales .
we carried out the observations with the nobeyama 45 m telescope and the nobeyama millimeter array ( nma ) in the @xmath0@xmath1 line emission .
observations with the 45 m telescope trace the overall density and kinematic structures of the star - forming core . for the high - resolution observations with nma
, we combined the data with the 45 m telescope data to image the structures of the dense core over a wider spatial frequency range down to the inner dense envelope scales .
the mapping observations of the @xmath0@xmath1 line emission at @xmath59 toward the b335 region were carried out during 2006 march may .
we used the 25 beam array receiver system ( bears ; * ? ? ?
* ; * ? ? ?
* ) , which consists of superconductor insulator superconductor ( sis ) receivers on a @xmath60 grid with a separation of @xmath61 , in the double sideband ( dsb ) mode .
the observations were made in the on - the - fly ( otf ) mode of the nobeyama radio observatory ( nro ) 45 m telescope @xcite .
the mapping center was placed at the iras source 19347 + 0727 associated with b335 , @xmath62 , @xmath63 ( b1950 ) , and the mapping region was @xmath64 .
we obtained five otf - scan data over this region in the r.a . and decl .
directions in each and merged them to create a final map cube . at @xmath65 , the half - power beam width and main - beam efficiency were @xmath66 and @xmath67 , respectively . as the back end
, we used 25 sets of @xmath68 channel autocorrelators ( acs ) , with a frequency resolution of @xmath69 corresponding to a velocity resolution of @xmath70 .
the system noise temperatures were between @xmath71 and @xmath72 .
we took the off position of @xmath73 , @xmath74 ( b1950 ) and used the standard chopper wheel method to convert the receiver output into @xmath75 intensity scale .
the telescope pointing was checked once an hour by five - point observations of the sio maser from rt aql in the @xmath76 using the sis receiver ( s40 ) .
the pointing accuracies were within @xmath77 for the @xmath78 observations . for the @xmath78 data obtained with the otf observations , idl - based reduction software , nostar ( nobeyama otf software tools for analysis and reduction ) , was used for flagging , baseline subtraction , and making the map cube .
we corrected for relative gain differences among the 25 beams using correction factors provided by nro , and then made the baseline fitting and subtraction .
we made a three - dimensional image cube that has an effective spatial resolution of @xmath79 and an achieved noise level in channel images ( @xmath80 ) of @xmath81 in @xmath82 with a velocity resolution of @xmath70 .
we also observed three inversion spectra of @xmath83 , @xmath84 , and @xmath85 at @xmath86 toward the iras source associated with b335 using the hemt receiver ( h22 ) during 2006 march may . at @xmath87 , the half - power beam width and main - beam efficiency were @xmath88 and 0.82 , respectively .
we used eight sets of 2048 channel acousto - optical spectrometers ( aoss ) that have a frequency resolution of @xmath89 , corresponding to the velocity resolution of @xmath90 .
the typical system noise temperatures were in a range of @xmath91@xmath72 during the observations .
the achieved noise level in the spectrum was about @xmath92 in @xmath82 , with a velocity resolution of @xmath90 .
the pointing accuracies were within @xmath93 for the @xmath20 observations .
the observational parameters are summarized in table [ tbl:1 ] for the 45 m telescope observations .
the aperture synthesis observations were carried out using the six - element nma .
we observed the @xmath0@xmath1 line in the period from 2005 december to 2006 april .
the phase tracking center for our observations was set at the position of the iras source in b335 .
the field of view ( i.e. , the primary beam size ) of the @xmath94 dishes was about @xmath95 at @xmath65 .
our observations were conducted with two array configurations ( d and c configurations ) . for the maximum scale
detectable in our observations , @xmath96 , the expected brightness recovered at the phase center is @xmath97 @xcite when a gaussian brightness distribution having a fwhm of @xmath98 is assumed .
we used the sis receivers as the front end and the digital spectro - correlator , new - fx , which has 1024 channels and bandwidth of @xmath99 ( achieved velocity resolution was @xmath70 ) , as the back end .
we also obtained the continuum data simultaneously with the other spectro - correlator , ultra - wide band correlator ( uwbc ; * ? ? ?
* ) , which has 128 channels in the @xmath100 bandwidth mode .
the system temperatures were typically in the range of @xmath101@xmath71 in dsb .
we used 3c345 and 3c454.3 as passband calibrators and b1923 + 210 as a gain calibrator .
the absolute flux densities were calibrated using the flux density of b1923 + 210 determined by the bootstrapping method with uranus and neptune .
the measured flux density of b1923 + 210 was @xmath102@xmath103 at @xmath65 during the observation period .
the uncertainty in our flux calibration was expected to be less than 10% .
the nma visibility data were processed ( calibration , flagging , and continuum subtraction ) using the uvprocii package . for the continuum data
, we merged the visibilities in each sideband obtained with the uwbc to construct an image .
the calibrated visibility data set was processed through imaging and image reconstruction with the miriad package .
the synthesized beam size is @xmath104 ( position angle ( p.a . )
@xmath105 ) at @xmath65 with natural weighting . imaging noise levels ( @xmath80 ) of @xmath78 and @xmath65 continuum data are @xmath106 , with a velocity resolution of @xmath70 and @xmath107 , respectively .
the parameters for the nma observations are summarized in table [ tbl:2 ] .
we applied a data combining technique of interferometer and single - dish data in the fourier domain ( @xmath108@xmath109 domain ) for our nma and 45 m telescope @xmath78 data .
we made the 45 m telescope image cube using a gridding kernel of spheroidal @xcite with a grid size of @xmath110 , which is the nyquist spacing of a dish diameter of @xmath111 at @xmath65 .
the velocity separation between the image channels is @xmath70 , which is the same as that of the nma observations .
the short spacing information in the @xmath108@xmath109 domain including the zero - baseline within the diameter of the primary antenna of nma ( @xmath112 ) , which can not inherently be obtained with the nma observations , can be complemented with the 45 m telescope data .
the fundamental theory and detailed algorithm of data combining are described in @xcite . in this method , we generated pseudo - visibilities from the fourier transformed single - dish image data , which are deconvolved by a gaussian beam with an fwhm of @xmath79 and multiplied by a primary beam of nma approximated by a gaussian function with an fwhm of @xmath95 .
the single - dish visibilities that were generated were combined with the nma data in the @xmath108@xmath109 domain . as a result
, we can make synthesis images for a high spatial dynamic range and a high spatial resolution while recovering missing large - scale fluxes .
we applied the data optimizations to sensitivities and relative weights between 45 m and nma data @xcite .
we finally obtained a combined image of b335 in the @xmath78 line emission with a synthesized beam size of @xmath113 ( p.a .
of @xmath114 ) and with a total flux of @xmath115 that is comparable to that of the 45 m telescope image , @xmath116 .
furthermore , y. kurono et al .
( 2013 , in preparation ) describe the schemes that are applied to our observed data , and demonstrate the importance of relative ( 1 ) flux scaling , ( 2 ) sensitivity , and ( 3 ) weighting between interferometer and single - dish data in order to obtain reliable results .
the spectrum of the @xmath83 transition , which consists of five hyperfine groups , and the main component of the @xmath84 transition were obtained with good signal - to - noise ratios for the 45 m telescope .
figure [ fig:1 ] shows the spectra of the @xmath83 and @xmath84 transitions obtained at the position of the iras source in b335 .
no emission in the @xmath85 transition was detected . to estimate the optical depth and gas kinetic temperature
, we analyzed the hyperfine structures that are caused by the electric quadrupole moment of the nitrogen nucleus .
we derived the peak main - beam temperatures ( @xmath117 ) and intrinsic velocity widths ( @xmath118 ) by fitting each line component with a gaussian function .
we assumed that all hyperfine components have equal beam filling factors and excitation temperatures .
we estimated the optical depth of the @xmath119 main component [ @xmath120 , the rotational temperature [ @xmath121 that describes the relative population between the @xmath122 and @xmath84 levels , and the kinetic temperature ( @xmath123 ) by following the analysis in a previous work by @xcite . with the parameters of the best - fit results ( summarized in table [ tbl:3 ] ) shown by the blue curves in figure [ fig:1 ] , we obtained @xmath124 , @xmath125 , and @xmath126 .
the rotational temperature is almost equal to the kinetic temperature for the range of @xmath127 @xcite .
therefore , we determined that the b335 core has a mean kinetic temperature of @xmath128 over the beam area of the 45 m telescope ( @xmath129 ) , and used this temperature to estimate the isothermal sound speed and the column density from the molecular line data .
we detected @xmath65 continuum emission with the nma toward the iras source in b335 , as shown in figure [ fig:2 ] .
the peak position was measured to be @xmath130 , @xmath131 ( b1950 ) , which is consistent with the peak position of the @xmath132 continuum images @xcite .
the @xmath65 continuum image shows an elongated emission structure perpendicular to the outflow lying from east to west @xcite .
furthermore , a cavity - like distribution in the red - lobe side of the molecular outflow can be seen which is quite similar to that in the @xmath132 continuum image shown by @xcite and supports their suggestion that this distribution traces the wall of the outflow cavity .
the beam deconvolved size , peak intensity , and total flux density ( above the @xmath133 contour ) are @xmath134 ( corresponding to @xmath135 ) , @xmath136 , and @xmath137 , respectively .
as mentioned in @xcite , contribution from free free emission is considered to be negligible when taking into account the extrapolation from the flux at @xmath138 .
thus , we determine that the @xmath65 continuum emission comes from a dust envelope surrounding a protostellar object . under the condition of being optically thin for thermal dust emission , the dust envelope mass ( @xmath139 )
was estimated using the equation @xmath140 , where @xmath141 is the total flux density , @xmath142 is the dust mass opacity coefficient , @xmath143 is the dust temperature , @xmath144 is the source distance , and @xmath145 is the planck function . by combining our @xmath65 and image band @xmath146 measurements with the flux densities at millimeter wavelengths estimated by @xcite , @xcite , and @xcite , we obtained a spectral index of @xmath147 for @xmath148 .
the spectral index gives @xmath149 for the emissivity law , @xmath150 , using the approximated relation , @xmath151 , which is valid for millimeter wavelengths @xcite .
thus , given @xmath152 @xcite , we obtain a dust mass opacity of @xmath153 .
the dust envelope mass is estimated to be @xmath154 with a dust temperature of @xmath155 @xcite .
the panels of figure [ fig:3 ] show the integrated intensity images of the @xmath156@xmath1 line emission over the lsr velocity range from @xmath157 to @xmath158 made using the 45 m telescope data ( left ) and the combined 45 m telescope and nma data ( right ) .
we should note that this combined map includes the effect of the primary beam attenuation of the nma so that extended emission toward the outside of the image seen in the 45 m telescope image is not reproduced in the combined image .
the @xmath78 emission in the 45 m telescope map has a single - peaked spatial distribution and shows an elongation from north to south .
the size of the @xmath78 core above the @xmath133 level is @xmath159 , with a p.a . of @xmath160 . the 45 m telescope plus nma combined map shown in the right panel of figure [ fig:3 ] clearly depicts the detailed structure of the inner core including large - scale flux distributions with a high resolution .
the higher contours of @xmath161 show an elongated distribution from north to south with a size of @xmath162 , which is believed to be an inner dense envelope associated with a central stellar source .
the envelope has a double peak near the center , and the @xmath65 continuum source is located between the peaks .
the elongation of the core and inner envelope is perpendicular to the molecular outflow axis ( e.g. , * ? ? ?
* ) . in the 45 m telescope map , there are faint ridges from the center along the p.a . of @xmath163 , @xmath164 , and @xmath165 ,
and they can be seen more clearly in the combined map . taking the outflow direction and opening angle into account ( @xmath49 ; * ? ? ?
* ) , we believe that these ridges are related to the outflow activity . in the velocity channel maps obtained with the 45 m telescope shown in figure [ fig:4 ] ,
the emission peaks are located at the north of the @xmath65 continuum source in the velocity range @xmath166@xmath167 , and at the south in the velocity range @xmath168@xmath169 .
this velocity gradient is perpendicular to the outflow axis and can be interpreted as the rotational motion of the b335 core . on the other hand , in the channel maps of combined data shown in figure [ fig:5 ] , it is difficult to identify the corresponding velocity gradient because of the complicated emission distribution .
more detailed kinematics of the b335 core are discussed in section [ subsubsec:3.3.3 ] and [ subsec:4.2 ] using position
velocity ( pv ) diagrams made from the 45 m telescope and combined images . in order to estimate the column density of the b335 protostellar core
, we analyzed the @xmath78 images made from the 45-m telescope and the combined nma plus 45 m telescope data .
under the local thermodynamic equilibrium ( lte ) assumption , the @xmath170 column density can be calculated using the following formula : @xmath171 where @xmath172 is the excitation temperature , @xmath173 is the optical depth of the @xmath78 line , and @xmath174 is the fractional abundance of @xmath78 .
when deriving of equation ( [ eq:1 ] ) , we used the permanent dipole moment @xmath175 @xcite and the rotational constant @xmath176 . to obtain a complementary expression for the combined synthesized image
, we convert the antenna temperature into the flux density @xmath141 in the @xmath177 through @xmath178 , where @xmath179 is the beam solid angle given by @xmath180 .
thus , we have @xmath181 we assumed the excitation temperature of @xmath182 from the @xmath183 analysis ( section [ subsec:3.1 ] ) and the @xmath78 line of the optically thin limit .
we assumed @xmath174 to be @xmath184 @xcite . the derived total mass from the 45 m telescope map ( the left panel of figure [ fig:3 ] ) is @xmath185 .
we derived the radial column density profile in the b335 core from the column density map calculated by the above formula .
the column densities were calculated from re - gridded images with cell sizes corresponding to the spatial resolutions of the images , i.e. , @xmath186 ( @xmath187 ) for the combined image and @xmath188 ( @xmath189 ) for the 45 m image .
the column density profile was made as a function of the radius from the @xmath65 peak using the distributions of the estimated column densities over the re - gridded cells . in order to calculate the column density , the 45 m plus nma image needed to be corrected for the primary beam attenuation , which increases the noise level in the outer region of the image .
hence , the column density profile in the outer region of the core was derived using the 45 m telescope data , and the 45 m plus nma combined data were used to fill in the inner region of @xmath190 where the data can not be obtained from 45 m telescope data .
figure [ fig:6 ] represents the resulting @xmath170 column density profiles obtained in the above procedure .
as suggested in section [ subsubsec:3.3.1 ] , the b335 core could be affected by the outflow . to examine this effect we also made the column density profile by masking out the regions with p.a .
of @xmath191@xmath192 and @xmath193@xmath194 .
the masking angle of @xmath195 was chosen to match the opening angle of the outflow : @xmath49 by @xcite and @xmath196 by @xcite .
figure [ fig:6](a ) and ( b ) show the @xmath170 column density profile without and with the masking , respectively . for both profiles , we can see that the column densities estimated from the 45 m and combined data are smoothly connected around the radius of @xmath197 , and that the profiles in the inner radius are shallower than those in the outer radius .
the column density profiles were fitted by two power - law functions of @xmath198 , where @xmath199 is the power - law index and @xmath200 is the column density at @xmath201 . since figure [ fig:6 ] clearly shows two different slopes between the inner and outer regions , we estimated the turnover radius at which the power - law index changes .
we evaluated the correlation coefficient of the power - law fitting in the range of @xmath202@xmath203 , where the inner radius ( @xmath202 ) was variable and the outer radius ( @xmath203 ) was fixed to @xmath204 . as a result
, we found that the correlation coefficient of the fitting decreased exceedingly when the fitting inner radius was taken inside of @xmath205 .
thus , we adopted a turnover radius of @xmath205 in this paper and performed the power - law fittings in the two regions of @xmath206@xmath205 and @xmath207@xmath208 separately .
for the inner region ranging from @xmath206 to @xmath205 , we obtained @xmath209 without masking , and @xmath210 with masking .
for the outer region from @xmath207 to @xmath208 , we obtained @xmath211 without masking and @xmath212 with masking .
the column density profile obtained with masking was estimated to be slightly shallower than that without the masks . from our fitting result ,
the column density at @xmath213 is @xmath214 , which is consistent with the estimate by @xcite , @xmath215 . using our results , we examined the density distribution of the b335 core ( discussed in section [ subsubsec:4.1.1 ] ) and estimated the total mass of gas associated with the b335 core to be @xmath216 within the radius of @xmath217 .
to examine the overall velocity structure in the b335 core , we made pv diagrams using the combined @xmath0@xmath1 line data as shown in figure [ fig:7 ] , with black contours and gray scales .
we chose the two axes of @xmath218 ( perpendicular to the outflow axis ) and @xmath219 ( along the outflow axis ) , passing through the @xmath65 continuum source .
the comparison with the diagrams made from the low - spatial - resolution 45 m data overlaid with blue contours demonstrates a drastic effect of data combining . along the p.a .
of @xmath220 , the 45 m telescope data show a slight velocity gradient , @xmath221 over a scale of @xmath8 .
this gradient is comparable to the typical value of the cores in the taurus dark cloud @xcite .
if the gradient represents a solid - body - like rotation of the core , then it is converted into an angular velocity of @xmath222 . on the other hand , the diagram of p.a . of @xmath223
does not show a conspicuous velocity gradient .
the diagrams of the 45 m and nma combined data show line broadening and a double - peaked intensity profile at the core center ( offset @xmath224 ) . with a broadening line width toward the core center , the diagram along the p.a
. of @xmath220 shows an overall symmetrical diamond - like structure ( represented by a dashed line in light gray ) .
we calculated intensity - weighted first and second moments defined by @xmath225 and @xmath226 , respectively .
the second moment at the center ( offset of @xmath227 ) was @xmath228 , whereas that at the offset of @xmath229 was @xmath230 .
the double - peaked profile at the core center is also slightly visible in the diagram made from the 45-m telescope data .
this profile has asymmetrical features that are more enhanced in the blueshifted peak than in the redshifted one .
this blue - skewed profile is maybe due to the infall motion of the core @xcite because the @xmath231 line emission is moderately optically thick around the systemic velocity ( see section [ subsubsec:4.1.3 ] ) .
the center of the overall symmetrical emission distribution ( lowest contour ) , as represented with a dashed diamond in figure [ fig:7](a ) , is approximately @xmath232 and @xmath227 , which is considered to be the dynamical center of the b335 core .
since the lsr velocity of @xmath233 is consistent with the peak velocities of the @xmath20 main components ( section [ subsec:3.1 ] ) , we adopt an lsr velocity of @xmath233 as the systemic velocity of the protostellar system .
when a molecular cloud core has a density distribution of @xmath234 , the profile of the column density integrated along the line of sight has a relation of @xmath235 .
hence , our obtained indices of @xmath236 in the inner part of the b335 core and of @xmath237 in the outer part correspond to @xmath5 and @xmath3 , respectively .
we note that these indices barely varied even if we considered the influence of the cavity in the core made by the outflow .
the column density profile derived from our 45 m and combined data with masking ( section [ subsubsec:3.3.2 ] ) can be converted into the number density distribution : @xmath238 most previous studies have derived density profiles for the b335 star - forming core similar to our observational results : inner and outer indices of @xmath239 and @xmath240 , respectively , with a turnover radius of @xmath241 . a power - law index of @xmath242 for the density profile was estimated from single - dish observations of the @xmath243@xmath1 and @xmath0@xmath1 line emissions ( radius range of @xmath244@xmath245 ) by @xcite . from near - infrared extinction measurements , @xcite suggested that the b335 core has a constant power - law index of @xmath246 over the region @xmath247@xmath248 , and otherwise displays inside - out collapse with an infalling radius ( see section [ subsec:4.2 ] ) of @xmath249 ( corresponding to @xmath250 ) .
@xcite also suggested a single power law with @xmath251 within the inner @xmath252 region from interferometric observations of @xmath253 and @xmath254 continuum .
these results agree well with our estimate of the density profile from the @xmath0@xmath1 data
. meanwhile , detailed analysis of scuba @xmath255 and @xmath256 continuum maps by @xcite showed that the model with a power - law index of @xmath257 well describes the data as a best - fit result , and suggested an infalling radius of @xmath258 , which is smaller than estimates in our as well as previous studies .
@xcite recently conducted an unbiased fitting to the dust continuum observations toward b335 .
for the power - law density distribution , they obtained @xmath259@xmath260 throughout the envelope , although they did not find strong evidence of inside - out collapse with an infalling radius of @xmath261 .
we confirmed the above estimate of the density profile from our combined @xmath0@xmath1 line data using an analysis of the visibility data of the @xmath65 continuum emission observed with the nma .
this approach of directly examining in the @xmath108@xmath109 domain enables us to avoid possible artifacts caused by the deconvolution of interferometric images . when discussing the emission distribution of the inner region of an envelope observed with interferometers , we should consider the temperature distribution of the envelope .
we consider the observed intensity as a function of the impact parameter , @xmath262 , from an optically thin dust envelope that has a spherical density distribution , @xmath14 , and a dust temperature , @xmath263 .
if the density and temperature follow radial power laws , @xmath264 and @xmath265 , and if we assume that the opacity does not vary along the line of sight , then the intensity profile also has a power - law profile , @xmath266 , in the rayleigh
jeans regime .
we assumed that the extent of source intensity @xmath262 is sufficiently compact in the primary beam .
the visibility amplitude as a function of @xmath108@xmath109 distance ( @xmath267 ) can be given by @xmath268 for @xmath269 ( see the appendix for details ) .
figure [ fig:8 ] shows a plot of binned visibility amplitudes as a function of the @xmath108@xmath109 distance for the @xmath65 continuum data .
the binning is logarithmic , and the amplitudes are obtained as the vectorial average of the complex visibilities in each bin .
the decrease in visibility amplitude with @xmath108@xmath109 distance can be interpreted as a power law .
we fitted the visibility amplitude profile with a power - law function of the form @xmath270 .
the best - fit power - law index is @xmath271 ( solid line in the plot ) , which can be converted into a power - law index with an intensity distribution of @xmath272 .
the power - law index of dust temperature distribution is expected to depend on the opacity , @xmath273 , where @xmath274 @xcite .
thus , the power - law index of the density distribution is estimated to be @xmath275 for @xmath276 ( see section [ subsec:3.2 ] ) , which shows a good agreement with the estimate from the combined @xmath78 data ( section [ subsubsec:4.1.1 ] ) .
there are several uncertainties included in the derived density distribution . in this section ,
we examine the effect of opacity , assumption of uniform temperature , index conversion from column density to volume density profiles , and @xmath78 fractional abundance .
first , we investigate the effect of the optical depth of @xmath78 line emission , because the decrement of derived column densities in the inner radii might be due to larger optical depth within the @xmath277 beam .
@xcite indicated that the @xmath78 emission is optically thin even at the center of an @xmath278 beam from the measured @xmath279 to @xmath78 peak intensity ratio of @xmath280 and expected abundance ratio , @xmath281 , of @xmath282 .
furthermore , the integrated intensity ratio of @xmath279 to @xmath78 implies that @xmath78 emission should be optically thin at the center . according to figure 1 in @xcite , the line ratio actually ranges from @xmath283 to @xmath284 across the velocity
; the lowest line ratio indicates that @xmath78 emission is marginally thick , @xmath285 at maximum in certain velocities .
this opacity enhancement is limited only to a narrow velocity range , so it does not introduce a significant error when estimating the column density near the center , i.e. , the underestimate .
therefore , the effect of the optical depth can not account for the turnover around the radius of @xmath286 .
second , we assumed that the core has a uniform temperature distribution that does not have much effect on the estimate of the column density .
if we adopt the temperature profile from @xcite , for example , then the difference in temperature between radii of @xmath187 and @xmath8 should be a factor of @xmath287 . from equation ( [ eq:2 ] )
, the uncertainty of the column density estimate expected by this temperature variation could be derived as @xmath288 at most .
third , the derivation of the density profile of the core from the column density distribution depends on the radial finiteness of the core traced by the @xmath78 molecular line emission . for a spherical core with a size of @xmath289 and a power - law density profile of @xmath290 ,
the column density that is given by integrating the densities along the line of sight at the impact parameter @xmath291 can be expressed as @xmath292 where @xmath293 and the variable @xmath294 is an angle between a radial vector and the plane of the sky .
therefore , the coefficient of @xmath200 is also a function of @xmath291 so that the column density profile deviates from a power - law of the form @xmath295 . moreover , if the core has different power - law dependencies between the inner and outer regions , then the simple prediction of the column density profile in equation ( [ eq:4 ] ) is not exactly valid .
figure [ fig:9 ] shows the column density profiles of cores that have power - law density distributions with a cutoff radius @xmath289 .
the dashed lines indicate the profiles for the predictions of @xmath296 . for the case of a single power - law density distribution of @xmath297
as shown in figure [ fig:9](a ) , the resulting column density coincides well with the dependency of @xmath298 at radii smaller than @xmath299 .
however , it tends to depart from the power - law dependency and become steeper with increasing radius .
figure [ fig:9](b ) shows the column density profile for the core with density distributions of @xmath300 for @xmath301 and @xmath297 for @xmath302 .
this plot demonstrates that the column density profile of a core that has inner and outer regions with different power - law density distributions can be estimated to be steeper than the simple prediction of @xmath295 .
we examined the uncertainties caused by these effects when converting the column density profile into the density profile , and we found that the true density profile is likely to be shallower by @xmath303 at most in the power - law index than for our estimates . finally , as for the abundance of @xmath78 , recent studies of chemical evolution in star - forming cores have shown that the abundance of @xmath304 , which is a daughter species of @xmath305 , decreases at radii that are smaller than the @xmath305 sublimation radius @xcite .
nevertheless , it is shown that the abundance of @xmath304 at the inner radii increases with the evolution of core collapse after a central stellar object is born .
@xcite simulated a large number of molecular line profiles from b335 using various physical models . by the use of a self - consistent chemical model with core evolution ,
the result showed that the abundance of @xmath304 hardly changes along the core radius . the shallower profile at smaller radii can be affected by the radial distribution of the @xmath78 abundance , although it is not expected to be dominant .
the possible effects of radial dependence in the @xmath78 abundance and the finite core radius oppose one another .
it is difficult to estimate how they contribute to our data .
nevertheless , by taking into account the agreement of the column density profiles in the inner region of the b335 core between the derivations from the combined @xmath78 and dust continuum data , we determine that the effects of variation in optical depth and radius dependence in the @xmath78 abundance are likely to be negligible .
moreover , the agreement between the profile indices derived from our data with those from the extinction analysis of near - infrared data indicates that the effect of these uncertainties , especially in the temperature and fractional abundance of @xmath78 , is not very considerable . in other words ,
our @xmath78 data well represent the column density structure of the core , which mostly ensures the validity of our analysis for the physics of star formation .
we performed model calculations of the pv diagrams and investigated which model reproduces the observed signatures of the pv diagram in the @xmath78 line well . for comparison with the observed results ,
the model calculations were performed with different radial distributions of the infalling velocities and also for three different assumed central stellar masses .
we assume a contracting spherical star - forming core with rotation as described in detail below .
most of the parameters for the calculations simulating pv diagrams were taken from our observational results and previous studies .
the core has a size of @xmath8 and the radial density profile in equation ( [ eq:3 ] ) shows power - law dependencies ( @xmath264 ) of @xmath306 in the inner ( @xmath307 ) and @xmath308 in the outer ( @xmath309 ) regions .
such a difference in power - law indices is naturally expected from the isothermal collapse model because of the boundary of the inner free - falling region and the outer region in which the condition at the stage of the formation of a central stellar object is expected to be conserved .
the boundary radius is referred to as the infalling radius ( @xmath310 ) , and therefore our derived density profile indicates @xmath311 .
we also introduced the rotational motion of the core .
the inclination angle of the rotation axis from the plane of the sky should be considered because it affects the line - of - sight velocity of rotational motion .
we assumed that the rotation axis corresponds to the outflow axis whose inclination angle for b335 was estimated to be @xmath312 @xcite , @xmath313 @xcite , and @xmath314 @xcite .
we adopted an inclination angle of @xmath312 for the rotation axis in our calculations . in the free - fall region with @xmath315 , we adopted the increasing rotational velocity which is inversely proportional to the radius owing to the angular momentum conservation during collapse .
such a velocity field in infalling envelopes was indeed suggested observationally ( e.g. , l1551-irs5 by * ? ? ?
the overall velocity gradient from the 45 m telescope observations , @xmath316 ( @xmath317 ) , can be regarded as the initial angular momentum of the core , which is still preserved in the outer radius of @xmath318 . for the thermal line broadening , we set the line - of - sight velocity width of @xmath319 for @xmath78 molecules at @xmath320 .
we calculated radiative transfer equations to estimate the relative intensity distributions and verify the effect of optical depth .
assuming a two - level state , the populations in rotational levels of an @xmath78 molecule were calculated using the einstein @xmath321 coefficient and the collisional rate coefficients with @xmath170 from and @xcite .
the final model pv diagrams were smoothed with gaussian functions whose fwhms are the same as actual resolutions of the combined image for both the spatial and velocity directions .
we conducted simulations by varying the parameters of the inward velocity in the outer region ( @xmath322 ) of the core , @xmath323 , and the mass of the central stellar object to compare with observational results .
we adopted the inward velocities in the outer region of @xmath324 , @xmath39 , and @xmath325 .
the central stellar mass ( @xmath326 ) for b335 was estimated in a wide range in previous studies , e.g. , @xcite estimated it to be @xmath327 while @xmath328 was suggested by @xcite . in our model calculations
, we used three values of the central stellar mass : @xmath329 , @xmath330 , and @xmath331 .
figure [ fig:10 ] shows the calculated pv diagrams perpendicular to the rotation axis passing through the core center . as for
the inward velocities in the outer region of the core , the calculated pv diagrams for @xmath324 well reproduce the features in the observed pv diagram .
the inward velocity in the outer region affects the overall line width in the pv diagram .
although the pv diagrams calculated with @xmath332 are also similar to the observed result , the overall velocity widths seem to be larger than that of the observed one .
the simulated diagrams with @xmath333 undoubtedly disagree with the observation .
we found that the central stellar mass mainly contributes line broadening at the core center , which means that the infalling motion is more dominant than the spin - up rotation owing to the conservation of angular momentum around the core center .
for the diagrams with @xmath324 , it seems that line wings at the center position are not enough in the case of @xmath329 while they are excessive in the case of @xmath334 .
consequently , we determine in our model calculations that the pv diagram with @xmath324 and @xmath335 well represents the observed pv diagram .
the calculated model pv diagram with @xmath324 and @xmath335 shows that the peak is skewed toward the blueshifted velocity and has a shoulder in the profile around the systemic velocity at the center position .
these features in the model diagram are due to the @xmath78 optical depth and are similar to the observational result .
hence , the suggestion discussed in section [ subsubsec:4.1.3 ] that the @xmath156@xmath1 line is marginally optically thick is probably reasonable .
in addition , we can obtain the radial distribution of the specific angular momentum from the profile of rotation velocity in the model calculations of pv diagrams .
figure [ fig:11 ] shows the radial distribution of specific angular momentum in the model that well represents the observation : the case of @xmath324 and @xmath335 .
we can see the spin - up rotation within the infall radius of @xmath311 as deviations from the solid - body rotation with an angular velocity of @xmath336 . in previous observational studies ,
the specific angular momenta of the core and envelope in b335 have been measured at several radii from velocity gradients across the outflow axis .
@xcite derived the specific angular momenta of @xmath337 at a radius of @xmath338 and @xmath339 at @xmath340 from velocity gradients seen in the @xmath341@xmath33 and @xmath342@xmath35 envelopes , respectively .
they discussed the evolution of specific angular momentum , including the measurements by @xcite of @xmath343 at @xmath258 and @xmath344 at @xmath8 .
these measurements are also indicated in figure [ fig:11 ] as open squares and open triangles .
this plot shows that our model calculations successfully derived the radial distribution of the specific angular momentum consistent with measurements from previous observational studies .
in particular , specific angular momenta that were recently estimated from high - resolution observations in submillimeter wavelengths at radii of @xmath340 and @xmath338 show good agreement with the results of our model calculations .
the density and velocity structures of a collapsing molecular cloud core are crucial to distinguish theoretical models of gravitational collapse .
we examine the observational results by comparing them with the key properties of the two star formation models , i.e. , the @xcite and larson penston solutions ( * ? ? ?
* ; * ? ? ?
* hereafter lp solution ) .
although the shu and lp pictures of core contraction are similar in terms of the power - law dependencies of density distribution and inward motion , it is possible to distinguish between the two models by quantitative comparisons .
the power - law dependency of the density profile in the b335 core qualitatively matches the similarity solutions for an isothermal spherical cloud in its post - protostar formation stage @xcite , which has also been demonstrated by numerical simulations ( e.g. , * ? ? ?
* ) ; i.e. , @xmath345 in the dynamical free - fall region and @xmath346 in the outer region where the condition at the protostar formation stage should be conserved . the absolute density of the core is one of the key parameters used to discriminate between isothermal collapse models .
we derived the number density of the b335 core as @xmath347 at the radius of @xmath205 . in shu s inside - out picture ,
the density distribution in the outer region of the core is expected to correspond to the singular isothermal sphere , @xmath348 . from this model ,
we obtain @xmath349 with @xmath350 . on the other hand ,
the `` runaway collapse '' of the lp solution during the core formation has a density that is @xmath351 times higher than that in shu s solution , so that we obtain @xmath352 at @xmath353 .
our derived density is comparable to that predicted by shu s solution with a factor of @xmath354 , however , it is considerably smaller than that of the lp solution .
the inward velocity in the outer region from our model calculations also supports shu s solution . as described above our model calculations of pv diagrams
successfully explain the observed features within the uncertainties , in which the mass of the central stellar source and the inward velocity in the outer core region are estimated . here
, we focus on the inward motion in the outer region , which is one of the key characteristics to distinguish the gravitational collapse models .
the static outer region of the core ( @xmath355 ) represents the velocity structure of shu s solution . on the other hand ,
numerical simulations of isothermal collapse models , which have been conducted to compare with similarity solutions , have showed that the inward velocity in the runaway collapse phase is not a constant but a function of age and radius .
these inward velocities can be observed as intermediate pictures between those of the lp and shu models , and are important characteristics of the collapse of an isothermal sphere .
the infall velocity in the runaway collapse phase is determined by the initial ratio of gravitational force to pressure force @xcite .
the cloud core that begins to collapse from a marginally unstable isothermal gas sphere has a subsonic inward velocity in the outer region .
even from our model comparisons with the pv diagram of the observed data , we can not exactly clarify if the outer region of the core has a subsonic inward velocity .
however , it is plausible that there is an inward velocity @xmath356 in the outer region , such as @xmath333 which represents lp solution , can be ruled out .
consequently , we suggest that the b335 core has initiated its gravitational collapse from a quasi - static initial condition similar to shu s model . otherwise , it is also possible to explain our observed results by means of an isothermal collapse of a cloud core that has a mass slightly larger than the bonner
ebert mass @xcite . in shu
s similarity solutions , an infalling radius provides a rough age of the cloud after a point source is formed at the core center , because the boundary between the infalling inner region and the outer region propagates outward as a rarefaction wave with a velocity of the isothermal sound speed . in our case
, we take the turnover radius to be the infalling radius and obtain an age of @xmath357 which is comparable to the order of the age of the class 0 phase . adopting the mass infall rate predicted by shu s solution , @xmath358 , along with the estimated age
, we obtain a central stellar mass of @xmath359 .
the estimate of the central stellar mass from our model calculation is closer to the prediction from shu s solution than that from the lp model , which expects a 48 times higher mass infall rate .
therefore , the above comparisons between our results and the theoretical models indicate that the picture of shu s model is more preferable than that of the lp solution for the b335 core .
we presented a study of the dense molecular cloud core harboring the low - mass protostar , b335 in the @xmath360 molecular line emission using the nobeyama 45 m telescope and the nma .
our main findings are summarized as follows .
the single - dish observations revealed a dense core with a size of @xmath159 .
our analysis using a combining technique of single - dish and interferometer data revealed the structure of the inner dense envelope within the core with a high spatial resolution of @xmath361 .
the envelope size is @xmath362 .
both of them have an elongated distribution toward the north south direction , perpendicular to the outflow axis .
the mass of the core is estimated to be @xmath363 \2 .
we determined the radial column density profile of the b335 core and found a reliable difference between the power - law indices of the outer and inner regions of the dense core .
the turnover radius is considered to be @xmath241 , which is consistent with the infalling radius estimated in previous work . our derived density profile , @xmath364 for @xmath365 and @xmath366 for @xmath367 ,
is better explained , both qualitatively and quantitatively , in the picture of shu s self - similar solution than in that of the lp solution .
the dense core shows a slight overall velocity gradient of @xmath7 over the scale of @xmath8 across the outflow axis .
this velocity gradient is considered to represent a solid - body rotation and corresponds to an angular velocity of @xmath368 .
our combined image also revealed detailed velocity structures in the dense core with a high resolution .
the velocity structure of the b335 core can be well explained in terms of the collapse of an isothermal sphere , in which the core has an inner free - fall region and an outer region preserving the condition at the stage of protostar formation .
we performed simple model calculations of pv diagrams to examine the observed diagrams .
the model calculations successfully reproduce observational results , while suggesting a central stellar mass of @xmath369 and a small inward velocity of @xmath370 in the outer region of the core @xmath11 .
quantitative comparisons of density and velocity structures from the observational results with theoretical models show an agreement with shu s quasi - static inside - out star formation .
furthermore , it is possible for the outer region of the b335 core to have a subsonic inward velocity .
we concluded that a picture of shu s solution or an isothermal collapse of a marginally stable bonnor
ebert sphere is suitable for the gravitational collapse of the b335 core .
this study was based on observations at the nobeyama radio observatory ( nro ) , which is a branch of the national astronomical observatory of japan , national institutes of natural sciences .
the authors are grateful to the staff at the nro for operating the nma and the 45 m telescope , helping us with the data reduction .
we also thank an anonymous referee whose comments significantly improved the paper . y.k .
thanks kenichi tatematsu and the staff at the naoj chile observatory for their helpful comments , continuing interest , and encouragement .
in section [ subsubsec:4.1.2 ] , we discuss the emission distribution of the inner region of the envelope using interferometric data in the @xmath108-@xmath109 domain .
the detailed expression of the analysis is described in this paper . for optically thin dust emission , the observed intensity from an envelope that has a spherical density distribution , @xmath14 , and a dust temperature , @xmath263 , as a function of the impact parameter , @xmath291 ,
is written as @xmath371 where @xmath372 is the outer radius of the envelope . if the density and temperature follow radial dependencies of power - laws , @xmath264 and @xmath265 , and if we assume that the opacity does not vary along the line of sight , then the intensity also has a power - law profile , @xmath266 , in the rayleigh
jeans regime .
we assume that the intensity distribution , @xmath262 , is more compact in extent than the primary beam of interferometric observations , and gain variations during the observations are properly corrected .
the visibility as a function of @xmath108@xmath109 distance , @xmath373 , can be given by the hankel transform of the intensity distribution , @xmath374 where @xmath375 is a zeroth - order bessel function .
equation ( [ eq : b2 ] ) is rewritten as a function of @xmath108@xmath109 distance @xmath267 , @xmath376 d\theta r_{\rm b } d r_{\rm b } , \label{eq : b3}\ ] ] where @xmath377 and @xmath378 . by definition ,
a zeroth - order bessel function is given by @xmath379 so we obtain @xmath380 this is the hankel transform of the intensity distribution .
we expect the intensity distribution to have a power - law dependency , this integral has a solution of the form @xcite @xmath381 for @xmath382 where @xmath383 is the gamma function : @xmath384 therefore , we obtain @xmath385 for @xmath386 the visibilities of an intensity distribution with a spherically symmetric power law , @xmath387 for @xmath388 , are a power law in the @xmath108@xmath109 domain , @xmath389 .
aikawa , y. , wakelam , v. , garrod , r. t. , & herbst , e. 2008 , , 674 , 984 andr , p. 1994 , in proc .
28th rencontre de moriond , the cold universe , ed
. t. montmerle et al .
( gif - sur - yvette , france : editions frontieres ) , 179 andre , p. , ward - thompson , d. , & barsony , m. 1993 , , 406 , 122 andre , p. , ward - thompson , d. , & motte , f. 1996 , , 314 , 625 barsony , m. 1994 , clouds , cores , and low mass stars , 65 , 197 beckwith , s. v. w. , henning , t. , & nakagawa , y. 2000 , protostars and planets iv , 533 cabrit , s. , goldsmith , p. f. , & snell , r. l. 1988 , , 334 , 196 caselli , p. , benson , p. j. , myers , p. c. , & tafalla , m. 2002 , , 572 , 238 chandler , c. j. , gear , w. k. , sandell , g. , et al . 1990 , , 243 , 330 chandler , c. j. , & sargent , a. i. 1993 , , 414 , l29 chen , x. , launhardt , r. , & henning , t. 2007 , , 669 , 1058 choi , m. , evans , n. j. , ii , gregersen , e. m. , & wang , y. 1995 , , 448 , 742 clemens , d. p. , & barvainis , r. 1988 , , 68 , 257 danby , g. , flower , d. r. , valiron , p. , schilke , p. , & walmsley , c. m. 1988 , , 235 , 229 doty , s. d. , & leung , c. m. 1994 , , 424 , 729 doty , s. d. , tidman , r. , shirley , y. , & jackson , a. 2010 , , 406 , 1190 evans , n. j. , ii , lee , j .- e . , rawlings , j. m. c. , & choi , m. 2005 , , 626 , 919 flower , d. r. 1999 , , 305 , 651 foster , p. n. , & chevalier , r. a. 1993 , , 416 , 303 frerking , m. a. , & langer , w. d. 1982 , , 256 , 523 frerking , m. a. , langer , w. d. , & wilson , r. w. 1987 , , 313 , 320 furuya , r. s. , kitamura , y. , & shinnaga , h. 2006 , , 653 , 1369 gerin , m. , goicoechea , j. r. , pety , j. , & hily - blant , p. 2009
, , 494 , 977 goodman , a. a. , benson , p. j. , fuller , g. a. , & myers , p. c. 1993 , , 406 , 528 gradshteyn , i. s. , & ryzhik , i. m. 1994 , table of integrals , series , and products ( 5th ed . ; new york : academic ) , 668 haese , n. n. , & woods , r. c. 1979 , chemical physics letters , 61 , 396 harvey , d. w. a. , wilner , d. j. , lada , c. j. , et al .
2001 , , 563 , 903 harvey , d. w. a. , wilner , d. j. , myers , p. c. , tafalla , m. , & mardones , d. 2003 , , 583 , 809 harvey , d. w. a. , wilner , d. j. , myers , p. c. , & tafalla , m. 2003 , , 596 , 383 hasegawa , t. i. , rogers , c. , & hayashi , s. s. 1991 , , 374 , 177 hirano , n. , kameya , o. , nakayama , m. , & takakubo , k. 1988 , , 327 , l69 hirano , n. , kameya , o. , kasuga , t. , & umemoto , t. 1992 , , 390 , l85 huard , t. l. , sandell , g. , & weintraub , d. a. 1999 , , 526 , 833 hunter , c. 1977 , , 218 , 834 keene , j. , davidson , j. a. , harper , d. a. , et al .
1983 , , 274 , l43 kurono , y. , morita , k .-
i . , & kamazaki , t. 2009 , , 61 , 873 larson , r. b. 1969 , , 145 , 271 lee , c. w. , myers , p. c. , & tafalla , m. 1999 , , 526 , 788 lee , c. w. , myers , p. c. , & tafalla , m. 2001 , , 136 , 703 lee , c. w. , myers , p. c. , & plume , r. 2004 , , 153 , 523 lee , j .- e . , bergin , e. a. , & evans , n. j. , ii 2004 , , 617 , 360 mangum , j. g. , wootten , a. , & mundy , l. g. 1992 , , 388 , 467 menten , k. m. , harju , j. , olano , c. a. , & walmsley , c. m. 1989 , , 223 , 258 momose , m. , ohashi , n. , kawabe , r. , nakano , t. , & hayashi , m. 1998 , , 504 , 314 moriarty - schieven , g. h. , & snell , r. l. 1989 , , 338 , 952 ogino , s. , tomisaka , k. , & nakamura , f. 1999 , , 51 , 637 okumura , s. k. , momose , m. , kawaguchi , n. , et al .
2000 , , 52 , 393 penston , m. v. 1969 , , 144 , 425 saito , m. , sunada , k. , kawabe , r. , kitamura , y. , & hirano , n. 1999 , , 518 , 334 sawada , t. , ikeda , n. , sunada , k. , et al .
2008 , , 60 , 445 shirley , y. l. , evans , n. j. , ii , rawlings , j. m. c. , & gregersen , e. m. 2000 , , 131 , 249 shirley , y. l. , evans , n. j. , ii , & rawlings , j. m. c. 2002 , , 575 , 337 shu , f. h. 1977 , , 214 , 488 stutz , a. m. , rubin , m. , werner , m. w. , et al .
2008 , , 687 , 389 sunada , k. , yamaguchi , c. , nakai , n. , et al .
2000 , , 4015 , 237 tafalla , m. , mardones , d. , myers , p. c. , et al .
1998 , , 504 , 900 takakuwa , s. , ohashi , n. , bourke , t. l. , et al .
2007 , , 662 , 431 tobin , j. j. , hartmann , l. , chiang , h .- f . , et al .
2011 , , 740 , 45 velusamy , t. , kuiper , t. b. h. , & langer , w. d. 1995 , , 451 , l75 ward - thompson , d. , scott , p. f. , hills , r. e. , & andre , p. 1994
, , 268 , 276 ward - thompson , d. , motte , f. , & andre , p. 1999
, , 305 , 143 wilner , d. j. , & welch , w. j. 1994 , , 427 , 898 wilner , d. j. , myers , p. c. , mardones , d. , & tafalla , m. 2000 , , 544 , l69 whitworth , a. , & summers , d. 1985 , , 214 , 1 yamaguchi , c. , sunada , k. , iizuka , y. , iwashita , h. , & noguchi , t. 2000 , , 4015 , 614 yen , h .- w . , takakuwa , s. , & ohashi , n. 2010 , , 710 , 1786 yen , h .- w . , takakuwa , s. , & ohashi , n. 2011 , , 742 , 57 zhou , s. , evans , n. j. , ii , koempe , c. , & walmsley , c. m. 1993 , , 404 , 232 zhou , s. , evans , n. j. , ii , koempe , c. , & walmsley , c. m. 1994 , , 421 , 854 cccccccccccccccc @xmath83 & @xmath390 & h22 & @xmath391 & @xmath392 & @xmath393 & @xmath394 & ps & c + @xmath395 & @xmath396 & h22 & @xmath391 & @xmath392 & @xmath397 & @xmath398 & ps & c + @xmath399 & @xmath400 & h22 & @xmath391 & @xmath392 & @xmath401 & @xmath402 & ps & c + @xmath0@xmath1 & @xmath403 & bears & @xmath404 & @xmath67 & @xmath405 & @xmath406 & otf & @xmath407 + [ tbl:1 ] ( upper panel ) and @xmath416 ( lower panel ) line profiles ( black histogram ) observed with the nobeyama 45 m telescope toward the b335 .
blue curves represent the best - fit results of gaussian fitting ( section [ subsec:3.1 ] ) .
the rms noise levels are @xmath417 in @xmath117 for both spectra . ]
continuum image toward the iras source in b335 observed with the nma .
contours start from the @xmath133 level with a @xmath133 interval , where @xmath418 . dashed contours show the @xmath419 level .
solid and dashed arrows in the image indicate the outflow directions of blue and red lobes , respectively .
the ellipse at the bottom right corner indicates the synthesized beam size ( @xmath420 with a position angle of @xmath421 ) . ]
@xmath1 line emission obtained with the 45 m telescope ( left ) and by combining the 45 m telescope and nma data ( right ) .
contour intervals are @xmath133 , starting from the @xmath133 levels of the images where @xmath80 noise levels are @xmath422 in @xmath117 for the 45 m telescope map and @xmath423 for the combined map .
dashed contours show the @xmath419 level .
the open star in each map is the peak position of the @xmath65 continuum emission observed with the nma .
the beam size for each map is shown as a filled circle or filled ellipse at the bottom right corner .
dotted circle in the combined image indicates the field of view , i.e. , fwhm primary beam size of the nma observations . ]
@xmath1 line emission taken with the 45 m telescope .
the central lsr velocity in @xmath424 for each channel is shown in the upper left corner .
contour intervals are the same as in figure [ fig:3 ] , and the @xmath80 noise level is @xmath425 in @xmath117 .
the open star , filled circle in each panel , and dotted circle are the same as in figure [ fig:3 ] . ]
@xmath1 line emission obtained by combining the 45 m telescope and nma data .
the central lsr velocity in @xmath424 for each channel is shown in the upper left corner .
contour intervals are the same as in figure [ fig:3 ] and @xmath80 noise level is @xmath426 .
the open star , filled ellipse in each panel , and dotted circle are the same as in figure [ fig:3 ] . ]
column density in the b335 core made from the @xmath0@xmath1 image obtained with the 45 m telescope only ( filled circles ) and the combined 45 m telescope and nma data ( filled triangles ) .
panel ( a ) shows the profile without the mask in the cavity regions and panel ( b ) shows that with the mask .
dot - dashed and dashed lines in the plot indicate the best - fit results of power - law fitting with the fitting range of @xmath206@xmath205 and @xmath207@xmath208 , respectively .
dotted curves indicate approximated beam patterns of 45 m telescope image and the combined image , gaussian functions with the fwhm of @xmath427 and @xmath428 , respectively . ]
@xmath1 line emission for the b335 core made from the 45 m telescope image data ( blue contours ) and the combined data ( gray scale with black contours ) .
panel ( a ) and ( b ) , respectively , show the diagrams along @xmath429 ( perpendicular to the outflow axis ) and @xmath430 ( along the outflow axis ) , passing through the @xmath65 continuum source . ]
continuum data taken with the nma as a function of @xmath108@xmath109 distance .
binning is logarithmic and amplitudes are obtained as vectorial average of the complex visibilities in each bin . a solid line shows the best - fit power - law function . ] .
panel ( a ) shows the case of a density distribution of @xmath297 and panel ( b ) shows the case of a density distribution of @xmath431 ( @xmath301 ) and @xmath297 ( @xmath302 ) .
dashed lines indicate column density profiles expected from simple predictions of @xmath298 for ( a ) , and @xmath432 ( @xmath433 ) and @xmath298 ( @xmath302 ) for ( b ) . ] , @xmath39 , and @xmath325 with the central stellar mass of @xmath329 , respectively .
panels ( d)(f ) show the diagrams in the cases of @xmath335 and panels ( g)(i ) show the cases of @xmath334 .
pv diagram of the @xmath0@xmath1 line emission from b335 perpendicular to the outflow axis is shown in panel ( j ) with black contours .
contour intervals of model pv diagrams relative to the peaks are the same as those for the observed diagram . ] ) for the case of @xmath324 and @xmath335 ( cross marks with solid line ) as a function of radius .
open squares represent specific angular momenta measured from velocity gradients seen in the @xmath341@xmath33 and @xmath342@xmath35 envelopes by @xcite , and open triangles denote specific angular momenta measured from @xmath243@xmath1 core and @xmath0@xmath1 envelope by @xcite .
a dashed line indicates the profile of specific angular momentum of @xmath434 . ] | we present an observational study of the protostellar core b335 harboring a low - mass class 0 source .
the observations of the @xmath0@xmath1 line emission were carried out using the nobeyama 45 m telescope and nobeyama millimeter array .
our combined image of the interferometer and single - dish data depicts detailed structures of the dense envelope within the core .
we found that the core has a radial density profile of @xmath2 and a reliable difference in the power - law indices between the outer and inner regions of the core : @xmath3 for @xmath4 and @xmath5 for @xmath6 .
the dense core shows a slight overall velocity gradient of @xmath7 over the scale of @xmath8 across the outflow axis .
we believe that this velocity gradient represents a solid - body - like rotation of the core .
the dense envelope has a quite symmetrical velocity structure with a remarkable line broadening toward the core center , which is especially prominent in the position velocity diagram across the outflow axis .
the model calculations of position velocity diagrams do a good job of reproducing observational results using the collapse model of an isothermal sphere in which the core has an inner free - fall region and an outer region conserving the conditions at the formation stage of a central stellar object .
we derived a central stellar mass of @xmath9 , and suggest a small inward velocity , @xmath10 in the outer core at @xmath11 .
we concluded that our data can be well explained by gravitational collapse with a quasi - static initial condition , such as shu s model , or by the isothermal collapse of a marginally critical bonnor
ebert sphere . |
the main strategy behind the modelling approach we follow is to first calculate the collapse and merging history of individual dark matter halos , which is governed purely by gravitational interactions , and secondly to calculate the more complex physics of the baryons inside these dark matter halos , including e.g. radiative cooling of the gas , star formation and feedback from supernovae by simplified prescriptions on top of the dark matter evolution .
each of the dark matter halos consist of three main components which are distributed among individual galaxies inside them . a stellar , cold and hot gas component , where the latter is only attributed to _
central _ galaxies , the most massive galaxies inside individual halos . in the following sections
, we will describe briefly the recipes used to calculate these different components which are mainly based on recipes presented in kauffmann et al.@xcite ( hereafter , k99 ) and springel et al.@xcite ( hereafter , s01 ) , and we refer readers for more details on model implementations to their work and references therein . throughout this paper
we use the following set of cosmological parameters : @xmath24 , @xmath25 , @xmath26 , @xmath27 and @xmath28 .
we calculate the merging history of dark matter halos according to the prescription presented in somerville et al.@xcite .
this approach has been shown to produce merging histories and progenitor distributions in reasonable agreement with result from n - body simulations of cold dark matter structure formation in a cosmological context@xcite .
the merging history of dark matter halos is reconstructed by breaking each halo up into progenitors above a limiting minimum progenitor mass @xmath29 .
this mass cut needs to be chosen carefully as it ensures that the right galaxy population and merging histories are produced within the model .
progenitor halos with masses below @xmath29 are declared as _ accretion _ events and their histories are not followed further back in time .
progenitors labelled as accretion events should ideally not host any significant galaxies in them and be composed mainly of primordial hot gas at the progenitor halo s virial temperature . to achieve a good compromise between accuracy and computational time , we use @xmath30 @xmath31 , which ensures that the results presented here are unaffected by numerical resolution effects .
as mentioned above , once the merging history of the dark matter component has been calculated , it is possible to follow the evolution of the baryonic content in these halos forward in time .
we assume each halo consists of three components : hot gas , cold gas and stars , where the latter two components can be distributed among individual galaxies inside a single dark matter halo .
the stellar components of each galaxy are further divided into bulge and disc , to allow morphological classifications of model galaxies . in the following ,
we describe how the evolution of each component is calculated .
each branch of the merger tree starts at a progenitor mass of @xmath29 and ends at a redshift of @xmath32 .
initially , each halo is occupied by hot primordial gas which was captured in the potential well of the halo and shock - heated to its virial temperature @xmath33 ^ 2 $ ] k , where @xmath34 is the circular velocity of the halo@xcite .
subsequently this hot gas component is allowed to radiatively cool and settles down into a rotationally supported gas disc at the centre of the halo , which we identify as the central galaxy@xcite .
the rate at which hot gas cools down is estimated by calculating the cooling radius inside the halo using the cooling functions provided by @xcite and the prescription in s01 . in the case of a merger between halos we assume that all of the hot gas present in the progenitors gets shock - heated to the virial temperature of the remnant halo , and this gas can cool down only onto the new central galaxy which is the central galaxy of the most massive progenitor halo . the central galaxy of the less massive halo will become a satellite galaxy orbiting inside the remnant halo . in this way , a halo can host multiple satellite galaxies , depending on the merging history of the halo , but will always only host one central galaxy onto which gas can cool .
the cold gas content in satellite galaxies is given by the amount present when they first became satellite galaxies and does not increase , but decreases due to ongoing star formation and supernova feedback .
in the simplified picture adopted above , the amount of gas available to cool down is limited only by the universal baryon fraction @xmath35@xcite . however , in the presence of a photoionising background the fraction of baryons captured in halos is reduced @xcite and we use the recipe of somerville et al.@xcite , which is based on a fitting formulae derived from hydrodynamical simulations@xcite , to estimate the amount of baryons in each halo . for the epoch of reionisation , we assume @xmath36 , which is in agreement with observations of the temperature - polarisation correlation of the cosmic microwave background@xcite .
once cooled gas has settled down in a disc , we allow for fragmentation and subsequent star formation according to a parameterised global schmidt - kennicutt law@xcite of the form @xmath37 , where @xmath38 is a free parameter describing the efficiency of the conversion of cold gas into stars , and @xmath39 is assumed to be the dynamical time of the galaxy and is approximated to be 0.1 times the dynamical time of the dark matter halo@xcite . as in k99 we allow star formation only in halos of @xmath40 km / s to avoid excessively - massive central galaxies in clusters .
feedback from supernovae plays an important role in regulating star formation in small mass halos and in preventing excessively - massive satellite galaxies from forming .
we implement feedback based on the prescription presented in k99 with @xmath41 here we introduce a second free parameter @xmath42 which represents our ignorance about the efficiency with which the energy from supernovae reheats the cold gas .
the expected number of supernovae per solar mass of stars formed is given by @xmath43 , taken as the value for the scalo initial mass function@xcite , and @xmath44 erg is the energy output from each supernova .
we take @xmath34 as the circular velocity of the halo in which the galaxy was last present as a central galaxy .
we allow for mergers between galaxies residing in a single halo .
as mentioned earlier , each halo is occupied by one central galaxy and a number of satellite galaxies depending on the past merging history of the halo .
whenever two halos merge , the galaxies inside them are going to merge on a time - scale which we calculate by estimating the time it would take the satellite to reach the centre of the halo under the effects of dynamical friction .
satellites are assumed to merge only with central galaxies and we set up their orbits in the halo according to the prescription of k99 , modified to use the coulomb logarithm approximation of s01 .
if the mass ratio between the two merging galaxies is @xmath45 ( @xmath46 ) we declare the event as a _
major _ merger and the remnant will be an elliptical galaxy and the stellar components and the gas will be treated according to the prescriptions below . in the case of _ minor _ merger @xmath47 the cold gas in the disc of the smaller progenitor
is assumed to settle down in the gas disc of the remnant and its stars contribute to the bulge component of the remnant@xcite .
it is suggested that major mergers will lead to the formation of elliptical galaxies@xcite .
indeed detailed numerical simulations in the last decade seem to support this hypothesis@xcite , and we wil assume in the following that major mergers disrupt the discs in progenitor galaxies as seen in various numerical simulations and relax to a spheroidal distribution . during the merger , any cold gas in the discs of the progenitor galaxies
is assumed to be funnelled into the centre of the remnant where it ignites a starburst which transforms all of the cold gas into stars contributing to the spheroidal component@xcite .
the second assumption is certainly a simplification of what might happen since we neglect the possibility that not all of the cold gas is funnelled to the centre but some fraction of it may for example settle down onto an extended disk which continues to grow inside - out by fresh supply of gas from tidal tails@xcite .
the results of barnes@xcite indicate that @xmath48 of the initial gas in the disc could end up in the central region of the remnant and be consumed in a starburst .
the exact amount is somewhat dependent on the merger geometry and on the mass ratio of the merger .
unfortunately , a sufficiently large survey investigating the gas inflow to the centres of merger remnants is not available yet , so we use the simplified approach of assuming that all cold gas gets used up in the central starburst .
this prescription for the fate of the cold gas results in an overestimate of the spheroid masses and an underestimate of the secondary disc components in our model .
this is not very significant for massive ellipticals since they are mainly formed in relatively non - dissipative mergers@xcite .
we here follow the model introduced in kauffmann & haehnelt@xcite in which super - massive black holes get formed and fed during major mergers of glaxies .
the assumption is to say that a fraction of the the cold gas available in the progenitor discs , that is funnelled into the centre of the remnant , will be accreted onto the black hole .
kaufmann & haehnelt introduced the following scaling law for the effectivness of this procces : @xmath49 where @xmath50 is the amount of cold gas available in the progenitor disks , @xmath34 the circular velocity of the dark matter halo and @xmath51 a free parameter .
this scaling provides a good fit to the observed relation between the mass of the super - massive black hole and the velocity dispersion of a galaxy@xcite .
following croton et al.@xcite we here assume the velocity dispersion of a galaxy to be identical to the circular velocity of the dark halo the galaxy is embedded in
. this however is not strictly true as e.g. in the case of an isothermal sphere @xmath52 .
several authors have investigated the correlation between @xmath34 and @xmath3@xcite and found different correlations .
we here note that using a different scaling between @xmath34 and @xmath3 in our models still produces the same @xmath53relation if we adjust the free parameter @xmath51 accordingly .
we here introduce a new prescription for the modelling of feedback from super - massive black holes , which is based on the results presented in this paper . according to our analyses a critical black hole mass exists for a given galaxy velocity dispersion @xmath3 , at
which feedback is so strong that it reheats all the available cold gas and hence prevents further star formation .
this critical black hole mass is derived from our empirical fit to the residual star formation fraction found in our data as @xmath54 in galaxies with black holes more massive than the corresponding critical black hole mass we stop any cooling of gas and star formation .
we callibrate this relation to give the right fraction of rsf galaxies as seen in figure 1 .
we normalise our two model parameters for the star formation efficiency @xmath38 and supernova feedback efficiency @xmath55 by matching the @xmath56band tully - fisher relation of giovanelli et al.@xcite and requiring that spiral central galaxies of halos with circular velocity @xmath57 km / s have on average @xmath58 m@xmath59 of stars and @xmath60 m@xmath59 of cold gas@xcite .
the third free parameter @xmath51 , which regulates the black hole growth is set to match the observed @xmath53relation and is @xmath61 .
we derive the @xmath62 colour-@xmath3 relation for the virgo cluster down to much smaller values of @xmath3 than in our galex - sdss sample by combining the near - uv and optical photometry from boselli et al.@xcite with velocity dispersion measurements from the goldmine database .
the purpose of looking at the virgo cluster galaxies is to constrain the low end of the @xmath3 and so get an idea of the point at which all galaxies are rsf galaxies and thus below critical . since the optical photometry of the virgo galaxies in this sample are in v - band , we approximately convert the rsf criterion of _ _ nuv - r__=5.4 to _ nuv - v_. we use simple stellar populations of solar metallicity and three ages ( 3,6 and 9 gyr ) to derive the _ nuv - v _ rsf criterion ; these are _ _ nuv - v__= 5.16 , 5.12 and 5.09 respectively . in all three cases ,
the point at which no virgo early - type galaxy is above this cutoff - i.e. all show some signs of young populations - is at @xmath63 .
we use _ nuv - r _ colour as a discriminant between those galaxies which are quiescent and those which show signs of recent star formation . besides agn
, there is one further effect which can mimic young stars in the near - uv .
extremely old stellar populations can give rise to the uv upturn phenomenon in some early - type galaxies@xcite .
we choose the _ nuv - r _ colour of one of the strongest uv upturn galaxies ngc 4552@xcite to be the boundary between galaxies with no young component on the one hand and those that are so blue they must have some young stars in them .
although still quite limited , theoretical population synthesis models which are supported by empirical data ( lee et al.@xcite ) also suggest that the nuv - r of passively evolving etgs could not be much bluer than 5.4 . using this criterion , we compute the fraction of galaxies that must have experienced some star formation within the last billion years .
the ages and mass fractions of these young stellar components are generally 300 - 500 myr and 1 - 3% by mass@xcite . due to the high sensitivity to dust extinction , these fractions are a lower limit as cold gas and dust have been detected in many early - type galaxies@xcite . _ nuv - r _ probes recent star formation for up to 1 gyr and so traces different time scales than those of agn feedback .
so there may be a certain amount of time lag between the uv emissions and the processes underlying feedback .
the process and criteria of the visual inspection are described in detail in schawinski et al.@xcite . in summary , we find early - type galaxy ( etg ) candidates from the sdss database using @xmath64 parameter , which is the weight of the de vaucouleurs component in the two - component ( de vaucouleurs and exponential disc ) fits .
we use @xmath65 , which is a highly conservative criterion , hoping to exclude as many spiral interpolers as possible . despite this , still some spiral bulges remain in our sample .
hence , we remove them via visual inspection on @xmath66 bands@xcite .
since visual inspections are also subject to errors , we investigat out to what redshift and apparent magnitude visual inspection based on sdss images is reliable by comparing them with one of the combo-17 fields overlapping sdss .
the combo-17 image is significantly deeper and has much better seeing with 0.7 `` as opposed to 1.4 '' which are typical for sdss . by comparing our classification based on these two different data sets
, we concluded that visual inspection of morphology using sdss was possible to z @xmath67 0.13 and r @xmath68 .
we settle on more conservative limits of z @xmath69 0.1 and r @xmath69 16.8 .
the presence of late - type interlopers with star - forming disks or spiral arms is therefore not a significant concern and does not give rise to the correlations between @xmath70 , @xmath3 and nuv - r color .
a selection of sample images with their velocity dispersions are shown in figure [ vi_ex ] . | detailed high resolution observations of the innermost central parts of nearby galaxies have revealed the presence of supermassive black holes@xcite .
these black holes may interact with their host galaxies by means of feedback in the form of energy and material jets ; this feedback affects the evolution of the host and gives rise to the observed relations between the black hole and the host . here
we report observations of the ultraviolet emissions of massive early - type galaxies .
we derive an empirical relation for a critical black - hole mass ( as a function of velocity dispersion ) above which the outflows from these black holes suppress star formation in their hosts by heating and expelling all available cold gas .
supermassive black holes are negligible in mass compared to their hosts but nevertheless seem to play a critical role in the star formation history of galaxies .
department of physics , university of oxford , oxford ox1 3rh , uk center for space astrophysics , yonsei university , seoul 120 - 749 , korea laboratoire dastrophysique de marseille , 13376 marseille cedex 12 , france california institute of technology , mc 405 - 47 , pasadena , ca 91125 laboratory for astronomy and solar physics , nasa goddard space flight center , greenbelt , md 20771 department of astronomy , columbia university , mc 5246 , new york , ny 10027 department of physics and astronomy , johns hopkins university , baltimore , md 21218 observatories of the carnegie institution of washington , 813 santa barbara st . , pasadena , ca 91101 department of physics and astronomy , ucla , los angeles , ca 90095 - 1562 the near - uv ( nuv ) detector of the galaxy evolution explorer satellite _ _ galex__@xcite covers a range in wavelength between @xmath0 and @xmath1 and is extremely sensitive to young stellar populations . with it
, we can detect small mass fractions of 1 - 3% of young stars formed within the last billion years@xcite .
this high sensitivity allows us to trace ongoing residual star formation in present day early - type galaxies ( etgs ) .
we select a volume - limited sample of visually inspected etgs from the sloan digital sky survey@xcite that have been observed by galex .
we remove active galactic nucleus ( agn ) candidates from our sample to avoid confusion between the observed uv emissions from the agn and from young stars @xcite .
we use @xmath2 colour as a probe of small amounts of recent star formation ( rsf ) and find that the fraction of etgs that show signs of rsf is strongly correlated with the stellar velocity dispersion @xmath3 , ( hereafter `` rsf-@xmath3 relation '' ) , but not with the luminosity of the galaxy ( see figure 1 ) . the correlation with @xmath3
can not be explained by a systematic trend of @xmath3 with internal extinction , because such a trend would be exact opposite@xcite to that observed .
we use a semi - analytic model of galaxy evolution in the @xmath4cdm paradigm to interpret these results@xcite ( for model details see supplements ) .
our simulation includes all the systematic processes that affect galaxy evolution and takes into account the different merger and star formation histories of galaxies@xcite .
when we run this simulation and investigate the predicted amount of recent star formation in low redshift etgs , we find that all galaxies underwent enough star formation within the last billion years to be detected by the extremely sensitive galex nuv detector .
massive galaxies continued to gain cold gas as fuel for star formation from mergers with satellites and through the cooling of hot gas present in those galaxies .
this overproduction of stars in massive galaxies is a known problem in models of galaxy evolution . in order to prevent star formation ,
a shutdown mechanism is required : cold gas must be heated up and expelled to prevent it from turning into stars .
feedback from supernovae ( sne ) was once a prime suspect , but it turned out to be too low to be effective in massive galaxies which have deep potential wells@xcite .
all the systematics implicit in the @xmath4cdm paradigm , including feedback from sne have already been included in our simulation , so we turn to feedback from active galactic nuclei ( agn ) powered by supermassive black holes ( smbh ) as the remaining most likely mechanism sufficiently powerful to shut down star formation in massive galaxies to reproduce the results of our observations@xcite .
energetic outflows caused by gas falling onto the supermassive black holes thought to reside at the centres of galaxies can self - regulate the growth of the black hole and the star formation in the galaxy by heating the available gas and blowing it out .
numerical simulations have shown that the energy input from a black hole in such an agn phase is capable of doing this@xcite .
as sne feedback is incapable of producing the quiescent galaxies we observe with galex , we conclude that agn feedback is the most likely process by which the available gas is heated and expelled in massive etgs .
in addition to this , the nature of the observed correlation with sigma ( see figure 1 ) allows us to infer more about the way in which agn feedback regulates galaxy evolution .
the masses of black holes correlate strongly with various properties of the host galaxies and the relation thought to have the smallest intrinsic scatter is that between @xmath5 and the velocity dispersion @xmath3 as a proxy for the depth of the gravitational potential@xcite . by considering the maximum strength of an active galactic nucleus given in terms of the eddington luminosity @xmath6 and expressing the gravitational potential depth by the stellar velocity dispersion @xmath3 , it is possible to derive a minimum black hole mass for which the outflow is capable of depleting all gas from the potential . according to our observations , massive etgs which have large values of @xmath3 , are on average less likely to have undergone recent star formation compared to less massive etgs : the `` rsf-@xmath3 relation '' .
thus , for a given velocity dispersion , we assume that there is a _
critical supermassive black hole mass _ above which the feedback powered by the black hole is sufficiently powerful to heat up and/or expel all gas in the host galaxy , thus preventing it from forming any stars even at the 1% level .
the observed intrinsic scatter in the @xmath5@xmath7@xmath3 relation gives a range of black hole masses for a given value of @xmath3 .
hence , in order to reproduce the decreasing fraction of rsf etgs with increasing @xmath3 , the @xmath8@xmath7@xmath3 relation should be less steep than the @xmath5@xmath7@xmath3 relation itself .
we can derive an empirical @xmath9@xmath7@xmath3 relation from our observed trend assuming a slope and scatter for the @xmath5@xmath7@xmath3 relation .
the @xmath10 relation is usually expressed as a power law of the form @xmath11 , so we also try to find such parameters @xmath12 and @xmath13 in the new @xmath8@xmath7@xmath3 relation that result in the observed `` rsf-@xmath3 relation '' .
we generate black hole masses for the galaxies in our sample by putting their velocity dispersions into the @xmath10 relation given by tremaine et al.@xcite and perturbing the black hole masses by the scatter of the relation ; we do this ten times for each galaxy ( figure 2 ( a ) ) .
then we vary @xmath12 and @xmath13 until we find the best match to the observed rsf-@xmath3 relation .
we find that with 95% confidence @xmath14 . as we are studying a volume and magnitude - limited sample , the @xmath8@xmath7@xmath3 relation derived from the monte - carlo simulation is only an approximation and we need to perform full simulations , as they are complete in @xmath3 as well as in absolute magnitude , to derive the actual @xmath8@xmath7@xmath3 relation .
we perform a more realisitc simulation using our semi - analytic models of galaxy evolution in order to test the plausibility of the hypothesis of a critical black hole mass controlling star formation via agn feedback .
this is a standard semi - analytic simulation in the @xmath15 paradigm that already includes various heating and cooling processes ( e.g. , sne ) .
our model does indeed reproduce the observed trend , but only when agn feedback is included , and the best agreement with the observations ( figure 1 ) is achieved by the relation @xmath16 ( figure 2b ) , which is in good agreement with our initial guess .
the agreement in the lowest @xmath17bin is poorer than in the others due to an increased scatter in our simulated faber - jackson relation relative to the observed one@xcite .
the supermassive black holes grown in our simulation have an intrinsic scatter of a factor of 1.5 in mass and are in good agreement with the relation suggested by tremaine et al.@xcite as our @xmath18 relation is less steep than the @xmath10 relation we expect an increasing fraction of star forming early - type galaxies with decreasing velocity dispersion until eventually no agn will be sufficiently powerful to shut down all star formation .
boselli et al.@xcite find that the _ nuv - v _ ( optical ) and _ nuv - h _
( near ir ) colour - magnitude relations for the early - type population of the virgo cluster shows precisely this kind of trend : low luminosity - and therefore less massive - virgo early - type galaxies are increasingly bluer and eventually all lie below the rsf cut - off .
we use stellar models@xcite to convert our rsf criterion to the _ nuv - v _ colours of the virgo galaxies and find that below @xmath19 all early - types are classified as rsf .
our simulations agree with this cutoff very well .
the possible implications of the @xmath18 relation on galaxy formation and the growth of black holes can be summarized by three main regions ( see figure 3 ) : first , the low @xmath3-region , in which the majority of black holes are too small to be able to halt star formation in their host galaxies by feedback during their initial accretion phase . during this phase , black holes
grow mainly by accreting gas at the eddington rate and are only limited in their growth by the amount of available fuel .
star formation on the other hand is regulated by the energy output from supernovae into the interstellar medium and will be proportional to the surface density of the gas@xcite .
this region extends up to @xmath20 as suggested earlier .
the next phase is at @xmath21 .
black holes are now becoming on average massive enough to be able to terminate star formation by their feedback and at the same time limit their available fuel .
the mass growth of the black holes is still dominated by accretion but mergers between black holes start playing an increasingly important role .
most host galaxies will be identified as agn or qso when observed during the accretion phase onto the black hole .
we sub - divide this region into two : at the lower end , a minority of galaxies will have critical black holes ; the black hole still grow mostly via accretion . at the high end , most black holes are critical and
so most of the mass growth now occurs via mergers .
the boundary between these two is at the intersection between the @xmath10 and the critical relation , i.e. @xmath22 .
finally , for galaxies with @xmath23 the vast majority of black holes is massive enough to halt all star formation and so star formation becomes virtually prohibited in such massive galaxies .
both the galaxy and the supermassive black holes now mainly grow via dry mergers@xcite .
further observations of nearby galaxies hosting supermassive black holes in combination with star formation estimates will help to shed more light into the scenario we propose and will place a strong test for the existence of `` a critical black hole mass '' in galaxies . 10 url # 1`#1`urlprefix[2]#2 [ 2][]#2 _ et al . _ .
_ _ * * , ( ) . & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
& . _
_ * * , ( ) . & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _ _ ( ) . .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ ( ) . .
_ et al . _ . _
_ * * , ( ) .
& .
_ _ * * , ( ) . , , & .
_ _ * * , ( ) . & .
_ _ * * , ( ) . & .
_ _ ( ) . .
& . _
_ * * , ( ) . , & .
_ _ * * , ( ) .
_ et al . _ .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) . , & .
_ _ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ et al . _ . _
_ * * , ( ) .
_ _ * * , ( ) . .
_ _ * * , ( ) . & .
_ _ * * , ( ) .
we would like to thank john magorrian for discussions and comments .
galex ( galaxy evolution explorer ) is a nasa small explorer , launched in april 2003 .
we gratefully acknowledge nasa s support for construction , operation , and science analysis for the galex mission , developed in cooperation with the centre national detudes spatiales of france and the korean ministry of science and technology .
this work was supported by grant no .
r01 - 2006 - 000 - 10716 - 0 from the basic research program of the korea science & engineering foundation to s.k.y .
the authors declare that they have no competing financial interests .
k.s . ,
sadegh khochfar , sugata kaviraj , s.k.y .
have performed the data sampling , analysis , interpretation , model construction , and writing of the manuscript .
a.b .
supplied the virgo galaxy data , and the rest of the authors contributed to the data acquisition using the galex satellite .
correspondence and requests for materials should be addressed to s.k.y .
( email : [email protected] ) . |
blazars are well known for their variability and power of polarized radiation across a wide range of the electromagnetic spectrum @xcite . in some cases
, the flux can vary on timescales as short as an hour or less ( see e.g. , * ? ? ?
* ; * ? ? ?
blazars exhibit a doubly - peaked spectral energy distribution ( sed ) , in which the low - energy component could extend from radio through uv or x - rays while the high - energy component extends from x - rays to @xmath0-rays .
the low - frequency component of the sed is almost certainly due to synchrotron emission from nonthermal , ultra - relativistic electrons .
the high - frequency component , on the other hand , is a result of inverse compton scattering of seed photons by the same ultra - relativistic electrons producing synchrotron emission ( in a leptonic jet scenario ) . in this case
, the seed photons could be the synchrotron photons produced within the jet ( synchrotron self compton , ssc ) @xcite , and/or external photons entering the jet from outside ( the ec process ) ( e.g. , * ? ? ? * and references therein ) .
the spectral variability patterns and seds are key ingredients in determining the acceleration of particles and the time - dependent interplay of various radiation mechanisms responsible for the observed emission .
another defining characteristic of blazars is the high degree of linear polarization at optical wavelengths .
many bright @xmath0-ray blazars that are in the _ fermi - lat _
bright @xmath0-ray source list @xcite have shown spectral and linear polarization variability @xcite .
linear polarization at mm , ir , and optical wavelengths tends to exhibit similar position angles and sometimes correlation across these wavebands , often with some time delay @xcite .
such correspondence between the variation in polarization and flux across a wide range of the electromagnetic spectrum , combined with vlbi imaging , can be used to identify the location of variable emission at all wavebands and shed light on the physical processes responsible for the variability @xcite .
the blazar 3c 279 , located at a redshift of 0.538 @xcite , is one of the most prominent and well - studied blazars .
this is due to its highly variable nature ( change in magnitude @xmath1 at optical bands ) at all wavelengths and high optical polarization up to 45.5% observed in the u band @xcite .
intensive multiwavelength campaigns ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and theoretical efforts ( e.g. , * ? ? ?
* ; * ? ? ?
* ) have led to some important conclusions about the physical properties of 3c 279 .
chatterjee et al . ( 2008 ) showed that the flux variability in 3c 279 has been found to be significantly correlated at x - rays , optical r band , and 14.5 ghz wave bands , which also suggests that nearly all x - rays are produced in the jet .
the x - ray flux has also been associated with superluminal knots , as suggested by correlation with the flux of the core region in the 43 ghz vlba images @xcite .
nevertheless , the nature and origin of its high - energy emission and the relationship of its behavior to the physical aspects of the jet remain elusive @xcite .
in addition , the correspondence between high- and low - energy emission is also not very well understood . here
, we aim to understand the physical state of 3c 279 at different flux levels and look for a correspondence between the flux and polarization variation observed during the flaring state of november 2008 and quiescent state of may 2010 .
this can be achieved by carrying out a detailed spectral analyses of 3c 279 under the internal shock scenario , and understanding the role of intrinsic parameters and interplay of synchrotron and inverse compton radiation processes in shaping the corresponding spectra of the two states . as a first step toward comparing the physical state of 3c 279 at different flux levels , we use the multi - slice time - dependent leptonic jet model of joshi & bttcher ( 2011 ) ( in the framework of internal shock scenario ) with radiation feedback to simulate the sed of 3c 279 corresponding to the optical high state of early 2006 .
the broadband emission of 3c 279 for this state indicates suppressed external compton emission , which makes it an ideal candidate for simulation using a synchrotron - ssc model .
we use physical jet parameters obtained from the vlba monitoring to guide our modeling efforts and discuss the role of various intrinsic parameters and radiation processes in producing the resultant sed .
we briefly describe the model of joshi & bttcher ( 2011 ) in [ model ] .
we discuss our findings about the connection between the flux and polarization variation observed in 3c 279 during flaring and quiescent periods of november 2008 and may 2010 , respectively , in [ 3c2 ] .
we discuss our first results from this study in [ results ] .
we summarize and give a brief description of future work in [ summary ] .
the mode of acceleration of particles to highly relativistic energies and its location in the jet is not yet completely understood . a colliding shell or
an internal shock model offers a way to gain insight into the physics of particle acceleration . under this model
, the central engine ( black hole + accretion disk ) is assumed to spew out shells of plasma with different velocity , mass , and energy . an inner shell , ejected at a later time , and travelling with a higher speed catches up to a slower moving outer shell that is ejected at an earlier time . the subsequent collision results in an emission region , as shown in figure [ emission_region ] , with two internal shocks ( reverse ( rs ) and forward ( fs ) ) separated by a contact discontinuity ( cd ) and traveling in opposite directions to each other in the frame of the shocked fluid .
as the shocks propagate through the emission region , they convert the ordered bulk kinetic energy of the plasma into the magnetic field energy and random kinetic energy of the particles .
the highly accelerated particles then radiate and produce the emission observed from the jet .
the treatment of shell collision and shock propagation is hydrodynamic and relativistic in nature @xcite . &
@xmath2 , respectively , such that @xmath3 .
the primed quantities refer to the comoving frame and unprimed refer to the lab ( agn ) frame . the comoving pressures @xmath4 and @xmath5 of the shocked fluids across the cd
are considered equal . @xmath6 and @xmath7 are the widths of the inner and outer shells , respectively , after the collision .
the quantities have been obtained by using hydro - dynamic and relativistic jump conditions across the shocks and cd @xcite.,width=283 ] [ emission_region ] the model follows the evolution of the electron and photon populations inside the emission region in a time - dependent manner .
the equations governing this evolution are , @xmath8 + q_{e } ( \gamma , t ) - \nonumber\\ \frac{n_{e } ( \gamma , t)}{t_{e , esc}}~\end{aligned}\ ] ] and @xmath9 here , @xmath10 is the particle ( referred to as electrons in the rest of the text ) energy loss rate due to synchrotron and inverse compton losses .
the quantity @xmath11 is the sum of external injection of electrons and intrinsic @xmath12 pair production rate , and @xmath13 is the electron escape time scale .
the quantities @xmath14 and @xmath15 are the subsequent photon emission and absorption rates , and @xmath16 is the photon escape timescale ( see [ scheme ] ) .
the radiative energy loss rates and photon emissivities are calculated using the time - dependent radiation transfer code of joshi & bttcher ( 2011 ) .
the model reproduces the nir to @xmath0-ray emission from blazars as it follows the evolution of the emission region out to sub - pc scales only .
thus , it simulates the early phase of @xmath0-ray production . during this time
, the radiative cooling is strongly dominant over adiabatic cooling and the emission region is highly optically thick at ghz radio frequencies .
hence , the simulated radio flux is well below that of the actual radio data .
we calculate the resultant spectrum from the collision , for a cylindrical emission region , in a time - dependent manner .
we consider the inhomogeneity in the photon and electron energy density throughout the emission region by dividing the region into multiple zones and providing a fraction of the photon flux from each zone to its adjacent zones in the two directions , as shown in figure [ zone_dia ] .
values represent the photon densities in the forward and backward direction , respectively , in the comoving frame of the emission region .
the rest of the unprimed quantities refer to the lab frame.,width=302 ] [ zone_dia ] the fraction of photon flux from a zone or , in other words , the photon escape rate of a zone in a particular direction is calculated using the photon density of that zone and the probability of escape , p , for a photon from that zone in that direction ( forward , backward , or sideways ) , according to the equation @xmath17 where @xmath18 is the volume - and - angle - averaged photon escape timescale for a cylindrical region that is calculated semi - analytically in the model .
we use this framework to calculate the radiation transfer within each zone and in between the zones @xcite .
analytical studies of the seds of 3c 279 , collected during the lifetime of cgro , indicate that the ssc mechanism is responsible for producing x - rays @xcite .
intensive multiwavelength campaigns ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and theoretical efforts ( see e.g. * ? ? ?
* ; * ? ? ?
* ) have provided insights on the physical properties of 3c 279 .
but many aspects of the radiation mechanism responsible for high - energy emission still remain unclear .
also , no consistent trends of cross - correlations have been found between optical , x - ray , and @xmath0-rays @xcite although the source , in early 2009 , exhibited coincidence of a @xmath0-ray flare with that in the optical and a dramatic change of optical polarization angle during that time @xcite .
such behavior poses a serious challenge to the single - zone leptonic jet model approach where the electron population is assumed to be homogeneous and is responsible for producing synchrotron emission as well as high - energy emission through inverse compton scattering @xcite . in order to improve our insight into the behavior of 3c 279
, we aim to understand the evolution of the source at different flux levels by simulating the corresponding seds , using a multi - zone leptonic jet model , and looking for connections between the evolution of spectral states and the behavior of the corresponding polarization at optical and radio wavelengths .
figure [ 3c279lcs ] shows the longterm @xmath0- and x - ray lightcurves ( lcs ) of 3c 279 from august 2008 till april 2011 . from these lcs ,
we have extracted two 1-month time periods : f1 ( 11/08/2008 - 12/08/2008 ) and q1 ( 05/22/2010 - 06/26/2010 ) , corresponding to a @xmath0-ray flaring and a quiescent period , as shown in figures [ f1lcs ] & [ q1lcs ] .
- and x - ray energy bands .
the top panel corresponds to the data compiled from rxte while the bottom panel refers to the fermi data .
long - term variability can be seen in the x - rays & @xmath0-rays.,width=283 ] [ 3c279lcs ] as can be seen from fig .
[ 3c279lcs ] , long - term variability exists both in the x - rays & @xmath0-rays . on the other hand , in case of f1 ( see fig .
[ f1lcs ] , more complicated features exists for such shorter time - periods that need to be considered in detail . while no correlation is apparent between the flux and polarization behavior at optical and radio wavelengths for period f1 , the vlba data on 06/14/2010 ( period q1 ) resulted in a degree of polarization @xmath19 and a radio polarization angle @xmath20 , which agrees very well with the corresponding values of _ p _
( @xmath21 ) and _ evpa _ ( @xmath22 ) at optical r band .
this suggests a common source of origin for the two .
[ f1lcs ] [ q1lcs ] the seds of 3c 279 corresponding to states f1 and q1 are shown in fig .
[ vlbaim ] .
the data have been compiled from observations with fermi , swift , vlba , and various ground - based optical and radio telescopes .
the optical and near - ir data were dereddened using the galactic extinction coefficients of schlegel et al .
( 1998 ) and the magnitudes were converted to fluxes using the zero - point normalizations of bessel et al .
the shape of the optical and near - ir spectra indicates that the synchrotron component for both states peaks in the near - ir regime with the @xmath23 value of f1 higher than that of q1 by almost an order of magnitude .
the x - ray flux of f1 remains in a low state coinciding with that of state q1 .
the shape of the optical spectrum suggests that the synchrotron component extends into the uv regime with a turnover to the high - energy component taking place at @xmath24 hz .
the @xmath0-ray spectrum for both states indicates a higher total energy output in the high - energy component compared to the low - energy one , which is typical for the class of flat spectrum radio quasars @xcite .
[ vlbaim ] the blazar 3c 279 was observed in an extensive multi - frequency campaign in january 2006 @xcite .
the object went into an optical flaring state on 01/15/2006 .
the interesting feature of this state is that the x - ray flux remained in a low state during that period , while coinciding with the historical x - ray low - state of june 2003 @xcite .
almost a month later , in februrary 2006 , it was detected by major atmospheric gamma - ray imaging cherenkov telescope ( magic ) in a very high energy ( vhe ) @xmath0-ray state @xcite .
as a first step toward simulating the seds of 3c 279 in order to understand the evolution of spectral states and compare the flaring states of the source in november 2008 and january 2006 , we have used the time - dependent leptonic jet model of joshi & bttcher ( 2011 ) that calculates the synchrotron and ssc radiation to reproduce the observed sed of 3c 279 corresponding to the optical high state of 01/15/2006 .
figures [ 3c279inst ] & [ 3c279int ] show the instantaneous and time - integrated simulated seds of 3c 279 , respectively , for this day .
the model independent parameters @xcite estimated using the sed , vlba observations @xcite , and variability on 1-day timescale were used to develop an initial set of input parameters : @xmath25 here , @xmath26 is the doppler boosting factor .
the symbol r is the radius of the emission region , b is the comoving magnetic field of the region , and @xmath27 is the equipartition parameter between the magnetic field and the electron energy density , assumed to be equal to 1 here .
the quantities @xmath28 & @xmath29 refer to the low and high energy cutoffs of the electron energy distribution . the spectral index of the electron population , at the time of injection ,
is given by _
q _ and _ p _ is the equilibrium spectral index obtained using the optical synchrotron spectrum , which is given by @xmath30 for strongly cooled electrons . the quantity , @xmath31 is the observing angle inferred from vlba observations .
the entire emission region has been divided into 100 zones ( 50 in the forward and 50 in the reverse emission region ) to analyze the observed sed of the source .
the initial set of parameters was modified to reproduce the state of 3c 279 as observed on 01/15/2006 .
various simulations were carried out by varying input parameters one at a time to obtain a reasonable fit to the observed sed of 3c 279 .
a fit is considered successful if it passes through the observed data points , corresponding to 01/15/2006 , without over - producing the x - ray flux .
an observer s line of sight angle , @xmath32 , and a cross - sectional radius , @xmath33 cm , for the cylindrical emission region , have been chosen to obtain the current best fit .
table [ paramlist ] lists the rest of the parameters used for obtaining the current fit of the observed sed .
these parameters result in @xmath34 for the entire emission region , @xmath35 and @xmath36 for both the forward and reverse emission regions , and @xmath37 & @xmath38 for the forward and reverse emission regions , respectively . as can be seen from figures [ 3c279inst ] & [ 3c279int ] , the observed sed for 01/15/2006 shows a high - energy bump that is indicative of a dominant ssc component and a suppressed ec component .
the time - integrated simulated sed of the successful model passes very close to the ir and optical data points , indicating that the synchrotron component is responsible for the lower energy bump of the sed .
the spectral upturn takes place in the soft x - rays at @xmath39 0.1 kev due to the presence of the ssc component in the simulation .
the lower - energy part of the ssc component reproduces the x - ray data quite well , suggesting the dominance of the ssc component in producing this part of the high - energy bump . as can be seen in figure [ 3c279int ]
, there is a break in the low - energy part of the x - ray spectrum of january 2006 at @xmath40 hz , which could be due to the presence of a low - energy cutoff of the electron distribution at ultrarelativistic energies @xcite .
[ 3c279inst ] [ 3c279int ] .model parameters used to reproduce the state of 3c 279 as observed on 01/15/2006 .
+ [ cols="<,^,^,^,^,^,^,^",options="header " , ] + @xmath41 : luminosity of the injected electron population in the blob , @xmath42 : blfs of the inner and outer shells before collision , q : particle spectral index , @xmath43 : ratio of electron and shock energy density , @xmath44 : ratio of magnetic field and shock energy density , @xmath45 : fraction of accelerated electrons [ paramlist ]
we have presented first results of our study on the behavior of 3c 279 at different spectral states , and the connection between the flaring and quiescent states and the behavior of its polarization at optical and radio wavelengths . in order to carry this out ,
we have extracted a @xmath0-ray flaring ( f1 ) and a quiescent ( q1 ) state from the long - term lcs of 3c 279 shown in fig .
[ 3c279lcs ] . in case of f1 ,
a clear cross - correlation can not be seen at optical frequencies over a shorter time - period such as a month ( see fig . [ f1lcs ] ) and some complicated features exist that need to be considered in detail . in case of q1 , there is an anti - correlation between @xmath0- and x - ray , with x - rays mildly flaring during the @xmath0-ray quiescent state , but the r - band flux shows a strong relationship to the optical polarization ( see fig .
[ q1lcs ] ) . on the other hand , while no correlation can be seen between the flux and polarization behavior at optical and radio wavelengths for the period f1 , the vlba data during the q1 period agrees very well with the corresponding values of _ p _ and _ evpa _ at optical r band .
this suggests a common source of origin for the two during this state . in order to understand the behavior of the source at different spectral states , it is necessary to simulate the corresponding seds of the source . to that end , we have simulated the optical high state of 3c 279 observed in january 2006 using the time - dependent multi - zone leptonic jet model of @xcite .
the model , in its current form , calculates radiation resulting from synchrotron and ssc processes .
the spectral state of 3c 279 observed in january 2006 makes a good candidate for this model , as the @xmath0-ray emission is suppressed in this case and the x - ray emission is generally associated with ssc scattering @xcite . our current model of the sed results in a high value of @xmath46 compared with that of collmar et al .
this is reasonable , as the internal shock model is expected to have low acceleration efficiency , and thus would need a higher kinetic luminosity to achieve the desired acceleration .
the current fit results in a value of magnetic field that agrees reasonably well with the model independent parameter estimate and that of collmar et al .
the value of electron energy index , _ q _ , used to obtain the current fit is almost equal to the model independent parameter _
p _ and is comparatively higher than that of collmar et al .
this indicates that the synchrtron spectrum in the current fit does not come from a cooled population of electrons .
the value of _ q _ needs further adjustment in order to obtain a satisfactory fit . with a successful fit ,
we hope to gain insights on the acceleration efficiency , the relative speeds with which the two shocks are moving away from each other ( in the frame of the shocked fluid ) , and the mass of each plasma shell needed to reproduce an optical high state of 3c 279 such as that of 01/15/2006 . as part of our future work , the external compton component due to photons entering the jet from the blr and dusty torus
is being incorporated in the existing model of @xcite to reproduce more fully the seds of blazars , especially for flat spectrum radio quasars ( joshi et al .
2012 , in prep . ) , and to analyze a wider variety of blazar spectral states .
we plan to use the extended version of the model to simulate the seds of 3c 279 for a typical period of ~ 10 days corresponding to f1 and q1 and compare the results with that of the january 2006 optical high state of 3c 279 .
further , we plan to incorporate the effects of magnetic field orientation , as inferred from polarization monitoring programs , on the resultant spectral variability and seds of blazars .
this would aid us in the study of intrinsic parameter differences between various blazar subclasses that arise from the orientation of the magnetic field in the jet . | we have observed 3c 279 in a @xmath0-ray flaring state in november 2008 .
we construct quasi - simultaneous spectral energy distributions ( seds ) of the source for the flaring period of 2008 and during a quiescent period in may 2010 .
data have been compiled from observations with fermi , swift , rxte , the vlba , and various ground - based optical and radio telescopes .
the objective is to comprehend the correspondence between the flux and polarization variations observed during these two time periods by carrying out a detailed spectral analyses of 3c 279 in the internal shock scenario , and gain insights into the role of intrinsic parameters and interplay of synchrotron and inverse compton radiation processes responsible for the two states . as a first step ,
we have used a multi - slice time - dependent leptonic jet model , in the framework of the internal shock scenario , with radiation feedback to simulate the sed of 3c 279 observed in an optical high state in early 2006 . we have used physical jet parameters obtained from the vlba monitoring to guide our modeling efforts .
we briefly discuss the effects of intrinsic parameters and various radiation processes in producing the resultant sed . |
although observations of acoustic peaks in the cosmic microwave background ( cmb ) fluctuation spectra are sensitive to the curvature of spacetime , and appear to require some form of cosmological dark energy , they may be less useful for discriminating among different equations of state for the dark energy .
various studies ( e.g. ) have concluded that the signatures of different types of spatially smooth , evolving dark energy on the cmb spectra are relatively undiscriminating , in part because cmb fluctuations were formed at high redshift , when the dark energy was not a prominent constituent , although spatial fluctuations in the dark energy could lift the degeneracy considerably @xcite . by contrast
, observations of sources at small to moderate redshifts ( @xmath0 ) would probe epochs where the dark energy is prominent , leading to suggestions that observations of type ia supernovae or of galaxy counts could be used to determine the nature of the dark energy most effectively . however , the efficacy of these observational programs is controversial , as there is considerable degeneracy among the predictions of different dark energy models for , for example , the luminosity distance as a function of redshift@xcite .
one approach to the analysis of low redshift data would be to presume particular classes of models , perhaps parametrized by a series expansion of @xmath2 , and then attempt to constrain the model parameters by a likelihood or bayesian method ( e.g. ) .
a second approach is to fit the data by a parametric representation of , for example , the luminosity distance as a function of redshift , and then analyze the results to constrain the properties of the dark energy indirectly ( e.g. ) . although the first approach is preferable ( because it generally avoids difficulties associated with differentiating data , and allows a simpler assessment of uncertainties in derived parameters , in addition to outlining its implicit assumptions more clearly ) , the second approach is more useful for understanding whether or not such observations can ever yield a unique solution for the dark energy equation of state . in this short note ,
the question considered is : suppose one could analyze data from @xmath3 to determine the expansion rate of the universe as a function of scale factor , @xmath4 , _
exactly_. could one then determine the equation of state of the dark energy exactly from these measurements alone ?
the answer , as we shall see , is _ no _ : a _ separate _ determination of the fraction of the universe in the form of nonrelativistic particles is needed @xcite .
the necessity of separately determining the density of nonrelativistic particles in order to deduce the equation of state of the dark energy can be demonstrated most easily by constructing a specific example .
suppose the data show that the universe is expanding according to the simple relationship h^2=h_0 ^ 2(_1+_2a^3 ) , [ expand ] where @xmath5 .
one ( tempting ) interpretation of this result would be that the universe consists of two components , nonrelativistic matter contributing a fraction @xmath6 of the closure density , and a cosmological constant contributing a fraction @xmath7 . however , this interpretation is not unique : there are models involving a scalar field @xmath8 with nonconstant effective potentials @xmath9 that can lead to eq .
( [ expand ] ) . to see this
, we can construct an explicit model with both a scalar field and nonrelativistic matter , so that h^2=8g_3+h_0 ^ 2_ma^3 with @xmath10 in general .
the scalar field energy density is therefore _ = _
0(_1+_2-_ma^3 ) , where @xmath11 . differentiate @xmath12 with respect to @xmath13 to find = -3_0(_2-_m)a^4 = -3(_+p_)a [ conserve ] where the second equality follows from conservation of energy for the scalar field .
thus , we find simply that @xmath14 , and ^2&=&_+p_=_0(_2-_m)a^3 + v()&=&_-p_2=_0(_1 + ( _ 2-_m)2a^3 ) . [ phidentify ] note that the first of eqs .
( [ phidentify ] ) requires that @xmath15 .
we can determine @xmath9 by solving the first of eqs .
( [ phidentify ] ) for @xmath16 , inverting to find @xmath17 , and substituting into the second of eqs .
( [ phidentify ] ) . combining eqs .
( [ expand ] ) and the first of eqs .
( [ phidentify ] ) implies a^2(dda)^2=^2h^2 = 3(_2-_m)8g_2(1+_1a^3/_2 ) .
define x=_1a^3_2_g^2=_2-_m24g_2 ; [ xdef ] then if @xmath18 x^2(ddx)^2=11+x , which has the general solution = _ 2 ^ -1(1 ) , where @xmath19 is a constant of integration .
the sign in this solution can be chosen arbitrarily ; chosing the @xmath20 sign gives a solution in which @xmath21 increases with increasing @xmath13 . if we let @xmath22 today , then we find = _ 0 + 2 ^ -1 -2 ^ -1 , [ psisol ] which we may invert to find = ^2 , where @xmath23 .
the second of eqs .
( [ phidentify ] ) then determines the potential to be @xcite v()=_0\{_1 + 12(_2-_m ) ^2 } .
[ conspirepot ] note that @xmath24 and @xmath25 asymptotically . also , when @xmath26 , the potential becomes exponential : @xmath27 .
the key question here revolves around prior assumptions .
the `` simplest '' interpretation of eq .
( [ expand ] ) is that the energy density of the universe is the sum of contributions due to nonrelativistic matter and a cosmological constant .
however natural this interpetation may seem to be , it only follows if one assumes _ either _ that the dark energy is in the form of a cosmological constant , or that the mass density in nonrelativistic form has closure parameter @xmath6 .
if one is somewhat more agnostic on either of these points , then a range of possibile interpretations accounts for the data equally well .
the degeneracy is not just a property of models for which @xmath28 is given by eq .
( [ expand ] ) .
if the data were to favor a model of the form h^2(a)=h_0 ^ 2 , with @xmath29 some function , and @xmath30 , then without additional prior assumptions the most one can say for sure is that @xmath31 $ ] , which leaves open the possibility of a range of different solutions for @xmath9 .
this model illustrates that if one makes no prior assumptions about the form of the quintessence potential @xmath9 , then it is impossible to fit low redshift data alone to find @xmath9 uniquely unless @xmath1 is known independently in this example , a range of models , parametrized by the value of @xmath32 , will fit equally well .
constraints on the value of @xmath33 limit the range of acceptable models .
although minimally @xmath34 , we know that this model would be unable to account for the formation of large scale structure if @xmath33 is too small . if we assume that only the nonrelativistic matter can clump ( i.e. that the dark energy and relativistic matter remain smooth ) , then , in linear theory , the density contrast @xmath35 obeys the equation ( e.g. @xcite ) + ( 3+dhda ) 1addda-4g_mh^2a^2d=0 ; [ linear ] eq . ( [ linear ] ) has both growing and shrinking modes , @xmath36 and @xmath37 , respectively .
if we include radiation ( which is a known component of the universe , and subdominant today , but important in determining linear theory growth rates ) , then eq .
( [ expand ] ) must be altered to h^2(a)=h_0 ^ 2(_1+_2a^3 + _
rada^4 ) , [ hrad ] where @xmath38 and , for three low mass or massless neutrinos , @xmath39 if @xmath40 . the growing mode mainly amplifies perturbations between @xmath41 and @xmath42 . during this phase , @xmath43 , and @xmath44 , where _ + ( f)=14(-1 ) .
the overall linear theory growth factor is @xmath45 , so the ratio of the growth factor for @xmath46 to its value for @xmath47 is approximately ( a_1a_eq)^14(-5 ) ( a_1a_eq)^-3(1-f)/5 , where the approximation assumes @xmath48 .
thus , we estimate that for the growth factor to be within a factor of two of its value for @xmath47 , we must have @xmath49 .
the solid line in fig .
1 shows the ratio of the growth factor for @xmath50 to its value for @xmath51 , that is r_/ as a function of @xmath1 for @xmath52 , where @xmath53 is the linear theory growth factor evaluated at scale factor @xmath13 , given @xmath1 and @xmath6 .
these results were found by solving eq .
( [ linear ] ) numerically from @xmath54 to the present day , assuming identical initial perturbation amplitudes .
( relativistic particles were included as a smooth component via eq .
[ [ hrad ] ] . )
the numerical results show that for @xmath55 , then we must require @xmath56 , corresponding to @xmath57 , approximately the same as the analytic estimate of the previous paragraph .
( the estimate is accurate to about 20% for @xmath58 . )
linear theory also predicts velocity fluctuations proportional to the peculiar velocity factor @xmath59 . the dashed line in fig .
1 shows r_v = t_+(1;_m,_2)t_+(1;_2,_2 ) , the ratio of the peculiar velocity factor at @xmath60 for @xmath50 to its value for @xmath61 .
although there is some dependence of @xmath62 on @xmath1 , it is not as extreme as for @xmath63 .
thus , primarily from the requirement that linear density perturbations can grow substantially from small values in the early universe , we know that if eq .
( [ expand ] ) were truly exact , @xmath33 must be close to one . imposing limits on the range of plausible @xmath1 from independent , physical and phenomenological arguments
amounts to using prior information to restrict the possibilities to be compared with the data from observations at @xmath0 .
( the use of prior information from large scale structure and cmb anisotropies has been discussed in e.g. . )
this does not eliminate the degeneracy implicit in eq .
( [ expand ] ) altogether , but does diminish its significance by constraining the range of acceptable values of @xmath33 severely .
however , the fact remains that if one were to try to determine the equation of state of the dark energy from observations that , hypothetically , yield eq .
( [ expand ] ) , then one can never hope to find a unique result from an analysis that ignores constraints from the development of large scale structure ( or any other considerations that yield independent information about @xmath1 , such as peculiar velocities , cmb fluctuations and weak lensing ) .
moreover , even if such restrictions are imposed , unless @xmath1 is determined _ precisely _ by them , some degeneracy remains in the construction of @xmath9 from observations at @xmath64 .
if eq . ( [ expand ] ) with @xmath6 substantially different from @xmath1 is an excellent approximation at low to moderate redshifts , it can _ never _ be exact . in this case
, there must be a transition in @xmath28 , so that the component of the density that scales like @xmath65 is predominantly due to nonrelativistic particles for a period sufficient to grow large scale structure .
the implicit deviations from eq .
( [ expand ] ) would contain information on @xmath1 as well as @xmath6 , and probably additional parameters as well .
in this situation , it might be possible to ascertain the equation of state of the dark energy from measurements at low to moderate redshifts alone , provided that the imprint of @xmath1 and the additional parameters in @xmath28 can be discerned from these observations .
the question is , if eq .
( [ expand ] ) is an accurate ( but inexact ) fit at @xmath0 , and @xmath6 differs substantially from @xmath1 , how well could we determine @xmath1 , and hence break the degeneracy , observationally ? to make these issues more concrete ,
suppose in reality eq .
( [ expand ] ) is actually the large @xmath13 limit of a more general , exact relationship , say h^2(a)=h_>^2(a)f(aa_t)+h_<^2(a ) , [ newexpand ] where @xmath66 as @xmath67 and @xmath68 as @xmath69 , with @xmath70 a transition value of the scale factor , @xmath71 given by eq .
( [ expand ] ) , and , for example , h^2_<(a)=h_0 ^ 2(_<+_ma^3 ) ; since the model is flat , 1=(_1+_2)f_0 + ( _ < + _
m)(1-f_0 ) , [ flatcond ] where @xmath72 .
such a model could be consistent with the growth of large scale structure , but might still be hard to distinguish from eq .
( [ expand ] ) based on observations at low to moderate redshift , depending on the value of @xmath73 , and the form of @xmath74 .
( [ newexpand ] ) , although admittedly _ ad hoc _ , is useful for studying the extent to which observations of various sorts could discern that @xmath75 , and pin down the difference sufficiently to break the degeneracy discussed above . for a flat cosmology , eq .
( [ newexpand ] ) implies that the radial coordinate for a source at @xmath76 is r_s(a;_m,_2,f)&=&(1)-(a)= _ a^1dbb^2h(b ) + & = & h_0 ^ -1_a^1 db [ sourcecoord ] where @xmath77 , and @xmath78 .
( here , the notation @xmath79 means the radial coordinate as a function of @xmath13 given @xmath1 and @xmath15 , as well as a set of parameters required to define @xmath74 . )
if @xmath74 does not vary much over the range of @xmath13 covered by the observations , then the data would only determine @xmath80 accurately , and would yield little useful information on @xmath6 or @xmath1 : only the combination @xmath81 could be deduced .
this situation is possible provided that @xmath73 is fairly small , below the range covered by observations , and @xmath74 varies relatively slowly as a function of @xmath82 for large values of @xmath82 . in this case
, we can not expect to eliminate the degeneracy among models directly from observations of low to moderate redshift sources alone .
constraints from other sorts of observations that probe @xmath1 directly would be needed to deduce the quintessence model as accurately as possible .
more quantitative statements only can be made in terms of specific @xmath74 , and require some exploration of parameters .
let us examine one choice : @xmath83 , in which case eq .
( [ flatcond ] ) is satisfied exactly , and eq .
( [ sourcecoord ] ) becomes r_s(a;_2,_m , f)=h_0 ^ -1_a^1db .
[ simplerad ] = 5.5 in eq .
( [ simplerad ] ) may be used to compute magnitude differences among models with given @xmath6 , @xmath1 and @xmath74 ; for illustrative purposes , let us choose @xmath84 .
2 depicts the magnitude difference m(z;_2,_m , a_t)=5.0_10relative to the model with @xmath85 as a function of @xmath86 assuming @xmath87 ( solid ) and @xmath88 ( dashed ) for ( from top to bottom in each case ) @xmath89 and @xmath90 .
not surprisingly , the magnitude differences are smaller for the larger value of @xmath73 , and decrease as @xmath91 .
nevertheless , the _ largest _ magnitude difference shown for @xmath92 is 0.29 mag , for @xmath93 ; for @xmath94 , the difference is 0.077 mag .
= 5.5 in fig .
3 shows results for ( _ m;_2,a_t)=-1 , [ rsmodel ] the fractional difference between the angular size of the ( sonic ) horizon at @xmath95 as a function of @xmath96 and its value for @xmath61 for the two different values of @xmath73 .
( deviations of the sound speed from @xmath97 have not been included . )
larger fractional deviations arise for smaller @xmath1 and @xmath73 .
the largest deviation shown in the two figures is 12.7% for @xmath98 ; the largest fractional deviation for @xmath88 is about 1.3% , and occurs near @xmath99 . although this is only one example , based on a particular ( _ ad hoc _ ) choice of @xmath74 , it illustrates that while data at both low to moderate @xmath86 as well as cmb observations can yield information that lifts the degeneracy between @xmath6 and @xmath1 , considerable precision may be needed to determine the equation of state of the dark energy accurately .
note that no attempt has been made to fit simulated data based on eq .
( [ rsmodel ] ) to figure out how well @xmath4 could be determined in practice ; the comparisons made in figs .
2 and 3 are based on exact functions .
the observational challenge of distinguishing among models at these levels could be even greater than these relatively small differences would indicate . on the other hand , figs .
2 and 3 were constructed based on a particular ( _ ad hoc _ ) model for @xmath28 designed for a phenomenological study of how easily information on @xmath1 could be gleaned from luminosity distances and cmb acoustic peaks .
the task could be easier or harder in cosmologies based on particular physical theories for the dark energy , depending on the details of the model .
to conclude , the point of this paper is _ not _ to suggest that it is absolutely inconceivable that the equation of state of the dark energy could be measured , but rather to show , via a specific example , how measurements at @xmath100 admit a continuum of interpetations in terms of evolving dark energy fields _ unless _ @xmath1 can be determined separately somehow .
the specific example , eq .
( [ expand ] ) , can be intepreted `` naturally '' in terms of a universe containing a mixture of nonrelativistic particle dark matter plus a cosmological constant , but can be interpreted equally well in terms of a quintessence model with a range of potentials , eq .
( [ conspirepot ] ) .
this specific example sheds some light on why simulated analyses of low to medium redshift data appear to allow degenerate interpretations : it is always possible that the component of the total energy density of the universe that evolves like @xmath65 is only partially due to nonrelativistic particles , with the rest arising from quintessence .
unique interpretations can only be obtained in terms of specific models for the quintessence that forbid such a conspiracy _ a priori _ , by either specifying the form of @xmath9 or fixing the value of @xmath1 .
realistically , the extent to which the @xmath101 constituent must be due to nonrelativistic particles is constrained by the requirement that large scale structure formation evolve `` normally , '' that is , unimpeded by the existence of a component with density proportional to @xmath65 that is incapable of clustering . if eq .
( [ expand ] ) were truly exact , then @xmath6 would have to be close to @xmath1 for large scale structure to grow , but even in this case , a limited degeneracy remains in the determination of @xmath9 .
if @xmath6 and @xmath1 differ substantially , then eq . ( [ expand ] ) can not be exact , raising the question of how well one could discern the two parameters separately from observations . to examine this issue , a modified ( phenomenological ) expansion law , eq .
( [ newexpand ] ) , that encodes information on both @xmath6 and @xmath1 was introduced . in the context of a particular model that is consistent with large scale structure formation , but reduces to eq .
( [ expand ] ) with @xmath15 at large enough @xmath13 , we have seen that information about @xmath1 could be gleaned both from measurements of the positions of cmb peaks and from luminosity distance determinations .
however , the deviations among models with various @xmath50 can be quite small , which would still pose a substantial challenge for programs that aim to determine the equation of state of the dark energy . of course , in the context of
_ specific _ models for the quintessence field , embodied in particular forms for @xmath9 , the analyses may not encounter pronounced degeneracies . at present , though , there is little compelling reason to assume any particular @xmath9 _ a priori_. when analyzing data for a particular form of @xmath9 , one is primarily engaged in estimating the parameters of the model , as well as @xmath1 .
thus , one might be able to find the best fit model of a particular type ( e.g. @xmath102 ) , without being able to tell if the underlying model would be favored by the data if other possibilities were admitted .
if the allowed ranges of parameters for a particular quintessence model do not shrink with the accumulation of data , one can conclude that the model is inadequate with confidence . however , there is no guarantee that a given model is correct even if the data seem to converge on a unique set of parameters , as eqs .
( [ expand ] ) and ( [ conspirepot ] ) illustrate .
the need for precise independent knowledge of @xmath1 for `` reconstructing '' the dark energy equation of state has also been stressed in @xcite in the context of particular choices for comparison quintessence models .
quintessence models with double exponential potentials have been treated e.g. by sen , a. a. & sethi , s. 2001 , gr - qc/0111082 , without stressing their relationship to the implicit degeneracy problem posed by incomplete knowledge of @xmath1 . | using a specific model for the expansion rate of the universe as a function of scale factor , it is demonstrated that the equation of state of the dark energy can not be determined uniquely from observations at redshifts @xmath0 unless the fraction of the mass density of the universe in nonrelativistic particles , @xmath1 , somehow can be found independently . a phenomenological model is employed to discuss the utility of additional constraints from the formation of large scale structure and the positions of cmb peaks in breaking the degeneracy among models for the dark energy . epsf = msbm10 = msbm7 = msbm5 = = = # 1#1 5125 ! |
if dark matter is produced at a hadron collider , its likely signature will be missing transverse momentum . to help determine the underlying origin of the observed dark - matter , it is important to measure the masses of the new particle states .
mass determination is a key part to identifying the underlying theory which lies beyond the standard model .
newly discovered particles could be kaluza - klein ( kk ) states from extra - dimensions , supersymmetric partners of known states , technicolor hadrons , or something else that we have not anticipated .
models predict relationships between parameters : supersymmetry relates the couplings of current fermions to the couplings of new bosons and the supersymmetric particle masses reflect the origin of supersymmetry breaking ; masses of kk states tell us about the size of the extra dimensions . in general
, mass determination of new particle states is central to discerning what lies beyond the standard model .
if dark - matter particle states are discovered at the lhc , how will the masses of these new particles be measured ? determining the mass of the dark - matter particle ( or other states which decay to dark matter ) will be difficult because we do nt expect the dark - matter particles to leave tracks or have a displaced vertex , and because we do not know the rest frame of the initial parton collision . if one has confidence about the specific model and process responsible for the observed events , then the magnitude of the process s cross - section constrains the mass .
if one does not yet know the model , then model - independent techniques will be needed to determine properties of the newly discovered particles such as couplings , spin , phases , mixing angles , and masses .
proposed methods for finding the gluino phase @xcite or spin @xcite , rely on our ability to make model - independent kinematic mass measurements .
thus , model determination needs a suite of model - independent tools to provide initial constraints on the mass .
there has been much recent work in developing model - independent mass determination tools .
edges in invariant mass combinations provide information about mass differences or mass squared differences @xcite .
if the hadron collider accesses many new states , then we may be able to determine the masses by combining the relationships provided by many different invariant - mass edges of a cascade decay @xcite .
there is also a series of approaches called mass shell techniques ( mst ) where one uses an assumption about the topology and on - shell conditions to solve for the unknown masses .
one mst variant assumes a long symmetric cascade decay chain and counts which multiplets of masses have solution to the most events @xcite , another assumes the masses in two events must be equal @xcite , and another hybrid combines a mst with the information from the many edges in cascade decays @xcite .
variable is a simple example of such a hybrid technique .
] there has been extensive work using the so - called ` stransverse mass ' variable , @xmath6 , to determine the mass _ difference _ between parent particle s and a dark - matter candidate particle s mass given an assumed mass for the dark - matter candidate ( refs .
@xcitehave more than 45 citations ) .
if the final decay to the lightest state involves a three - body decay - body decay is a sufficient but not necessary condition .
two - body decays can also display kinks @xcite provided the decaying particles have sufficiently large transverse boosts . ] as shown in fig .
[ figeventtopology ] , then a ` kink ' in the @xmath2@xcite , will occur at a position which indicates the invisible particle mass , as described by cho , choi kim , park ( cckp ) @xcite and corroborated in refs .
the ` kink ' in @xmath2 can be quantified by the constrained mass variable @xmath0 that we introduced in a previous letter @xcite . even in models where new invisible particles are nearly massless ( like the gravitino studied in @xcite )
, one would rather not just assume the mass of the lightest supersymmetric particle ( lsp ) , which is needed as an input to the traditional @xmath6 analysis , without measuring it in some model independent way .
which approach turns out to be best is likely to depend on what scenario nature hands us , since the various techniques involve different assumptions . having different approaches
also offers the advantage of providing a system of redundant checks . in ref .
@xcite , we introduced the @xmath0 kinematic variable which gives an event - by - event lower bound on the dark - matter particle s absolute mass given the mass difference between the dark matter candidate and its parent . in this paper , we introduce a complementary variable @xmath1 which gives an event - by - event _ upper _ bound on the same absolute mass .
our study shows that this technique rivals other invisible - particle mass determination techniques in precision and accuracy . in this paper
, we provides a demonstration of the use of the variable @xmath0 and @xmath1 in lhc conditions .
the variables @xmath7 and @xmath8 give an event - by - event lower - bound and upper - bound respectively on the mass of @xmath9 assuming the topology in fig . [ figeventtopology ] and the mass difference @xmath10 . to get the mass difference , we use events where @xmath9 decays into @xmath11 and two visible states via a three - body decay in which we can easily determine the mass difference from the end point of the visible - states invariant - mass distribution , @xmath12 .
one might also conceive of a situation with @xmath0 supplementing an alternative technique that gives a tight constraint on the mass difference but may have multiple solutions or a weaker constraint on the mass scale @xcite@xcite . given this mass difference and enough statistics , @xmath0 s endpoint
gives the mass of @xmath9 .
however the main advantage of the @xmath0 method is that it does not rely simply on the position at the endpoint but it uses the additional information contained in events which lie far from the endpoint . as a result it gives
a mass determination using significantly fewer events and is less sensitive to energy resolution and other errors . to illustrate the method , in this paper we study in detail the performance of the @xmath0 constrained mass variable in a specific supersymmetric model .
we study events where each of the two branches have decay chains that end with a @xmath13 decaying to a @xmath14 and a pair of opposite - sign same - flavor(ossf ) leptons .
thus the final states of interest contain four isolated leptons ( made up of two ossf pairs ) and missing transverse momentum .
fig [ figeventtopology ] defines the four momentum of the particle states with @xmath15 , @xmath16 , and the ossf pairs forming the visible particles @xmath17 .
any decay products early in the decay chains of either branch are grouped into @xmath18 which we generically refer to as upstream transverse momentum ( utm ) .
nonzero @xmath18 could be the result of initial state radiation ( isr ) or decays of heavier particles further up the decay chain .
events with four leptons and missing transverse momentum have a very small standard - model background . to give a detailed illustration of the @xmath0 methods , we have chosen to analyze the benchmark point p1 from @xcite which corresponds to msugra with @xmath19 gev , @xmath20 gev , @xmath21 , @xmath22 , @xmath23 .
our susy particle spectrum was calculated with isajet @xcite version 7.63 .
we stress that the analysis technique employed applies generically to models involving decays to a massive particle state that leaves the detector unnoticed .
a powerful feature of the @xmath0 distribution is that , with some mild assumptions , the shape away from the endpoint can be entirely determined from the unknown mass scale and quantities that are measured .
the ideal shape fit against early data therefore provides an early mass estimate for the invisible particle .
this study is meant to be a guide on how to overcome difficulties in establishing and fitting the shape : difficulties from combinatoric issues , from differing energy resolutions for the leptons , hadrons , and missing transverse momentum , from backgrounds , and from large upstream transverse momentum ( utm ) .
as we shall discuss , utm actually provides surprising benefits .
the paper is structured as follows : in section [ secub ] , we review @xmath0 and introduce the new observation that , in addition to an event - by - event lower bound on @xmath24 , large recoil against utm enables one also to obtain an event - by - event _ upper _ bound on @xmath24 .
we call this quantity @xmath1
. section [ secmodeling ] describes the modeling and simulation employed .
section [ secshapefactors ] discusses the implications of several effects on the shape of the distribution including the @xmath25 ( in our case @xmath26 ) distribution , the utm distribution , the backgrounds , combinatorics , energy resolution , and missing transverse momentum cuts . in section [ secperformance ]
, we put these factors together and estimate the performance .
we conclude in section [ secconclusions ] with a discussion about the performance in comparison to previous work .
we will now review the definition of @xmath0 as providing an event - by - event lower bound on @xmath24 . in generalizing this framework
, we find a new result that one can also obtain an upper bound on the mass @xmath24 when the two parent particles @xmath9 recoil against some large upstream transverse momentum @xmath27 .
fig [ figeventtopology ] gives the relevant topology and the momentum assignments .
the visible particles @xmath28 and @xmath29 and invisible particle @xmath11 are labeled with with momentum @xmath30 and @xmath31 ( which we group into @xmath32 ) and @xmath33 , respectively @xmath34 and @xmath35 in the other branch .
we assume that the parent particle @xmath9 is the same in both branches so @xmath36 .
any earlier decay products of either branch are grouped into the upstream transverse momentum ( utm ) 4-vector momentum , @xmath18 .
we assume the two decay chains share a common end - state given in this diagram .
all previous decay products are grouped into the upstream transverse momentum , @xmath18.,width=288 ] in previous work with events of this topology , we @xcite showed how to find an event - by - event lower bound on the true mass of @xmath37 and @xmath24 .
we assume that the mass difference @xmath38 can be accurately measured from the invariant mass edges @xmath39 or @xmath40 . for each event ,
the variable @xmath0 is the minimum value of the mass of @xmath9 ( the second lightest state ) after minimizing over the unknown division of the missing transverse momentum @xmath41 between the two dark - matter particles @xmath11 : @xmath42 there are eight unknowns corresponding to the four momentum components of @xmath33 and @xmath35 and five equations of constraint . because the true @xmath33 and @xmath35 must be within the domain over which we are minimizing , the function is guaranteed to be less than or equal to @xmath24 .
one way of calculating @xmath0 for an event is to use @xmath43 @xcite , which provides a lower bound on the mass of @xmath9 for an assumed mass @xmath44 of @xmath11 .
the true mass of @xmath9 lies along the line @xmath45 where we use @xmath46 to denote the possible masses of @xmath9 and to distinguish it from the true mass of @xmath9 denoted with @xmath24 .
thus we can see that for @xmath44 to be compatible with an event , we must have @xmath47 . for a given event ,
if one assumes a mass @xmath44 for @xmath11 , and if the inequality @xmath48 is satisfied , then there is no contradiction , and the event is compatible with this value of @xmath44 .
if however , @xmath49 , then we have a contradiction , and the event excludes this value @xmath44 as a viable mass of @xmath11 . using this observation ,
@xmath0 can be found for each event by seeking an intersection between @xmath43 and @xmath50 @xcite .
the lower bound on @xmath24 is given by @xmath51 where @xmath52 is a zero of @xmath53 in the case @xmath54 , the extreme events analyzed in cckp @xcite demonstrate that @xmath55 will only have one positive zero or no positive zeros , and the slope at a zero will always be negative . for no positive zeros , the lower
bound is the trivial lower bound given by @xmath56 .
note that a lower bound on the value of @xmath24 corresponds to a lower bound on the value of @xmath37 .
the appendix in ref .
@xcite shows that at the zeros of @xmath57 which satisfy eq([eqgprime ] ) , the momenta satisfy eqns([eqc1]-[eqc4 ] ) .
if there is large upstream transverse momentum ( utm ) ( @xmath58 ) against which the system recoils , then we find a new result .
using the @xmath6 method to calculate @xmath0 gives one the immediate ability to see that @xmath24 can also have an upper bound when requiring eqns([eqc1]-[eqc4 ] ) .
this follows because for large utm the function @xmath57 may have two zeros has more than two zeros , but we have not encountered such cases in our simulations . ] which provides both an upper and a lower bound for @xmath24 from a single event .
we have also found regions of parameter space where @xmath55 has a single zero but @xmath59 corresponding to an upper bound on the true mass of @xmath37 ( and @xmath24 ) and only the trivial lower bound of @xmath60 .
we can can obtain some insight into the cases in which events with large utm provides upper bounds on the mass by studying a class of extreme event with two hard jets , @xmath61 and @xmath62 against which @xmath9 recoils ( @xmath63 ) .
we will describe this extreme event and solve for the regions of parameter space for which one can analytically see the intersection points giving a lower bound and/or an upper bound .
the event is extremal in that @xmath43 , which gives a lower bound on @xmath24 , actually gives the true value of @xmath24 when one selects @xmath44 equal to the true mass @xmath37 .
the ideal event we consider is where a heavier state @xmath64 is pair - produced on shell at threshold . for simplicity
we assume the lab - frame is the collision frame .
assume that the @xmath64s , initially at rest , decay into visible massless jets @xmath61 , @xmath62 and the two @xmath9 states with the decay product momenta @xmath65 and @xmath66 .
both jets have their momenta in the same transverse plane along the negative @xmath67-axis , and both @xmath9 s momentum are directed along the @xmath67-axis . finally , in the rest frame of the two @xmath9s , both decay such that the decay products visible states have their momentum @xmath68 and @xmath69 along the @xmath67-axis and both invisible massive states @xmath11 have their two momenta along the negative @xmath67-axis .
in the lab frame , the four - vectors are given by @xmath70 for the event given by eqns([eqevent1a]-[eqevent1c ] ) , we can exactly calculate @xmath43 : @xmath71 this is found by calculating the transverse mass for each branch while assuming @xmath44 to be the mass of @xmath11 .
the value of @xmath72 is chosen so that the transverse masses of the two branches are equal .
substituting this value back into the transverse mass of either branch gives @xmath43 .
fig [ figextrememt2 ] shows @xmath57 , given in eq([eqgchi ] ) , for several choices of @xmath73 for the process described by eqs([eqevent1a]-[eqevent1c ] ) with @xmath74 and @xmath75 . because @xmath64 is the parent of @xmath9
, we must have @xmath76 . if @xmath77 , then @xmath78 is larger than @xmath79 up until their point of intersection at @xmath80 . in this case
their point of intersection provides a lower bound as illustrated by the dotted line in fig .
[ figextrememt2 ] for the case with @xmath81 . for @xmath82
there are two solutions @xmath83 when @xmath84 , the function @xmath57 has only one zero with the lower bound equalling the upper bound at @xmath37 . the solid line in fig .
[ figextrememt2 ] shows this case .
between @xmath85 we again have two solutions but this time with @xmath86 the dashed line in fig .
[ figextrememt2 ] shows this case with @xmath87 . for @xmath73
greater than this , we have @xmath88 and @xmath89 .
shows @xmath57 for the extreme event in eq([eqevent1a]-[eqevent1c ] ) with @xmath74 and @xmath75 .
the dotted line has @xmath81 and shows an event providing a lower bound on @xmath24 .
the dashed line @xmath87 and shows an event with both a lower bound and an upper - bound on @xmath24 .
the the solid line shows @xmath84 where the lower bound equals the upper bound.,width=384 ] this example illustrates how @xmath0 can provide both a lower - bound and an upper - bound on the true mass for those events with large utm .
the upper - bound distribution provides extra information that can also be used to improve early mass determination , and in what follows we will refer to the upper bound as @xmath1 .
we now move on to discuss modeling and simulation of this new observation .
as a specific example of the application of the @xmath0 method , we have chosen a supersymmetry model msugra , @xmath19 gev , @xmath20 gev , @xmath21 , @xmath22 , @xmath23 . from
@xcite which we also used in @xcite . ] the spectrum used in the simulation has @xmath90 and @xmath91 .
we have employed two simulation packages .
one is a mathematica code that creates the ` ideal ' distributions based only on very simple assumptions and input data .
the second is herwig @xcite which simulates events based on a susy spectrum , mssm cross sections , decay chains , and appropriate parton distribution functions .
if the simple mathematica simulator predicts ` ideal ' shapes that agree with herwig generator , then one has reason to believe that the all the relevant factors relating to the shape are identified in the simple mathematica simulation .
this is an important check in validating the benefits of fitting the @xmath0 and @xmath1 distribution shape as a method to measure the mass of new invisible particles produced at hadron colliders .
our ` ideal ' distributions are produced from a home - grown monte carlo event generator written in mathematica .
this generater serves to ensure that we understand the origin of the distribution shape .
it also ensures that we have control over measuring the parameters needed to determine the mass without knowing the full model , coupling coefficients , or parton distribution functions .
we also use this simulation to determine on what properties the ideal distributions depends .
the simulator is used to create events satisfying the topology shown in fig [ figeventtopology ] for a set of specified masses .
we take as given the previously measured mass difference @xmath92 , which we use in all our simulations .
we neglect finite widths of the particle states as most are in the sub gev range for the model we are considering .
we neglect spin correlations between the two branches .
we perform the simulations in the center - of - mass frame because @xmath0 and @xmath1 are transverse observables and are invariant under longitudinal boosts .
the collision energy @xmath93 is distributed according to normalized distribution @xmath94 unless otherwise specified .
the @xmath13 is produced with a uniform angular distribution , and all subsequent decays have uniform angular distribution in the rest frame of the parent . the utm is simulated by making @xmath27 equal to the utm with @xmath95 ( unless otherwise specified ) , and boosting the other four - vectors of the event such that the total transverse momentum is zero .
as we will show , these simple assumptions capture the important elements of the process .
being relatively model independent , they provide a means of determining the mass for various production mechanisms .
if one were to assume detailed knowledge of the production process , it would possible to obtain a better mass determination by using a more complete simulation like herwigto provide the ` ideal ' distributions against which one compares with the data . here
we concentrate on the more model independent simulation to demonstrate that it predicts the @xmath0 and @xmath1 distributions well - enough to perform the mass determination that we demonstrate in this case - study .
the @xmath0 and @xmath1 distributions of herwig events before smearing ( to simulate detector resolution ) is applied .
the distributions end - points show @xmath96 gev .
the top thick curve shows the net distribution , the next curve down shows the contribution of only the signal events , the bottom dashed curve shows the contribution of only background events.,title="fig:",width=297 ] the @xmath0 and @xmath1 distributions of herwig events before smearing ( to simulate detector resolution ) is applied .
the distributions end - points show @xmath96 gev .
the top thick curve shows the net distribution , the next curve down shows the contribution of only the signal events , the bottom dashed curve shows the contribution of only background events.,title="fig:",width=297 ] in order to obtain a more realistic estimate of the problems associated with collision data , we generate samples of unweighted inclusive supersymmetric particle pair production , using the herwig monte carlo program with lhc beam conditions .
these samples produce a more realistic simulation of the event structure that would be obtained for the supersymmetric model studied here , including the ( leading order ) cross sections and parton distributions .
it includes all supersymmetric processes and so contain the relevant background processes as well as the particular decay chain that we wish to study .
charged leptons ( @xmath97 and @xmath98 ) produced in the decay of heavy objects ( susy particles and @xmath99 and @xmath100 bosons ) were selected for futher study provided they satisfied basic selection criteria on transverse momentum ( @xmath101 gev ) and pseudorapdity ( @xmath102 ) .
leptons coming from hadron decays are usually contained within hadronic jets and so can be experimentally rejected with high efficiency using energy or track isolation criteria .
this latter category of leptons was therefore not used in this study .
the acceptance criterion used for the hadronic final state was @xmath103 .
the detector energy resolution functions used are described in section [ sec : detector ] .
there are several factors that control or affect the shape of the @xmath0 and @xmath1 distributions .
we divide the factors into those that affect the in - principle distribution and the factors that affect the observation of the distribution by the detector like energy resolution and selection cuts .
the in - principle distribution of these events is influenced by the presence or absence of spin - correlations between the branches , the the @xmath26 distribution of the visible particles , any significant upstream transverse momentum ( utm ) against which the system is recoiling ( _ e.g. _ gluinos or squarks decaying further up the decay chain ) , and background coming from other new - physics processes or the standard model . as all these processes effectively occur at the interaction vertex , there are some combinatoric ambiguities .
these are the factors that influence the in - principle distribution of events that impinges on the particle detector .
the actual distribution recorded by the detector will depend on further factors .
some factors we are able to regulate for example cuts on the missing transverse momentum .
other factors depend on how well we understand the detector s operation such as the energy resolution and particle identification . where the effect of such factors is significant , for example for the @xmath25 , @xmath27 , and background distributions , our approach has been to model their effect on the ideal distributions by using appropriate information from the ` data ' , much as one would do in a real lhc experiment . for the present
our ` data ' are provided by herwig , rather than lhc events , but the principle is the same . * * mass difference and mass scale * the end - point of @xmath0 and @xmath1 distributions give the mass of @xmath13 .
therefore the mass scale , @xmath104 , is a dominant factor in the shape of the ` ideal ' distribution .
this is the reason we can use these distributions to determine the mass scale .
fig [ figlotsofidealcurves ] shows the @xmath0 and @xmath1 distributions for five choices of @xmath104 assuming the herwig generated @xmath26 and utm distributions .
we show the @xmath0 and @xmath1 ideal distributions for five choices of @xmath104 assuming the herwig generated @xmath26 and utm distributions.,width=384 ] how does the shape change with mass scale ?
the shape is typically sharply peaked at @xmath105 followed by a tail that ends at the mass of @xmath104 .
the peak at @xmath56 is due to events that are compatible with @xmath106 .
we say these events give the trivial constraint . because we bin the data , the height of the first bin depends on the bin size .
as @xmath107 becomes larger , then the non - trivial events are distributed over a wider range and the endpoint becomes less clear .
in general if all other things are equal , the larger the mass , the more events in the first bin and a longer flatter tail .
the distribution also depends on the mass difference @xmath108 which we assume has been determined .
we expect that experimentally one should be able to read off the mass difference from the @xmath26 kinematic end - point with very high precision .
gjelsten , miller , and osland estimate this edge can be measured to better than @xmath109 @xcite using many different channels that lead to the same edge , and after modeling energy resolution and background .
errors in the mass determination propagated from the error in the mass difference in the limit of @xmath110 are given approximately by @xmath111 where @xmath112 is the error in the determination of the mass difference @xmath108 .
an error in @xmath108 will lead to an @xmath0 distribution with a shape and endpoint above or below the true mass in the direction indicated by eq([eqdeltammerroreffects ] ) . to isolate this source of error from the uncertainty in the fit , we assume that the mass difference is known exactly in our stated results . in our case
an uncertainty of @xmath113 would lead to an additional @xmath114 to be added in quadrature to the error from fitting . * * spin correlations * there are no spin correlation effects relevant to @xmath0 if the @xmath13 pairs are directly produced and the @xmath13 three - body decay is dominated by the @xmath100 boson ( _ e.g _ the sleptons are much heavier that @xmath13 ) @xcite .
there are also no spin correlations if the @xmath13 parents are part of a longer decay chain which involves a scalar at some stage . in the model we are studying ,
there are enough vertices between the two @xmath13 decays that any correlation is very likely washed out , and we can treat their decays as uncorrelated . in the simple mathematica simulations , we have assumed no spin dependence in the production of the hypothetical ideal distribution . *
* input @xmath25 distributions * the @xmath26 distribution affects the @xmath0 distribution .
fig [ figmlldep ] shows two @xmath26 distributions and the corresponding @xmath0 distributions with @xmath110 ( no utm ) .
the solid lines show the case where the three - body decay from @xmath13 to @xmath14 is completely dominated by a @xmath100 boson .
the dashed line shows the case where the @xmath26 distribution is extracted from the ` realistic ' herwig simulation .
we can see that the @xmath0 distribution is affected most strongly in the first several non - zero bins . if we were to determine the mass only from the shape of these first several bins using only the @xmath100 contribution for the @xmath26 difference , we would estimate of the mass to be about @xmath115 gev below the true mass .
this can be understood because the the shape change of the @xmath26 distribution effectively took events out of the first bin and spread them over the larger bins simulating the effect of a smaller mass .
dependence of @xmath0 distribution on the @xmath26 distribution with @xmath54 . * left : * the @xmath26 distributions .
* right : * the corresponding @xmath0 distributions .
the solid curves show the case where the @xmath26 distribution when the three - body decay is dominated by the @xmath100 boson channel , and the dashed curves show the case where the @xmath26 distribution is taken directly from the herwig simulation.,title="fig:",width=288 ] dependence of @xmath0 distribution on the @xmath26 distribution with @xmath54 .
* left : * the @xmath26 distributions . *
right : * the corresponding @xmath0 distributions .
the solid curves show the case where the @xmath26 distribution when the three - body decay is dominated by the @xmath100 boson channel , and the dashed curves show the case where the @xmath26 distribution is taken directly from the herwig simulation.,title="fig:",width=288 ] * * input upstream transverse momentum distribution * as we discussed in section [ secub ] , if there is a large upstream transverse momentum ( utm ) against which the two @xmath13 s recoil , then we have both an upper and lower bound on the mass scale .
the left frame of fig . [ figm2cisrideal ] shows the utm distribution observed in the ` realistic ' herwig data . the right frame of fig .
[ figm2cisrideal ] shows the @xmath0 and @xmath1 distributions for fixed utm ( @xmath27 ) of @xmath116 , @xmath117 , @xmath118 , @xmath119 , @xmath120 , and @xmath121 @xmath122 all with @xmath95 .
as we discuss under the next bullet , we also find the distribution is not sensitive to the value of @xmath123 .
for @xmath124 , these curves begin to approach a common shape .
these are ideal @xmath0 upper and lower bound distributions where @xmath125 and @xmath126 .
notice that there is no upper - bound curve for the case with zero @xmath27 utm .
the utm makes the distribution have a sharper endpoint and thereby make the mass easier to determine .
this is equivalent to having a sharper kink in @xmath2 in the presence of large utm @xcite .
how do we determine @xmath27 from the data ? because we demand exactly four leptons ( two ossf pairs ) , we assume all other activity , basically the hadronic activity , in the detector is utm . the shape used in the ` ideal ' distribution is a superposition of the different fixed utm distributions , shown on right frame of fig .
[ figm2cisrideal ] , weighted by the observed utm distribution , shown on the left frame of fig .
[ figm2cisrideal ] .
equivalently , we obtain the ideal distribution by selecting @xmath27 in the mathematica monte carlo according to the observed utm distribution .
* left : * the utm distribution observed in the herwig simulation .
* right : * ideal @xmath0 upper bound and lower bound distribution for a range of upstream transverse momentum ( utm ) values ( @xmath127 ) where @xmath128 gev and @xmath129 gev . ,
title="fig:",width=297 ] * left : * the utm distribution observed in the herwig simulation . *
right : * ideal @xmath0 upper bound and lower bound distribution for a range of upstream transverse momentum ( utm ) values ( @xmath127 ) where @xmath128 gev and @xmath129 gev . ,
title="fig:",width=297 ] * * shape largely independent of parton distributions and collision energy * in the limit where there is no utm , then @xmath0 is invariant under back - to - back boosts of the parent particles ; therefore , @xmath0 is also invariant to changes in the parton distribution functions .
how much of this invariance survives in the presence of large utm ?
the answer is that it remains largely independent of the parton collision energy and largely independent of the mass @xmath123 as shown in fig [ figdiff ] numerically . on the left frame ,
we show three distributions and in the right frame their difference with @xmath130 error bars calculated from @xmath131 events .
the first distribution assumes @xmath132 , @xmath95 , @xmath93 distributed via @xmath133 .
the second distribution assumes @xmath132 , @xmath134 , @xmath93 distributed via eq([eqsdep ] ) .
the third distribution assumes @xmath132 , @xmath95 , and a fixed collision energy of @xmath135 .
shows that even with large utm , the distribution is independent of @xmath123 and the parton collision energy . shown
are three distributions and their difference calculated from @xmath131 events .
( 1 ) @xmath132 , @xmath95 , @xmath93 distributed via eq([eqsdep ] ) .
( 2 ) @xmath132 , @xmath134 , @xmath93 distributed via eq([eqsdep ] ) .
( 3 ) @xmath132 , @xmath95 , @xmath135.,title="fig:",width=297 ] shows that even with large utm , the distribution is independent of @xmath123 and the parton collision energy .
shown are three distributions and their difference calculated from @xmath131 events .
( 1 ) @xmath132 , @xmath95 , @xmath93 distributed via eq([eqsdep ] ) .
( 2 ) @xmath132 , @xmath134 , @xmath93 distributed via eq([eqsdep ] ) .
( 3 ) @xmath132 , @xmath95 , @xmath135.,title="fig:",width=297 ] * * backgrounds * the invariant mass of the ossf leptons from both branches of forming a dalitz - like wedgebox analysis .
the events outside the @xmath136 signal rectangle provide control samples from which we estimate the background shape and magnitude .
the dark events are signal , the lighter events are background.,width=480 ] backgrounds affect the shape and , if not corrected for , could provide a systematic error in the estimated mass . in section [ secperformance ] we will see that the position of the minimum @xmath137 in a fit to @xmath138 is barely affected by the background .
the main effect of the background is to shift the parabola up , giving a worse fit . to improve the fit
, we may be able to estimate the @xmath0 and @xmath1 distribution and magnitude of the background from the data itself .
we first discuss the sources of background , and then we describe a generic technique using a dalitz - like wedgebox analysis to estimate a background model which gives approximately the correct shape and magnitude of the background . one reason we study the four - lepton with missing transverse momentum channel is because of the very - low standard - model background @xcite . a previous study @xcite estimates about @xmath139 standard - model four - lepton events ( two ossf pairs ) for @xmath140 with a @xmath141 gev cut .
they suggest that we can further reduce the standard - model background by requiring several hadronic jets .
because we expect very - little direct @xmath13 pair production , this would have very little effect on the number of signal events .
also , because @xmath100s are a part of the intermediate states of these background processes , very few of these events will have @xmath26 significantly different from @xmath142 .
what is the source of these standard - model backgrounds ?
about @xmath143 is from @xmath100-pair production events with no invisible decay products , in which the missing transverse momentum can only arise from experimental particle identification and resolution errors .
this implies that a slightly stronger @xmath41 cut could further eliminate this background .
another @xmath144 are due to @xmath145 production .
not explicitly discussed in their study but representing another possible souce of backgrounds are events containing heavy baryons which decay leptonically .
if we assume @xmath146-quark hadrons decay to isolated leptons with a branching ratio of @xmath147 , then lhc @xmath148 production will lead to about @xmath149 events passing these cuts for @xmath140 where both ossf leptons pairs have @xmath150 .
tau decays also provide a background for our specific process of interest .
the process @xmath151 will be misidentified as @xmath152 or @xmath153 about @xmath154 of the time . because the @xmath155 decays introduce new sources of missing transverse momentum ( @xmath156 )
, these events will distort the @xmath0 calculation .
this suggests that the dominant background to the @xmath157 hadrons will be from other susy processes .
we now create a crude background model from which we estimate the magnitude and distribution of the background using the ` true ' herwig data as a guide .
we follow the suggestion of ref .
@xcite and use a wedgebox analysis plotting the invariant mass @xmath158 against @xmath159 to supplement our knowledge of the background events mixed in with our signal events .
this wedgebox analysis , seen in fig . [ figmlldalitzlike ] for our herwig simulation , shows patterns that tell about other susy states present .
the presence of the strips along @xmath160 gev indicate that particle states are being created that decay to two leptons via an on - shell @xmath100 .
the observation that the intensity changes above and below @xmath161 gev shows that many of the states produced have one branch that decays via a @xmath13 and the other branch decays via an on - shell @xmath100 .
the lack of events immediately above and to the right of the @xmath162 coordinate but below and to the left of @xmath163 coordinate suggest that symmetric process are not responsible for this background .
we also see the density of events in the block above @xmath164 and to the right of @xmath164 suggest a cascade decay with an endpoint near enough to @xmath165 that it is not distinguishable from @xmath142 . following this line of thinking , we model the background with a guess of an asymmetric set of events where one branch has new states @xmath64 , @xmath166 and @xmath11 with masses such that the @xmath26 endpoint is @xmath167 and the other branch is our @xmath13 decay .
the masses one chooses to satisfy this edge did not prove important so long as the mass differences were reasonably sized ; we tried several different mass triplets ending with the lsp , and all gave similar answers .
we now describe the background model used in our fits .
one branch starts with a massive state with @xmath168 which decays to a lepton and a new state @xmath169 which in - turn decays to to a lepton and the lsp .
the second branch has our signal decay with the @xmath13 decaying to @xmath14 and two leptons via a three - body decay .
we added utm consistent with that observed in the events . by matching the number of events seen outside the @xmath170 region
, we estimate the number of the events within the signal cuts that are due to backgrounds .
we estimate @xmath171 of the events with both ossf pairs satisfying @xmath170 are background events .
the model also gives a reasonable distribution for these events .
inspecting the actual herwig results showed that actual fraction of background events was @xmath172 .
if we let the fraction be free and minimize the @xmath137 with respect to the background fraction , we found a minimum at @xmath173 .
our background model is simplistic and does not represent the actual processes , but it does a good job of accounting for the magnitude and the shape of the background mixed into our signal distribution .
most of the herwig background events came from @xmath99 and charginos which introduce extra sources of missing transverse momentum .
never the less , the shape fit very accurately and the performance is discussed in section [ secperformance ] .
it is encouraging that our estimate of the background shape and magnitude is relatively insensitive to details of the full spectrum . even ignoring the background , as we will see in section [ secperformance ] , still leads to a minimum @xmath137 at the correct mass . * * combinatoric ambiguities * if we assume that the full cascade effectively occurs at the primary vertex ( no displaced vertices ) , then the combinatoric question is a property of the ideal distribution produced in the collisions .
there are no combinatoric issues if the two opposite - sign same - flavor lepton pairs are each different flavors .
however if all four leptons are the same flavor , we have found that we can still identify unique branch assignments 90% of the time .
the unique identification comes from the observation that both pairs must have an invariant mass @xmath26 less than the value of the @xmath174 edge . in @xmath175 of the events
, there is only one combination that satisfies this requirement .
this allows one to use @xmath176 of the four lepton events without ambiguity .
the first @xmath177 are identified from the two ossf pairs being of different flavors and @xmath175 of the remaining can be identified by requiring both pairs satisfy @xmath178 on which branch .
the events which remain ambiguous have two possible assignments , both of which are included with a weight of @xmath179 in the distribution .
as just described , the ` ideal ' in - principle distribution is created from the observed @xmath26 distribution and the observed utm distribution .
we include combinatoric effects from events with four leptons of like flavors .
last , we can estimate the magnitude of background events and their @xmath0 and @xmath1 shape .
we now modify the in - principle distribution to simulate the effects of the particle detector to form our final ` ideal ' distribution that includes all anticipated effects .
the two main effects on the @xmath0 and @xmath1 distributions are the energy resolution and the @xmath41 cuts . *
* shape dependence on energy resolution * energy resolution causes the @xmath0 and @xmath1 distributions to be smeared .
here we assume the angular resolution is negligible .
for both the mathematica monte carlo model and the herwig events we simulate the detector s energy resolution by scaling the four vectors for electrons , muons , and hadrons by @xmath180 respectively @xcite@xcite .
a more detailed detector simulation is of course possible , but since we do not know the true behavior of any lhc detector until the device begins taking data , a more sophisticated treatment would be of limited value here . in practice the dependence of the ideal distribution shapes on
the missing transverse momentum resolution should reflect the actual estimated uncertainty of the missing transverse momentum of the observed events .
smearing of the distributions decreases the area difference between two normalized distributions , thereby decreasing the precision with which one can determine the mass from a given number of signal events .
this expanded uncertainty can be seen in section [ secperformance ] .
the @xmath0 calculations depend on the mass difference , the four - momenta of the four leptons , and the missing transverse momentum .
as the lepton energy resolution is very tight , the missing transverse momentum s energy resolution is dominated by the hadronic energy resolution .
we model the energy resolution of the utm as a hadronic jet .
this significantly increases the uncertainty in the missing transverse momentum because hadrons have about five times the energy resolution error . in our mathematica model
, we represent the utm as a single four - vector @xmath18 , but in reality it will be the sum of many four - vectors . because we apply the energy resolution smearing to @xmath18 , if @xmath18 is small the simple mathematica model will have a smaller missing energy resolution .
however , an events with almost @xmath116 utm could have a large missing momentum energy resolution if it has a lot of hadronic jets whose transverse momentum mostly cancels .
fig [ figm2cisrideal ] shows that most of the time we have considerable hadronic utm , so this effect is a minor correction on our results . * * shape dependence on missing transverse momentum cuts * a key distinguishing feature of these events is missing transverse momentum . to eliminate the large number of standard model events with four - lepton and with no @xmath41
, we will need to cut on this parameter .
[ figptcutdependence ] shows the herwig simulation s missing transverse momentum versus the @xmath0 .
a non - trivial @xmath0 requires substantial @xmath41 .
small @xmath41 of less than about @xmath181 only affects the @xmath0 shape below about @xmath182 .
the shape of the @xmath183 therefore will require a higher fidelity model from which to train the shapes .
instead , we just choose to not fit bins with @xmath184 .
all events near the end of @xmath1 distribution require significant @xmath41 , therefore @xmath41 cuts will not affect the part of this distribution which we fit .
the number of events with no non - trivial upper - bounds will also be affected by @xmath41 cuts .
we only fit the @xmath1 distribution up to about @xmath185 . the missing transverse momentum vs @xmath0 values for herwig data .
this shows that a @xmath186 cut would not affect the distribution for @xmath187.,width=528 ]
the result of @xmath137 fits to the data with differing assumptions for @xmath3 ( left pane ) and @xmath188 ( right pane ) .
the thick line with filled squares shows the final result with all cuts , resolution error , combinatorics , and backgrounds included and estimated in the shape fitting .
this gives us @xmath189 with @xmath190 events ( signal or background ) representing @xmath3 .
after @xmath188 this improves to @xmath191 .
the error - free best case gives @xmath192 .
the correct value is @xmath90 .
, title="fig:",width=297 ] the result of @xmath137 fits to the data with differing assumptions for @xmath3 ( left pane ) and @xmath188 ( right pane ) .
the thick line with filled squares shows the final result with all cuts , resolution error , combinatorics , and backgrounds included and estimated in the shape fitting .
this gives us @xmath189 with @xmath190 events ( signal or background ) representing @xmath3 .
after @xmath188 this improves to @xmath191 .
the error - free best case gives @xmath192 .
the correct value is @xmath90 .
, title="fig:",width=297 ] determining the mass based on the shape of the distribution enables one to use all the events and not just those near the end point .
we fit both upper - bound and lower - bound shapes to the data as described in the appendix [ appendixchisqfitting ] .
as one expects , fitting the lower - bound shape more tightly constrains the mass from below and fitting the upper - bound shape more tightly constrains the mass from above . combining the two gives approximately even uncertainty .
we calculate ideal distributions assuming @xmath138 at five values @xmath193 , @xmath194 , @xmath195 , @xmath196 , and @xmath197 gev .
we then fit a quadratic interpolation through the points .
our uncertainties are based on the value where @xmath137 increases by @xmath28 from its minimum of this interpolation .
this uncertainty estimate agrees with about @xmath198 of the results falling within that range after repeated runs .
our uncertainty estimates do not include the error propagated from the uncertainty in the mass difference ( see eq([eqdeltammerroreffects ] ) ) .
we present results for an early lhc run , about @xmath199 , and for the longest likely lhc run before an upgrade , about @xmath200 .
after about @xmath201 , we have @xmath190 events ( about @xmath4 signal and @xmath202 background ) .
after @xmath200 , we have about @xmath203 events ( about @xmath204 signal and @xmath205 background ) .
only @xmath115 events out of @xmath204 are from direct pair production .
most of our signal events follow at the end of different decay chains starting from gluinos or squarks .
the upstream decay products produce significant utm against which the two @xmath13 parent particles recoil .
first for the ideal case .
after @xmath206 , using only signal events and no energy resolution , the @xmath137 fits to the predicted shapes give @xmath192 ( filled circles in fig .
[ figperformanceplots ] ) .
this mass determination can practically be read off from the endpoints seen in fig .
[ figherwigresults ] ; the @xmath0 endpoint is near @xmath207 and subtracting the mass differences gives @xmath208 .
we now explore how well we can do with fewer events and after incorporating the effects listed in section [ secshapefactors ] .
how does background affect the fit ? if we ignore the existence of background in our sample , and we fit all the events to the signal - only shapes , then we find a poor fit shown as the empty circle curve in fig .
[ figperformanceplots ] . by poor fit ,
we mean the @xmath137 is substantially larger than the @xmath209 bins being compared ( @xmath210 bins from each the upper - bound and lower - bound distributions ) . despite this worse fit ,
the shape fits still give a very accurate mass estimate : @xmath211 after @xmath199 and @xmath212 after @xmath200 . at this stage
, we still assume perfect energy resolution and no missing transverse momentum cut .
next , if we create a background model as described in section [ secshapefactors ] , we are able to improve the @xmath137 fit to nearly @xmath28 per bin ; the mass estimate remains about the same , but the uncertainty increases by about @xmath213 .
we find a small systematic shift ( smaller than the uncertainty ) in our mass prediction as we increase the fraction of the shape due to the background model vs the signal model .
as we increased our fraction of background , we found the mass estimate was shifted down from @xmath214 at @xmath215 background to @xmath216 when we were at @xmath217 background .
the best @xmath137 fit occurs with @xmath218 background ; which is very close to the @xmath219 we use from the estimate , but farther from the true background fraction of about @xmath220 . with @xmath188 of data ,
the systematic errors are all but eliminated with the endpoint dominating the mass estimate .
these fits are shown as the triangles with dashed lines and give @xmath221 which after the full run becomes @xmath222 . including energy resolution as described in section [ secshapefactors ] shows a large increase in the uncertainty .
the dashed - line with empty square markers shows the @xmath137 fit when we include both a background model and the effect of including energy resolution .
these fits are shown as the empty squares with dashed lines and give @xmath223 which after the full run becomes @xmath224 . the final shape factor that we account for are the cuts associated with the missing transverse momentum .
after we apply cuts requiring @xmath225 and fit only @xmath187 we have our final result shown by the thick lines with filled squares .
this includes all cuts , resolution error , combinatorics and backgrounds .
we find @xmath189 with @xmath190 events ( signal or background ) representing @xmath3 , and after @xmath188 this improves to @xmath191 .
the true mass on which the herwig simulation is based is @xmath226 , so all the estimates are within about @xmath227 of the true mass . herwig data for @xmath3 ( thick line ) and the smooth ideal expectation assuming @xmath228 generated by mathematica with all resolution , background , and combinatoric effects included ( thin line ) .
the @xmath137 of this curve to the herwig gives the solid - black square on the left frame of fig .
[ figperformanceplots].,width=384 ] fig [ figherwigwithexpectation ] shows the ideal curve expected if @xmath228 including all effects from energy resolution , background , combinatoric , and @xmath41 cuts .
the @xmath137 corresponds to the solid square on the left pane of fig .
[ figperformanceplots ] .
the error in mass determination obtained with limited statistics can be estimated using poisson statistics . in our studies
we find that , as one would expect , increasing the number of events by a factor of four , we bring down our error by about a factor of two .
this means that one could expect @xmath229 after about @xmath230 which represents @xmath231 signal events and @xmath117 background events .
despite adding some of the complicating effects one would encounter with real data , we have discovered other factors which demonstrate one could obtain an even better precision than we originally reported in ref .
there , we used only our simple mathematica model that assumed @xmath54 and neglected most sources of realistic uncertainty .
we assumed all the events could be modeled as being direct production without spin - correlations . with these simplifications ,
we argued the mass could be determined to @xmath232 using @xmath233 signal events . in this paper , we performed a case study to show that that the relevant @xmath0 and @xmath1 shapes can be successfully determined from the mass difference @xmath56 , the @xmath26 distribution observation , the upstream transverse momentum ( utm ) distribution observation .
we included and accounted for many realistic effects : we modeled the large energy - resolution error of hadronic jets .
we also included the effects of backgrounds , @xmath41 cuts , and combinatorics .
our signal and backgrounds were generated with herwig .
we discussed how a dalitz - like plot can estimate the background fraction and shape .
observed inputs were used in a simple model to determine ideal distribution shapes that makes no reference to parton distribution functions , cross sections , or other model - dependent factors that we are not likely to know early in the process . despite these extra sources of uncertainty , we found a final mass determination of @xmath234 with about @xmath4 signal events which is still better than the appropriately scaled result from ref .
the sources of the mass determination improvement are twofold : ( 1 ) the prediction and fitting of upper - bound distribution , and ( 2 ) the sharper end - point in the presence of large utm . under equivalent circumstances ,
the sharper endpoint is enough to give a factor of @xmath29 improvement in the uncertainty over the direct production case assumed in @xcite .
fitting the upper bound tends to improve the determination by an additional factor of @xmath235 .
this improvement is then used to fight the large hadronic - jet energy resolution and background uncertainty .
mass determination using @xmath0 and @xmath1 applies to many other processes .
we have focused on cases where the mass difference is given by the end - point of an @xmath26 distribution involving a three - body decay .
if there is not a three - body decay , then the mass difference may be found by applying other mass determination techniques like the mass shell techniques ( mst ) @xcite or edges in cascade decays @xcite or @xmath6 at different stages in symmetric decay chains @xcite .
how does our method s performance compare to previous mass determination methods ?
firstly , this technique is more robust than the @xmath2 ` kink ' because in fitting to the shape of the distribution , it does not rely entirely on identification of the events near kinematic boundary
. one can view @xmath0 and @xmath1 as variables that event - by - event quantify the ` kink ' .
other than the ` kink ' technique , the previous techniques surveyed in the introduction apply to cases where there is no three - body decay from which to measure the mass difference directly .
however , each of those techniques still constrains the mass difference with great accuracy .
the technique of @xcite which uses edges from cascade decays determines the lsp mass to @xmath236 with about 500 thousand events from @xmath237 .
the approach of @xcite assumes a pair of symmetric decay chains and assumes two events have the same structure .
they reach @xmath238 using @xmath190 signal events after @xmath237 , but have a @xmath239 systematic bias that needs modeling to remove . by comparison , adjusting to @xmath190 signal events we achieve @xmath240 without a systematic bias after propagating an error of @xmath109 in the mass difference and with all discussed effects .
uncertainty calculations differ amongst groups , some use repeated trial with new sets of monte carlo data , and others use @xmath137 . without a direct comparison under like circumstance , the optimal method is not clear ; but it is clear that fitting the @xmath0 and @xmath1 distributions can determine the mass of invisible particles at least as well , if not better than the other known methods in both accuracy and precision . in summary , we have developed a mass determination technique , based on the constrained transverse mass , which is able to determine the mass of a dark - matter particle state produced at the lhc in events with large missing transverse momentum .
the @xmath0 method , which bounds the mass from below , was supplemented by a new distribution @xmath1 which bounds the mass from above in events with large upstream transverse momentum .
a particular advantage of the method is that it also obtains substantial information from events away from the end point allowing for a significant reduction in the error .
the shape of the distribution away from the end - point can be determined without detailed knowledge of the underlying model , and as such , can provide an early estimate of the mass .
once the underlying process and model generating the event has been identified the structure away from the end - point can be improved using , for example , herwig to produce the process dependent shape .
we performed a case - study simulation under lhc conditions to demonstrate that mass - determination by fitting the @xmath0 and @xmath1 distributions survives anticipated complications . with this fitting procedure
it is possible to get an early measurement of the mass - with just 400 signal events in our case study we found we would determine @xmath241 .
the ultimate accuracy obtainable by this method is @xmath242 .
we conclude that this technique s precision is as good as , if not better than , the best existing techniques .
we also want to thank chris lester for very helpful conversations and comments on the manuscript , and laura serna for reviewing the manuscript .
ms acknowledges support from the united states air force institute of technology .
this work was partly supported by the science and technology facilities council of the united kingdom .
the views expressed in this paper are those of the authors and do not reflect the official policy or position of the united states air force , department of defense , or the us government .
in order to determine @xmath138 we perform a @xmath243 fit between ideal distributions and the herwig data . first for definitions : we define @xmath244 as the number of @xmath0 events in the region to be fit , and likewise @xmath245 is the number of @xmath1 events in the region to be fit .
the @xmath0 of the events are grouped into bins ; @xmath246 counts the events in the @xmath247 bin .
the variable @xmath248 is the normalized @xmath0 distribution of ideal events expected in bin @xmath249 as calculated with an assumed @xmath138 , the measured @xmath56 , the observed @xmath26 distribution , the observed utm distribution , and the appropriate detector simulator .
we likewise define the upper bound distribution to be @xmath250 .
we also define the background distribution for lower - bound and upper - bound distributions to be @xmath251 and @xmath252 and the fraction of the total events we estimate are from background @xmath253 . assuming a poisson distribution , we assign an uncertainty , @xmath254 , to each bin @xmath249 given by @xmath255 and likewise for the upper - bound distribution .
the second term has been added to ensure an appropriate weighting of bins with very few events that does not bias the fit towards or away from this end - point . in bins with few counts , normal poisson statistics
does not apply has a large percent error when used as a predictor of the number of counts @xmath246 when @xmath256 is less than about 5 . ] .
the @xmath243 is given by @xmath257 we calculate ideal distributions for @xmath258 .
we fit quadratic interplant through the points .
the minimum @xmath259 of the interplant is our estimate of @xmath138 .
the amount @xmath138 changes for an increase in @xmath243 by one gives our @xmath260 uncertainty , @xmath261 , for @xmath138 @xcite .
b. c. allanach , c. g. lester , m. a. parker , and b. r. webber , _ measuring sparticle masses in non - universal string inspired models at the lhc _ , _ jhep _ * 09 * ( 2000 ) 004 , [ http://xxx.lanl.gov/abs/hep-ph/0007009[hep-ph/0007009 ] ] .
b. c. allanach , j. p. conlon , and c. g. lester , _ measuring smuon - selectron mass splitting at the cern lhc and patterns of supersymmetry breaking _ , _ phys .
* d77 * ( 2008 ) 076006 , [ http://xxx.lanl.gov/abs/0801.3666[0801.3666 ] ] . c. g. lester ,
_ constrained invariant mass distributions in cascade decays : the shape of the m(qll)-threshold and similar distributions _ , _ phys .
_ * b655 * ( 2007 ) 3944 , [ http://xxx.lanl.gov/abs/hep-ph/0603171[hep-ph/0603171 ] ] .
b. k. gjelsten , d. j. miller , p. osland , and a. r. raklev , _ mass determination in cascade decays using shape formulas _ , _ aip conf .
* 903 * ( 2007 ) 257260 , [ http://xxx.lanl.gov/abs/hep-ph/0611259[hep-ph/0611259 ] ] .
m. m. nojiri , g. polesello , and d. r. tovey , _ a hybrid method for determining susy particle masses at the lhc with fully identified cascade decays _ , _ jhep _ * 05 * ( 2008 ) 014 , [ http://xxx.lanl.gov/abs/0712.2718[0712.2718 ] ] .
c. g. lester and d. j. summers , _ measuring masses of semi - invisibly decaying particles pair produced at hadron colliders _
lett . _ * b463 * ( 1999 ) 99103 , [ http://xxx.lanl.gov/abs/hep-ph/9906349[hep-ph/9906349 ] ] .
a. j. barr , b. gripaios , and c. g. lester , _ weighing wimps with kinks at colliders : invisible particle mass measurements from endpoints _ , _ jhep _ * 02 * ( 2008 ) 014 , [ http://xxx.lanl.gov/abs/0711.4008 [ 0711.4008 ] ] .
w. s. cho , k. choi , y. g. kim , and c. b. park , _ measuring superparticle masses at hadron collider using the transverse mass kink _ , _ jhep _ * 02 * ( 2008 ) 035 , [ http://xxx.lanl.gov/abs/0711.4526[0711.4526 ] ] .
s. moretti , k. odagiri , p. richardson , m. h. seymour , and b. r. webber , _ implementation of supersymmetric processes in the herwig event generator _ , _ jhep _ * 04 * ( 2002 ) 028 , [ http://xxx.lanl.gov/abs/hep-ph/0204123[hep-ph/0204123 ] ] .
g. marchesini _ et .
_ , _ herwig : a monte carlo event generator for simulating hadron emission reactions with interfering gluons .
version 5.1 - april 1991 _ , _ comput . phys .
* 67 * ( 1992 ) 465508 .
m. bisset , n. kersting , j. li , f. moortgat , and q. xie , _ pair - produced heavy particle topologies : mssm neutralino properties at the lhc from gluino / squark cascade decays _
phys . j. _ * c45 * ( 2006 ) 477492 , [ http://xxx.lanl.gov/abs/hep-ph/0501157[hep-ph/0501157 ] ] .
collaboration , s. akhmadalev _ et .
_ , _ hadron energy reconstruction for the atlas calorimetry in the framework of the non - parametrical method _ , _ nucl . instrum .
_ * a480 * ( 2002 ) 508523 , [ http://xxx.lanl.gov/abs/hep-ex/0104002 [ hep - ex/0104002 ] ] . | we develop techniques to determine the mass scale of invisible particles pair - produced at hadron colliders .
we employ the constrained mass variable @xmath0 , which provides an event - by - event lower - bound to the mass scale given a mass difference .
we complement this variable with a new variable @xmath1 which provides an additional _ upper _ bound to the mass scale , and demonstrate its utility with a realistic case study of a supersymmetry model .
these variables together effectively quantify the ` kink ' in the function @xmath2 which has been proposed as a mass - determination technique for collider - produced dark matter .
an important advantage of the @xmath0 method is that it does not rely simply on the position at the endpoint , but it uses the additional information contained in events which lie far from the endpoint .
we found the mass by comparing the herwig generated @xmath0 distribution to ideal distributions for different masses .
we find that for the case studied , with @xmath3 of integrated luminosity ( about @xmath4 signal events ) , the invisible particle s mass can be measured to a precision of @xmath5 .
we conclude that this technique s precision and accuracy is as good as , if not better than , the best known techniques for invisible - particle mass - determination at hadron colliders . |
intra - cluster medium ( icm ) , i.e. , the hot plasma confined within the gravitational potential of clusters of galaxies , constitutes the most dominant _ known _ form of baryons .
the icm was thought to cool over the hubble time by emitting optically - thin thermal x - rays , because its radiative cooling time is estimated to be @xmath5 yr at the center of many `` cd clusters '' ( those hosting cd galaxies at their centers ) .
this inspired so - called cooling flow ( hereafter cf ) hypothesis ( e.g. * ? ? ?
the idea was apparently supported by several soft x - ray features observed around cd galaxies , including general decreases of the icm temperature , strong excess x - ray surface brightness , and excess photoelectric absorption up to a few times @xmath6 @xmath7 . with the first imaging capability over a broad x - ray energy band up to 10 kev and a much improved energy resolution than was available before , _
@xcite has provided a number of new results that altogether cast serious doubt upon the reality of cfs .
these include ; a shortage of the predicted cool gas in the hydra - a cluster @xcite ; the presence of uncooled icm in 3-dimensional core regions of the centaurus cluster @xcite ; a central excess brightness of abell 1795 that does not depend strongly on the x - ray energy @xcite ; hierarchical potential structures in several galaxy groups @xcite ; and systematic differences in the icm chemical composition between the central and outer regions of cd clusters @xcite .
we compiled all these results in ( * ? ? ? *
hereafter paper i ) , and argued that the classical cf hypothesis needs a significant revision .
the _ asca _
suggestions have been reconfirmed and reinforced by _
chandra _ and _ xmm - newton _ ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
although the cf scenario in its original form is no longer considered valid , the icm temperature of a cd cluster does decrease toward the center .
furthermore , the deprojected radial icm temperature profiles of a fair number of clusters all reduce to a single `` universal '' profile @xcite , which reach , at the cluster center , a minimum value of @xmath8 @xcite . here , @xmath9 is the icm temperature observed outside @xmath10 kpc .
it is yet to be explained how the intriguing universal temperature profile and the scaling of eq.([eq : tcth ] ) are produced , and how they are related to mechanisms which suppress cfs .
the concept of the universal temperature profile is based on a `` single - phase '' ( hereafter 1p ) picture , that the icm at a given three - dimensional radius is represented by a single dominant temperature which monotonically decreases toward the center .
in contrast , the original _ asca _ results on cd clusters ( paper i ) are based primarily on a `` two - phase '' ( hereafter 2p ) view ; the icm consists of a `` hot phase '' and a `` cool phase '' , intermixed together , with the volume filling factor of the latter increasing toward the center .
this view was at the beginning motivated by the complex angular response of the _ asca _ telescopes , but was reinforced _
a posteriori _ by the fact that 2p fits to the _ asca _ spectra of a cluster generally give a pair of well - defined temperatures regardless of the two - dimensional radii used to extract the signal photons ( e.g. fukazawa et al .
1994 ; paper i ) .
in other words , the 2p modeling is likely to be more than a mere convention .
generally , the hot - phase and cool - phase temperatures , derived under the 2p assumption , are respectively close to @xmath11 and @xmath9 of eq.([eq : tcth ] ) obtained through the 1p analysis .
furthermore , eq.([eq : tcth ] ) holds for the 2p results as well @xcite .
the 1p and 2p approaches are thus consistent in the sense that both yield essentially the same two characteristic temperatures , @xmath11 and @xmath9
. nevertheless , their physical implications are much different . the 1p situation would require a fine tuning between the radiative cooling and the putative heating mechanism , or
quasi - hydrostatic " gas cooling @xcite . to realize the 2p condition , in contrast
, the cool phase must be thermally insulated from the hot phase , and heated in a stable manner , so that it should neither evaporate @xcite , nor collapse due to cooling .
yet another possibility is that the icm temperature takes a range of values even at a single radius @xcite ; this is to be called `` multi - phase '' picture . in order to identify the cf - suppression mechanism ,
it is of basic importance to clarify which of the 1p and 2p pictures ( or the multiphase view ) is closer to the reality . in spite of its importance
, there have been few attempts to address this issue .
we hence analyze in the present paper the _ xmm - newton _ data of the centaurus cluster .
this object is best suited to our purpose , because it harbors the very prominent cool component and has extensively been studied with _ asca_. our strategy is to examine whether or not deprojected spectra of this cluster require the presence of multiple temperatures at each 3-dimensional radius .
the data will also allow us to reconfirm the lack of emission with temperature below the value of eq.([eq : tcth ] ) . in 2 , we briefly summarize previous x - ray results on the centaurus cluster , and describe the _ xmm - newton _ observations in 3 .
section 4 , which forms the core of the present paper , is devoted to the description of data analysis and results .
the results are discussed in 5 , followed by a summary in 6 . throughout the paper ,
the hubble constant is expressed as @xmath12 km @xmath13 mpc@xmath13 .
the three - dimensional radius and the projected radius are denoted as @xmath14 and @xmath15 , respectively .
errors represent 90 % confidence limits unless stated otherwise .
located at a redshift of @xmath16 = 0.0104 @xcite where @xmath17 corresponds to 12.5 kpc , the centaurus cluster ( abell 3526 ) is one of the most well studied objects among nearby clusters .
it exhibits a roughly round shape , and its cd galaxy , ngc 4696 , does not show significant nuclear activity . in x - rays ,
the centaurus cluster has been observed extensively , with _ uhuru _
@xcite , _ oso8 _ @xcite , _ heao1 _
@xcite , _ einstein _
@xcite , _ exosat _ @xcite , and _ ginga _ @xcite .
these observations yielded an icm temperature of @xmath0 kev , except in the central region ( @xmath18 ) where a very prominent cool component and an excess brightness are observed .
@xcite studied the centaurus cluster using the data from the _ rosat _ pspc .
they found some evidence of metallicity increase and excess absorption at the center . through a deprojection analysis , they also derived 3-dimensional properties of the icm under the 1p modeling , reporting that the icm is cooled down to @xmath19 kev toward the center . using _ asca _ with the much improved energy resolution and the wider energy band ,
@xcite successfully resolved both the fe - k line and low - energy emission lines , and detected hot and cool components with a temperature of @xmath0 and @xmath19 kev , respectively .
the strong metallicity increase toward the center was reconfirmed .
they also discovered that the 3-dimensional core region ( @xmath20 ) of the cluster is filled mostly with the hot component , whereas the cool component occupies a minor fraction of the volume . through a more detailed analysis of the _ asca _ data
, @xcite showed that the 1p density and temperature profiles derived with _
@xcite underpredict the hard x - ray flux from the central region , and hence fail to explain the 0.510 kev _ asca _ data .
@xcite instead found that the _ asca _ spectra from various regions near the center can be described adequately by the 2p formalism employing the two components identified by @xcite .
this reinforced the presence of the hot phase in the three - dimensional core region , and justified the 2p approach .
the 2p formalism on the centaurus cluster was further refined and reinforced by @xcite . incorporating two temperatures of @xmath21 kev and @xmath22 kev , together with a central metallicity increase
, they successfully constructed a 3-dimensional icm model that can simultaneously explain the _ rosat _ surface brightness and the _ asca _ annular spectra , namely the best - quality data sets available at that time . the excess absorption at the center , clamed by @xcite based on the _ asca _ data , was not reconfirmed by @xcite . utilizing the unprecedented angular resolution of _ chandra _
, @xcite detected a `` plume '' like x - ray feature near the center of the centaurus cluster , reconfirming a previous observation with the _ rosat _ high resolution imager @xcite . within @xmath23 of the center including the `` plume '' region , a cool plasma component with a temperature of @xmath2 kev was clearly detected @xcite . employing the 1p formalism , @xcite also derived the icm temperature distribution near the center ( @xmath24 ) , but the field of view of _ chandra _ was not wide enough to determine the global temperature structure over the entire `` cool '' region ( @xmath18 ) .
there are some indications that a subcluster centered on the elliptical galaxy ngc 4709 is merging with the main body .
this region exhibits an excess x - ray surface brightness and a temperature increase to @xmath25 kev @xcite .
however , these effects , seen @xmath26 off the center , are considered to be small in regions which we analyze ( @xmath27 ) .
in fact , using _ suzaku _ , @xcite showed that the icm within the central @xmath28 can be regarded as hydrostatic : specifically , they set an upper limit of @xmath29 km s@xmath13 on any large - scale ( @xmath10 kpc ) bulk motion in the icm over this region , and ruled out a previously claimed detection of significant bulk motions @xcite .
the centaurus cluster was observed with _
xmm - newton _ on 2002 january 3 for a gross exposure of 45 ksec .
the boresight was set to ( @xmath30 , @xmath31 ) = @xmath32 .
the epic pn and mos were operated both in the full frame mode with the thin filters @xcite , while the rgs in spectroscopy mode @xcite .
the epic dataset obtained in the same observation were already analyzed by @xcite mainly for metallicity distributions .
the present paper emphasizes the temperature structure .
similarly , @xcite utilized the rgs data from the same observation , as well as those from an additional 110 ks observation .
the epic data were extracted using the standard sas software of version 5.4.1 .
we selected events with _ pattern _
@xmath33 for pn and _ pattern _
@xmath34 for mos , with which most of the valid x - ray events are accepted .
according to instructions by the detector team , we further discarded events with bad flags , e.g. , those out of the field of view and those detected at bad ccd pixels .
soft protons often cause the epic background to increase suddenly by up to two orders of magnitude .
these sporadic `` flares '' must be removed carefully when analyzing extended sources .
we accordingly produced 27 kev band light curves of the present data , excluding a central region ( @xmath35 ) where the cluster emission dominates .
this particular energy band was shown by @xcite to serve as a good measure of the epic background , and is roughly optimized in the present case to provide a high signal - to - ratio for proton flares ; below @xmath36 kev the cluster signals become significant , while above @xmath37 kev the fluorescent background lines in the pn spectra become a nuisance .
the results are shown in figure [ fig : lc ] .
following @xcite and @xcite , we then discarded those time periods when the count rates deviate by more than @xmath38 from those in quiescent periods .
after this screening , the net exposure time became 31 ksec with pn and 43 ksec with mos .
several faint point sources are found in the field of view .
we excluded regions where their signals are significant , typically within @xmath39 of each source .
since the cluster emission fills nearly the entire epic field of view , background spectra must be constructed using some other `` blank sky '' data . for this purpose
, we utilized the observation of pg 1115 + 080 for pn ( obsid=0082340101 ) , and that of vik 59 for mos ( obsid=0107860501 ) , because these data sets showed a close similarity to the present centaurus data in terms of high - energy ( 1114 kev for pn and 1012 kev for mos ) count rates accumulated over peripheral regions of the fields of view . after removing proton flares and celestial point x - ray sources in the same manner , we extracted background events from these data sets , using the same regions on the detectors as employed to accumulate the on - source data .
the background derived in this way is estimated to be accurate generally to within @xmath40 , except at some particular spectral regions @xcite , as detailed in [ subsub : syserr ] , where we employ 8% .
this accuracy is sufficient for our purpose , because the cluster emission is much brighter than the background , e.g. , by more than a factor of five , at least up to @xmath41 from the aimpoint in energies below 5 kev .
thus , the background uncertainty therein is at most a few percent of the cluster signal . to discriminate the 1p and 2p conditions
, it is essential to remove the projection effects , using so - called deprojection procedure . to create deprojected spectra
, we employed the standard `` onion peeling '' method ( e.g. * ? ? ?
* ) , assuming that the object is spherically symmetric , and the plasma is uniformly distributed in each thin shell region .
no model spectra were assumed in our deprojection procedure .
specifically , at each energy @xmath42 , we calculated the deprojected spectrum @xmath43 in the @xmath44-th shell as @xmath45 where @xmath46 is the projected spectrum accumulated over the @xmath47-th annulus , @xmath48 is a triangular matrix derived by inverting the projection matrix , and @xmath49 denotes the outermost annulus .
below , we utilize the data within @xmath50 ( 150 kpc ) of the x - ray centroid at @xmath51 , which coincides with the ngc 4696 nucleus .
the emission outside @xmath50 , @xmath52 , was calculated assuming that the plasma is isothermal and isoabundance at @xmath53 , and that the icm density profile is expressed by a @xmath54 model determined jointly with _
asca _ and _ rosat _ , namely @xmath55 and the core radius of @xmath56 @xcite .
these parameters can reproduce the _ xmm - newton _ images themselves @xcite .
actual matrix elements of @xmath57 are numerically given in appendix a. in order to take into account calibration uncertainties , conservative systematic errors of 2% @xcite were assigned to the on - source spectra in energies above 0.6 kev . below that energy , the systematic error was increased to 5% to represent uncertainties in the quantum efficiency around the oxygen absorption edge ( e.g. * ? ? ?
* ) . to further consider systematic uncertainties involved in the background subtraction process , we assigned to the background data the systematic errors as given in table [ tbl : syserror ]
the errors are thus taken to be 8% ( [ subsub : epic_bkgd ] ) in those energy bands where the background is dominated by instrumental fluorescence lines , because their intensities are known to vary rather independently of that of the background continuum . in energies below @xmath19 kev , the background spectra
are significantly contributed by diffuse soft x - ray background , of which the brightness varies from sky to sky .
there , we hence employed the same 8% systematic error , although the background in this energy range is generally negligible compared to the bright cluster emission . in other energy regions , we employed a nominal value of 3% ( [ subsub : epic_bkgd ] ) . again using the sas software , we processed the rgs data , and extracted the first and second order spectra within @xmath58 ( in the cross - dispersion direction ) of the center .
the spectral range of 623 was analyzed , in order to utilize many atomic lines therein .
the rgs background was derived from blank sky fields which were made available by the instrument team @xcite . because these authors reported that the rms fluctuation of the rgs background is not more than 30% over the full spectral range
, we assigned a systematic error of 30% to the rgs background . in reality , the emission from the centaurus cluster exceeds the background by more than an order of magnitude over the interested spectral range ; therefore , the background uncertainties have little effects on the results .
figure [ fig : rawpspec ] shows the pn spectra of the centaurus cluster , obtained by accumulating the on - source events over a series of concentric annular regions centered on the ngc 4696 nucleus ( [ subsub : epic_process ] ) , and then subtracting the background described in [ subsub : epic_bkgd ] .
the radii defining these annuli are given in the figure caption .
the figure also shows their ratios to those averaged over an outer region of @xmath59 , where the cool component is hardly seen .
the spectra from these regions do not bear strong atomic lines except the fe - k line at @xmath60 kev , and their spectral shapes are very similar to one another .
therefore , these regions at @xmath61 must be filled with an approximately isothermal hot plasma .
the figure also confirms correct background subtraction , because wrong subtraction would make the hardest - band ( @xmath62 kev ) spectra significantly depend on the radius . toward the center , in contrast
, we can see prominent increases in the line intensities ; the fe - k line enhancement is caused by the metallicity increase , while those of si - k ( at rest - frame energies of 1.86 and 2.01 kev ) , s - k ( 2.46 and 2.62 kev ) , and fe - l ( @xmath19 kev ) lines are mainly due to the emergence of the cool component . at the very central region ( @xmath63 ) , hereafter called `` cd region '' , fe - l lines with energies below 1 kev become prominent , indicating the presence of a still cooler plasma with a temperature of @xmath64 kev .
the same inference can be obtained from the drastic increases in the h - like to he - like intensity ratios of the si - k and s - k lines .
we therefore qualitatively reconfirm the _
detection of the coolest ( @xmath2 kev in temperature ) component at the center @xcite .
applying the procedure described in
[ subsub : depro ] to the spectra in figure [ fig : rawpspec ] , we derived deprojected spectra from 11 concentric shells .
some of them are presented in figure [ fig : rawdspec ] , in comparison with the non - deprojected spectra .
the deprojection obviously degrades the data statistics , but does not alter the spectral shape very much ; the only eye - catching difference is the reduction in the hard x - ray ( @xmath65 kev ) flux of the innermost spectra ( denoted [ cd ] ) , caused by the removal of contributions from the foreground / background hot emission .
as first - cut evaluation of the annular spectra of figure [ fig : rawpspec ] , we conducted a conventional single - temperature analysis .
this is meant to approximate 1p conditions , but only crudely , because three effects , coupled with the temperature and abundance gradients , would make each spectrum deviate from an exact isothermality ; the most dominant one is the projection effects , while the less important ones are the finite angular resolution of the x - ray telescope which mixes up signals from adjacent shells , and the finite thickness of each annulus .
thus , the single - temperature representation should be regarded as a crude approximation here .
we fitted the three projected epic spectra from each annulus with a single - temperature apec model @xcite with photoelectric absorption , simultaneously but allowing the model normalization to differ among the three detectors .
the abundances of o , ne , mg , si , s , ar , ca , fe , and ni were all left free , while those of c and n were fixed at the solar values by @xcite .
the redshift was fixed at 0.0104 , and the hydrogen column density at the galactic - line - ofsight value , @xmath66 @xmath7 , because @xcite found the absorption to be grossly consistent with it .
since some systematic residuals were observed around the fe - k line region , we incorporated small gain correction factors , + 0.3% to pn , + 0.6% to mos1 , and + 0.3% to mos2 , all within the reported calibration uncertainties .
these improved the fit chi - squared , without significantly affecting the best - fit parameters . as expected from figure [ fig : rawpspec ]
, fits to the cd region spectra left large residuals around the fe - l energies .
we therefore added , in these cases , another cool apec component with a temperature ( left free ) of @xmath2 kev , of which the abundances are tied to that of the major component . as a result
, the fits were much improved , in agreement with the report by @xcite .
this modeling should be distinguished from the 2p approach to be examined later .
the 1p model ( plus the 0.7 kev component ) reproduced the spectra moderately well , although the fits are not necessarily acceptable .
figure [ fig:1prad ] ( filed circles ) shows radial profiles of the obtained parameters .
thus , the plasma temperature approaches a constant value of @xmath67 3.8 kev toward outer regions , while drops inside @xmath68 , in agreement with the inference from figure [ fig : rawpspec]b and with the previous results ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
as mentioned above , the @xmath2 kev plasma component was needed at @xmath69 . in figure [ fig:1prad ] , the abundances of fe and
si are seen to increase toward the center as first revealed with _
@xcite , while that of oxygen is spatially rather constant .
these results are consistent with the single - temperature analysis results from @xcite . although @xcite reported a _
_ detection of a metallicity decrease within @xmath70 ( down to @xmath71 solar at @xmath72 ) , such an effect , if any , is less significant in the epic data .
we performed the same model fitting to the deprojected spectra .
since the spectra are now free from the projection effects , the single - temperature modeling is equivalent to 1p conditions , assuming that each deprojected shell is thin enough and the signal mixing due to the telescope response is negligible .
this analysis yielded the data points represented in figure [ fig:1prad ] by open diamonds .
thus , the overall results are very similar to those obtained with the annular spectra .
the fits became generally better , due to the reduced statistics of the deprojected spectra , and to the removal of fore- and background contributions .
the latter also makes the deprojected temperatures systematically lower ( typically by 0.30.5 kev ) than the non - deprojected values . as an alternative approach
, we fitted the epic spectra by a 2p model , i.e. , a sum of two apec components with different temperatures , plus the @xmath2 kev component at @xmath73 . for each annulus ,
the cool component temperature @xmath11 was left free , while the abundance of each element was assumed to be the same among the components .
like in the 1p analysis ( [ subsub : ana1p ] ) , the c and n abundances were fixed ; so were the redshift and the photoelectric absorption .
when we first let the hot component temperature @xmath9 float as well , it became rather unconstrained in some outer regions , and scattered over a range of @xmath74 to @xmath37 kev .
this is due to coupling with @xmath11 @xcite .
however , the value was generally consistent , within rather large errors , with 3.84.0 kev obtained in past observations .
we hence fixed it at @xmath75 kev after the _ asca _ results @xcite and our 1p fits ( figure [ fig:1prad ] ) . by this 2p model ,
both the non - deprojected and deprojected spectra were reproduced reasonably well in all regions .
the obtained parameters , shown in figure [ fig:2prad ] , are generally consistent with the two - temperature results by @xcite within some differences in the modeling .
thus , the fit goodness is comparable to , or slightly better than , those obtained with the 1p analysis .
the derived abundance profiles are essentially the same as the 1p results , and the radial profile of fe ( fig .
[ fig:2prad]c ) agrees very well with the _ asca _ 2p results @xcite .
the temperature differences between the projected and non - deprojected spectra became smaller than in the 1p analysis , and often insignificant within errors .
this is consistent with a 2p condition , in which an annulus and the corresponding deprojected shell are expected to share nearly the same two temperatures .
figure [ fig : vfill]a shows radial profiles of the emission measures ( per unit volume ) of the hot and cool components , denoted @xmath76 and @xmath77 , respectively .
thus , the cool component is required within @xmath78 , where the 1p temperature decreases significantly in our 1p result ( figure [ fig:1prad ] ) .
the derived @xmath11 stays at @xmath36 kev over @xmath79 ( though rather poorly determined outside @xmath80 ; figure [ fig:2prad]a ) , while it decreases down to @xmath81 kev at the center .
the latter agrees with the central 1p temperature from our 1p fits ( figure [ fig:1prad ] ) and from _ rosat _ @xcite . at the cd region ,
the third 0.7 kev component is again required .
in contrast , neither the ring - sorted spectra nor the deprojected thin - shell ones required the cool component in regions outside @xmath78 .
accordingly , figure [ fig:2prad ] outside @xmath78 simply reproduces the 1p results .
incidentally , the 2p analyses of the _ asca _ data yielded somewhat lower values of @xmath11 ; @xmath19 kev @xcite , or @xmath82 kev @xcite .
this is partially because _ asca _ was not able to separately detect the 0.7 kev component .
the remaining difference may be due to the different plasma codes @xcite , the mekal and apec codes used in the _ asca _ study and the present analysis , respectively . to characterize the obtained 2p solution , let us introduce , after @xcite , the volume filling factor @xmath83 of the cool phase . using the emission measures , this quantity is defined as @xmath84 where @xmath85 and @xmath86 are the densities of the hot and cool phases , respectively , while @xmath87 is a factor of order unity reflecting the metallicity . assuming a pressure balance between the two phases @xcite , namely , @xmath88 we can calculate @xmath83 as @xmath89^{-1}. \label{eq : filling_factor}\ ] ] figure [ fig : vfill]b shows the radial profile of @xmath83 , thus calculated using eq.([eq : filling_factor ] ) ; in deriving this result , we fixed @xmath11 at 2.0 kev in shell regions outside @xmath90
. the figure reconfirms the _ asca _ results @xcite , that the hot phase dominates the volume down to a radius rather close ( @xmath91 or 18 kpc ) to the center .
the filling factor implied by the _ xmm - newton _ data is higher by a factor of @xmath74 than the _ asca _ determination ; as can be understood via eq.([eq : filling_factor ] ) , this is due partially to the higher value of @xmath11 . as we have seen in
[ sub : epic_ana ] , the two descriptions , 1p and 2p , are consistent with each other in the sense that both find essentially the same set of three plasma temperatures , @xmath92 kev , 1.72.0 kev , and @xmath2 kev .
outside @xmath93 ( @xmath94 kpc ) , the icm can be approximated as isothermal at @xmath92 kev . within @xmath93 toward the cluster core , the icm becomes gradually cooler down to @xmath36 kev ( if employing the 1p description ) , or the @xmath36 kev plasma becomes intermixed with the hot 3.8 kev component with a progressively larger @xmath83 ( if employing the 2p description ) . in the cd region ,
both scenarios require the additional 0.7 kev plasma . up to this stage , we are unable to tell which of the two approaches is more favored by the data .
although the 2p fit to the deprojected spectra often gives a slightly lower reduced chi - squared than the 1p fit ( figure [ fig:1prad]b vs. figure [ fig:2prad]b ) , the difference is not necessarily statistically significant , and could be due to the finite temperature gradient within each shell . in order to better distinguish the 1p and 2p modelings under higher statistics , we took the deprojected spectra from consecutive five thin shells covering altogether @xmath95 ( [ e ] though [ i ] in figure [ fig : rawpspec ] ) , which have 1p temperatures of 2.03 , 2.48 , 2.65 , 3.10 , and 3.30 kev respectively ( [ subsub : ana1p ] ) .
we then summed them into a single spectrum which now represents a thicker 3-dimensional shell in the cool core region ; this is hereafter called `` shell c '' .
the cd region , together with the plume " structure , is excluded to avoid complexity introduced by the 0.7 kev component .
since the statistical errors associated with the adjacent thin - shell spectra are not mutually independent , we estimated the errors to be assigned to the thick - shell spectra by properly considering error propagation utilizing eq.([eq : deprojection])@xcite .
the shell c spectra , one for each detector , should contain the 3.8 kev component if the 2p scenario is correct .
if instead the 1p picture is more appropriate , the shell c spectra would rule out contributions from such a high temperature component , because the 1p temperature changes across shell c from 2.0 kev to 3.3 kev but not higher ( figure [ fig:1prad ] ) .
we expect that the improved data statistics , achieved by the spectral summation , allow us to distinguish these two cases .
in addition to the improved statistics , analyzing a thick shell has two more advantages : the signal mixing caused by the finite telescope resolution is reduced , and the errors due to spectral changes within individual thin shells , if any , nearly cancel out . as an instructive exercise of the latter effect ,
let us assume , for example , the 3rd and 4th annuli constituting shell c to have a difference by @xmath96 in their fe - k line equivalent widths , and examine what would happen if we artificially assigned the two annuli with a same value @xmath97 that is the emission - measure - weighted average between the two shells .
this exaggerates the procedure of averaging out a radial spectral change over a finite shell thickness .
the errors caused by the averaging are expected to propagate through deprojection into the 1st through 4th thin shells . however , as shown in appendix a , a mathematical estimate using actual matrix elements of eq.([eq : deprojection ] ) confirms that the errors in the four thin shells roughly cancel out , and hence the final thick shell - c spectrum suffers in the present case no larger errors in the line equivalent width than @xmath98 . since the same estimate applies to a temperature gradient , which is at most @xmath99 in each thin shell ( figure [ fig:1prad ] ) over @xmath100 , the residual temperature errors in the shell c spectrum are estimated to be at most @xmath101 , and hence negligible .
let us examine the shell c spectra against the 1p view .
we hence took the best - fit single - temperature apec model for each of the 5 constituent thin shells , and convolved it with a response of each detector that is weighted by the surface brightness of the projection of the relevant thin shell . in the same way as the actual thick - shell data
, we then summed up the 5 simulated thin - shell spectra of each detector into a single one , to be called `` synthetic 1p spectrum '' .
figure [ fig : regccmpr]a directly compares these synthetic 1p spectra with the actual shell c spectra of the corresponding detectors .
thus , we observe a moderately good agreement , but the actual shell c spectra exhibit some excess above the synthetic 1p model toward higher energies .
in fact , their difference , in terms of chi - squared summed over the three detectors ( pn , mos1 and mos2 ) , becomes @xmath102 as summarized in table [ tbl : chisquares ] . the data vs. model comparison in figure [ fig : regccmpr]a incorporates no adjustment , since the synthetic 1p spectra are uniquely specified by the best - fit models describing the 5 constituent thin shells ( [ subsub : ana1p ] ) .
however , the agreement might be improved by adjusting the model parameters .
we therefore allowed to vary freely the normalizations of the five apec components which constitute the synthetic 1p spectrum .
then , the agreement ( now to be called `` a fit '' ) was improved to @xmath103 ( table [ tbl : chisquares ] ) ; the normalizations of the five components changed by a factor of 1.65 , 0.25 , 0.84 , 1.10 , and 1.26 in the increasing order of the temperature , while the summed normalization remained unchanged within @xmath104 .
thus , the coolest and hottest of the five components tend to be enhanced , implying a 2p - like condition . to retain a stable fit convergence ,
we did not attempt to let the 5 temperatures float . in order to next examine the 2p approach , we fitted the actual shell c spectra with a sum of two apec models , exactly in the same manner as in [ subsub : ana2p ] ; the two temperatures and the two normalizations were left free , as were the abundances of o , ne , mg , si , s , ar , ca , fe , and ni which are assumed to be common to the two components . the relative normalizations among the 3 detectors were left free as before . as a result , the degree of freedom decreased by 15 compared to the synthetic 1p modeling .
the two components were convolved with the same response , weighted by the x - ray surface brightness . as shown in figure [ fig : regccmpr]b and table [ tbl : chisquares ] , the obtained fit ( @xmath105 ) is significantly better than that with the synthetic 1p model , even considering the reduced degree of freedom .
in fact , the high energy excess , observed in figure [ fig : regccmpr]a , has been reduced in figure [ fig : regccmpr]b .
the two temperatures have been obtained as @xmath106 kev and @xmath107 kev . while these values agree well with the 2p results from the individual thin shells ( figure [ fig:2prad ] ) , the obtained @xmath9 is significantly higher than the temperature range ( up to 3.30 kev , with a typical error of @xmath108 kev ) involved in the 1p description of shell c ( figure [ fig:1prad ] ) .
although the error ranges associated with the derived @xmath109 and @xmath11 are somewhat ( typically by @xmath110 ) under - estimated due to rather large values of the reduced chi - squared ( @xmath111 ) , the above inference remains valid even if the error ranges are enlarged by @xmath110 . just for consistency , we applied in the same way the 2p fit to the synthetic 1p spectra , to which poissonian noise was added .
this yielded @xmath112 kev and @xmath113 kev , although the associated errors are rather difficult to estimate .
these two temperatures are both contained within the 1p temperature distribution across shell c ( figure [ fig:1prad ] ) , from 2.03 kev to 3.30 kev , that we employed in constructing the synthetic 1p spectra .
this difference between the actual shell c spectrum and the sythetic 1p spectrum suggests that the hot 3.8 kev component was already present in the 5 thin - shell spectra , and has become more significant by the spectral summation . in the above 2p fit to the actual shell c spectra
, the two components were treated to have the same abundances , and were convolved with the same response .
however , the true 2p condition would be somewhat different , since the cool component , more weighted toward the metal - enriched inner regions than the hot one , must have higher abundances when averaged over shell c. to better express this condition , we returned to the original best - fit 2p models to the five thin - shell spectra , convolved them with projected responses of the respective thin shells , and summed the results over the 5 shells .
these simulated data , involving two fixed temperatures ( 2.0 and 3.8 kev ) and properly considering the radial abundance changes , are to be called `` synthetic 2p spectra '' , because they are constructed exactly in the same way as their 1p counterparts .
although these synthetic 2p spectra do not involve any adjustable free parameter , they have reproduce the shell c spectra with @xmath114 , even better than the above 2p fit , and also than the synthetic 1p modeling . from the statistical viewpoint , even the synthetic 2p simulation obtained in
[ subsub : thick_shell_2p ] does not give an acceptable reproduction of the data ( table [ tbl : chisquares ] ) .
this hampers correct error evaluations , and makes it difficult to quantitatively compare different modelings . in figure
[ fig : regccmpr ] , fit residuals are observed around strong emission lines , such as si - k , s - k , and ar - k .
therefore , the inadequate fit goodness is likely to be caused mainly by subtle calibration inaccuracy in reproducing the detailed line profiles , rather than model inappropriateness . under the higher data statistics ,
the calibration uncertainties presumably exceeded the systematic errors of 2% already assigned in
[ subsub : syserr ] .
although our data analysis utilized rather old sas version ( 5.4.1 ) , a reanalysis of a few representative spectra from the central region , using an updated version 7.0.0 sas software ( processing and responses ) , showed rather insignificant improvements of these problems .
furthermore , the fitting results derived with the two versions agreed within errors . accordingly
, we retain the version 5.4.1 analysis , and assigned 5% of the source counts as a new systematic errors to the whole energy range , and repeated the overall analyses .
this procedure is meant to make the fit acceptable when using the model that is best preferred by the data , and to make securer the distinction among different models . by this error renormalization ,
the values of @xmath115 have reduced appreciably , while the radial profiles of various plasma parameters remained unchanged .
specifically , the comparison of the shell c spectra with the remade synthetic spectra gives @xmath116 and 381 for 1p and 2p , respectively , both with @xmath117 ( table [ tbl : chisquares ] ) .
the latter is statistically acceptable , and gives a chi - squared which is smaller by 30 than the former .
thus , the 2p modeling of the shell c spectra is statistically superior to the 1p one . in a similar way
, we derived epic spectra from another thick shell , called shell p , which sums up three consecutive thin shells covering @xmath118 .
this shell p is peripheral to the central cool region .
the summed shell p spectra were then compared with the corresponding synthetic 1p and 2p spectra , constructed in the same manner as before from the best - fit solutions to the 3 constituent thin shells .
the comparison of the actual spectra with the synthetic 1p and 2p spectra gave @xmath119 and 496 ( without the error renormalization ) , respectively , both with @xmath117 .
the difference in chi - squared , though smaller than in the shell c case , amounts to 15 , and remains @xmath120 even when the systematic error is renormalized as in [ subsub : renormalize ] .
we also fitted the shell p spectra with the 2p model , in the same way as for the shell c case but fixing @xmath9 at 3.8 kev as we did in [ subsub : ana2p ] .
we then obtained @xmath121 kev , together with @xmath122 .
this value of @xmath11 again agrees with those obtained by the 2p fits to the constituent thin - shell spectra ( figure [ fig:2prad ] ) , while it is significantly below the distribution of the 1p temperature across shell p , 3.03.5 kev ( figure [ fig:1prad ] ) .
therefore , the cool ( @xmath67 2.0 kev ) component is inferred to be present even in this relatively outer region . as a consistency check , we fitted the synthetic 1p spectra in shell p with the same 2p model , and obtained @xmath123 kev which agrees with the input 1p temperature range .
we have so far found that the two sets of thick - shell spectra both prefer the 2p view to the 1p scenario . however , the results could be subject to various artifacts .
an immediate suspect is that the appearance of the hot component in shell c is an artifact caused by insufficient background subtraction .
however , this would not explain the presence of the cool component in shell p. furthermore , the constant spectral shape in @xmath124 ( [ subsub : epic_spec ] ; figure [ fig : rawpspec ] ) argues against this possibility . for a more quantitative examination , we remade the deprojected spectra with the background artificially under - subtracted , by 3% which is comparable to the background systematic errors estimated in [ sub : epic ] .
in fact , this little affected the results obtained so far : a comparison of the remade shell c spectra with the 2p synthetic spectra ( [ subsub : thick_shell_2p ] ) gives nearly the same @xmath115 as before , changing only by @xmath67 2 .
thus , our results are robust against the background uncertainty .
another concern is possible deviations from the so - far assumed spherically symmetry .
indeed , the icm temperature of the centaurus cluster exhibits significant non - axi - symmetric distributions @xcite .
then , even if the icm were in a 1 t condition , shell c would sample a wider range of temperatures , to mimic a 2 t condition .
however , as far as we consider an annular region of @xmath125 that is crucial to this issue , the icm temperature ( approximated as 1 t ) is azimuthally constant typically within @xmath126 @xcite . to examine this issue using the present data , we derived a temperature map , presented in figure [ fig : tmap ] , by analyzing the projected epic spectra in four sectors by a single - temperature model .
it reconfirms a mild temperature decrease from northwest to southeast directions , with an amplitude comparable to those measured previously . according to a simple estimation , a temperature variation of this amount is insufficient , by at least a factor of two , to explain the difference between the two values of @xmath106 kev and @xmath127 kev , obtained via the 2 t fit to the shell c spectrum and its synthetic 1 t data , respectively .
thus , the preference for the 2 t view remains intact .
the finite angular resolution of the _ xmm - newton _ x - ray telescope could also blur the shell c boundary , and cause contaminations by signals from the outer hotter region .
however , shell c is much thicker ( @xmath128 ) than the blurring width ( @xmath129 ) , and the surface brightness decreases outwards . as a result
, the contamination of the outer hotter emission into shell c is estimated to be at most @xmath104 , which is within the renormalized systematic uncertainty of 5% introduced in [ subsub : renormalize ] . in the deprojection process ( [ subsub : depro ] ) , we assumed the region outside @xmath50 ( 150 kpc ) to have a constant spectrum that is identical to that measured just inside it .
this is based on the fact that the azimuthally - averaged icm temperature of this cluster , measured with the _ asca _ gis utilizing its wide field of view and low background , is essentially constant at @xmath130 kev from @xmath131 up to @xmath132 ( 0.5 mpc ; * ? ? ?
* ) . of course ,
deviations from this assumption would in principle affect , via eq.([eq : deprojection ] ) , the deprojected spectrum in every shell .
however , our numerical estimate in appendix a indicates that the emission from the @xmath133 region contributes only @xmath134 to the shell c spectrum , and hence any realistic error associated to it is considered negligible .
furthermore , the icm temperature has been confirmed in many clusters to decrease beyond @xmath135 times the virial radius ( e , g .
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , or @xmath136 mpc in the present case .
then , we are likely to be overestimating the outermost icm temperature , and hence over - subtracting the hot ( 3.8 kev ) emission as foreground and background contributions .
this makes our argument simply more conservative .
a more difficult issue is asymmetry along the line of sight .
if the object has a prolate shape in the depth direction , the deprojected shell c spectra would sample effectively outer ( hence hotter ) regions , and would involve the 3.8 kev component even if it is in a 1p condition . similarly , an oblate 1p condition would make shell p sample the cool emission .
however , neither case can explain the two thick shells simultaneously . finally in [ subsub : thick_shell_spec ] ,
possible spectral changes within each thin shell was already shown to become negligible through the construction of thick shells . as an overall confirmation to address this issue , in appendix b
we have numerically simulated a 1p cluster , and analyzed the fake data in the same manner as for the actual data .
then , the simulated shell c spectra have been reproduced by the 1p view , without demanding the 2p formalism . as a logical contraposition , the preference of the 2p modeling by the actual data demands the actual object not to be in the 1p condition . from these examinations , we conclude that the data favor the presence of the hot @xmath92 kev component in the 3-dimensional core region ( @xmath95 ) , and the cool @xmath137 kev component in the peripheral region ( @xmath118 ) . in other word , the 2p picture is considered to describes the actual physical state better than the 1p view . at the center of the centaurus cluster ,
some previous works reported the presence of excess x - ray absorption , up to a few times @xmath138 cm@xmath13 ( depending on the modeling ) , above the galactic column @xcite ; this was taken as evidence for the cf .
however , the effect was not confirmed in other works @xcite .
likewise , we have here experienced no problems in our spectral analysis with the absorption fixed at the galactic value .
generally , a somewhat higher absorption is often derived when a two- ( or multi- ) temperature emission is fitted with a 1p model ( e.g. , * ? ? ?
. however , some of the previously reported excess column densities in the centaurus cluster considerably exceed such modeling uncertainties .
we examined this issue using the non - deprojected and deprojected spectra from regions within @xmath23 and @xmath140 of the center , respectively .
specifically , we refitted these epic data using the 2p plus 0.7 kev model , applying an intrinsic absorption factor to both components in addition to that due to the galactic column .
( the hot component can be neglect here . ) as a result , the fits were little improved in either case .
the obtained _ excess _ column density in the rest frame is @xmath141 @xmath7 for the non - deprojected spectra within @xmath23 , and @xmath142 @xmath7 for the deprojected spectra within @xmath140 .
these are significantly lower than the previously claimed values , @xmath143 @xmath7 , and the case of the deprojected spectrum is consistent with no excess absorption .
then , what is responsible for this discrepancy ? regardless of the modeling ( 1p or 2p ) , the _ xmm - newton _ data reveal that oxygen in the central region is deficient relative to iron and silicon ( [ subsub : ana1p ] , [ subsub : ana2p ] ) .
nevertheless , the previous works usually assumed metals in the icm to obey the solar abundance ratios , simply because of inadequate energy resolution .
this presumably caused data deficits around the oxygen line ( @xmath144 kev ) relative to the model predictions , and forced previous investigators to introduce an artificial excess absorption in an attempt to suppress the over - predicted o - k line flux in their models .
actually , if we constrain the icm to follow the solar abundance ratios , the present data clearly require an excess absorption by @xmath145 @xmath7 , although the fit becomes worse ( with @xmath146 ) compared to the case of the galactic absorption and free abundance ratios ( @xmath147 ) .
thus , the reported excess absorption in the centaurus cluster is considered an artifact , caused mainly by the non - solar abundance ratios around the cd galaxy .
because the surface brightness of the centaurus cluster is strongly peaked at the center , we can utilized the rgs data to better constrain the temperature structure of the icm in the cd region , although the available spatial information is limited to 1-dimensional projections .
we extracted the first and second order spectra from rgs1 and rgs2 , over a strip of @xmath148 in width centered on the ngc 4696 nucleus .
the cross - dispersion direction is at a position angle of @xmath149 , so that the data accumulation strip partially covers the plume .
the background was subtracted as described in
[ sub : rgsana ] .
figure [ fig : rgs]a shows the obtained rgs spectrum , which combines the 1st and 2nd order spectra from the two rgs units . although the energy resolution is somewhat degraded due to the finite angular extent of the source , we clearly observe many atomic emission lines . in particular , the spectrum bears a strong k - line at @xmath150 from h - like oxygen ions ( oviii ) , but lacks those from he - like ones ( ovii ) which would be emitted strongly at about 22 by plasmas with a temperature lower than 0.3 kev .
this indicates that the cluster core region is devoid of such very cool plasmas , contrary to the prediction by the cf hypothesis .
since we already know that at least two plasma temperatures ( 0.7 kev and 1.72.0 kev ) are required to reproduce the emission from the cd region , we fitted the rgs spectra with a simple 2p plasma emission model ; we hereafter call this particular modeling `` quasi-2p fit '' , because it indeed invokes two apec components but the employed temperatures are different from those composing the 2p model employed in
[ subsub : ana2p ] and [ sub : thick_shell ] .
the four spectra ( 1st- and 2nd - order spectra from rgs1 and rgs2 ) were prepared separately , and fitted jointly .
we neglected for the moment the hot ( 3.8 kev ) component , because its high temperature and large angular extent make low - energy lines rather weak , and hence its parameters are difficult to constrain with the rgs data .
the abundances of o , ne , mg , si , fe , and ni were left free , but constrained to be common between the two components .
after the epic results @xcite , we fixed the s , ar , and ca abundances at 1.7 , 1.5 , and 2.4 solar , respectively , because k - lines of these elements are outside the rgs wavelength range .
the absorption is assumed to be galactic , as confirmed in [ sub : excess ] . to deal with the spatial extent , we blurred each spectral component with a gaussian of a free width @xmath151 .
the quasi-2p fit to the rgs spectra is shown in panels ( b ) and ( c ) of figure [ fig : rgs ] , and the obtained parameters are summarized in table [ tab : rgsfit ] .
thus , the overall emission - line features have been reproduced to a reasonable extent .
although the fit is not yet fully acceptable , this is mainly due to some unessential factors , such as our simple spectral blurring , the projection effects , and calibration uncertainties .
the obtained two temperatures , 0.8 kev and 1.7 kev , agree well with the lowest two characteristic values measured with the epic from the cd region .
the gaussian widths of the 0.8 kev and 1.7 kev components , @xmath152 and @xmath153 respectively , can be understood as reflecting the angular extent of the @xmath2 kev and @xmath154 kev component detected with the epic .
the rgs - determined abundances of mg , si , and fe are lower by a factor of @xmath36 than the values obtained with the epic data .
the discrepancy is at least partially relaxed by considering the hot ( 3.8 kev ) component , which does not emit prominent lines , but is estimated to contribute to the continuum up to 30% and reduce equivalent widths of the lines by that amount .
the remaining abundance discrepancy could be due to the abundance drop at the very center @xcite . in any case
, the temperatures , determined primarily by the line intensity ratios , are not affected .
the rgs spectra may be examined for the presence of other temperature components , besides the @xmath155 kev and 1.7 kev ones constituting the quasi-2p fit .
the rgs spectra were therefore analyzed with the multi - temperature fit procedure of @xcite , which is a kind of differential emission measure analysis like those in @xcite and @xcite .
specifically , we prepared seven plasma emission components of which the temperatures are given as @xmath156 , @xmath157 , @xmath158 , .. , and @xmath159 , with @xmath156 being the lowest temperature .
we then allowed to vary the normalizations of the seven components , as well as @xmath156 to which the remaining 6 temperatures scale .
the abundances of o , ne , mg , si , fe , and ni were left free , but constrained to be common among the seven components . each spectral component was blurred with a gaussian , of which the width is left free but constrained to be less than @xmath160 as obtained in the previous fit ( table [ tab : rgsfit ] ) .
as shown in figure [ fig : rgsdem ] , the best fit model obtained from this analysis is physically very close to that obtained with the quasi-2p model .
the fit has been improved little ( @xmath161 ; cf .
table [ tab : rgsfit ] ) by considering extra temperature components , with only the two components ( @xmath155 kev and 1.7 kev ) remaining significant .
note that these two temperatures are constrained to have a ratio of 1.5@xmath162 , while their absolute values are left free .
the 1.1 kev component has turned out to be weak in contrast to the two adjacent ones , implying that the significantly detected two components are discrete entities rather than representing a continuous temperature distribution .
the 2.55 kev component is not significant , either .
the rgs results constrain the 3.8 ( = @xmath163 ) kev component to have an emission measure which is @xmath164 of that of the 1.7 kev component . employing eq.([eq : filling_factor ] ) , the volume filling factor of this component is then estimated as @xmath165^{-1 } = 0.6 $ ] ( or equivalently @xmath166 ) .
this is consistent with figure [ fig : vfill ] .
figure [ fig : rgsdem ] also reconfirms that any cool emission below 0.7 kev is insignificant .
the upper limits are by more than an order of magnitude below the prediction of the isobaric cf model @xcite , if the model normalization is adjusted to the data at 1.7 kev .
therefore , we conclude that the icm around ngc 4696 is not cooling to temperatures much below @xmath2 kev . analyzing the rgs data from the present and an additional 110 ks observation
, @xcite also confirmed that the emission measure from the central region decreases more steeply toward lower temperatures than in a simple cf picture .
we analyzed the _ xmm - newton _ epic data of the central region ( @xmath167 or @xmath168 kpc ) of the centaurus cluster , assuming a spherical symmetry .
the icm has been characterized by three representative temperatures ; @xmath169 kev , @xmath170 found within @xmath93 ( 75 kpc ) , and @xmath2 kev localized within @xmath171 ( 12 kpc ) of the center .
the values of @xmath11 and @xmath9 are consistent with the _ asca _ measurements @xcite , and satisfy eq.([eq : tcth ] ) .
similarly , the detection of the 0.7 kev component reconfirms the _ chandra _ result @xcite .
the rgs data from the core region reinforced the presence of these discrete temperatures , and gave tight upper limits to emission with a temperature of @xmath172 kev .
as far as the deprojected thin - shell spectra are individually analyzed , the data had no preference between the 1p and 2p approaches .
however , the two sets of epic spectra from 3-dimensional thicker shells have revealed that the hot component is present even in a very inner ( @xmath20 ) region , and the cool component protrudes out beyond @xmath173 ( [ sub : thick_shell ] ) .
the rgs results from the cd region ( [ sub : rgsresult ] ) provides an important support to this view .
@xcite found that _ xmm - newton _ spectra of the central regions of abell 1795 also prefer the 2p formalism to the 1p modeling . in principle , the icm could be in multi - temperature conditions @xcite .
since a spectrum resulting from such a superposition of a range of temperatures is rather insensitive to the way of superposing them @xcite , an apparently 2p plasma may well be in a multi - temperature condition , and vise versa , particularly when the energy resolution is not too high .
indeed , the measured value of @xmath11 appears to distribute from 1.7 to 2.0 kev , depending on the radius . nevertheless , the rgs data preferred the presence of discrete temperatures , to a continuous temperature distribution ( [ sub : rgsresult ] ) .
therefore , a multi - phase modeling of the icm is considered inappropriate , at least the cd region of the centaurus cluster is concerned .
based on these results , we suggest that the plasmas in the central @xmath94 kpc of the centaurus cluster are better envisaged by invoking a few ( two to three ) discrete temperature components , rather than assuming a continuous radial temperature gradient or a condition of differential emission measure .
these results support and extend our view of cd clusters developed with _ asca _
( paper i ) .
below , we adopt the 2p view as our working hypothesis .
when integrated up to @xmath174 ( 150 kpc ) , the hot component , the cool component , and the 0.7 kev one have bolometric luminosities of @xmath175 erg s@xmath13 , @xmath176 erg s@xmath13 , and @xmath177 erg s@xmath13 , respectively . according to @xcite , @xmath178 increases by a factor of 2 if integrated further out to @xmath179 ( 375 kpc ) .
we have determined the radial abundance profiles of fe , si and o in the icm , assuming that the different plasma components share the same abundances at a given radius .
those of the other elements are much more uncertain @xcite . as first discovered by @xcite , and
already reported by @xcite by analyzing the present _ xmm - newton _ data , the fe and si abundances have been reconfirmed to increase significantly toward the center ( figure [ fig:2prad ] ) , and stay rather high even at the very center . the si / fe abundance ratio at the center , about 1.3 ( figure [ fig:2prad ] ) , agrees with the _ asca _ measurements @xcite .
in contrast to the behavior of fe and si , the o abundance is radially more constant , implying a decrease in the o / fe ratio toward the center , in agreement with @xcite . as argued by various authors ( paper i ;
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , this can naturally be understood as a result of varying contributions from type ia and type ii suparnovae : the outer - region icm is contributed significantly by oxygen - rich products from type ii supernovae , while that in the inner region by iron - rich products from type ia supernovae that occurred in the cd galaxy .
so far , a fair number of other clusters have been known to exhibit abundance profiles similar to those of the centaurus cluster , with a central increase of iron and silicon , and a flat distribution of oxygen ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these include in particular recent _ suzaku _ observations of the fornax cluster @xcite , abell 1060 @xcite , and awm 7 @xcite .
therefore , the relative oxygen deficit in the cluter core region may be a common phenomenon .
the present results argue against the cf scenario from several independent aspects .
these include ; the lack of cooling component ( [ sub : rgsresult ] ) ; the lack of excess absorption ( [ sub : excess ] ) ; and the radial changes in the abundance ratios ( [ subsub : abundances ] ) which rule out large - scale inflows of the icm .
below , we discuss the first two issues in some more details .
the cf scenario relates the bolometric luminosity , @xmath180 , of a `` cooling '' portion of icm with the mass deposition rate , @xmath181 , as @xcite @xmath182 where @xmath183 is the mean molecular weight and @xmath184 is the proton mass . by substituting the measured value of @xmath185 erg s@xmath13 ( 5.1 ) into @xmath180
, we obtain @xmath186 yr@xmath13 .
this in itself is consistent with the previous results from _ rosat _ @xcite and _ asca _ ( paper i ) .
however , we have found no evidence of emission from any plasma component with a temperature @xmath187 kev , with upper limits much tighter than the cf - predicted emission measure ( figure [ fig : rgsdem ] ) .
therefore , the cf interpretation fails to explain the strong cool emission of the centaurus cluster , as well as in other clusters ( paper i ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the value of @xmath181 should not b e taken as an actual mass - deposition rate .
we have shown in [ sub : excess ] that the previously reported excess absorption , at the center of the centaurus cluster , is an artifact caused by the relative oxygen deficit .
in fact , when the non - solar abundance ratios are properly considered , the absorption became consistent with being galactic only , with any additional column at most several times @xmath188 @xmath7 .
as discussed in
[ subsub : abundances ] , the relative oxygen depletion in the cluster center appears to be a common feature .
then , the excess absorption , claimed previously in many other clusters ( e.g. * ? ? ? * ) , will be explained away as well . indeed , using the rgs spectra , @xcite also confirmed the lack of excess absorption from several other clusters .
given these results , there is practically no evidence for the x - ray absorbing materials at least in nearby well - studied clusters .
let us briefly consider the nature of the 0.7 kev component , which is localized to the cd galaxy , ngc 4696 .
similar cool emission comonents are seen around other cd galaxies , including m87 in particular , but their interpretation has generally been unsettled ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
judging from the angular extent ( @xmath3 kpc ) and the bolometric luminosity ( @xmath189 erg s@xmath13 ) , this 0.7 kev component could be hot interstellar medium ( ism ) associated with the cd galaxy .
its temperature is consistent with the stellar velocity dispersion of ngc 4696 , 274 km s@xmath13 .
the plasma emitting this component ( including both the central concentration and the plume ) has an estimated mass of a few times @xmath190 @xcite , which is also consistent with this ism interpretation .
although the estimated cooling time of this plasma is as short as @xmath5 yr @xcite , the radiative energy loss could be supported by such processes in ngc 4696 as the supernova heating ( a few times @xmath191 erg s@xmath13 ) , and the accumulation of stellar mass loss which is dynamically `` hot '' on the galaxy scale .
in contrast to the above arguments , the complex filamentary morphology of this component @xcite , including the plume like elongated tail structure , can not easily be reconciled with the hot ism interpretation , even considering interactions of the cd galaxy with the icm .
similar problems with the cool gas in m87 ( the virgo cd galaxy ) were pointed out by @xcite , who showed its close association with the synchrotron radio arms .
furthermore , as exploited in paper i , the hot ism interpretation could be applied better to the cool ( 2 kev ) component . considering these
, the 0.7 kev plasma could alternatively be associated with some magnetic structures around the cd galaxy , which in turn could be a result of past nuclear activity in ngc 4696 .
further discussion of this issue is beyond the scope of the present paper .
the successful 2p picture ( [ sub : thick_shell ] , [ subsub:2tvs1 t ] ) and the absence of significant cfs ( [ subsub : anticf ] ) in the centaurus cluster suggest that its core region ( within @xmath94 kpc ) hosts a cool plasma phase which is stably intermixed with the hot phase .
while this reconfirms a prediction made in paper i , it poses a series of questions to be answered ; ( i ) how such a 2p configuration is realized ; ( ii ) how the cool phase is heated against the radiative cooling ; ( iii ) how the heating and cooling be balanced in a thermally stable manner ; ( iv ) what determines the value of @xmath11 ; and ( v ) what produces the scaling of eq.([eq : tcth ] ) .
below , we examine them one by one .
an immediate issue associated with the 2p scenario is ( i ) how to realize such a configuration .
we may suppose that the two phases are intermixed on a typical length scale of @xmath192 kpc , as indicated by various sharp structures seen in x - ray images of some clusters @xcite , including the `` cold front '' structure @xcite .
then , employing the classical spitzer heat conductivity , @xmath193 where @xmath194 is the temperature in kelvin , the time scale of conductive heat exchange between the two phases would be very short , as @xmath195 @xcite .
the two phases would quickly become isothermal , unless the heat conduction between them is suppressed @xcite .
the most natural way to suppress the heat conduction is to invoke magnetic fields , as argued , e.g. , by @xcite to explain the cool plasma in m87 .
in fact , the icm is known to be generally magnetized to a few @xmath183 g , and the aforementioned sharp x - ray features themselves are indicative of ordered strong ( @xmath196 g ) magnetic fields @xcite . in the particular case of the centaurus cluster , a magnetic field strength up to @xmath197 g is reported from radio observations @xcite .
although these field strengths are not necessarily higher than the equipartition value ( @xmath198 g ) , they are more than sufficient to completely suppress the heat transport across field lines , because the gyration radius of 4 kev thermal electrons in a 1 @xmath183 g field , @xmath5 cm , is more than 10 orders of magnitude smaller than their classical mean free path for coulomb scattering which also determines eq.([eq : conduction ] ) .
in contrast , the heat conductivity along field lines may well be approximated by the spitzer value , eq.([eq : conduction ] ) , unless the field lines are highly tangled @xcite .
based on solar analogy , magnetic field lines around a cd galaxy may be classified grossly into `` open '' and `` closed '' ones @xcite , corresponding to coronal holes and coronae , respectively .
this view relies only on a general and simple classification of the magnetic field - line topology , without invoking any particular magnetic configuration .
then , open - field regions , connected to the outer cluster volume , must be filled with the hot - phase icm , and kept nearly isothermal by the efficient field - aligned heat conduction of eq.([eq : conduction ] ) .
in contrast , closed - field domains , connected to the cd galaxy and thermally insulated from the surrounding hot phase , can take independent ( possibly lower ) temperatures of its own .
in addition , the loop interior plasma is expected to be metal enriched by type ia supernovae in the cd galaxy . from these considerations , we speculate that the plasma , filling numerous magnetic loops anchored to the cd galaxy , is observed as the cool phase of the centaurus cluster ( and of other similar objects ) . in other words ,
the cool phase may be regarded as a `` magnetosphere '' associated with the cd galaxy ( paper i ; * ? ? ?
the magnetic fields , while working as a thermal insulator between the two phases , need not to be as strong as the equipartition value , since the loop - interior plasma is confined primarily by external pressure from the hot phase .
this `` magnetosphere '' picture was already invoked successfully in paper i to explain the _ asca _ results . indeed , h@xmath199-emitting filamentary features
are optically observed in central regions of some clusters on a spatial scale of a few to few tens kpc ( e.g. , * ? ? ?
* ) , and the brightest of them can even be identified in some _
chandra _ x - ray images ( e.g. * ? ? ?
the best example is the recent _ chandra _ result on m87 @xcite , which reveals rich filamentary structures with a typical length of 10 to 50 kpc . in the centaurus cluster as well , regions of strong magnetic fields are reported to be associated with filamentary structures that emit h@xmath199 photons and enhanced soft x - rays @xcite . while these filaments could be limb - brightened edges of bubbles blown by the active galactic nucleus of m87 @xcite , they can alternatively trace magnetic structures such as assumed there .
in addition to these x - ray and radio results , magnetohydrodynamic numerical simulations by @xcite , including both the radiative cooling and the field - alighned heat conduction , show the emergence of low - temperature regions along the loop - shaped magnetic field lines . from these arguments , we believe we have found a viable answer to the issue ( i ) raised at the beginning of the present section .
we may assume that the cluster core region is fed with a constant heating lumiosity @xmath200 , via some mechanism to be specified later in [ sub : heating ] .
in such a case , the volume heating rate will be generally proportional to the local icm density , while the volume emissivity is obviously proportional to the density squared .
therefore , denser regions would preferentially cool and become even denser , leading to a thermal instability : this is the issue ( iii ) . if , however , adopting the magnetosphere picture , the cool phase can be thermally stabilized by a built - in feedback mechanism , hereafter called rtv mechanism , originally developed by @xcite to explain quiet solar coronae .
the rtv mechanism holds for a thin magnetic flux tube hydrostatically immersed in an external pressure , with the two loop footpoints anchored to a cool matter reservoir .
the loop - interior plasma is assumed to be heated with a constant luminosity @xmath200 , either uniformly along the loop or at the loop summit .
the deposited heat @xmath200 is assumed to flow toward the footpoints due to loop - aligned conduction of eq.([eq : conduction ] ) , and ultimately radiated away from various heights of the loop .
suppose that @xmath200 was initially balanced by the radiative cooling , but some perturbation caused @xmath200 to decrease .
then , the loop loses its internal pressure , and becomes thinner under the external compression , thus reducing the conductive heat flow along it . at the same time , some portion of the plasma flows into the matter reservoir , and the consequent decrease in the emission integral reduces the radiative energy loss .
these concordant effects bring the loop into a new steady state corresponding to a lower value of @xmath200 .
thus , the rtv mechanism thermally stabilizes the cool phase , and gives an answer to the issue ( iii ) .
following the original work by @xcite , @xcite used the _ yohkoh _ spacecraft to examin the rtv mechanism .
they confirmed that it is likely to be actually working in quiescent solar coronae , which keep emitting x - rays in an apparently steady manner for a time much longer than the nominal cooling time ( @xmath201 minutes ) .
the rtv mechanism further allows us to answer ( iv ) , namely , to explain the absolute values of @xmath11 .
let us assume that the matter reservoir at the footpoints is much cooler than the loop interior , so that heat inflow to the reservoir can be neglected .
under such boundary conditions , the temperature of the loop - confined plasma is shown to take the maximum value of @xmath202 at the loop top @xcite . here , @xmath203 ( cm ) is the loop half length , and @xmath204 is the loop - confining external pressure in units of dyn @xmath7 .
this is so - called rtv temperature scaling , which is actually confirmed in quiescent solar coronae @xcite .
the range of the coefficient , 1.1 to 1.4 , reflects differences in the assumed heat input distribution along the loop .
the index of scaling in eq.([eq : rtv1 ] ) depends weakly on the form of the cooling function @xcite ; the particular value of 1/3 is valid over a temperature range of 0.52 kev and near - solar abundances .
most importantly , @xmath205 in eq.([eq : rtv1 ] ) is determined solely by the product @xmath206 , without depending on @xmath200 .
the loop - ineterior density @xmath207 becomes also independent of @xmath200 , because @xmath208 is assumed to be constant .
changes in @xmath200 affect only the loop cross section @xmath209 , and hence the cool - phase x - ray luminosity which is by definition equal to @xmath200 .
more specifically , eq.([eq : conduction ] ) and eq.([eq : rtv1 ] ) , together with @xmath210 , can be combined into another scaling relation as @xmath211 thus , the loop becomes thicker under higher values of @xmath200 .
the loop also gets thicker if we decrease @xmath204 while keeping @xmath200 constant , but in this case , @xmath205 automatically decreases via eq.([eq : rtv1 ] ) , so as to reduce the conductive heat flux ( per unit cross section ) @xmath212 and hence keep @xmath200 unchanged . in the present case of centaurus , we may identify @xmath204 with the hot - phase pressure in the core region , namely @xmath213 dyn @xmath7 with @xmath214 @xmath215 @xcite , and rewrite eq.([eq : rtv1 ] ) as @xmath216^{1/3 } ~~{\rm kev}~. \label{eq : rtv2}\ ] ] here , we assumed that loop semi - lenght to be comparable to the narrower core radius , @xmath201 kpc , of the double-@xmath54 modeling for the gravitational potential in the centaurus cluster @xcite .
this is because each magnetic loop , with its interior having a higher density than the surrounding hot phase , may not become taller than the flat core radius of the central gravitational potential .
the assumed length is comparable to those of the soft x - ray filaments in m87 @xcite .
when such a loop is observed as an integrated entity , we expect to measure an x - ray temperature averaged over the loop length .
assuming the footpoint temperature to be 0.7 kev ( [ sub:0.7kev ] ) , a simple numerical solution to the rtv equation predicts this average to be @xmath217 , or @xmath218 kev from eq.([eq : rtv2 ] ) .
this is in an excellent agreement with @xmath11 actually measured in the centaurus cluster , giving an answer to the issue ( iv ) . in reality ,
the loop length may well scatter , with longer loops tending to be hotter according to eq.([eq : rtv2 ] ) .
then , we expect the spherically averaged loop temperature to slightly increase outward , because each loop is hottest at its top region , and a larger radii would sample longer loops .
this agrees with what we actually observe in figure [ fig:2prad ] . taking it for granted that @xmath11 can be identified with @xmath217 ,
let us consider the issue ( iv ) , namely , to explain the scaling of eq.([eq : tcth ] ) .
the luminosity of the hot phase in a cluster is empirically known to scale with its temperature as @xmath219 ( e.g. , * ? ? ?
* ) . on the other hand , we generally have @xmath220 , where @xmath221 is a typical radius of the hot phase .
elimination of @xmath222 from these two equations gives @xmath223 , and hence @xmath224 . substituting this into eq.([eq : rtv2 ] ) , we obtain @xmath225 assuming that @xmath226 is relatively constant , @xmath11 indeed becomes roughly proportional to @xmath9 , providing an explanation to eq.([eq : tcth ] ) .
as shown so far , the rtv mechanism can account for the stable coexistence of the cool and hot phases in the cluster core region .
the scenario is independent of the amount of available heat input , because a lower value of @xmath200 simply reduces the cool - phase luminosity .
furthermore , the rtv scaling can quantitatively explain the absolute values of @xmath11 , as well as its ( near ) proportionality to @xmath9 expressed by eq.([eq : tcth ] ) .
thus , three of the four issues raised at the beginning of [ sub : magnetosphere ] have been answered .
we therefore propose the magnetosphere model , incorporating the rtv mechanism , as a promising hypothesis to explain the plasma physics in the central regions of cd clusters in general . to complete our new viewpoint
, we may ask ourselves why the bright cool emission is observed predominantly around cd ( or xd ) galaxies ( * ? ? ? * figure 1 of paper i ) , including ngc 4649 .
conversly , non - cd galaxies , like ngc 4472 in the virgo cluster , generally lack such bright cool x - ray emission , even if they are optically as luminous as their cd counterparts .
an outstanding characteristic of a cd galaxy is that it is nearly in a standstill at the bottom of the gravitational potential .
then , we may speculate that a substantial magnetosphere can develop around a galaxy in a cluster environment only when it is at rest with respect to the icm .
when a galaxy is moving through the icm , its magnetosphere will be disrupted by the icm ram pressure , which can be one to two orders of magnitude higher than the magnetic pressure of the assumed magnetosphere ( paper i ) .
such strong interactions are actually revealed by _
observations in nearby clusters ( e.g. , * ? ? ?
* ; * ? ? ?
furthermore , optical observations of distant clusters reveal numerous blue spiral galaxies with distorted morphology ( e.g. , * ? ? ?
* ) , which we suggest to be interacting with the icm .
having answered four out of the 5 questions raised at the beginning of [ sub : magnetosphere ] , the only remaining issue is ( ii ) , i.e. , to seek for possible heating mechanisms to sustain the cool - phase luminosity . in the case of the centaurus cluster ,
the target value is @xmath227 erg s@xmath13 .
after makishima ( 1997 ) , paper i , and the discussion conducted so far , we retain our magnetosphere viewpoint . the argument in
[ subsub : interaction ] indicates that moving galaxies in a cluster interact with the icm , and transfer their kinetic energies to the icm .
this process is estimated to proceed at a rate of @xmath228 @xcite , where @xmath49 is the number of moving galaxies , @xmath229 is the average icm number density , @xmath230 is the average galaxy velocity , and @xmath231 is the interaction radius for each galaxy .
we may use @xmath232 km s@xmath13 which is @xmath233 times the velocity dispersion of the centaurus cluster ( 586 km s@xmath13 ) after @xcite , @xmath234 @xmath215 which is an average over the central 300 kpc @xcite , and @xmath235 .
furthermore , a relatively large value of @xmath231 , say , @xmath25 kpc , is suggested by the argument in [ subsub : interaction ] , and by some x - ray observations @xcite . through eq.([eq : interaction ] ) , these numbers yield @xmath236 erg s@xmath13 , which well exceeds @xmath237
erg s@xmath13 .
therefore , the galaxy - icm interaction becomes a promising candidate for the icm heating , and for the suppression of cfs .
these effects have been reproduced successfully by numerical simulations @xcite . how large is the kinetic energy @xmath238 stored in the moving galaxies ?
the stellar mass of the centaurus cluster within the central 300 kpc , but excluding the cd galaxy , is estimated as @xmath239 @xcite after correcting for the different values of the hubble constant .
we have then @xmath240 ergs , which is likely to be a lower limit because the employed galaxy mass assumes a conservative mass - to - light ratio of 8 . as a result
, the time scale of the energy tansfer from the moving galaxies to the icm is estimated as @xmath241 yr .
therefore , the kinetic energy in the moving galaxies is large enough to sustain @xmath242 over the hubble time .
the galaxy - to - icm energy transfer is expected to cause galaxies to gradually fall to the potential center .
there are several pieces of observational evidence suggesting that this effect is actually taking place .
in fact , the stellar - mass distributions in many clusters are known to be much more centrally peaked than that of the icm . furthermore ,
clear central decreases in `` iron - mass to light ratio '' , observed from a fair number of clusters and groups ( paper i ; * ? ? ?
* ; * ? ? ?
* ) , suggest that galaxies used to be distributed more widely than today , and polluted the icm with metals out to the cluster edges .
the energy released by the moving galaxies may be stored once in the icm in the form of turbulence and bulk flows . however , according to the _ suzaku _ observation @xcite , these kinetic energies in the icm of the centaurus cluster , as estimated from the width and position - dependent centroid shifts of the fe - k line , can not largely exceed its thermal energy . therefore , the turbulent and bulk - flow energies must be dissipated and thermalized efficiently in the icm , with a rate comparable to @xmath242 . if a fair fraction of the dissipated energy is deposited onto the magnetosphere , the available heating luminosity of the cool phase becomes of the order of @xmath243 erg s@xmath13 which can just sustain the observed @xmath244 .
we would rather say that @xmath244 is self - adjusted , via the rtv mechanism , to match the predetermined energy dissipation rate .
the remaining portion of @xmath242 will heat the hot phase , to make it spatially more extended than the total gravitating mass .
the actual dissipation of turbulence and bulk - flow energies may take place through magnetohydrodynamic processes .
for example , moving galaxies will pick up magnetic field lines in the icm and stretch them , exciting alfvn waves and causing field lines to reconnect .
when the magnetic reconnection occurs between two closed magnetic loops , the released energy will be spent in heating the cool phase .
that between closed and open field lines will expel a small potion of the metal - rich cool - phase plasma into the hot phase .
the ejected plasmoid will become quickly isothermal with the hot phase , in terms of the electron temperature via eq.([eq : conduction ] ) , as well as in terms of the ion ionization temperature ( in a few times @xmath245 yr ; * ? ? ?
the resulting highly ionized heavy ions will remain there due to their slow diffusion @xcite , and increase the metallicity of the hot phase .
this can explain why the central metallicity increase is observed in both phases .
in addition to the above mechanism , there can be another icm heating mechanism localized around each cd galaxy ; namely , a shrink of its self - gravitating core .
since this has already been discussed in paper i , we simply mention that this mechanism can also account for a heating luminosity of @xmath246 erg s@xmath13 .
analyzing the _ xmm - newton _ epic and rgs data of the centaurus cluster , we have obtained the following results . 1 .
the icm in the central @xmath94 kpc can be described better by the two - phase view ( plus the 0.7 kev component at @xmath247 kpc ) than by the single - phase picture .
2 . there is no evidence of cooling flows .
the iron and silicon abundances increases significantly in the central region , while that of oxygen is radially constant .
the previously reported excess x - ray absorption disappears when the central oxygen deficit is considered .
the overall results agree with those by @xcite who used _ asca _ and _ rosat_. to explain these observations , we have developed a working hypothesis ( after paper i ) , in which the cool phase is thought to constitute a magnetosphere associated with the cd galaxy . incorporating the rosner - tucker - vaiana mechanism , it can account for the following essential features of the centaurus cluster , as well as of similar cd clusters . 1 . the hot and cool icm phases co - exsit in the cluster core region .
the icm is provided with a heating luminosity , which is high enough to sustain the x - ray emission against the radiative cooling .
the cool phase is kept thermally stable .
4 . the cool phase has a typical temperature of @xmath36 kev , with a mild outward increase .
5 . a good proportionality holds between the cool and hot temperatures
the icm metallicity becomes enhanced in both phases toward the center .
the present results thus provide a new insight into the physics of clusters of galaxies .
acknowledgements : this work was supported in part by grant - in - aid for scientific research ( s ) ( japanese ministry of education , culture , sports , scince & technology ) # 18104004 onstudy of interactions between galaxies and intra - cluster plasmas " , as well as by special research project for basic science ( institute of physical and chemical research ) on investigation of spontaneously evolving systems " .
this appendix is meant to provide numerical details of the deprojection process described in [ subsub : depro ] .
let us consider 11 annular regions , of which the outer radii are at @xmath248 , @xmath249 , @xmath250 , @xmath148 , @xmath80 , @xmath128 , @xmath68 , @xmath78 , @xmath251 , @xmath252 , and @xmath50 , and the corresponding 11 shell regions .
these divisions are the same as used in figures [ fig : rawpspec ] , [ fig:1prad ] , and [ fig:2prad ] , except that the innermost 4 regions therein are here re - arranged into two regions just for clearer presentation .
the reconstructed `` shell c '' spectrum , namely @xmath253 , is not affected by this simplified treatment of inner annuli / shells .
let @xmath254 be the projected spectra from the 11 annular regions at a given energy , while @xmath255 those from the 11 shells .
note @xmath256 has the outer boundary at @xmath50 , while @xmath257 includes emission outside that .
then , let @xmath258 denote the projection matrix which converts @xmath259 into @xmath260 . in the present geometrical setting
, it is numerically given as @xmath261 on the assumption that the emission is uniform within each individual shell , these matrix elements are uniquely determined by the geometry alone , without any dependence on the emission model or instrumental effects .
the only exception is the last column , namely @xmath262 ( @xmath263 ) , which assumes the emission outside @xmath50 as explained in [ subsub : depro ] .
although the matrix elements are shown with only 3 digits below decimal points , this is again for simplicity ; the calculation is preformed with double precision . by inverting this matrix
, we can calculate the deprojection matrix @xmath57 defined in eq.([eq : deprojection ] ) .
numerically , it is calculated as @xmath264 from eq.([eq : depronum ] ) , the shell c spectrum @xmath265 at each energy can be expressed as @xmath266 thus , the outermost annulus @xmath256 , which is most subject to the background uncertainty as well as to that of the emission outside @xmath174 , contributes _ geometrically _ only less than 5% to the shell c spectrum , compared to the individual shells @xmath267 through @xmath268 .
in addition , the flux in @xmath256 is @xmath74 to @xmath201 times lower than those in the 3rd to 7th annuli ( figure [ fig : rawpspec ] ) . considering these
, we estimate that the emission from the @xmath133 region contributes _ physically _ by no more than @xmath134 to the shell c spectrum .
therefore , any systematic error associated with @xmath256 is concluded to be negligible ( [ subsub : artifacts ] ) .
another exercise to be conducted using eq.([eq : depronum ] ) is the issue raised in
[ subsub : thick_shell_spec ] , i.e. , how the assumption of emission constancy within each thin shell affects our shell c spectrum .
as mentionned there , let us examine what happens if we purposely neglected spectral differences between @xmath269 and @xmath270 ( the 3rd and 4th thin shell constructing shell c , respectively ) , where the 1p properties change rather steeply with @xmath14 .
taking the fe - k line equivalent width as an example ,
we may write the fluxes in the fe - k line energy region as @xmath271 and @xmath272 , where @xmath273 represents continuum and @xmath274 the line equivalent width . using the difference @xmath275 and the emission - measure - weighted mean @xmath276 , we may rewrite as @xmath277 $ ] and @xmath278 $ ] . substitution of this into eq.([eq : shellcnum ] ) readily yields @xmath279 where @xmath280 represents terms unrelated to @xmath269 or @xmath270 , while @xmath281 is a numerical factor .
thus , the negligence of @xmath96 between the two annular spectra has only an effect of at most @xmath98 in the shell c spectrum .
the essence is that the deprojection procedure has mathematically differential " nature exaggerating small - scale features , while this problem is largely removed by constructing a thick shell wherein a process of integration is involved .
this is evidenced by eq.([eq : shellcnum ] ) , where @xmath265 is close to a simple summation of @xmath267 through @xmath268 , except that @xmath267 must be given a 1.7 times larger weight and @xmath282 ( the shell just outside shell c ) must be subtracted with a relatively large weight of 1.5 .
in order to assess the validity of our thick shell " method , the following numerical simulation was performed .
we started from the best - fit 1p numerical models ( figure [ fig:1prad ] ) obtained for the 13 deprojected thin shell regions which are defined in figure [ fig : rawpspec ] .
then , to represent the radial spectral changes more smoothly , each of the 5 shells constituting shell c ( covering @xmath283 ) was subdivided into inner and outer halves , and appropriate temperature and emission measure were assigned to the inner / outer pair .
these model spectra , now comprising 18 shells , were projected numerically onto the corresponding 18 annuli , to form numerical spectral models for the 18 annular regions . at this stage
, the inner / outer pair were numerically recombined together .
the derived 13 model spectra were convolved with the epic energy responses , and were given poisson errors simulating the actual observation .
then , in the same manner as the actual data , we analyzed these 13 annular spectra via deprojection .
figure [ fig : fake1p ] shows the simulated shell c spectra , compared with the
sysnthetic 1p spectra " introduced in [ subsub : thick_shell_1p ] .
the comparison is fully satisfactory with @xmath284 . for reference ,
a 2p fit with two apec components , in the same manner as explained in
[ subsub : thick_shell_2p ] , gave @xmath285 , which is similarly acceptable .
in other words , the 1p and 2p interpretations degenerate unlike the case of the actual data . furthermore , unlike the 2p fit to the actual shell c spectra which revealed the @xmath286 kev component , the 2p fit to the simulated data yielded @xmath287 kev and @xmath288 kev , which are both consistent with the assumed 1p temperature range ( 2.03 to 3.30 kev ) .
lcc parameter & comp . 1 & comp . 2 + temperature & @xmath289 & @xmath290 + gaussian @xmath151 & @xmath291 & @xmath292 + + o & + ne & + mg & + si & + fe & + ni & + @xmath115/d.o.f . & + | x - ray data of the centaurus cluster , obtained with _ xmm - newton _ for 45 ksec , were analyzed .
deprojected epic spectra from concentric thin shell regions were reproduced equally well by a single - phase plasma emission model , or by a two - phase model developed by _
asca _ , both incorporating cool ( 1.72.0 kev ) and hot ( @xmath0 kev ) plasma temperatures .
however , epic spectra with higher statistics , accumulated over 3-dimentional thick shell regions , were reproduced better by the two - phase model than by the singe - phase one .
therefore , hot and cool plasma phases are inferred to co - exist in the cluster core region within @xmath1 kpc .
the iron and silicon abundances of the plasma were reconfirmed to increase significantly towards the center , while that of oxygen was consistent with being radially constant .
the implied non - solar abundance ratios explains away the previously reported excess x - ray absorption from the central region .
although an additional cool ( @xmath2 kev ) emission was detected within @xmath3 kpc of the center , the rgs data gave tight upper limits on any emission with a tempeartures below @xmath4 kev .
these results are compiled into a magnetosphere model , which interprets the cool phase as confined within closed magnetic loops anchored to the cd galaxy . when combined with so - called rosner - tucker - vaiana mechanism which applies to solar coronae
, this model can potentially explain basic properties of the cool phase , including its temperature and thermal stability . |
for many non - perturbative problems in quantum field theory , numerical simulations on the lattice offer the only known way to compute various quantities from first principles .
much progress has been made during recent years @xcite , e.g. in calculating the light hadron spectrum , the light quark masses , the running coupling constant @xmath0 or observables in heavy quark physics .
further - on , lattice simulations allow the study of phenomena , like chiral symmetry breaking and confinement , or of phase transitions in the early universe .
the computer resources required for such simulations are huge and critically depend on the physical parameters , like quark masses , and the formulation of the theory on the lattice , e.g. with improved chiral properties . to make the necessary resources available , various research groups engage in the development of massively parallel computers which are specifically optimized for this kind of applications .
one of these projects is ape @xcite which is currently developing its fourth generation of machines , , within the framework of an european collaboration by infn ( italy ) , desy ( germany ) and the university of paris sud ( france ) .
is designed as a massively parallel computer following the single program multiple data ( spmd ) programming model .
all machine functionalities , including the network and memory interfaces , are integrated into one single custom chip running at a clock frequency of 200 mhz .
the nodes run asynchronously and are implicitly synchronized by communications .
the processor core is a 64-bit architecture and all floating point operations use ieee double precision format
. a so - called _ normal _ operation @xmath1 , with @xmath2 , @xmath3 , and @xmath4 complex numbers , can be started at each clock cycle .
the peak performance of each node is therefore 1.6 gflops .
each node is a fully independent processor with 2561024 mbytes private memory ( standard ddr - sdram with ecc ) storing both program and data .
since has very long instruction words ( vliw ) , microcode de - compression and instruction buffers are used to reduce conflicts between data and instruction loading .
a large number of 256 @xmath5-bit registers allows efficient data re - use .
the high - performance communication network has a 3-d torus topology .
each link moves one byte per clock cycle and the startup latency of about 20 clock cycles , i.e. 100 ns , is very short .
communications and arithmetic operations can be carried out in parallel . by loading local or remote data through dedicated pre - fetch queues one can
almost perfectly exploit the concurrency between memory accesses , network transfers , and arithmetic operations . on an processing board
16 nodes are mounted , and 16 boards are interconnected on the backplane within a crate .
larger systems are assembled by connecting together several crates using external cables .
thus , a single rack system with 2 crates has 512 nodes and provides a peak performance of 0.8 tflops .
the footprint is about 1 m@xmath6 only and the estimated power consumption is @xmath7 kw . due to this very moderate power dissipation air cooling is still possible .
systems are accessed from a front - end pc via a custom designed host - interface pci - board .
it uses a simple i2c link for bootstrapping and basic operating system requests , while i / o operations are handled via high - speed data links . by connecting one or more processing boards of an machine with host - interface boards the overall i / o bandwidth
can be adjusted to the user needs .
two high - level programming languages , tao and c ( with suitable extensions for parallelisation ) are supported on .
after further optimisation of the intermediate assembly , the microcode is generated , scheduled and compressed .
the design of has been finished and prototype processors are to be delivered in autumn 2003 . all other hardware components exist and have been tested .
the final procurement costs are estimated to be 0.5 /mflops peak performance .
the full vhdl design of the machine has been used for benchmarking .
key lattice gauge theory kernels were measured to run at a sustained performance of @xmath8 or more .
while the design is optimized for simulating lattice qcd , we expect the architecture to be suitable also for other applications .
european research groups in lattice gauge theory aim for large dedicated installations of massively parallel computers during the next years . in edinburgh
( uk ) a 10 tflops qcdoc machine will be installed by 2004 and in italy a 10 tflops installation is planned . in germany researchers from various groups
joined the ( latfor ) initiative .
they have set up a broad research program @xcite , assessed the required computing resources , and proposed a 15 and 10 tflops installation of hosted by nic at desy ( zeuthen ) and gsi ( darmstadt ) .
99 for an overview see `` proceedings of the xxth international symposium on lattice field theory , '' nucl .
b ( proc . suppl . ) 119 ( 2003 ) .
r. alfieri _ et al . _
[ ape collaboration ] , arxiv : hep - lat/0102011 ; r. ammendola _ et al .
_ , arxiv : hep - lat/0211031 ; f. bodin _ et al .
_ , arxiv : hep - lat/0306018 .
r. alkofer _ et al . _
, http://www-zeuthen.desy.de/latfor . | we present the ( array processor experiment ) project for the development of dedicated parallel computers for numerical simulations in lattice gauge theories .
while is a production machine in today s physics simulations at various sites in europe , a new machine , , is currently being developed to provide multi - tflops computing performance . like previous machines ,
the new supercomputer is largely custom designed and specifically optimized for simulations of lattice qcd .
13 = 14.5pt |
the layered disc model was proposed by @xcite to account for accretion - related phenomena in t tauri stars .
he assumed that the angular momentum is transported by the magneto - rotational instability , commonly referred to as the mri @xcite .
however , in the outer disc ( beyond @xmath0 au ) the temperature and the ionization degree is so low that the gas is not well coupled to the magnetic field and the mri decays .
there , the only parts of the disc in which the mri can operate are the surface layers that are ionized by cosmic rays ( ionization due to x - ray quanta emitted by the central star was also considered ; see @xcite ) .
sandwiched between the active surface layers is an mri - free , and , consequently , non - viscous area near the mid - plane of the disc , commonly referred to as the dead zone .
an interesting property of layered discs is that , in general , the accretion rate @xmath1 is a function of the radius @xmath2 .
the specific form of this function depends on the mass - weighted opacity .
however , @xmath1 increases with @xmath2 if the opacity does not depend on the density and increases with the temperature not faster than @xmath3 .
this is true in the range @xmath4 k @xmath5 @xmath6 @xmath5 @xmath7 k , where the opacity is dominated by ice - free grains , and also between @xmath8 k and @xmath4 k , where the opacity drops due to the sublimation of ices as @xmath6 grows . at temperatures below @xmath8 k
the opacity varies as @xmath3 , and the accretion rate is constant as a function of @xmath2 ( in this paper we use opacities of gas and dust mixture according to @xcite ) . when @xmath1 increases with @xmath2 , an annulus centred at @xmath9 receives more mass per unit time from the outer disc @xmath10 than that it loses to the inner disc @xmath11 .
as a result , the mass accumulates in the dead zone .
eventually , at one or more locations in the dead zone the surface density becomes so large that the heat released by the accretion of the accumulated matter pushes the temperature to a level at which the collisional ionization can restore the coupling between the gas and the magnetic field .
in such case a triggering event , e.g. perturbation due to the passage of the companion star , heat flux from the inner active disc or gravitational instability of the dead zone @xcite , could start the accretion and `` ignite '' the dead zone , making at least part of it active .
consequently , the rate at which mass is accreted by the central star could increase dramatically .
this mechanism was suggested by @xcite to explain the fu ori type outbursts @xcite .
the layered disc model was further developed by @xcite , who noted that the non - zero thickness of the dead zone is an important factor influencing the structure of the active surface layers ( this is because their structure depends on the vertical component of gravity , which in turn depends on the distance from the mid - plane ) .
on the other hand , detailed mhd simulations of surface layers performed in a 3d shearing box approximation by @xcite showed that mri - driven , non - axisymmetric density waves can propagate far into the dead zone . as a result ,
a purely hd turbulence is excited there , providing an effective viscosity .
our original intention was to study how the evolution of the dead zone may be affected by the radiative heating from the inner active part of the disc . to that end
, we performed extensive two - dimensional ( axisymmetric ) simulations of the evolution of layered discs , allowing for radiative energy transfer in both radial and vertical directions .
unexpectedly , we found that the dead zone is unstable in a way that has not been reported before ; namely , it tends to decompose into rings . in the present paper
we illustrate this `` ring '' instability with numerical simulations , and we discuss it analytically , providing a physical explanation of the observed phenomena .
the remaining results of our simulations will be reported in a forthcoming paper .
the outline of the present paper is as follows . in
2 we briefly describe the numerical code and we list the assumptions underlying our simulations .
a layered - disc model whose dead zone decomposes into rings is presented in 3 .
4 contains an analytical discussion of the ring instability .
finally , in 5 we summarize our results and discuss the effects of the ring instability in more sophisticated disc models .
the simulations are performed with the help of the 3-d code tramp @xcite .
the equations of continuity @xmath12 momentum conservation @xmath13 and internal energy @xmath14 = -p\nabla\cdot\mathbf{v } + \mathbf{t}:(\nabla\mathbf{v})\ ] ] are solved in spherical coordinates @xmath15 using an explicit operator - splitting method .
@xmath16 is here the gravitational potential , @xmath17 is the centrifugal force , @xmath18 is the viscous stress tensor , @xmath19 is the rate of strain tensor composed of derivatives of velocity components , @xmath6 is the temperature ( assumed to be the same for gas , dust and radiation ) , @xmath20 is the specific heat , and the colon denotes a double scalar product of two tensors .
the self - gravity of the disc is neglected , so that @xmath21 where @xmath22 is the mass of the central star .
we use the equation of state of the ideal gas @xmath23 where @xmath24 is boltzmann constant , @xmath25 is the average molecular weight of the disc gas , and @xmath26 is the proton mass . the radiation transport is treated at the end of each hydrodynamic time - step by solving the equation @xmath27 where @xmath28 is the radiation energy density , and @xmath29 is the radiative flux . we adopt @xmath30 where @xmath31 is the rosseland mean opacity for the gas - dust mixture @xcite , and @xmath32 is the flux limiter used to interpolate between optically thin and optically thick cases @xcite . equations ( [ ttrte ] ) and ( [ ttrflux ] ) are solved implicitly using an iterative sor method .
after the convergence has been achieved , the updated temperature is calculated from the updated radiation energy density .
the advection routine for mass , momentum and energy is based on a second - order monotonic transport scheme originally introduced by @xcite and optimized by @xcite . for more details about the numerical code we refer to @xcite
. the models are axially symmetric .
we allow for the flow through the midplane of the disc , i.e. we do not impose a reflecting boundary condition there .
grid points are spaced logarithmically in @xmath2 , resulting in a progressively smaller radial extent of grid cells near the centre of the disc , where their vertical extent is also progressively smaller .
thus , the shape of grid cells is more nearly regular throughout the grid .
in the inner active part of the disc , approximate initial distributions of density and temperature are obtained from analytical @xmath33-disc models @xcite assuming a constant temperature profile in the direction perpendicular to the midplane . the outer , layered part of the disc
is initiated with the analytical model of @xcite .
the two solutions merge at the radius @xmath34 , where the mid - plane temperature of the layered part reaches @xmath35 k. the accretion rate of the layered part at @xmath34 defines the accretion rate in the inner active part .
similarly , the surface density of the dead zone is calculated assuming continuity of the total surface density at @xmath34 . initially , the surface density of the dead zone is constant in @xmath2 .
the initial rotational velocity is chosen so as to balance the gravity reduced due to the radial pressure gradient ( in a thin disc the rotation is nearly keplerian ) .
other components of velocity are set to zero . at the outer edge of the disc mass
is injected at every time - step at a fixed rate by the following procedure : 1 .
densities and temperatures are copied into the ghost zone from the adjacent active cells 2 .
densities in the ghost zone are normalized to the initial surface density 3 . a uniform radial velocity is set across the ghost zone such that the accretion rate is the same as in the analytical model .
the angular rotational velocity at both inner and outer boundary is extrapolated using a @xmath36 power - law . at the inner edge of the disc , and at the upper and lower boundary of the grid
, a free outflow boundary condition is imposed .
a constant temperature of @xmath37 k is maintained at the upper and lower boundary . at each time
- step the location of the dead zone is found based on two conditions that have to be fulfilled simultaneously : 1 .
@xmath38 , where @xmath39 is the minimum temperature at which the coupling between the gas and the magnetic field still operates 2 .
@xmath40 , where @xmath41 is the column density integrated from the surface of the disc along a line perpendicular to the midplane , and @xmath42 is the column density ionized by the cosmic rays . beyond the dead zone the viscosity coefficient is defined according to the @xmath33-prescription of @xcite @xmath43 where @xmath33 is a dimensionless parameter , @xmath44 is the sound speed and @xmath45 is the thickness of the active layer which is determined from the vertical distribution of density for each radius at each time - step .
we chose this particular form of shakura - sunyaev prescription because it is closer to the analytical , vertically averaged model . within the dead zone we set @xmath46 . to avoid numerical problems ,
we introduce a density limiter : whenever the density falls below @xmath47 ( which may be different for different models ) , it is doubled , and the temperature is adjusted so as to keep the pressure unchanged .
this procedure leads to the formation of a low - density `` atmosphere '' surrounding the disc . to prevent it from collapsing onto the disc , we artificially damp vertical and radial velocities in this region
however , this artificial readjustment only affects a negligible number of grid cells in the `` atmosphere '' , and the overall evolution of the disc is not influenced .
we obtained a broad sample of models with different parameters , which will be presented elsewhere . here
we describe a representative model with `` canonical '' parameters @xmath48 , @xmath49 k , @xmath50 and @xmath51 .
note that at @xmath52 k the ionization degree increases by five orders of magnitude due to ionization of potassium @xcite , and the magnetic reynolds number , which is directly proportional to the ionization degree , exceeds unity .
@xmath53 is the standard stopping column density of cosmic rays @xcite , and 0.01 is a value of the viscosity parameter often adopted for protoplanetary discs ( e.g. @xcite ) .
the simulations were performed on a grid of @xmath54 points in ( @xmath2,@xmath55 ) , with @xmath2 varying between 0.05 and 0.35 au and @xmath55 varying between @xmath56 and @xmath57 .
the model was integrated for @xmath58 yr , i.e. for 600 orbits at the outer edge of the disc .
the initial model is far from thermal equilibrium , however it quickly relaxes due radiative heat diffusion . the density and mass flux across the computational domain shortly after the relaxation , at the beginning of the instability and at the end of the simulation
are shown in fig .
[ rtmfdz ] : one clearly sees how the dead zone successively breaks into more and more rings . as we have not imposed any explicit perturbations , there is a good reason to believe that the rings result from a linear instability of layered discs .
the mechanism of this instability is illustrated in fig .
[ ringmech ] .
assume a small axially symmetric enhancement of the surface density at a radius @xmath9 ( which , of course , must be accompanied by an increase of the dead zone thickness @xmath59 ) .
since at higher distances from the mid - plane the vertical component of gravity is stronger , the active layer ( whose column density is at all times fixed by cosmic rays ) must get thinner at @xmath9 . in the standard prescription ,
the viscosity coefficient is proportional to the local thickness of the active zone @xmath45 ( this is motivated by the idea that @xmath45 is a natural scale that limits the maximum size of the edies of the mhd turbulence ) . therefore , smaller @xmath45 results in a lower viscosity in the elevated part of the active layer . the accretion rate , which depends on the derivative of viscosity ( see eq .
[ efar ] below ) , decreases at the inner edge of the ring .
this causes a bottle - neck effect in the accretion flow , and the mass accumulates in the ring . on the other hand
, the higher accretion rate at the outer edge of the ring makes the ring more compact .
this positive feedback between the dead zone thickness and the mass accumulation rate leads to the formation of the rings .
the rings formed in the simulation are shown in the radial profile of the surface density in fig .
[ rp - om ] .
the accretion rate profile exhibits the described minima at the inner edges of the rings and maxima at the outer ones .
the plot also shows the deviations from the keplerian angular velocity it can be seen that the inner edges of the rings rotate at a super - keplerian velocity .
thus , the rings may capture and concentrate the radially drifting boulders of meter - size , preventing them from accretion onto the central star ( e.g. @xcite ) .
such concentration of solids in peaks of gas density was observed by @xcite and @xcite in the case of spiral density waves formed by the gravitational instability .
smaller than the unperturbed value @xmath60
. therefore also @xmath61 .
the radial profiles of average viscosity and accretion rate are shown at the bottom of the figure . ]
( top ) , the accretion rate @xmath62 ( middle ) and difference to the keplerian angular velocity @xmath63 ( bottom ) . ]
let us consider a layered disc consisting of a dead zone with thickness @xmath64 and two active surface layers with thickness @xmath45 each .
the surface layer accretes at a rate @xmath65 where @xmath42 is the column density of the surface layer , @xmath66 is the kinematic viscosity , and @xmath2 is the cylindrical radial coordinate . equation ( [ efar ] ) is a direct consequence of angular momentum conservation @xcite . assuming that the accretion energy is radiated locally , we obtain the standard formula for the effective temperature @xmath67 @xmath68 where @xmath69 is stefan - boltzmann constant . in the optically thick approximation the temperature @xmath70 at the boundary between the dead zone and the active layer
is given by the formula @xmath71 where @xmath72 is the optical depth of the active layer , and @xmath31 is the rosseland mean opacity @xcite . in general , the opacity coefficient is a complex function of density and temperature . in a protoplanetary disc
it can be approximated by a set of power laws @xmath73 where the constants @xmath74 , @xmath75 and @xmath76 have different values in different opacity regimes @xcite . at
temperatures lower than @xmath7 k ( i.e. nearly everywhere in a protoplanetary disc ) @xmath31 practically does not depend on density , so we assume @xmath77 . as before , we assume that the viscosity coefficient is given by equation ( [ s - s ] ) . let @xmath78 be the ratio of the total disc half - thickness to the active layer thickness @xmath45 : @xmath79 from the equation of hydrostatic equilibrium in the direction perpendicular do the midplane of the disc we get an approximate formula @xmath80 where @xmath81 , @xmath82 and @xmath83 is , respectively , pressure , density and sound speed at the boundary between the dead zone and the active layer . combining ( [ vedz ] ) with equations ( [ ldis ] ) and ( [ hubeny ] ) , and using ( [ s - s ] ) with @xmath84 we get @xmath85^\frac{1}{3-b } \sigma_a^\frac{2}{3-b } \alpha^\frac{1}{3-b } \delta^{-\frac{1}{2(3-b ) } } \omega^\frac{1}{3-b } \ . \label{ti}\ ] ] inserting equations ( [ ldis ] ) ( [ ti ] ) into equation ( [ efar ] ) , we arrive at the following formula for the accretion rate in the active layer : @xmath86 where @xmath87^\frac{1}{3-b } \frac{k_b}{\mu m_\mathrm{h}}^\frac{4-b}{3-b } \sigma_a^\frac{5-b}{3-b } \alpha^\frac{4-b}{3-b } \ .\ ] ] in the following , we shall derive an equation which relates the angular rotational velocity @xmath88 to the thickness of the disc .
let us assume that the disc is thin ( @xmath89 ) , and neglect the dependence of @xmath88 on @xmath90 ( see e.g. @xcite ) .
we may write @xmath91 where @xmath92 , @xmath93 and @xmath83 are , respectively , mid - plane values of pressure , density , and sound speed . note that because there is no heat generation in the dead zone , the mid - plane sound speed is the same as at the boundary between the dead zone and the active layer ( i.e. at each @xmath2 the dead zone is isothermal along a line perpendicular to the midplane ) .
the mid - plane density @xmath93 is given by the vertical hydrostatic equilibrium @xmath94 and for @xmath95 we have @xmath96 where it was assumed that the mass accumulated in dead zone is already large ( i.e. @xmath97 ) , which allows us to neglect terms of the second order in @xmath98 .
inserting @xmath99 and @xmath100 into equation ( [ omper_bas ] ) we obtain @xmath101 the sound speed @xmath83 only weakly depends on disc thickness ( @xmath102 for @xmath103 ) .
the dependence of @xmath82 on @xmath78 is even weaker ( @xmath104 for @xmath105 ) .
therefore , we assume that for a small perturbation of @xmath78 the last two rhs terms in equation ( [ omd_full ] ) , which are proportional to derivatives of @xmath106 and @xmath107 , are small compared to the second rhs term , which is directly proportional to the derivative of @xmath78 . the final relation between @xmath88 and @xmath78 is @xmath108 equations ( [ ardo ] ) and ( [ omd ] ) will be used in the next subsections to derive the dispersion relation for the perturbations of the disc .
let us perturb @xmath78 at a radius @xmath9 , assuming that in a small region @xmath109 , @xmath110 @xmath78 consists of the unperturbed part @xmath111 ( which in that region may be regarded as independent of @xmath2 ) plus a cosine term with a wavenumber @xmath112 @xmath113 where @xmath114 is the amplitude of the perturbation . according to ( [ omd ] ) , the angular velocity consists of the keplerian part @xmath115 and the pressure - correction @xmath116 .
neglecting terms of the second order in @xmath116 , and using equation ( [ omd ] ) we obtain @xmath117 the first and second derivatives of @xmath78 and @xmath88 are @xmath118 since we only want to obtain a rough idea about the growth rate of the perturbation with a specific @xmath112 , we may approximate @xmath119 with 1 and @xmath120 with 0
. then the disc thickness @xmath78 , the angular velocity @xmath88 and their derivatives are : @xmath121 @xmath122 due to mass accumulation , the surface density of the dead zone @xmath123 grows at a rate @xmath124 inserting equation ( [ ardo ] ) into ( [ dsdm ] ) and using approximations ( [ delta_pert ] ) and ( [ omega_pert ] ) we obtain @xmath125\end{aligned}\ ] ] the linearized unperturbed part of the last equation is @xmath126 and the linearized equation which describes the evolution of surface density perturbation with wavenumber @xmath112 is @xmath127 \ .\end{aligned}\ ] ] the surface density of the dead zone @xmath128 is related to the disc thickness @xmath78 through @xmath129 where we used the mid - plane density given by ( [ rhomp2 ] ) .
the integral can be evaluated using an error function to yield @xmath130 remembering that @xmath97 we get @xmath131 differentiating the last equation with respect to time leads to a formula connecting @xmath123 with @xmath132 @xmath133 whose unperturbed part and equation ( [ sdz0 ] ) can be combined into an equation @xmath134 which describes the evolution of the unperturbed disc thickness due to the accumulation of mass .
next , from equation ( [ sdzk ] ) , and equation ( [ dsdzd ] ) written for a specific wavenumber @xmath112 we get the dispersion relation which describes the growth rate of the perturbation with the wavenumber @xmath112 .
@xmath135\end{aligned}\ ] ] the perturbation growth rate diverges at short wavelengths .
this is because our simple analysis does not include any damping mechanisms .
in reality , however , thin rings would diffuse due to the thermal motion of particles .
note also that that the radial pressure scale height can not become smaller than the vertical one .
therefore , rings with circular r - z profiles should form , which is in agreement with our numerical model ( see fig .
[ rtmfdz ] ) . in comparison to growth rates calculated from equation ( [ disprel ] ) .
growth rates in the numerical simulation were determined by measuring @xmath78 and @xmath136 ( @xmath137 ) . ] fig .
[ od - dr1 ] shows growth rates of the disc thickness @xmath78 in the four inner - most rings .
they are compared to the growth rates given by the dispersion relation ( [ disprel ] ) .
the growth rates in the analytical model are more than one order of magnitude higher .
the possible explanation of this discrepancy can be that the analytical model does not allow for heat transfer from the ring into the active layer above it or for convective flows that mix the mass inside the rings .
the radiative transfer in the radial direction may also be important .
all those processes make the dip in the average viscosity shallower and effectively decrease the growth rate of the instability . in this section
we estimate how strongly the ring instability is affected by the radiation flux from the star . in the preceding section the stellar flux
was not included self - consistently since the linear analysis presented there serves as the comparison model for the numerical simulation which does not include irradiation .
on the other hand , with the irradiation included explicitly the calculations would become too complex ( or just impossible ) to perform .
firstly , we estimate how important the irradiation is for the unperturbed disc .
we do it by computing the change of the temperature at the boundary between the dead zone and the surface layer ( from which the thickness of the surface layer and the viscosity are determined ) .
stellar heating can be included in eq .
( [ hubeny ] ) in the following way @xmath138 where @xmath67 is the effective temperature resulting _ solely _ from the viscous dissipation , @xmath139 is the temperature of the stellar photosphere , and @xmath140 is the dilution factor which depends on stellar radius , distance from the star , and geometry of the disc @xcite . for a conical disc ( in which the aspect ratio @xmath141 does not depend on @xmath2 ) and @xmath142 we have @xmath143 ( see e.g. @xcite ) . evaluating the rhs terms in eq .
( [ irr_hubeny ] ) for our disc model and typical parameters of a t tauri star ( @xmath144 r@xmath145 , @xmath146 k ) we find that the first ( viscous ) term always dominates .
more susceptible to effects of irradiation are flaring discs , in which @xmath147 . the flaring index @xmath148 depends on the model , e.g. a vertically isothermal model in which the intercepted stellar flux is equal to the blackbody emission from the disc yields @xmath149 @xcite .
however , even in such model the irradiation term dominates only for ( @xmath150 au ) , where the grazing angle becomes sufficiently large .
effects of irradiation may be more important for the ring instability itself .
the inner edge of a growing ring is more strongly heated by stellar radiation , because the grazing angle @xmath151 is larger there . as a result ,
the temperature and the thickness of the active layer increase locally , leading to an increased viscosity .
then , the dip in the accretion rate , which is responsible for the ring growth , becomes shallower or it may even disappear entirely . the importance of this effect may be roughly estimated by comparing the amplitude of viscosity increase due to stellar irradiation to the amplitude of viscosity decrease due to the stronger vertical component of the gravitational force ( the latter effect is explained in fig .
[ ringmech ] ) .
let us assume a ring - like perturbation of the disc thickness with an amplitude @xmath114 and a wavelength @xmath152 .
since the most unstable wavelength is @xmath153 , let us parametrize the perturbation wavelength by the dimensionless value @xmath154 .
the grazing angle at which the stellar radiation strikes the inner edge of the ring is @xmath155 evaluating the dilution factor @xmath156 , inserting it into eq .
( [ irr_hubeny ] ) , and combining it with eqs .
( [ ldis ] ) , ( [ vedz ] ) and ( [ s - s ] ) we obtain the viscosity in a form @xmath157^{1/4 } % n_1(r)\delta^{\frac{-4+b}{2(3-b ) } } + % n_2(r)\delta^{-1/2 } \left [ \frac{\delta_k}{l\delta } + n_3(r)\right]^{1/4}\end{aligned}\ ] ] where @xmath158 @xmath159 and @xmath160 initially , the function @xmath161 always grows as @xmath114 increases .
however , depending on the remaining parameters ( @xmath111 , @xmath2 and @xmath162 ) , it may start to decrease and quickly drop below the initial value @xmath163 .
therefore , a very small perturbation of the disc thickness is always stabilized by the stellar irradiation , but if the amplitude of the perturbation reaches some value , the ring instability may start to work .
this critical value @xmath164 ( @xmath165 ) strongly depends on @xmath162 : it is smaller for higher @xmath162 , where the grazing angle is smaller .
the dependence on the remaining parameters ( @xmath2 and @xmath111 ) for @xmath166 is shown by fig .
[ fig5 ] .
we see that the stabilizing effect of the irradiation may become unimportant for broad rings ( large @xmath162 ) , small @xmath111 ( less mass accumulated in the dead zone ) and/or at small radii .
for which @xmath165 . beyond their endpoints
the solution becomes unphysical ( @xmath167 , i.e. the perturbed disk has regions with negative thickness ) . ]
we described a new instability in layered accretion discs .
the accretion rate in such discs is in general a function of radius . as a result
, mass may accumulate in the non - viscous dead zone near the mid - plane of the disc .
however , a small ring - like perturbation of the dead zone thickness leads to the deviation of the rotational velocity which results in non - uniform accumulation rate of the mass supporting growing of that ring perturbation .
this ring instability may eventually lead to a decomposition of the dead zone into rings .
we observed the formation of such rings in the 2d axially symmetric radiation - hydrodynamic simulation of the layered disc . to illustrate how the instability works , we give its approximate analytical description .
according to the analytical results , the narrowest rings grow most rapidly .
therefore , we may expect the formation of the radially thin rings whose radial extent will be given just by the thermal pressure of the gas .
the comparison of ring sizes shows a reasonable agreement between the numerical simulation and the analytical model . the irradiation by the central star may substantially decelerate or even stop the growth of the instability in some regions of the disc
however , broad rings are less affected , and the effects of irradiation become less important in thin discs and/or at small distances from the star . on the other hand , once the innermost ring has developed , and the flaring index @xmath148 is not too high , the disc at larger radii is shadowed and more rings may develop there . if the disc is truncated ( e.g. by the magnetosphere of the star ) the shadowing effect may also be caused by its inner edge , allowing for an efficient growth of the instability . the rings created by this instability may be important in terms of planet formation , because they can be the places where the solid particles ( gravel and boulders ) accumulate .
it is because of the higher rotational velocity at the inner edge of the ring and the lower one at the outer edge .
the drag force which makes the grains to move inwards is smaller at the inner edge of the ring and higher at the outer edge .
the solid particles may merge into larger bodies ( planetesimals ) necessary for the formation of planets .
massive rings are subject to a hydrodynamical instability in 3d , e.g. with respect to non - axisymmetric perturbations @xcite .
this instability occurs if the rotational angular velocity decreases with radius steeper than @xmath168 . in our simulation
, this occurs for the innermost ring at a time of @xmath169 yr . in the non - viscous case
such ring would most likely decay into so - called `` planets '' , i.e. large scale vortices , as found in numerical simulations by @xcite , and further investigated by @xcite .
the fate of such a ring in 3d viscous hydro - simulations including the effects of layered accretion still have to be performed .
our model assumes zero viscosity in the dead zone .
however , there are some indications @xcite that even in the dead one there might be some turbulence present , induced by waves originating in the active parts of the disc . in such a case
, the excess of surface density could result in a higher accretion rate in the rings , and the growth of the rings would be stopped or even they might not form at all .
this issue will be addressed in a forthcoming paper .
this research was supported in part by the european community s human potential programme under contract hprn - ct-2002 - 00308,planets .
rw acknowledges financial support by the grant agency of the academy of sciences of the czech republic under the grants avoz 10030501 and b3003106 .
mr acknowledges the support from the grant no . 1 p03d 026 26 from the polish ministry of science . | in this work we use the radiation hydrodynamic code tramp to perform a two - dimensional axially symmetric model of the layered disc . using this model
we follow the accumulation of mass in the dead zone due to the radially varying accretion rate .
we found a new type of instability which causes the dead zone to split into rings .
this `` ring instability '' works due to the positive feedback between the thickness of the dead zone and the mass accumulation rate .
we give an analytical description of this instability , taking into account non - zero thickness of the dead zone and deviations from the keplerian rotational velocity .
the analytical model agrees reasonably well with results of numerical simulations .
finally , we speculate about the possible role of the ring instability in protoplanetary discs and in the formation of planets .
[ firstpage ] solar system : formation , accretion discs , hydrodynamics , instabilities . |
quantum systems when placed in low dimensional lattices typically exhibit strongly correlated effects driving them towards regimes with no classical analog .
many properties of these regimes or quantum phases @xcite depend in turn on the properties of their ground state and low lying energy excitations @xcite .
a problem of particular interest in the field of strongly correlated systems is the emergence of critical phases in a system where the generic behaviour as coupling constants are varied is to be a gapped system , although those gapped phases may be of different nature . in this paper
we address this problem by selecting a system of quantum spins that allows us to perform a detailed study of critical and non - critical phases on equal footing , i.e. , without any bias towards an a priori preferred phase . for reasons explained in sect.[sect_model ] ,
the quantum spins are arranged in a 2-leg ladder lattice @xcite with anti - ferromagnetic heisenberg couplings along the legs while rung couplings are ferromagnetic .
in addition , we also introduce an explicit dimerization coupling in the hamiltonian along the leg directions , which can be varied from zero to strong values .
this coupling plays a major role in order to create the aforementioned critical phases out of a system with only gapped phases .
this particular type of 2-leg ladder system has a number of open problems such as the precise location of critical phases in the phase diagram of the coupling constants , and the nature of the gapped phases it exhibits .
our study is complete enough so as to be able to solve for these problems in a very precise manner .
the understanding of these purely quantum effects is usually a hard problem .
perturbative and variational methods in quasi - one dimensional systems like chains and ladders are not well suited to uncover the physics in the whole range of coupling constants involved in the description of the interactions in the system . on the contrary
the dmrg method @xcite , @xcite , @xcite , @xcite , @xcite allows us to identify the critical phases clearly and without any bias .
this is so because the method is non - perturbative and allows a controllable management of errors .
our studies are also of interest since experiments on ladder materials have revealed a very complex behaviour , such as an interplay between a spin - gapped normal states and superconductivity @xcite . moreover , a new field of study for these complex effects has been opened by the simulation of strongly correlated systems in optical lattices @xcite , in particular quantum spin chains and ladders @xcite .
this paper is organized as follows : in sect.[sect_model ] we introduce the model hamiltonian describing a 2-leg ladder lattice of spins @xmath3 with columnar bond - alternating antiferromagnetic couplings in the horizontal direction and ferromagnetic couplings in the vertical direction , see fig.[ladderpic ] .
we can identify some particular behaviours in appropriate weak and strong coupling limits , but not for generic values of the couplings .
in sect.[sect_critical ] we point out the rich physical effects posed by open boundary conditions in these 2-leg ladders with finite length , although it also implies an a priori analysis in order to find out which low - lying states contribute to the gap of the system in the thermodynamic limit .
this we can be done with the dmrg method by targeting several states and measuring their magnetization properties in the bulk and at the ends .
then , we compute numerically the gap and we establish the existence of a critical line in the quantum phase diagram of the model
. a numerical fit of this critical curve is also given . in sect.[sect_haldanedimer ] we determine the structure of the phase diagram by identifying the type of gapped phases occurring at each side of the critical line found in the previous section .
they correspond to haldane and dimer phases .
they are identified by measuring the string order parameter and the dimerization parameter with the dmrg method .
we complete our study of these phases measuring different observables .
sect.[sect_conclusions ] is devoted to conclusions .
competing ferromagnetic versus antiferromagnetic spin interactions may give rise to critical phases if they are appropriately arranged in certain quasi - one dimensional lattices .
one emblematic example of this phenomenon is a lattice of quantum spins with the shape of a 2-leg ladder such that there are antiferromagnetic couplings along the legs and ferromagnetic interactions along the rungs connecting both legs .
in addition , the antiferromagnetic couplings are bond - alternating in a columnar fashion .
dimerization interactions in the hamiltonian are also known as staggered interactions .
this configuration is shown in fig.[ladderpic ] .
more precisely , this configuration of heisenberg - like interactions is associated with the following quantum hamiltonian . in the horizontal direction ( legs )
, we picture the bond alternation with strong links @xmath4 and weak links @xmath5 . in the vertical direction ( rungs ) , the system is arranged in the form of a columnar dimerization : strong links are parallel to one another , and similarly for weak links in the lattice . ]
@xmath6 where @xmath7 are quantum spin @xmath3 operators located at site @xmath8 of the leg @xmath9 , and @xmath10 , @xmath11 , @xmath12 $ ] are the antiferromagnetic , ferromagnetic and staggering couplings , respectively , as mentioned above .
notice that several known regimes can be reached by tuning the coupling constants towards particular values . in the weak coupling limit , making @xmath13 we end up with a system consisting on two effectively decoupled @xmath14 heisenberg chains with bond alternation ( bahc ) , which are known to be gapped for every value of the dimerization parameter @xmath15 @xcite , except for the point @xmath16 . in the strong coupling limit , making @xmath17 the system can be effectively described by a @xmath18 spin chain with bond alternation , which is predicted to be gapped for all values of @xmath15 except for a critical point at a non - zero value @xmath19 @xcite .
these predictions are based on an approximate mapping onto the o(3 ) @xmath20 model @xcite at topological angle @xmath21 .
this yields a critical value of @xmath22 when @xmath23 , and similarly another symmetric critical value at @xmath24 .
thus , we shall always concentrate in the region @xmath25 , due to the symmetry @xmath26 in the hamiltonian . this non - linear sigma model ( nl@xmath20 m )
prediction misses the correct location of the critical point due to the approximations involved in that mapping .
the exact location of this point has been widely studied @xcite and results slightly varied depending on the approach , however modern studies place it at @xmath27 @xcite@xcite , also compatible with fig.[l500vsl2x140](_lower _ ) which gives @xmath28 for a chain of @xmath29 sites .
these studies also conclude that the region @xmath30 corresponds to a haldane phase while for @xmath31 we move to a dimer phase .
the emergence of a dimerized @xmath18 spin chain in the strong coupling limit can be explained by noting that as the rung coupling is ferromagnetic and strong @xmath32 , the two spins @xmath3 in each rung find energetically favorable to form a spin triplet .
ladder at the point @xmath33 .
_ up _ : mean magnetization @xmath34 in the states with @xmath35 and @xmath36 , as explained in the text . _
inset _ : cumulative sum of the magnetization over the whole extent of the ladder and the states with @xmath37 and @xmath38 .
the order of the sites in the @xmath39-axis corresponds in this case to the path used to traverse the ladder in a dmrg sweep .
_ down _ : difference of energy density ( doe ) of the excited states with @xmath37 and @xmath38 : @xmath40 .
.the scale on the right axis corresponds to the cumulative sum .
see text for more explanations .
, title="fig : " ] ladder at the point @xmath33 .
_ up _ : mean magnetization @xmath34 in the states with @xmath35 and @xmath36 , as explained in the text .
_ inset _ : cumulative sum of the magnetization over the whole extent of the ladder and the states with @xmath37 and @xmath38 .
the order of the sites in the @xmath39-axis corresponds in this case to the path used to traverse the ladder in a dmrg sweep .
_ down _ : difference of energy density ( doe ) of the excited states with @xmath37 and @xmath38 : @xmath40 .
.the scale on the right axis corresponds to the cumulative sum .
see text for more explanations .
, title="fig : " ] for generic values of the coupling constants in the hamiltonian , this model has been the subject of a series of conjectures based on exact diagonalization numerical studies @xcite in the absence of dimerization @xmath16 and analytical studies using bosonization and nl@xmath20 m mapping @xcite in the presence of dimerization @xmath41 .
those numerical methods only allowed to reach ladder lengths typically of @xmath42 or so , which prevents from reaching any definitive conclusion on the bulk properties of the system in the thermodynamic limit .
as for the analytical studies , they conjectured the existence of a possible critical region , but due to the nature of the methods it is not possible to give its location in terms of the original coupling constants in the model hamiltonian .
computed on a @xmath43 ladder for different values of the parameter @xmath44 .
each minimum in the gap belongs to the critical line . ]
one of the main issues in this model is whether it exhibits a critical line in the quantum phase diagram of @xmath44 vs. @xmath15 .
we solve this open problem in the positive by using the dmrg method in finite version algorithm which provides us with better accuracy values than the infinite method version , although at the expense of more demanding time computing resources .
the performance of the finite dmrg algorithm is characterized by the following parameters : the number of states @xmath45 retained in the truncation process of the rg method , the weight of the discarded states @xmath46 which is a measure of the dmrg error , the number of sweeps @xmath47 or iterations of the method after the initial warm - up process and the tolerance @xmath48 of the target state energy which controls the average number of iterations that will need the diagonalization algorithm ( lanczos in our case ) to compute the target state .
we shall provide values of these parameters in our numerical computations below . before applying the finite - size dmrg method ,
two important remarks are in order : i/ as we shall always work with a fixed value of @xmath49 the length of the lattice , the gap @xmath50 is always finite and only in the thermodynamic limit @xmath51 it may vanish for certain values of @xmath44 and @xmath15 which define the critical line we are searching for . thus , the signature of a gap in @xmath50 for fixed @xmath44 and varying @xmath15 will show up as a minimum in the dimerization parameter . upon increasing the value of @xmath49
, we shall obtain more robust estimations of the critical value @xmath52 from the minima @xmath53 .
this is a finite - size scaling analysis of the dmrg numerical data . .
the points of the critical line correspond to the coordinates that minimize @xmath54 .
the solid line is only a guide for the eye._bottom _ : the value of the minimum gap for the ladder with @xmath43 sites is very similar to the corresponding @xmath18 bahc with @xmath55 but the value of @xmath19 that minimizes this gap is still a bit shifted , which constitutes a signal that @xmath56 is still a low value to accurately mimic the limit bahc behaviour .
the computations for the @xmath57 bahc were performed storing @xmath58 eigenvectors of the density matrix .
, title="fig : " ] .
the points of the critical line correspond to the coordinates that minimize @xmath54 .
the solid line is only a guide for the eye._bottom _ : the value of the minimum gap for the ladder with @xmath43 sites is very similar to the corresponding @xmath18 bahc with @xmath55 but the value of @xmath19 that minimizes this gap is still a bit shifted , which constitutes a signal that @xmath56 is still a low value to accurately mimic the limit bahc behaviour .
the computations for the @xmath57 bahc were performed storing @xmath58 eigenvectors of the density matrix .
, title="fig : " ] ii/ the physics of this 2-leg ladder is richer when the lattice has open boundary conditions .
moreover , the numerical performance of the dmrg method is also better under these conditions .
however , open boundary conditions must be handled with care in order to identify the gap @xmath50 we are after .
we shall provide ways to do this identification by targeting appropriate low - lying states and measuring convenient observables with them . in particular
, using open boundary conditions we have found that the first excited state lies within the sector with total z - spin angular momentum @xmath37 , but in the haldane phase it converges to the ground state that has @xmath59 as we take larger sizes of the system .
we have then to consider the next excited state in the sector with @xmath38 to compute the gap of the spectrum as @xmath60 fits very well to a potential function of the form @xmath61 , with @xmath62 and @xmath63 . ]
the reason for considering @xmath64 instead of @xmath65 as the gap of the system , can be justified as follows : in the complete dimerized limit @xmath66 , it is clear that the difference in energy between two arbitrary consecutive levels is the same , and corresponds exactly to the energy needed to promote one singlet bond to a triplet . the argument for the limit @xmath16 makes use of the properties of haldane phases , where it is known to appear a non bulk excitation due to the existence of virtual spins at the end of the chain .
our conclusion is that the lowest lying state with @xmath67 consists on a superposition of two kind of excitations , namely the haldane non - bulk triplet mentioned before and the bulk itself , also giving a triplet . considering this scheme , in order to obtain the gap related to the bulk excitations we have to substract the non bulk excitations present in the lowest lying states of sectors @xmath68 and @xmath67 . in fig.[obcexcitations ]
we show rigourous comparations of these two states in the haldane limit @xmath16 .
computations have been done on a @xmath69 ladder at the point @xmath33 . on the _ up _ part of the figure , we plot the mean magnetization @xmath34 in the states with @xmath35 and @xmath36 , computed in one leg of the ladder , since due to the symmetry of the hamiltonian , the magnetization is the same in both legs . as a check of the accuracy of our computations we observed that the results in both legs are the same up to the fifth or sixth decimal digit . in the _ inset _ of that figure , we show the cumulative sum of the magnetization over the whole length of the ladder and the states with @xmath37 and @xmath38 .
the order of the sites in the @xmath39-axis corresponds in this case to the path used to traverse the ladder in a dmrg sweep . in
the _ down _ part of this figure , we plot the difference of energy density of the excited states with @xmath37 and @xmath38 : @xmath40 the difference has been divided into three contributions : the contribution labelled with _ even _ stands for links involving sites in the same leg and the even sublattice ( @xmath70 ) , _ odd _ involves links joining sites in the same leg and the odd sublattice ( @xmath71 ) , and _
perpendicular _ denotes links among legs ( @xmath72 ) .
the cumulative sum of the difference of the various contribution , measured in the right axis scale , is also shown .
interestingly enough , we can observe the magnetization pattern at the ends being almost identical in the states with @xmath73 and @xmath74 .
the contribution to the @xmath75-axis projection of the spin coming from the ends is equal to 1 in both cases .
notice also that the difference of the density of energy between these states is close to zero at the ends , while it becomes clearly apreciable in the bulk .
all these facts strongly support the picture of a non - bulk triplet excitation with the same nature in both states , that leaves the bulk of the chain with a neat value of the projection equal to @xmath76 and @xmath77 and gives a strong hint on the equivalence of @xmath78 with periodic boundary conditions and @xmath64 for open systems . sites .
the sop has been computed forming the triplets with adjacent @xmath14 spins located in different legs .
clearly the value of this parameter is non - vanishing in the low region of @xmath15 , where the system is in the haldane phase .
the dimer phase is nonetheless characterized by a vanishing sop._inset _ : sop computed for a @xmath18 bahc .
the resemblance between both systems is evident in the region with @xmath79 . ] after this previous analysis to identify the states needed to target the gap of the system , we present in fig.[gap21 ] some values of the gap @xmath64 for a ladder consisting on @xmath43 sites , at different regions of the parameter space .
computations have been performed retaining @xmath80 states of the density matrix and the grid used to explore the phase diagram is @xmath81 $ ] , and @xmath82 .
the existence of a set of minima in the function @xmath54 is clear in this graph , although they shall become more distinguishable as we move to higher values of @xmath83 . as an instance of the accuracy of our results , we point out that a systematic examination of the error in each of the truncations of our dmrg computations reveals that the highest values in the whole process are of the order of @xmath84 , and
mostly they are of order @xmath85 . to obtain a suitable acuracy in the results
we have set the number of sweeps @xmath86 and the tolerance to @xmath87 . to compute with enough precision the critical value @xmath52 that minimizes the gap it becomes necessary to use large amount of data . on this regard
, we have used interpolated values resulting from the dmrg computations .
computed in the middle of a ladder of @xmath88 sites .
since we are explicitly introducing some staggering in the hamiltonian , the dimerization parameter is non - vanishing even in the haldane phase .
however , the shape of the graphs seem to have an inflexion at the critical point .
the graph corresponding to the @xmath18 bahc , has been scaled down by a factor @xmath89 due to the effective coupling constant of the ladder , which is known to be half the constant corresponding to the bahc . ]
now , we can detect the presence of a critical line in the quantum phase diagram separating gapped phases . in fig.[l500vsl2x140 ] we plot the critical region consisting of the coordinates for each minimum in fig.[gap21 ] . in earlier studies
@xcite , we placed the critical point of the @xmath18 bond - alternating heisenberg chain ( bahc ) at @xmath27 .
the curve shown in fig.[l500vsl2x140 ] shows a vertical asymptota that is still a bit off from this limiting value corresponding to the region @xmath79 , but this is simply because we have chosen a value of @xmath90 which is still not big enough and also due to finite - size effects on the 2-leg ladder . in the lower plots of fig.[l500vsl2x140 ]
we address these possibilities by comparing our ladder in the strong ferromagnetic limit with a pure @xmath18 bahcs with different sizes .
two parameters are important in this discussion , namely , the value @xmath52 that minimizes the gap , and the value of the gap itself at this point @xmath91 .
as we can observe in fig.[gap21 ] , the value of @xmath91 does not strongly depend on the particular choice of the coupling constant ratio @xmath44 , while it is definitely influenced by the size of the system .
in fig.[l500vsl2x140](lower ) it is shown that the shift of @xmath19 computed for two @xmath18 bahcs with different sizes , but still large enough both , is less noticeable than the difference in their value of @xmath91 .
it is clear the similarity of this magnitude in the case of the ladder and the corresponding bahc , as well as the shift in the value of @xmath19 .
all this make us conclude that in order to attain a better convergence with the @xmath18 bahc and a better estimate of the critical asymptota @xmath27 , we shall increase the strength of the ferromagnetic coupling rather than the size of the system .
as we have a set of numerical data from the finite - size analysis of the critical line , we can also make a numerical estimation of the criticality curve . in fig .
[ fit ] we present a fit of the critical curve in the region close to @xmath92 .
we choose as trial function for this fitting an inverse power law with some coefficients and exponents that are fixed by our numerics , namely , @xmath93 the fitting yields the following estimations for the values of the parameters @xmath94 and @xmath95 that best fit the data : @xmath62 and @xmath96 , and for simplicity the value of @xmath27 is taken for fixed . in the ground state of a @xmath69 sites ladder and @xmath56 .
the plot on the left shows the correlation value for some arbitrary values of @xmath15 .
the plot on the right shows the value of this magnitude in the whole region of the parameter @xmath15 . ]
once we have established the existence of a critical line in the quantum phase diagram of the model , it is natural to wonder about the two gapped phases that this line separates .
more specifically , whether they are different or not and their identification as quantum phases in the framework of strongly correlated systems
. the possible nature of those phases can be guessed from the strong ferromagnetic limit @xmath97 of the ladder , effectively leading to the @xmath18 bahc .
the phases of this chain are known to be the massive haldane phase , separated by a critical point from the also massive dimer phase . to test the nature of each phase
, we will resort to two different order parameters .
the haldane phase is known to exhibit a particular hidden order that can be measured by the string order parameter ( sop ) @xcite , @xcite .
the definition of this operator for a spin-1 chain is as follows : @xmath98 this operator acting on our ground state measures how far it is from a spin liquid nel state consisting on a sequence of @xmath18 spins such that every spin with projection @xmath99 is followed by @xmath100 and @xmath101 for every @xmath102 . when we deal with @xmath14 particles , to compute the sop we have to define the pairs of particles which are most likely to couple to give a triplet and compute the sop along the path connecting them .
in our case , the existence of a ferromagnetic coupling clearly suggests that the triplets will result via this coupling .
it is also worth recalling that the sop is a parameter suited to work with translational invariant systems . in order to correctly estimate the sop in open systems
, we must restrict the computation to a region shorter than the whole length of the chain where end - effects are negligible and only bulk physics is relevant . in fig.[sop ] we show the sop computed traversing the path shown in fig.[laddersop ] .
we can observe a non - vanishing sop in the haldane region , while it rapidly decays to zero in the dimer phase .
the inset shows the sop computed for a @xmath18 bahc and the resemblance between both systems is apparent .
therefore , the phase below the critical line in the numerical phase diagram of fig.[l500vsl2x140 ] is a gapped haldane phase . in the first lying excited state in the sector with @xmath37 of a @xmath69 ladder and @xmath56 . ] in the first lying excited state in the sector with @xmath38 of a ladder consisting on @xmath88 sites and @xmath56 . ]
as for the region above the critical line in fig.[l500vsl2x140 ] , we have guessed from the strong coupling limit that it may be a dimer phase .
the structure of a dimer phase is such that full translational invariance symmetry of the system is broken by one unit cell of the lattice .
this situation can be detected by means of the dimerization parameter , which can be defined for our particular 2-leg ladder as @xmath103 ladder for the lowest lying states in the sectors @xmath37 ( _ top _ ) , and @xmath38 ( _ bottom _ ) . the curves are separated in the different sublattices consisting on the sites ocuppying odd or even positions .
notice that both states present a peaked magnetization at the ends for @xmath104 well into the haldane phase , while it vanishes in the dimer phase with @xmath105 .
, title="fig : " ] ladder for the lowest lying states in the sectors @xmath37 ( _ top _ ) , and @xmath38 ( _ bottom _ ) . the curves are separated in the different sublattices consisting on the sites ocuppying odd or even positions .
notice that both states present a peaked magnetization at the ends for @xmath104 well into the haldane phase , while it vanishes in the dimer phase with @xmath105 .
, title="fig : " ] the subindex @xmath8 is necessary since open systems are intrinsically not translationally invariant . in fig.[dimerization ] it is plotted the dimerization parameter of the ladder measured in the middle of the chain . since staggering is explicitly introduced into the hamiltonian , the order parameter vanishes only at @xmath16 , but is finite even in the haldane phase .
nevertheless , our plots clearly exhibit different behaviours related to the convexity of the parameter at each phase .
this observation indicates that an accurate estimation of the point of inflexion in the dimerization parameter could be used as a measure of the critical point separating both phases .
we have also performed some measurements in the ladder to give more hints to understand the nature of both phases .
figures [ colormap0 ] , [ colormap1 ] , and [ colormap2 ] show the correlation @xmath106 between sites in the perpendicular rungs .
the pattern of the correlation can be understood by noticing that the correlation between two isolated @xmath14 spins coupled to give a singlet is @xmath107 while it equals @xmath108 if the spins form a triplet . from these values
we observe that the perpendicular rungs in the ground state are forming triplets and the distribution is uniform all along the ladder . in the excited states
however , the triplet strength of some rungs is weakened , signaling the presence of magnon - like excitations , also apparent in fig.[szgraph ] .
the nature of the non - bulk excitation present in the haldane phase is not magnon - like and that explains the different number of kinks in the haldane and dimer phase in fig.[colormap1 ] and fig.[colormap2 ] .
we have determined the existence of a critical line in the quantum phase diagram of a 2-leg ladder with columnar dimerization and ferromagnetic vs. antiferromagnetic couplings . in this study
, we use the finite - size system dmrg method which allows us to give the location of that critical curve .
moreover , we have clearly identified the two phases separated by the critical line to be a haldane phase and a dimer phase .
this identification is carried out by measuring the string order parameter and the dimerization order parameter in the whole range of values of the coupling constant ratio @xmath44 and dimerization parameter @xmath15 . as a byproduct
, we have introduced a systematic analysis of the spins at the borders of the open 2-leg ladder lattice .
our model is based on @xmath3 spins , then these end - chain spins exhibit physical effects of their own .
they are real spins unlike the virtual spins appearing in integer spin chains or ladders .
their physics is specially interesting when the system size is finite , and even during the process of reaching the thermodynamic limit they produce non - trivial finite - size effects along the way .
these facts difficult the technical analysis of the opening or closing of a gap in the low - lying spectrum of a 2-leg ladder with open boundary conditions .
we have solved these difficulties by analyzing the ground state and low - lying energy excitations with respect to their bulk and boundary properties such as local magnetization and the like . with this information , it is possible to identify which states contribute to the gap in thermodynamic limit . these low - lying states have a definite total spin @xmath109 and they can be targeted with the dmrg method . in this fashion , we have been able to identify the gapped or gapless behaviour of the model within the framework of the dmrg with open boundary conditions .
m. kohno , m. takahashi , m. hagiwara , `` low - temperature properties of the spin-1 antiferromagnetic heisenberg chain with bond alternation '' , physical review b , * 57 * , 1046 - 1051 ( 1998 ) .
h. watanabe , `` numerical diagonalization study of an s=1/2 ladder model with open boundary conditions '' ; phys .
b * 50 , ( 13442 - 13448 1994 ) . | we obtain the phase diagram in the parameter space @xmath0 and an accurate estimate of the critical line separating the different phases .
we show several measuments of the magnetization , dimerization , nearest neighbours correlation , and density of energy in the different zones of the phase diagram , as well as a measurement of the string order parameter proposed as the non vanishing phase order parameter characterizing haldane phases .
all these results will be compared in the limit @xmath1 with the behaviour of the @xmath2 bond alternated heisenberg chain ( bahc ) .
the analysis of our data supports the existence of a dimer phase separated by a critical line from a haldane one , which has exactly the same nature as the haldane phase in the @xmath2 bahc . |
hii galaxies , a subset of blue compact galaxies ( bcg ) , are gas rich dwarf galaxies whose optical spectra are dominated by strong and narrow emission lines .
they are currently experiencing intense star formation in small volumes .
the emission lines are produced by gas ionized by the young massive stars .
observations indicate that hii galaxies are in general metal poor systems .
this fact in addition to their very young stellar populations has been known for some time and has led to the proposal that these systems are very young , suffering their first burst of star formation @xcite .
it seems , however , that these galaxies are not as young as it was originally thought , since recent photometric observations indicate that they host stellar populations which are at least @xmath0 a. old , reaching in some cases a few ga @xcite .
there is now wide agreement about the existence of underlying populations with ages of several gas in bcgs and/or hii galaxies . even in the most metal - poor galaxy
known , izw18 , the best candidate to _ primordial galaxy _ , there is an underlying population of at least some 10@xmath110@xmath2 a @xcite .
however there is no consensus about their history of star formation .
two scenarios are postulated : \a ) the star formation takes place in short but intense episodes separated by long quiescent periods of null or low activity @xcite .
\b ) the star formation is continuous and of low intensity along the galaxy life , with superimposed sporadic bursts @xcite . in both scenarios ,
a stellar population of intermediate age contributes to the observed luminosity .
indeed , recent work takes into account the existence of an underlying stellar population and combine photo - ionization codes with evolutionary synthesis models ( hereinafter esm ) .
examples of this procedure can be found in @xcite , @xcite , @xcite and @xcite . in general the esm
so far do not include the chemical evolution .
given the low metal content of hii galaxies the metallicity effect can be of considerable relevance .
@xcite have shown that the equivalent width of the h@xmath3 line , ew(h@xmath3 ) , which can be taken as an age indicator for the ionizing population , and the ( u - v ) colour in hii galaxies show an anti - correlation which can not be interpreted in terms of a single stellar population ( ssp ) . at a given ew(h@xmath3 )
the data are displaced to colours redder than those predicted by the models .
the most likely explanation is that the ionizing population , that is the most recent burst of star formation , is overimposed on an older stellar population evolved enough as to produce a redder ( u - v ) colour .
there is also an inverse correlation between ew(h@xmath3 ) and the gas oxygen abundance , @xmath4 , which could be interpreted in terms of galactic chemical evolution : as the average age and the total mass of the stellar population increases , the ew(h@xmath3 ) decreases and the mean metallicity of the interstellar medium increases .
a number of works ( * ? ? ?
* and references therein ) have computed purely chemical evolution models ( hereinafter cem ) for the study of bcg and dwarf irregular galaxies .
most of them assume that the star formation occurs in bursts and include the effects of galactic winds and/or gas infall .
however most of them limit the study to the evolution of nitrogen and oxygen abundances @xcite and/or the luminosity - metallicity relation @xcite .
only @xcite used the information coming from the chemical evolution models from @xcite to perform the next step and combine chemical and spectral evolution .
their models exclude however the early stages of evolution , i.e. during the nebular stage when the most massive stars dominate the energy output . a code combining chemical evolution , evolutionary synthesis and photo - ionization models has not yet been developed and used for the spectral analysis of bcg and hii galaxies . in this work ,
we include the chemical evolution model results in the computation of the spectral energy distributions which are used as ionizing source for a photo - ionization code .
our method allows the simultaneous use of the whole available information for the galaxy sample concerning on the one hand the ionized gas emission lines intensities and equivalent widths , elemental abundances , gas densities which defines the present time state of the galaxy , and on the other hand spectro - photometric parameters colours , absorption spectral indices defined by the stellar populations and their evolution with time and thus related to the galaxy star formation history .
this is done in a self - consistent way , that is using the same assumptions regarding stellar evolution , model stellar atmospheres and nucleosynthesis , and using a realistic age - metallicity relation .
the basic codes employed are described in section 2 .
section 3 shows the corresponding results of each kind of model , and the whole combined results about galaxy evolution .
these results are discussed in section 4 .
finally , our conclusions are summarized in section 5 .
the code used for this work is a generalization of the code used for the solar neighborhood by @xcite and applied to the whole galaxy disc in @xcite .
that model described the galaxy as a two - zone system ( halo and disc ) in which the disc is a secondary structure formed by the gravitational accumulation of gas from the halo . in the present work , however , all the gas is assumed to be within an only region from @xmath5 , that is , the collapse time scale has been eliminated as an input parameter of the code .
the matter is assumed to be in different phases : * a stellar population , where we distinguish the stars able to create and eject heavy elements to the interstellar medium ( @xmath6 ) , and the stars which only eject hydrogen and he ( @xmath7 ) . * stellar remnants , as the endpoint of star evolution , that act as a matter sink , removing mass from chemical evolution .
* interstellar diffuse gas out of which stars are forming following a simple schmidt law .
no molecular component is explicitly considered in the present work .
a bursting star formation has been assumed taking place in successive bursts along the time evolution . at any time , the available gas for each burst is the sum of the gas left after the previous burst of star formation and the gas ejected by massive stars . according to this description , the time dependence of the total mass fraction in each phase and the chemical abundances in the ism are determined by the interactions between these phases and , as a consequence of the evolution , we get the star formation rate .
the basic equations to study the behavior of the mass of stars and gas are , therefore : @xmath8 where @xmath9 is the mass in stars , m@xmath10(t ) is the mass of gas , @xmath11 is the total mass of the system , @xmath12(t ) the star formation rate and e(t ) the ejection rate of mass from the stars to the ism .
a part of the restituted gas consists of enriched material , e@xmath13(t ) , which is created in the interior of the stars and is ejected when they die . to compute both quantities , ejected gas and element production
, we use the _ @xmath14 matrix _ formalism introduced by @xcite and widely used in classical chemical evolution models .
it is very useful to treat elemental abundances which increase at different rate in the interstellar medium , that is , the non - solar elemental ratios @xcite . in the present work
we adopt the same matrix prescription as in @xcite , which is an updated version from the one given in @xcite and @xcite calculated following the method described by @xcite . in order to calculate these matrices
it is necessary to consider the nucleosynthesis of stars , their initial mass function ( imf ) and their end either as type i or ii supernovae , or through quiet evolution .
the mean stellar lifetimes , final states , and nucleosynthesis products , depend on the stellar mass which makes the imf to play an important role in the chemical evolution model .
the adopted imf is assumed to be universal in space and constant in time @xcite and taken from @xcite .
this imf is very similar to scalo s law @xcite and in good agreement with the recent expressions from @xcite and @xcite .
nucleosynthesis yields for massive stars have been taken from @xcite . for low and intermediate mass stars ,
we have used the yields from @xcite .
the combination of these sets of stellar yields , with our assumed imf producing the required yields per stellar generation , has been completely revised and calibrated with the mwg in @xcite , where other stellar yields sets have been used and analyzed too .
it also reproduces observations of spiral and irregular galaxies @xcite , obtaining reasonable results even for nitrogen abundances in low - metallicity objects @xcite . for type i supernova explosion releases , model w7 from @xcite , as revised by @xcite , has been taken , while the rate of this type of snia explosions are included through a table given by ruiz - lapuente ( private communication ) , following works of @xcite .
we have assumed that the gas is used to form stars with a given efficiency called h : @xmath15 . if there would not be gas ejection from massive stars , as it would be the case for instance in the firsts moments of the evolution , @xmath16 , then : @xmath17 that is
, the consumed gas rate would be a function that depends on the available gas at a given time , this expression is modified , as we will show in section 3 . ]
being a decreasing function of time through the parameter h , which defines the star formation efficiency . then : @xmath18 a change in the star formation rate implies a change in the value of the parameter h or efficiency : @xmath19 we have run models considering the star formation as a set of successive bursts in a region with a total mass of gas of 100@xmath2010@xmath21 m@xmath22 and a radius of 500 pc ( see section 3.2 for details ) .
every galaxy experiences 11 star formation bursts along its evolution of 13.2 ga , that is one every 1.3 ga . in each burst a certain amount of gas is consumed to form stars .
we have computed two types of models . in the first one , which we call _ burst _ models ( models b ) , the same efficiency h is assumed for every burst while in the second one ( models a ) we assume an attenuated bursting star formation mode , _ i. e. _ a variable h is assumed along the time evolution .
for each type we have computed two models with different initial star formation efficiency : 1 .
model 1 , that uses @xmath23 64 per cent of the gas for star formation in each time step .
thus , following eq.(8 ) , @xmath24 ma 2 .
model 2 , that uses @xmath23 33 per cent of the gas to form stars in each time step , which implies @xmath25 ma these percentages are maintained in models b , which means that every burst will transform the corresponding percentage of the available gas ( taking into account the gas ejected by massive stars ) in stars . in models a , these values correspond to the initial efficiencies and are reduced in the successive bursts by a factor @xmath26 , where @xmath26 is the number of the burst .
the code solves the chemical evolution equations to obtain , in each time step , the abundances of 15 elements : h , d , @xmath27 , @xmath28 , c , @xmath29 , o , n , ne , mg , si , s , ca , fe , and _ nr _ where _ nr _ are the isotopes of the neutron rich elements , synthesized from @xmath30 , @xmath29 , @xmath31n and @xmath32 inside the co core . we have taken a time step , @xmath33t= 0.5 ma , from the initial time , @xmath5 , up to the final one , @xmath34 ga .
also , at each time step , the star formation rate and the mass in each phase low mass , massive stars and remnants , total mass in stars created , and mass of gas are computed .
this amounts to a total of 26348 stellar _ generations _ or ssps , each one with its corresponding age and metallicity .
once the chemical evolution models have been calculated , the spectrophotometric properties can also be obtained . in this work
we have used the code from @xcite as updated in @xcite .
the required inputs are the stellar isochrones and the stellar atmosphere models for the individual stars . in that work we used the set of 50 isochrones from the padova group from @xcite for ages between 4 ma and 20 ga , for 5 metallicities between 1/50 and 2.5 z@xmath22 ( [email protected] , 0.004 , 0.008 , 0.02 and 0.05 ) .
to this set of isochrones we have added other 18 isochrones from @xcite and @xcite with ages from 0.5 to 4 ma , used in @xcite and @xcite for the youngest stellar populations of 4 metallicities [email protected] , 0.008 , 0.02 and 0.05 .
that means that for the lowest metallicity , @xmath36 , only isochrones older than 4 ma are available . in both cases ,
isochrones give the number of stars in each phase assuming that stellar clusters were formed with a standard salpeter imf with @xmath37 m@xmath22 and m@xmath38 m@xmath22 .
this imf , although using the same lower and upper limits , is not exactly the one used in the multiphase chemical evolution code , what implies some differences in the resulting spectra , mainly for the oldest and the youngest stellar populations .
they are , however , smaller than 10 per cent ( garca - vargas et al .
2008 , in preparation ) , and therefore we assume that this inconsistency will not produce very important differences in the final results .
the emergent spectral energy distribution ( sed ) was synthesized by calculating the number of stars in each point of the h - r diagram and assigning to it the most adequate stellar atmosphere model .
the models of @xcite and @xcite have been used for stars with @xmath39 50000 k ( last evolutionary stages of massive stars ) and 2500 @xmath40 k , respectively . to each sed computed for a stellar cluster
, it was added a continuum nebular emission contribution as explained in @xcite .
the gas is assumed to have an electron temperature , t@xmath41 which depends on z. the assumed values are t@xmath4218000 , 11000 , 9000 , 6500 and 4000 k for [email protected] , 0.004 , 0.008 , 0.02 and 0.05 , respectively , selected according to the average value obtained by @xcite .
the free - free , free - bound emission by hydrogen and neutral helium , and the two photon hydrogen - continuum have been calculated by means of the atomic data from @xcite and @xcite .
the seds for the complete sets of 68 ( 50 for z=0.0004 ) isochrones were calculated with the same code as in @xcite0.0004 , were not used ] for the cited metallicities .
therefore we finally have a set of 68 seds for ssp of different ages for 4 metallicities each , and 50 for ages older than 4 ma for [email protected] , which we use as a spectral library .
once the seds for ssps are computed , they must be convolved with the star formation history ( sfh ) in order to calculate the corresponding sed for a region where more than a burst take place : @xmath43 where @xmath44 is the age of the stellar population created in a time @xmath45 and @xmath46 being the sed for each ssp of age @xmath47 and metallicity z reached in that time @xmath45 .
a sed from the ssp library , @xmath46 , must assigned to each time step according to its corresponding age and metallicity taking into account the sfh , @xmath48 , and the amr , @xmath49 , obtained from the chemical evolution model , to finally calculate l@xmath50 by the above integration .
however the metallicity changes continuously while the available seds have only 4 or 5 possible values .
we have , therefore , interpolated logarithmically between the two ssp of the same age @xmath47 closest in metallicities to z(t ) to obtain the corresponding @xmath51 .
the final result is the total luminosity at each wavelength @xmath52 . when this process is applied to the stellar population created by the first burst two problem
arise . on the one hand
the initial metallicity is z@xmath53 and , although during the first ma after the creation of the first stars the metallicity increases , it does not reach the minimum [email protected] , and therefore we must extrapolate with the available seds of the youngest ssps . due to the uncertainties in the evolution of the _ quasi_-zero - metallicity stars , we have preferred to assign to this first burst metallicity a minimum value of 0.0028 , smaller than the one reached at the end of the first burst but still valid for extrapolating without creating instability numerical problems . on the other hand we must to use two different pairs of metallicities to extrapolate with the seds of ssps : [email protected] and 0.008 before 4 ma and [email protected] and 0.004 after 4 ma , since seds for the most - metal - poor and youngest ssps are not available .
we have checked that this method does not produce discontinuities and that the resulting spectra behave smoothly . in this way
we obtain the sed corresponding to the whole stellar population , including the ionizing continuum proceeding from the last formed stellar population .
the photo - ionization code cloudy , in its 96.0 version @xcite , has been used in order to obtain the emission lines .
the gas is assumed to be ionized by the massive stars belonging to the current burst of star formation whose sed has been previously calculated by the combination of the chemical and evolutionary synthesis models .
the shape of the ionizing continuum is defined by the pair of values ( @xmath54(ryd ) , log@xmath55 ) ( eq .
[ sp ] ) .
we have assumed the emitting gas to be located at a distance @xmath56 pc at the beginning of the evolution .
this size is characteristic of hii galaxies @xcite .
this radius is subsequently adjusted as to keep a constant stellar density configuration .
this results in all cases in a plane - parallel geometry .
a given model is characterized by its ionization parameter u defined as : @xmath57 where n@xmath58 is the electron density , assumed constant for simplicity and equal to 100 @xmath59 , and @xmath60 is the number of lyman ionizing photons , calculated from the cluster sed , which reach the gas at a velocity @xmath61 .
finally , the gas elemental abundances are those reached at the end of the starburst previous to the current one , as calculated from the chemical evolution model , except for na , ar and ni which are not computed in the model and are scaled to the solar ratio @xcite .
the star formation rate is one of the main results of the different models since it drives the behaviour of all the other quantities . in fig .
[ sfr ] the star formation history for each of the 4 computed models is shown . in all of them
the first burst is strong , while the successive bursts are less intense since the amount of gas available has decreased in spite of the ejected gas produced by the massive stars . in models a ( right panels ) , the first burst has the same star formation rate as in models b ( left panels ) , since the initial efficiency is the same for both .
the successive star formation episodes have a smaller star formation rate in models a than the corresponding ones in models b due to the attenuation factor included in the inputs .
@xcite have estimated the star formation rates for the sample of 39 local hii galaxies of different morphologies from @xcite , which were selected from the most luminous of the @xcite catalog and with h@xmath3 equivalent widths between 30 and 280 .
these star formation rates were obtained from the h@xmath62 luminosity , by assuming t@xmath63 k and case b recombination .
they are in the range from 0.5 to 7 m@xmath64 , marked by dotted lines in fig .
[ sfr ] , indicating that these hii galaxies have high star formation rates .
these findings are in agreement with those from @xcite who found star formation rates in the range of 10@xmath65 to 10 m@xmath66 .
our models produce sfr between the observed values after the sixth burst , which corresponds to 6.5 ga after the beginning of star formation , except model b2 , where every burst before the ninth - tenth show higher sfr than observed . since the star formation consumes gas , the averaged gas mass decreases with time following eq.(6 ) , as can be seen in fig .
the chemical evolution code , however , takes into account the mass of gas ejected from massive stars during their evolution and , therefore , the gas mass increases during a given burst . on the other hand , the mass of stars which increases in each burst and globally due to the star formation , decreases during each burst as the most massive stars end their lives
it is interesting to note that models with smaller efficiencies , h , in the lower panels of figs .
[ sfr ] and [ mass ] , consume less gas in each burst and , as a consequence , the star formation rate remains higher , and declines more slowly , than in models with higher values of h shown in the upper panels .
the element content of the gas is usually measured through the oxygen abundance , given as @xmath67 .
oxygen contributes more than 40 per cent to the mass in metals and is mainly created within massive stars . since these stars are very short lived , their products are ejected very rapidly to the interstellar medium ( ism ) and hence the value of its oxygen abundance increases with time , as shown in fig .
[ oh ] . in that figure
we see how o / h increases abruptly when a burst takes places remaining more or less constant between each two of them .
observational data from @xcite and @xcite show that most of the hii galaxies have metallicity distributions between @xmath68 .
these limits are shown as dashed lines in fig .
the first limit is reached with the first burst in all cases .
it can be seen that models of type 1 ( dotted lines ) show oxygen abundances higher than the observed upper limit .
the efficiency of the first burst is so high that a large number of massive stars are formed , and hence the oxygen abundance reaches rather high values in a very short time .
also the oxygen abundance of model b2 reaches values higher than shown by data from the third burst onwards ( @xmath69 ) .
only model a2 shows oxygen abundances within the range of hii galaxies during the whole evolution since the attenuation of the bursts keeps the star formation lower than in the other three cases . the star formation efficiency taken for model a2 results to be an upper limit for hii galaxies in oxygen abundances and it may reproduce the most metal rich systems .
abundance ratios also give information about the star formation process , essentially the time scale or duration of the bursts , when the production of the elements involved takes place in stars in different mass ranges .
this is the case for the n / o ratio .
oxygen is very quickly ejected by the most massive stars .
nitrogen , however , may be created in stars of all masses . moreover
, the production of nitrogen may take place in two ways : 1 ) the main process requires another element like carbon or oxygen , to synthesize n through the cno cycle , thus n shows , at least in part , a secondary behavior . in that case
the n abundance has to be proportional to the initial abundance of heavy elements and hence the n / o ratio grows with metallicity .
there is , however , a proportion of nitrogen which must have a primary origin , as suggested by the data in fig . [ no ]
@xcite shown by ( green ) dots which exhibit a constant n / o ratio .
nitrogen may be produced by low and intermediate mass stars @xcite and therefore its contribution may appear in the ism a certain time after the massive stars have died . in fact , n grows more slowly than the oxygen abundance in the range @xmath70 , which may be explained by this primary behavior ( see * ? ? ? * for details ) . in fig .
[ no ] we can see the oscillating behavior of the n / o ratio shown by our computed models due to the successive bursts of star formation followed by quiescent periods . at the beginning of a given burst , the n / o ratio decreases , as the o is created and ejected by the most massive stars in the burst
; then it increases during the quiescent periods between bursts as the low and intermediate mass stars die and eject their synthesized nitrogen while the oxygen abundance remains constant .
according to the results of this section , model a2 is the one better reproducing the observations , since the total mass , the gas fraction , the star formation rate and the oxygen abundances are within the range shown by data .
therefore , from now onwards , in what regards the ionizing continuum and emission line properties only the a2 model will be analyzed .
the convolution of the results of the chemical evolutionary code with the single stellar populations of the synthesis evolutionary code gives us directly the sed , @xmath71 , in time steps of logt@xmath35 0.05 . from this
sed the number of hydrogen ionizing photons , coming from hot and young stars of the first spectral types ( o - b ) , can be calculated as : @xmath72 where l@xmath73 is the continuum luminosity at frequency @xmath54 .
[ foton ] shows the number of hydrogen ionizing photons as a function of time for model a2 . in the first burst , with a total mass of @xmath74 m@xmath22 forming stars ,
the number of massive ionizing stars is high and so it is the number of ionizing photons .
the number of ionizing photons per unit stellar mass in our models is @xmath75 at the starting of the first burst , corresponding to a zero age stellar population ( zasp ) of very low metallicity and then decreases as the cluster ages .
this initial number of ionizing photons per unit stellar mass is almost the same for the successive bursts although , decreasing slightly as the abundance increases . within each burst the number of ionizing photons decreases by almost two orders of magnitude from 0.5 to 10 ma , as show in fig
. [ foton - burst ] , and by almost 8 orders of magnitude by 100 ma .
essentially no ionizing photons are available during the quiescent inter - burst periods . at the beginning of each successive burst
, the number of ionizing photons raises abruptly but reaching a value progressively lower as less gas is available to form stars .
also , as the mean gas metallicity increases with time , the number of ionizing photons per unit stellar mass decreases @xcite .
this effect increases with the age of the ionizing cluster reaching a factor of about 5 at 10 ma .
the average number of ionizing photons observed for local hii galaxies is @xmath76 @xcite .
this number of ionizing photons , marked as a dotted line in fig .
[ foton ] , is reached by model a2 at around the fifth or sixth burst of star formation , that is for the burst occurring 5.2 - 6.5 ga after the beginning of the formation of the galaxy , when the predicted star formation rate also fits the data ( see fig .
[ sfr ] ) . on the other hand ,
the average derived ionization parameter for the local hii galaxy sample is @xmath77 which in our models corresponds to an age of the ionizing population of about 6.5 ma .
it should be noted that during the first 7 ma of a burst the number of ionizing photons remain essentially constant ( see fig .
[ foton - burst ] ) .
therefore , if the current burst is identified with redshift @xmath78 , the formation of the galaxy would have taken place at a redshift @xmath79 .
also the observed number of ionizing photons for lcbgs at a redshift @xmath80 , estimated as @xmath81 @xcite , would correspond to the value of the first burst of our models . according to these results
, our model seems consistent with observations at these intermediate redshifts reproducing at the same time those corresponding to the more luminous local hii galaxies .
the question of if these galaxies are young galaxies , experiencing their first burst of star formation or if , on the contrary , there exists an underlying population formed some gas ago , may be addressed by multi - colour photometry .
although a direct detection of the underlying population in lcbg or hii galaxies is complicated because the observed light is dominated by massive stars , the effect of an underlying stellar population should be easily seen in the observed colours of these galaxies , since any star forming burst previous to the currently observed one will contribute substantially to the total continuum luminosity at the different wavebands . in fig .
[ colors ] we represent the time evolution of the continuum u - b , b - v , v - r and v - i colours for our model a2 along the first 10 ma after each burst .
the first burst solid black thick line follows the expected evolution of a single stellar population of very low metallicity , remaining bluer than during the subsequent bursts where the effect of the accumulated continua from the previous star formation episodes makes colours become redder than expected for a ssp . even though they are also starbursts in themselves , their star formation efficiency is much lower than that of the first burst , and so is their contribution to the total continuum luminosity which results in redder colours .
furthermore , the metallicity , almost zero for the first burst , increases up to @xmath82 already during the second burst which also has an reddening effect over the colours .
the calculated sed in each time step can be used as input ionizing sources in the photo - ionization code to predict the emission lines intensities which provide information about the youngest stellar population which dominates the final spectra .
[ o3o2 ] shows the relation between the @xmath83\lambda\lambda$ ] 4959,5007/h@xmath3 and [ oii]@xmath84 3727,3729/h@xmath3 ratios , which constitutes an excitation diagnostics for ionized nebulae . in this diagram ,
the most excited objects lie up and to the left while the objects with the lowest excitation are at the bottom right .
the data shown in the graph have been extracted mainly from two main sources .
first , the compilation from @xcite which provides emission line measurements , corrected for extinction , published for local hii galaxies .
the sample comprises 450 objects and constitutes a large sample of local hii galaxies with good - quality spectroscopic data .
the sample is rather inhomogeneous in nature , since the data proceed from different instrumental setups , observing conditions and reduction procedures , but have been analysed in a uniform way .
data for these sample objects include the emission line intensities of : [ oii]@xmath85 3727,29 , [ oiii]@xmath85 4959,5007 , and [ nii]@xmath85 6548,84 , all of them relative to h@xmath3 , and the equivalent widths of the [ oii ] and [ oiii ] emission lines ew([oii ] ) ew([oiii ] ) , and the h@xmath3 line , ew(h@xmath3 ) . as a second source , we have used the metal poor galaxy data from the data release 3 of sloan digital sky survey , taken from @xcite .
the sloan digital sky survey @xcite constitutes a large data base of galaxies with well defined selection criteria and observed in a homogeneous way .
the sdss dr3 @xcite provides spectra in the wavelength range from 3800 to 9300 for @xmath23 530000 galaxies , quasars and stars . @xcite
extracted @xmath23 2700 spectra of non - active galaxies with the [ oiii]@xmath524363 emission detected above 1@xmath86 level .
this initial sample was further restricted to the objects with an observed flux in the h@xmath3 emission line larger than @xmath87 and for which accurate abundances could be derived .
they have also excluded all galaxies with both [ oiii]@xmath524959/h@xmath3 @xmath88 0.7 and [ oii]@xmath523727/h@xmath3 @xmath89 1.0 . applying all these selection criteria
, they obtain a sample of @xmath23 310 sdss objects .
data for these sample objects include the emission line intensities of : [ oiii]@xmath524959,5007 and [ nii]@xmath526584 relative to h@xmath3 and the equivalent width of h@xmath3 .
they also include the intensity of the [ oii ] @xmath85 3727,29 emission line for the lowest redshift objects . the very young stellar population during the first ma after each burst produces a high ionization , so the [ oiii]/h@xmath3 ratio is high , then it decreases raising again at 4 ma due to the presence of wolf rayet stars that produces a harder continuum @xcite . after 5 ma
, the burst evolves to lower values of @xmath90/h\beta$ ] .
diagonal lines in this plot correspond to nearly constant ionization parameter and are swept along by the different successive bursts of a given age .
high excitation objects , with high @xmath90/h\beta$ ] and low @xmath91/h\beta$ ] ratios , have low metallicities and they are not reproduced by our model , which fits better the high metallicity sample , in the middle to right side of the panel .
it can be better seen in fig .
[ nho3 ] , which shows the relation between the excitation parameter , @xmath92/h\beta)$ ] , and the metallicity indicator @xmath93/h\alpha)$ ] which gives information about the younger populations plus the star formation rate @xcite . at the beginning of the evolution , as we saw in previous sections , the star formation rate is high , and the h@xmath62 emission of our model is strong . as the galaxy evolves , the h@xmath62 emission decreases while the [ nii ] emission increases due to the growth of metallicity .
our models reproduce the general trend shown by data but can not reproduce the lower metallicity ( low @xmath94/h_{\alpha}$ ] ratio ) objects to the left of the graph .
this is due to the fact that the efficiency of the first burst , 33.1 per cent , produces a metallicity which is already too high . a lower efficiency is therefore required to reproduce the observations of the less metallic objects . once the continuum spectral energy distributions and the different emission line fluxes are computed
, we can calculate the contribution of these emission lines to the different broad band filters and then synthesize the colours of our model galaxies .
these are the colours which are readily observable through integrated photometry .
we have done so taking into account only the strongest emission lines that contribute to the colour in each broad - band spectral interval at redshift zero .
these are : [ oii ] @xmath85 3727 , 3729 in u , h@xmath3 @xmath52 4861 in b , [ oiii]@xmath85 4959 , 5007 in v , h@xmath62 @xmath52 6563 in r and [ siii ] @xmath85 9069 , 9532 in i. the continuum colours and the colours including the emission line contribution at redshift zero are given in columns 3 to 10 of table 1 for each of the models characterized by their burst number and age ( columns 1 and 2 ) .
columns 11 to 15 of the same table give the contribution , in percentage , by the emission lines to the different continuum fluxes .
the necessary information to calculate these contributions for any other redshifts in the nearby universe ( z @xmath950.1 ) , is given in table 2 which lists for each of the models characterized by their burst number and age ( columns 1 and 2 ) , the continuum fluxes in each of the u , b , v , r and i ( columns 3 to 7 ) and the fluxes in the [ oii ] , h@xmath3 , [ oiii ] , h@xmath62 and [ siii ] lines ( columns 8 to 12 ) . figure [ colors ] shows , with dotted lines , the new computed colours including the contribution by the different emission lines . as it can be seen the colours do not change by a large amount in the case of u - b or v - r .
however the effect in the b - v and v - i colours is rather dramatic .
figure [ colours - lines ] shows the b - v vs v - i colour - colour diagram for model a2 .
light blue circles show the continuum colours ( solid lines in figure [ colors ] ) while black solid ones show the colours computed including the contribution by the stronger emission lines given in table 1 .
the inclusion of the contribution by the emission lines to the continuum colours shifts the position of the model points almost perpendicularly to the originally computed ones .
the location of blue points is mainly determined , even at zero redshift , by the contribution of strong [ oiii]4959,5007 emission lines to the v band . in the same diagram
we show the data by @xcite on bcg as red squares with error bars .
these correspond to readily observed colours therefore including both continuum and line emission .
no reddening correction has been applied .
it can be seen that some of the data are impossible to reproduce by the models which do not take into account the contribution by emission lines , even if some amount of reddening is invoked ( the arrow in the diagram shows the correction corresponding to 1 mg visual extinction ) , while a certain amount of reddening would bring most of the data in agreement with the models . on
he other hand the data by @xcite of resolved locations in mrk 370 , shown in the figure as green triangles , correspond to data that have been corrected for the contribution by emission lines and therefore correspond to pure continuum colours . in these case , the agreement with the models is excellent .
the evolution of hii galaxies must be considered by looking simultaneously at both continuum and emission line properties , the first one providing information on a long time scale evolution , of the order of ga , and the second one being related to the evolution in a much shorter time scale , of the order of ma .
@xcite suggested the use of the equivalent width of h@xmath3 , ew(h@xmath3 ) , as an age estimator for hii regions , and applied this method to rank the ages of hii regions in the magellanic clouds .
if a ssp is considered , the h@xmath3 emission is very high at the beginning of the burst , while the continuum at h@xmath3 , dominated by the most massive and luminous stars , is low hence producing a large value of ew(h@xmath3 ) .
as the burst evolves , the h@xmath3 luminosity decreases as does the number of ionizing photons , and the contribution to the continuum increases thus lowering the value of ew(h@xmath3 ) .
this case corresponds to the thick ( black ) line in fig .
[ hbeta ] . the evolution of the equivalent width in our model is shown in fig . [ hbeta ] .
it is seen that the first burst , that corresponds to a ssp , has an initial equivalent width slightly greater than the successive bursts , which decreases with time reaching very low values ( @xmath88 40 ) 6 ma after the beginning .
when a new burst takes place , new massive ionizing stars form and the ew(h@xmath3 ) goes up again , although not as much as in the previous one due to the decreasing gas mass involved in the burst , increased metallicity and the accumulated contribution from the continua from the previous bursts . at any rate , even taking into account this contribution , the ew(h@xmath3 ) remains a good age indicator for the age of the current burst , inside 10 ma . on the other hand
it is not possible to detect the presence of the underlying stellar populations for a given galaxy solely on the basis of this parameter .
the distribution of the ew(h@xmath3 ) in hii galaxies shows that most of them have values lower than 150 @xcite which , in our models , corresponds to ages greater than 3 - 4 ma .
@xcite , from the application of inversion methods , have shown that , globally , this distribution is inconsistent with what is expected from a single burst scenario and resembles more the distribution obtained for a succession of short bursts separated by quiescent periods with little or no star formation .
these results are not that strong when samples showing a restricted metallicity range are analyzed .
however , as we have seen before , metallicity increases with time in a way which constrains the sfh which is best reproduced by our a2 model .
[ hbeta_ox ] shows the evolution of the @xmath96 with oxygen abundance . during the duration of the first burst the @xmath96 decreases while the oxygen abundance increases from zero to @xmath97 .
the successive bursts start with this higher abundance and each of them evolves vertically in a 10 ma time scale with the @xmath96 decreasing rapidly at almost constant oxygen abundance .
this occurs because the oxygen is produced by the most massive stars in a very short time and therefore its abundance changes very little during the next 10 ma after each burst ( see fig .
it is clear that our model is able to reproduce the most metallic galaxies in the samples taken from @xcite and @xcite where this high abundance is explained as the consequence of the gas being processed by one or more previous generations of stars .
an alternative scenario to produce such high oxygen abundances would be the occurrence of a very intense initial star formation burst , with a sfr of the order of 100 m@xmath22 ma@xmath98 , out of the range observed in local hii galaxies ( see figures [ oh ] and [ sfr ] ) .
on the other hand , galaxies @xmath99 are not reproduced by our model . in order
to fit these data , a model with a star formation efficiency lower than assumed , that is , @xmath100 33 per cent , should be used . the most informative data to uncover the presence of underlying populations in hii galaxies consist of the combination of a line emission parameter , which characterizes the properties of the current burst of star formation , and a continuum colour which represents better the sfh in a longer time scale .
[ beta - cont ] , left panel , shows the equivalent width of h@xmath3 _ vs _ the u - v colour along the evolution of the galaxy for the successive stellar bursts . when a given burst occurs the ew(h@xmath3 )
is high and decreases as the stellar population ages .
we have over - plotted the results for the ssps from the model starburst99 ( * ? ? ?
* sb99 ) with two different metallicities as labelled . as expected , our results agree better with the higher metallicity sb99 model .
the black line in the left panel , that represents a metal - poor ssp , is too blue compared with the data for a given ew(h@xmath3 ) . in order to decrease ew(h@xmath3 ) and
move the model colour to the red , a more metal - rich ssp was selected ( dotted line at the left panel ) , but even such unrealistic high abundance does not reproduce the observations , as we have shown in fig .
therefore , the observed trend can not be explained as an age effect since ew(h@xmath3 ) for low metallicity ssp does not reach values lower than 100 and does not reproduces the colors , neither as a metallicity effect because such high abundances are not observed in hii galaxies .
it is therefore necessary to include both effects simultaneously , as we have done , to see how the effect of an underlying older population contributing to the color of the continuum makes it redder .
from the observational point of view ,
u - v colours for hii galaxies are scarce , therefore we have compared our results with those from @xcite that include galaxies taken from @xcite and @xcite providing the equivalent widths of both the [ oii ] and [ oiii ] lines .
we have then computed pseudo - colours from the intensities of the adjacent continua of [ oii]@xmath523727 and [ oiii]@xmath525007 lines as : @xmath101\lambda 3727}(\rm \aa)= \frac{[oii]\lambda 3727 ( erg s^{-1})}{i_{cont}(3730)(erg s^{-1}{\aa}^{-1})}\ ] ] and @xmath102\lambda 5007}(\rm \aa)= \frac{[oiii]\lambda 5007(erg s^{-1})}{i_{cont}(5010)(erg s^{-1}{\aa}^{-1})}\ ] ] @xmath103\lambda 3727 } { [ oiii]\lambda 5007 } \frac{ew_{[oiii]\lambda 5007}}{ew_{[oii]\lambda 3727 } } \right)\ ] ] the right panel of fig .
[ beta - cont ] shows the equivalent width of h@xmath3 as a function of the pseudo - colour defined above for model a2 together with the observational data .
it should be noted that no correction for extinction associated with the continuum light has been made .
this correction amounts to about 0.15 dex for a standard reddening law and therefore a ssp would always be unable to reproduce the observed colours . yet
our model a2 produce uv continuum colours which are too blue as compared with the bulk of the observations which indicates that our assumed degree of attenuation is too low .
we have explored the viability of a kind of a theoretical model in order to understand how the star formation takes place in hii galaxies .
our work involves the combination of three different tools : a chemical evolution code , an evolutionary synthesis code and a photo - ionization code , in a self - consistent way , _
i. e. _ taking the same assumptions about stellar evolution and nucleosynthesis , and takes into account the resulting metallicity in every time step .
the chemical evolution code provides the star formation history of the galaxies as well as the time evolution of the abundances in their interstellar medium , that can be confronted with observations .
the evolutionary synthesis code provides the spectrophotometric evolution of the integrated stellar population .
finally , the photo - ionization code uses the properties of the ionizing stellar population to calculated the emission line spectra of the ionized gas .
each galaxy is modelled assuming an initial amount of unprocessed gas of 10@xmath1 m@xmath22 in a 1 kpc diameter .
the evolution is computed along a total duration of 13.2 ga during which 11 successive starbursts , separated by 1.3 ga , take place .
different types of models have been computed with different starbursting properties : equal and attenuated bursts , and different values of the initial star formation efficiency .
the only model which has proved to be able to fit the observational data is model a2 , which consists of attenuated bursts with an initial star formation efficiency of 33.1 per cent .
this type of model reproduces the star formation rate estimates of local hii galaxies and the oxygen abundances of the average metallicity objects although fails to account for the most metal deficient ones which would require models with lower star formation efficiencies than assumed .
we have computed the time evolution of the continuum colours with and without the inclusion of the emission lines .
it is shown that both nebular continuum and line emission must be taken into account in the photometric studies of the underlying stellar populations of hii galaxies .
the effect of the emission lines are more pronounced in the b - v and v - i colours .
the combination of parameters which characterize the current star formation on a time scale of ma , such as the equivalent width of h@xmath3 , and the sfh over a time scale of ga , such as broad - band continuum colours , is found to provide an effective means to uncover the presence of underlying stellar populations .
the comparison of models and observations show that in most hii galaxies ssp are unable to reproduce the relatively red colours shown by data with the observed h@xmath3 equivalent width values requiring the contribution of previous stellar generations .
our model , however , produces u - v colours that are too blue when compared with observations , therefore implying that the successive bursts that take place in hii galaxies should be even more attenuated than has been assumed in our work .
this work has been partially supported by dgicyt grant aya-2004 - 02860-c03 .
aid acknowledges support from the spanish mec through a sabbatical grant pr2006 - 0049 . also , partial support from the comunidad de madrid under grant s-0505/esp/000237 ( astrocam ) is acknowledged .
support from the mexican research council ( conacyt ) through grant 19847-f is acknowledged by rt .
we thank the referee for all the comments and suggestions for a better understanding of this work .
we thank elena terlevich for a careful reading of this manuscript and the hospitality of the institute of astronomy in cambridge where this paper was partially written . | we have computed a series of realistic and self - consistent models of the emitted spectra of hii galaxies .
our models combine different codes of chemical evolution , evolutionary population synthesis and photoionization .
the emitted spectrum of hii galaxies is reproduced by means of the photoionization code cloudy , using as ionizing spectrum the spectral energy distribution of the modelled hii galaxy , which in turn is calculated according to a star formation history ( sfh ) and a metallicity evolution given by a chemical evolution model that follows the abundances of 15 different elements .
the contribution of emission lines to the broad - band colours is explicitly taken into account .
the results of our code are compared with photometric and spectroscopic data of hii galaxies .
our technique reproduces observed diagnostic diagrams , abundances , equivalent width - colour and equivalent width - metallicity relations for local hii galaxies .
[ firstpage ] galaxies : abundances galaxies : evolution galaxies : starburst galaxies : stellar content |
our knowledge of background cosmology has recently improved dramatically due to new supernovae and cosmic microwave background data .
current observations favor a flat universe with matter density @xmath4 @xcite and the remaining contribution in the form of cosmological constant @xcite or some other form of dark energy .
the models with cosmological constant are known , however , to suffer from two major problems .
one is related to the origin of the constant - it can not be explained in terms of the vacuum energy since its energy is orders of magnitude smaller .
the other is the lack of explanation why the present densities in matter and cosmological constant are comparable
. a new class of models that solve these problems and also satisfy present observational constraints has been proposed a few years ago @xcite . in these models
the cosmological constant is replaced with a new energy component , called quintessence , characterized by the equation of state @xmath5 .
the component can cluster on largest scales and therefore affect the mass power spectrum @xcite and microwave background anisotropies @xcite .
the investigations of the physical basis for the existence of such component are now more than a decade old @xcite .
one of the promising models is based on so - called tracker fields " that display an attractor - like behavior causing the energy density of quintessence to follow the radiation density in the radiation dominated era but dominate over matter density after matter - radiation equality @xcite .
it is still debated , however , how @xmath3 should depend on time , and whether its redshift dependence can be reliably determined observationally @xcite .
a considerable effort has gone into attempts to put constraints on models with quintessence and presently the values of @xmath6 seem most feasible observationally @xcite . here
i review the main characteristics of structure formation in the quintessential universe which may provide constraints on the equation of state . in the last section
i discuss the current status of observational limits on @xmath3 and future perspectives .
quintessence obeys the following equation of state relating its density @xmath7 and pressure @xmath8 @xmath9 the case of @xmath10 corresponds to the usually defined cosmological constant . the evolution of the scale factor @xmath11 ( normalized to unity at present
, @xmath12 is the redshift ) in the quintessential universe is governed by the friedmann equation @xmath13 where @xmath14^{-1/2}\ ] ] and @xmath15 is the present value of the hubble parameter .
the quantities with subscript @xmath16 here and below denote the present values .
the parameter @xmath17 is the standard measure of the amount of matter in units of critical density and @xmath18 measures the density of quintessence in the same units @xmath19 the einstein equation for acceleration @xmath20 shows that @xmath21 is needed for the accelerated expansion to occur .
the left panel of fig .
[ aodtage ] shows the evolution of scale factor in different models .
the right panel presents the dependence on @xmath3 of the present age of universe @xmath22 solving the equation for the conservation of energy @xmath23 with condition ( [ q1 ] ) we get the evolution of the density of quintessence which for @xmath24 , the case considered in this paper , reduces to @xmath25 the evolution of @xmath17 and @xmath18 with scale factor is given by @xmath26 while the hubble parameter itself evolves so that @xmath27 $ ] .
the linear evolution of the matter density contrast @xmath28 is governed by equation @xcite @xmath29 where dots represent derivatives with respect to time . for flat models and arbitrary @xmath3 an analytical expression for @xmath30 , the growing mode of the time - dependent part of @xmath31 ,
was found @xcite . with our notation and the normalization of @xmath32 for @xmath33 and
@xmath34 it becomes @xmath35\ ] ] where @xmath36 is a hypergeometric function .
the solutions ( [ th8 ] ) for different @xmath3 and cosmological parameters @xmath4 and @xmath37 are plotted in fig .
[ doda ] .
the peculiar velocity field in linear perturbation theory is obtained from @xcite @xmath38 where @xmath39 is the peculiar gravitational acceleration and @xmath40 is the dimensionless velocity factor @xmath41 for flat models this formula can be evaluated analytically using eq .
( [ th8 ] ) .
the dependence of @xmath40 on @xmath42 at present ( @xmath43 ) for flat models with different @xmath3 is shown in the left panel of fig .
we see immediately that the dependence on @xmath3 is very weak . however , as shown in the right panel of fig .
[ ffz ] , when going to higher redshifts we find that the velocity factor is much more sensitive to @xmath3 which gives some hope for applying it to determine @xmath3 from local peculiar velocity field .
the simplest model of formation of bound objects ( called the spherical or top hat model ) @xcite describes the nonlinear evolution of a uniform spherical density perturbation in an otherwise homogeneous universe . if the overdensity is big enough the perturbed region will slow down the expansion and eventually turn around and collapse .
the evolution of the size of such a perturbation can be described @xcite by the einstein equation @xmath44.\ ] ] together with the friedmann equation ( [ th1 ] ) it can be solved to obtain the ratio of the cluster density to the background density at the time of turn - around @xmath45 where the last formula is the approximation given in @xcite .
one of the most useful quantities one can derive from the spherical top hat model is the density of the virialized halo in units of the critical density at the time of collapse corresponding to redshift @xmath46 @xmath47 the redshifts @xmath46 and @xmath48 are related by the assumption that the collapse time @xmath49 , the time corresponding to @xmath50 , is twice the turn - around time @xmath51 in reality the object does not decrease its size to zero but virializes to reach the effective final radius @xmath52 .
the ratio by which the halo collapses is estimated from the virial theorem which gives @xcite @xmath53 where @xmath54 figure [ dlwwz ] shows how @xmath55 depends on @xmath42 , @xmath3 and the redshift of collapse .
power spectrum @xmath56 is defined as the fourier transform of the correlation function of density fluctuations @xmath57 the spectra for universe dominated by cold dark matter ( cdm ) have been widely discussed in the literature , e.g. @xcite . for the present time
@xmath58 the power spectrum is usually written in the form @xmath59 where @xmath60 measures the slope of the primordial power spectrum ( we will assume @xmath61 ) , @xmath62 is the transfer function and @xmath63 is a normalization constant . in the presence of cosmological constant ( @xmath64cdm )
the transfer function @xmath65 can be approximated by @xcite @xmath66^{-1/2},\ ] ] where @xmath67 , @xmath68 $ ] and @xmath69 km/(s mpc ) ] @xmath70 .
we assume the barion contribution to the density parameter @xmath71 . for flat models with quintessence
the modification of the transfer function has been proposed by ma _
_ @xcite . the transfer function in equation ( [ q15 ] )
is then @xmath72 , where @xmath73 is approximated by fits given in @xcite .
the present linear power spectra obtained for @xmath1 and @xmath2 assuming cobe normalization are shown in the left panel of fig .
[ pofksig ] .
the rms density fluctuation , @xmath74 , at comoving smoothing scale @xmath75 is given by @xmath76 where the smoothing is performed with the top hat filter @xmath77 the dependence of @xmath74 on smoothing scale for flat models with different @xmath3 is shown in the right panel of fig .
[ pofksig ] .
a particularly useful quantity , constrained by cluster abundance is the rms fluctuation at the scale of @xmath78 mpc .
due to cobe normalization its values turn out to depend strongly on @xmath3 and we get @xmath79 and @xmath80 for @xmath1 and @xmath2 respectively .
one of the most important measures of structure formation is provided by the mass function of bound objects . using the analytical prescription of press and schechter @xcite
, we can estimate the cumulative mass function ( the comoving number density of objects of mass grater than @xmath81 ) @xmath82 where @xmath83 is the number density of objects with mass between @xmath81 and @xmath84 @xmath85 in the expression above , @xmath86 is the background density , @xmath74 is the rms density fluctuation at comoving smoothing scale @xmath75 described in section 5 .
the mass is related to the smoothing scale by @xmath87 .
the parameter @xmath88 is the halo density at collapse as predicted by linear theory .
we adopt here its standard value @xmath89 ( exact for the einstein - de sitter universe ) since it depends very weakly on @xmath42 and @xmath3 @xcite .
figure [ cmfth ] shows the cumulative mass functions calculated from equation ( [ q12 ] ) with @xmath90 and @xmath37 for three models with @xmath91 and @xmath2 .
the strong dependence on @xmath3 is due to the cobe normalization of the power spectra .
the figure also shows data for rich clusters of galaxies from @xcite . for our choice of cosmological parameters
the value of @xmath92 is preferred . however , the press - schechter formulae are known to underestimate the mass functions on the scale of clusters of galaxies when compared to n - body simulations .
the mass functions are also very sensitive to the value of @xmath42 . therefore when
more exact predictions are used and @xmath42 is measured to be close to @xmath93 with a small error bar , we can expect higher values of @xmath3 to be preferred by the cluster data .
the primary constraint on the value of @xmath3 comes from the observations of accelerating expansion , which can only be obtained in models with @xmath94 .
age of universe is quite sensitive to @xmath3 and increases for lower @xmath3 , however the accuracy of our knowledge of both @xmath15 and @xmath95 is not good enough to put strong constraints on @xmath3 .
current estimates are consistent with @xmath96 .
one of the strongest arguments for the existence of dark energy comes from the studies of cosmic microwave background ( cmb ) .
although its power spectrum is weakly sensitive to @xmath3 ( e.g. the height of the first acoustic peak is somewhat increased and its position is shifted to higher multipoles for lower @xmath3 ) , it has already provided some limits . combining the data from cobe and recent balloon experiments balbi _ et al .
_ @xcite find @xmath97 while baccigalupi _
@xcite estimate the best - fitting value of @xmath3 to be @xmath98 .
the ongoing and future satellite experiments are expected to put even stronger limits on @xmath3 . among the promising probes of dark energy
are also supernovae ia .
the comoving distance they serve to measure is sensitive only to the interesting cosmological parameters and the errors related to supernova evolution or extinction are estimated to be small .
the existing data restrict @xmath3 only weakly @xcite , but future experiments like the supernova acceleration probe are expected to measure the value of @xmath3 with a few percent accuracy @xcite
. structure formation also offers methods to constrain the cosmic equation of state .
the suppression of linear growth of density fluctuations for higher @xmath3 alone shows that only for @xmath99 the structure observed today could have evolved from small initial perturbations deduced from cmb observations .
the same range of acceptable @xmath3 values follows from the behaviour of @xmath100 which is a strongly decreasing function of @xmath3 .
the most promising tests are based on the number counts of galaxy clusters .
it turns out @xcite that the slope of comoving abundance as a function of redshift depends sensitively on @xmath3 and therefore can be used to break degeneracies between @xmath3 and other cosmological parameters that appear e.g. in the analysis of cmb .
such measurements are expected to be performed using the proposed new x - ray and sunyaev - zeldovich effect surveys @xcite and the ongoing deep redshift survey @xcite .
the constraints from structure formation appear to be complementary to those from supernovae and cmb measurements .
s. m. harun - or - rashid , m. roos , _ astron . &
_ , * 373 * , 369 ( 2001 ) s. m. carroll , w. h. press , e. l. turner , _ annual review of astron . &
_ , * 30 * , 499 ( 1992 ) o. lahav , p. b. lilje , j. r. primack , m. j. rees , _ mnras _ , * 251 * , 128 ( 1991 ) r. r. caldwell , r. dave , p. j. steinhardt , phys
lett . , * 80 * , 1582 ( 1998 ) c. p.
ma , r. r. caldwell , p. bode , l. wang , _ astrophys . j. _ , * 521 * , l1 ( 1999 ) a. balbi , c. baccigalupi , s. matarrese , f. perrotta , n. vittorio , _ astrophys . j. lett .
_ , * 547 * , 89 ( 2001 ) c. baccigalupi , a. balbi , s. matarrese , f. perrotta , n. vittorio , _ phys . rev .
d _ , * 65 * , 63520 ( 2002 ) b. ratra , p. j. e. peebles , _ phys . rev .
* 37 * , 3406 ( 1988 ) i. zlatev , l. wang , p. j. steinhardt , _ phys .
_ , * 82 * , 896 ( 1999 ) p. j. steinhardt , l. wang , i. zlatev , _ phys
d _ , * 591 * , 270 ( 1999 ) v. barger , d. marfatia , _ phys .
b _ , * 498 * , 67 ( 2001 ) i. maor , r. brustein , p. j. steinhardt , _ phys . rev .
_ , * 86 * , 6 ( 2001 ) j. weller , a. albrecht , _ phys . , rev
_ , * 86 * , 1939 ( 2001 ) l. wang , r. r. caldwell , j. p. ostriker , p
. j. steinhardt , _ astrophys .
j. _ , * 530 * , 17 ( 2000 ) d. huterer , m. s. turner , _ phys .
d _ , * 64 * , 123527 ( 2001 ) p. j. e. peebles , _ principles of physical cosmology _ , princeton univ .
press , princeton 1993 v. silveira , i. waga , _ phys .
d _ , * 50 * , 4890 ( 1994 ) j. e. gunn , j. r. gott , _ astrophys
. j. _ , * 176 * , 1 ( 1972 ) j. e. gunn , _ astrophys .
j. _ , * 218 * , 592 ( 1977 ) e. l. okas , y. hoffman , _ the identification of dark matter _
3rd international workshop , eds n. j. c. spooner , v. kudryavtsev , world scientific , singapore 2001 , p. 121 e. l. okas , y. hoffman , astro - ph/0108283 l. wang , p. j. steinhardt , _ astrophys . j. _ , * 508 * , 483 ( 1998 ) n. sugiyama , _ astrophys . j. _ , * 471 * , 542 ( 1995 ) w. h. press , p. schechter , _ astrophys .
j. _ , * 187 * , 425 ( 1974 ) m. girardi , s. borgani , g. giuricin , f. mardirossian , m. mezzetti , _ astrophys .
* 506 * , 45 ( 1998 ) s. perlmutter , m. s. turner , m. white , _ phys .
_ , * 83 * , 670 ( 1999 ) s. c. c. ng , d. l. wiltshire , _ phys .
d _ , * 64 * , 123519 ( 2001 ) z. haiman , j. j. mohr , g. p. holder , _ astrophys . j. _ , * 553 * , 545 ( 2001 ) j. a. newman , m. davis , _ astrophys . j. lett .
_ , * 534 * , l11 ( 2000 ) | i review the main characteristics of structure formation in the quintessential universe . assuming equation of state @xmath0const i provide a brief description of the background cosmology and discuss the linear growth of density perturbations , the strongly nonlinear evolution , the power spectra and rms fluctuations as well as mass functions focusing on the three values @xmath1 and @xmath2 .
finally i describe the presently available and future constraints on @xmath3 . |
in this paper we study physical layer authentication over a noisy multiple access channel and no pre - shared key .
such scenarios may occur when a network is established in a hostile environment .
being able to trust the information observed is just as important as actually observing it .
it also stands to reason , that always having a pre - shared key seems restrictive as it is dependent on a method to generate and secure the key .
we model this scenario by considering two transmitters ( alice and neer - do - well eve ) and one receiver ( bob ) .
alice and eve share a discrete memoryless multiple access channel ( dm - mac ) , so that not only may eve try to forge messages to bob , she may try to modify the ones alice transmits . to further complicate the matter , we assume that eve not only knows the codebook which alice and bob share , but the message alice will send as well .
our goal is for bob to be able to validate the authenticity of the sender ( or alternatively the integrity of the information ) , as well as be able to decode the information given it is authentic .
simmons first studied physical layer authentication in @xcite .
simmons allowed for alice and bob to have a pre - shared key , which eve did not have access to .
this key , which can be made small relative to the number of symbols it protects , in essence partitions the entire state of messages into sets which are valid and ones that are not .
any attempt by eve to modify the message then results , with high probability , in a message included in the invalid set . physical layer authentication , without the use of pre - shared key ,
was considered by maurer in @xcite . instead of a pre - shared key , alice , bob , and eve
are given given access to different random variables , which share some arbitrary correlation . by using their individual random variables ,
alice and bob proceed in public discussion over a perfect channel to generate a secret key .
this is related to authentication in that the channel being perfect allows eve the ability to perfectly spoof any packet from alice .
hence , the distributions of the random variables play a pivotal role in establishing the authenticity of any message . in this model
the three nodes must share copies of a random variable generated from an independent and identically distributed ( iid ) source . and
furthermore the variables that alice and bob share must not be `` simulatable '' by eve given her random variable .
use of an iid source though guarantees no underlying code structure , which could leak information about the random variables alice and bob view .
it is also impractical in scenarios not concerned with secrecy to first generate a secret key for authentication purposes in later transmissions . because of these limits
others have studied authentication in noisy channels .
works include yu et al .
who in @xcite used spread spectrum techniques in addition to a covert channel to ensure authentication @xcite . and
xiao et al in @xcite , who used the unique scattering of individual users in indoor environments to authenticate packets .
for purpose of discussion we focus on two works in particular .
first , lai et al . , who in @xcite considered the problem of authentication in a noisy channel model when alice and bob have a pre - shared key . in their model
the link between bob and alice is noisy while the link between eve and bob is noise - free , a worst case scenario .
much like the work of simmons , the key alice and bob share is used to label some transmitted sequences as valid and not . and much like simmons work , it is then nearly impossible for eve to pick a transmitted message that would be valid .
the second work , is that of jiang @xcite .
jiang in specific considered the case where either alice or eve may transmit to bob , but not simultaneously . by assuming the channels distinct
, jiang showed that even without a pre - shared secret key it was possible to guarantee authentication .
this was done by considering the intersection of the distributions alice could induce at bob and the distributions eve could induce at bob had only marginal overlap if the channel from alice to bob was not simulatable using the channel from eve to bob .
in contrast , by allowing simultaneous communication , we allow more possibilities for eve s attack .
this is because eve does not need to attack the entire message transmitted , only a smaller portion , and thus only slightly disturb the distributions , which as mentioned previously were critical to ensuring authentication .
much like jiang , we do this by showing that the distributions induced by the legitimate party , and the ones by the illegitimate party have marginal overlap .
unlike jiang though , we approach the problem using traditional random coding methods and techniques .
we derive a new inner bound , which is dependent on the channel not having a simulatable - like property .
unlike simulability though , this property can be controlled in part by alice and her choice of distribution .
random variables will be denoted by upper case letters , their alphabets will be the script version of that letter , and particular realizations of that random variable will be denoted in lower case . in other words , @xmath0 is a random variable , and @xmath1 is a particular realization of @xmath0 .
the @xmath2-th extension of a random variable is denoted @xmath3 , and particular realizations of that random variable are denoted similarly by @xmath4 .
to represent a particular value in the @xmath2-th extension we use subscripts , such as @xmath5 is the value in the @xmath6-th position of @xmath4 .
when the need arises for discussion of the actual distribution of the random variable over their particular alphabet , we will use @xmath7 with that related random variable as a subscript ( @xmath8 is the distribution @xmath0 has over @xmath9 ) .
when the distribution is clear from context , we drop the subscript . by @xmath10
we denote when @xmath11 form a markov chain @xmath12 . by @xmath13 } } $ ] we define the set @xmath14 , where @xmath15 is the induced type of @xmath4 . generally , @xmath13 } }
$ ] is referred to as a strongly typical set .
we assume that @xmath16 follows the delta convention outlined in ( * ? ? ?
* convention 2.11 ) .
furthermore by @xmath17 and @xmath18 we define the entropy ( conditional entropy resp . ) of random variable @xmath0 ( @xmath0 given @xmath19 resp . ) , where @xmath20 and @xmath21 . finally by @xmath22
we denote the mutual information of random variables @xmath0 and @xmath19 , where @xmath23 .
for our channel model , graphically represented in figure [ fig : v1 ] , alice and eve share a dm - mac @xmath24 with bob .
alice wishes to transmit a message @xmath25 , while eve attempts to modify its content so that bob will believe alice transmitted some @xmath26 that is not equal to @xmath27 . by @xmath2
we denote the number of symbols that will be transmitted over the channel .
alice uses a possibly stochastic function @xmath28 to generate her sequence which she transmits .
eve simultaneously chooses a sequence @xmath29 as a function of the message and encoder , which she transmits concurrently with alice s transmission .
we assume that there exists an element @xmath30 representative of the state when eve decides not to transmit ( hence eve not attacking is denoted @xmath31 . )
because the distribution associated with the output of the mac when eve is silent is of distinct importance , we refer to it specifically by @xmath32 . in general
, bob will receive a value @xmath33 , and then will use a decoder @xmath34 to either output an estimate of the message , or instead to declare that eve is attempting to intrude .
we assume that alice , bob and eve have access to the encoder @xmath35 and the decoder @xmath36 .
we expect the code to decode to the correct message when eve is not attacking ( @xmath31 ) , and to either decode to the correct message or detect manipulation when eve is attacking ( @xmath37 ) .
these expectations lead directly lead to the following definition for our physical layer authentication code .
a @xmath38-physical layer authentication code is any encoder - decoder pair @xmath39 , for which the encoder takes @xmath40 messages and the decoder satisfies @xmath41 a code which meets these two requirements , can both decode and authenticate with high probability . we do not concern ourselves with the optimization of these two probabilities , and instead focus first on guaranteeing existence . as we will show using random coding ,
there exist codes such that both probabilities go to zero as @xmath2 grows to infinity .
section [ sec : achiev ] establishes the following inner bound on the rate of a physical layer authentication code .
@xmath42 where @xmath43 contains all @xmath44 such that * @xmath45 * @xmath46 for all distributions @xmath47 such that @xmath48 . if the distribution @xmath49 which achieves capacity is a member of @xmath43 , then the inner bound is also an outer bound . in general though , this is not the case
. the necessity of auxiliary random variable @xmath44 can be demonstrated through a simple example .
let @xmath50 , @xmath51 and @xmath52 , and define the dm - mac by @xmath53 where the realizations of @xmath54 are enumerated @xmath55 .
for this channel though , choosing @xmath56 results in @xmath57 thus @xmath58 for all distributions @xmath49 .
now , instead lets introduce the auxiliary random variable @xmath44 , and set @xmath59 , with @xmath60 and @xmath61 .
it is clear that if @xmath62 then bob can detect eve , which corresponds to the case @xmath63 . in specific the value @xmath64 results with probability
@xmath65 if @xmath66 equal @xmath67 or @xmath68 thus eve must have @xmath69 . clearly , assuming @xmath70 , @xmath71 and @xmath72 .
thus it follows that @xmath73 . by adding an auxiliary random variable , we have now made the inner bound non - empty and thus increased our inner bound region . in this example
it is not hard to see that in fact we can let @xmath74 and approach capacity .
finally , in this work we assume that eve knows the message @xmath27 and encoding function @xmath35 .
why not allow for eve to know @xmath4 as well ? while this does in fact make eve more powerful , it actually simplifies the math . indeed allowing eve knowledge of @xmath4 results in the property that for every @xmath38-physical layer authentication code there exist a corresponding deterministic @xmath75 physical layer authentication code such that @xmath76 .
this follows because for any @xmath77 @xmath78 thus the existence of a sequence of rate @xmath79 codes such that @xmath80 and @xmath81 converge to @xmath82 , implies the existence of a sequence of rate @xmath79 deterministic codes for which @xmath83 and @xmath84 converge to @xmath82 as well . hence considering deterministic encoders
is sufficient for determining the maximum rate . restricting the codes in section [ sec : achiev ] results in the same conditions except without the auxiliary @xmath44 , with @xmath0 instead replacing it .
our inner bound then is strictly smaller , as one would expect , due to the arguments of our example .
fix a distribution @xmath85 . define @xmath86 , and for all @xmath25 randomly and independently select @xmath87 . for transmission
, alice chooses a value of @xmath4 according to the distribution @xmath88 , and sends the value over the dm - mac , @xmath89 .
bob , who receives a sequence @xmath90 from the channel , uses the decoder defined by @xmath91 } } ( u^n(m ) ) \\ & \text{and } y^n \notin \bigcup _ { \hat m
( \neq m ) \in { \mathcal{m } } } { t^{n}_{\left [ y|u,\emptyset \right ] } } ( u^n(\hat m ) ) \\ { \boldsymbol{!}}&\text { otherwise } , \end{cases}\end{aligned}\ ] ] to estimate the message or declare intrusion . for the remainder of the paper , by @xmath92 we will denote the random variable over the codebooks .
any particular @xmath93 is itself a function which defines a distribution over @xmath4 as given earlier . given that @xmath31 , an error will occur if either @xmath94 } } ( u^n(m))$ ] or if @xmath95 } } ( u^n(\hat m))$ ] . by using the union bound we may individually consider the probabilities of the two events . for the first event there exists an @xmath96 such that @xmath97 } } ( u^n(m ) )
\middle| v^n = \emptyset^n \right\ } \leq \epsilon_n , \ ] ] by ( * ? ? ?
* lemma 2.12 ) . for the second , the existence of an @xmath98 for which @xmath99 } } ( u^n(\hat m ) ) \right\ } \leq \epsilon_n'\end{aligned}\ ] ] whenever @xmath100 , for some @xmath101 , can be shown through traditional random coding arguments . for any codebook
, the value of @xmath102 is fixed .
we want to show that the average over all codebooks results in @xmath103 as @xmath2 increases . if the average probability of error for both the cases when eve attacks and when she does not , converge to @xmath82 , then there must exist a sequence of codes where the individual errors converge to @xmath82 as well . for the probability of error given eve attacking @xmath104 } } ( u^n(\hat m ) )
\middle| v^n \neq \emptyset^n \right\}\notag \\ & = e_m \!\ !
\pr \!\!\left\ { \!\ ! y^n \!\!\in \!\ ! \bigcup_{\hat m \neq m } { t^{n}_{\left [ y|u,\emptyset \right ] } } ( u^n(\hat m ) ) \middle| v^n \neq \emptyset^n , m = m \right\}.\end{aligned}\ ] ] by symmetry we need only consider a particular value of @xmath105 , and hence we focus on @xmath106 } } ( u^n(\hat m ) ) \middle| m = m , v^n \neq \emptyset^n \right\}. \label{eq : d_1}\end{aligned}\ ] ] the probability of error in this case is dependent on what eve decides to transmit , which as mentioned previously is dependent upon the codebook , @xmath93 , and the actual message transmitted @xmath27 .
for every codebook then , it is important to consider the probability of a value particular @xmath90 occurring given the codebook , message , and sequence transmitted .
this is made explicit by writing equation as @xmath107 } } ( u^n(\hat m))}\right ) } \label{eq : d_2}.\end{aligned}\ ] ] next we seek to determine the probability of error given a particular codebook and @xmath108 .
this action is equivalent to summing over @xmath90 , for equation can be written @xmath109 } } ( u^n(\hat m))}\right ) } \label{eq : d_2b}.\end{aligned}\ ] ] consider @xmath110 , for a fixed value @xmath111 , there exists an @xmath96 such that @xmath112 } } ( u^n(m),v^n ) \middle|
u^n(m),v^n \right\ } \leq \epsilon_n,\ ] ] due to ( * ? ? ?
* lemma 2.12 ) . furthermore because the distribution @xmath113 is considered fixed , and because the @xmath114 defining the typical set ( recall section [ sec : notation ] ) , there exists a @xmath101 such that @xmath115 where @xmath116 are random variables that are jointly distributed according to the type induced from @xmath111 , due to ( * ? ? ? * problem 2.5.b ) . because the distribution of @xmath111 is not fixed over all values of @xmath117 and @xmath108 , to efficiently apply equations and we must partition the values of @xmath118 so that over every partition we only need worry about a single distribution . to this end ,
let the set @xmath119 contain all possible joint distributions on @xmath120 . with a slight abuse of notation , a particular realization of @xmath119 will be denoted @xmath121 in order to emphasize that @xmath121 is representative of a certain distribution over @xmath120 . introducing @xmath121 and applying equations and , we may upper bound equation by @xmath122 } } ( u^n(\hat m ) ) \cap { t^{n}_{\left [ y|u , v \right ] } } ( u^n(m),v^n)}\right ) } \label{eq : d_3}.\end{aligned}\ ] ] we have introduced the notation @xmath123 into equation to make explicit the dependence of the entropy on the distribution of @xmath118 .
having dealt with the term @xmath110 , we turn towards completing the summation . to this end
, @xmath124 is an upper bound of equation .
equation follows first from @xmath125 } } ( u^n(\hat m ) ) \cap { t^{n}_{\left [ y|u , v \right ] } } ( u^n(m),v^n)}\right ) } \notag \\ & \leq \sum_{\hat m \neq m } { \mathfrak{1}\left({{t^{n}_{\left [ y|u,\emptyset \right ] } } ( u^n(\hat m ) ) \cap { t^{n}_{\left [ y|u , v \right ] } } ( u^n(m),v^n)}\right ) } .
\label{eq : d_4b}\end{aligned}\ ] ] and secondly from ( * ? ? ?
* problem 2.10 ) , @xmath126 } } ( u^n(\hat m ) ) \cap { t^{n}_{\left [ y|u , v \right ] } } ( u^n(m),v^n)}\right ) } \notag \\ & \leq \max _ { \begin{matrix } ( u',u , v ) : \\ ( u^n(\hat m ) , u^n(m ) , v^n ) \in { t^{n } _ { u',u , v } } \\ p_{y|u ' } = p_{y|u , v = \emptyset } \\ p_{y| u , v } = p_{y|u ,
v,{\mathcal{q } } = q } \end{matrix } } \hspace{-30pt } 2^ { n ( h(y| u',u , v ) + \lambda_n ' ) } \label{eq : sumy}.\end{aligned}\ ] ] while it may appear that the bound induced from equation introduces a large amount of error , due to channel coding it does not .
recall that @xmath127 } } ( u^n(\hat m))$ ] are representative of the region for which the decoder outputs @xmath128 .
these regions should rarely overlap , and thus it is most likely that for a @xmath90 , there is only a single value of @xmath27 for which @xmath129 } } ( u^n(\hat m ) ) \cap { t^{n}_{\left [ y|u , v \right ] } } ( u^n(m),v^n)}\right ) } $ ] is non zero . letting @xmath130 be the set of all distributions on @xmath131 which may meet the requirements presented in the max of equation , and @xmath132 be the number of @xmath128 such that @xmath133
, we can more succinctly write equation by @xmath134 where @xmath135 .
the @xmath121 term has been removed since it is uniquely determined by @xmath136 .
we now turn our attention to giving an upper bound @xmath132 , which will complete the proof .
in fact , there exists a @xmath137 such @xmath138 due to the hoeffding bound , alternatively ( * ? ? ? * lemma 17.9 ) .
this is because @xmath139 , and hence since each codeword is chosen independently from the distribution , @xmath140 is simply the probability that the sum of @xmath141 independent and identically distributed binomial random variables is greater than @xmath142 . in this case
the probability of the random variable can be upper bounded by @xmath143 for some @xmath144 .
therefore there exists a @xmath145 , @xmath146 such that @xmath147 upper bounds since the maximum value of @xmath148 , and @xmath149 . by summing over all @xmath150 and then recognizing since @xmath119 is at most polynomial in @xmath2
, we obtain an @xmath151 such that @xmath152 upper bounds equation .
equation leaves us with a bound dependent on the value of @xmath153 when compared to @xmath79 for the maximum value of @xmath154 .
consider the case where @xmath155 , where equation becomes @xmath156 for @xmath157 this term converges to @xmath82 .
indeed , recall that @xmath158 for all @xmath154 , and that @xmath159 conceptually , this bound relates to attacks where eve does not care what value of @xmath27 bob chooses , as long as it is not @xmath128 .
now , consider the alternative situation in which @xmath160 , for which equation reduces to @xmath161 which clearly goes to @xmath82 if @xmath162 .
but @xmath163 if and only if @xmath164 for the condition implies that @xmath165 . combining back the two situations , and excluding the distributions of @xmath85 such that there exists a @xmath166 satisfying equation gives that on average for our coding scheme @xmath167 . which as discussed prior proves the existence of a set of codes which satisfy our inner bound
of critical importance in our result is the ability of eve to find a distribution @xmath166 such that @xmath168 . by fixing @xmath169
, this implies @xmath170 , which implies the channel must be `` simulatable '' ( see @xcite ) given @xmath169 in order for eve to manipulate those values of @xmath44 . on the other hand
fixing @xmath171 gives that the channel must be manipulable ( see @xcite ) for all values of @xmath166 eve chooses to use . | if alice must communicate with bob over a channel shared with the adversarial eve , then bob must be able to validate the authenticity of the message .
in particular we consider the model where alice and eve share a discrete memoryless multiple access channel with bob , thus allowing simultaneous transmissions from alice and eve . by traditional random coding arguments
, we demonstrate an inner bound on the rate at which alice may transmit , while still granting bob the ability to authenticate .
furthermore this is accomplished in spite of alice and bob lacking a pre - shared key , as well as allowing eve prior knowledge of both the codebook alice and bob share and the messages alice transmits . |
solar flares are the result of sudden energy release , often exceeding @xmath1 ergs , from magnetic reconnection @xcite .
the main manifestation of flares is extensive emission over much of the electromagnetic spectrum .
intriguingly , under some special circumstances , negative flare contrasts ( decrease of intensity ) are occasionally reported in stellar observations in visible continuum wavelengths @xcite . in those cases , a 20% negative contrast in the optical continuum
is typically detected prior to the initial brightening @xcite .
the widely accepted postulate for generating negative flares is the h@xmath2 absorption model @xcite , in which a bombarding beam of electrons penetrates to the lower and cooler atmosphere and causes enhanced collisional ionization of hydrogen , and leads to an increase of electron density and eventually an increase of h@xmath3 opacity . as a result ,
the formation layer of the visible continuum , though still in the photosphere , is shifted slightly upward and becomes cooler .
analogously , intensity drops associated with flares can occur on the sun , but involving a variety of features and formation mechanisms .
the continuum dimming caused by the enhanced h@xmath3 opacity was rarely observed and the diminution effects are weaker on the sun than that of the stellar events @xcite .
so far only two events have been reported , with negative contrasts of 5% at 5500 @xcite and 1% @xmath4 2% at 8500 @xcite .
however , due to the measurement uncertainty and low resolution , both events are regarded as non - compelling detections @xcite . in microwave observations ,
negative contrasts of solar radiation during flares have also been reported @xcite .
the term `` negative flare '' was used but the dimming in radio flux was caused by absorption of intervening filament material , or a similar mechanism , rather than a change in the local emission process itself . besides the visible continuum and microwave observations
, negative flares can also be found in helium - line observations .
solar observations by @xcite attracted attention to the d3 line at 5876 , by which the helium element was first discovered in 1868 .
@xcite confirmed the occurrence of enhanced d3 darkening and discussed fine structures of the dark - bright pattern of flare ribbons .
owing to electrons orbiting the nucleus , a atom comprises two different kinds of energy states : the singlet ( para - helium ) and the triplet ( ortho - helium ) states .
since the ground state is a singlet one ( @xmath5 @xmath6 ) , radiative transitions to the higher triplet states ( @xmath7 @xmath8 and @xmath9 @xmath10 ) responsible for the d3 line are forbidden .
these states are then populated through collisional transitions or photoionization from the ground state followed by recombinations , which require energy from thermal ( or nonthermal ) electrons and euv irradiation , respectively .
whether the d3 line appears in emission or absorption depends on whether the line source function in the line forming region is larger or smaller than the nearby continuum intensity .
the line source function is in turn determined by the ratio of the populations at the two energy states @xcite . in the case that these energy states are populated mainly through thermal collisional transitions , the plasma temperature is the key factor in determining whether the d3 line appears in absorption or emission @xcite .
similar to the d3 line , another helium line , at 10830 , is formed by the transition between @xmath11 @xmath12 and @xmath7 @xmath8 of the helium triplet .
a negative flare was reported by @xcite , in 10830 using spectrographic data .
however , due to the lack of spatial information , the possibility of enhanced absorption of plage or dark filament material moving into the spectral slit can not be ruled out .
two mechanisms , namely photoionization - recombination and collisional ionization - recombination , that can populate the helium triplet and thus enable the transitions between @xmath11 @xmath12 and @xmath7 @xmath8 states , were studied by @xcite .
as expected , the euv radiation turns the 10830 line to emission via photoionization - recombination , which has been confirmed by @xcite .
notably , the collisional ionization - recombination can enhance the absorption at the beginning of the flare and generate stronger emission at flare maximum than the photoionization - recombination effect @xcite .
as such , under collisional bombardment , the intensity in a flaring region could first decrease and then increase afterward . in this study , we present the very first , unambiguous evidence of negative contrast observed during solar flares and discuss the morphological and spectral properties of the negative flares .
the key observations in 10830 presented in this study , were obtained using the newly commissioned 1.6-m new solar telescope ( nst ; * ? ? ?
* ) at big bear solar observatory .
equipped with a high - order adaptive optics system using 308 sub - apertures @xcite , nst provides unprecedented high spatio - temporal resolution .
for instance , the image scale at 10830 achieves 0.085 ( 61.6 km ) per pixel .
the near infrared ( 10830 ) imaging system is based on a liquid - nitrogen - cooled hgcdte focal plane cmos array and a lyot - filter @xcite , which was tuned to the blue wing of the line at 10830.05 with a passband of 0.5 . the typical field - of - view ( fov ) is about 85 and the effective cadence is 15 s after performing the speckle reconstruction @xcite .
supportive h@xmath0 imaging spectroscopic data were obtained simultaneously by the fast imaging solar spectrograph ( fiss ; * ? ? ?
* ) , which has a smaller fov of 40 by 60 with a scanning cadence as fast as 20 s and a spectral resolving power ( @xmath13/@xmath14 ) of @xmath15 .
complementary full - disk intensity maps are obtained from two major instruments onboard the solar dynamics observatory ( sdo ; * ? ? ?
* ) , the helioseismic and magnetic imager ( hmi ; * ? ? ?
* ) provided the visible continuum maps and the atmospheric imaging assembly ( aia ; * ? ? ?
* ) provided uv / euv maps .
the spatial resolution of sdo / hmi and sdo / aia is lower than that of the nst , but they provide context information in locating the nst target on the solar disk and comparing flare emissions in different wavelengths . in the past several decades , taking advantage of rapid developments of focal plane instrumentation , substantial flare observations have been carried out by 2-d imaging to provide spatially resolved information .
however , there was no high - resolution imaging spectroscopy for solar flares in the visible or uv wavelengths from space @xcite . the newly launched nasa mission , iris @xcite , fills this gap by providing imaging spectroscopy in both near ultraviolet ( nuv ) and far ultraviolet ( fuv ) bands .
the nuv window of iris has a main passband at 2800 , and the fuv window covers the spectra near 1400 .
the iris raster size is chosen depending on the program .
the spectrograph may scan a fov of 130 by 175 using a slit with dimension of 0.33 by 175 for slow observations @xcite and may speed up with 1 - 16 step rasters for flares or faster programs .
pre - flare spectra served as reference and for the study of pre - flare conditions .
the slit position is visible in the slit - jaw images as a dark vertical line .
it is notable that previous d3 observations of solar flares are rare and the data were recorded on film @xcite . to our knowledge , no negative flare has ever before been clearly recorded using an imaging system in 10830 , probably due to the low sensitivity of detectors and coarse observing resolutions ( spatial and temporal ) . using the unprecedented 100-km resolution in 10830 , we report the first imaging observation of negative flares . as listed in table [ nf_list ] ,
two events were observed , on 2013 august 17 ( m1.4 ) and 2014 august 1 ( m1.5 ) , and presented in this study .
lccccr + date & goes sxr & nst 10830 & fiss / nst h@xmath0 & iris & rhessi + & magnitude & & & & + 2013 aug .
17 & m1.4 & yes & yes & no & no + 2014 aug .
01 & m1.5 & yes & no & yes & no + on 2013 august 17 , we observed a solar flare with goes soft - x - ray - class m1.4 at about 18:43 ut in active region noaa 11818 . a filament eruption , or possibly eruption of a magnetic flux rope ,
was associated with this flare and led to a halo coronal mass ejection ( cme ) .
however , the filament disappeared early in the flare period and therefore has no effect on the 10830 dark flare ribbon found much later at the flare footpoints .
figure [ fulldisk ] presents the hmi and nst images taken around 19:12 ut showing a negative flare ribbon in the 10830 indicated by the black arrow . in the left panel ,
a full - disk hmi continuum map is displayed , on which a small area is highlighted by a red box corresponding to the nst s fov .
the right panel shows the nst 10830 map .
this flare has a typical two - ribbon morphology .
one ribbon is located inside the sunspot penumbra - umbra region with a stronger magnetic field and the other is located in the surrounding granular area with a weaker magnetic field .
we focus on the ribbon outside the sunspot as it is the one associated most clearly with negative contrasts .
it is notable that all the nst images are speckle reconstructed to achieve diffraction - limited resolution .
the speckle reconstruction could in principle add artificial features to a region with a strong gradient in brightness . in order to confirm that the negative flare signature is real , we verified that the dark ribbon is also clearly identified in the raw data prior to speckle reconstruction .
therefore , we are confident that the observed flare ribbon with a negative contrast is not an artificial feature introduced by the speckle reconstruction .
one example of the 10830 image taken at 18:56:14 ut is shown in figure [ spacetime]a .
the bright flare source is clearly seen .
owing to nst s unprecedented high resolution , a dark edge is clearly visible on the upper boundary of this flare ribbon .
note that the ribbon is moving upward in the fov and thus the dark edge is actually the moving front that represents the footpoints of newly reconnected flare loops .
in other words , the ribbon with a negative contrast corresponds to the latest electron beams penetrating from the solar corona , where the magnetic reconnection occurred .
the bright part of the ribbon behind the dark frontier is the signal of the flare heating immediately following the enhanced line absorption .
such a leading front due to electron penetration has been reported in h@xmath0 observations as narrow bright features with red - shifted spectra , for instance by @xcite using the multi - slit spectrograph developed by lockheed solar observatory @xcite . during our observing run , using the imaging spectroscopy by fiss , we were able to identify the red - shift in h@xmath0 spectra at the same location where the enhanced 10830 absorption occurs .
for illustration , as shown in figure [ ha ] , a cross is put on a representative location of the ribbon front , which shows emission in h@xmath0 and euv 304 , but absorption in 10830 . on the upper right panel , the h@xmath0 spectra on the cross and
the light curves of 10830 and euv 304 are plotted . at an early time of 19:04:00 ut
, the flare emission had not yet propagated to the cross and a typical broad h@xmath0 absorption line is seen ( black dotted curve in the lower right panel ) .
later on , at 19:06:28 ut , the ribbon - front had arrived at the cross location , indicating the start of precipitation characterized by a strong red - wing enhancement in the h@xmath0 line profile ( blue curve in the lower right panel ) .
after 19:07:28 ut , the emission dominates in all wavelengths and we see an emission profile in h@xmath0 with a typical central reversal , resulting from self - absorption @xcite . this strong red wing emission in h@xmath0 , appearing in the early phase of the flare , indicates the existence of large scale mass motions caused by impulsive heating , consistent with the most recent radiative hydrodynamic simulation of an electron beam heated atmosphere @xcite .
therefore , a close relationship is built up between the negative flare ribbon and the precipitation of energetic particles .
the above description of the event agrees with the scenario predicted by @xcite , in which the electron beams bombarding on cooler plasma over - populate the @xmath11 @xmath12 state via collisional ionization - recombination , resulting in the absorption in 10830 .
later on , when this layer of the solar atmosphere is efficiently heated to a higher temperature , the @xmath7 @xmath8 state is also sufficiently populated and emission is produced due to both photoionization - recombination and collisional ionization - recombination , by enhanced euv irradiation and nonthermal electrons , respectively .
the transition from absorption to emission is likely a result of the line source function varying from a lower value ( in a cooler atmosphere ) to a higher one ( in a hotter atmosphere ) .
we select a slit cutting through the dark ribbon ( see figure [ spacetime]a ) and plot its time - space diagram in figure [ spacetime]b . as such ,
the motion of the dark ribbon and the expansion of the following bright flare ribbon are clearly identified . by measuring
the distance traveled within the time period from 18:56:14 ut to 19:14:44 ut , the average speed of the ribbon motion is calculated to be 3.7 km s@xmath16 , which is significantly slower than some strong flares ( e.g. , * ? ? ? * ) but within a typical range for moderate flares .
in the bottom panels of figure [ spacetime ] , we see an emission area in uv 1700 , which is almost identical to 10830 but obviously without the dark fringe .
the temporal variation of intensity is illustrated in figure [ lc ] .
on the left panel , a small slit on the path of the dark ribbon is chosen , and a square region is selected as a reference area .
the light curves of the averaged intensity in the slit and the large box are plotted in the right panel in red and green colors , respectively . as expected , when the ribbon sweeps through the slit , we see an obvious drop of the intensity compared to the initial stage .
afterwards , the slit region is dominated by the subsequent flare heating and thus a steady increase of the 10830 intensity is observed .
the error bars represent 1@xmath17 deviation of the measurements of the contrasts in both the reference area and the slit .
therefore , we are confident that the intensity variation , especially the decrease , is due to neither the variation of the background ( green curve ) nor the seeing fluctuation . the brightness , relative to the reference area , of the slit area drops from about 5.7% to @xmath18 .
this can be translated to a negative contrast of @xmath19 relative to the pre - flare condition .
the slit light curve also provides the time scale of the dark flare ribbon .
a gaussian fit of the dimming period shows a full - width - at - half - maximum ( fwhm ) of about 91 s. another important parameter is the characteristic size ( width of the ribbon ) , which is crucial for the estimation of electron flux used in the numerical simulation of flare absorption and emission .
we measure the size following the method of @xcite . as shown in figure [ width ] , four slits are selected at different positions and different times across the dark flare ribbon . in the bottom panels , the intensity spatial profile of each slit is fitted using a gaussian function plus a constant at the pre - flare level .
the fwhm measurements give a width of the dark ribbon in a range from 340 km to 510 km .
a wider range of the width is possible by selecting more slit positions .
however , the four - point measurements are within the best seeing conditions and therefore are more representatively meaningful .
note that the dark flare ribbon is fully resolved during the entire period of observation , and thus the lower limit of the width is at least greater than 120 km ( two pixels ) . in the second event on 2014 august 1 , the interface region imaging spectrograph ( iris )
@xcite observed the same region as nst .
the high resolution uv spectra provide supportive information with respect to the negative flare ribbon observed by nst . while not as striking as the earlier event , the main features of the negative contrast front are the same in the two events . at about 18:15 ut
, the iris spectral slit was placed at a position that cuts through the flare ribbon perpendicularly , which is perfect to study the different components of the flare ribbon ( see figure [ 2ndsji ] ) . in figure
[ 2ndspe ] , and spectra of this slit position are plotted .
recall that the negative flare edge shows negative contrast in 10830 and the following part is in emission .
therefore , we anticipate that we will see different spectral characteristics at the front edge of the ribbon compared to the areas behind .
two line groups , the lines at 2796.33 and 2803.51 and the lines at 1334.53 and 1335.71 , are studied .
the line profiles , of and , in representative pixels of the front ( corresponding to the dark edge in 10830 ) and the middle ( emission in all uv lines and 10830 ) of the northern flare ribbon , are plotted in figure [ spemg ] and figure [ spec ] . as expected , the front edge ( blue ) of the northern ribbon shows significant abnormal features compared to the rest of this ribbon body ( orange ) , which is expected because negative contrast is observed in 10830 .
several important characteristics are inferred : 1 ) in general , the spectra of the ordinary flare ribbon behind the front are brighter than those on the negative front .
2 ) strong , mainly red - shifted emission and line broadening ( with a fwhm near or greater than 0.8 ) are found in the negative front relative to the ribbon behind .
3 ) both the and show a central reversal pattern in the ribbon front .
this is a common feature for quiet sun spectra @xcite but unusual for flare regions @xcite .
furthermore , we classified the profile shapes into groups of self - similar profiles using a machine learning algorithm @xcite .
a total of 30 groups are classified according to their characteristics , for instance the line broadening and the central reversal .
the northern ribbon , with a dark front observed in 10830 , appeared with a special type of profile , including the broadest profiles of the whole fov .
a selection of these profiles is shown in figure [ ml ] .
the two images on the left show the iris raster in two different wavelengths at 2792.37 and 2796.34 .
overplotted are the color - coded locations of special profiles , whose shapes are shown in the panels on the right .
the average profile of each colored group is shown in black and is used to determine the fwhm .
the widest profiles of the fov with a fwhm of 1 are shown in green and from the magnified inset it can be seen that these profiles lie just north of the bright flare ribbon , corresponding to the dark ribbon observed in 10830 .
the southern ribbon did not show any especially broad profiles and it also did not show any dark ribbon front in 10830 .
the prevalence of the negative flare ribbons observed in 10830 provides important and unique constraints in modeling solar flares .
the negative contrasts are detected on the advancing edge of the flare ribbon , representing the footpoints of newly reconnected flare loops according to the ` standard ' flare model @xcite .
in other words , the ribbon with a negative contrast corresponds to the latest electron beams penetrating from the solar corona , where the magnetic reconnection occurred .
the bright part of the ribbon behind the dark frontier is the signal of the flare heating immediately following the enhanced line absorption .
such a leading front due to electron penetration has been reported in h@xmath0 observations as narrow bright features with red - shifted spectra , for instance by @xcite using the multi - slit spectrograph developed by lockheed solar observatory @xcite .
spectroscopic observations in h@xmath0 from nst / fiss and uv lines from iris reveal distinct features on the ribbon front from that behind the front , namely a strongly enhanced red wing and line broadening , respectively . in newly formed flare ribbons ,
the red shift is usually the dominant velocity pattern , which is considered an unambiguous signature of mass motion , possibly due to chromospheric evaporation resulting from the electron precipitation . according to previous modeling ,
two mechanisms are plausible in explaining the enhanced absorption in 10830 . in the collisional ionization - recombination model ,
the helium atom is excited , through collisional ionization from the ground state by energetic electrons followed by recombinations to triplet states .
our observation can be explained by this model since we are able to correlate the electron - precipitation site with the dark edge of the flare ribbon in 10830 . on the other hand
, the photoionization - recombination model assumes that the helium atom is excited by the euv radiation @xcite . any euv radiation with a wavelength shorter than 304 can generate excitation of atoms via the photoionization - recombination process .
one recent simulation reproduces a much deeper absorption profile in 10830 caused by x - ray and euv radiation @xcite .
in such a scenario , the flare ribbon front is heated by the x - ray and euv radiation and enhanced absorption is seen .
as the heating accumulates , for the region behind the front , emission becomes dominant .
therefore , either model can explain the mixture of negative and positive contrasts observed in one flare ribbon .
however , as the soft x - ray and euv radiation is not so strong in the beginning , and rises less impulsively than the non - thermal electron flux , it is hard to explain the narrowness of the dark ribbon .
also , it does not explain the associated strong red - shifts and broadening .
therefore , we think that the first scenario , collisional ionization - recombination , is more promising in explaining the observed narrow negative flare front . in conclusion , observations of darkening in 10830 can be used to provide strong constraints in modeling the nonthermal effects of solar flares . by contrast
, both collisional ionization - recombination and photoionization - recombination mechanisms can contribute to the emission that appears in the following bright ribbon .
the decisive physical parameter in determining whether collisional ionization - recombination leads to an absorption or emission is the local plasma temperature . during a strong flare ,
the heating is more efficient and the lower layers are heated to a higher temperature very quickly . under these conditions ,
it is less likely that a negative flare in 10830 will be observed .
however , weak flares , like the low m - class flares studied , have a more moderate heating pattern .
therefore , we are able to see the dark flare source before the lower atmosphere is heated to a high temperature .
although the exact value of the critical temperature is unknown at this moment , a hint is provided by the d3 observations , in which the threshold is about 25000 k @xcite . a more detailed study , especially a correlative numerical simulation , is unavoidably hampered in these events due to the lack of rhessi hard x - ray coverage , which is needed to provide quantitative measurements of the electron energy distribution
nevertheless , this is a significant and pioneering observation providing a rare but promising opportunity to understand the non - thermal effects in solar flares .
furthermore , this unique nst observation provides a solid scientific cornerstone for the future observations with the 4-meter daniel k. inouye solar telescope that will have even higher resolution .
we would like to thank the referee for valuable comments .
the data used in this paper were obtained with nst at big bear solar observatory , which is operated by new jersey institute of technology .
obtaining the excellent data would not have been possible without the help of the bbso team .
bbso operation is supported by njit , us nsf ags-1250818 and nasa nnx13ag14 g , and nst operation is partly supported by the korea astronomy and space science institute and seoul national university , and by strategic priority research program of cas with grant no .
iris is a nasa small explorer mission developed and operated by lmsal with mission operations executed at nasa ames research center and major contributions to downlink communications funded by the norwegian space center ( nsc , norway ) through an esa prodex contract .
the supportive white light data is taken by sdo / hmi .
this work is supported by nsf grants ags-1153424 , ags-1250374 , ags-1408703 , ags-1348513 , and nasa grants nnx-11aq55 g , nnx-13af76 g , nnx-13ag13 g and nnx-11ao70 g to njit .
w. cao was supported by nsf ags-0847126 and nsfc-11428309 .
h. ji was supported by cas xiandao - b with grant xdb09000000 , nsfc-11333009 , nsfc-11173062 , nsfc-11221063 , and the 973 program with grant 2011cb811402 . the work of k .- h .
cho and j. chae was supported by the national research foundation of korea ( nrf - 2012 r1a2a1a 03670387 ) . , l. , hudson , h. , cauzzi , g. , getman , k. v. , giampapa , m. , hawley , s. l. , heinzel , p. , johnstone , c. , kowalski , a. f. , osten , r. a. , & pye , j. 2011 , in astronomical society of the pacific conference series , vol .
448 , 16th cambridge workshop on cool stars , stellar systems , and the sun , ed . c. johns - krull , m. k. browning , & a. a. west , 441 , p. r. , & cao , w. 2012 , in astronomical society of the pacific conference series , vol
. 463 , second atst - east meeting : magnetic fields from the photosphere to the corona . ,
t. r. rimmele , a. tritschler , f. wger , m. collados vera , h. socas - navarro , r. schlichenmaier , m. carlsson , t. berger , a. cadavid , p. r. gilbert , p. r. goode , & m. knlker , 357 725 km ) .
the active region noaa 11818 locates to the southwest from the disk center .
the red box outlines the area of interest ( aoi ) where the negative flare appeared .
right : nst 10830 observation taken at 19:11:59 ut registered with the sdo / hmi map in heliographic coordinates .
the black arrow points to the flare ribbon of interest with a dark frontier .
an image of the earth is displayed in the corner for comparison of size .
the spot on the upper right corner masks a group of hot pixels .
[ fulldisk ] ] line center , 10830 image and sdo / aia 304 image taken almost simultaneously .
the cross located on the dark flare ribbon indicates the position where the h@xmath0 spectra are obtained .
upper right panel : h@xmath0 spectra and the light curves of 10830 and sdo / aia 304 of the cross center .
the h@xmath0 line profiles at 19:06:28 ut and 19:07:28 ut ( see the blue and green lines ) are plotted in the same colors in the lower right panel .
lower right panel : h@xmath0 line profile at three different times , representing the pre - flare ( 19:04:00 ut ) , electron precipitation ( 19:06:28 ut ) and evaporation ( 19:07:28 ut ) stages at the location of the cross .
[ ha ] ] is the light curve of the flare , from which the sudden dip represents the perturbation of leading dark edge .
please note that the second dip is not a dimming but a relative weak emission .
the blue gaussian curve provides an estimate of the length of time that the dark front spends passing across the slit .
the estimate of the maximum decrease in intensity is about @xmath19 , comparing to the pre - flare condition . | solar flares are sudden flashes of brightness on the sun and are often associated with coronal mass ejections and solar energetic particles which have adverse effects in the near earth environment . by definition , flares are usually referred to * bright * features resulting from excess emission . using the newly commissioned 1.6 m new solar telescope at big bear solar observatory , here we show a striking `` negative '' flare with a narrow , but unambiguous `` * * dark * * '' moving front observed in 10830 , which is as narrow as 340 km and is associated with distinct spectral characteristics in h@xmath0 and lines .
theoretically , such negative contrast in 10830 can be produced under special circumstances , by nonthermal - electron collisions , or photoionization followed by recombination .
our discovery , made possible due to unprecedented spatial resolution , confirms the presence of the required plasma conditions and provides unique information in understanding the energy release and radiative transfer in astronomical objects . |
compared to heat transfer by phonons in crystalline materials , heat transfer in amorphous materials is complicated by the existence of three regimes of vibrational modes.@xcite low - frequency propagons are delocalized and have a well - defined wave vector and group velocity,@xcite similar to phonons in crystals , while high - frequency locons are localized and contribute negligibly to thermal conduction.@xcite diffusons have intermediate frequencies and are delocalized , but do not have well - defined wave vectors or group velocities .
the contribution of diffusons to thermal conduction can be notable , however , as they occupy the majority of the vibrational spectrum.@xcite from kinetic theory,@xcite the contribution of an individual phonon or propagon mode to thermal conductivity is proportional to its mean free path ( mfp ) . because diffusons do not have a well - defined group velocity , it is not clear if they have a mfp or how it can be defined .
their contribution to thermal conductivity can be predicted using their diffusivity , which is well - defined and can be calculated from allen - feldman theory.@xcite nevertheless , it would be insightful to identify a frequency - dependent length scale for propagons and diffusons describing the decay of the heat flux at each vibrational frequency .
such a definition would lift the ( fundamentally ) arbitrary distinction between propagons and diffusons and enable a unified description of heat transfer at all vibrational frequencies . in this paper , we apply the spectrally - decomposed mfp method@xcite to probe the non - equilibrium mfps of vibrational heat carriers in amorphous silicon ( a - si ) .
this method is based on calculating the spectrally - decomposed heat current ( sdhc)@xcite in systems of different lengths using non - equilibrium molecular dynamics ( nemd ) simulations .
the mfps are determined from the variation of the sdhc as a function of system length at each vibrational frequency.@xcite we previously used the spectrally - decomposed mfp method to calculate the non - equilibrium mfps in low - dimensional systems such as carbon nanotubes@xcite and anharmonic chains.@xcite we demonstrated that the non - equilibrium mfps transparently describe the ballistic - to - diffusive transition in the length - dependence of thermal conductivity .
compared to previous calculations for a - si , the spectrally - decomposed mfp method has several advantages . unlike in modal life - time calculations,@xcite we do not need to estimate the group velocities of individual modes to calculate their mfps .
we also do not need to distinguish between propagons and diffusons@xcite nor resort to the harmonic approximation.@xcite in contrast to recent calculations studying the spectral conductivity of a - si in fixed - size systems , @xcite we focus on the mfps of heat carriers and the system - size dependence of thermal conductivity .
the rest of the paper is organized as follows .
the calculation methods are presented in sec .
[ sec : setup ] and the numerical results are discussed in sec . [
sec : results ] .
we also introduce a simple method for the quantum correction of thermal conductivity from classical md simulations , based on weighting the sdhc by the quantum occupation function . because this quantum correction method operates at the frequency level
, it is more reasonable than quantum - correction methods that operate at the system level ( see ref . and references therein ) and allows us to compare our predictions to experimental measurements .
nm . the spectral heat flux @xmath1 is calculated at the cross - section in the middle of the structure ( dashed line ) .
the length @xmath2 between the langevin heat baths is varied to extract the vibrational mode mfps based on the decrease of @xmath1 as a function of @xmath2.,width=325 ] all simulations are carried out using the lammps package@xcite with a time step of 2.5 fs .
the si - si interactions are modeled by the stillinger - weber potential@xcite with the parameters of ref . .
the nemd simulation geometry is shown in fig .
[ fig : geom ] .
the atomic coordinates for a - si are generated by following the melt - quench procedure of ref .
and the final density is 2,291 kg / m@xmath3 .
after the equilibration of the quenched system , atoms located within a distance @xmath4 nm from the left and right edges of the structure are coupled to langevin heat baths at temperatures @xmath5 and @xmath6 with bath relaxation times of 1 ps . to prevent sublimation , atoms at the far left and right edges are fixed to their equilibrium positions .
periodic boundary conditions are applied at the boundaries transverse to the current flow .
the width of the system cross - section is @xmath7 nm .
system lengths @xmath2 ( i.e. , the region between the baths ) between 1 and 10 nm at intervals of 1 nm are considered .
the sdhc is calculated through the plane of decomposition located halfway between the hot and cold baths ( dashed line in fig .
[ fig : geom]).@xcite the sdhc @xmath8 between particles @xmath9 and @xmath10 located on opposite sides of this plane is given by the pair - wise sdhc equation @xcite @xmath11 where @xmath12 is the simulation time , @xmath13 is the angular frequency , and the interatomic force constant @xmath14 is defined as @xmath15 the velocities @xmath16 and @xmath17 are the discrete fourier transforms of the atomic velocities @xmath18 and @xmath19 ( the exact definitions are in ref . ) , where @xmath20 and @xmath21 are the displacements of atoms @xmath9 and @xmath10 from their equilibrium positions in directions @xmath22,@xmath23 . in eq . , @xmath24 is the interatomic potential energy function .
the spectral flux through the plane of decomposition is obtained from eq . by summing over all pairs of atoms ( one of the left side , denoted by @xmath25 , and one on the right side , denoted by @xmath26 ) within the potential cut - off distance of each other and dividing by the interface area @xmath27 : @xmath28 while eq .
is the first - order approximation to the inter - particle sdhc,@xcite we have confirmed that the contribution of higher - order terms is negligible for a - si by comparing the integral of eq . , which we denote as @xmath29 , to the total flux determined from the work done by the heat baths .
the two results agree within 4% .
we attribute this good agreement to the stiffness of the interatomic bonds in a - si , which ensures that the first - order term in the current ( proportional to the harmonic force constants @xmath30 ) dominates the higher - order terms that are related to anharmonic force constants .
this restriction to the first - order current term at the plane of decomposition does not , however , mean that anharmonic scattering in the bulk is neglected , because all anharmonic effects are included in the nemd simulations.@xcite frequency - dependent mfps @xmath31 are calculated by determining @xmath32 for different system lengths @xmath2 and fitting the length - dependent @xmath32 to the equation @xcite @xmath33 } , \label{eq : mfp}\ ] ] where @xmath34 is the spectral flux when the baths are in contact ( @xmath35 ) .
both @xmath34 and @xmath31 are determined from the fitting procedure . while @xmath34 depends on the details of the heat baths ,
the mfps extracted from the length - dependence are not expected to depend on the bath details.@xcite the frequency - dependent mfps determined from eq .
are mode - averaged and projected along the direction of heat transfer.@xcite the mfps are independent of the system length and therefore correspond to the bulk values.@xcite we note that the mfps determined from eq .
correspond to the decay length of the heat flux .
this definition does not necessarily coincide with the conventional definition of the mfp as the decay length of a wave packet.@xcite this distinction is important for diffusons , which do not have a well - defined wave - vector so that the traditional definition can not be applied .
once the spectral mfps are determined , the thermal conductivity @xmath36 for length @xmath2 can be determined from@xcite @xmath37}. \label{eq : kappa_est}\ ] ] the length - dependence of the thermal conductivity can be intuitively understood by writing eq . in the equivalent form @xmath38 where the `` effective '' mfp @xmath39 has been introduced , which is similar to the well - known matthiessen rule.@xcite the effective mfp accounts for boundary scattering through the additional @xmath40 term and is limited to below this value .
finally , eq . allows for a simple quantum correction to the thermal conductivity prediction , because the contributions of different frequencies can be weighted by the vibrational mode energy and occupation , as in the landauer - bttiker formalism @xcite .
we define the quantum corrected thermal conductivity as @xmath41 } \frac{\hbar \omega}{k_\mathrm{b } } \frac{\partial f_{\mathrm { be}}(\omega , t)}{\partial t } , \label{eq : kappa_quantum}\ ] ] where @xmath42^{-1}$ ] is the bose - einstein distribution function , @xmath43 is the boltzmann constant , and @xmath44 is the planck constant divided by @xmath45 .
by defining the dimensionless , length - dependent bath - to - bath transmission function as @xmath46},\ ] ] eq .
can be written in the familiar landauer - bttiker form as @xcite @xmath47 the proposed quantum correction accounts for the quantum specific heat of the modes at each frequency , but does not account for quantum effects in the dynamics .
the method is thus similar to the one recently introduced by lv and henry,@xcite who weight the modal contributions to the equilibrium green - kubo thermal conductivity by the quantum population function .
all our nemd simulations are performed at a mean temperature of @xmath48 k with temperature bias @xmath49 k. choosing a relatively large temperature bias allows for very good signal - to - noise ratio in the spectral heat flux , suppressing the statistical noise .
we checked that halving the bias to @xmath50 k does not change the spectral mfps .
in addition , we checked that the heat flux is not sensitive to the exact arrangement of atoms arising from the melt - and - quench procedure , which we attribute to the large cross - section of the system giving rise to spatial averaging in the spectral currents .
therefore , we performed a single melt - quench for each system length .
for different system lengths @xmath2 . increasing the system length reduces the heat flux , especially at high frequencies , where the mfps are shorter . at frequencies below 2 thz ,
the heat current is independent of system length , suggesting ballistic transport.,width=325 ] the spectral heat flux @xmath1 for selected system lengths @xmath2 as a function of frequency @xmath51 is plotted in fig .
[ fig : flux ] . the spectral distribution of the heat flux for @xmath52 nm was recently analyzed in detail by zhou and
hu,@xcite so we focus here on its length - dependence . as expected , increasing the system length reduces the heat current throughout the whole frequency range because of increased phonon - phonon scattering .
the reduction is strongest at high frequencies , where the mfps are shorter compared to low frequencies . at frequencies less than 2 thz , the spectral current is nearly independent of system length .
such nearly ballistic conduction suggest that the low frequency mfps are notably longer than the system sizes considered here .
equation suggests that the inverse of the spectral flux will be linearly proportional to the system length , with the slope given by @xmath53^{-1}$ ] . to determine the mfps
, we calculated the spectral flux for system sizes @xmath54 nm , plotted @xmath55 versus @xmath2 and fitted a linear function at each frequency using least squares fitting.@xcite a linear function accurately reproduces the length - dependence of @xmath55 ( not shown ) , as previously also observed for other systems.@xcite because of the high computational cost associated with calculating the spectral heat fluxes for large systems , we limited our study to systems at most 10 nm long .
this restriction precludes extracting mfps longer than 10 nm accurately , limiting the current analysis to frequencies greater than 2 thz .
the mfps @xmath31 extracted from the linear fitting procedure are shown in a log - log plot in fig .
[ fig : mfps ] . at frequencies below 5 thz ,
the mfps obey a power - law scaling @xmath56
. this scaling agrees with modal life - time calculations on a - si.@xcite at frequencies greater than 5 thz , the power - law scaling breaks down and the mfps increase with increasing frequency , giving rise to a local maximum around 8 thz .
a similar maximum for a - si has been observed in effective mfps @xcite and in lifetimes.@xcite at higher frequencies , the mfps decrease again and fall below 1 nm , which is on the order of the silicon - silicon bond length . at such high frequencies ,
the uncertainty is large because of the sensitivity of the spectral flux to the system size .
larkin and mcgaughey reported a propagon - diffuson transition frequency of 1.8 thz,@xcite such that the frequency range considered in fig .
[ fig : mfps ] mostly corresponds to diffuson - like modes . in the analysis
below , we assume that the scaling @xmath56 ( solid line in fig . [
fig : mfps ] ) remains valid at frequencies below 2 thz .
with such scaling , the mfps exceed 100 nm below 530 ghz and 1 @xmath57 m below 170 ghz .
while the @xmath58 scaling may break down in real situations because of defects , boundary scattering , or even the onset of a rayleigh - like @xmath59 scaling at very low frequencies , @xcite we assume it to hold for simplicity .
we now investigate the length - dependence of the thermal conductivity using eq . .
the calculated thermal conductivity ( continuous line ) is compared with that determined directly from nemd simulations ( data points ) for lengths up to @xmath60 nm in fig .
[ fig : kappa ] . in the evaluation of the integral in eq .
, the mfps have been assumed to scale as @xmath61 at frequencies below 2 thz .
these calculations were carried out without the quantum correction as the nemd simulations are classical .
equation combined with the mfp data of fig .
[ fig : mfps ] reproduces the length - dependence of thermal conductivity up to lengths @xmath62 nm to within 2% . this close agreement ( i )
supports the assumption of @xmath61 scaling at low frequencies , ( ii ) shows that the mfp data in fig . [ fig : mfps ] , which were determined from simulations of systems shorter than @xmath63 nm , can be reliably used to estimate the relative contributions of different vibrational frequencies to thermal transport in much larger systems , and ( iii ) provides support for the accuracy of eq . in describing the length - dependence . because the debye temperature of a - si is @xmath64 k @xcite , which is well above room temperature , we need to apply the quantum correction to compare the predicted temperature - dependence of thermal conductivity to experimental data .
to do so , we use eq . and evaluate the integral as a function of temperature using the mfps from fig .
[ fig : mfps ] , again assuming the scaling @xmath61 at frequencies below 2 thz .
a full quantum - corrected analysis would require determining the mfps at each temperature . for simplicity and based on the recent results of lv and henry @xcite
, we assume that the mfps calculated at a temperature of 300 k remain valid at other temperatures .
below frequencies of 2 thz and to be independent of temperature .
also plotted is the thermal conductivity of a 520 nm thick hydrogenated a - si thin film measured by cahill _
@xcite.,width=325 ] the quantum - corrected thermal conductivity [ eq . ] is plotted as a function of temperature for system lengths of 50 , 250 , and 1000 nm in fig .
[ fig : kappa_qm ] . as noted above ,
assuming finite @xmath2 in eq .
limits the mfps to @xmath40 .
experimental data from cahill _
et al . _ for a 520 nm thick film of
hydrogenated a - si with one atomic percent hydrogen content are also plotted.@xcite because available thermal conductivity measurements for a - si contain significant scatter,@xcite we use the data of cahill _ et al .
_ to check the trend of our predictions , but do not expect agreement .
differences may also exist due to the use of the stillinger - weber potential and the classical nature of the nemd simulations .
the increase of thermal conductivity with increasing temperature is well described by the quantum - corrected thermal conductivity . at temperatures higher than 300 k , the experimentally measured thermal conductivity increases slightly slower as a function of temperature than our prediction , but
this disagreement may be related to our approximation that the mfps are independent of temperature . at such high temperatures ,
anharmonic scattering will reduce mfps and therefore decrease the thermal conductivity . without the quantum - correction ,
the predicted thermal conductivity would depend only very weakly on temperature ( due to the weak temperature - dependence of the mfps ) , precluding reasonable agreement with the trend of the experimental data .
we caution that this quantum correction has only been examined for a - si here and that its application to other systems warrants further investigation .
we investigated vibrational heat transfer in a - si by determining the sdhc and mfps from nemd simulations .
the calculated mfps directly reflect the decay of the heat flux at each vibrational frequency and do not rely on the existence of a well - defined modal wave vector , thereby avoiding the separate treatment of diffusons and propagons . as shown in fig .
[ fig : mfps ] , the mfps exhibit @xmath58 scaling at frequencies above 2 thz and below 5 thz . at
frequencies higher than 10 thz , the mfps fall below 1 nm , corresponding to strongly localized vibrations .
the length - independent mfps can be used to accurately predict the thermal conductivity in systems as long as 100 nm ( fig .
[ fig : kappa ] ) . weighting the sdhc by the frequency - dependent quantum occupation function
provides a simple method for a quantum - correction of thermal conductivity and is able to reproduce the experimentally measured temperature - dependence of thermal conductivity , as shown in fig .
[ fig : kappa_qm ] . in the future , it would be useful to calculate the sdhc for systems longer than those considered here , enabling direct extraction of mfps at frequencies below 2 thz .
such an analysis could inform the ongoing discussion@xcite of the low - frequency scaling of mfps in a - si .
it will also be important to compare the non - equilibrium mfps to those calculated from equilibrium molecular dynamics simulations .
this work was initiated during the research visit of k.s .
to carnegie mellon university .
the computational resources were provided by the finnish it center for science and aalto science - it project .
the work was partially funded by the aalto energy efficiency research programme ( aef ) and the academy of finland .
23ifxundefined [ 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 10.1080/13642819908223054 [ * * , ( ) ] link:\doibase 10.1103/physrevb.48.12589 [ * * , ( ) ] link:\doibase 10.1103/physrevb.64.094201 [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1103/physrevb.48.12581 [ * * , ( ) ]
link:\doibase 10.1103/physrevb.91.115426 [ * * , ( ) ] link:\doibase 10.1103/physrevb.90.134312 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.245411 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.89.144303 [ * * , ( ) ] http://stacks.iop.org/1367-2630/18/i=1/a=013028 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.195205 [ * * , ( ) ] link:\doibase 10.1103/physrevb.79.224305 [ * * , ( ) ] \doibase http://dx.doi.org/10.1006/jcph.1995.1039 [ * * , ( ) ] link:\doibase 10.1103/physrevb.31.5262 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/s0022-3093(01)00342-8 [ * * , ( ) ] http://stacks.iop.org/0953-8984/26/i=5/a=055011 [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1103/physrevlett.81.232 [ * * , ( ) ] \doibase doi : 10.1006/spmi.1997.0561 [ * * , ( ) ] link:\doibase 10.1103/physreve.88.012128 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0038-1098(84)90589-1 [ * * , ( ) ] link:\doibase 10.1103/physrevb.50.6077 [ * * , ( ) ] | the frequency - dependent mean free paths ( mfps ) of vibrational heat carriers in amorphous silicon are predicted from the length dependence of the spectrally decomposed heat current ( sdhc ) obtained from non - equilibrium molecular dynamics simulations .
the results suggest a ( frequency)@xmath0 scaling of the room - temperature mfps below 5 thz .
the mfps exhibit a local maximum at a frequency of 8 thz and fall below 1 nm at frequencies greater than 10 thz , indicating localized vibrations .
the mfps extracted from sub-10 nm system - size simulations are used to predict the length - dependence of thermal conductivity up to system sizes of 100 nm and good agreement is found with separate molecular dynamics simulations . weighting the sdhc by the frequency - dependent quantum occupation function provides a simple and convenient method to account for quantum statistics and provides reasonable agreement with the experimentally - measured trend and magnitude . |
the measurement of spin observables gives very important information on the structure of the strong interactions @xcite .
recently , the hermes collaboration has found a significant ( @xmath9 20@xmath10 ) double - spin asymmetry in elastic vector meson electroproduction at energy @xmath11 gev @xcite .
this result is quite intriguing since it was not expected within models of the vector meson production processes based on convenient mesonic and pomeron exchanges .
there are two different approaches to describe the electromagnetic production of light vector mesons at intermediate energies @xcite and @xcite .
the first assumes that only @xmath12 and @xmath13 exchanges are relevant for explaining the @xmath0 and @xmath14 cross sections and their relation .
the second involves the pomeron , and the @xmath12 , @xmath13 and @xmath15secondary reggeon exchanges . in both of the approaches the main contribution to the @xmath0 production cross section comes from the natural parity exchanges , e.g. pomeron , @xmath16 and @xmath1 . in spite of the fact that the @xmath17 coupling constant is very large , the contribution of unnatural parity @xmath12meson exchange to the elastic cross - section is small and falls off rapidly with energy . at hermes energies its value is negligible in both models . for spin observables the parity of the exchanges plays a crucial role .
for example , the double - spin asymmetry will not vanish only if there is a considerable interference between the exchanges with natural and unnatural parities .
it turns out , that the hermes data are impossible to describe by interference of the pion exchange with any existing natural parity exchanges .
recently we suggested a new unnatural parity anomalous @xmath2 trajectory , with a very high intercept @xmath18 and a small slope @xmath19 , to explain the behavior of the elastic proton - proton , proton - antiproton and vector meson photoproduction cross - sections at large energies and momentum transfers @xcite .
the connection of this new trajectory with spin physics has been stressed , and its importance for the low @xmath20 behavior of the nucleon spin - dependent structure function @xmath21 and the double spin asymmetries in diffractive reactions at large energies has been shown @xcite .
in here we analyze the contributions of the @xmath2 exchange to the double - spin asymmetries in @xmath0-meson electromagnetic production at intermediate energies . and @xmath22-meson production is straightforward . ]
the kinematics of the @xmath0-meson electromagnetic production off the nucleon is shown in fig.[diag ] .
the invariant variables of the reaction are @xmath23 and @xmath24 , where @xmath25 and @xmath26 are the momenta of the initial photon and proton , @xmath27 and @xmath28 are the momenta of the final @xmath0-meson and proton . at hermes
energies the mechanism for vector meson production is rather complicated .
we will take into account the contributions to the amplitude related to the @xmath29 channel pomeron and the anomalous @xmath30 exchanges @xcite , with a weak energy dependence , together with the secondary regge @xmath12 , @xmath13 and @xmath15 exchanges , which have a rather strong energy dependence - pole exchange related to axial anomaly @xcite .
its contribution is important at rather large @xmath31 .
hermes only measures elastic vector meson production cross section at low @xmath32 . ] . for the pomeron exchange
we use the donnachie - landshoff model @xcite .
the pomeron exchange amplitude is @xmath33 where @xmath34 is the mass of the @xmath0 meson and @xmath35 is its leptonic width .
the quark - pomeron coupling @xmath36 can be fixed from fits @xcite to total hadron - hadron cross sections . for the mass scale @xmath37 = 4 gev@xmath8
one obtains @xmath36 = 2.0 gev@xmath38 .
the pomeron trajectory is given by @xcite @xmath39 the form factor of the pomeron - nn vertex is @xcite @xmath40 it is also necessary to include an additional factor which takes into account the nonlocality of the pomeron vertex @xcite , @xmath41 where @xmath42 = 1.1 gev@xmath8 .
this formula was obtained under the assumption that the effective quark - quark scattering amplitude induced by the pomeron has the following form @xmath43 which is supported by the landshoff - nachtmann two - gluon model of the pomeron @xcite . since at hermes the energy is not very high , it is also necessary to take into account the contribution from secondary reggeons .
the most important reggeons in the reaction under consideration are the @xmath44 , @xmath16 and @xmath1-trajectories .
the usual approach to estimate their contribution to vector meson production is to fix their parameters from some low energy decay amplitudes and from fitting the total and differential elastic cross sections for vector meson production @xcite and @xcite .
therefore the predictive power of such approaches is rather low . to avoid this problem
, we calculate the contribution from secondary regge exchanges using only information extracted from proton - proton and proton - antiproton scattering in analogy to the donnachie - landshoff approach to the calculation of the pomeron contribution to vector meson production @xcite . within our approach
one determines the parameters of the secondary regge exchanges between two quarks from the hadron - hadron cross - sections .
then by using the assumption on a nonrelativistic shape for the vector meson wave function , normalized to the leptonic decay width @xmath35 , one calculates the contribution of any exchanges to the vector meson production amplitude free of parameters . within this model , using the following forms for effective @xmath44-quark and @xmath16-quark interactions
@xmath45 we obtain for the amplitudes @xmath46 where @xmath47 is the value of the coupling of the @xmath48 meson to quarks , and we use for the pion - nucleon coupling @xmath49 @xcite .
the @xmath44-meson form factor of the nucleon is given by @xcite @xmath50 with @xmath51 .
for the @xmath16 exchange one has @xmath52 the parameters of the @xmath16 exchange are taken from @xcite : @xmath53=0.55 gev , @xmath54 , @xmath55 = 10.2 , @xmath56 with @xmath57 . in ( [ pion ] ) and ( [ sigma ] ) the regge factor
@xmath58 is @xcite @xmath59 where @xmath60 and @xmath61 is the mesonic regge trajectory @xmath62 with slope @xmath63 .
we will assume with refs .
@xcite and @xcite that the coupling of @xmath1 to the quarks has the same form as that of the pomeron ( [ pom ] ) .
therefore the formula for the @xmath1 amplitude is similar to ( [ pomampl ] ) changing the parameters to : @xmath64 = 3.5 gev@xmath38 , @xmath65 = 1 gev@xmath8 , and the trajectory is now @xmath66 .
the values of parameters have been obtained from a fit to the total hadron - hadron cross - sections @xcite . in a similar manner
one can obtain the unnatural parity @xmath30trajectory contribution to the @xmath0-meson production amplitude . the @xmath2 interaction with the quarks
is @xmath67 by using this coupling we have for the amplitude , @xmath68 in ( [ f1 ] ) the coupling of @xmath2 to the quarks , @xmath69 = 2.5 , is fixed by using a constituent quark model and the result of the analysis of the proton spin problem within @xmath2 anomalous exchange model @xcite .
the form factor is given by @xmath70 with @xmath71 .
the result of the calculation of @xmath44- , @xmath16- , @xmath15 , @xmath30 contributions to the total elastic @xmath0 photoproduction cross - section as a function of the photon energy @xmath72 are shown in fig.([crs ] ) .
our model describes the experimental data well without tuning the parameters .
it is evident that for hermes kinematics , @xmath73 @xcite , the main contribution to the cross section comes from the pomeron and @xmath1 exchange and the contributions from @xmath44 and @xmath16 exchanges are very small and therefore can be neglected .
the contribution of the @xmath2 exchange near the @xmath0 meson production threshold is smaller then the @xmath44 and @xmath16 contributions .
but due to the very high intercept , the @xmath2 anomalous trajectory , has a non vanishing contribution at hermes energies . due to its unnatural parity and negative signature , the corresponding amplitude is proportional to the product of helicities of the photon and the proton
. therefore the interference of the @xmath2 amplitude with the natural parity amplitudes leads to a nonzero double spin asymmetry , @xmath74 where @xmath75 are the cross - sections for @xmath0-meson production from different photon - proton helicity states . due to the large intercept @xmath18 and negative signature ,
the amplitude of the @xmath2 exchange is real with high accuracy .
therefore one can expect that its interference with the nearly imaginary pomeron amplitude is very small and its contribution to the double - spin asymmetry is also small . at the same time
the @xmath1 amplitude contains a significant real part and , as consequence , the main contribution to the double - spin asymmetry should come from the interference between the @xmath1 and the @xmath2 exchanges . within our model
all amplitudes have the same @xmath76 dependence and therefore the double - spin asymmetry does not depend on the value of @xmath76 . in fig.[asymm ] we compare the contribution to the asymmetry from the @xmath2-pomeron ( dotted line ) , @xmath2-@xmath1 ( dashed - dot line ) interferences and the total result ( solid line ) with hermes data for an average photon virtuality @xmath77 = 1.7 gev@xmath8 . in the same figure
we show the result for a slightly different set of @xmath1 trajectory parameters @xcite by a dashed line .
as one can see , the contribution of the @xmath2-pomeron interference is negative and rather small .
the main contribution comes , as it was to be expected , from the @xmath1-@xmath2 interference and the total contribution explains the hermes data within error bars .
we propose a new mechanism to explain the double - spin asymmetry in @xmath0-meson electromagnetic production off the nucleon at intermediate energies .
this mechanism is related to the existence of a new anomalous unnatural parity , negative signature @xmath2 regge trajectory which was introduced recently in @xcite to explain features of diffractive reactions at large momentum transfers and the anomalous behavior of the spin - dependent structure function @xmath21 .
the interference of the @xmath2 exchange with the natural parity @xmath1 secondary reggeon exchange gives the very large contribution to the double spin asymmetry .
the calculated value for the total asymmetry is @xmath78 , which is compatible with the measured value 24@xmath7911% . for @xmath22- and @xmath80 mesons
the main contribution to the double - spin asymmetry should come from interference of pomeron and the @xmath2 trajectory , because the @xmath1 couplings with these mesons vanish in ozi limit .
therefore one can expect small and negative double - spin asymmetries for both mesons .
we have shown that the @xmath2 anomalous exchange gives the biggest contribution to the spin - dependent cross sections of vector meson production , not only at very large energies @xcite , but also in intermediate energy range .
we are grateful to a.borissov and k.lipka for useful discussion .
this work was partially supported by seui - bfm2001 - 0262 , rfbr-01 - 02 - 16431 , intas-2000 - 366 grants and bk21 and heisenberg - landau programs .
the work of av , nk and dpm is also supported in part by kosef 199 - 2 - 111 - 005 - 5 and krf 2001 - 015-dp0085 .
99 r.l.jaffe , hep - ph/0101280 .
a. airapetian _ et al . _
[ hermes collaboration ] , phys .
* b513 * , 301 ( 2001 ) . b. friman and m. soyeur , nucl .
* a600 * , 477 ( 1996 ) . y. oh , a. i. titov and t. s. lee , nucl - th/0004055 . j. m. laget and r. mendez - galain , nucl .
* a581 * , 397 ( 1995 ) , + j. m. laget , phys .
* b489 * , 313 ( 2000 ) . n. i. kochelev , d. p. min , y. oh , v. vento and a. v. vinnikov , phys .
* d61 * , 094008 ( 2000 ) .
y. oh , n. i. kochelev , d. p. min , v. vento and a. v. vinnikov , phys . rev . * d62 * , 017504 ( 2000 )
. m. guidal , j. m. laget and m. vanderhaeghen , nucl .
* a627 * , 645 ( 1997 ) . n. i. kochelev and v. vento , phys .
* b515 * , 375 ( 2001 ) ; + hep - ph/0110268 . v. g. stoks , r. timmermans and j. j. de swart , phys .
* c47 * , 512 ( 1993 ) .
r. machleidt , adv .
* 19 * , 189 ( 1989 ) .
a. donnachie and p. v. landshoff , phys .
* b185 * , 403 ( 1987 ) .
a. donnachie and p. v. landshoff , phys .
* b296 * , 227 ( 1992 ) p. v. landshoff and o. nachtmann , z. phys .
* c35 * ( 1987 ) 405 .
d. e. groom _ et al . _
[ particle data group collaboration ] , eur .
j. * c15 * , 231 ( 2000 ) .
w. struczinski _ et al . _
[ aachen - hamburg - heidelberg - munich collaboration ] , nucl .
* b108 * , 45 ( 1976 ) , + y. eisenberg _ et al . _ ,
* b42 * , 349 ( 1972 ) , + j. breitweg _ et al .
_ [ zeus collaboration ] , eur . phys .
j. * b2 * , 247 ( 1998 ) ; + y. a. aleksandrov _ et al .
_ , yad . fiz .
* 32 * , 651 ( 1980 ) ; + r. m. egloff _ et al .
_ , phys .
lett . * 43 * , 657 ( 1979 ) ; + j. barth _ et al .
_ [ saphir collaboration ] , acta phys .
* b29 * , 3321 ( 1998 ) . | we calculate the contribution of meson and pomeron exchanges to the double - spin asymmetry in @xmath0-meson electromagnetic production at hermes energies .
we show that the observed double - spin asymmetries , which are large , can be explained by the interference between the natural parity @xmath1-secondary reggeon and the unnatural parity anomalous @xmath2 exchanges .
170 mm 250 mm -5 mm -10 mm * a mechanism for the double - spin asymmetry in electromagnetic @xmath0 production at hermes * n.i .
kochelev,@xmath3 d .- p .
min,@xmath4 v. vento,@xmath5 and a.v .
vinnikov@xmath6 _ _ @xmath7 bogoliubov laboratory of theoretical physics , jinr , dubna , moscow region , 141980 russia + @xmath8 institute of physics and technology , almaty , 480082 , kazakhstan + @xmath4 school of physics and center for theoretical physics , + seoul national university , seoul 151 - 747 , korea + @xmath5 departament de fsica terica and institut de fsica corpuscular , universitat de valncia - csic e-46100 , burjassot ( valencia ) , spain |
wide range equation of state ( eos ) for hydrogen or its isotopes is of crucial interest for inertial confinement fusion ( icf ) and astrophysics @xcite . in the traditional central - hot - spot ignition designs of icf
, a deuterium - tritium ( d - t ) capsule is assumed to be imploded to high density either directly by high power laser pulses @xcite or indirectly by x rays generated in the hohlraum @xcite . due to the fact that the compressibility of the capsule is determined by eos ,
high precision eos of the d - t fuel is essential for hydrodynamic simulations and ignition facility designs . in astrophysics ,
the giant jovian planets , such as jupiter , saturn , uranus , and neptune , are composed primarily of hydrogen and helium .
the knowledge on the size and mass distribution of giant jovian planets are sensitive to the eos of hydrogen in a wide range @xcite .
the eos of hydrogen has been probed through gas gun @xcite , converging explosive @xcite , magnetically driven flyer @xcite , and high power laser - driven experiments @xcite , where a pressure - temperature thermodynamical domain with amplitudes of megabar ( mbar ) and electron volt ( ev ) has been reached .
theoretical approximations , such as classical molecular dynamics based on interatomic potentials @xcite , linear mixing method @xcite , fluid variational theory ( fvt ) @xcite , path integral monte carlo ( pimc ) @xcite , and quantum molecular dynamics ( qmd ) @xcite , have already been employed to study high - pressure behaviors of hydrogen and its isotopes .
although it has been pursued over decades , there are still some fundamental issues worth to be noticed .
for instance , the first - order phase transition from molecular to atomic fluid transition is being under intense discussion , and plasma phase transition ( ppt ) characterized by electronic ionization still needs to be clarified @xcite . high precision , wide range eos for hydrogen are of particularly importance for hydrodynamic simulations in icf , especially at densities from @xmath4 to @xmath5 g/@xmath2 and temperatures up to 10@xmath6 k , or even higher .
currently , sesame - eos table @xcite for hydrogen describes chemical species , such as , h@xmath7 molecules , h atoms , and free protons and electrons based on chemical models @xcite , which are only expected to work well in the weak coupling limit .
recently , a new eos table based on pimc simulations has been built , however , pimc results are not consistent with experiments at pressures below 50 gpa along the hugoniot curve @xcite . as a consequence , wide range eos for hydrogen
is highly recommended to be constructed from other promising ways for comparisons and applications . in the present work , a combined density functional theory ( dft )
method of first - principles molecular dynamics ( fpmd ) and orbital - free molecular dynamics ( ofmd ) has been used to construct wide range eos for fluid hydrogen with a temperature range @xmath8 k and density range @xmath9 g/@xmath2 . in fpmd
the electrons are treated quantum mechanically through finite - temperature dft ( ftdft ) with the only approximation of exchange - correlation functional .
due to the fermi - dirac distribution of the electronic states , at extremely high temperatures a huge number of occupational bands have to be introduced , and fpmd simulations are then restricted . as a consequence ,
ofmd simulations , where the electronic kinetic energy is expressed as a functional of the local electronic density and possibly of its gradient , have been adopted to avoid the limitation .
the rest of this paper is organized as follows .
section ii describes the computational methods with respect to fpmd and ofmd . in section iii
, we discuss the eos in detail , and finally we get our conclusions in section iv .
in this section , we briefly describe the basic formalism employed to explore thermodynamic properties of fluid hydrogen .
that is , two basic quantum - mechanical dft approaches , one based on kohn - sham ( ks ) formula and the other based on orbital - free method .
then the simulation parameters are presented in detail .
our fpmd simulations for fluid hydrogen have been performed by using abinit code @xcite . in these simulations ,
the electrons are fully quantum mechanically treated by employing a plane - wave ft - dft description , where the electronic state occupations follow the fermi - dirac distribution .
the ions move classically according to the forces from the electron density and the ion - ion repulsion .
we employed the nvt ( canonical ) ensemble , where the number of particles @xmath10 and the volume are fixed @xcite .
the system was assumed to be in local thermodynamic equilibrium with the electron and ion temperatures being equal ( @xmath11 ) . in these calculations ,
the electronic temperature was been kept constant according to the fermi - dirac distribution , while the ionic temperature was controlled by the noe thermostat . at each step during md simulations ,
a set of electronic state functions @xmath12 for each @xmath13-point were determined within ks construction by @xmath14 with @xmath15 in which the four terms respectively represent the kinetic contribution , the electron - ion interaction , the hartree contribution , and the exchange - correlation functional .
the electronic density was obtained by @xmath16 then by applying the velocity verlet algorithm , based on the force from interactions between ions and electrons , a new set of positions and velocities were obtained for ions .
ofmd simulations @xcite , where the kinetic energy of the electrons is treated semiclassically , have also been used to investigate the wide range eos for fluid hydrogen under extreme conditions .
the orbital - free electronic free energy can be expressed as @xmath17\}+\int drh(n)\frac{|\bigtriangledown n|^{2}}{n}\\ & + f_{xc}[n]+\frac{1}{2}\int\int drdr'\frac{n(r)n(r')}{|r - r'|}+\sum_{\ell=1}^{n_{\alpha}}z_{\ell}\int dr\frac{n(r)}{r - r_{\ell}}\\ & -\mu\int dr[n(r)-\sum_{\ell=1}^{n_{\alpha}}z_{\ell } ] , \end{split}\ ] ] where @xmath18 is the fermi integral of order @xmath19 , and the screened potential @xmath20 is related to the electronic density by @xmath21).\ ] ] the first integral in eq .
( [ equation_ofenergy ] ) , which depends only on the local electronic density in the true spirit of the hohenberg - kohn theorem , is the well - known finite - temperature thomas - fermi expression @xcite . the second term in eq . ( [ equation_ofenergy ] )
denotes the von werzscker correction . in the present simulations we have omitted this gradient term and worked in a thomas - fermi - dirac form using the formula proposed by perrot @xcite to deal with the kinetic - entropic part .
the orbital - free procedure treats all electrons on an equal footing , albeit approximately , with no distinction between bound and ionized electrons .
except for that , the ofmd simulation procedure is similar to that of fpmd . using the above - mentioned dft formalisms ( namely , fpmd and ofmd )
, we aim to build a wide range dft - eos table of data points for liquid hydrogen with the density ranging from @xmath0 to @xmath1 g/@xmath2 and temperature from @xmath5 to @xmath3 k. generally , the coulomb liquids can be characterized by two non - dimensional parameters .
that is , the ionic coupling parameter and electronic degenerate parameter . for liquid hydrogen , the former one is commonly defined as @xmath22 , which presents the ratio of the mean electrostatic potential energy and the mean kinetic energy of the ions .
the degeneracy parameter @xmath23 is the ratio of the temperature to the fermi temperature @xmath24 . within the fpmd formalism
, the electronic states are occupied according to the fermi - dirac distribution .
thus , our fpmd simulations have been restricted to temperatures lower than @xmath25 ( @xmath26 ) at @xmath27 g/@xmath2 . for lower densities ,
our fpmd simulations have been performed up to a temperature of 15.682 ev . to overcome the computational cost limit ,
ofmd was used in the same simulated conditions ( density and temperature ) as those in fpmd , and explored to extend to higher temperatures .
the results indicate that both of the pressure and internal energy difference are better than 2% between qmd and ofmd simulations as @xmath28 ( see fig . [ fig_qmd_ofmd_compare ] ) .
( red open circles ) , 7.8545 g/@xmath2 ( red open squares ) , and 42.095 g/@xmath2 ( red open diamonds ) .
qmd results have been plotted as the black dashed line .
each curve corresponds to an isochore .
each curve has been shifted by 1.0 from the previous one for clarity.,height=226 ] in general , fpmd and ofmd simulations based on dft have introduced pseudopotentials to reduce the computational cost and ensure the accuracy at moderate densities .
however , the pseudo - core approximation fails at high densities , where the interatomic distance is comparable with or smaller than the cutoff radius of the pseudopotential , due to pressure - induced delocalization of the core electrons and the overlapping of the pseudization spheres . in order to avoid the limitations introduced by pseudopotential approximation , a columbic pseudopotential with a cutoff radius of 0.001 a.u .
has been built @xcite . as the energy dependence is better than 1% between projector augmented wave ( paw ) potentials and columbic potential ( see fig .
[ fig_pot ] ) , we explore the eos of hydrogen into high density ( @xmath29 g/@xmath2 ) by using a short cutoff radius columbic potential .
we have considered a total number of 8 @xmath30 512 atoms ( corresponding to expanded and ultra dense regimes ) in a series of volume - fixed supercells , which are repeated periodically throughout the space . only @xmath31 point is used to sample the brillouin zone in molecular dynamic simulations , because the selection of higher number of @xmath13 points modifies the eos within 3% .
each system was assumed to be in local thermodynamic equilibrium with the electron and ion temperatures being equal ( @xmath32 ) . in order to balance the pseudopotential approximation in the high density regime and the computational cost ,
two potentials have been adopted in both fpmd and ofmd simulations .
that is , the paw ( with @xmath33 a.u . )
pseudopotential ( @xmath34 30 g/@xmath2 ) and short cutoff radius coulombic potential ( @xmath35 20 g/@xmath2 ) , where the plane wave cutoff energy is set to 200 ha , respectively .
the exchange - correlation functional is determined by local density approximation ( lda ) with teter - pade parametrization @xcite , and the temperature dependence of exchange - correlation functional , which is convinced to be as small as negligible , is not taken into account .
n@xmath36 = 6000 has been used in the molecular dynamic simulations , and the time steps are selected with considering different density and temperature @xcite .
the eos are obtained as running average of the last 1000 steps of molecular dynamic simulations .
additionally , in fpmd simulations , sufficient electronic states have been adopted to secure the occupational number below @xmath37 .
a wide range dft - eos ( listed in table i ) has been constructed by data obtained from fpmd ( for @xmath38 ) and ofmd ( @xmath39 ) simulations .
results are compared with previous theoretical and experimental ones in this section .
r c c + & & + continued from previous page + & & + + & @xmath40g/@xmath2 & + 1.348 & 0.138 & 6.290 + 2.695 & 0.351 & 13.021 + 5.391 & 0.864 & 26.066 + 8.215 & 1.393 & 35.787 + 10.781 & 1.870 & 46.596 + 15.682 & 2.698 & 62.313 + 21.563 & 3.862 & 80.142 + 34.500 & 6.316 & 119.051 + 43.125 & 7.950 & 144.951 + 86.250 & 16.114 & 274.346 + 172.500 & 32.441 & 533.100 + 345.000 & 65.097 & 1050.593 + 690.000 & 130.407 & 2085.570 + 1293.750 & 244.701 & 3896.774 + 4312.500 & 816.153 & 12952.800 + + & @xmath41g/@xmath2 & + 1.348 & 0.212 & 6.113 + 2.695 & 0.542 & 12.339 + 5.391 & 1.367 & 25.339 + 8.215 & 2.239 & 35.584 + 10.781 & 3.084 & 45.612 + 15.682 & 4.431 & 61.651 + 21.563 & 6.107 & 79.572 + 34.500 & 10.004 & 118.522 + 43.125 & 12.598 & 144.430 + 86.250 & 25.561 & 273.838 + 172.500 & 51.483 & 532.598 + 345.000 & 103.324 & 1050.092 + 690.000 & 207.002 & 2085.071 + 1293.750 & 388.453 & 3896.275 + 4312.500 & 1295.667 & 12952.301 + + & @xmath42g/@xmath2 & + 1.348 & 0.359 & 5.927 + 2.695 & 0.908 & 11.683 + 5.391 & 2.298 & 24.074 + 8.215 & 3.827 & 34.673 + 10.781 & 5.188 & 44.264 + 15.682 & 7.739 & 60.708 + 21.563 & 10.493 & 78.773 + 34.500 & 17.231 & 117.794 + 43.125 & 21.715 & 143.715 + 86.250 & 44.118 & 273.146 + 172.500 & 88.912 & 531.914 + 345.000 & 178.498 & 1049.412 + 690.000 & 357.665 & 2084.391 + 1293.750 & 671.210 & 3895.596 + 4312.500 & 2238.934 & 12951.622 + + & @xmath43g/@xmath2 & + 1.348 & 0.676 & 5.732 + 2.695 & 1.710 & 10.860 + 5.391 & 4.316 & 22.440 + 8.215 & 7.278 & 33.087 + 10.781 & 9.943 & 41.650 + 15.682 & 14.981 & 59.245 + 21.563 & 20.322 & 77.561 + 34.500 & 33.495 & 116.710 + 43.125 & 42.257 & 142.658 + 86.250 & 86.019 & 272.126 + 172.500 & 173.515 & 530.908 + 345.000 & 348.498 & 1048.412 + 690.000 & 698.457 & 2083.394 + 1293.750 & 1310.879 & 3894.600 + 4312.500 & 4373.030 & 12950.626 + + & @xmath44g/@xmath2 & + 1.348 & 1.463 & 5.066 + 2.695 & 3.884 & 9.320 + 5.391 & 9.334 & 18.976 + 8.215 & 16.559 & 30.688 + 10.781 & 22.833 & 39.750 + 15.682 & 34.787 & 56.895 + 21.563 & 49.460 & 75.601 + 34.500 & 80.720 & 115.000 + 43.125 & 101.500 & 141.001 + 86.250 & 204.800 & 270.546 + 172.500 & 410.900 & 529.354 + 345.000 & 823.100 & 1046.868 + 690.000 & 1655.046 & 2081.854 + 1293.750 & 3106.100 & 3893.061 + 4312.500 & 10365.235 & 12949.087 + + & @xmath45g/@xmath2 & + 1.348 & 2.472 & 4.493 + 2.695 & 6.411 & 8.595 + 5.391 & 16.101 & 18.019 + 8.215 & 27.862 & 29.063 + 10.781 & 38.592 & 37.668 + 15.682 & 59.760 & 55.062 + 21.563 & 84.600 & 73.978 + 34.500 & 138.700 & 113.586 + 43.125 & 174.600 & 139.635 + 86.250 & 353.100 & 269.251 + 172.500 & 709.600 & 528.081 + 345.000 & 1425.665 & 1045.603 + 690.000 & 2859.130 & 2080.591 + 1293.750 & 5367.606 & 3891.798 + 4312.500 & 17910.085 & 12947.823 + + & @xmath46g/@xmath2 & + 1.348 & 4.264 & 4.712 + 2.695 & 12.140 & 8.083 + 5.391 & 30.746 & 16.663 + 8.215 & 52.677 & 27.011 + 10.781 & 73.257 & 35.436 + 15.682 & 112.734 & 52.641 + 21.563 & 162.600 & 71.692 + 34.500 & 268.500 & 111.558 + 43.125 & 339.100 & 137.679 + 86.250 & 682.388 & 267.396 + 172.500 & 1382.377 & 526.260 + 345.000 & 2782.234 & 1043.791 + 690.000 & 5581.767 & 2078.781 + 1293.750 & 10481.126 & 3889.987 + 4312.500 & 34977.421 & 12946.009 + + & @xmath47g/@xmath2 & + 1.348 & 5.218 & 4.035 + 2.695 & 14.462 & 7.844 + 5.391 & 33.584 & 17.101 + 8.215 & 62.068 & 26.480 + 10.781 & 86.394 & 34.845 + 15.682 & 133.412 & 50.345 + 21.563 & 192.000 & 71.017 + 34.500 & 318.500 & 110.942 + 43.125 & 401.800 & 137.082 + 86.250 & 809.780 & 266.829 + 172.500 & 1641.220 & 525.702 + 345.000 & 3303.920 & 1043.235 + 690.000 & 6629.240 & 2078.225 + 1293.750 & 12448.600 & 3889.430 + 4312.500 & 41545.000 & 12945.451 + + & @xmath48g/@xmath2 & + 0.259 & 2.694 & 0.854 + 0.431 & 3.948 & 1.153 + 0.518 & 4.584 & 1.320 + 0.690 & 5.537 & 1.717 + 0.863 & 7.924 & 1.836 + 1.348 & 12.306 & 3.131 + 2.695 & 28.755 & 7.031 + 5.391 & 71.782 & 14.766 + 8.215 & 121.424 & 23.631 + 10.781 & 167.717 & 32.418 + 15.682 & 260.606 & 47.720 + 21.563 & 375.300 & 67.930 + 34.500 & 627.300 & 107.984 + 43.125 & 794.300 & 134.196 + 86.250 & 1609.411 & 264.053 + 172.500 & 3268.481 & 522.956 + 345.000 & 6586.363 & 1040.493 + 690.000 & 13222.325 & 2075.477 + 1293.750 & 24835.110 & 3886.676 + 4312.500 & 82900.032 & 12942.686 + + & @xmath49g/@xmath2 & + 0.259 & 4.773 & 0.752 + 0.431 & 6.561 & 1.122 + 0.518 & 7.336 & 1.311 + 0.690 & 9.403 & 1.733 + 0.863 & 10.704 & 2.281 + 1.348 & 18.789 & 3.426 + 2.695 & 45.908 & 6.683 + 5.391 & 110.191 & 14.078 + 8.215 & 185.120 & 22.566 + 10.781 & 253.965 & 30.910 + 15.682 & 395.081 & 45.969 + 21.563 & 569.300 & 65.689 + 34.500 & 955.800 & 105.708 + 43.125 & 1212.000 & 131.937 + 86.250 & 2489.000 & 261.841 + 172.500 & 5013.911 & 520.751 + 345.000 & 10110.722 & 1038.280 + 690.000 & 20305.244 & 2073.253 + 1293.750 & 38145.083 & 3884.443 + 4312.500 & 127346.277 & 12940.439 + + & @xmath50g/@xmath2 & + 0.259 & 7.194 & 0.779 + 0.345 & 8.599 & 0.958 + 0.431 & 9.692 & 1.133 + 0.518 & 10.432 & 1.363 + 0.690 & 13.249 & 1.753 + 0.863 & 15.283 & 2.267 + 1.348 & 26.465 & 3.124 + 2.695 & 60.048 & 6.458 + 5.391 & 135.419 & 14.176 + 8.215 & 235.860 & 21.950 + 10.781 & 320.826 & 30.080 + 15.682 & 499.007 & 44.982 + 21.563 & 720.600 & 64.364 + 34.500 & 1209.000 & 104.306 + 43.125 & 1537.000 & 130.534 + 86.250 & 3158.000 & 260.445 + 172.500 & 6396.000 & 519.350 + 345.000 & 12849.073 & 1036.869 + 690.000 & 25809.147 & 2071.833 + 1293.750 & 48491.275 & 3883.014 + 4312.500 & 161900.916 & 12938.999 + + & @xmath51g/@xmath2 & + 0.259 & 11.871 & 0.798 + 0.345 & 13.700 & 0.975 + 0.431 & 15.444 & 1.162 + 0.518 & 16.105 & 1.451 + 0.690 & 18.874 & 1.889 + 0.863 & 22.934 & 2.267 + 1.348 & 36.898 & 3.296 + 2.695 & 82.165 & 6.293 + 5.391 & 186.413 & 13.224 + 8.215 & 298.231 & 21.654 + 10.781 & 414.411 & 29.201 + 15.682 & 643.852 & 43.913 + 21.563 & 924.300 & 62.908 + 34.500 & 1562.000 & 102.714 + 43.125 & 1984.000 & 128.927 + 86.250 & 4094.000 & 258.819 + 172.500 & 8293.000 & 517.709 + 345.000 & 16672.990 & 1035.209 + 690.000 & 33498.380 & 2070.157 + 1293.750 & 62944.837 & 3881.326 + 4312.500 & 210176.125 & 12937.294 + + & @xmath52g/@xmath2 & + 0.259 & 22.170 & 0.881 + 0.345 & 24.459 & 1.064 + 0.431 & 24.239 & 1.349 + 0.518 & 26.790 & 1.617 + 0.690 & 31.156 & 1.949 + 0.863 & 37.359 & 2.291 + 1.348 & 57.070 & 3.257 + 2.695 & 117.535 & 6.153 + 5.391 & 255.253 & 12.911 + 8.215 & 414.266 & 20.725 + 10.781 & 566.886 & 28.198 + 15.682 & 852.768 & 42.767 + 21.563 & 1220.000 & 61.324 + 34.500 & 2066.000 & 100.926 + 43.125 & 2627.000 & 127.094 + 86.250 & 5430.000 & 256.931 + 172.500 & 10984.218 & 515.784 + 345.000 & 22178.636 & 1033.253 + 690.000 & 44572.273 & 2068.174 + 1293.750 & 83762.886 & 3879.325 + 4312.500 & 279722.953 & 12935.268 + + & @xmath53g/@xmath2 & + 0.086 & 28.457 & 0.666 + 0.259 & 36.339 & 0.954 + 0.431 & 36.579 & 1.480 + 0.690 & 47.351 & 2.009 + 0.863 & 55.779 & 2.342 + 1.348 & 81.633 & 3.291 + 2.695 & 156.889 & 6.117 + 5.391 & 326.268 & 12.694 + 8.215 & 522.548 & 20.335 + 10.781 & 685.171 & 27.621 + 15.682 & 1058.104 & 41.926 + 21.563 & 1515.000 & 60.154 + 34.500 & 2562.000 & 99.553 + 43.125 & 3267.000 & 125.671 + 86.250 & 6744.000 & 255.430 + 172.500 & 13690.000 & 514.238 + 345.000 & 27562.538 & 1031.671 + 690.000 & 55403.077 & 2066.565 + 1293.750 & 104128.019 & 3877.696 + 4312.500 & 347756.729 & 12933.615 + + & @xmath54g/@xmath2 & + 0.086 & 43.727 & 0.739 + 0.259 & 49.036 & 1.148 + 0.431 & 53.691 & 1.583 + 0.690 & 68.576 & 2.092 + 0.863 & 79.054 & 2.419 + 1.348 & 110.256 & 3.352 + 2.695 & 203.690 & 6.226 + 5.391 & 403.036 & 12.590 + 8.215 & 634.218 & 20.070 + 10.781 & 854.595 & 27.285 + 15.682 & 1293.154 & 41.537 + 21.563 & 1799.000 & 59.237 + 34.500 & 3020.820 & 98.444 + 43.125 & 3844.600 & 124.507 + 86.250 & 7984.900 & 254.174 + 172.500 & 16287.600 & 512.930 + 345.000 & 32907.200 & 1030.326 + 690.000 & 66155.400 & 2065.190 + 1293.750 & 124347.000 & 3876.302 + 4312.500 & 415320.000 & 12932.195 + + & @xmath55g/@xmath2 & + 1.348 & 259.313 & 3.792 + 2.695 & 404.291 & 6.492 + 5.391 & 718.712 & 12.638 + 8.215 & 1077.005 & 19.843 + 10.781 & 1423.714 & 26.707 + 15.682 & 2108.910 & 40.544 + 21.563 & 2905.275 & 57.192 + 34.500 & 4831.500 & 95.738 + 43.125 & 6136.150 & 121.574 + 86.250 & 12721.700 & 250.794 + 172.500 & 25961.000 & 509.293 + 345.000 & 52483.001 & 1026.515 + 690.000 & 105556.001 & 2061.257 + 1293.750 & 198451.502 & 3872.289 + 4312.500 & 662975.008 & 12928.083 + + & @xmath56g/@xmath2 & + 1.348 & 384.866 & 4.176 + 2.695 & 560.854 & 6.845 + 5.391 & 944.470 & 12.880 + 8.215 & 1382.305 & 19.957 + 10.781 & 1796.059 & 26.734 + 15.682 & 2632.108 & 40.258 + 21.563 & 3678.614 & 57.308 + 34.500 & 5871.477 & 93.996 + 43.125 & 7456.522 & 120.184 + 86.250 & 15556.243 & 252.157 + 172.500 & 31880.886 & 512.898 + 345.000 & 64517.773 & 1030.922 + 690.000 & 129799.546 & 2065.866 + 1293.750 & 244056.648 & 3876.955 + 4312.500 & 815392.161 & 12932.766 + + & @xmath57=1.2263 g/@xmath2 & + 1.348 & 591.578 & 4.774 + 2.695 & 811.125 & 7.441 + 5.391 & 1282.332 & 13.347 + 8.215 & 1816.492 & 20.228 + 10.781 & 2324.122 & 26.844 + 15.682 & 3353.252 & 40.091 + 21.563 & 4638.863 & 56.811 + 34.500 & 7391.677 & 92.514 + 43.125 & 9309.797 & 118.716 + 86.250 & 19393.895 & 250.119 + 172.500 & 39770.389 & 510.828 + 345.000 & 80525.778 & 1028.729 + 690.000 & 162054.556 & 2063.622 + 1293.750 & 304757.918 & 3874.666 + 4312.500 & 1018329.725 & 12930.412 + + & @xmath57=1.5591 g/@xmath2 & + 2.695 & 1217.205 & 8.329 + 5.391 & 1802.722 & 14.093 + 8.215 & 2471.084 & 20.813 + 10.781 & 3107.909 & 27.280 + 15.682 & 4394.780 & 40.431 + 21.563 & 6008.367 & 56.715 + 34.500 & 9711.442 & 94.365 + 43.125 & 11868.865 & 117.384 + 86.250 & 24627.130 & 248.026 + 172.500 & 50504.260 & 508.563 + 345.000 & 102305.520 & 1026.416 + 690.000 & 205947.039 & 2061.235 + 1293.750 & 387371.949 & 3872.216 + 4312.500 & 1294596.497 & 12927.871 + + & @xmath57=2.0241 g/@xmath2 & + 2.695 & 1914.054 & 9.701 + 5.391 & 2653.269 & 15.315 + 8.215 & 3499.908 & 21.844 + 10.781 & 4303.718 & 28.115 + 15.682 & 5940.816 & 40.924 + 21.563 & 8003.076 & 56.885 + 34.500 & 12739.349 & 93.854 + 43.125 & 15981.851 & 119.108 + 86.250 & 31973.372 & 245.712 + 172.500 & 65497.743 & 505.963 + 345.000 & 132718.486 & 1023.755 + 690.000 & 267252.973 & 2058.470 + 1293.750 & 502770.574 & 3869.360 + 4312.500 & 1680568.581 & 12924.885 + + & @xmath57=3.6956 g/@xmath2 & + 2.695 & 5561.156 & 15.030 + 5.391 & 6851.582 & 20.465 + 8.215 & 8282.303 & 26.503 + 10.781 & 9683.735 & 32.433 + 15.682 & 12471.118 & 44.273 + 21.563 & 16109.020 & 59.548 + 34.500 & 24440.344 & 94.995 + 43.125 & 30151.959 & 119.149 + 86.250 & 58805.086 & 239.239 + 172.500 & 119458.718 & 500.123 + 345.000 & 241975.435 & 1017.331 + 690.000 & 487450.871 & 2051.576 + 1293.750 & 917350.383 & 3862.109 + 4312.500 & 3067567.944 & 12917.150 + + & @xmath57=5.0 g/@xmath2 & + 2.695 & 9549.150 & 19.264 + 5.391 & 11289.594 & 24.705 + 8.215 & 13149.412 & 30.532 + 10.781 & 14947.198 & 36.149 + 15.682 & 18635.588 & 47.707 + 21.563 & 23350.696 & 62.524 + 34.500 & 33750.403 & 94.172 + 43.125 & 41345.121 & 119.872 + 86.250 & 80760.602 & 244.172 + 172.500 & 162974.601 & 496.972 + 345.000 & 329461.530 & 1026.672 + 690.000 & 662036.697 & 2067.672 + 1293.750 & 1243990.000 & 3869.013 + 4312.500 & 4152780.000 & 12918.432 + + & @xmath57=7.8545 g/@xmath2 & + 5.391 & 23919.083 & 32.364 + 8.215 & 26366.144 & 35.213 + 10.781 & 28417.191 & 40.201 + 15.682 & 34567.155 & 50.461 + 21.563 & 40887.110 & 63.836 + 34.500 & 57821.094 & 96.066 + 43.125 & 69405.461 & 118.905 + 86.250 & 129424.016 & 240.005 + 172.500 & 257431.789 & 492.021 + 345.000 & 517526.963 & 968.205 + 690.000 & 1040160.000 & 2052.492 + 1293.750 & 1953890.000 & 3857.032 + 4312.500 & 6522980.000 & 12881.855 + + & @xmath58 g/@xmath2 & + 5.391 & 52334.071 & 45.782 + 8.215 & 56804.007 & 50.849 + 10.781 & 60424.429 & 54.502 + 15.682 & 68219.429 & 62.075 + 21.563 & 77112.745 & 74.889 + 34.500 & 103991.965 & 105.034 + 43.125 & 121078.467 & 127.234 + 86.250 & 213335.193 & 245.844 + 172.500 & 412180.825 & 487.534 + 345.000 & 823130.585 & 1015.634 + 690.000 & 1652820.000 & 2028.342 + 1293.750 & 3103200.000 & 3841.490 + 4312.500 & 10356600.000 & 12898.888 + + & @xmath59 g/@xmath2 & + 5.391 & 132965.313 & 69.516 + 8.215 & 140631.941 & 75.069 + 10.781 & 147606.997 & 79.707 + 15.682 & 161508.892 & 91.188 + 21.563 & 176692.441 & 98.682 + 34.500 & 214260.781 & 124.801 + 43.125 & 244807.793 & 144.762 + 86.250 & 397368.524 & 255.511 + 172.500 & 733042.082 & 498.257 + 345.000 & 1436410.000 & 1015.776 + 690.000 & 2860880.000 & 2037.937 + 1293.750 & 5365630.000 & 3849.903 + 4312.500 & 17899300.000 & 12890.991 + + & @xmath60 g/@xmath2 & + 5.391 & 420619.812 & 120.275 + 8.215 & 436741.892 & 121.507 + 10.781 & 445038.475 & 125.749 + 15.682 & 477946.889 & 134.127 + 21.563 & 500374.227 & 144.655 + 34.500 & 574053.275 & 169.598 + 43.125 & 621244.176 & 187.496 + 86.250 & 900873.051 & 288.390 + 172.500 & 1516040.000 & 519.066 + 345.000 & 2861940.000 & 996.990 + 690.000 & 5626920.000 & 2054.507 + 1293.750 & 10503500.000 & 3853.605 + 4312.500 & 34970300.000 & 12898.971 + + & @xmath61 g/@xmath2 & + 5.391 & 1380830.000 & 197.188 + 8.215 & 1405720.000 & 199.217 + 10.781 & 1437320.000 & 206.732 + 15.682 & 1491360.000 & 217.589 + 21.563 & 1546100.000 & 223.268 + 34.500 & 1657160.000 & 252.920 + 43.125 & 1765120.000 & 269.112 + 86.250 & 2250500.000 & 359.484 + 172.500 & 3390870.000 & 572.109 + 345.000 & 5962050.000 & 1047.419 + 690.000 & 11406900.000 & 2049.414 + 1293.750 & 21127400.000 & 3824.306 + 4312.500 & 69971100.000 & 12895.733 + + & @xmath62 g/@xmath2 & + 5.391 & 4470890.000 & 348.905 + 8.215 & 4531710.000 & 353.258 + 10.781 & 4567730.000 & 357.249 + 15.682 & 4706480.000 & 365.018 + 21.563 & 4754720.000 & 374.545 + 34.500 & 5038440.000 & 396.244 + 43.125 & 5235950.000 & 411.271 + 86.250 & 6139410.000 & 492.918 + 172.500 & 8286640.000 & 684.142 + 345.000 & 13024000.000 & 1128.948 + 690.000 & 23505400.000 & 2101.594 + 1293.750 & 42790700.000 & 3685.375 + 4312.500 & 140264000.000 & 12902.503 + + & @xmath63 g/@xmath2 & + 5.391 & 14545500.000 & 588.613 + 8.215 & 14665000.000 & 591.546 + 10.781 & 14723000.000 & 599.508 + 15.682 & 14995000.000 & 602.836 + 21.563 & 15179700.000 & 616.630 + 34.500 & 15789200.000 & 636.221 + 43.125 & 16093600.000 & 650.506 + 86.250 & 17859500.000 & 726.080 + 172.500 & 21530900.000 & 897.457 + 345.000 & 30504600.000 & 1301.652 + 690.000 & 50560500.000 & 2224.865 + 1293.750 & 87731800.000 & 3952.434 + 4312.500 & 281715000.000 & 12945.568 + + & @xmath64 g/@xmath2 & + 5.391 & 47188700.000 & 985.291 + 8.215 & 47374300.000 & 989.586 + 10.781 & 47542300.000 & 993.478 + 15.682 & 47866200.000 & 1000.964 + 21.563 & 48258700.000 & 1010.034 + 34.500 & 49138100.000 & 1030.355 + 43.125 & 49735000.000 & 1044.162 + 86.250 & 52836300.000 & 1115.850 + 172.500 & 59610600.000 & 1272.441 + 345.000 & 75258300.000 & 1634.074 + 690.000 & 112078000.000 & 2485.051 + 1293.750 & 183785000.000 & 4142.350 + 4312.500 & 566346000.000 & 12930.233 + + & @xmath65 g/@xmath2 & + 5.391 & 152939000.000 & 1629.645 + 8.215 & 153307000.000 & 1633.926 + 10.781 & 153649000.000 & 1637.809 + 15.682 & 152494000.000 & 1645.251 + 21.563 & 155067000.000 & 1654.214 + 34.500 & 156786000.000 & 1674.139 + 43.125 & 157956000.000 & 1687.602 + 86.250 & 163941000.000 & 1756.777 + 172.500 & 176641000.000 & 1903.589 + 345.000 & 204903000.000 & 2230.145 + 690.000 & 271115000.000 & 2995.092 + 1293.750 & 405010000.000 & 4541.812 + 4312.500 & 1154030000.000 & 12958.110 + high precision eos data are essential for understanding target implosion process in icf .
we first examine the present dft - eos theoretically through the rankine - hugoniot ( rh ) equations , which follow from conservation of mass , momentum , and energy across the front of the shock wave .
the locus of points in ( @xmath66 , @xmath67 , @xmath68)-space described by rh equations satisfy @xmath69 @xmath70 @xmath71 where subscripts 0 and 1 represent the initial and shocked state , and @xmath66 , @xmath67 , and @xmath68 denote internal energy , pressure , and volume , respectively . @xmath72 is the particle velocity of the material behind the shock front and @xmath73 is the shock velocity . along the hugoniot curve of the liquid hydrogen ,
the starting point with a density of 0.0855 g/@xmath2 and a temperature of 23 k has been selected , where the relative internal energy has been set to zero and the pressure is considered as small as negligible .
smooth functions have been adopted to fit dft - eos in the relative density and temperature regime .
our dft - based hugoniot curve with pressure up to @xmath74 gpa and temperature up to @xmath3 k is presented in fig .
[ fig_hugoniot1 ] .
our simulation results indicate that the maximum shock compression ratio is 4.5 with a pressure around 40 gpa , at which the system is governed by gradual dissociation of molecules . with the increase of pressure , the compression ratio decreases and
then reaches a value of 4.23 below 950 gpa .
this hardening behavior of the hugoniot can be attributed to the formation of mono - atomic fluid . however , as the pressure exceeds @xmath5 gpa ( temperature above 19 ev ) , the compression ratio merges into 4.0 , which indicate a full ionization of the liquid hydrogen .
on the other side , sesame hugoniot @xcite plotted in fig .
[ fig_hugoniot1 ] shows a maximum compression ratio around 4.5 , but the corresponding pressure is much too high with respect to our dft - hugoniot . for comparison ,
previous results from pimc simulations @xcite are also plotted in fig .
[ fig_hugoniot1 ] , which shows consistency with our dft results at pressures beyond 50 gpa .
however , pimc simulations have failed to reproduce the experimental results below 50 gpa . over the past ten years
, the hugoniot of hydrogen or deuterium has been experimentally explored up to @xmath30200 gpa .
the latest set of data points were obtained by two - stage light gas gun @xcite , explosive - driven compression @xcite , z - pinch - driven compression @xcite , where the compression ratio @xmath75 shows a maximum close to 4.3 , or by laser - driven compression with the nova laser and the omega laser ( the eos would possibly be corrected by introducing the quartz standard ) , which suggest a stiff behavior ( @xmath76 ) below 100 gpa and become softer ( @xmath77 ) at higher pressures @xcite . to clearly show the comparison between the present dft - eos and those previous results , fig .
( [ fig_hugoniot2 ] ) plots the hugoniot curve below 250 gpa .
the present hugoniot curve from dft - eos with accounting for the ionic quantum zero - point energy ( zpe ) shows better accordance with experimental data . at pressures below 100 gpa , both of the curves , as discussed above , exhibit a maximum compression ratio of 4.5 , which is accordant with previous qmd results and experimental data obtained by gas gun , converging explosives and magnetically driven flyer .
however , as pressures go beyond 100 gpa , @xmath78 indicates the agreement with high power laser experiments with the quartz standard .
predictions from various chemical models are also shown for comparison , for instance , kerley 2003 @xcite , linear mixing model @xcite , and the fluid variational theory ( fvt ) @xcite .
those chemical methods generally predict larger maximum compression ratio into higher pressures , except for the kerley 2003 eos , which is in better agreement with experiments .
( d@xmath7 ) samples at 297 k have been studied : 0.16 gpa , 0.7 gpa , and 1.48 gpa.,height=226 ] recently , laser - driven shock compressions on h@xmath7 or d@xmath7 precompressed in diamond anvil cells from 0.16 to 1.6 gpa have been proved to provide visible ways to generate shock hugoniot data over a significantly broader thermodynamical regime than previous experiments @xcite .
these experimental data are highly valuable for examining various theoretical models . in the present work ,
we have shown the hugoniot data for initial pre - compressions of 0.16 gpa , 0.7 gpa , and 1.48 gpa in fig .
[ fig_hugoniot3 ] . as a consequence
, the data covers a density range @xmath30 2 times greater than previous investigations limited to the principal hugoniot alone . as shown in fig .
[ fig_hugoniot3 ] , good agreement has been gained between the present dft - eos pre - compressed hugoniot data and laser - driven experimental results .
the maximum compression ratio along a given hugoniot has been observed to strongly depend on the initial density .
that is , with increasing initial density , the compression ratio decreases . for molecular fluid in the warm dense regime , which consists of atoms , molecules , nuclei , and electrons ,
the free energy can be expressed as @xmath79 where @xmath80 and @xmath81 are the ideal free energies for ions and electrons , respectively , @xmath82 is the excess free energy , while @xmath83 denotes the contribution from molecular dissociation with the following form @xcite : @xmath84 here the dissociation ratio @xmath85 is used as an adjustable function of density and temperature , with an assumed fermi - function form @xmath86 + 1},\ ] ] where @xmath87 and @xmath88 .
using the present dft - eos data , we have determined the value @xmath89 , @xmath90 , @xmath91 , and @xmath92 . .
for comparison , previous theoretical results @xcite have also been plotted.,height=226 ] at temperatures below 10000 k , molecular dissociation governs the first - order phase transition , which is important in determining the nonmetal - to - metal transition . in this work ,
we have introduced a fermi formula to fit our dft - eos in warm dense region by using eq .
( [ equation_alpha ] ) , and the dissociation fraction has been plotted in fig .
[ fig_dissociation ] .
@xcite have introduced a criteria to estimate the fraction of molecular hydrogen by counting the number of atoms located within a radius , which last for a time greater than ten vibrational periods .
@xcite have used a coordination number , @xmath93 to determine @xmath85 , where @xmath94 is the pair correlation function to present the possibility of finding a particle from a reference atom .
results from those different assumptions are compared with the present work in fig .
[ fig_dissociation ] .
they yield similar tendency with temperature at the sampled density .
the dissociation fraction from vorberger _
strongly depends on the definition of the molecule in this region , and shows abrupt increase as temperature increases .
however , qmd method gives a smoother behavior of @xmath85 as indicated in fig .
[ fig_dissociation ] . , respectively .
the blue open diamonds denote pimc data .
each curve corresponds to an isochore .
each curve has been shifted by 1.0 from the previous one for clarity.,height=226 ] denotes data from holst _ et al .
_ @xcite ; ( b ) pressure difference as a function of temperature , @xmath95 are obtained from hu _ et al .
each curve has been shifted by 1.0 from the previous one for clarity.,height=226 ] in this section , the present dft - eos data have been systemically compared with previous theoretical predictions . at densities from @xmath96 g/@xmath2 to @xmath97 g/@xmath2
, we have shown the pressure and internal energy difference between our dft - eos and those obtained by pimc simulations ( see fig .
[ fig_expanded ] ) .
the results indicate a maximum of 7% difference for the pressure and 15% for the energy over the temperatures we explore . at temperatures above @xmath30 30 ev
, the distinction can be viewed as small as negligible between the two methods . in the warm dense regime , which is highlighted at densities between 0.2 and 3.0 g/@xmath2 , very good agreement has been found between our dft - eos data and those fitted qmd results at the temperature domain 2000 @xmath30 10000 k ( left panel in fig .
[ fig_e1 ] ) . at temperature beyond 10000 k , results from pimc simulations by militzer _
@xcite , chabrier model @xcite , and qmd simulations @xcite have been plotted in the right panel in fig .
[ fig_e1 ] .
the pimc method is suitable for investigating many - body quantum systems at high temperatures . in this method ,
electrons and ions are treated on equal footing as paths .
the model of chabrier and potekhin considers a fully ionized plasma , which is reliable at high temperature and low density region . as
shown in fig .
[ fig_e1 ] , the present results are in accordance with those numerical simulations and theoretical models .
the isotherms of the pressure have been observed to show a systematic behavior in terms of the density and temperature ( see fig .
[ fig_mid_p ] ) . in this region
, we do not find any signs for @xmath98 , which would indicate another first - order phase transition ( the so - called ppt ) .
ppt is usually considered in chemical models such as fluid variational theory @xcite or liquid state perturbation theory @xcite . in these chemical models ,
minimization of the free energy for a mixture consists of atoms , molecules , and plasma in equilibrium .
relations between different particles are described by effective potentials . as we explore to a higher temperature region ( right panels in fig .
[ fig_mid_p ] and fig .
[ fig_eos_dense ] ) , pimc data by hu _
@xcite are shown for comparison .
it is clearly indicated that the pressure given by dft - eos is in good agreement with pimc calculations up to @xmath99 k for the densities concerned , and this agreement extends toward lower temperatures when the density decreases .
in summary , we have constructed a wide - range dft - eos by means of fpmd and ofmd simulations . after building short cutoff radius columbic potential
, we have the ability to explore the eos into ultra - dense region .
the present dft - eos is valid at densities from @xmath0 to @xmath1 g/@xmath2 with the temperature up to @xmath3 k. available experimental data and theoretical models have been introduced to compare with current dft - eos .
we have found good agreement between our results and those data probed by gas gun , chemical explosive , and magnetic driven plate flyer experiments , which indicate a maximum compression ratio of 4.5 around 40 gpa . at higher pressures ,
our data show stiff behavior and validates the high power laser experiments with quartz standard .
the principal hugoniot curve is also accordant with previous qmd simulation results .
agreement has also been found between our dft - eos hugoniot data and those obtained by pre - compressed laser - driven shock wave experiments , which provide visible ways to generate eos in a broader density and temperature regime . as density and temperature
enter into a denser and hotter regime , where experimental detections are prohibited , the present results are compared with those predicted by chemical model and pimc simulations .
the present dft - eos covers typical states as can be reached in icf and will be applied in hydrodynamic simulations in the future work .
this work was supported by nsfc under grants no . 11275032 , no .
11005012 and no . 51071032 , by the national basic security research program of china , and by the national high - tech icf committee of china .
s. atzeni and j. meyer - ter - vehn , _ the physics of inertial fusion : beam plasma interaction , hydrodynamics , hot dense matter _ , international series of monographs on physics ( clarendon press , oxford , 2004 ) .
the time steps have been taken as @xmath100 , where @xmath101 is the ionic sphere radius ( @xmath102 is the ionic number density ) , @xmath103 presents the kinetic energy , and @xmath104 is the ionic hydrogen . | wide range equation of state ( eos ) for liquid hydrogen is ultimately built by combining two kinds of density functional theory ( dft ) molecular dynamics simulations , namely , first - principles molecular dynamics simulations and orbital - free molecular dynamics simulations .
specially , the present introduction of short cutoff radius pseudopotentials enables the hydrogen eos to be available in the range @xmath0 to @xmath1 g/@xmath2 and up to @xmath3 k. by comprehensively comparing with various attainable experimental and theoretical data , we derive the conclusion that our dft - eos can be readily and reliably conducted to hydrodynamic simulations of the inertial confinement fusion . |
the existence of some sort of hierarchical order is an apparently widespread feature of many complex systems , including gene @xcite and human brain @xcite networks , ecosystems @xcite , social and urban structures @xcite , the internet @xcite or open - source communities @xcite .
the presence of such underlying order in the multiscale organization of complex systems is a long standing hypothesis @xcite giving rise to the idea of hierarchy as a central concept -see also @xcite . although usually treated only in qualitative terms , some formal approaches to the problem have been proposed .
the efforts towards a well - defined quantification of hierarchical order have been improving by means of complex networks theory . as a key part of their organization
, dedicated efforts have been made towards a proper identification of hierarchical trends .
one outcome of these efforts has been a number of powerful , heuristic measures @xcite .
often , a nested organization -formally identical to order , in set - theoretical terms @xcite- can be identified as the basis of hierarchical order .
if we think in hierarchy in these terms , we might agree with herbert simon that `` it is a commonplace that nature loves hierarchies '' -cited in @xcite .
many examples belong to this picture . within the context of matter organization ,
molecules are made of atoms , which result from the combination of elementary particles , some of which having also internal subunits ( as quarks ) .
similarly , the relation of characteristic scales of organization by inclusion of one in another , like chinese boxes or matryoshka dolls , has been seen as a sort of hierarchical organization .
another relevant example is the case of fractal structures which naturally define a hierarchy of self - similar objects .
finally , within the context of taxonomy , spin glasses or optimization theory , the use of _ ultrametricity _ has also allowed to define hierarchical order @xcite .
biological hierarchies have also evolved through time as part of a process that generates high - level entities out of nesting lower - level ones @xcite
. however , taxonomic classification trees are probably the most obvious representation of a hierarchy based on inclusion . in biology ,
living beings are individuals grouped in _ taxa
_ according their characteristics . starting from species ,
a nested hierarchy is defined where species belong to genera , which are included within families forming orders and so forth . here
, every organism is , in principle , unambiguously classified and therefore no uncertainty can be associated to the process of classification . alternatively to the view of nestedness
, a hierarchical organization can be defined in a structure of causal relations .
paradigmatic examples are the flowchart of a company or the chain of command in the army , where the authority concept defines a particular case of causal relation .
causality induces an asymmetrical link between two elements , and this asymmetry , either by inclusion or through any kind of causal relation , defines an order .
we observe that any circular or symmetrical relation between two elements violates the concept of order , thereby intuitively loosing its hierarchical nature . from this perspective ,
any feedforward relation is potentially hierarchical .
such feedforward structures pervade a diverse range of phenomena and structures , from phylogenetic trees to river basins .
a common feature of most of the approaches mentioned above is that the notion of _ hierarchy _ is basically identified with the concept of _ order_. order is a well defined concept in mathematics @xcite but , is order enough to grasp the intuitive idea of hierarchy ?
can we actually define what is _ hierarchy _ ? quoting herbert simon , @xcite .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ( ... ) a hierarchical system -or hierarchy- can be defined in principle as a system that is composed of interrelated subsystems , each of the latter being also hierarchic in structure until the lowest scale is reached . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ this definition does not provide a clear formalization of hierarchy as a measurable feature , although it certainly grasps the intuitive idea of hierarchy . how can such a general measure be defined ?
it is reasonable to assume that we have a hierarchy if there is no ambiguity ( or uncertainty ) in the chain of command followed for any individual to the chief in the flowchart .
we shall call this feature the _ definiteness condition_. this is also valid for nested structures .
one might think that this condition is justified by a single chain of command or in the case of matryoshka dolls .
however , such structures are already defined within order theory as _ totally ordered _
( see fig .
[ fig1]a ) .
order and hierarchy are closely related but they are not essentially the same . in this paper
we reserve the word _
hierarchy _ to designate a concept that goes beyond the definition of order .
we argue that the difference stems from the fact that a hierarchical structure must also satisfy a _
pyramidal organization _
constraint -see fig .
( [ fig1]b ) . in other words ,
the lower the layer of organization , the larger the number of entities it contains .
but , what happens when this pyramid structure is inverted ? intuitively , they would not be hierarchical but anti - hierarchical . in this work
we show how information theory naturally provides the suitable framework to characterize hierarchy in causal structures . within this theoretical apparatus
we provide the rigorous definition of the hierarchy index for causal structures and how it is applied in some illustrative examples establishing a distinction among hierarchical , non - hierarchical and anti - hierarchical .
in this section we will present the basic theoretical background used in this paper , grounded both in order and graph theories . at the end of this section
we will formally define the _ causal graph _
, the key concept of this theoretical framework where the proposed hierarchy measures are applied . hierarchy is undoubtedly tied to order
this is why we make a brief review of order theory highlighting some features that have been commonly attributed to hierarchy .
the first task is to define an ordered pair between two elements @xmath0 of a given set @xmath1 , to be written as @xmath2 , @xmath3 or , in a formally equivalent way : @xmath4 this latter formalization explicitly defines order from an inclusion relation @xcite .
this immediately connects to standard views of hierarchical systems , as we already mentioned , in which inclusion relations are considered essential . having defined an ordered pair
, we define an order relation .
let @xmath5 be a countable , finite set and @xmath6 a relation .
such a relation is an _ order relation _
-rigorously speaking , a _ strict partial order_- if the following condition holds : @xmath7 we finally define two subsets of @xmath1 from the definition of order relation which will be useful to characterize the kind of structures studied in this paper .
the set of maximal elements of @xmath1 , to be written as @xmath8 , is defined as : @xmath9 similarly , the set of minimal elements , to be written as @xmath10 is defined as : @xmath11 let @xmath12 be a directed graph , being @xmath13 , the set of nodes , and @xmath14 the set of arcs -where the order , @xmath15 implies that there is an arc in the following direction : @xmath16 . given a node @xmath17
, the number of outgoing links , to be written as @xmath18 , is called the _ out - degree _ of @xmath19 and the number of ingoing links of @xmath19 is called the _ in - degree _ of @xmath19 , written as @xmath20 .
the _ adjacency matrix _ of a given graph @xmath21 , @xmath22 is defined as @xmath23 ; and @xmath24 otherwise . through the adjacency matrix , @xmath25 and @xmath26
are computed as @xmath27 furthermore , we will use the known relation between the @xmath28-th power of the adjacency matrix and the number of paths of length @xmath28 going from a given node @xmath19 to a given node @xmath29 specifically , @xmath30 is the number of paths of length @xmath28 going from node @xmath19 to node @xmath29 @xcite .
it is said that @xmath19 _ dominates _
@xmath31 if @xmath32 . a _ feed - forward _ or _ directed acyclic graph _ ( dag ) is a directed graph characterized by the absence of cycles : if there is a _ directed path _ from @xmath19 to @xmath31 ( i.e. , there is a finite sequence @xmath33 ) then , there is no directed path from @xmath31 to @xmath19 .
conversely , the matrix @xmath34 depicts a dag with the same underlying structure but having all the arrows ( and thus , the causal flow ) inverted .
the _ underlying graph _ of a given dag @xmath21 , to be written as @xmath35 , is the undirected graph @xmath36 obtained by substituting all arcs of @xmath37 , @xmath38 by edges giving the set @xmath39 .
a dag @xmath21 is said to be _ connected _ if for any pair of nodes combination @xmath40 there is a finite sequence of pairs having the following structure @xmath41 being @xmath42 .
given the acyclic nature of a dag , one can find a finite value @xmath43 as follows : @xmath44 it is easy to see that @xmath45 is the length of the longest path of the graph . borrowing concepts from order theory @xcite , we define the following set : @xmath46 to be named the set of _ maximal nodes _ of @xmath21 , by which @xmath47
. additionally , one can define the set of nodes @xmath48 as @xmath49 to be referred as the set of _ minimal nodes _ of @xmath21 .
the set of all paths @xmath50 , @xmath51 , from @xmath52 to a given node @xmath53 is indicated as @xmath54 .
given a node @xmath53 , the set of all paths from @xmath52 to @xmath19 is written as @xmath55 furthermore , we will define the set @xmath56 as the set of all nodes participating in this path , except the maximal one .
conversely , the set @xmath57 is the set of all nodes participating on this path , except the minimal one . attending to the node relations depicted by the arrows , and due to the acyclic property , at least one node ordering can be defined , establishing a natural link between order theory and dags
this order is achieved by labeling all the nodes with sequential natural numbers and obtaining a configuration such that : @xmath58 the existence of this labeling connects order relations with directed acyclic graphs .
finally , throughout this paper we reserve the word _
tree _ to refer to those graphs where all nodes excluding the maximal one have @xmath59 and all nodes except the minimal ones display @xmath60 .
therefore , we distinguish between chains ( all nodes with @xmath61 excluding the minimal one ) and trees . in a causal graph
we only consider immediate relations between elements i.e. two elements are causally related if there exists just one cause - effect event relating them .
we explicitly neglect those relations between nodes which can only be derived by transitivity .
a causal relation can be illustrated by genetic inheritance in a genealogy .
offspring s characters come from its parents and indirectly from its grandparents .
therefore , no direct causal relation can be defined between grandparents and grandsons .
however , it is true that grandparents indirectly determine the characters of grandsons , due to the transitive nature of the genetic relations . in this work
we will restrict the use of the term _ causal relation _ to refer to direct relations such as direct parent - sons relations , as described in the above example . a causal graph
@xmath62 is a directed graph where @xmath63 are the elements of a set ( the members of a family , in the above described example ) and @xmath37 are the _ causal relations _ that can be defined between the members of @xmath63 . in this work ,
we restrict the term _ causal graphs _ to graphs being _ acyclic _
( i.e. , dags ) and _ connected_. the former property avoids conflicts in the definition of the causal flow .
the latter property assumes that two non - connected causal structures have no relation among them , and therefore must be considered as two independent systems .
hereafter , we will refer to the set of paths @xmath54 as the set of _ causal paths_.
in this section we propose the basis for a rigorous evaluation of hierarchy .
we begin by defining the features of what we consider as the _ perfect hierarchical structure_. as will be shown below , our proposed definition of hierarchy matches with an ordered structure with special features , thereby making an explicit difference between order and hierarchy .
therefore , we reserve the term hierarchy to refer to a special class of order . within the framework of graph theory , the required conditions naturally match those displayed by a tree - like feedforward graph .
then , as we shall see , the estimator we propose identifies the feed - forward tree topology as perfectly hierarchical .
the main point of the section is devoted to the definition of a quantitative estimator of hierarchy based on two entropy measures that captures the intuitive ideas described in the introductory section : the definiteness and pyramidal organization condition .
we stress that the forthcoming formalism applies only to the class of _ causal graphs_. we are going to refer to a system as _ perfectly hierarchical _ if it satisfies the following conditions .
let us consider a system depicted by a _ causal graph _ @xmath64 .
we say that this graph @xmath65 will be perfectly hierarchical if the following two conditions hold : 1 . _
definiteness condition.- _ for every element @xmath66 there is only one element @xmath17 , @xmath67 such that @xmath68 .
a straightforward consequence of this condition is that @xmath69 2 .
_ pyramidal condition.- _ there is a partition @xmath70 of the set @xmath63 , i.e. , @xmath71 by which : @xmath72 a direct consequence of the above is that @xmath73 which reflects the pyramidal structure of the graph .
a measure of hierarchy must properly weight the deviations of the studied graph from the above requirements .
one could add another condition , by imposing that every node in a given layer dominates the same number of nodes contained in the next downstream layer . in formal terms , this implies that , in addition to the above conditions , we can add a third one , namely : \3 .
_ symmetry condition_. it is established by means of @xmath74 which actually corresponds to a so - called complete @xmath28-ary tree ( gross and yellen , 1999 ) . therefore , in those cases where symmetry is considered as an inherent feature of ideal hierarchy , deviations from symmetrical configurations must also be taken into account in our quantitative approximation .
maximals exhibiting and inverted pyramidal condition ( note that inverting the arrows we would have a tree - like graph ) e ) an ordered but no hierarchical linear chain violating the pyramidal condition and f ) an inverted tree - like structure where no definiteness and pyramicity conditions are satisfied onwards but completely satisfied backwards.,width=264 ] let us summarize the above statements @xmath75 , @xmath76 and @xmath77 and their consequences . the so - called _ definiteness condition
_ implies that _ there is no uncertainty _ in identifying the premise of a given causal relation , i.e. , the node that immediately governs a given node .
taking into account the definition of causal graph , ( essentially , a connected dag ) this first statement restrict to tree structures the number of candidates for perfectly hierarchical structures .
note that these trees -including linear chains- have a single root node , i.e. , @xmath78 .
the _ pyramidal condition _
rules out from the set of perfectly hierarchical structures those dags having linear chains in their structure . in other words , nodes displaying @xmath79 are not allowed .
we observe that , according to the pyramidal condition , @xmath80 therefore , it is straightforward to conclude that the most simple representation of an ideal hierarchical structure is a binary tree -see fig ( [ fig2]a)- , in which the above inequality becomes equality for all successive layers .
this is consistent with standard graph theory and the definition of perfect binary tree .
finally , the _ symmetry condition _ , optional for our definition of hierarchy , rules out those trees which are not perfectly symmetrical . whereas trees shown in figs .
( [ fig2]a ) and ( [ fig2]b ) can be considered hierarchical by virtue of conditions @xmath75 and @xmath76 , condition @xmath77 makes a distinction between them , being only perfectly hierarchical the first one . as a final remark ,
let us note that one can build non - hierarchical and anti - hierarchical structures , by simply violating some of the conditions we stated above , namely the pyramidal condition alone -fig .
( [ fig2]e)- or both the definiteness and the pyramidal condition , -see fig .
( [ fig2]d and [ fig2]f ) .
it is easy to see that a quantitative estimator of hierarchy should account for limit cases and place as intermediate point structures such as the one depicted in fig .
( [ fig2]c ) .
the above features describe the _ perfect _ hierarchical structure . in this section
we go further and we provide the basis for a definition of a _ hierarchical index _ of a causal graph grounded in the framework of information theory .
this index provides a quantitative estimation of how far is a given causal graph from the conditions of a perfect hierarchy . in the following subsections
we will define two entropies for a causal graph , attending to the top - down and bottom - up observations of the causal graph according to the onward and backward flows in the graph .
the aim of this mathematical formalism is to quantify the impact of the number of pathways in the causal graph .
specifically , we will consider the balance between the _ richness _ of causal paths ( a top - down view ) versus the _ uncertainty _ when going back reversing the causal flow ( i.e. , a bottom - up perspective ) .
then , attending to the direction of the flow we interpret the top - down view as a richness whilst the bottom - up as an uncertainty in terms of topological reversibility as recently introduced in @xcite . arguably , the larger is the number of decisions going down , the higher is the richness of causal paths .
similarly , the larger the number of alternative pathways to climb up , the larger will be the uncertainty in recovering the causal flow . in the following subsections , we will explore , within the framework of information theory , the relationship between diversity and uncertainty and the their impact in the fulfillment of the hierarchy conditions .
we begin this section with a brief revision of the core concepts of information theory , to be used as our theoretical framework . according to classical information theory @xcite ,
let us consider a system @xmath81 with @xmath82 possible states , whose occurrences are governed by a random variable @xmath83 with an associated probability mass function formed by @xmath84 . according to the standard formalization , the _ uncertainty _ or _ entropy _ associated to @xmath83 , to be written as @xmath85 , is :
@xmath86 which is actually an average of @xmath87 among all events of @xmath81 , namely , @xmath88 , where @xmath89 is the _ expectation _ or average of the random quantity between parentheses .
analogously , we can define the _
conditional entropy_. given another system @xmath90 containing @xmath91 values or choices , whose behavior is governed by a random variable @xmath92 , let @xmath93 be the conditional probability of obtaining @xmath94 if we already know @xmath95 .
then , the conditional entropy of @xmath92 from @xmath83 , to be written as @xmath96 , is defined as : @xmath97 the first task is to study the degree of reversibility of causal paths , thereby considering the role of the _ definiteness condition_. this will be evaluated by computing the uncertainty in reversing the process starting from a given node in @xmath48 . the formalism used in this section is close to the one developed in @xcite .
we first proceed to define the probability distribution from which the entropy will be evaluated .
accordingly , the probability to chose a path @xmath98 from node @xmath99 by making a random decision at every crossing when _ reverting _ the causal flow is : @xmath100 the conditional entropy obtained when reverting the flow from @xmath53 will be : @xmath101 the overall uncertainty of @xmath21 , written as @xmath102 , is computed by averaging @xmath103 over all minimal nodes , i.e : @xmath104\nonumber\\ & = & \sum_{v_i\in\mu}q(v_i ) \sum_{v_j\in v(\pi(v_i)}\log ( k_{in}(v_j))\left[\sum_{\pi_k : v_j\in v(\pi_k)}\mathbb{p}(\pi_k|v_i)\right]\nonumber\\ & = & \sum_{v_i\in\mu}q(v_i ) \sum_{v_k\in v\setminus m}\phi_{ik}({\cal g})\log k_{in}(v_k ) . \label{entropia1}\end{aligned}\ ] ] where we assume , unless indicated : @xmath105 instead of a vector , now we construct a @xmath106 matrix , @xmath107 accounting for the combinatorics of paths and how they contribute to the computation of entropy .
@xmath108_{ij}\equiv\phi_{ij}({\cal g})=\sum_{\pi_k : v_j\in v(\pi_k)}\mathbb{p}(\pi_k|v_i).\nonumber\end{aligned}\ ] ] this represents the probability to reach @xmath29 starting from @xmath19 .
now we derive the general expression for @xmath109 . to compute the probability to reach a given node
, we have to take into account the probability to follow a given path containing such a node , defined in ( [ pprod ] ) . to rigorously connect it to the adjacency matrix
, we first define an auxiliary , @xmath110 matrix @xmath111 , namely : @xmath112 where @xmath113 . from this definition ,
we obtain the explicit dependency of @xmath109 from the adjacency matrix , namely , @xmath114^k({\cal g})\right)_{ij}. \label{phi(b)}\ ] ] and accordingly , we have @xmath115 ^ 0({\cal g})\right)_{ii}=1.\ ] ] therefore , we already obtained the explicit form of such a conditional entropy , namely : @xmath116 assuming equiprobability , the above expression leads to : @xmath117 let us now estimate the topological richness of a causal graph , i.e. , the average amount of information needed to describe a given top - down path within the structure .
let us observe that the kind of question we are trying to answer is the same than the one explored above , but considering the top - down approach .
therefore , the mathematical form of this quantity , to be referred as @xmath118 , will be formally analogous to the previous one , but considering that we are going onwards according to the causal flow . thus : @xmath119 where @xmath120 is the cardinality of @xmath52 -the set of maximal nodes .
@xmath121 is analogous to @xmath122 of equation ( [ hgeneral ] ) . in this case , elements @xmath123 of matrix @xmath124 represent the probability to cross node @xmath31 departing from @xmath125 according to the causal flow .
the explicit expression of @xmath124 is defined from matrix @xmath126 : @xmath127 then , @xmath128^k({\cal g})\right)_{ij}. \label{phi(b)}\ ] ] and as above , we have @xmath129 ^ 0({\cal g})\right)_{ii}=1.\ ] ] as we shall see , the above definition of information will bring us the ingredients to define a hierarchy index according to the list detailed in section [ section : conceptual ] . roughly speaking , what we propose in the following lines
is to evaluate the balance between the pyramidal structure of the graph against the degree of reversibility of the paths it generates , i.e , the balance between @xmath130 and @xmath131 .
however , in order to rigorously characterize hierarchy , we need to properly treat the studied graph attending to the different layers of its feedforward structure .
the analysis of the graph structure allows us to identify and quantify deviations from the perfect structure at any level of the graph .
the starting point will involve the characterization of a layered structure within the graph defining a partition @xmath132 of the set of nodes .
involved in the evaluation of hierarchy . @xmath133 and @xmath134 are obtained through successive application of bottom - up leaf removal algorithm , which implies that we remove all nodes having @xmath135 . @xmath136 and @xmath137 are obtained by the successive application of a top - down leaf removal algorithm , thereby removing the nodes having @xmath138 .
we observe that the generation of @xmath139 and @xmath140 implies the breaking of the net -see text.,width=283 ] given the dag @xmath141 , let us define two partitions of @xmath63 , @xmath70 and @xmath142 .
the members of such partitions are the layers of the dag by either performing a top down or bottom up _ leaf removal algorithm _ @xcite .
specifically , the first members of such partitions are defined as : @xmath143 and @xmath144 we observe that @xmath145 and that @xmath146 . with the above subsets of @xmath63 we can define the graphs @xmath147 , and @xmath148 in the following way : @xmath149 and @xmath150 respectively .
similarly , we build @xmath151 as : @xmath152 and @xmath153 respectively .
we therefore defined two sequences of subgraphs of @xmath21 ordered by inclusion , namely : @xmath154 and @xmath155 where it is easy to observe that : @xmath156 in fig .
( [ fig : leafremovals ] ) we describe the generation of these subsets of graphs for a given toy model of dag . in summary , we constructed two collections of subsets by finding the layered structure using a bottom - up leaf removal algorithm ( pruning the elements having @xmath135 successively ) and a top - down leaf removal algorithm ( i.e. , pruning the elements having @xmath138 ) .
notice that , even if @xmath157 one can not assume @xmath158 except in symmetrical cases . in order to generate a normalized estimator @xmath159 ( between @xmath160 and @xmath161 ) accounting for the balance between @xmath162 and @xmath102 we will define it as
: @xmath163 since both the layered structure and its pyramidal composition must be taken into account , the hierarchical index of the graph must be weighted taking into account the successive layers of the system .
this avoids to identify as completely hierarchical those structures not perfectly satisfying the _ pyramidal condition_. therefore , the _ hierarchical index of a feed - forward net _ , to be indicated as @xmath164 will be the average among the @xmath165 subgraphs @xmath166 , the @xmath165 subgraphs @xmath167 and @xmath21 itself -note that we average between @xmath168 objects , i.e. : @xmath169 it is strictly necessary to take into account all these subgraphs in order to identify any violation of the hierarchy conditions at any level of the structure . we can go a step further by imposing symmetry in the pyramidal structure as suggested above , to distinguish among different topologies such as those displayed in figure ( [ fig2]a ) .
let us indicate by @xmath170 the set of paths from @xmath52 to @xmath48 present in the graph @xmath171 .
the so - called _
jensen s inequality _
@xcite provides an upper bound of the information content which , in our case reads : @xmath172 we observe that a ) is only achieved when all @xmath173 from @xmath52 to @xmath48 are equiprobable , being this equiprobability an indicator of symmetry .
the same applies to b ) , but now we consider the bottom - up estimator , when paths are considered from @xmath48 to @xmath52 . in this case , attending to the symmetric condition we can define and estimator analogous to @xmath174 , namely , @xmath175 accordingly , the symmetrized version of the hierarchy index , @xmath176 , would be : @xmath177 to ensure the consistency of @xmath178 and @xmath179 we must overcome the last conceptual problems .
when performing the leaf removal operation one can break the graph in many connected components having no connections among them .
let us indicate as @xmath180 the set of components of our graph @xmath181 , each of them obtained from the set @xmath182 of nodes and their . in this case ,
the natural way to proceed is to average the individual contributions of the different connected components of @xmath183 or @xmath184 according to the number of nodes they have against @xmath185 or @xmath186 , leading to : @xmath187 the same applies for @xmath188 for the computation of @xmath179 . finally , we impose , for both mathematical and conceptual consistency that : @xmath189 furthermore , if @xmath190 , ( i.e. , the case where the graph consists of a single node ) : @xmath191 according to this formulation some scenarios would lead to @xmath192 .
the simplest one is the just mentioned by definition , consisting of a single node .
another one is the linear feed - forward chain having @xmath193 or more linked nodes .
it is clear that in these cases , @xmath194 .
it is worth to stress that this particular situation matches with the causal graph of a total order relation , and therefore , we have the way to differentiate this particular graph from other structures having null hierarchy .
finally , a third class of structures belongs to the family of non - hierarchical graphs .
they give @xmath195 since @xmath196 .
this is the case of erds rnyi dags or dag cliques . in these cases ,
the causal graph is not hierarchical because all the diversity of paths generated when crossing the causal flow downwards is neutralized by the uncertainty in recovering any causal path backwards . and @xmath197 .
for both graphs we add , in an iterative way , arcs at random until we reach the feed forward clique configuration .
note that the link addition to the two initial tree - like structures converge in the same full connected configuration which contains @xmath198 links .
for both experiments squares represent @xmath178 values , and triangles @xmath199 . as expected , @xmath200 , except in limit cases .
the small graphs provide a visual clue for type of structures we are obtaining through the arc enrichment process . for every point in the chart , mean and standard deviation were calculated from @xmath201 replicas .
entropies were computed considering @xmath202.,width=321 ] in this section we evaluated the hierarchy of several toy models in order to intuitively grasp the scope of the measure . in fig .
( [ figgrafets ] ) we evaluated the hierarchy index ( both the raw one and the symmetrical one ) for several structures leading to hierarchical , anti - hierarchical and non - hierarchical structures .
the figure illustrates the impact of number of maximals and minimals and the multiplicity of pathways in relation to the existence of a pyramidal and predictable structure .
we observe that deviations from tree and inverted tree configurations lead to a non binary interpretation of hierarchy .
furthermore , we measured ( fig .
( [ figevolution ] ) the impact in terms of hierarchy of arc addition preserving the acyclic character .
staring from two extreme tree graphs ( the feedforward and the inverted ones , respectively ) we add arcs at random until we reach a fully connected feed - forward structure in both situations .
we consider the starting point of our numerical experiment a binary tree , @xmath203 containing @xmath204 nodes .
we construct an inverted binary tree @xmath205 by the transposition of the adjacency matrix of @xmath203 . in both graphs
we say that , consistently with the ordering property of dags , given an arc @xmath206 then @xmath207 . in an iterative process we construct two new dags @xmath208 where @xmath209 labels the number of additions of new arcs to the underlying @xmath203 and @xmath205 .
the process ends when graphs achieve the directed acyclic clique condition , i.e. , the linearly ordered graph @xmath210 containing @xmath211 nodes : @xmath212 for statistical significance , we performed @xmath201 replicas of the numerical experiment .
then @xmath213 and @xmath214 and their respective standard deviations were calculated for each set of iterations .
( [ figevolution ] ) shows that starting from an initial value of @xmath215 for @xmath203 and @xmath216 for @xmath205 , the addition of feed - forward arcs causes a decrease in absolute value of the hierarchical indexes until @xmath217 corresponding to a total linear ordered structure where every possible feed - forward path is included in the graph . as expected , @xmath200 except in the extreme full - connected cases .
we finally test the case of a directed acyclic erds rnyi ( er ) graph @xmath218 .
this is an interesting example of a topologically homogeneous dag with non - correlation in terms of @xmath25 and @xmath26 @xcite .
graphs were obtained by the construction of an undirected er graph @xmath219 where @xmath220 is the set of edges ( undirected links ) . directed acyclic condition was obtained by a process of random numbering of nodes @xcite .
the direction of the arrows was defined attending @xmath221 ( i.e. , condition depicted in eq .
( 6 ) ) fig .
[ er ] shows a representative behavior of the null hierarchical character of @xmath218 ensembles .
notice that the normal distribution is centered for @xmath178 and @xmath222 at zero values , indicating that such random structures have not any hierarchical organization . and
@xmath222 for an ensemble of @xmath223 replicas of directed acyclic er graphs of @xmath224 and @xmath225 caption .
numerical results show gaussian - like distributions centered at zero .
notice that @xmath222 distribution displays a narrower variation than the @xmath178 one in agreement to the @xmath226 inequality.,width=302 ]
hierarchical patterns are known to pervade multiple aspects of complexity . in spite of their relevance ,
it is not obvious in general how to formalize them in terms of a quantitative theory .
this paper presents a definition of hierarchy to be applied to the so - called _ causal graphs _ ,
i.e. , connected , directed acyclic graphs where arcs depict some direct causal relation between the elements defining the nodes .
it is therefore a measure of hierarchy over the structure of the causal flow .
the conceptual basis of this measure is rooted in two fundamental features defining hierarchy : the absence of ambiguity in recovering the causal flow and the presence of a pyramidal structure .
the hierarchy index presented here weights the deviations from such general properties .
the specific expression for this index is derived using techniques and concepts from information theory .
it is shown , thus , that the requirements of hierarchy naturally fit the tension between richness in causal paths against the uncertainty in recovering them depicted by a balance between two conditional entropies . under our previous assumptions
, we have shown that the feed - forward tree is the structure that fully satisfies the conditions for a perfect hierarchical system .
interestingly , trees as perfect representations of hierarchies is a long - standing idea @xcite . in this way
, our mathematical formalization establishes a bridge between the qualitative idea of hierarchy and its quantification .
our approach allows to measure the hierarchy of any system provided that it can be represented in a feedforward causal graph . throughout the paper we emphasized that although hierarchy is deeply tied to order , there are strong reasons to go beyond it .
the most obvious one is that _ order _ is a well established concept and therefore , there is no need for the use of a different word , if we identify it with hierarchy . but another important issue must be taken into account tied to the intuitive notion of hierarchy we propose .
we propose that hierarchy must feature the pyramidal nature of the connective patterns and that this pyramidal structure must be in agreement with the top - down nature of the feed - forward flow of causality .
information theory reveals extremely suitable to define a hallmark to study hierarchy in the terms described in this paper : the richer the structure ( but at the same time , reversible , in topological terms ) , the more hierarchical it is . in this way , since the conditions we defined for a system to be perfectly hierarchical lead us to conclude that a feed forward tree is the perfect hierarchical structure since maximizes the richness without loss of predictability .
it is worth to note that precisely , the pyramidal condition is the key point to guarantee the the predictability .
the extreme case is the feedforward clique - see fig .
[ figgrafets ] ) .
although richness can be increased through pathway redundancy , this effect cancels out due to decreased predictability , leading to a non hierarchical structure .
a particular case is the linear chain .
this representation of a total order relation has null values of both entropies .
in other words , a perfect predictable system but without richness . it is worth to note that both cases are not pyramidal structures .
by contrast , anti - hierarchical ones exhibit an inverted pyramidal structure leading to a different effect . from the perspective of our formalism
, the anti - hierarchical organization occurs by minimizing richness and predictability . in consequence , other structure
different than an inverted tree will be less anti - hierarchical .
therefore it is easy to see that the hierarchical index in absolute value measures the closeness of an ( anti)-hierarchical tree structure capturing somehow the path complexity of the structure .
further work should explore the relation of this theoretical achievement within the framework of a formal measure of complexity .
additionally , this research could be expanded to a more general class of directed graphs containing cycles .
this latter point would be achieved by properly defining a measure of _ how well ordered _ is a net , for it is clear that the presence of cycles will generate conceptual problems in the identification of the causal flow .
this work was supported by the eu @xmath227 framework project complexdis ( nest-043241 , crc and jg ) , the ute project cima ( jg ) , fundacin marcelino botn ( crc ) , the james mcdonnell foundation ( bcm and rvs ) and the santa fe institute ( rvs ) .
we thank complex system lab members for fruitful conversations and an anonymous reviewer for his / her constructive suggestions .
d. s. bassett , e. bullmore , b. a. verchinski , v. s. mattay , d. r. weinberger , and a. meyer - lindenberg .
hierarchical organization of human cortical networks in health and schizophrenia .
, 28:92399248 , 2008 . | in this paper we explore the concept of hierarchy as a quantifiable descriptor of ordered structures , departing from the definition of three conditions to be satisfied for a hierarchical structure : _ order _ , _ predictability _ and _ pyramidal structure_. according to these principles we define a hierarchical index taking concepts from graph and information theory .
this estimator allows to quantify the hierarchical character of any system susceptible to be abstracted in a feedforward causal graph , i.e. , a directed acyclic graph defined in a single connected structure .
our hierarchical index is a balance between this predictability and pyramidal condition by the definition of two entropies : one attending the onward flow and other for the backward reversion .
we show how this index allows to identify hierarchical , anti - hierarchical and non hierarchical structures .
our formalism reveals that departing from the defined conditions for a hierarchical structure , feedforward trees and the inverted tree graphs emerge as the only causal structures of maximal hierarchical and anti - hierarchical systems , respectively .
conversely , null values of the hierarchical index are attributed to a number of different configuration networks ; from linear chains , due to their lack of pyramid structure , to full - connected feedforward graphs where the diversity of onward pathways is canceled by the uncertainty ( lack of predictability ) when going backwards .
some illustrative examples are provided for the distinction among these three types of hierarchical causal graphs .
* the idea of hierarchy has been largely attributed to a disparate number of systems and , although easily perceived , its quantification is not a trivial issue . in this work
we quantify the hierarchy of a given causal structure with a feedforward structure . starting with the representation of a system of causal relations as a graph
, we define a non heuristic measure of hierarchy having strong grounds on the principles of information theory .
we depart from the definition of the conditions for a system to be considered perfectly hierarchical : a pyramidal structure with a completely predictable reversion of the causal flow . in this context , a hierarchy index is defined by weighting how far is a given feedforward structure from these conditions .
as we shall see , structures that fully satisfy this property belong to a special class of trees .
our estimator allows to establish a quantitative criterion for the definition of hierarchic , non - hierarchic and anti - hierarchic networks . * |
our goal in developing v - t theory is to apply the established techniques of many body physics to the mechanical problem of the motion of atoms in a monatomic liquid .
the many body formulation begins with an approximate hamiltonian @xmath8 composed of a complete orthogonal set of excitations , bosons or fermions , whose exact statistical mechanics is known .
@xmath8 is complemented with an interaction hamiltonian @xmath9 , expressing the key effect missing from @xmath8 , and often but not always treated as a perturbation .
the approach was developed to treat the wide variety of physical behaviors observed in condensed matter , and is well illustrated in the monographs of pines @xcite and kittel .
@xcite the principles are evident in boltzmann s theory for a gas of freely moving atoms which interact via collisions , @xcite and in the theory of born and coworkers for a crystal of harmonic phonons interacting via anharmonicity .
@xcite in the absence of such a many body formulation , liquid dynamics theory has been advanced by a series of conceptual developments .
an important early step was learning how to construct physically realistic interatomic potentials for nearly - free - electron metals from pseudopotential perturbation theory .
@xcite it was shown that md calculations using these potentials give an excellent account of experimental data for elemental condensed systems , for example : for thermodynamic properties of crystals @xcite and liquids , @xcite and for the liquid structure factor @xcite and dynamic structure factor .
@xcite _ ab initio _
md was introduced , @xcite and has since become the method of choice for reliable calculations on many liquid types .
the resulting physical picture of liquids as nuclei moving on the ground state adiabatic potential , and subject to conventional statistical mechanical averaging , is the basis of our theoretical work .
formal derivation of the corresponding condensed matter hamiltonian is reviewed in ch . 1 of ref . .
the conceptual contribution of v - t theory is to classify the valleys that comprise the liquid s many - body potential surface as random or symmetric .
the random valleys all have the same statistical mechanical averages , and together they dominate the potential surface in the thermodynamic limit .
@xcite originally a hypothesis , this classification has been numerically verified for various representative systems .
@xcite the atomic motion is vibrations in one ( any ) random valley , interspersed by transits , which carry the system between random valleys .
accordingly , we define the extended random valley as the harmonic extension to infinity of any random valley , and take the hamiltonian @xmath10 for motion in one extended random valley as our zeroth order liquid hamiltonian . to study any particular function ,
we first calculate the function assuming the hamiltonian is @xmath11 and then we examine the ( often small ) remainder to see how models of transit motion can account for it .
( studies of liquid motion in terms of hops between potential valleys date back at least to stillinger and weber @xcite and zwanzig , @xcite but without the benefit of the v - t classification , which makes clear the choice of initial hamiltonian . )
transits have been observed in md calculations for liquid na and ar , at very low temperatures where single transits are well resolved .
@xcite these transits occur in the highly correlated motion of a small local group of atoms . at and above melting
, transits proceed at a high rate throughout the liquid . in its initial formulation ,
v - t theory incorporated an empirical melting - entropy constant to represent the transit contribution to the entropy of elemental liquids .
@xcite this formulation requires extension in two ways : in order to treat _ all _ thermodynamic properties , the transit entropy theory must be replaced by a free energy theory ; and in order to make a purely _ liquid _ theory , all sensitivity to the nature of the crystal or the melting process must be removed .
these extensions were carried out through an analysis of the temperature dependence of experimental entropy for elemental liquids , @xcite plus a statistical mechanics free energy model calibrated to the entropy results .
@xcite _ a priori _ density functional theory ( dft ) calculations then verified the theory to high accuracy for thermodynamic properties of liquid na and cu .
@xcite in nonequilibrium problems , v - t theory has achieved success in two applications to dynamic response .
first , it was found that an _ a priori _ calculation of the brillouin peak dispersion curve , based on the vibrational motion alone , is in essentially perfect agreement with md calculations and with experimental inelastic scattering data for liquid na .
@xcite second , in comparison with the benchmark theories of generalized hydrodynamics and mode coupling , a near-_a priori _ theory of self dynamic response was found to have significantly improved analytic properties and modestly improved accuracy .
@xcite the purpose of the present work is to test the viability of v - t theory for the distinct part of the density autocorrelation function .
hansen and mcdonald @xcite define time correlation functions related to self and distinct contributions to dynamic response , and discuss the self dynamic structure factor , including its gaussian approximation .
these authors also point out it is possible in principle to measure separately the self and distinct parts .
we have not found studies dedicated to the distinct part , presumably because the self part and the total function cover the entire theoretical problem .
however , the distinct part must be studied separately , because the strong difference between self and distinct correlations implies a similar difference in their decorrelation processes .
formulation of dynamic response theory is a quintessential quantum mechanics problem , analyzed for crystals by maradudin and fein , @xcite ambegaokar , conway and baym , @xcite cowley , @xcite and ashcroft and mermin , @xcite and for crystals and liquids by lovesey @xcite and glyde .
@xcite however , our focus is the elemental liquids in general , so we omit the few quantum liquids and work in classical statistical mechanics .
we study the density autocorrelation function @xmath12 , also called the intermediate scattering function , for all wave vectors @xmath2 and time @xmath3 .
expansions in powers of scattering events are inefficient , and we must work with the full theory , correct to all powers of @xmath2 ( see appendix n of ref . ) .
we apply the primitive - lattice harmonic crystal analysis from the above references to a system moving in an extended random valley , then extract the classical limit to obtain the zeroth order liquid function @xmath13 ( details may be found in ref . ) . in the present paper we begin with the distinct part @xmath14 , and introduce the decorrelating effects of transits to damp @xmath14 to the function representing v - t theory , @xmath15 .
ii sets out the standard plan " for constructing the theoretical function @xmath15 .
the perfect vibrational system provides equations for @xmath14 and for the motional and structural correlations .
transits are responsible for all damping beyond the dephasing already contained in @xmath14 . the transit contribution to @xmath16
is assigned to the structural correlation and is calibrated from md .
a crossover time @xmath4 is defined , such that motional correlation damps to zero on @xmath17 , and structural correlation damps to zero on @xmath18 . in sec .
iii , the massively averaged motion due specifically to transits on @xmath17 is modeled as the transit drift , and a theory is made for its contribution to motional decorrelation . in sec .
iv , the transit random walk theory from self dynamic response @xcite is extended to the structural decorrelation .
the agreement of theory with md is at the remarkable level of @xmath19 , for all @xmath3 , and all @xmath2 for which the standard plan applies .
v examines the validity of the standard decorrelation plan as function of @xmath2 . in sec .
vi , we review our main accomplishment : an expression @xmath20 that includes the physical ideas behind v - t theory .
we also summarize the physical nature of the standard plan and the current status of transit modeling , and identify the path to making the present theory fully _ a priori_. in this work , we frequently use md results to validate or parametrize our theory .
our computational system represents liquid na at 395 k , a bit above the melting temperature @xmath21 k. the system is a cube containing @xmath22 atoms , with periodic boundary conditions .
the md time step is 7 fs .
the interatomic potential is based on pseudopotential theory , @xcite and has produced excellent agreement with a wide range of experimental data .
we study the distinct autocorrelation function , @xcite defined by @xmath23 for a system of atoms @xmath24 , located at @xmath25 at time @xmath3 .
the average may be evaluated analytically for simple systems ( like the vibrational system below ) or numerically for a single md system , on an equilibrium trajectory , and includes an average over the star of @xmath26 when periodic boundary conditions are used ( a star is the set of all wave vectors related by the cubic point group ) .
the function measures static pair configurational correlations in terms of the structure factor @xmath27 , and measures their statistical average time decay at @xmath28 . to evaluate eq .
( [ eq1 ] ) for the vibrational system , we write @xmath29 where @xmath30 is the random valley equilibrium position ( structural site ) and @xmath31 is the motional displacement .
the vibrational average is then @xcite @xmath32 where @xmath33 , @xmath34 is the debye - waller factor , @xmath35 and the motional time - correlation functions are given by @xmath36 the normal modes are labeled @xmath37 , the three zero - frequency modes being omitted , mode @xmath38 has frequency @xmath39 , and eigenvector @xmath38 has cartesian vector @xmath40 at atom @xmath41 . the right side of eq .
( [ eq3 ] ) is also averaged over each @xmath42 star . to calibrate the vibrational hamiltonian for liquid na
, we quench the na computational system to a random structure , @xcite and there evaluate the equilibrium positions and the vibrational frequencies and eigenvectors .
@xcite these na parameters are then used to evaluate the vibrational functions in this paper , and all other vibrational functions for liquid na .
any one random structure is suitable for this purpose because of the statistical similarity of the random valleys in the thermodynamic limit .
equation ( [ eq3 ] ) for @xmath14 makes use of three data sets , the structural positions @xmath30 , the debye - waller factors in eq .
( [ eq4 ] ) , and the motional time correlation functions in eq .
( [ eq5 ] ) .
these data sets are all strongly coupled inside the sums in eq .
( [ eq3 ] ) , and that equation admits of no acceptable decoupling approximation .
the key to analysis is in the time dependence .
@xmath14 is subject to natural vibrational decorrelation , or vibrational dephasing , of the @xmath43 factors in eq .
( [ eq5 ] ) , as @xmath3 increases from zero .
this process is always present , and has the effect of reducing the time correlation functions to zero as @xmath44 .
what remains of @xmath14 as @xmath44 is therefore given by @xmath45 because the debye - waller factors are positive and increasing with @xmath2 , eq .
( [ eq4 ] ) , @xmath46 as @xmath2 increases .
@xmath14 accounts for all correlations in a perfect vibrational system .
these correlations can be classified as motional and structural .
we shall follow this intuitive notation , in a modification designed for v - t theory .
a. motional correlation is that contained in the time correlation functions with @xmath47 in eq .
( [ eq3 ] ) .
motional correlation is in the normal mode motion and resides in the eigenvectors , eq .
( [ eq5 ] ) .
b. for vibrational motion in a single random valley , each atom @xmath41 remains within a small volume around its equilibrium position @xmath48 .
this constraint describes the structural correlation , as the term is used here , and it is contained in eq .
( [ eq6 ] ) . fig .
[ fig2 ] compares the distinct functions @xmath49 and @xmath50 , which describe the initial correlations for all the allowed @xmath2 in our system .
fluctuations are larger in the vibrational curve because it is calculated for a single random valley while the md data averages over random valleys . the difference @xmath51 is formally identified in v - t theory as the transit contribution .
[ h ! ] at allowed @xmath2 ( large circles identify the test @xmath2 from table i ) and the line is @xmath50.,title="fig:",scaledwidth=45.0% ] as the figure shows , the vibrational contribution alone produces the correct peak structure , and it correctly locates the first peak .
it also gets the width of the peak correct and provides about two thirds of its height .
in fact , the transit contribution @xmath52 is a small correction everywhere except the first peak tip , where it is large and negative .
( we found a similar situation with the brillouin peak dispersion curve @xcite : the vibrational part alone gets the location of the peak in @xmath53 but not its value . )
while we do not yet have a microscopic theory of how transits affect initial correlations , the fact that @xmath52 is large only at the nearest neighbor distance suggests that it is structural in the sense defined above and is thus contained in @xmath54 .
accordingly we write @xmath55 where @xmath56 has no dependence on @xmath57 .
the formation of excess correlation by transits is one matter , but the transit - induced process that cause those correlations to decay is another .
we do have a microscopic theory for those processes , and we begin to lay the groundwork for that theory in the next section . [ cols=">,>,>,>,>,<,^,^,^",options="header " , ] [ table ii ] success of the standard plan in the first peak regime encourages application of the same procedure to the brillouin peak regime .
the challenge here is the brillouin peak oscillation .
the oscillation itself is present in both @xmath58 and @xmath14 , but it requires an additional damping in the vibrational function to agree with the md function at @xmath59 .
however , in the standard plan , the oscillation is completely removed in @xmath15 , via eq .
( [ eq32 ] ) , so the entire oscillation in @xmath58 appears as an error in @xmath15 at @xmath59 .
this is not a problem for the standard plan , since the oscillation amplitude is generally less than our theoretical error .
comparison of theory with md in the brillouin peak regime is shown in fig .
[ fig11 - 12]a and [ fig11 - 12]b for @xmath60 and @xmath61 , respectively .
in spite of missing the oscillation , the standard plan remains accurate to @xmath62 in the brillouin peak regime ( fig .
[ fig9]b ) .
it is significant that we can set @xmath63 for both test @xmath2 in the brillouin peak regime ( see table ii ) .
[ h ! ] : comparison of the complete v - t theory ( line ) with md ( circles ) , for two @xmath2 in the brillouin peak regime .
inset shows crossover and the brillouin peak oscillation .
, title="fig:",scaledwidth=35.0% ] : comparison of the complete v - t theory ( line ) with md ( circles ) , for two @xmath2 in the brillouin peak regime .
inset shows crossover and the brillouin peak oscillation .
, title="fig:",scaledwidth=35.0% ] the @xmath64 data exhibit a long time tail at @xmath65 and @xmath66 , shown in fig .
[ fig13 ] at @xmath67 ( see also fig . [
we are investigating the source of this feature .
[ fig13 ] also shows the good agreement of theory with md as the curves approach zero , before the tail begins .
: long - time data showing the md tail at @xmath68 ( circles ) , not yet addressed in v - t theory ( line ) .
, title="fig:",scaledwidth=35.0% ] in the large @xmath2 regime , at @xmath2 beyond the first peak , three @xmath2 were chosen for study , located respectively at the first minimum , second maximum , and second minimum of @xmath50 ( see fig .
[ eq1 ] and table i ) . at @xmath69 , shown in fig . [ fig14 ] , it is perhaps surprising that the standard plan still works well . both motional and structural decorrelation are accurately accounted for , the theoretical slope discontinuity at @xmath70 is insignificant , and we are able to set @xmath71 in the structural damping . the deviation @xmath72 in fig .
[ fig9 ] is well below @xmath73 .
[ h ! ] [ fig14 ] : comparison of the complete v - t theory ( line ) with md ( circles ) , for @xmath74 , at the minimum after the first peak .
vertical line is at @xmath75.,title="fig:",scaledwidth=35.0% ] as @xmath2 increases further , two important changes occur .
first , @xmath54 goes to zero , as noted following eq .
( [ eq6 ] ) , and as seen in the last three entries of table i. second , the decorrelation process becomes faster , with @xmath4 falling below the first - transit period of @xmath76 , again as seen in table i. these changes mark the ultimate convergence of the distinct autocorrelation function toward zero as @xmath2 increases .
both trends are also found in the self autocorrelation function as @xmath2 increases toward the free particle limit .
@xcite as a result , at @xmath77 and @xmath78 , the total process resembles a single motional decorrelation of an unusual shape .
the indication is that @xmath79 does not measure structural correlation , and measures motional correlation in a way different from what we have seen at smaller @xmath2 .
there are several @xmath2-segments where @xmath50 and/or @xmath49 are of small magnitude , say @xmath80 , where analysis is difficult for two reasons . first , the values of the md and vibrational autocorrelation functions are small enough to be compromised by finite-@xmath81 error , at our present @xmath22 .
second , the md and vibrational functions cross zero at different @xmath2-values , a phase shift not treated in the standard plan .
it makes sense to leave these effects until the dominant part of the theory is developed .
this is where @xmath50 and @xmath49 are positive in the first peak .
this regime provided the test case from which the standard plan was developed . here
the @xmath58 curves have uniform character across the @xmath2-range , and so do the @xmath14 curves . since distinct and self contributions are both positive , no cancellation complications arise .
this is where the brillouin peak oscillation is apparent in @xmath58 ; in our system it runs from @xmath82 , the smallest allowed @xmath2 , to somewhat beyond @xmath83 . here
, @xmath50 and @xmath49 are around @xmath84 ; on this scale the standard plan is accurate , as shown in fig .
[ fig9 ] , and may be applied . in this regime ,
the self and distinct autocorrelation functions nearly cancel for md data , and also for vibrational data .
the brillouin peak oscillation appears entirely in the distinct function , and contains important physics not addressed in the standard plan , namely the inelastic scattering cross section as function of @xmath2 and @xmath85 . for our system
, the brillouin peak oscillation appears in @xmath14 with the same frequency as it appears in @xmath58 , for all @xmath2 in the regime .
hence @xmath14 already contains the brillouin peak dispersion curve , _ a priori _ and to very high accuracy .
@xcite what is still required for a complete theory of dynamic response is an extremely accurate decorrelation theory for the brillouin peak oscillation . beyond the first peak , near the first and second minima and the second maximum , where @xmath86 is large enough to justify an exploratory investigation , the character of the decorrelation process changes under two influences : the motional decorrelation becomes faster , and @xmath54 goes to zero ( see table i ) . at @xmath69 ,
the standard plan still applies and is still accurate , but a practiced eye will see the beginning of change . at @xmath77 and @xmath78 , the entire process has the appearance of a single motional decorrelation , an appearance we expect to remain as @xmath2 increases further , and the md and vibrational autocorrelation functions converge to zero ( see the discussion at the end of sec .
the standard plan does not apply at these @xmath2 .
this work is an investigation into the validity of v - t theory for the distinct density autocorrelation function of a monatomic liquid .
our main result is eqs .
( [ eq20 ] ) and ( [ eq32 ] ) , which together accurately model @xmath87 over the broad range of @xmath2 for which the standard decorrelation plan applies . that result depends on a theoretical notion , the transit drift introduced in sec .
iiia , which is the basis of our theories of motional and structural decorrelation in @xmath87 . in this study
, the vibrational contribution is parametrized by appeal to a realistic many - body potential for na , and the various transit parameters are determined from md simulations using that potential , but we emphasize that this is not necessary : once one understands the physical meaning behind the parameters , their values may be acquired by the method of one s choice . with that in mind , we summarize the physical considerations that led to the model and give the parameters their meaning . at the outset
, the two classes of configurational correlation , motional and structural , are precisely defined by the perfect vibrational system .
the motional correlation is measured by the time correlation functions in eq .
( [ eq5 ] ) , and the motional decorrelation ( due to vibrations and transits ) carries these functions from their @xmath88 values to zero . in the process
, @xmath14 goes from @xmath49 to @xmath54 according to eqs .
( [ eq3 ] ) and ( [ eq6 ] ) .
the structural correlation is measured by the set of equilibrium positions @xmath89 , and the debye - waller factors @xmath90 , in eq .
( [ eq6 ] ) for @xmath54 .
the @xmath48 remain fixed during the motional damping , then the @xmath48 begin to move and proceed to damp @xmath54 to zero .
this outline of the process is defined entirely in terms of the theoretical function @xmath14 , the definition applying to all monatomic liquids .
the three functions @xmath52 , @xmath91 , and @xmath92 that enable calculation of the detailed time dependence are transit properties . in physical meaning , @xmath52 is the transit contribution to the static structure factor .
@xmath52 is supposed to be mainly structural , hence is modeled as a multiple of @xmath54 .
the crossover time @xmath4 is the time required to damp the motional correlation to zero , under the simultaneous actions of natural dephasing and transit - induced decorrelation .
the physical key to @xmath4 is the condition that long - time motional correlation due to low - lying normal modes is damped out by transits in the liquid .
finally , the structural damping coefficient @xmath92 expresses the transit random walk , and an _ a priori _ zeroth - order theory for @xmath92 is already in place .
@xcite all these properties are attributed to the same transit motion in every monatomic liquid .
the transit drift is the massively averaged motion resulting specifically from transits in the liquid state , and present in addition to the normal mode motion . in sec .
iii , in constructing a model equation for the drift - induced motional decorrelation , in the period @xmath5 , three predictions are made about characteristic properties of the drift .
the predictions are expressed in terms of the transit parameters @xmath93 and @xmath94 , previously calibrated independently of any study of @xmath12 , with values given in eq .
( [ eq14 ] ) .
the predictions are listed in eq .
( [ pinco ] ) .
a. the first prediction is @xmath95 .
this is qualitatively correct .
specifically , @xmath96 ps , while from table i the average @xmath4 for seven @xmath2 in the standard plan is @xmath97 ps
. b. the second prediction is @xmath98 .
this is qualitatively correct .
specifically , @xmath99 , while from table i the average @xmath100 for seven @xmath2 in the standard plan is @xmath101 . c. the third prediction is @xmath102 for @xmath5 . the linear @xmath3 dependence is used in evaluating the motional decorrelation , eq .
( [ eq22 ] ) , and agreement of theory with md in fig .
[ fig567 ] verifies the @xmath3 dependence within the observed errors .
the level of agreement is excellent for nonequilibrium data ; further , the systematic character of the v - t theory errors shown in fig .
[ fig9]a indicates that the motional decorrelation theory of eqs .
( [ eq19 ] ) and ( [ eq20 ] ) can still be improved .
the transit drift is a steady state motion , hence is always present in the liquid .
however , in order to address the structural decorrelation at @xmath103 , it is necessary to resolve the drift into its effective motion of the equilibrium positions .
this resolution yields the transit random walk .
indeed , the time required after @xmath88 for the random walk to come into effect provides the physical explanation for the delay to @xmath4 of the onset of structural decorrelation , as it appears in eq .
( [ eq32 ] ) .
simple exponential decay , beginning after a delay period , is a hallmark of md data for self and distinct density autocorrelation functions . in self decorrelation ,
the theoretical damping coefficient @xmath104 is in excellent agreement with md data for all @xmath2 .
the result constitutes a unification within v - t theory : one can not calculate the density autocorrelation damping coefficient @xmath105 from the self diffusion coefficient @xmath106 , @xcite but one _ can _ calculate both @xmath104 and @xmath106 from the same random walk of the equilibrium positions . finally , we shall make a highly approximate model for the anistropic damping coefficient @xmath107 . consider a representative pair of atoms @xmath108 which are nearest neighbors at @xmath88 . because the atomic dynamics enforces a minimum distance between atoms ,
the first - transit drift of atom @xmath41 will not be isotropically distributed about @xmath109 , but will move preferentially away from @xmath110 .
the transit random walk will not develop about @xmath109 , but about @xmath111 , where @xmath112 is a small displacement in the direction of @xmath113 .
the picture can be developed from eq .
( [ eq17 ] ) .
the resulting @xmath107 will be positive for @xmath2 on the first - peak leading edge , and negative at the first - peak tip .
this is what we see in sec .
moreover , since the effect appears only for nearest neighbors , @xmath114 will be significant only for @xmath2 in the first peak regime . even for @xmath2 in the first peak ,
@xmath114 is small because the correlation is weak after the first transit .
all these details of @xmath114 are confirmed by the last column of table ii .
the theoretical challenge now is a more quantitative understanding of @xmath114 .
our expressions for @xmath115 rely on the vibrational parameters @xmath116 and three transit functions @xmath117 , and @xmath92 . in this study ,
the vibrational parameters were determined from a single random valley of the na pseudopotential , and the transit functions were determined from combined md and vibrational data .
this is done in order to show the extreme accuracy of the calibrated theory ; we believe this high - accuracy capability gives credence to the formulation . on the other hand ,
the vibrational parameters can be determined from first principles : examine a random valley on a potential surface calculated from dft .
the remaining functions are determined by transit dynamics and can in principle be calculated from a transit model .
( the current transit model already provides reasonably accurate estimates @xmath118 for @xmath4 , and @xmath104 for @xmath92 . ) an improved transit model , in addition to providing better estimates for @xmath4 and @xmath92 , would ideally ( a ) describe the transit decorrelation processes that dominate for @xmath2 outside the standard plan and ( b ) explain the origin of the initial transit - based correlations , given by @xmath52 . | we examine the distinct part of the density autocorrelation function @xmath0 , also called the intermediate scattering function , from the point of view of the vibration - transit ( v - t ) theory of monatomic liquid dynamics .
a similar study has been reported for the self part , and we study the self and distinct parts separately because their damping processes are not simply related .
we begin with the perfect vibrational system , which provides precise definitions of the liquid correlations , and provides the vibrational approximation @xmath1 at all @xmath2 and @xmath3 .
two independent liquid correlations are defined , motional and structural , and these are decorrelated sequentially , with a crossover time @xmath4 .
this is done by two independent decorrelation processes : the first , vibrational dephasing , is naturally present in @xmath1 and operates to damp the motional correlation ; the second , transit - induced decorrelation , is invoked to _ enhance _ the damping of motional correlation , and then to damp the structural correlation .
a microscopic model is made for the transit drift , " the averaged transit motion that damps motional correlation on @xmath5 .
following the previously developed self - decorrelation theory , a microscopic model is also made for the transit random walk , " which damps the structural correlation on @xmath6 .
the complete model incorporates a property common to both self and distinct decorrelation : simple exponential decay following a delay period , where the delay is @xmath4 , the time required for the random walk to emerge from the drift .
our final result is an accurate expression for @xmath0 for all @xmath2 through the first peak in @xmath7 .
( a modification will be required at @xmath2 where @xmath7 converges to zero . )
the theory is calibrated and tested using molecular dynamics ( md ) calculations for liquid na at 395 k ; however , the theory itself does not depend on md , and we consider other means for calibrating it . |
gravitational lensing of electromagnetic waves is a well known phenomenon predicted by the general theory of relativity ( for a review on this issue see schneider , ehlers and falco @xcite ) . in principle
, gravitational lensing of gravitational waves should occur in the same way as it does for light .
the most obvious difference is that gravitational wave propagation is not disturbed by dust grains , as happens for light , so that the central part of our galaxy may be investigated by using the next generation of gravitational wave detectors . in a very interesting paper ruffa ( @xcite ) , assuming that the mass of the galactic center is in the form of a massive black hole with mass @xmath0 m@xmath1 ( for a galactic center overview see eckart , genzel , ott and schoedel @xcite ) , the gravitational wave lensing problem was studied by a typical fraunhoffer diffraction approach .
it was pointed out ( ruffa @xcite ) that extra galactic sources can be amplified by a factor of about @xmath2 and galactic neutron stars by over @xmath3 .
the author also argued that the earth would take about 10.1 days to traverse the focused region of the extra galactic sources . for galactic bulge sources the focused region
would be scaled down by a factor of nearly 3 and hence the observing time would be reduced to about 3 days .
the question that naturally arises is `` how likely is it that we see the proposed enhancement ? '' to answer this question we need to estimate the number of sources that could be expected to be observed , both galactic and extra - galactic . in principle , of course , there are many extra - galactic sources as there are @xmath4 galaxies .
however , we need to limit the number , applying a cut - off by requiring that the expected intensity ( after amplification ) be greater than the sensitivity of the detector . as will be seen , the chances of seeing the dramatic enhancement calculated by ruffa are extremely small .
this is due to the fact that ruffa assumes a very special geometry , with the source , lens and observation point aligned . whereas ruffa considered fraunhoffer diffraction , an analysis using `` geometric optics '' for gravitational waves had been undertaken , in which the special geometry of ruffa was not assumed ( de paolis , ingrosso and nucita @xcite ) .
this analysis gave a much higher probability but a lower enhancement .
in this paper , we evaluate the probability of enhancement of the gravitational wave signals as a consequence of diffraction by massive compact objects in the highly aligned geometry .
we also consider the lensing effect due to a possible companion star of the gravitational wave source in the framework of the classical microlensing theory .
in addition to this case we consider the gravitational wave lensing due to the massive black hole at the galactic center .
as we shall see , the gravitational wave signal amplification due to the companion object ( or the black hole at the galactic center ) might be detectable by the virgo detector for some orbital parameters of the binary system .
let us assume that the observer , the lens and the source ( a spinning pulsar ) are on the same line . the lens will focus the gravitational waves that pass through the einstein radius @xmath5 to the earth in a region orthogonal to the line of sight . here
@xmath6 is the distance between the observer and the lens ( @xmath7 kpc ) and @xmath8 is the lens - source distance .
using diffraction theory , the focused area is determined by @xmath6 and the diffraction pattern for the circular annular region bounded by radii @xmath9 and @xmath9+@xmath10 .
note that a gravitational wave that passes at the distance @xmath9+@xmath10 from the lens is focused in @xmath11 .
it is possible to show ( by evaluating the 3 db points of the signal , see ruffa @xcite ) that the radius of the focused region is given by @xmath12 where @xmath13 is the gravitational wave length .
hz , the emitted gravitational waves has frequency @xmath14 hz ( see e.g. shapiro and teukolsky @xcite ) .
consequently , the gravitational wave length turns out to be @xmath15 cm . ]
as pointed out by ruffa , the increase in intensity depends on the ratio ( known as the intensity ratio ) of the annular region bounded by the radii @xmath9 and @xmath9+@xmath10 to the focused region area .
this can be determined by the system of equations cm and , from eq .
( [ rhod ] ) , @xmath16 . ]
@xmath17 we note that the first equation is ruffa s eq .
( 4 ) at first order approximation ( since @xmath16 ) and the second one derives from differentiating the expression for the einstein radius @xmath9 with respect to @xmath6 . solving for @xmath18 , we finally obtain @xmath19 which reduces to @xmath20 for @xmath21 .
taking into account the previous equations and the expression of the einstein radius , the intensity ratio turns out to be @xmath22 implying a gravitational wave amplification factor is proportional to the square root of the energy flux , the enhancement in @xmath23 is proportional to the square root of the intensity ratio . ]
@xmath24 [ cols="<,<,<,<,<,<,<",options="header " , ] assume now that ruffa s treatment still holds if the source is not fully aligned with the observer - lens line but forms at least an angle @xmath25 with it .
consequently , the probability density of finding the source within a circular region of radius between @xmath26 and @xmath27 is proportional to the ratio between that circular area and the surface of the cylinder centered at the galactic center and having radius @xmath8 and height @xmath28 . from simple geometric considerations , this probability density , after integration over @xmath29 ,
gives the following probability of finding a source within the radius @xmath30 @xmath31 on the other hand , the real probability to observe a gravitational wave enhancement in the case of the observer , lens and source on the same line of sight has to take into account also the number of expected sources behind the central black hole . from an extrapolation of the neutron star ( ns ) birthrate ( narayan and ostriker @xcite ) yr ( tammann , loeffler and schroeder @xcite ) .
these estimates do not take into account any sn rate evolution with time and the fact that only about @xmath32 of the historical sne have been close and bright enough to have been detectable by the naked eye , and therefore have to be considered as a lower limit to the inferred present galactic sn number .
a more realistic estimate of the sn rate for our galaxy is about one over 20 years ( panagia @xcite ) .
observations show that large late - type spiral galaxies produce about 10 sne per century ( tammann , loeffler and schroeder @xcite ) . ] and from the number of supernovae required to account for the heavy element abundance in the milky way ( arnett , schramm and truran @xcite ) , we expect that our galaxy contain @xmath33 nss . for simplicity
we assume that the ns population has a usual disk - like shape , i.e. the number density profile is given by @xmath34 where @xmath35 m@xmath1 pc@xmath36 , @xmath37 m@xmath1 , @xmath38 kpc and @xmath39 kpc is the typical length scale of the galactic disk . in order to consider the distribution of matter behind the black hole , equation ( [ probdistanza ] ) has to be multiplied by the factor @xmath40 , @xmath41 being the total volume of the disk . consequently , the probability to have an enhancement of the gravitational wave signal according to the mechanism proposed by ruffa is @xmath42 in fig .
[ fig1 ] we give the probability @xmath43 as a function of the source distance @xmath8 from the galactic center .
as one can see , the probability to have a gravitational wave enhancement _ a la ruffa _ by the black hole at the galactic center from a pulsar in the line of sight to the galactic center is as low as @xmath44 per year .
the total probability ( assuming that there are no pulsars closer than @xmath45 kpc to the galactic center ) is @xmath46 per year . in other words
, we could expect to wait for a hundred times the age of the universe to see one !
let us see now what happens in the case of a gravitational wave signal due to a supernova explosion in other galaxies behind the galactic black hole .
since @xmath21 , in this case @xmath47 .
the probability of a galaxy being within the focused region at a distance @xmath8 is simply @xmath48 .
thus the total probability will go as @xmath49 . since @xmath50 , we only need to find the probability of a galaxy at the closest distance to be on the line of sight , i.e. the number density of galaxy does not matter .
it is known that supernovae explode in typical galaxies less frequently than once every @xmath51 years .
consequently , the relevant probability is @xmath52 per year .
taking @xmath53 as 1 mpc , the probability is less than @xmath54 per year ! as such , there is no real likelihood of being able to see the ruffa enhancement .
of course , there are also other expected sources of bursts of gravitational waves from gamma - ray bursts ( see e.g. ruffert and janka @xcite ) or binary systems ( see e.g. de paolis , ingrosso and nucita @xcite and references therein ) that could in principle be detectable by virgo ( jaranowski and krlak @xcite and passaquieti @xcite ) , ligo ( barish @xcite ) or lisa ( hiscock , larson , routzahn and kulick @xcite and vecchio @xcite ) experiments . however , from the discussion above , it can easily be calculated that the expected probability of seeing a gravitational wave amplification by the central black hole of one of such event is even less than @xmath54 per year .
let us consider now the enhancement of the gravitational waves emitted by a rotating neutron star due to the microlensing effect of a companion star moving around the compact source on a circular orbit with radius @xmath55 .
let @xmath56 and @xmath57 be the masses of the pulsar and of the companion star , respectively .
we further assume that the orbital plane of the binary star is inclined at an angle @xmath58 with respect to the line of sight .
the companion star coordinates , with respect to a reference frame with @xmath59 and @xmath60 axes on the orbital plane , are given by @xmath61 where @xmath62 is the orbital angle measured with respect to the positive direction of the @xmath59 axis pointing towards the observer .
the distance between the companion star @xmath57 and the line of sight to the gravitational wave source turns out to be @xmath63 for a source - observer distance greater than the distance between the source and the lens ( @xmath64 ) , the einstein radius is @xmath65 .
consequently , as in usual microlensing theory , the flux amplification factor is given by ( see e.g. schneider , ehlers and falco @xcite ) @xmath66 with @xmath67 . obviously , since the impact parameter @xmath68 depends on the position of the companion star on the orbit , we expect that the flux amplification @xmath69 is a periodic function of time with period @xmath70 .
we also note that the effect of the amplification is maximal for @xmath71 .
the expected gravitational wave amplitude enhancement is @xmath72 . in table
[ tabella ] , assuming @xmath73 , the maximum flux amplification @xmath74 , the gravitational wave signal amplification @xmath75 and the microlensing event time - scale @xmath76 are shown for different values of the orbital parameters ( @xmath55 , @xmath43 , @xmath58 and @xmath57 ) . as
one can note , the amplification effect decreases for increasing inclination angles while the event time - scale @xmath76 goes up .
since the gravitational wave amplification is the same for a compact or a normal companion star , assuming that virgo will be able to detect only @xmath77 of the pulsars with spin period @xmath78 ms within about @xmath51 kpc from earth in three years of integration , one expects @xmath79 detectable binaries .
therefore , the number of binary systems with plane inclination angle @xmath80 is easily found to be about 10 .
of course , since the probability of having high enough amplification of the signal for long enough is expected to be rather small , the chances of having a really detectable amplified signal from a binary system are low .
some higher probability is expected if the lens object is the black hole at the galactic center , due to its higher mass . in this case , as one can see from table [ tabella ] , both the signal amplification and duration are expected to be high enough to have a signal detectable by virgo . for details about the detectable gravitational wave amplitude see gourgoulhon and bonazzola @xcite . finally , we want to note that the effects of the gravitational light deflection and amplification in eclipsing binary stars has been considered in detail ( marsh @xcite , beskin and tuntsov @xcite ) and by schneider ( @xcite ) for the binary pulsar 1957 + 20 as a way for determining the binary system parameters . actually ,
the gravitational wave amplification is , if gravitational wave astronomy becomes practicable in the future , a much more powerful tool since gravitational waves are never shielded by the lens object ( as happens for light rays ) giving the unique possibility to trace the sources even in regions ( like the galactic center ) where light suffers strong absorption .
1989 arnett w.d . , schramm d.n . and
truran j.w . , 1989 , apj , 339 , l25 barish b.c . ,
2000 , advances in space research , volume 25 , 1165 beskin g.m . and tuntsov a.v . , 2002 , preprint astro - ph/0208095 de paolis f. , ingrosso g. and nucita a.a . , 2001 ,
a&a 366 , 1065 de paolis f. , ingrosso g. and nucita a.a . , 2002 ,
a&a 388 , 470 eckart a. , genzel r. , ott t. and schoedel r. , 2002 , preprint astro - ph/0201031 gourgoulhon e. and bonazzola s. , 1996 , proceedings of the international conference on gravitational waves : sources and detectors , italy : world scientific hiscock w.a .
, larson s.l . ,
routzahn j.r . and kulick b. , 2000 , apj 540 , l5 jaranowski p. and krlak a. , 1999 , phys.rev .
d59 , 063003 marsh t. r. , 2001 , mnras 324 , 547 narayan r. and ostriker j.o.m . , 1990 ,
apj , 352 , 222 panagia n. , 1999 , proceedings of the international summer school on _ experimental physics of gravitational waves _ ( eds .
barone m. et al . ) , world scientific , pag .
107 passaquieti r. , 2000 , nuclear physics b 85 , 241 ruffa a.a .
apj 517 , l31 , 1999 ruffert m. and janka h.t . , 1998 , a&a 338 , 535 schneider j. , 1989 , a&a 214 , 1 schneider p. , ehlers j. and falco e.e . , 1992 , gravitational lenses , cambridge : springer - verlag shapiro s.l .
and teukolsky s.a . , 1983 , _ black holes , white dwarfs , and neutron stars _ , john wiley and & sons , new york tammann g. a. , loeffler w. and schroeder a. , 1994 , apjs 92 , 487 vecchio a. , 1999 , in : _ gravitational waves , third e. amaldi conference _ , aip conference proceedings , vol . 523 p. 238
, aip press | in a recent paper ( ruffa @xcite ) it was proposed that the massive black hole at the galactic center may act as a gravitational lens focusing gravitational wave energy to the earth .
considering the gravitational wave signal emitted by galactic spinning pulsars , an enhancement in the gravitational wave intensity by a factor of a few thousand is found . for galactic and extra - galactic sources the intensity enhancement can be as high as 4,000 and 17,000 , respectively . in this note
we consider the probability of significant signal enhancement from galactic and extra - galactic pulsars by the proposed mechanism and find that it is actually negligible . the lensing effect due to a possible companion object ( a star or the galactic center black hole ) of the gravitational wave source is also investigated in the framework of the classical microlensing theory . |
the standard picture for the evolution of cataclysmic variables ( cv ) , now nearly 2 decades old , has been reasonably successful as a framework for interpreting the phenomenology of cv . in this interpretation ,
angular momentum loss by gravitational waves and magnetic braking causes cataclysmic binaries to evolve from longer to shorter orbital periods ( @xmath0 ) , with magnetic braking dominating above the ` period gap ' ( @xmath1 hr ) , and a lower braking rate below this gap ( @xmath2 hr ) .
this picture explains some overall statistical properties , for example the existence of the period gap itself and the higher mass transfer rate above the gap compared to short period systems .
an additional element of the phenomenology , independent of the magnetic braking scenario but not in conflict with it either , is the existence of outflows from the accretion disk .
evidence for outflows comes from the profiles in uv - lines seen in outbursts of dwarf novae ( e.g. woods et al . 1992 ) , possibly the single - peaked line profiles of the novalike ( nl ) and sw sex binaries ( see however hellier 2000 for an alternative explantation of these line profiles ) , and other indications for outflow ( e.g. long et al .
a number of observations , however , are hard to fit into the standard picture : * 1 a major problem is the very large spread in inferred mas transfer rates at a given orbital period ( patterson 1984 ; warner 1987 ) .
the braking scenario predicts a close relation between mass transfer rate and orbital period . *
2 an extreme form of the spread in mass transfer rate is shown by the short - period super soft sources ( sss ) . though some sss have long orbital periods and their large transfer rates
can be explained as the result of expansion of the secondary by nuclear evolution , several have short orbital periods , in the same range as the novalike systems .
the secondaries at such periods are too small to have significantly evolved .
an unknown process causes these binaries to transfer mass at rates sufficient ( @xmath3 ) to sustain the observed steady burning of hydrogen at the surface of the white dwarf primary star . *
3 a second unexpected fact is the distribution of the novalike systems , which cluster at periods ( around 4 - 6 hr ) just above the period gap , whereas the braking scenario predicts them to predominate at longer periods , where the predicted mass transfer rate are highest . *
4 the single peaked profiles of the emission lines in the sss , the sw sex stars , the am cvn stars and many nl systems show that in some cases the emission lines are not predominantly formed in the accretion disk , but in optically thin material of large emission measure somewhere else in the system .
the phenomenon is particularly striking in some am cvn systems , ( cf .
warner 1995 ) , where a narrow emission line is seen on top of a broad double peaked emission line from the accretion disk .
this phenomenon has been interpreted in terms of circumbinary material already by solheim and sion ( 1994 ) .
a similar phenomenon may be the low - velocity emission peaks in supersoft sources like qr and ( deufel et al .
* 5 ` additional light ' in systems with parameters such that mid - eclipse of the primary star and accretion disk should be total . in some of these systems
the light remaining at mid - eclipse can not be explained by the spectral type of the secondary star . in the case of uu aqr
the spectrum of the additional light has been determined ; it shows the characteristics of optically thin emission ( baptista et al .
* 6 comparison of the spectral types of the secondary stars with theoretical models ( baraffe and kolb 2000 ) shows them to be significantly cooler , on average , than main sequence stars , as if they were less massive and overexpanded as a result of a recent prolonged period of large mass transfer .
* 7 superhumps above the period gap .
the so - called superhumps , a feature in the lightcurve that drifts in orbital phase , is thought to be due to a resonance in the accretion disk .
this resonance can only occur at low mass ratios , such as occur in the short - period systems .
they are also seen in several novalike systems , however ( patterson , 1999 ) .
this would be another indication that the mass of the secondary in such systems above the period gap is often smaller than expected from the standard braking scenario . *
8 pile up at the minimum period .
the standard braking scenario predicts that the orbital evolution of cvs slows down as they approach the minimum orbital period , after which their orbital period increases again at a very slow rate .
most old cvs are predicted to end up somewhere near this minimum period .
the observed period distribution does not show such a spike in the distribution ( cf .
kolb & baraffe 1999 ) .
though the observations are probably biased against detection of these systems because of their low quiescent luminosity and the long recurrence time of their outbursts , the discrepancy is large , and can probably not be explained entirely by these selection effects ( kolb 2001 ; marsh et al .
2001 ; and references therein ; gnsicke et al .
* 9 value of the minimum period .
the theoretically predicted value of the minimum orbital period of cvs ( excepting thiose with helium star companions ) is around p=70 min , the observed value around 80 min .
stellar models are believed to be sufficiently accurate , but agreement with observations would be obtained if the mean mass transfer rate of the short - period systems were at least a factor 3 higher than assumed in the standard braking scenario . * 10 the rate of magnetic braking .
the standard formulas used for angular momentum loss by magnetic stellar winds is somewhat of an extrapolation , based on loss rates inferred from stars rotating much more slowly than the secondaries of cvs .
recent work by andronov , pinsonneault and sills ( 2001 ) indicates that this extrapolation overestimates the angular momentum loss rate in cvs by as much as 2 orders of magnitude .
if this is correct , the observed mass transfer rates in cvs can not be due to magnetic braking .
though some of the anomalies listed are possibly red herrings which may find unrelated explanations , they all fit qualitatively into the suggestion made here , a cb disk , i.e. a disk surrounding the binary .
if these exist , they would be a major , thus far unrecognized component of cataclysmic binary systems .
a cb disk would be an additional source of angular momentum loss for the binary .
the tides raised in the cb disk by the orbiting binary transfers angular momentum to it , and prevent it from spreading inward . depending on the surface density of the cb disk , the angular momentum loss from the binary would differ , and with it the mass transfer rate
. the sss would , in this interpretation , be the relatives ( greiner et al .
2000 ) of nl and sw sex systems , the difference being the presence of a more massive cb disk .
the increased mass transfer rates caused by cb disks would be systematic , since a cb disk would dissipate only very slowly by spreading .
this might explain the apparent overexpansion of the secondary stars . a higher mass transfer rate would also help to explain the discrepancy between the observed and predicted values of the period minimum .
the obvious question is how a cb disk would be formed .
one possibility would be that it is a ` fossil ' : a remnant from a late stage of the common envelope ( ce ) evolution phase in which the cataclysmic binary was formed ( for a review see taam & sandquist 2000 ) .
this could happen as a ` fall back ' process at the time the common envelope becomes optically thin .
given the difficulty of computing the ce evolution , it is hard to estimate the likelihood of such a formation channel for cb disks .
a second possibility is that cb disks form as a by - product of mass loss from the system .
the single peaked line profiles in sw sex stars indicate outflow velocities up to 2000 km / s , which is well above the escape speed from the binary .
there is large optical depth in these lines at lower velocities as well , however , so that a significant outflow component at velocities of the order of the escape speed from the binary ( a few 100 km / s ) or lower may also exist .
this would be material that would ` hang around ' the binary rather than being ejected .
it would naturally form a cb disk by cooling and settling to the equatorial plane .
it is hard to quantify from the observations how much material might condense into a cb disk in this way .
instead , we have used model calculations in which this uncertainty is parametrized .
these show that a relatively minor amount would suffice to build up substantial cb disk .
for the model calculations ( spruit and taam 2001 , taam and spruit 2001 , dubus et al . 2001 ) we have made use of the observation that evidence for outflows is strongest in systems with high mass transfer rates : dwarf novae in outburst , the sw sex stars , supersoft sources and the am cvn stars .
this suggests that mass loss by outflows increases with mass transfer rate .
we have made the simple assumption that the two are proportional .
in addition , we have assumed that of the outflow , a fixed fraction settles into a cb disk .
the mass input rate @xmath4 into the cb disk is thus a given fraction @xmath5 of the mass transfer rate @xmath6 .
the consequence of the assumed proportionality between mass input into the cb disk and the mass transfer rate is a positive feedback : once a cb disk is present , the angular momentum it extracts from the binary enhances the mass transfer rate from the secondary to the primary .
this increases the mass input rate into the cb disk , increasing its surface density . with this higher density the increasing torque on the binary in turn increases the mass transfer rate .
though the mass transfer might initially be caused by the standard magnetic braking process alone , the mass building up in the cb disk would eventually make this feedback strong enough to become unstable . from this point on
, the mass transfer rate and the input rate into the cb disk would increase in step exponentially . in this way the large mass transfer rates in for example the short - period supersoft sources could develop . in our interpretation , these would be descendants of novalike systems , as suggested before by greiner et al .
( 2000 ) .
we call the phase of high mass transfer resulting from this instability ` accelerated mass transfer ' induced by the cb disk .
the calculations show that the orbital period typically ` bounces ' shortly after accelerated mass transfer sets in ( if the secondary is not significantly evolved ) . the more rapid mass loss from the secondary causes it to expand , forcing the orbital period to increase .
this contrasts with the evolution under standard braking , which results in an essentially monotonically decreasing orbital period . whether the effect of a cb disk on the binary becomes significant depends mostly on the mass input rate into it .
analytic results based on the above assumptions ( spruit and taam 2001 ) suggested that a minimum value of @xmath5 around @xmath7 would be sufficient to cause accelerated mass transfer to occur .
the precise value is unfortunately somewhat sensitive to the physics determining the viscous spreading of the disk .
remarkably , the uncertain value of the viscosity itself , as parametrized through a constant @xmath8-viscosity , has only a weak influence .
this is because the torque exerted by the disk is proportional to @xmath9 , where @xmath10 is the surface density at the inner edge of the disk .
a larger value of @xmath8 is compensated largely by the more rapid spreading of the disk , which lowers @xmath10 .
more important than @xmath8 itself is the relative variation of the viscosity through the disk .
calculations using more detailed physics for the cb disk ( taam and spruit 2001 ) indicated that a rather high value of @xmath5 , around @xmath11 , would be needed for accelerated mass transfer , but these did not include the effect of convection on the structure of more massive cb disks .
this was included by dubus et al .
( 2001 ) , together with a more sophisticated method for calculating the viscous evolution of the cb disk .
in these so far most realistic calculations values of @xmath5 around @xmath12 were found to be sufficient to cause a rapid acceleration to high transfer rates to happen within a hubble time .
in the standard magnetic plus gravitational radiation braking scenario , the mass transfer rate declines rapidly with decreasing secondary mass , such that complete transfer of the secondary mass does not happen within a hubble time . with torques due to a cb disk the evolution of the secondary is more interesting . in the ` accelerated ' transfer phase , the angular momentum loss is determined by the surface density of the cb disk , rather than by the secondary star .
hence the mass transfer rate stays finite as the secondary mass vanishes . as a consequence ,
the secondary dissolves completely within a finite , relatively short time once a phase of accelerated mass transfer has set in , and leaves a single white dwarf .
this would have significant effects on the distribution of systems with orbital period .
if a substantial fraction of all systems go through a cb - induced phase of accelerated mass transfer , they would disappear from the cv population , and the pile - up of systems towards the period minimum around 80 min .
would be replaced by a gradual decline of the number of systems .
this would be more in line with current observations ( cf kolb 2001 ) .
after dissolution of the companion , the white dwarf primary would be left as a ` white widow ' , with a cb disk still surrounding it .
the mass in this disk could be substantial , perhaps as much as @xmath13 .
as this disk cools , it may well form planets in the same way as a protoplanetary disk .
it remains to be seen whether such planets would be detectable , for example through doppler shifts of the nlte cores of the balmer lines of the wd .
the cb interpretation given here implies a large systematic spread in mass transfer rates .
[ as opposed to cyclic mass transfer variations such as proposed by king et al .
( 1995 ) . in this mechanism
the average mass transfer rate does not spread much ] . though this spread explains a number of facts such as the observed varying degree of overexpansion of the secondaries , it is not easily compatible with the explanation of the period gap in terms of disrupted magnetic braking .
the period gap gets ` washed out ' if the long - term transfer rate differs too much between systems .
it remains to be seen whether a plausible alternative explanation of the gap can be found within the cb scenario .
one may wonder how it is possible that something as serious as circumbinary disks around cvs could have escaped detection so far . on closer inspection , however , it turns out that such disks would in fact be rather difficult to observe .
on the one hand , the luminosity of a cb disk is low .
though it extracts a lot of angular momentum , this happens at a large distance in a shallow gravitational potential , and is thus accompanied by little energy dissipation .
this makes them inconspicuous against the very bright accretion disk , which processes material deep in the potential well of the primary .
secondly , the size of the cb disk is such that its spectral energy distribution peaks in the poorly accessible mid ir ( at 3 - 10@xmath14 m , cf .
the models of dubus et al . , 2001 ) .
finally , model calculations show that the cb disk probably lies in the shadow of the outer part of the accretion disk .
enhanced brightness through illumination of the disk by the bright central regions of the accretion disk is thus also unlikely .
baptista , r. , silveira , c. , catalan , s. , steiner , j.e .
, & horne , k. 1998 , ix latin american regional iau meeting , `` focal points in latin american astronomy '' , eds : aguilar , a. ; carraminana , a. ; revista mexicana de astronomia y astrofisica serie de conferencias .
, e26 patterson , j. 1999 , in disk instabilities in close binary systems , disk instabilities in close binary systems .
, eds . s. mineshige and j.c .
wheeler , frontiers science series no .
universal academy press , p.61 | observations of cataclysmic variables show a number of phenomena that do not fit easily into the standard magnetic braking scenario .
these include the large spread in mass transfer rates , the low surface temperatures of many of the companion stars , and evidence for material at low velocities .
we propose that these anomalies have a common cause : the presence of a circumbinary ( cb ) disk .
this may be a significant component of mass transferring binaries in general .
direct detection of such cb disks may be possible but difficult , because of their low luminosity and spectral energy distribution peaking in the mid - ir .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
one of the central problems of particle physics is to understand how the electroweak scale associated with the mass of the w boson is generated , and why it is so small as compared to the planck scale associated with the newton s constant . in the standard model ( sm ) the electroweak scale is generated through the vacuum expectation value ( vev ) of the neutral component of a higgs doublet @xcite .
apart from the fact that this vev is an arbitrary parameter in the sm , the mass parameter of the higgs field suffers from quadratic divergences , making the weak scale unstable under radiative corrections .
supersymmetry is at present the only known framework in which the weak scale is stable under radiative corrections @xcite , although it does not explain how such a small scale arises in the first place . as such
, considerable importance attaches to the study of supersymmetric models , especially the minimal supersymmetric standard model ( mssm ) , based on the gauge group @xmath1 , with two higgs doublet superfields .
it is well known that , because of underlying gauge invariance and supersymmetry ( susy ) , the lightest higgs boson in mssm has a tree level upper bound of @xmath2 ( the mass of z boson ) on its mass @xcite .
although radiative corrections @xcite to the tree level result can be appreciable , these depend only logarithmically on the susy breaking scale , and are , therefore , under control .
this results in an upper bound of about @xmath3 gev on the one - loop radiatively corrected mass @xcite of the lightest higgs boson in mssm gev as compared to its one - loop value @xcite . ] . because of the presence of the additional trilinear yukawa couplings , such a tight constraint on the mass of the lightest higgs boson need not _ a priori _ hold in extensions of mssm based on the gauge group @xmath1 with an extended higgs sector .
nevertheless , it has been shown that the upper bound on the lightest higgs boson mass in these models depends only on the weak scale , and dimensionless coupling constants ( and only logarithmically on the susy breaking scale ) , and is calculable if all the couplings remain perturbative below some scale @xmath4 @xcite .
this upper bound can vary between @xmath5 gev to @xmath6 gev depending on the higgs structure of the underlying supersymmetric model .
thus , nonobservation of such a light higgs boson below this upper bound will rule out an entire class of supersymmetric models based on the gauge group @xmath7 .
the existence of the upper bound on the lightest higgs boson mass in mssm ( with arbitrary higgs sectors ) has been investigated in a situation where the underlying supersymmetric model respects baryon ( @xmath8 ) and lepton ( @xmath9 ) number conservation .
however , it is well known that gauge invariance , supersymmetry and renormalizibility allow @xmath8 and @xmath9 violating terms in the superpotentioal of the mssm @xcite .
the strength of these lepton and baryon number violating terms is , however , severely limited by phenomenological @xcite , and cosmological @xcite constraints . indeed ,
unless the strength of baryon - number violating term is less than @xmath10 , it will lead to contradiction with the present lower limits on the lifetime of the proton .
the usual strategy to prevent the appearance of @xmath8 and @xmath9 violating couplings in mssm is to invoke a discrete @xmath11 symmetry @xcite known as matter parity , or r - parity .
the matter parity of each superfield may be defined as @xmath12 the multiplicative conservation of matter parity forbids all the renormalizable @xmath8 and @xmath9 violating terms in the superpotential of mssm .
equivalently , the r - parity of any component _ field _ is defined by @xmath13 , where @xmath14 is the spin of the field . since @xmath15 is conserved in any lorentz - invariant interaction , matter parity conservation and r - parity conservation are equivalent .
conservation of r - parity then immediately implies that superpartners can be produced only in pairs , and that the lightest supersymmetric particle ( lsp ) is absolutely stable .
although the minimal supersymmetric standard model with r - parity conservation can provide a description of nature which is consistent with all known observations , the assumption of @xmath16 conservation appears to be _ ad hoc _ , since it is not required for the internal consistency of mssm .
furthermore , all global symmetries , discrete or continuous , could be violated by the planck scale physics effects @xcite .
the problem becomes acute for low energy supersymmetric models , because @xmath8 and @xmath9 are no longer automatic symmetries of the lagrangian , as they are in the standard model .
it is , therefore , more appealing to have a supersymmetric theory where r - parity is related to a gauge symmetry , and its conservation is automatic because of the invariance of the underlying theory under this gauge symmetry .
fortunately , there is a compelling scenario which does provide for exact r - parity conservation due to a deeper principle .
indeed , @xmath16 conservation follows automatically in certain theories with gauged @xmath17 , as is suggested by the fact that matter parity is simply a @xmath11 subgroup of @xmath17 .
it has been noted by several authors @xcite that if the gauge symmetry of mssm is extended to @xmath18 , or @xmath19 , the theory becomes automatically r - parity conserving .
such a left - right supersymmetric theory ( slrm ) solves the problems of explicit @xmath8 and @xmath9 violation of mssm , and has received much attention recently @xcite .
of course left - right symmetric theories are also interesting in their own right , for among other appealing features , they offer a simple and natural explanation for the smallness of neutrino mass through the so called see - saw mechanism @xcite .
such a naturally r - parity conserving theory necessarily involves the extension of the standard model gauge group , and since the extended gauge symmetry has to be broken , it involves a `` new scale '' , the scale of left - right symmetry breaking , beyond the susy and @xmath7 breaking scales of mssm . in @xcite
we showed that in the slrm with minimal particle content the upper bound on the mass of the lightest neutral higgs boson depends only on gauge couplings and those vevs which break the @xmath20 symmetry . here
we will present a detailed analysis of the higgs sector of the left - right supersymmetric models , and consider some of the distinguishing features of the lightest higgs boson in these models . in the slrms
there are typically also other light higgs particles .
the light doubly charged higgs boson , which we will consider in detail as well , should provide a clear signal in experiments .
the plan of the paper is as follows . in section [ sec : tree ]
, we review the various left - right supersymmetric models that we consider in this paper . in section [ sec : higgs ] , we obtain the tree level upper bound on the mass of the lightest cp - even higgs boson for the various slrms considered in section [ sec : tree ]
. we use a general procedure to obtain this ( tree - level ) upper bound on the mass of the lightest cp - even higgs boson in models with extended gauge groups , such as slrms .
we show that the upper bound so obtained in the renormalizable models is independent of the supersymmetry breaking scale , as well as the left - right breaking scale . in the case of models containing non - renormalizable terms , although the upper bound depends on the left - right breaking scale , the dependence is extremely weak , being suppressed by powers of planck mass . in section [ sec : rad ]
we calculate the radiative corrections to the upper bound on the mass of the lightest higgs boson , and show that the most important radiative corrections arising from quark - squark loops are of the same type as in the mssm based on @xmath21 .
the radiatively corrected upper bound so obtained is numerically considerably larger than the corresponding bound in the mssm , but for most of the parameter space is below 200 gev . in section [ sec : decoup ]
we consider the branching ratios of the lightest neutral higgs boson , and find that its couplings are similar to the corresponding higgs couplings in the standard model in the decoupling limit . in section [ sec
: double ] we discuss the mass of the lightest doubly charged higgs boson and the possibility of its detection at colliders . in section [ sec : conclusion ] , we present our conclusions . the full scalar potential of the minimal supersymmetric left - right model is presented in the appendix a.
in this section we briefly review the minimal supersymmetric left - right model , and then discuss models which have an extended higgs sector , and finally discuss models with non - renormalizable interaction terms in the superpotential .
the minimal slrm is based on the gauge group @xmath22 .
the matter fields of this model consist of the three families of quark and lepton chiral superfields with the following transformation properties under the gauge group : @xmath23 where the numbers in the brackets denote the quantum numbers under @xmath22 .
the higgs sector consists of the bidoublet and triplet higgs superfields : @xmath24 there are two bidoublet superfields in order to implement the @xmath7 breaking , and to generate a nontrivial kobayashi - maskawa matrix .
furthermore , two @xmath25 higgs triplet superfields @xmath26 and @xmath27 with opposite @xmath17 are necessary to break the left - right symmetry spontaneously , and to cancel triangle gauge anomalies due to the fermionic superpartners .
the gauge symmetry is supplemented by a discrete left - right symmetry under which the fields can be chosen to transform as @xmath28 thus , the @xmath29 triplets @xmath30 and @xmath31 are needed in order to make the lagrangian fully symmetric under @xmath32 transformation , although these are not needed phenomenologically for symmetry breaking , or the see - saw mechanism . the most general gauge invariant superpotential involving these superfields can be written as ( generation indices suppressed ) @xmath33 the scalar potential can be calculated from @xmath34 where @xmath35 , @xmath36 , and @xmath37 represent the contribution of @xmath38-terms , the @xmath39-terms , and the soft susy breaking terms , respectively . the full scalar potential can be found in appendix a. the general form of the vacuum expectation values of the various scalar fields which preserve the @xmath40 gauge invariance can be written as @xmath41 we note that the triplet vacuum expectation values @xmath42 and @xmath43 represent the scale of @xmath25 breaking and are , according to the lower bounds @xcite on heavy w- and z - boson masses , in the range @xmath44 tev .
these represent a new scale , the right - handed breaking scale .
we note that @xmath45 and @xmath46 contribute to the mixing of the charged gauge bosons and to the flavour changing neutral currents , and are usually assumed to vanish . furthermore , since the electroweak @xmath47 parameter is close to unity , @xmath48 @xcite , the triplet vacuum expectation values @xmath49 and @xmath50 must be small .
the yukawa coupling @xmath51 is proportional to the neutrino dirac mass @xmath52 .
the light neutrino mass in the see - saw mechanism is given by @xmath53 , where @xmath54 is the majorana mass .
the magnitude of @xmath51 is not accurately determined given the present upper limit on the light neutrino masses .
on the other hand the yukawa coupling @xmath55 is proportional to the electron mass and is , thus , small . in the minimal model described above
, parity can not be spontaneously broken at the renormalizable level without spontaneous breaking of @xmath56-parity .
this may be cured by adding more fields to the theory . in @xcite
it was suggested that a parity - odd singlet , coupled appropriately to triplet fields , be introduced so as to ensure proper symmetry breaking .
this leads to a set of degenerate minima connected by a flat direction , all of them breaking parity .
when soft susy breaking terms are switched on , the degeneracy is lifted , but the global minimum that results breaks @xmath40 .
because of the flat direction connecting the minima , there is no hope that the fields remain in the phenomenologically acceptable vacuum , which rolls down to global minimum after susy is softly broken .
the only option that is left is to have a relatively low @xmath25 breaking scale , with spontaneously broken @xmath56-parity ( @xmath57 is non - zero ) .
we note that present experiments allow for a low @xmath25 breaking scale .
there is an alternative to the minimal left - right supersymmetric model which involves the addition of a couple of triplet fields , @xmath58 and @xmath59 , instead of a singlet higgs superfield , to the minimal model @xcite . in these extended models
the breaking of @xmath25 is achieved in two stages . in the first stage
the gauge group @xmath19 is broken to an intermediate symmetry group @xmath60 , and at the second stage @xmath61 is broken to @xmath62 at a lower scale . in this theory there is only one parity - breaking minimum , in contrast to the minimal model , which respects the electromagnetic gauge invariance .
the superpotential for this class of models obtains additional terms involving the triplet fields @xmath63 and @xmath64 : @xmath65 where @xmath66 is the superpotential ( [ eq : superpotential ] ) of the minimal left - right model . in these models
the see - saw mechanism takes its canonical form with @xmath67 , where @xmath52 is the neutrino dirac mass . in this case
the low - energy effective theory is the mssm with unbroken @xmath56-parity , and contains besides the usual mssm states , a triplet of higgs scalars much lighter than the @xmath68 breaking scale .
a second option is to add non - renormalizable terms to the lagrangian of the minimal left - right supersymmetric model , while retaining the minimal higgs content @xcite the superpotential for this class of models can be written as @xmath69 \nonumber \\ & & + { \lambda_{ijkl}\over m}{\rm tr}\,i\tau_2 \phi_i^t i\tau_2 \phi_j{\rm tr}\,i\tau_2 \phi_k^t i\tau_2 \phi_l + { \alpha_{ijl}\over m } { \rm tr}\ , \delta_l \delta_l \phi_i i\tau_2 \phi_j^t i\tau_2 \nonumber\\ & & + { \alpha_{ijr}\over m } { \rm tr}\ , \delta_r \delta_r i\tau_2 \phi_i^t i\tau_2 \phi_j + { 1 \over m}{\rm tr}\,\tau_2 \phi_i^t \tau_2 \phi_j [ \beta_{ijl } { \rm tr}\ , \delta_l \delta_l + \beta_{ijr } { \rm tr}\ , \delta_r \delta_r ] \nonumber \\ & & + { \eta_{ij } \over m } { \rm tr}\,\phi_i \delta_r i\tau_2
\phi_j^t i \tau_2 \delta_l + { \overline \eta_{ij } \over m } { \rm
tr}\,\phi_i \delta_r i\tau_2 \phi_j^t i\tau_2 \delta_l \nonumber \\ & & + { k_{ql } \over m}q^t i\tau_2 l\,q^{ct } i\tau_2 l^c + { k_{qq } \over m}q^t i\tau_2 q\,q^{ct } i\tau_2 q^c + { k_{ll } \over m}l^t i\tau_2 l\,l^{ct } i\tau_2 l^c \nonumber \\ & & + { 1\over m } [ j_l q^t i\tau_2 q\,q^t i\tau_2 l + j_r q^{ct } i\tau_2 q^c \,q^{ct } i\tau_2 l^c ] .
\label{nonsuperpot}\end{aligned}\ ] ] it has been shown that the addition of non - renormalizable terms suppressed by a high scale such as planck mass , @xmath70 gev , with the minimal field content ensures the correct pattern of symmetry breaking in the supersymmetric left - right model .
in particular the scale of parity breakdown is predicted to be in the intermediate region @xmath71 gev , and @xmath56-parity remains exact .
this theory contains singly charged and doubly charged higgs scalars with a mass of order @xmath72 , which may be experimentally accessible .
however , what is different is the nature of see - saw mechanism .
whereas in the renormalizable version the see - saw mechanism takes its canonical form , in the non - renormalizable case it takes a form similar to what occurs in the non - supersymmetric left - right models , with the neutrino mass depending on the unknown parameters of the higgs potential .
this in general leads to different neutrino mass spectra , which can be experimentally distinguished .
given the fact that the higgs sector of slrm models contain a large number of higgs multiplets , and the vevs of some of the higgs fields involve possibly large mass scales compared to the electroweak and susy breaking scales , it is important to ask what is the mass of the lightest higgs boson in these models .
the upper bound on the lightest higgs boson mass in the minimal model was derived in @xcite in the limit when @xmath73 , using the fact that for any hermitean matrix the smallest eigenvalue must be smaller than that of its upper left corner @xmath74 submatrix . in the basis
in which the first two indices correspond to ( @xmath75,@xmath76 ) , we find for matrix elements @xmath77 ( see appendix [ sec:1 ] ) @xmath78 it follows that the upper bound on the lightest higgs boson mass in the minimal supersymmetric left - right model can be written as @xcite : @xmath79 where @xmath80 .
the upper bound ( [ eq : treeupper1 ] ) is not only independent of the supersymmetry breaking parameters ( as in the case of supersymmetric models based on @xmath1 ) , but it is also independent of the @xmath25 breaking scale , which , _ a priori _ , can be large .
the upper bound is controlled by the weak scale vacuum expectation value , @xmath81 , and the dimensionless gauge couplings ( @xmath82 and @xmath83 ) only . since the former is essentially fixed by the electroweak scale , the gauge couplings @xmath82 and @xmath83 determine the bound .
we will see below that even when @xmath84 are non - zero , the upper bound on the lightest higgs mass at the tree level does not depend on either the right - handed breaking scale or the susy breaking scale .
a general method to find an upper limit for the lightest higgs mass in models based on @xmath20 and maximally quartic potentials was presented in @xcite .
we will apply this method to the case of slrms , with possible nonrenormalizable terms , in an appropriate manner .
consider a set of scalar fields @xmath85 transforming under @xmath29 , and define a discrete transformation @xmath86 of these fields , where @xmath87 is the dimension of @xmath29 representation @xcite , but @xmath88 provides the best limit in the case of slrm . ] . by setting the @xmath88-even fields to their vevs
, a normalized field can be defined in the direction of @xmath29 breaking , @xmath89 , where @xmath90 , and all the fields orthogonal to @xmath91 have zero vevs .
since the original lagrangian is invariant under the transformatiom @xmath88 , the potential must be invariant under @xmath88 , so that the potential contains only even powers of @xmath91 , @xmath92 we take into account only the leading nonrenormalizable terms and thus @xmath93 is the largest power in the potential @xmath94 . if @xmath91 were the mass eigenstate , then by using the minimization condition the higgs mass would be @xmath95 . in the general case , this expression provides an upper bound on the mass of the lightest higgs boson , @xmath96 since only the @xmath29 doublet fields are relevant for obtaining this bound , adding extra singlets or triplets in the model has no effect on the bound .
thus , there are three separate cases to be considered for deriving an upper bound on the mass of the lightest higgs boson in the models discussed in the last section , namely : ( a ) @xmath56-parity is broken ( sneutrinos get vevs ) , ( b ) @xmath56-parity is conserved because there are additional triplets , and ( c ) @xmath56-parity is conserved because there are nonrenormalizable terms .
let us first consider the situation in the minimal model , case ( a ) .
we define a new neutral scalar field in the direction of the breaking in the @xmath29 doublet space : @xmath97 where @xmath98 all the @xmath29 doublet fields which are orthogonal to @xmath99 have vanishing vevs .
we calculate next the quartic term @xmath100 in the potential .
it has contributions from both the f - terms and the d - terms , and we find that the upper bound on the mass of the lightest higgs boson in the minimal slrm is @xmath101 , \label{eq : treeupper2}\ ] ] where @xmath102 as is evident , the upper bound ( [ eq : treeupper2 ] ) is independent of susy and right - handed breaking scales , and depends only on the dimensionless gauge and yukawa couplings , and vacuum expectation values which are determined by the weak scale : @xmath103 the triplet vevs @xmath49 and @xmath50 must be small in order to maintain @xmath104 . in the limit when @xmath105 , the bound ( [ eq : treeupper2 ] ) reduces to the upper bound ( [ eq : treeupper1 ] ) .
it is obvious that the addition of extra triplets does not change this bound .
thus , the bound for the case ( b ) , the slrm with additional triplets to ensure that @xmath56-parity is not spontaneously broken , can be obtained from ( [ eq : treeupper2 ] ) by taking the limit @xmath106 .
the total number of nonrenormalizable terms in case ( c ) is rather large .
all the coefficients in nonrenormalizable terms are proportional to inverse powers of a large scale .
thus the largest contribution comes from those terms which have the smallest number of large scales and the largest number of potentially large vevs from @xmath25 triplets .
we recall that the terms of the form @xmath100 and @xmath107 in the potential are needed to determine the contributions to the mass bound .
the nonrenormalizable terms have no effect on the contribution from @xmath39-terms .
the leading terms in the superpotential which can give a @xmath100 and @xmath107 type f - term contribution are of the type @xmath108 i.e. one term with two and another with four bidoublet fields . here
@xmath109 . with @xmath29 singlets
fixed to their vevs , the corresponding @xmath38-terms are @xmath110 the contribution to the higgs mass bound from these nonrenormalizable terms is @xmath111 if the vev @xmath112 gev in these models , the contribution is numerically negligible .
therefore the upper bound for this class of models is essentially the same as in the case ( b ) .
since it is known that the radiative corrections to the lightest higgs mass are significant in the mssm , as well as its extensions based on the @xmath1 , it is important to consider the radiative corrections to the upper bound on the lightest higgs boson mass obtained in the previous section . in this section
we discuss the one - loop radiative corrections to the upper bound on the lightest higgs mass in the minimal supersymmetric left - right model , which was obtained in last section .
we shall use the method of one - loop effective potential @xcite for the calculation of radiative corrections , where the effective potential may be expressed as the sum of the tree - level potential plus a correction coming from the sum of one - loop diagrams with external lines having zero momenta , @xmath113 where @xmath114 is the tree level potential ( [ eq : scalarpotential ] ) evaluated at the appropriate running scale @xmath115 , and @xmath116 is the one loop correction given by @xmath117 where @xmath118 is the field dependent mass eigenvalue of the @xmath119th particle of spin @xmath120 . the dominant contribution to ( [ eq : deltav1 ] ) comes from top - stop ( @xmath121 ) system .
however , under certain conditions the contribution of bottom - sbottom ( @xmath122 ) can be nonneglible
. we shall include both these contributions in our calculations of the radiative corrections . in order to evaluate the contributions of top - stop and bottom - sbottom to ( [ eq : deltav1 ] )
, we need the stop and sbottom mass matrices for the slrm . from ( [ eq : fterms ] ) , ( [ eq : dterms ] ) and ( [ eq : softterms ] ) , it is straightforward to calculate squark mass matrices @xcite . ignoring the interfamily mixing ,
the part of the potential containing the stop and sbottom mass terms can be written as @xmath123 where the mass matrix elements for the stop are @xmath124^*,\nonumber\\ ( \tilde m_u ) _ { u_r^ * u_r } & = & m_{q^c}^2 + m_u^2 + \frac 14 g_r^2(\omega^2_\kappa -2\omega^2_r ) + \frac 16 g_{b - l}^2 \omega^2_r , \end{aligned}\ ] ] while for the sbottom these are @xmath125^*,\nonumber\\ ( \tilde m_d ) _ { d_r^ * d_r } & = & m_{q^c}^2 + m_d^2 -\frac 14 g_r^2(\omega^2_\kappa -2\omega^2_r ) + \frac 16 g_{b - l}^2 \omega^2_r , \end{aligned}\ ] ] where top and sbottom squared masses are given by @xmath126 and @xmath127 , respectively , @xmath128 , @xmath129 , @xmath130 and @xmath131 are soft supersymmetry breaking parameters ( see eq .
( [ eq : softterms ] ) ) , and @xmath132 in order that @xmath133 is unbroken , none of the physical squared masses of squarks can be negative .
necessarily then all the diagonal elements of the squark mass matrices should be non - negative . combining the diagonal elements of the stop and sbottom mass matrices leads to the inequality @xmath134 where we have ignored terms which are of the order of the weak scale or less .
the eigenvalues @xmath135 of the stop and sbottom mass squared matrices are given by ( @xmath136 ) @xmath137 where @xmath138 , \nonumber \\
\delta^2_{\tilde t } & = & \frac 12 \left\ { \left [ m ^2_q- m ^2_{q^c } + \frac 14 ( g_l^2-g_r^2 ) \omega_\kappa^2 + \frac 12 g_r^2 \omega_r^2- \frac 13 g_{b - l}^2 \omega_r^2 \right ] ^2 \right .
\nonumber \\ & & \left .
+ 4 \left [ h_{\chi q } a_t \kappa_2 - h_{\chi q}\tilde\mu_t \kappa_1 \right]^2 \right\ } ^{\frac 12 } , \label{eq : squark2}\end{aligned}\ ] ] and @xmath139 , \nonumber \\
\delta^2_{\tilde b } & = & \frac 12 \left\ { \left [ m ^2_q- m ^2_{q^c } - \frac 14 ( g_l^2-g_r^2 ) \omega_\kappa^2 - \frac 12 g_r^2 \omega_r^2- \frac 13 g_{b - l}^2 \omega_r^2 \right ] ^2 \right .
\nonumber \\ & & \left .
+ 4 \left [ h_{\phi q } a_b \kappa_1 - h_{\phi q}\tilde\mu_b \kappa_2 \right]^2 \right\ } ^{\frac 12 } .
\label{eq : squark3}\end{aligned}\ ] ] here we have defined @xmath140 using eqs .
( [ eq : squark1 ] ) , ( [ eq : squark2 ] ) and ( [ eq : squark3 ] ) in ( [ eq : deltav1 ] ) , we have calculated the radiatively - corrected expressions for the matrix elements of the upper left corner @xmath74 submatrix of the @xmath141 cp - even higgs mass matrix . after imposing the appropriate one - loop minimization conditions , we find the following form for the radiatively corrected upper left corner @xmath74 submatrix of cp - even higgs mass matrix : @xmath142 where @xmath143 , \label{eq : radelementsa}\end{aligned}\ ] ] @xmath144 and @xmath145 \nonumber \\ & & + \frac{m_b^4}{\cos^2 \beta } \frac{a_b^2 \left ( a_b - \tilde \mu_b \tan \beta \right ) ^2 } { \left ( m^2_{\tilde b _ 1 } - m^2_{\tilde b _ 2 } \right ) ^2 } g ( m^2_{\tilde b _ 1 } , m^2_{\tilde b _ 2 } ) + \frac{m_t^4}{\sin^2 \beta } \frac{\left ( a_t - \tilde\mu_t \cot \beta \right ) ^2 \tilde\mu_t ^2 } { ( m^2_{\tilde t _ 1 } - m^2_{\tilde t _
2})^2 } g ( m^2_{\tilde t _ 1 } , m^2_{\tilde t _ 2 } ) , \nonumber\\ \label{eq : radelementsb}\end{aligned}\ ] ] @xmath146 \nonumber \\ & & + \frac{m_t^4}{\sin^2 \beta } \frac{a_t^2 \left ( a_t - \tilde \mu_t \cot \beta \right ) ^2}{\left ( m^2_{\tilde t _
1}-m^2_{\tilde t _ 2 } \right ) ^2 } g ( m^2_{\tilde t _
1},m^2_{\tilde t _ 2 } ) + \frac{m_b^4}{\cos ^2 \beta } \frac { \left ( a_b - \tilde \mu_b \tan \beta \right ) ^2 \tilde\mu_b ^2 } { \left ( m^2_{\tilde b
_ 1}- m^2_{\tilde b _ 2 } \right ) ^2 } g(m^2_{\tilde b _ 1 } , m^2_{\tilde b _ 2 } ) , \nonumber\\ \label{eq : radelementsc}\end{aligned}\ ] ] @xmath147 \nonumber \\ & & + \frac{m_b^4}{\cos^2 \beta } \frac { \left ( a_b - \tilde\mu_b \tan \beta \right ) \left ( -\tilde\mu_b \right ) } { m^2_{\tilde b _ 1}-m^2_{\tilde b _ 2 } } \times \left [ \ln \left ( \frac{m^2_{\tilde b _ 1}}{m^2_{\tilde b _ 2 } } \right ) + \frac{a_b \left ( a_b- \tilde\mu_b \tan \beta \right ) } { m^2_{\tilde b _ 1}-m^2_{\tilde b _ 2 } } g(m^2_{\tilde b _ 1},m^2_{\tilde b _ 2 } ) \right ] , \nonumber\\ \label{eq : radelementsd}\end{aligned}\ ] ] with @xmath148 we have neglected d - terms in the squark masses , because these are small , and , since we are including only the quark - squark contributions to @xmath116 , in order to gain approximate independence of the renormalization scale @xmath115 ( see also the inequality ( [ eq : susyineq ] ) ) . using eqs .
( [ eq : squarkmass ] ) and ( [ eq : radelements ] ) , we obtain the one - loop radiatively corrected upper bound on the lightest higgs boson mass in the slrm : @xmath149\nonumber\\ \label{eq : upperbound1}\end{aligned}\ ] ] for @xmath150 , one can neglect the b - quark contribution in the radiative corrections .
then , in the approximation @xcite @xmath151 the upper bound ( [ eq : upperbound1 ] ) on the lightest higgs mass reduces to @xmath152 \label{eq : upperbound2}\end{aligned}\ ] ] where @xmath153 , and @xmath154 is the supersymmetry breaking scale ( @xmath155 ) . in this limit
the upper bound eq .
( [ eq : upperbound2 ] ) on the lightest higgs mass in the supersymmetric left - right model differs in form from the corresponding mssm upper bound only because of @xmath156 being nonzero .
the upper bound is maximised by @xmath157 for a given value of @xmath156 .
the radiatively corrected upper bound ( [ eq : upperbound1 ] ) on the mass of the lightest higgs boson is plotted in fig.[mh1 ] as a function of the large scale @xmath158 up to which the supersymmetric left - right model remains perturbative .
the upper bound comes from the requirement that all the gauge couplings of the slrm remain perturbative below the scale @xmath158 .
we have taken into account the dominant radiative corrections coming from the quark and squark loops in our calculations . in fig.[mh1 ] we have taken two values of @xmath159 and @xmath160 . in the figure the upper bound is shown for two different values of the @xmath25 breaking scale , @xmath161 tev and @xmath162 gev , respectively , and for two values of soft supersymmetry breaking mass parameter , @xmath163 tev and @xmath164 tev .
it is seen from this figure that if the difference between the @xmath25 breaking scale and the large scale @xmath4 is more than two orders of magnitude , the radiatively corrected upper bound on the mass of the lightest higgs boson remains below 250 gev . for large values of @xmath158
the upper bound is below 200 gev . the upper bound increases with increasing @xmath165 and with increasing soft supersymmetry breaking parameters .
it is considerably larger than the corresponding upper bound in the mssm .
in order to study the phenomenology of the lightest higgs boson in the slrm , we must obtain its couplings to the fermions and see how these differ from the corresponding mssm couplings .
the major difference between the two models arises due to the triplet higgs couplings to leptons .
as discussed earlier , the lightest neutral higgs is composed mainly of the bidoublet fields , and we do not expect the triplet couplings to have a large effect on the lightest neutral higgs branching ratios . another difference in the higgs couplings in the slrm and the mssm arises because of the mixing between the higgs(ino ) and lepton sectors in models where @xmath56-parity is spontaneously broken . since the lightest chargino contains gaugino and higgsino admixture , one might expect that in some region of the parameter space the couplings are significantly different from the couplings in the mssm .
we will take the lightest chargino to correspond to the @xmath166-lepton .
it will be denoted by @xmath167 in the following .
thus , it is important to study in detail the couplings of the lightest higgs boson to the @xmath167 . in figure [ tau2 ]
we have plotted the branching ratio of the lightest neutral higgs boson to @xmath168 , @xmath169 , and @xmath170 pair both in the mssm and in the slrm for large @xmath171 and a tree - level mass @xmath172 gev as a function of the second lightest neutral higgs boson mass .
we note that for the second lightest neutral higgs with mass @xmath173 gev , the branching ratio to bottom pair almost vanishes while the branching ratio to a @xmath167 pair is enhanced .
this behaviour occurs in a transition region where the composition of the light higgs bosons change rapidly , and the lightest higgs coupling to @xmath168 becomes negligible .
similar phenomenon occurs for small @xmath171 .
it is well known that in the left - right symmetric models problems with fcnc are expected if several light higgs bosons exist @xcite . in the supersymmetric case , the problem could be expected to be more severe , since the potential is more constrained .
furthermore , in the four higgs - doublet model @xcite , which is also the number of doublets in the supersymmetric left - right model , it was found that the potential is always real , thus preventing spontaneous violation of cp .
strong limits on the fcnc higgs were obtained in @xcite assuming that the only source for cp violation is the complex phase in the kobayashi - maskawa matrix and that the model is manifestly left - right symmetric . if these conditions are relaxed , for example by defining the symmetry transformation by eq .
( 4 ) , a weakened lower limit for second lightest neutral higgs is obtained from the neutral meson mass difference .
the limit is less severe for large values of @xmath171 , but remains of the same order than in the nonsupersymmetric left - right model .
taking into account the uncertainties in calculating the mass difference , @xmath174 tev ) .
thus the relevant limit to discuss is the one in which all the neutral higgs bosons , except the lightest one , are heavy . in order to consider in detail the neutral higgs couplings to the @xmath167-leptons ,
we ll need the interaction lagrangian .
we have seen that at the tree level the lightest higgs can be written as @xmath175 where @xmath176 are the scalar bidoublet fields and @xmath177 are all the other fields . in the decoupling limit @xmath178
, the relevant higgs - fermion interaction lagrangian can be written as @xmath179 where @xmath180 and @xmath181 .
the mass eigenstates @xmath182 are found by diagonalizing the mass matrices by unitary matrices @xmath183 @xcite , @xmath184 the diagonal chargino mass matrix is @xmath185 we decompose the mass matrix @xmath186 and the diagonalizing matrices @xmath187 and @xmath188 as @xmath189 where the contribution of the terms proportional to @xmath190 in determining the @xmath166 lepton eigenstate is small .
one can solve for the matrix elements to the order @xmath191 , @xmath192_{jk } } { ( m_0 ^ 2)_{jj}-(m_0 ^ 2)_{kk } } , \;j\neq k \end{array},\right . \\ z_{jk}&=&\left\{\begin{array}{l}0,\;j = k,\\ \frac{[u_0^*(x_0x_1^\dagger+x_1x_0^\dagger ) u_0^t]_{jk } } { ( m_0 ^ 2)_{jj}-(m_0 ^ 2)_{kk } } , \;j\neq k \end{array},\right . \end{aligned}\ ] ] where we have denoted @xmath193 . in terms of the mass eigenstates , the interaction lagrangian can be written as @xmath194 here we denote @xmath195 , where @xmath196 is the coupling matrix for charginos .
thus the diagonal couplings to @xmath167 are given by @xmath197 the chargino mass matrix can be written as @xmath198 , where @xmath199 and @xmath200 we have here assumed for simplicity that @xmath201 , and @xmath202 .
the matrix @xmath203 has one zero eigenvalue corresponding to the @xmath166 mass , @xmath204=0 .
the @xmath167 mass is @xmath205 for the sm one has @xmath206 .
thus , in the decoupling limit , we obtain the following result for the ratio of the yukawa couplings @xmath207 and @xmath208 : @xmath209 since the matrix @xmath210 is diagonal . even if the @xmath167 s contained large fraction of gauginos or higgsinos , the couplings responsible for higgs decays to the charged leptons or quarks are the same as in the standard model , since the physically relevant parameter region is close to the decoupling limit .
contrary to the mssm , one might have the lightest higgs decaying to a @xmath167-lepton and a heavier chargino . whether this decay mode is kinematically possible
is much more model dependent than the decay to leptons , and we can not say anything general about it .
we have seen that the neutral higgs sector contains one relatively light higgs boson , which however may be heavier as compared to the lightest higgs in the mssm , and which has mssm like couplings to the fermions of the model .
thus , one can not tell from the properties of the lightest higgs boson , about the nature of the model . if a light higgs is found before any supersymmetric particles
are observed , one would not even know whether it is the higgs boson of a supersymmetric model .
altogether there are four doubly charged higgs bosons in the slrm , of which two are right - handed and two left - handed .
the mass matrices of the left - handed triplets depend on the soft susy breaking parameters , left - triplet vevs , and the weak scale .
thus , their masses are expected to be of the same order as the soft terms .
the mass matrix for the right - handed scalars depends on the right - triplet vev instead of the left - triplet vev .
nevertheless , it was noticed in @xcite that in the slrm with broken r - parity the lightest doubly charged scalar tends to be light . also , it was shown in @xcite that in the nonrenormalizable case it is possible to have light doubly charged higgs bosons . on the other hand , in the nonsupersymmetric left - right model all the doubly charged scalars tend to have a mass of the order of the right - handed scale @xcite .
this is also true in the slrm with enlarged particle content @xcite .
thus a light doubly charged higgs would be a strong indication of a supersymmetric left - right model with minimal particle content . in the case of broken r - parity
, the mass matrix of the right - handed doubly charged scalars can be obtained from the potential in appendix a. the allowed parameter space is strongly constrained by demanding that all the eigenvalues of the mass squared scalar matrices remain positive . in order to find the relevant parameter region , we have studied the bounds on some of the masses in the model . although these may not be important as actual bounds , they restrict the parameter space . in the following we shall assume that @xmath211 . from the bound obtained from submatrix @xmath212 , we find the following constraint @xmath213 where @xmath214 . if the terms with bidoublet lepton yukawa couplings are ignored , and @xmath215 , eq . ( [ constr1 ] ) indicates that @xmath216 and thus @xmath217 .
other useful constraints follow from @xmath218 and @xmath219 mass matrices , @xmath220 we have studied in figure [ mpp_ad ] an example with the soft masses and right - handed breaking scale , as well as the @xmath221 parameter , of the order of 10 tev .
the maximum majorana yukawa coupling allowed by positivity of the mass eigenvalues in this case is @xmath222 .
for @xmath223 we have plotted the allowed doubly charged higgs mass @xmath224 as a function of @xmath225 for fixed @xmath226 gev@xmath227 tev .
it is seen that relatively narrow bands of @xmath228 and @xmath229 are allowed . even in the maximal case the mass of the doubly charged scalar @xmath230 tev and from the figure we see that the lightest of the doubly charged scalars can be as light or even lighter than the lightest neutral higgs boson .
the mass of the dangerous flavour changing higgs boson is @xmath231 .
the mass of the lightest doubly charged higgs , @xmath232 , in the case of nonrenormalizable couplings has been considered in @xcite . a general statement in these works is that the mass is @xmath233 , where @xmath234 is the scale of the nonrenormalizable terms .
relevant constraints on the parameters follow again from the positivity of the mass bounds . from the submatrix @xmath235 ,
a bound on @xmath236 can be obtained and we get @xmath237 this gives a condition @xmath238 . from the bound on @xmath239 from @xmath240 submatrix
, we find @xmath241\end{aligned}\ ] ] thus the @xmath39-term , @xmath242 has to be negative . from the mass bound on the doubly charged higgses , we get @xmath243/ ( v_{\delta_r}^2+v_{\delta_r}^2 ) + \frac 1 m 8b_rv_{\delta_r}v_{\delta_r}\mu_{2r } + \frac1{m^2}4b_r(2a_r+b_r)v_{\delta_r}^2v_{\delta_r}^2.\nonumber\\ \label{mppbound}\end{aligned}\ ] ] we see that the @xmath244-parameter must necessarily be nonvanishing . in fig .
[ nonren ] we have plotted the mass of the lightest doubly charged higgs boson as a function of the nonrenormalizable @xmath244-parameter for two values of the ratios , @xmath245 gev and @xmath246 gev .
one should note that at low energies @xmath244 is the only nonrenormalizable parameter , which is necessarily nonvanishing .
in particular one can find solutions with @xmath247 . in appendix
b we give examples of mass spectra using the tree level potential for the renormalizable and nonrenormalizable models .
one can expect that radiative corrections will change the spectrum somewhat , but qualitative features should remain unchanged .
the value of @xmath244 needed to produce a proper minimum depends on the other parameters in the model . in fig .
[ nonren ] a ) @xmath246 gev and @xmath248 is the same as the soft supersymmetry breaking parameters .
the mass of the flavour changing higgs is roughly proportional to the value of @xmath248 . for larger values of the @xmath39-term ,
larger @xmath244 s are needed to retain positive doubly charged higgs mass , as seen from eq .
( [ mppbound ] ) .
it is seen that the mass of the doubly charged higgs rises rapidly with @xmath244 to multi - tev region as is expected for large @xmath249 . in fig.[nonren ] b ) @xmath245 gev , and @xmath250 tev ( solid lines ) for all the values of the soft supersymmetry breaking parameters that are plotted .
this shows the dependence on the soft mass parameters . for the soft mass parameter of 10 tev we have plotted for comparison also the dependence on @xmath244 when @xmath251 tev ( dashed line ) .
it is seen that increasing the soft parameter , while the @xmath39-term remains constant , smaller @xmath244 s are needed to produce positive masses . in fig .
( [ nonren ] ) we have taken @xmath171 to be 50
. large @xmath171 somewhat increases the masses of the neutral and pseudoscalar higgses and decreases the mass of the lightest doubly charged higgs . with both values of @xmath249
it is possible to obtain a doubly charged higgs , which is light enough to be detected in future experiments , but @xmath252 increases with @xmath244 much faster for large @xmath249 . the collider phenomenology of the doubly charged scalars has been actively studied , since they appear in several extensions of the standard model , can be relatively light and have clear signatures . here
we will discuss the production of the doubly charged scalar in colliders @xcite .
if the light doubly charged higgs @xmath253 is lighter than any of the supersymmetric partners , the dominant decay mode of @xmath253 is to like sign leptons .
kinematically , production of a single doubly charged scalar would be favoured .
if the couplings between electrons and triplet higgses are large enough , the doubly charged particles can be detected in @xmath254 linear colliders in single production almost up to the kinematical limit .
the @xmath255 colliders are especially useful in studying the doubly charged higgs , since for nonzero triplet couplings it can be produced as an s - channel resonance .
the possibilities for single production in hadron colliders in @xmath256 fusion does not depend on the coupling to leptons , but it does depend , in addition to @xmath257 , on the vev of the triplet to which the doubly charged scalar belongs . if @xmath258 tev , one can detect doubly charged higgs bosons with @xmath259 tev at the lhc @xcite .
the advantage of pair production of the doubly charged scalars compared to the single production is that it is relatively model independent .
it can occur even if @xmath260 is very heavy , as in the nonrenormalizable case , or the triplet yukawa couplings are very small .
the doubly charged higgses can be produced in @xmath261 both at lepton and hadron colliders , if kinematically allowed . at lhc
this cross section falls off rapidly close to @xmath262 gev @xcite .
the pair production cross section at tevatron and at lep ii are given in @xcite .
for @xmath254 linear colliders the pair production cross section for @xmath263 gev is @xmath264 fb for @xmath265 gev .
thus the detection of doubly charged scalars in pair production is possible close to the kinematical limit .
supersymmetric left - right models are well motivated extensions of the mssm , since they conserve @xmath56-parity as a consequence of gauge invariance .
we have made a detailed study of the higgs sector in these models .
the lightest cp even higgs boson can be considerably heavier than in the mssm , but its couplings to fermions remain similar to the couplings of the standard model higgs boson . if a higgs , which nevertheless is too heavy to be the higgs boson of the mssm , is detected , one should consider extended supersymmetric models such as the ones studied in this paper . in the slrm with the minimal particle content one has typically also a light doubly charged higgs boson . if this particle is found , it is a strong indication of the slrm with minimal particle content .
we wish to emphasize the importance of studying the full mass matrices , including the soft mass parameters in determining the low energy mass spectrum , especially the masses of the light scalar particles .
one of us ( pnp ) would like to thank the helsinki institute of physics , where this work was started , and to desy , where it was completed , for hospitality .
the work of pnp is supported by the university grants commission , india under the project 10 - 26/98(sr - i ) .
the work of kh and kp is partially supported by the academy of finland ( no .
the different components of the scalar potential ( [ eq : scalarpotential ] ) for the minimal left - right supersymmetric model can be written as follows ( @xmath82 , @xmath83 and @xmath266 are the three gauge couplings ) : @xmath267 @xmath268 ^ 2 + \frac 18 g_r^2\sum_a \left [ -\rm{tr } ( \phi\tau_a\phi^\dagger ) - \rm{tr } ( \chi\tau_a\chi^\dagger ) \right .
\nonumber\\ & & \left .
+ 2\rm{tr } ( \delta_r^\dagger\tau_a\delta_r ) + 2\rm{tr } ( \delta_r^\dagger\tau_a\delta_r ) + l^{c\dagger}\tau_a l^c+q^{c\dagger}\tau_a q^c \right]^2 \nonumber\\ & & + \frac 18 g_{b - l}^2\left [ 2\rm{tr } ( -\delta_r^\dagger\delta_r+ \delta_r^\dagger\delta_r-\delta_l^\dagger\delta_l+ \delta_l^\dagger\delta_l)\right .
\nonumber\\ & & \left .
-l^\dagger l+l^{c\dagger } l^c + \frac 13 q^\dagger q-\frac 13 q^{c\dagger } q^c \right]^2 , \label{eq : dterms}\end{aligned}\ ] ] @xmath269 from ( [ eq : scalarpotential ] ) , ( [ eq : fterms ] ) , ( [ eq : dterms ] ) and ( [ eq : softterms ] ) it is straightforward to derive the mass matrix for the cp - even higgs scalars , whose eigenvalues will provide the masses of the physical scalar higgs bosons .
.mass spectrum of a renormalizable model with minimum particle content.[table1 ] [ cols="<,<,<",options="header " , ] here we give one example each of the mass spectrum of higgs bosons in renormalizable and nonrenormalizable suoersymmetric left - right models . in table
[ table1 ] we have shown the spectrum in the renormalizable model , and in table [ table2 ] the spectrum for the nonrenormalizable model with minimal particle content . in both examples the soft supersymmetry breaking parameters are @xmath270 tev . in table
[ table1 ] , the right handed scale is 10 tev , and in table [ table2 ] @xmath271 gev . we have denoted @xmath272 and @xmath273 . for reviews of supersymmetry see h.e . haber and g.l .
kane , phys .
( 1985 ) 75 ; h.p .
nilles , phys .
( 1984 ) 1 ; p. nath , r. arnowitt and a.h .
chamseddine , `` applied n=1 supergravity '' , world scientific , singapore ( 1984 ) .
y. okada , m. yamaguchi and t. yanagida , prog .
85 ( 1991 ) 1 ; j. ellis , g. ridolfi and f. zwirner , phys .
b257 ( 1991 ) 83 ; h.e .
haber and r. hempfling , phys .
66 ( 1991 ) 1815 ; r. barbieri , m. frigeni and f. caravaglios , phys .
b258 ( 1991 ) 167 .
r. hempfling and a hoang , phys .
b 331 ( 1994 ) 99 ; m. carena , j. espinosa , m. quiros , c. wagner , phys .
b 355 ( 1995 ) 209 ; h. haber , r. hempfling and a. hoang , z. phys .
c 75 ( 1997 ) 539 ; s. heinemeyer , w. hollik and g. weiglein , phys .
d 58 ( 1998 ) 091701 ; hep - ph/9812472 .
f. zwirner , phys .
lett . 132b ( 1983 ) 103 ; l.j . hall and m.
suzuki , nucl .
b231 ( 1984 ) 419 ; i.h .
lee , nucl .
b246 ( 1984 ) 120 ; s. dawson , nucl .
b261 ( 1985 ) 297 ; r. barbieri and a. masiero , nucl .
b267 ( 1986 ) 679 ; s. dimopoulos and l.j .
hall , phys .
lett . 207b ( 1988 ) 210 ; v. barger , g.f .
giudice and t. han , phys .
d40 ( 1989 ) 2987 ; h. dreiner and g.g .
ross , nucl .
b365 ( 1991 ) 597 .
giddings and a. strominger , nucl .
b307 ( 1988 ) 854 ; s. coleman , nucl .
b310 ( 1988 ) 643 ; j. preskill and l.m .
krauss , nucl .
b341 ( 1990 ) 50 ; r. holman et al .
b282 ( 1992 ) 132 ; m. kamionkowski and j. march - russell , phys .
b282 ( 1992 ) 137 .
m. gell - mann , p. ramond and r. slansky , in `` supergravity '' , eds .
p. van niewenhuizen and d.z .
freedman ( north holland , 1979 ) ( 1979 ) ; t. yanagida , proceedings of `` workshop on unified theory and baryon number in the universe '' , eds . o.sawada and a. sugamoto ( kek 1979 ) .
rizzo , phys.rev.d25 ( 1982 ) 1355 , phys.rev.d27 ( 1983 ) 657;m .
lusignoli , s. petrarca , phys .
b226 ( 1989 ) 397 ; m.d .
swartz , phys .
d40 ( 1989 ) 1521 ; j.a .
grifols , a. mendez , g.a .
schuler , mod .
lett . a4 ( 1989 ) 1485 ; n. lepore , b. thorndyke , h. nadeau , d. london , phys .
d50 ( 1994 ) 2031 ; e. accomando , s. petrarca , phys
b323 ( 1994 ) 212 ; g. barenboim , k. huitu , j. maalampi , m. raidal , phys .
b 394 ( 1997 ) 132 ; s. chakrabarti , d. choudhuri , r.m .
godbole , b. mukhopadhyaya , phys .
lett . b434
( 1998 ) 347 ; f. cuypers , m. raidal , nucl . phys .
b501 ( 1997 ) 3 ; m. raidal , phys . rev . d57 ( 1998 ) 2013 ; j.f .
gunion , int.j.mod.phys .
a13 ( 1998 ) 2277 . | we carry out a detailed analysis of the light higgs bosons in supersymmetric left - right models ( slrm ) .
this includes models with minimal particle content and those with additional higgs superfields .
we also consider models with non - renormalizable higher - dimensional terms .
we obtain an upper bound on the mass of the lightest @xmath0-even neutral higgs boson in these models .
the upper bound depends only on the gauge couplings , and the vacuum expectation values of those neutral higgs fields which control the spontaneous breakdown of the @xmath1 gauge symmetry .
we calculate the one - loop radiative corrections to this upper bound , and evaluate it numerically in the minimal version of the supersymmetric left - right model .
we consider the couplings of this lightest @xmath0-even higgs boson to the fermions , and show that in a phenomenologically viable model the branching ratios are similar to the corresponding branching ratios in the minimal supersymmetric standard model ( mssm ) .
we then study the most promising particle for distinguishing the slrm from other models , namely the doubly charged higgs boson .
we obtain the mass of this doubly charged higgs boson in different types of supersymmetric left - right models , and discuss its phenomenology .
hip-1999 - 16/th supersymmetry ; left - right symmetry ; higgs boson ; doubly charged higgs boson |
* use modern radiative transfer models of the earth s atmosphere for astronomical applications . * two uses : * * site testing for water vapor content . * * elimination of telluric lines from optical spectra without the need for contemporary telluric standards .
@xmath0the radiative transfer model ( rtm ) and the retrieval algorithm , incorporated in the sciatran 2.2 software package developed at the institute of remote sensing / institute of enviromental physics of bremen university ( germany ) , allows to simulate , among other things , radiance / irradiance spectra in the 2400 - 24000 range . in this work
we present the applications of rtm to two case studies . in the first case
the rtm was used to simulate direct solar irradiance spectra , with different water vapor amounts , for the study of the water vapor content in the sierra nevada observatory ( osn ) atmosphere .
simulated spectra were compared with those measured with a spectrometer operating in the 8000 - 10000 range . in the second case the rtm
was used to generate a telluric model to subtract the atmospheric contribution and correct high - resolution stellar spectra from atmospheric water vapor and molecular oxygen lines .
the results of both cases are discussed here .
@xmath0the osn is the southernmost high altitude ( 2900 m a.s.l . )
location in continental europe .
it is located in sierra nevada , granada , spain ( 37.06 n , 3.38 w ) , less than one km away from the iram 30 m telescope . given the dry climatic conditions it can be considered a very competitive location for mir - submm astronomical observations .
we obtained direct solar irradiance spectra with an array spectrometer in the 8000 - 10000 range ( 2.5 spectral resolution ) , during the 2007 - 2009 campaign , for a total of 248 days .
the wyoming university bi - daily radiosounding data were used to provide profiles of temperature ( t ) , pressure ( p ) and h@xmath1o volumetric mixing ratio ( vmr ) as a function of the geopotential height .
the 2003 - 2006 statistics of the radio balloon measurements from granada ( proxy gibraltar ) , canarias ( proxy gimar ) , and hawaii ( proxy hilo ) shows that osn is a very competitive site ( figs . 1 and 2 ) .
the rtm was used to simulate the direct solar irradiance spectra at osn and to compare them with those measured with the spectrometer . the equivalent width ( ew ) of measured spectra in the same water vapor absorption band
were also computed . o telluric lines in the h@xmath2 region .
data obtained with hrs / het at @xmath3 .
the red vertical bars mark the location of the strongest telluric lines in this wavelength range . ]
we used only the spectra observed @xmath41 hour from the gibraltar noon radiosounding ( 12:00 z ) . for each day , the profiles of p , t and h@xmath1o vmr provided by the radiosounding were used to define the model atmosphere in the rtm .
for each simulated spectrum the ew in the 9270 - 9700 water vapor absorption band was computed , and the corresponding pwv above osn was associated to it .
this provides the calibration shown in fig .
1 ( bottom ) , where blue dots are the individual models for 2007 - 2009 campaign , and the green line is a second order polynomial fit that goes through the `` origin '' ew .
the `` origin '' ew is measured for an atmosphere of zero h@xmath1o column and contributed by the other residual gases ( see the low ew pink line in the spectrum above ) .
notice that we have measured very dry epochs with pwv @xmath5 0.2 mm h@xmath1o .
we are carrying out an observing campaign to quantify in more detail the statistics of very dry epochs , pwv @xmath5 1 mm h@xmath1o , that are certainly not uncommon in sierra nevada .
o and o@xmath1 telluric lines in the 6300 region .
data obtained with feros at @xmath6 .
the vertical bars mark the location of the weak h@xmath1o ( cyan ) and strong o@xmath1 ( magenta ) telluric lines .
the horizontal orange bars indicate the location and approximate widths of the diffuse interstellar bands ( dibs ) in this region .
note that the model does not provide an accurate fit for the shortest wavelength ( 6277 ) telluric line . ]
* we generate two reference models for 3050 - 10000 . * * obtained for the sierra nevada observatory ( @xmath7 = 2900 m , @xmath8 = 151 mbar ) but applicable anywhere after rescaling . * * only contribution from lines in the output at very high ( @xmath9 ) spectral resolution . * * one model without h@xmath1o ( `` oxygen model '' ) . * * one model with 2.78 mm of h@xmath1o ( `` oxygen + water model '' ) .
* idl code to fit the telluric spectrum without contemporary telluric standards ( figs . 3 to 6 ) .
* * two ( o@xmath1 + h@xmath1o ) times three ( column , velocity , spectral resolution ) free parameters . * * possibility of defining different wavelength ranges to account for resolution differences in the spectrum . *
* tested with @xmath10 spectra from hermes ( mercator , la palma ) , fies ( not , la palma ) , feros ( 2.2 m , la silla ) , and hrs ( het , mcdonald ) .
o telluric lines in the region of the nai d1+d2 interstellar doublet .
data obtained with hrs / het at @xmath3 .
the red vertical bars mark the location of the strongest telluric lines in this wavelength range .
note the humid conditions under which the spectrum was obtained . ]
o telluric lines in the region of the heii @xmath118237 stellar absorption line .
this heii line is the best identifier for o stars in the 5500 - 10000 range and is located in the middle of a strong h@xmath1o telluric band .
data obtained with feros at @xmath6 .
the red vertical bars mark the location of the strongest telluric lines in this wavelength range . ] | the radiative transfer model ( rtm ) and the retrieval algorithm , incorporated in the sciatran 2.2 software package developed at the institute of remote sensing / institute of enviromental physics of bremen university ( germany ) , allows to simulate , among other things , radiance / irradiance spectra in the 2400 - 24000 range . in this work we present applications of rtm to two case studies . in the first case
the rtm was used to simulate direct solar irradiance spectra , with different water vapor amounts , for the study of the water vapor content in the atmosphere above sierra nevada observatory .
simulated spectra were compared with those measured with a spectrometer operating in the 8000 - 10000 range . in the second case
the rtm was used to generate telluric model spectra to subtract the atmospheric contribution and correct high - resolution stellar spectra from atmospheric water vapor and oxygen lines .
the results of both studies are discussed . |
by this time , the @xmath4 and @xmath5 antitriplet charmed baryon states ( @xmath6 , @xmath7 , @xmath8 and ( @xmath9 , @xmath10 , @xmath11 , and the @xmath4 and @xmath12 sextet charmed baryon states ( @xmath13 ) and ( @xmath14 ) have been observed , while the @xmath15-wave bottom baryon states are far from complete , only the @xmath16 , @xmath17 , @xmath18 , @xmath19 , @xmath20 , @xmath21 , @xmath22 , @xmath23 and @xmath24 have been observed @xcite . in 2002 ,
the selex collaboration reported the first observation of the doubly charmed baryon state @xmath25 in the decay @xmath26 @xcite , and confirmed it later in the decay @xmath27 @xcite .
however , the babar collaboration observed no evidence for the @xmath28 in the @xmath29 , @xmath30 decay modes and for the @xmath31 in the @xmath32 , @xmath33 decay modes , and the belle collaboration observed no evidence for the @xmath28 in the @xmath29 decay mode @xcite .
there are no experimental signals for the triply heavy baryon states , we expect that the large hadron collider ( lhc ) will provide us with the whole heavy , doubly heavy and triply heavy baryon states @xcite .
the triply heavy baryon states and heavy quarkonium states play an important role in understanding the heavy quark dynamics at the hadronic scale due to the absence of the light quark contaminations , and serve as an excellent subject in studying the interplay between the perturbative and nonperturbative qcd . on the other hand ,
the qcd sum rules is a powerful nonperturbative theoretical tool in studying the ground state hadrons @xcite . in the qcd sum rules ,
the operator product expansion is used to expand the time - ordered currents into a series of quark and gluon condensates which parameterize the long distance properties .
taking the quark - hadron duality , we can obtain copious information about the hadronic parameters at the phenomenological side @xcite .
there have been many works on the masses of the heavy and doubly heavy baryon states with the qcd sum rules @xcite .
it is interesting to study the mass spectrum of the triply heavy baryon states using the qcd sum rules . in ref.@xcite , we take the novel approach introduced by jido et al @xcite to study the positive - parity and negative - parity heavy and doubly heavy baryons in a systematic way by separating the contributions of the positive - parity and negative - parity baryons explicitly , as the interpolating currents have non - vanishing couplings to both the positive - parity and negative - parity baryons , there exist contaminations @xcite . before the work of jido et al , bagan et al take the heavy quark limit to separate the contributions of the positive - parity and negative - parity heavy baryons unambiguously @xcite . in this article , we extend our previous works to study the @xmath0 and @xmath34 triply heavy baryon states by subtracting the contributions from the corresponding @xmath2 and @xmath35 triply heavy baryon states with the full qcd sum rules .
the existing theoretical works focus on the heavy and doubly heavy baryon states , the works on the triply heavy baryon states are relatively few , for example , the effective field theory @xcite , the lattice qcd @xcite , the qcd bag model @xcite , the quark model estimation @xcite , the variational approach @xcite , the modified bag model @xcite , the relativistic three - quark model @xcite , the qcd sum rules @xcite , the non - relativistic three - quark model @xcite , potential non - relativistic qcd @xcite , the regge trajectory ansatz @xcite , etc .
the article is arranged as follows : we derive the qcd sum rules for the masses and the pole residues of the triply heavy baryon states in sect.2 ; in sect.3 , we present the numerical results and discussions ; and sect.4 is reserved for our conclusions .
the @xmath4 and @xmath12 triply heavy baryon states @xmath36 , @xmath37 and @xmath38 can be interpolated by the triply heavy quark currents @xmath39 , @xmath40 and @xmath41 , respectively , @xmath42 where the @xmath43 and @xmath44 represent the heavy quarks @xmath45 and @xmath46 , the @xmath47 , @xmath47 and @xmath48 are color indexes , and the @xmath49 is the charge conjunction matrix .
there are other currents such as the @xmath50 besides the ioffe currents @xmath51 to interpolate the @xmath52 triply heavy baryon states , @xmath53 the currents @xmath51 and @xmath50 correspond to the superimpositions @xmath54 and @xmath55 respectively , where the fundamental currents @xmath56 and @xmath57 are defined as @xmath58 we can take the simple replacements @xmath59 and @xmath60 to obtain the corresponding currents for the proton @xcite .
the convergent behavior of the current @xmath61 is not as good as the ioffe current @xmath62 , and appearance of chirality violation terms in the operator product expansion indicates that the current @xmath61 couples strongly both to the positive - parity and negative - parity baryon states @xcite .
we can also choose the most general current @xmath63 , @xmath64 and search for the ideal value @xmath65 .
detailed studies of all the octet baryon states based on the qcd sum rules indicate that the optimal value is about @xmath66 , if the experimental values of the masses are taken as the input parameters @xcite , i.e. the ioffe currents work well .
we expect that the conclusion survives in the case of the heavy quark systems . at the present time , no experimental data are available to be taken as the input parameters in searching for the optimal value of the @xmath65
. the corresponding @xmath5 and @xmath67 triply heavy baryon states can be interpolated by the currents @xmath68 and @xmath69 because multiplying @xmath70 to the currents @xmath71 and @xmath72 changes their parity @xcite , where the currents @xmath71 and @xmath72 denotes the triply heavy quark currents @xmath39 , @xmath40 and @xmath41 , respectively .
the correlation functions @xmath73 and @xmath74 are defined by @xmath75 the currents @xmath76 couple to the @xmath77 triply heavy baryon states @xmath78 , while the currents @xmath79 couple to both the @xmath80 triply heavy baryon states @xmath81 and the @xmath77 triply heavy baryon states @xmath78 @xcite , @xmath82 where @xmath83 the @xmath84 are the pole residues , the @xmath85 are the masses , and the dirac spinors @xmath86 and @xmath87 satisfy the following identifies , @xmath88 \ , .
\end{aligned}\ ] ] we insert a complete set of intermediate triply heavy baryon states with the same quantum numbers as the current operators @xmath76 and @xmath89 into the correlation functions @xmath73 and @xmath74 to obtain the hadronic representation @xcite . after isolating the pole terms of the lowest states of the positive - parity and negative - parity triply heavy baryons , we obtain the following results @xcite : @xmath90 in this article , we choose the tensor structure @xmath91 for analysis , the @xmath92 triply heavy baryon states have no contaminations .
we can take @xmath93 for the correlation functions @xmath94
( @xmath73 , @xmath95 ) , and obtain the spectral densities at the phenomenological side , @xmath96 where @xmath97 \ , , \nonumber \\ b(p_{0 } ) & = & { 1 \over 2 } \left [ \lambda_+^{2 } \delta(p_{0 } - m_+ ) - \lambda_-^{2 } \delta(p_{0 } - m_{-})\right ] \ , , \end{aligned}\ ] ] the combinations @xmath98 and @xmath99 contain the contributions from the positive - parity and negative - parity triply heavy baryon states , respectively . in the following ,
we briefly outline the operator product expansion performed at the large space - like region @xmath100 .
we contract the heavy quark fields in the correlation functions @xmath73 and @xmath74 with wick theorem , substitute the full heavy quark propagators into the correlation functions @xmath73 and @xmath74 firstly , then complete the integrals in the coordinate space and momentum space sequentially to obtain the correlation functions @xmath73 and @xmath74 at the quark level . in calculations ,
we take into account all diagrams like the typical ones shown in fig.1 .
once the analytical quark - level correlation functions @xmath73 and @xmath74 are obtained , we take the limit @xmath93 , and use the dispersion relation to obtain the qcd spectral densities @xmath101 and @xmath102 ( which correspond to the tensor structures @xmath103 and @xmath104 respectively ) .
finally we introduce the weight functions @xmath105 $ ] , @xmath106 $ ] , and obtain the following qcd sum rules , @xmath107&=&\int_{\delta}^{\sqrt{s_0}}dp_0\left[\rho^a(p_0)\pm\rho^b(p_0)\right]\exp\left[-\frac{p_0 ^ 2}{t^2}\right ] \ , , \end{aligned}\ ] ] @xmath108&=&\int_{\delta}^{\sqrt{s_0}}dp_0\left[\rho^a(p_0)\pm\rho^b(p_0)\right]p_0 ^ 2\exp\left[-\frac{p_0 ^ 2}{t^2}\right]\,,\end{aligned}\ ] ] where the @xmath109 are the continuum threshold parameters , the @xmath110 are the borel parameters , and the threshold parameters @xmath111 or @xmath112 .
the spectral densities @xmath101 and @xmath102 at the level of quark - gluon degrees of freedom are given explicitly in the appendix . we can obtain the masses @xmath85 and pole residues @xmath84 by solving above equations with simultaneous iterations .
the input parameters are taken as @xmath113 @xcite , @xmath114 and @xmath115 @xcite .
the value of the gluon condensate @xmath116 has been updated from time to time , and changes greatly @xcite .
the updated value @xmath117 @xcite and the standard value @xmath113 @xcite lead to a tiny difference as the gluon condensate makes tiny contribution .
the heavy quark masses appearing in the perturbative terms are usually taken to be the pole masses in the qcd sum rules , while the choice of the @xmath118 in the leading - order coefficients of the higher - dimensional terms is arbitrary @xcite . in calculations , we observe that the dominating contributions come from the perturbative term .
so we take the pole masses and neglect the uncertainties of the pole masses .
the integral intervals of the energy @xmath119 are rather small , variations of the threshold parameters @xmath120 or @xmath112 can lead to remarkable changes of the continuum threshold parameters @xmath121 , we can fix the @xmath122 and vary the @xmath121 . in the conventional qcd sum rules
@xcite , there are two criteria ( pole dominance and convergence of the operator product expansion ) for choosing the borel parameter @xmath110 and continuum threshold parameter @xmath109 .
we impose the two criteria on the triply heavy baryon states to choose the borel parameter @xmath110 and continuum threshold parameter @xmath109 . in our previous works on the heavy and doubly heavy baryon states ,
the pole contributions are taken as @xmath123 @xcite .
we can take the same pole contributions , then search for the continuum threshold parameters @xmath121 to reproduce the relation @xmath124 .
the borel parameters @xmath110 , continuum threshold parameters @xmath121 , masses , pole residues are shown in table 1 and figs.2 - 3 . in this article
, we have neglected the contributions of the perturbative @xmath125 corrections , which can be taken into account by introducing formal coefficient @xmath126 through the unknown function @xmath127 .
as the dominant contributions come from the perturbative term , we expect that the @xmath125 corrections to the perturbative term can not change the masses remarkably , those effects can be absorbed in the pole residues approximately .
if we choose the structures @xmath103 and @xmath104 to study the masses , there are contaminations of the negative- ( or positive- ) parity triply heavy baryon states to the positive- ( or negative- ) parity triply heavy baryon states , the corresponding fractions can be expressed as @xmath128\exp\left[-\frac{p_0 ^ 2}{t^2}\right ] } { \int_{\delta}^{\sqrt{s_0}}dp_0\left[\rho^a(p_0 ) \pm\rho^b(p_0)\right]\exp\left[-\frac{p_0 ^ 2}{t^2}\right ] } \ , .\end{aligned}\ ] ] in this article , we separate the contributions of the positive - parity and negative - parity triply heavy baryon states explicitly . in calculations , we have taken the pole masses .
on the other hand , we can take the @xmath129 masses @xmath130 , @xmath131 , as the @xmath129 masses are also used in the qcd sum rules , for example , in studying the @xmath132 form - factors @xcite .
we choose the same borel parameters and suitable continuum threshold parameters to reproduce the same pole contributions so as to obtain the ground state masses and pole residues , the predictions are presented in table 1 . from table 1
, we can see that the pole masses and the @xmath129 masses result in large discrepancies for the masses of the triply heavy baryon states .
if the @xmath129 masses are taken , the present predictions are compatible with the values from ref.@xcite within uncertainties . in ref.@xcite ,
the contributions of the positive - parity baryon states are not distinguished from that of the negative - parity baryon states .
for the established bottom baryon states @xmath133 with three stars , the masses are @xmath134 and @xmath135 respectively from the particle data group @xcite .
the prediction @xmath136 based on the pole mass @xmath137 is consistent with experimental data @xcite , while the prediction @xmath138 based on the @xmath129 mass @xmath139 underestimates the experimental data about @xmath140 , if we take the same values of other parameters ; so the pole masses are preferred . in the qcd sum rules ,
if the variations of the threshold parameters @xmath122 can lead to relatively large changes for the integral ranges @xmath141 , the predictions are sensitive to the masses @xmath118 . in the present case
, the mass uncertainty @xmath142 can result in uncertainty @xmath143 for the triply - bottom baryon state @xmath144 . the pole mass and
the @xmath129 mass correspond to quite different continuum threshold parameters @xmath109 , see table 1 . on the other hand ,
if the variations of the threshold parameters @xmath122 result in small values of the @xmath145 , the predictions based on the pole masses and the @xmath129 masses lead to small discrepancies .
irrespective of the pole masses and the @xmath129 masses , it would be better to understand the heavy quark masses @xmath118 as the effective quark masses ( or just the mass parameters ) .
our previous works on the mass spectrum of the @xmath146 and @xmath147 heavy and doubly heavy baryon states indicate such parameters ( the pole masses ) can lead to satisfactory results @xcite , we prefer the pole masses .
there are no experimental data for the masses of the triply heavy baryon states , the present predictions are compared with other theoretical calculations , such as the qcd bag model @xcite , the quark model estimation @xcite , the variational approach @xcite , the modified bag model @xcite , the relativistic three - quark model @xcite , the qcd sum rules @xcite , the non - relativistic three - quark model @xcite , see table 2 .
all those predictions should be confronted with the experimental data in the future .
the lhc will be the world s most copious source of the @xmath46 hadrons , and a complete spectrum of the @xmath46 and @xmath45 hadrons will be available through the gluon fusions . in proton - proton collisions at @xmath148 ,
the @xmath149 cross section is expected to be @xmath150 producing @xmath151 @xmath149 pairs in a standard year of running at the lhcb operational luminosity of @xmath152 @xcite . .
the borel parameters , continuum threshold parameters , pole contributions , masses and pole residues of the triply heavy baryon states .
the overline on the @xmath153 denotes the @xmath129 masses are used .
[ cols="^,^,^,^,^,^,^,^",options="header " , ]
in this article , we extend our previous works on the mass spectrum of the heavy and doubly heavy baryon states to study the @xmath0 and @xmath1 triply heavy baryon states by subtracting the contributions from the corresponding @xmath2 and @xmath3 triply heavy baryon states with the qcd sum rules , and make reasonable predictions for their masses .
the predictions can be confronted with the experimental data in the future at the lhc or used as basic input parameters in other theoretical studies .
this work is supported by national natural science foundation , grant number 11075053 , and the fundamental research funds for the central universities .
the spectral densities of the triply heavy baryon states at the level of quark - gluon degrees of freedom , @xmath154 where @xmath155\nonumber\\ & & \left[1+\frac{p_0}{4t}\right]\delta(p_0-\widetilde{m}_{qq'})-\frac{m_{q'}^2m_q^2}{192\pi^2p_0t}\langle\frac{\alpha_sgg}{\pi}\rangle \frac{1}{z^2}\delta(p_0-\widetilde{m}_{qq'})\nonumber\\ & & -\frac{m_{q}^4}{192\pi^2p_0t}\langle\frac{\alpha_sgg}{\pi}\rangle \left[\frac{z}{y^3}+\frac{z}{(1-y - z)^3}\right]\delta(p_0-\widetilde{m}_{qq'})\nonumber\\ & & + \frac{m_{q}^2}{32\pi^2}\langle\frac{\alpha_sgg}{\pi}\rangle \left[\frac{z}{y^2}+\frac{z}{(1-y - z)^2}\right]\delta(p_0-\widetilde{m}_{qq'})\nonumber\\ & & + \frac{p_0}{32\pi^2}\langle\frac{\alpha_sgg}{\pi}\rangle \left[y-(1-y - z)\right]\left[3+\frac{p_0}{2}\delta(p_0-\widetilde{m}_{qq'})\right]\nonumber\\ & & + \frac{m_q^2}{64\pi^2}\langle\frac{\alpha_sgg}{\pi}\rangle \left[\frac{1}{1-y - z}-\frac{1}{y}\right ] \delta(p_0-\widetilde{m}_{qq ' } ) \ , , \end{aligned}\ ] ] @xmath156\delta(p_0-\widetilde{m}_{qq'})\nonumber\\ & & -\frac{m_{q'}m_q^2}{96\pi^2}\langle\frac{\alpha_sgg}{\pi}\rangle \left[\frac{(1-y - z)}{y^2}+\frac{y } { ( 1-y - z)^2}\right]\left[\frac{1}{p_0}+\frac{1}{2t}\right]\delta(p_0-\widetilde{m}_{qq'})\nonumber\\ & & -\frac{m_{q'}m_q^2}{96\pi^2p_0 ^ 2t}\langle\frac{\alpha_sgg}{\pi}\rangle \left[\frac{m_{q'}^2}{z^3}+\frac{m_q^2}{(1-y - z)^3}+\frac{m_q^2}{y^3}\right]\delta(p_0-\widetilde{m}_{qq'})\nonumber\\ & & + \frac{m_{q'}}{8\pi^2}\langle\frac{\alpha_sgg}{\pi}\rangle \frac{y(1-y - z)}{z^2}\left[1+\frac{p_0}{4}\delta(p_0-\widetilde{m}_{qq'})\right]\nonumber\\ & & + \frac{m_{q'}m_q^2}{16\pi^2p_0}\langle\frac{\alpha_sgg}{\pi}\rangle \left[\frac{1}{y^2}+\frac{1}{z^2}+\frac{1}{(1-y - z)^2}\right]\delta(p_0-\widetilde{m}_{qq'})\nonumber\\ & & -\frac{m_{q ' } } { 16\pi^2}\langle\frac{\alpha_sgg}{\pi}\rangle \left [ 1-\frac{y}{z}+\frac{1-y - z}{z}\right ] \left[1+\frac{p_0}{4}\delta(p_0-\widetilde{m}_{qq'})\right]\nonumber\\ & & + \frac{m_{q'}m_q^2 } { 32\pi^2p_0}\langle\frac{\alpha_sgg}{\pi}\rangle \frac{1}{z } \left[\frac{1}{1-y - z}-\frac{1}{y } \right ] \delta(p_0-\widetilde{m}_{qq ' } ) \ , , \end{aligned}\ ] ] @xmath157\nonumber\\ & & \left[1+\frac{p_0}{2t}\right]\delta(p_0-\widetilde{m}_{qq'})-\frac{m_{q'}^2m_q^2}{384\pi^2p_0t}\langle\frac{\alpha_sgg}{\pi}\rangle \frac{1}{z^2 } \delta(p_0-\widetilde{m}_{qq'})\nonumber\\ & & -\frac{m_q^4}{384\pi^2p_0t}\langle\frac{\alpha_sgg}{\pi}\rangle \left [ \frac{z}{y^3 } + \frac{z}{(1-y - z)^3}\right]\delta(p_0-\widetilde{m}_{qq'})\nonumber\\ & & + \frac{m_q^2}{64\pi^2}\langle\frac{\alpha_sgg}{\pi}\rangle \left [ \frac{z}{y^2 } + \frac{z}{(1-y - z)^2}\right ] \delta(p_0-\widetilde{m}_{qq'})\nonumber\\ & & -\frac{p_0}{48\pi^2}\langle\frac{\alpha_sgg}{\pi}\rangle z\left[1+\frac{p_0}{8}\delta(p_0-\widetilde{m}_{qq'})\right]\ , , \end{aligned}\ ] ] @xmath158 \delta(p_0-\widetilde{m}_{qq'})\nonumber\\ & & + \frac{m_{q'}}{32\pi^2}\langle\frac{\alpha_sgg}{\pi}\rangle \frac{y(1-y - z)}{z^2 } \left[1+\frac{p_0}{2}\delta(p_0-\widetilde{m}_{qq'})\right]\nonumber\\ & & -\frac{m_{q'}m_q^2}{384\pi^2t}\langle\frac{\alpha_sgg}{\pi}\rangle \left [ \frac{y}{(1-y - z)^2}+\frac{1-y - z}{y^2}\right ] \delta(p_0-\widetilde{m}_{qq'})\nonumber\\ & & + \frac{m_{q'}m_q^2}{64\pi^2p_0}\langle\frac{\alpha_sgg}{\pi}\rangle \left[\frac{1}{y^2}+\frac{1}{(1-y - z)^2}+\frac{1}{z^2 } \right]\delta(p_0-\widetilde{m}_{qq'})\nonumber\\ & & + \frac{m_{q'}}{192\pi^2 } \langle\frac{\alpha_sgg}{\pi}\rangle \left[1+\frac{p_0}{2}\delta(p_0-\widetilde{m}_{qq'})\right]\ , , \end{aligned}\ ] ] m. a. gomshi nobary , phys . lett . *
b559 * ( 2003 ) 239 ; m. a. gomshi nobari , r. sepahvand , phys . rev . * d71 * ( 2005 ) 034024 ; m. a. gomshi nobary , r. sepahvand nucl . phys . *
b741 * ( 2006 ) 34 ; m. a. gomshi nobary , r. sepahvand phys . rev . * d76 * ( 2007 ) 114006 ; m. a. gomshi nobary , b. nikoobakht , j. naji , nucl . phys . * a789 * ( 2007 ) 243 ; y. q. chen , s. z. wu , arxiv:1106.0193 .
e. bagan , m. chabab , h. g. dosch , s. narison , phys . lett .
* b278 * ( 1992 ) 367 ; e. bagan , m. chabab , h. g. dosch , s. narison , phys . lett .
* b287 * ( 1992 ) 176 ; e. bagan , m. chabab , s. narison , phys . lett .
* b306 * ( 1993 ) 350 ; y. b. dai , c. s. huang , c. liu , c. d. lu , phys . lett . * b371 * ( 1996 ) 99 ; s. groote , j. g. korner , o. i. yakovlev , phys . rev . *
d55 * ( 1997 ) 3016 ; o. duraes , m. nielsen , phys . lett .
* b658 * ( 2007 ) 40 ; z. g. wang , eur . phys .
j. * c54 * ( 2008 ) 231 ; j. r. zhang , m. q. huang , phys .
* d78 * ( 2008 ) 094015 ; j. r. zhang , m. q. huang , phys .
rev . * d78 * ( 2008 ) 094007 ; z. g. wang , eur .
j. * c61 * ( 2009 ) 321 ; r. m. albuquerque , s. narison , m. nielsen , phys . lett . *
b684 * ( 2010 ) 236 ; z. g. wang , eur .
j. * a44 * ( 2010 ) 105 .
z. g. wang , phys .
b685 * ( 2010 ) 59 ; z. g. wang , eur . phys . j. * c68 * ( 2010 ) 479 ; z. g. wang , eur .
j. * c68 * ( 2010 ) 459 ; z. g. wang , eur . phys .
j. * a45 * ( 2010 ) 267 ; z. g. wang , eur . phys .
j. * a47 * ( 2011 ) 81 . | in this article , we study the @xmath0 and @xmath1 triply heavy baryon states in a systematic way by subtracting the contributions from the corresponding @xmath2 and @xmath3 triply heavy baryon states with the qcd sum rules , and make reasonable predictions for their masses .
+ zhi - gang wang + department of physics , north china electric power university , baoding 071003 , p. r. china pacs number : 14.20.lq , 14.20.mr key words : triply heavy baryon states , qcd sum rules |
ccd photometry was obtained in a field of 0.5 square degree in the bar of the lmc between 1991 - 1993 for eros .
about 2500 images spanning @xmath1130 days were taken in two broad bandpass filters @xmath15 and @xmath16 centered respectively on 490 and 670 nm in the 9192 campaign , and 5500 images were taken of the same field with a pair of very similar filters ( @xmath17 ) for the 1992 - 1993 campaign .
we have systematically searched the eros database for variable stars using the modified periodogram technique@xcite@xcite and the aov method@xcite . among the hundredth detected variable stars , we have discovered a bright lmc variable star ( @xmath18m@xmath19 , equinox 2000.0 ) with a period of @xmath0 days and a particular behaviour , a clear alternance between cycles with larger and smaller amplitudes ( fig 1 ) .
a fourier fit with 8 independent frequencies leads to a spectrum with two dominant frequencies at @xmath2=825.69 phz and at 1.5003@xmath3 ( fig.2 ) .
this suggests that the pulsation of this star is phase locked between two modes of frequencies @xmath2 and 1.5@xmath3 . from the photometry ,
assuming different values of reddening ( e(b - v ) = 0.10 - 0.30 ) and applying the temperature scale from kurucz s atmospheric models , we estimate the effective temperature of the star to be in the range @xmath20 .
low resolution spectroscopy in the wavelength range @xmath25 has been obtained at eso la silla in december 1995 with the eso 1.5 m equiped with a boller and chivens spectrograph .
the resolution was @xmath26 , and we get a signal to noise ratio of 50 at @xmath27
. this star belongs to the lmc .
its spectrum shows strong @xmath28 and @xmath5 emission lines , and neutral lines of helium indicating a spectral type between late o and early b ( fig 3 ) .
further observational information are : + ( 1 ) this star is located in the same area in the bar of the lmc , where 7 pre main sequence star candidates ( pmsc ) have been found@xcite . +
( 2 ) it presents strong balmer emission lines ( equivalent width of @xmath29 for @xmath30 ) .
+ ( 3 ) its colour and spectrum suggest a spectral type between late o and early b. + ( 4 ) we have no evidence for the presence of an extended @xmath31 region around this object .
+ ( 5 ) it is brighter than the pmsc which we discovered in the lmc .
+ ( 6 ) it presents an irregular photometric variability with a time scale of @xmath32 days
. however , the photometric variability seems to be due to a two mode phase - locked pulsation .
the observed large amplitude leads to favour a radial pulsation rather a non - radial one .
+ based on these arguments , we suggest that this star can be a pre main sequence object or a post - agb star . a companion poster @xcite examine the possible nature of this object by means of a linear stability analysis of hydrostatic envelopes , and computations of hydrodynamical models .
990.4 cm beaulieu j.p . , 1995 , in _ astrophysical application of stellar pulsation _ , stobie r.s . , whitelock p.a .
. ser . * 83 * , 260 beaulieu j.p . , lamers h. , et al .
, 1996,_science _ * 272 * , 995 cox j.p .
et al . , 1980 , _ space science reviews _ * 27 * , 529 .
grison p. , 1994
astrophys . _ * 289 * , 404 grison p. , et al . , 1995 , _
* 109 * , 447 kollth z. et al . , 1997 ,
this volume schwarzenberg - czerny , 1989 _ mon . not .
. soc _ * 241 * , 153 | using ccd photometry obtained by the eros collaboration in 1991 - 1993 , we have discovered an lmc variable star with a light curve that is oscillating with a mean period of @xmath0 days and an amplitude of @xmath1 0.3 mag .
the oscillations appear with irregular amplitude variations .
the fourier spectrum shows that the pulsation of this star is phase locked between two modes of frequencies @xmath2 and 1.5@xmath3 .
moreover , this object has strong @xmath4 and @xmath5 emission lines and neutral lines of helium that suggest a spectral type between late o and early b. in a preliminary analysis , we derive a luminosity of @xmath6 and an effective temperature in the range @xmath7 . * a @xmath0 days star with two phase - locked * 0.5 truecm * modes of pulsation in the eros database * 1.0 truecm 1.0 truecm authors : j.p .
beaulieu@xmath8 , r.buchler@xmath9 , m.j.goupil@xmath10 , z.kollath@xmath11 , @xmath12 kapteyn laboratorium , postbus 800 , 9700 av groningen , the netherlands .
+ @xmath13 institut dastrophysique de paris , cnrs , 98bis boulevard arago , f75014 paris , france .
+ @xmath9 physics department , university of florida , gainesville , fl 32611 , usa .
+ @xmath10 dasgal , observatoire de paris , meudon 92195 , france .
+ @xmath14 konkoly observatory , budapest , hungary . |
hypothesis testing is a very important component of statistical inference ; it helps us to systematically explore the veracity of an unsubstantiated claim on the basis of observed data in a real life experiment .
the philosophy of the statistical hypothesis testing was mainly developed in the early decades of twentieth century by fisher and neyman and pearson @xcite ; since then the theory has evolved in many directions through the contributions of several later researchers . yet
the classical likelihood ratio test ( lrt ) proposed by @xcite and formalized later by @xcite still remains our canonical hypothesis testing tool , which is used widely by the practitioners in all scenarios of human endeavor .
this test can be easily performed for most statistical models and satisfies several asymptotic optimality criteria . however , as in the case of maximum likelihood estimator ( mle ) in the estimation context , the lrt also has serious robustness problems under misspecification of models and/or presence of outliers .
there have been several attempts to develop robust tests of hypotheses having optimal properties similar to the lrt .
the lrt can be seen to be a special case of a large class of testing procedures , known as the disparity difference tests " which have strong robustness properties ( e.g. * ? ? ?
however , these tests involve kernel density estimators of the true density for continuous models and hence include all the complications associated with kernel smoothing .
motivated by the success of the minimum density power divergence estimators , which require no kernel smoothing , as an alternative to the class of minimum disparity estimators , @xcite developed a class of tests using the density power divergence .
this work was based on @xcite , which had developed the density power divergence measure , as well as the minimum density power divergence estimator ( mdpde ) .
consider the problem of testing a simple null hypothesis based on the random sample @xmath1 .
let @xmath2 represent the parametric family of densities ; suppose also that the true data generating density belongs to this model family .
let @xmath3 be any fixed point in the parameter space @xmath4 , which we believe to be the true value of the parameter .
based on the observed sample , we want to test for the simple hypotheses @xmath5 note that in general there is no uniformly most powerful test for this testing problem .
@xcite proposed to test this hypothesis by utilizing the minimum possible value of the density power divergence ( dpd ) measure between the data and a model density .
although they have empirically demonstrated some of strong robustness properties of the dpd based test , their paper had no concrete theoretical results on the robustness of the proposed tests . the present paper will focus on developing the theoretical robustness properties of the dpd based test .
however , instead of doing it simply for the dpd alone , we will prove all our results for a more general class of test statistics that contains the dpd based tests as a special case .
this general class of test statistics will be based on the recently developed family of @xmath0-divergences @xcite that contains both the pd ( power divergence ; cressie and read , 1984 ) and the dpd measure as its subfamilies .
for all the asymptotic results throughout the paper we need to assume some standard conditions of asymptotic inference , given by assumptions a , b , c and d of @xcite . in the rest of this paper , we will refer to these conditions simply as the _ lehmann conditions _ " .
similarly , we will also assume the conditions d1d5 of @xcite at the model , which we will refer to as the _ basu et al . conditions _ " .
both set of conditions , with a brief description of their implications and significances are provided in [ app : conditions ] .
there is some overlap among the conditions , but taken together they exhaust the necessary technicalities . to keep a clear focus in our discussion and to relate them directly to the empirical findings of @xcite
, we will restrict ourselves to a simple null hypothesis in this paper .
however , at the end of the paper , we will briefly indicate how the results can be extended to the case of composite null hypotheses .
the rest of the paper is organized as follows : in section [ sec:7sample1_simpletest ] we define the proposed test statistic based on the general family of @xmath0-divergence measures and provide its asymptotic properties .
several robustness measures of this test are derived in section [ sec:7sample1_simpletest_robust ] .
section [ sec:7example_simple ] presents a numerical illustration of all the theoretical results derived in this paper through the problem of testing for the normal mean with a known variance . in section [ sec : choice_tuning ] we present some general remarks integrating the theoretical results and numerical findings presented in this paper ; in this section we also give some guidance on the choice of appropriate tuning parameter for the proposed test .
section [ sec : composite ] briefly indicates the possible generalization to the case of the composite null hypotheses .
some concluding remarks are presented in section [ sec : conclusion ] .
the @xmath0-divergence measure , recently proposed by @xcite , is a general family of divergence measures ( between two density functions ) containing several popular divergences including the power divergence ( pd ) family of cressie and read ( 1984 ) and the density power divergence ( dpd ) family of @xcite .
the @xmath0-divergence between the densities @xmath6 and @xmath7 is defined in terms of two parameters @xmath8 $ ] and @xmath9 as @xmath10 where @xmath11 and @xmath12 . for the cases @xmath13 or @xmath14
, the corresponding @xmath0-divergence measure is defined by the respective continuous limits of the expression in equation ( [ eq : s_div_gen ] ) , which yield @xmath15 and @xmath16 interestingly , note that at @xmath17 the @xmath0-divergence measure becomes independent of the parameter @xmath18 and coincides with the squared @xmath19-distance . on the other hand ,
at @xmath20 it reduces to the pd family of cressie and read ( 1984 ) with parameter @xmath18 .
further , note that the @xmath0-divergence family also contains the dpd measure with parameter @xmath21 , given by @xmath22 as one of its special cases when @xmath23 .
therefore , it is natural to reconstruct the dpd based test statistics proposed by @xcite using this general family of @xmath0-divergences
. consider the problem of testing the simple null hypothesis as described in equation ( [ eq:7simple_hyp ] ) under the parametric set - up of section 1 .
we define the general test statistic based on the @xmath0-divergence with parameter @xmath24 and @xmath18 as @xmath25 where @xmath26 is the minimum dpd estimator ( mdpde ) of @xmath27 obtained by minimizing the dpd with tuning parameter @xmath28 .
the intuitive rationale behind construction of this test statistics is as follows : when the model is correctly specified and the assumed null hypothesis @xmath29 is correct , @xmath30 is the true data generating density and so it can be tested by considering the magnitude of the @xmath0-divergence measure between @xmath30 and @xmath31 for any consistent estimator @xmath32 of @xmath27 based on the observed sample . since the divergence employed in the definition of the statistic in equation ( [ eq : sdt ] ) is the @xmath0-divergence , the ideal choice for @xmath32 should be the minimum @xmath0-divergence estimator . the choice of @xmath26 , the mdpde of @xmath27 corresponding to tuning parameter @xmath28 , is preferred by us since the only subfamily in the @xmath0-divergence family that does not require the use of kernel density estimator is the dpd family .
note that , by putting @xmath23 this general test statistic coincides with the dpd based test statistic of @xcite , whereas with the choice @xmath33 it becomes equivalent to the lrt . in the rest of the paper
we refer to the test statistic @xmath34 in the singular , although in effect we are taking a family of test statistics that vary over the tuning parameters .
we will prove our results for a generic statistic , where the asymptotic distribution is a function of @xmath24 , @xmath18 and @xmath28 .
we first prove some asymptotic properties of this general @xmath0-divergence based test ( which we will refer to as the sdt ) to obtain the critical points and power approximation of the test .
for these , we require that the minimum dpd estimator used in constructing the test statistics is @xmath35-consistent and asymptotically normal , which hold under the basu et al .
conditions @xcite . then , the asymptotic variance - covariance matrix of the mdpde @xmath26 is @xmath36 where @xmath37 and @xmath38 we can now derive the asymptotic null distribution of the sdt as given in the following theorem .
suppose the model density satisfies the lehmann and basu et al
. conditions .
the asymptotic distribution of the test statistic @xmath39 , under the null hypothesis @xmath40 , coincides with the distribution of @xmath41 where @xmath42 are independent standard normal variables , @xmath43 are the nonzero eigenvalues of @xmath44 with @xmath45 and @xmath46 [ thm:7asymp_null_one ] the proof of the above theorem is a routine extension of the proof of the dpd based test provided by @xcite and is omitted .
note that the minimum @xmath0-divergence estimator corresponding to parameter @xmath28 and @xmath18 is first order equivalent to the minimum dpd estimator with parameter @xmath28 in the sense that they have the same asymptotic distribution @xcite .
so , if one were to replace the minimum dpd estimator @xmath26 in ( [ eq : sdt ] ) with the minimum @xmath0-divergence estimator corresponding to @xmath28 and @xmath18 , one would still get the same asymptotic null distribution as in theorem [ thm:7asymp_null_one ] ; this null distribution is independent of @xmath18 .
this allows us to study the class of tests in ( [ eq : sdt ] ) in a more general setting than what is observed at face value .
we could have made the set up fully general by actually substituting @xmath26 with the minimum @xmath0-divergence estimator at @xmath28 and @xmath18 in equation ( [ eq : sdt ] ) .
but the general @xmath0-divergence estimator involves the construction of a non - parametric density estimator and inherits its associated problems , and so the value addition due to the generality might be offset by its cost .
instead we stick to the formulation in ( [ eq : sdt ] ) , which gives us the same asymptotic distribution as the most general case , but also gives us a robust set of procedures without getting into the issue of non - parametric kernel density estimation .
it also allows us to study the theoretical properties of the tests of @xcite as a special case of our formulation .
one could also find the asymptotic critical values of the test in ( [ eq : sdt ] ) on the basis of the suggestion in remark 3 of @xcite .
our next theorem presents a power approximation of the above sdt , which can help us to obtain the minimum sample size required in order to achieve some pre - specified power of the test .
[ thm:7aprox_power_one ] suppose the model density satisfies the lehmann and basu et al
. conditions .
an approximation to the power function of the test statistic @xmath39 for testing the null hypothesis in equation ( [ eq:7simple_hyp ] ) at the significance level @xmath47 is given by @xmath48 where @xmath49 tends uniformly to the standard normal distribution function @xmath50 , @xmath51 is the @xmath52 quantile of the asymptotic distribution of @xmath39 and @xmath53 with @xmath54.\ ] ] here @xmath55 and @xmath56 , as defined in the statement of the @xmath0-divergence . for any @xmath57
, the probability of rejecting the null hypothesis @xmath29 at any fixed significance level @xmath58 with the rejection rule @xmath59 tends to @xmath60 as @xmath61 .
thus the test statistic @xmath34 is consistent .
now let us consider the robustness properties of the general test statistic based on the @xmath0-divergence family ( the sdt ) .
we will see that the robustness of the proposed test statistic originates from the robustness of the minimum dpd estimator used in the test statistic and hence the proposed test will be seen to be highly robust for @xmath62 . as a particular case , all the results to be covered in this section will also hold for the dpd based test developed in @xcite ; this will fill the gap of the absence of any theoretical robustness results in their paper .
let us first consider hampel s influence function of the test @xcite .
ignoring the multiplier @xmath63 in our test statistic we define the sdt functional as @xmath64 where @xmath65 is the minimum dpd functional defined in @xcite .
now consider the contaminated distribution @xmath66 with @xmath67 being the contamination proportion and @xmath68 being the degenerate distribution with all its mass at the contamination point @xmath69 .
then hampel s first - order influence function of the sdt functional is given by @xmath70 where @xmath71 is the influence function of the minimum dpd functional @xmath72 and @xmath73 .
in general the influence function of a test is evaluated at the null distribution @xmath74 . however , by the fisher consistency property of the functional @xmath75 , we get @xmath76 and @xmath77 so that the hampel s first - order influence function of our test statistic is zero at the null .
therefore , we need to consider higher order influence functions of the proposed sdt .
the second order influence function of our test statistic can be seen to have the form @xmath78 in particular , at the null distribution @xmath74 , the second order influence function of the sdt becomes @xmath79 thus the influence function of the @xmath0-divergence based test at the null hypothesis is independent of the parameter @xmath18 implying that the theoretical robustness of the proposed sdt will be independent of @xmath18 .
also , this influence function will be bounded if and only if the influence function of the minimum density power divergence functional is bounded .
@xcite derived the influence function of the mdpde which is bounded for all @xmath80 under most common parametric models ; however it is generally unbounded at @xmath81 . noting that the mdpde with @xmath81 is indeed the mle
, it shows that the use of a non - robust estimator like the mle leads to a non - robust overall test procedure as well .
next we consider the effect of contamination on the level and power of the proposed test which will give us a clearer picture about the robustness of the test . as
the test is consistent , we study its power under the contiguous alternative hypotheses @xmath82 with @xmath83 being a vector of positive reals having the same dimension as the parameter vector and @xmath84 . in order to explore the effect of contamination on the power and size of the test
, we also need to consider some contamination over these contiguous alternatives .
following @xcite , one must consider the contaminations such that their effect tends to zero as @xmath85 tends to @xmath3 at the same rate to avoid confusion between the null and alternative neighborhoods ( also see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
therefore , we consider the contaminated distributions @xmath86 and @xmath87 then the level influence function ( lif ) is given by @xmath88 and the power influence function ( pif ) is given by @xmath89 we begin with the derivation of the asymptotic power under contaminated distributions which is presented in the following theorem .
assume that the lehmann and basu et al .
conditions hold for the model density . then for any @xmath90 and @xmath91 , we have the following : * the asymptotic distribution of the @xmath0-divergence based test statistics @xmath39 under @xmath92 is the same as that of the quadratic form @xmath93 , where @xmath94 follows a @xmath95-variate normal distribution with mean @xmath96\ ] ] and variance - covariance matrix @xmath97 .
equivalently , this distribution is the same as that of @xmath98 where @xmath99 are the @xmath100 nonzero eigenvalues of @xmath44 as in theorem [ thm:7asymp_null_one ] and @xmath101 are independent non - central chi - square variables having degree of freedom one and non - centrality parameters @xmath102 respectively with @xmath103 and @xmath104 and @xmath105 is the matrix of normalized eigenvectors of @xmath44 . *
the asymptotic power of the proposed sdt under contaminated distribution @xmath92 is given by @xmath106 where @xmath107 denotes a chi - square random variable with @xmath95 degrees of freedom , @xmath108 is the minimum of @xmath109s for @xmath110 and @xmath111 with @xmath112 and @xmath113 ^ 2,\ ] ] for @xmath100 independent standard normal random variables @xmath114 .
[ thm:7asymp_power_one ] part ( ii ) follows from part ( i ) using the series expansion of the distribution function of a linear combination of independent non - central chi - squares in terms of central chi - square distribution functions as derived in @xcite .
interestingly note that the asymptotic power under contiguous alternatives and contiguous contamination is independent of the parameter @xmath18 ; thus asymptotically the power robustness of the sdt will be independent of @xmath18 .
further , substituting @xmath152 or @xmath153 in the above theorem , we will get several important cases ; these are presented in the following corollaries .
[ cor:7cont_power_null ] for @xmath154 , the distribution @xmath92 coincides with the null distribution .
putting @xmath154 in theorem [ thm:7asymp_power_one ] , we get @xmath155 so that @xmath156 for each @xmath157 and the asymptotic distribution of the sdt statistics given in the previous theorem coincides with its null distribution derived independently in theorem [ thm:7asymp_null_one ] .
so part ( ii ) of theorem [ thm:7asymp_power_one ] gives an alternative approximation to the critical values of the proposed sdt based on a suitable truncation of the infinite series .
error bounds in such truncation can also be calculated from remark [ rem : expansion_non - central_chi2 ] below or from the results of @xcite .
[ cor:7cont_power_one ] putting @xmath153 in theorem [ thm:7asymp_power_one ] , we get the asymptotic distribution of the proposed sdt under the contiguous alternative hypotheses @xmath158 as given in part ( i ) of theorem [ thm:7asymp_power_one ] with @xmath159 .
further , in this case part ( ii ) of the theorem yields the asymptotic power under contiguous alternative hypotheses given by @xmath160 putting @xmath152 in theorem [ thm:7asymp_null_one ] , we get the asymptotic distribution of the proposed sdt under the probability distribution @xmath161 as given in part ( i ) of the theorem with @xmath162 .
hence , the asymptotic level under contamination has the form @xmath163 [ cor:7asymp_level_one ] in the above theorem [ thm:7asymp_null_one ] , we have used a series expansion for the distribution of the linear combination of independent non - central chi - square random variables in terms of the central chi - square distributions from the work of @xcite .
there are in fact many more such expansions or approximations for the distribution of these random variables available in the literature ; for example see @xcite , @xcite and @xcite among many others . however , the particular expansion used here is specifically useful for deriving the power influence function , as we will see next .
further , for any practical usage , we can use a finite truncated sum as an approximation of the infinite series considered in ( [ eq:7asymp_power_cont_null ] ) . for the truncated series up to the @xmath164-th term
, the error in approximation can be bounded by @xmath165 see @xcite for more accurate error bounds for such approximations .
[ rem : expansion_non - central_chi2 ] [ thm:7pif_simpletest1 ] assume that the lehmann and basu et al .
conditions hold for the model density and the influence function @xmath178 of the minimum dpd estimator is bounded .
then , for any @xmath90 , the power influence function of the proposed sdt is given by @xmath179 p\left(\chi_{r+2v}^2 > \frac{t_\alpha^{\beta,\gamma}}{\zeta_{(1)}^{\gamma , \beta}(\theta_0)}\right)\right ) .
\nonumber\\ & & ~~ \label{eq:7pif_simpletest1 } \end{aligned}\ ] ] note that the quantity @xmath180
p\left(\chi_{r+2v}^2 > \frac{t_\alpha^{\beta,\gamma}}{\zeta_{(1)}^{\gamma , \beta}(\theta_0)}\right)\right ) = c^*$ ] , say , is independent of the parameter @xmath18 and the contamination point @xmath69 .
thus the above theorem shows that the power influence function is bounded whenever the influence function of the mdpde is bounded .
but the latter condition is satisfied by most standard parametric models as shown in @xcite .
this implies the power robustness of the proposed test statistics and the extent of theoretical robustness is again independent of the parameter @xmath18 .
further , combining equations ( [ eq:7asymp_power_cont_null ] ) and ( [ eq:7pif_00simpletest1 ] ) , we can get an interesting interpretation of the power influence function . substituting the value of @xmath181 from ( [ eq:7pif_00simpletest1 ] ) in to the expression ( [ eq:7asymp_power_cont_null ] ) and using corollary [ cor:7cont_power_one ]
, we get @xmath182 whenever the influence function of the mdpde is bounded .
thus , the power influence function gives a first order approximation for the change in asymptotic power under contiguous alternative hypotheses caused by a contiguous contamination in data .
this is in fact comparable with the interpretation of the influence function of any estimator as an indicator of the change in its bias under contaminated data . *
[ the case of @xmath184 * + all the results derived in this subsection including the power and level influence function can be further simplified for the cases of univariate parameters with @xmath185 . in this particular case
, it follows from theorem [ thm:7asymp_null_one ] that the asymptotic null distribution of the proposed sdt statistic is the same as @xmath186 , where @xmath187 follows a chi - square distribution with one degree of freedom .
thus , the critical value for a level @xmath47 sdt is given by @xmath188 , where @xmath189 is upper @xmath190 quantile of the distribution of @xmath187 .
in this section we will further explore the robustness of the sdt in terms of the stability of its limiting distribution under general contamination .
in contrast to the contiguous contamination considered in the preceding section , here we consider the fixed departure from our model density that is independent of the sample size @xmath202 . following @xcite ,
let us consider the unknown true distribution to be @xmath203 that may or may not be in our model family and let the null hypothesis of interest be @xmath204 then our @xmath0-divergence test statistics for testing the above can be written as @xmath205 where @xmath206 with @xmath207 being the empirical distribution function . for simplicity , we restrict our attention to the case of scalar @xmath27 only ; generalization to multivariate case can be made by a routine extension although the interpretation is more difficult for the multivariate case .
consider the above set - up with a scalar parameter @xmath27 and let @xmath6 denote the density function corresponding to @xmath203 . then under the null hypothesis in ( [ eq:7null_hyp_func ] ) , @xmath208 where @xmath209 with @xmath210 @xmath211 with @xmath212 and @xmath213 [ thm:7chi_infl_fact ] following the same proof as in theorem [ thm:7asymp_null_one ]
, we have that under the above null hypothesis , the asymptotic distribution of the test statistics @xmath39 is the same as that of @xmath214 ^ 2 $ ] .
however since @xmath215 , it follows that under @xmath6 , @xmath216 is asymptotically univariate normal with mean zero and variance @xmath217 so that the asymptotic distribution of @xmath214 ^ 2 $ ] is nothing but @xmath218 . as in @xcite for the disparity difference test , here also we refer @xmath219 as the _ chi - square inflation factor _ having exactly the similar interpretations
. however , the confidence interval of @xmath220 based on theorem [ thm:7chi_infl_fact ] will only be correct whenever @xmath221 ; it will be conservative provided @xmath222 and liberal otherwise .
clearly , @xmath219 is independent of the parameter @xmath18 and depends on the true density @xmath6 only through the parameter @xmath28 of the mdpde .
thus the stability of this sdt and corresponding confidence interval under the model misspecification solely depends on the robustness of the mdpde used in the test statistics .
as an illustration , we study in detail , the properties of this chi - square inflation factor @xmath219 for the particular case , @xmath223 . in this case
, the value of the quantity @xmath224 gives us the extent of stability of the test statistics under small contaminations with similar interpretation as that of the influence function .
whenever its value is bounded in @xmath69 , we get the stable test statistics under small contaminations and the magnitude of its supremum over all @xmath69 gives the extent of stability .
the following theorem gives an explicit form of this quantity ; the proof involves routine differentiation and is hence omitted .
[ thm:7chi_infl_fact_slope ] consider the above set - up with scalar parameter @xmath27 .
then we have , @xmath225,\end{aligned}\ ] ] where @xmath226 , @xmath227 and @xmath228 are independent of the contamination point @xmath69 .
let us now consider the most popular but simple hypothesis testing problem regarding the mean @xmath27 of an univariate normal with known variance @xmath229 ; so @xmath230 is the @xmath231 density with @xmath27 being the parameter of interest . based on a sample @xmath1 of size @xmath202 from this population ,
we want to test the simple hypothesis @xmath232 against the omnibus alternative .
let @xmath26 denote the mdpde of @xmath27 under this set - up . then , using the form of the normal density , the sdt has the simple form given by @xmath233 , ~~ a , b \neq 0,\ ] ] where @xmath234 .
whenever one of @xmath235 or @xmath236 is zero , the test statistic has the limiting form given by @xmath237 further note that at @xmath238 , we have @xmath239 , @xmath14 and @xmath240 , the mle of @xmath27 ; thus the sdt statistic becomes @xmath241 which is the ordinary likelihood ratio test statistic for the problem under consideration .
now , the asymptotic null distribution of the above sdt follows from theorem [ thm:7asymp_null_one ] ; the normal density family can be seen to satisfy the required assumptions .
recall from @xcite that the asymptotic distribution of @xmath242 is asymptotically normal with mean @xmath243 and variance @xmath244 under @xmath245 .
then the asymptotic null distribution of the sdt statistics @xmath34 is the same as that of @xmath246 , where @xmath187 follows a @xmath247 distribution with one degree of freedom and @xmath248 ; in other words , @xmath249 in particular when @xmath250 , we have @xmath251 as expected . further , we can compute an approximation to the asymptotic power of the above test at any alternative point @xmath252 using theorem [ thm:7aprox_power_one ] . to determine the power in the spirit of the contiguous alternatives , we choose the alternatives @xmath253 and plot the corresponding approximate power in the figure [ fig:7approx_power_comparison ] for @xmath254 , @xmath255 , @xmath256 , @xmath257 and different sample sizes @xmath258 .
it is clear from the figures that the approximate power of the proposed sdt is almost the same for different values of @xmath18 whenever the sample sizes are large enough or the values of the parameters @xmath259 is close to one ; note that the second case makes the the @xmath0-divergence measure to be almost independent of @xmath18 ( it is completely independent at @xmath17 ) .
however , for the small sample sizes with smaller values of @xmath259 , the approximate powers depend on @xmath18 significantly .
+ + + to explore how good this approximation is , we also perform a simulation study for exactly the same set - up of the normal model as above with 1000 replications and present the simulated powers in the same figure [ fig:7approx_power_comparison ] .
interestingly , the simulated powers turn out to be almost the same for different values of @xmath18 contradicting the corresponding results for approximate power in case of small samples sizes .
this illustrates that the power approximation derived above may not give the true picture about the power of @xmath0-divergence based test for such small sample sizes .
however , in case of large sample sizes the above power approximation works very well at the alternatives considered by us producing power values very close to the simulated power for any @xmath18 and @xmath254 .
note that , this fact is not very surprising as the above power approximation is only asymptotic in nature .
further , the simulated power at any @xmath259 values are also very much close to that obtained through the expression of the asymptotic power of sdt against contiguous alternative hypotheses in corollary [ cor:7cont_power_one ] ; the corresponding values are presented in table [ tab:7cont_power_norm_mu ] .
in fact , it follows from corollary [ cor:7cont_power_one ] that the asymptotic distribution of the sdt statistic under contiguous alternative hypotheses @xmath85 is the same as that of @xmath260 , where @xmath192 @xmath187 follows a non - central @xmath247 distribution with one degrees of freedom and non - centrality parameter @xmath261 ; hence its asymptotic power under contiguous alternatives @xmath85 turns out to be ( see remark [ rem:7sdt0_pif_1 ] ) @xmath262 where @xmath263 denotes the distribution function of @xmath192 and @xmath189 is the upper @xmath190 quantile of the central chi - square distribution with one degree of freedom .
note that the asymptotic power against the contiguous alternative hypotheses are independent of the parameter @xmath18 and decreases slightly as the parameter @xmath259 increases from @xmath243 to @xmath60 .
this shows that there is a loss in efficiency of the sdt with respect to the lrt at the pure data ; however the loss is not as significant to offset its strong robustness properties .
l| r r r r r r @xmath254 & 0 & 0.1 & 0.3 & 0.5 & 0.7 & 1 + power & 0.89 & 0.88 & 0.86 & 0.83 & 0.79 & 0.72 + [ tab:7cont_power_norm_mu ] now , we study the extend of robustness of the sdt in the present scenario .
we first consider the asymptotic results proved in section [ sec:7sample1_simpletest_robust ] , where we have seen that first order influence function of the sdt statistic and any order influence function of its level will be zero under any model .
therefore , we consider the second order influence function of the sdt statistic ; under the normal model it can be simplified as @xmath264 note that the above influence function is bounded whenever @xmath80 and unbounded at @xmath81 .
thus the sdt statistic should be always robust if we use a robust mdpde with @xmath80 ; but the use of the non - robust mle will make the test statistic non - robust also
. further the maximum possible value of the above influence function can be seen to be independent of parameter @xmath18 and decreases as the values of @xmath259 increases .
thus , the extent of the robustness of the sdt increases with the values of @xmath259 but remains unaffected as a function of @xmath18 .
figure [ fig:7if_normal_mu ] show the influence function of sdt for @xmath256 and @xmath255 .
we can also compute the power influence function of the sdt numerically for the normal model ; one particular case with @xmath256 , @xmath255 and @xmath265 is shown in figure [ fig:7if_normal_mu ] for the @xmath266 level of significance . from the robustness perspective , the inference derived from this power influence function is again similar to that from the if of the sdt statistic in figure [ fig:7if_normal_mu ] . finally to see the small sample robustness properties of the sdt and compare them with the nature of the ifs derived above ,
we have undertaken a simulation study based on @xmath267 replications with contaminated samples of several types .
for the simulation purpose , we have assumed @xmath256 and @xmath255 . the empirical size and power of sdt under some interesting contamination scenarios
are presented in figures [ fig:7cont_size_normal_mu][fig:7contamination_power_exact_normal_mu ] . in figure [ fig:7cont_size_normal_mu ]
we have presented the empirical size of the sdt for sample size @xmath268 at three different contamination levels ( @xmath269 ) when the data generating distribution is @xmath270 . to give a description of the full range of the observed sizes which may not be fully discernible from the figure
, we present the actual values of the sizes at some specific ( @xmath271 ) combinations in table [ tab:7sizes ] .
clearly , it follows that the size of the sdt remains more stable and closer to the nominal level of 5% for the larger values of @xmath259 compared to its smaller values like 0 or 0.1 . for small @xmath259 the empirical size of the sdt
is seen to somewhat depend on the values of @xmath18 ; the @xmath18 values closer to zero yields relatively more stable size closer to the nominal level of the test .
indeed , there are many members of the class of sdt that give more robust size compared to the dpd based test for testing the simple hypothesis under consideration .
+ l l| r r r r r r r & & + @xmath259 & @xmath67 & @xmath272 & @xmath273 & @xmath274 & 0 & 0.5 & 1 & 3 + 0 & 0 & 0.064 & 0.054 & 0.053 & 0.054 & 0.056 & 0.059 & 0.078 + & 0.05 & 0.092 & 0.078 & 0.077 & 0.078 & 0.079 & 0.082 & 0.111 + & 0.10 & 0.149 & 0.133 & 0.133 & 0.133 & 0.136 & 0.138 & 0.161 + 0.1 & 0.00 & 0.058 & 0.047 & 0.047 & 0.047 & 0.051 & 0.053 & 0.070 + & 0.05 & 0.087 & 0.081 & 0.080 & 0.081 & 0.081 & 0.082 & 0.097 + & 0.1 & 0.143 & 0.123 & 0.121 & 0.123 & 0.126 & 0.129 & 0.158 + 0.3 & 0 & 0.057 & 0.049 & 0.049 & 0.049 & 0.050 & 0.052 & 0.061 + & 0.05 & 0.076 & 0.069 & 0.069 & 0.069 & 0.069 & 0.072 & 0.080 + & 0.10 & 0.128 & 0.119 & 0.119 & 0.119 & 0.123 & 0.123 & 0.134 + 0.5 & 0 & 0.058 & 0.053 & 0.053 & 0.053 & 0.053 & 0.055 & 0.060 + & 0.05 & 0.068 & 0.065 & 0.065 & 0.065 & 0.066 & 0.066 & 0.070 + & 0.10 & 0.111 & 0.106 & 0.106 & 0.106 & 0.108 & 0.108 & 0.115 + 1 & 0 & 0.053 & 0.053 & 0.053 & 0.053 & 0.053 & 0.053 & 0.053 + & 0.05 & 0.066 & 0.066 & 0.066 & 0.066 & 0.066 & 0.066 & 0.066 + & 0.10 & 0.094 & 0.094 & 0.094 & 0.094 & 0.094 & 0.094 & 0.094 + [ tab:7sizes ] in figure [ fig:7contamination_power_contig_normal_mu ] ,
the power of the sdts are presented for the contiguous alternative hypotheses @xmath275 with @xmath276 under 10% contamination . in figures [ fig : power_cont_30_m1_1 ] and [ fig : power_cont_50_m1_1 ] , the true distribution is @xmath277 and the power are calculated at sample sizes @xmath278 and 50 respectively . on the other hand , in figures [ fig : power_cont_30_m2_1 ] and [ fig : power_cont_50_m2_1 ] the true distribution is @xmath279 and the powers are again computed at sample sizes 30 and 50 respectively .
the powers are practically constant over @xmath18 , although there is some slight variation over @xmath18 for small @xmath259 ( @xmath280 ) . also , the sdt with larger values of @xmath259 has quite high and stable power for any contamination scenarios compared to the lrt and other sdt with smaller @xmath259 values .
note that , although we have presented the results for only one particular value of @xmath257 , we have done the power calculation for several other values of @xmath83 ranging from @xmath281 to @xmath282 . in general , the overall representation is similar to figure [ fig:7contamination_power_contig_normal_mu ] , but the actual power increases slightly with @xmath83 . in figure
[ fig:7contamination_power_exact_normal_mu ] , we present the power of the sdt against the fixed alternative @xmath283 for sample sizes @xmath278 ( in the first column ) and @xmath284 ( in the second column ) under @xmath285 contamination .
we have considered three contamination distributions ( in three rows ) , namely @xmath286 , @xmath287 and @xmath288 .
clearly , for all the contamination scenarios the power of the sdts against the fixed alternative are almost one for @xmath289 .
however , the power of the sdts with smaller values of @xmath254 ( including lrt ) decreases from around 0.72 to 0.37 as the contamination distribution ( with means @xmath290 to @xmath291 ) moves away from the true distribution ( @xmath292 ) .
+ thus , the proposed sdt with larger @xmath259 gives us very useful alternatives to the lrt .
they have more stable size and contiguous power than the latter test and also have satisfactory power against any fixed alternative under contamination .
we will consolidate these empirical findings along with necessary theoretical results to suggest a suitable range of the tuning parameters @xmath18 and @xmath259 for practical usage in section [ sec : choice_tuning ] .
+ + .values of @xmath293 for different @xmath67 and @xmath294 under normal model with @xmath256 [ cols= " < ,
< , > , > , > , > , > , > " , ] [ tab:7chi - sq_infl ] finally , we complete this section with an illustration for the practical implication of our third robustness measure , the chi - square inflation factor ( derived in subsection [ sec : chi - square - inflation ] ) , studied here under the normal model . as the null mean is @xmath295 and the variance @xmath229 is known , we can easily compute the values of the the chi - square inflation factor @xmath296 under the contaminated model @xmath297 as defined in subsection [ sec : chi - square - inflation ] .
however , note that the chi - square inflation factor of the sdt is independent of the parameter @xmath18 as shown in theorem [ thm:7chi_infl_fact_slope ] .
tab:7chi - sq_infl ] presents the values of @xmath293 for different contamination proportions @xmath67 at the point @xmath298 for various values of @xmath294 and @xmath254 .
as mentioned earlier , larger the value of @xmath293 relative to @xmath60 , more liberal will the confidence interval based on the sdt be ; @xmath299 generates the confidence interval with the correct level of significance .
therefore , it follows from the table [ tab:7chi - sq_infl ] that the sdt with @xmath300 is highly non - robust producing extremely liberal confidence intervals even at small contamination proportions like @xmath301 .
however , the sdts with larger @xmath259 remain stable even under higher contaminations like @xmath302 or @xmath303 .
the chi - square inflation factor also depends on the parameters @xmath69 and @xmath294 ; note that @xmath3 is fixed by the null hypothesis .
also , it is clear from the theory discussed in section [ sec : chi - square - inflation ] that the effect of the contamination increases as its mass moves away from the true distribution , i.e. , as the values of @xmath69 increases in magnitude for the present case . to examine the effect of @xmath294 ,
the results for 3 particular values of @xmath294 are presented in table [ tab:7chi - sq_infl ] .
clearly , for any fixed contamination proportion @xmath67 at the point @xmath69 , the sdt generates more accurate confidence interval as the value of @xmath294 increases .
this fact is quite intuitive because the effective distance between the contamination distribution and the true distribution decreases as @xmath294 increases and hence the effect of contamination reduces .
further , for the normal model the slope of the chi - square inflation factor at the infinitesimal contamination can be computed easily using theorem [ thm:7chi_infl_fact_slope ] and has the simplified form given by @xmath304 again , it is clear from the above expression that the slope of the chi - square inflation factor remains bounded with respect to the contamination point for all @xmath80 implying the stability of the corresponding sdt . on the other hand , at @xmath81
it becomes unbounded in @xmath69 which further illustrates the non - robust nature of the corresponding sdt as observed above .
in previous sections , we have illustrated the performance of the proposed sdt both theoretically and empirically for different values of the tuning parameters @xmath305 and @xmath18 .
so , some guidance about the actual choice of the divergence within the large class of available members of the @xmath0-divergence family are necessary here .
the theoretical robustness of the proposed sdt is seen to depend solely on the mdpde used through the parameter @xmath28 ; as @xmath28 gets larger , the robustness increases .
however , the empirical power and size under contamination are seen to depend on @xmath18 at smaller values of @xmath306 though their actual values are unstable for such smaller values of @xmath259 . on the other hand , for larger values of @xmath307 ,
both the empirical power and size are quite stable and almost independent of @xmath18 ; in this region the powers are found to be satisfactory high values and sizes are quite close to their nominal levels .
however , a large value of @xmath254 leads to a loss in the power of test under pure data as we have seen in table [ tab:7cont_power_norm_mu ] ; but the loss is not very significant even at @xmath308 . besides the sdt does not necessarily have satisfactory sizes for all values of @xmath18 the observed sizes are generally close to nominal ones only for @xmath18 values near zero of the negative side .
therefore , our empirical calculations , along with our theoretical findings , indicate that the preferable region of tuning parameter combination would approximately be the rectangle @xmath309 $ ] and @xmath310 $ ] .
empirically , we have observed that this is the preferable region in the sense that they maintain levels close to the nominal level and exhibit reasonably high powers .
tests that can not maintain the nominal level are usually of little practical value as one does not know whether a higher observed power is an actual phenomenon or the consequence of an unstable size .
so , we do not advocate the use of such tests which have highly variable sizes in small sample .
although we have focused only on simple hypothesis testing throughout the present paper , it is worth noting that all the concepts and results derived for the simple null can be generalized for composite hypothesis testing .
this would indicate a wider scope for the proposals in this paper and their significance .
consider the set - up presented in the previous sections and let @xmath311 be a proper subset of the parameter space @xmath4 .
then , a composite hypothesis is given by @xmath312 note that , whenever @xmath313 contains only one element @xmath3 then the above composite hypothesis ( [ eq:7composite_hyp ] ) coincides with the simple null hypothesis ( [ eq:7simple_hyp ] ) . using a similar idea as in section [ sec:7sample1_simpletest ] , we can also construct a general family of test statistics based on the @xmath0-divergence measures .
suppose that @xmath26 denote the unrestricted mdpde of @xmath27 and @xmath314 denote the restricted mdpde obtained by minimizing the dpd with tuning parameter @xmath28 between the data and the model over the restricted subspace @xmath311 .
the asymptotic distribution and influence functions of the restricted mdpde can be found in @xcite and @xcite .
therefore , the general @xmath0-divergence based test for testing the composite hypothesis ( [ eq:7composite_hyp ] ) is given by @xmath315 note that if @xmath316 then @xmath317 for each @xmath28 and this general test statistic ( [ eq : sdt_composite ] ) coincides with our initial test statistics . all the results including the asymptotic distributions and robustness properties can easily be extended to the class of test statistics ( [ eq : sdt_composite ] ) ; the advantages are similar to the case of the simple null except for a few minor changes in the numerical values and constants due to the restrictions imposed by the null hypothesis . however , considering the length of the present paper , we have decided to consider the details of this extension in a future paper .
the excellent robustness properties of the tests of parametric hypotheses proposed by @xcite have already been empirically observed and heuristically argued for .
however , the literature remains incomplete unless the theoretical robustness properties of these tests are properly established .
these properties have been carefully assembled in this paper and the different theoretical properties of this testing procedure are now established .
the results theoretically conform the empirical observations of @xcite and make the credentials of the density power divergence test more complete .
we assume that the true density @xmath6 belongs to the model family with true parameter value @xmath3 , i.e. , @xmath318 and state the necessary conditions under this set - up .
* there is an open subset of @xmath319 of the parameter space @xmath4 , containing the true parameter value @xmath3 such that for almost all @xmath320 , and all @xmath321 , the density @xmath322 is three times differentiable with respect to @xmath27 . * the first and second logarithmic derivatives of @xmath230
satisfy the equations @xmath323
= 0,\ ] ] and @xmath324 = e_\theta\left[-\nabla^2 \log f_\theta(x)\right].\ ] ] * the matrix @xmath325 is positive definite with all entries finite for all @xmath321 , and hence the components @xmath326 are affinely independent with probability one .
* for all @xmath327 , there exists functions @xmath328 with finite expectation ( under true distribution ) such that @xmath329 * the support of the distribution function @xmath330 , i.e. , the set @xmath331 is independent of @xmath27 .
* there is an open subset of @xmath319 of the parameter space @xmath4 , containing the true parameter value @xmath3 such that for almost all @xmath320 , and all @xmath321 , the density @xmath322 is three times differentiable with respect to @xmath27 and the third partial derivatives are continuous with respect to @xmath27 .
* the integral @xmath332 can be differentiated three times with respect to @xmath27 , and the derivatives can be taken under the integral sign . *
the @xmath333 matrix @xmath334 , defined in section [ sec:7sample1_simpletest ] , is positive definite .
* there exist functions @xmath335 with finite expectation under the true distribution such that @xmath336 where @xmath337 is the density function of the degenerate distribution at @xmath338 .
the first two conditions ( d1)(d2 ) relate to our parametric assumptions , which are routine in asymptotic derivations and are satisfied by most parametric models .
the last three conditions depend both on the structure of the density power divergence as well as the value of @xmath28 and are necessary to establish the required asymptotics .
it is not difficult to verify that these conditions are satisfied by common parametric models like the normal , exponential , poisson , geometric etc . for all @xmath339 .
* acknowledgement : * + this research is partially supported by grant mtm 2012 - 331 - 40 .
the authors also gratefully acknowledge the comments of a referee which led to an improved version of the manuscript .
ghosh , a. , harris , i. r. , maji , a. , basu , a. , pardo , l. ( 2013 ) . a generalized divergence for statistical inference . , * biru/2013/3 * , bayesian and interdisciplinary research unit , indian statistical institute , kolkata , india . | the most popular hypothesis testing procedure , the likelihood ratio test , is known to be highly non - robust in many real situations .
@xcite provided an alternative robust procedure of hypothesis testing based on the density power divergence ; however , although the robustness properties of the latter test were intuitively argued for by the authors together with extensive empirical substantiation of the same , no theoretical robustness properties were presented in that work . in the present paper
we will consider a more general class of tests which forms a superfamily of the procedures described by @xcite .
this superfamily derives from the class of @xmath0-divergences recently proposed by @xcite . in this context
we theoretically prove several robustness results of the new class of tests and illustrate them in the normal model .
all the theoretical robustness properties of the @xcite proposal follows as special cases of our results .
hypothesis testing , robustness , @xmath0-divergence . |
one of the most interesting topics within modern condensed matter physics is that of phenomena in reduced dimensionality , resulting from some degree of spatial localization of the particles comprising a system.@xcite for example , chemists , materials scientists and physicists have created and explored numerous physical systems in which atoms and molecules are confined within quasi - one dimensional ( q1d ) environments .
the variety of these systems is remarkable , such as the peapod geometry , _ i.e. _ , a line of buckyballs within a carbon nanotube,@xcite fluids within artificial materials created by templating @xcite and q1d optical lattices created by laser fields.@xcite unfortunately , as far as we know , there exists no comprehensive review of this general problem , although many relevant subfields have been summarized.@xcite the present paper addresses a small subset of this exciting research field .
specifically , we consider problems involving fluids , both classical and quantum , confined within q1d channels , the focus of our group s research during the last decade.@xcite here , the term q1d refers to a system in which particles move in an external potential field @xmath0 which is either constant or slowly varying in _ one _ direction ( @xmath1 ) , while @xmath2(*r * ) is strongly localizing in the two other ( transverse ) directions . in the case of quantum particles , for which the transverse spectrum is discrete , one expects that the corresponding degrees of freedom are frozen out at low temperature ( @xmath3 ) ; transverse excitation does occur at higher @xmath3 , as determined by the gaps in the transverse spectrum of states .
this plausible expectation is borne out in some cases , but we shall see that there can be dramatic consequences of the transverse degrees of freedom in other cases , even at low @xmath3 .
one of the many exciting aspects of strictly 1d physics is its susceptibility to weak perturbations .
the reason for this behavior arises from the fact that ( in all practical situations)@xcite no phase transition can exist in a purely 1d system at any finite @xmath3 , even though the ground state may exhibit symmetry - breaking order . thus , there do exist transitions at @xmath3 identically equal to zero .
a familiar example is the 1d ising model , for which the correlation length and susceptibility both diverge as @xmath3 approaches 0 . since the purely 1d system has no finite @xmath3 transition , what brings about more interesting behavior in the q1d case ? as discussed below , the difference can arise from considering a set of parallel 1d systems which are weakly coupled .
alternatively , the behavior can happen because the system is only 1d insofar as the transverse dimensions are finite , unlike the length in the z direction , l , which achieves the thermodynamic limit ..... but the transverse dimensions _ are _ large enough to have an observable effect ( e.g. low energy gaps ) .
a third , more surprising , origin of interesting phenomena is when the system is a collection of noninteracting 1d systems , with `` quenched '' heterogeneity , which are coupled to a particle bath , so they possess a common chemical potential .
the next section discusses the conceptually and computationally simplest case of a q1d system : a low density , noninteracting gas within a single channel ; then , the solution of the one - particle schrdinger equation determines the physical behavior .
section 3 considers the case of many _ interacting _ particles within a single channel ; such a problem is often used as a model for fluids within regular or irregular porous materials .
section 4 considers the problem of q1d channels containing fluids that interact with one another as well as with fluids occupying other channels .
our first problem is conceptually and calculationally simple : a low density gas confined within a q1d geometry ; nevertheless , it provides interesting , and sometimes surprising , results .
one example that has been studied extensively is that of a gas inside single carbon nanotubes , _
e.g. _ quantum fluids or the `` peapod '' case of c@xmath4 molecules.@xcite another is the case of quantum gases inside templated regular pores.@xcite a third example is the so - called groove region between two nanotubes , _
e.g. _ on the outside of a bundle of nanotubes.@xcite a _ classical _ noninteracting gas has a kinetic energy per particle of ( 3/2 ) @xmath5 and a mean potential energy @xmath6 determined by its interaction with the environment . for the case of a particle localized near the z axis within a channel , @xmath7 , due to two transverse directions of excitation .
hence , the classical specific heat per particle is @xmath8_{classical } = ( 5/2 ) k_b$ ] . for a quantum gas , instead
, the transverse degrees of freedom are frozen out at low @xmath3 , so one expects @xmath8_{quantum } = ( 1/2 ) k_b$ ] .
the generalization to @xmath9 `` effective '' dimensions yields this expression for the dimensionless specific heat , @xmath10 , of a noninteracting boltzmann gas : @xmath11 fig .
[ fig : exp_c ] shows experimental results for @xmath10 in the case of @xmath12he inside of fsm-16 , a material consisting of straight hexagonal pores of cross - sectional distance of order 2 to 3 nm , pre - coated with a thin film of @xmath13he . at the low densities shown here
, the @xmath12he gas can be assumed to be noninteracting , although interaction effects appear at higher density.@xcite the behavior observed inside fsm-16 can be understood by analogy to calculations in fig . [
fig : theor_c ] , for a noninteracting gas of @xmath13he within a carbon nanotube .
note the overall similarity of these two figures .
he atoms within a single wall carbon nanotube , for cases @xmath14=0.8 nm ( dashes ) and 0.5 nm ( full curve ) .
quantum statistical effects are omitted from the calculation . adapted from ref .
@xcite.,height=336 ] in fig .
[ fig : theor_c],@xcite one observes that the low @xmath3 limit is @xmath15 , corresponding to a 1d classical gas , as expected . at high @xmath3 , instead , the limit is @xmath16 ; this limit is interpreted as that of a 2d gas moving on the inner surface of the nanotube .
the bump at intermediate @xmath3 is a general property found for spectra , such as that of the rigid rotor,@xcite for which the inter - level spacing increases with quantum number . in the present case , the relevant spectrum is that arising from the azimuthal kinetic energy , @xmath17 here @xmath18 is the azimuthal quantum number and @xmath19 is the radial coordinate , while @xmath20 is defined as a temperature characteristic of azimuthal excitation .
the peak in the specific heat occurs at a temperature near 3 @xmath20 for both radii considered in fig .
[ fig : theor_c ] , so its position serves as a benchmark from which one can determine the value of @xmath21 . note that inside a nanotube , the relevant value of @xmath21 is typically @xmath22 , where @xmath23 is the gas - surface hard core interaction length . the key qualitative difference between the behaviors seen in figs .
[ fig : exp_c ] and [ fig : theor_c ] is that the experimental data in fig .
[ fig : exp_c ] plunge to @xmath24 as @xmath25 .
this is an effect of quantum degeneracy , manifested in nernst s law , as found in explicit calculations which revise fig .
[ fig : theor_c ] by taking quantum statistics into account.@xcite there have been many theoretical studies of gas adsorption in the presence of nanotubes.@xcite in most treatments of these systems , one assumes that neighboring tubes are parallel . in that case , there exists a region of space- the so - called `` groove''- which is a 1d channel with a strongly attractive potential , created by the adjacent tubes .
the adsorbed gas then exhibits q1d behavior at low @xmath26 .
however , one can also inquire about the case when the tubes are not quite parallel , but instead diverge , leaving a particularly attractive region between them , with a minimum potential energy ( @xmath27 ) located at equilibrium position @xmath28 . in a forthcoming study,@xcite we will report remarkable results for this geometry , as exemplified in fig . [
fig : groove_c ] .
the low @xmath3 behavior is that of a gas in a 1d harmonic potential , @xmath29 , where @xmath30 is the force constant for particle motion parallel to the @xmath1 axis , midway between and nearly parallel to the tubes axes .
we introduce the characteristic temperature for this motion , @xmath31 and a reduced temperature @xmath32 . for @xmath33 ,
the specific heat is of arrhenius form , as seen in the figure , while for @xmath34 , @xmath35 , the specific heat of a 2d gas .
that behavior might not have been anticipated , at first glance , because for small divergence half - angle @xmath36 , one might have expected 1d behavior , i.e. , @xmath37 .
another surprise is the high @xmath38 limiting behavior , @xmath39 .
this peculiar result arises because the relevant particles motions are those in the plane perpendicular to the @xmath40 plane of the nanotubes , for which the potential variation is unusual- proportional to @xmath41 .
( solid - red ) and ne ( short - dashed - black ) between two nearly parallel nanotubes , each with @xmath42 nm , and a divergence half - angle @xmath43 degrees . from ref .
from the perspective of phase transitions , a channel of finite width is a q1d system , so that no true thermodynamic singularities can occur .
this means that many attempts to explore behavior in porous media with single channel models ( such as cylindrical and slit pores ) can not accurately describe phase transition behavior that is seen in genuinely 3d porous media
. nevertheless , these models may provide good semiqualitative predictive power , sufficient for most purposes since the fully 3d geometry is not known .
we have explored a variety of such models , with several different goals .
these include assessing the accuracy of simplifying models , such as mean - field - theory ( mft ) and the use of periodic boundary conditions .
both of these approximations are suspect , at first glance , due to the important role of fluctuations in 1d statistical physics .
[ fig : isotherms ] presents adsorption isotherms for uptake in a single cylindrical pore , described by a lattice gas model.@xcite in this model , continuous space is discretized into adsorption sites , which may be either occupied or empty in any microstate of the system . in the specific model used here ,
the set of sites ( within one transverse section ) consists of one axial site and seven `` cylindrical shell '' sites , corresponding to positions near the inner boundary of the nanotube .
the ensemble of sites include an infinite sequence of such layers of eight sites .
the energy of the system includes interactions between particles occupying these sites , plus interactions between particles and the substrate host .
[ fig : isotherms ] compares ( numerically ) exact results with those obtained from mft .
the results are quite similar , overall , given the highly expanded scale of reduced chemical potential ( @xmath44 ) .
note that the spurious transition seen in mft ( at reduced temperature @xmath45 ) is not very different from the nearly discontinuous isotherm seen in the exact results . as a function of the transverse interaction .
perturbation theory results and predictions from the anisotropic lattice gas model are compared . from ref .
consider a geometry consisting of a set of parallel q1d fluids , as in gases within or between nanotubes comprising a bundle of such tubes . while no transition can occur for an _ isolated _ q1d system , once coupling between such systems is present
, a finite temperature transition can occur .
[ fig : tc ] presents results for such a geometry.@xcite the critical temperature @xmath46 is shown as a function of the transverse coupling @xmath47 , for the case of buckyballs confined within parallel nanotubes . as indicated , rather similar results were obtained from exact solutions for a lattice - gas model and from a perturbation theory . in the latter case ,
the unperturbed equation of state was the exact 1d result for a model with lennard - jones interactions between the buckyballs and the perturbation was the weak van der waals interaction between balls in adjacent tubes . particularly striking in this figure
is the singular behavior for small @xmath47 , as blown - up in the inset .
this behavior is well - known for the lattice - gas , for which the transition temperature satisfies ( for weak coupling)@xcite @xmath48}\ ] ] here @xmath49 is the longitudinal interaction and @xmath50 is the anisotropy ratio of transverse to longitudinal interaction strengths .
the singular behavior reflects the divergent susceptibility of the 1d system at low t ; as the correlation length diverges , larger regions of fluid in adjacent channels are coupled , so increases rapidly with increasing @xmath51.the other notable result in this formula and the figure is that the characteristic energy and @xmath46 scale are given by the _ longitudinal _
coupling , which is about 500 k for the peapod case , whereas naively one might have expected to be proportional to @xmath51 .
we are grateful to doe and nsf for support of this research and to s. j. full and n. wada for help with figures .
j. vavro , m. c. llaguno , b. c. satishkumar , d. e. luzzi , and j. e. fisher , _ appl .
lett . _ * 80 * , 1450 ( 2002 ) ; d. e. luzzi , b. w. smith , _ carbon _ * 38 * , 1751 ( 2000 ) ; d. j. hornbaker et al .
, _ science _ * 295 * , 828 ( 2002 ) .
we exclude the case of an infinitely long - range interaction , for which mean - field theory correctly predicts a transition in 1d .
see m. kac , g. e. uhlenbeck and p. c. hemmer , _ j. math .
phys . _ * 4 * , 216 ( 1963 ) . | fluids confined within narrow channels exhibit a variety of phases and phase transitions associated with their reduced dimensionality . in this review paper
, we illustrate the crossover from quasi - one dimensional to higher effective dimensionality behavior of fluids adsorbed within different carbon nanotubes geometries . in the single nanotube geometry
, no phase transitions can occur at finite temperature .
instead , we identify a crossover from a quasi - one dimensional to a two dimensional behavior of the adsorbate . in bundles of nanotubes , phase transitions at finite
temperature arise from the transverse coupling of interactions between channels . |
experiments on the orbital angular momentum [ oam ] of light confirmed recently an uncertainty principle for angular position and angular momentum @xcite . whereas for separable quantum states uncertainty principles limit the accuracy for measurements of non - commuting observables , inseparable or entangled states may apparently overcome these limits .
this was first discussed by einstein , podolsky and rosen in their famous _ gedankenexperiment _
@xcite and led to the formulation of the epr paradox @xcite .
the implications of the epr paradox have been tested mainly on optical systems , e.g. on the polarisation of photons @xcite , quadrature phase components @xcite or directly on the optical version of epr s original example , the linear momentum and linear position of photons @xcite .
the relation between oam and its conjugate variable , the angular position , is fundamentally different from these systems , because oam is a discrete quantum observable of infinite dimension and the angular position is continuous and bounded .
the entanglement for oam of photon pairs generated in parametric down conversion has been confirmed both experimentally @xcite and theoretically @xcite .
it is therefore interesting to examine the possibility to demonstrate an ` angular ' epr paradox for this pair of observables , in particular as the necessary experimental techniques have already been employed in recent work in this field @xcite .
apart from the fundamental interest in angular epr correlations , the criterion for an epr paradox for oam and angular position of light provides a tool to characterise entanglement for these observables . using variances in violations of local uncertainty relations as entanglement criteria
has received renewed interest .
formerly applied to continuous variables on a specific example @xcite , it has been shown that variances of special observables can be used to detect entanglement in finite - dimensional system @xcite .
this approach has been generalised to arbitrary observables and led to entanglement criteria for finite dimensional systems @xcite which are known from the continuous variable regime @xcite .
a practical motivation for studying entanglement of optical oam arises from the possible advantages of oam in quantum information processes @xcite and quantum communication , as cryptographic schemes @xcite could profit from an enlarged basis of states .
oam of light is connected with the azimuthal phase structure of light beams : each photon in a beam with phase dependence @xmath0 carries an oam of @xmath1 @xcite .
the oam number @xmath2 can take on any integer number which leads to a discrete , infinite dimensional quantum system .
the conjugate variable is the angular or azimuthal position @xmath3 , which we choose to lie within the @xmath4 radian interval starting at angle @xmath5 .
the associated uncertainty relation is then @xcite @xmath6 where @xmath7 and @xmath8 is the angular probability density at the boundary of the chosen interval .
the topology of the basis sets is reflected in their fourier relation ; in contrast to the linear case a discrete fourier transform allows us to change from oam representation to angle representation : @xmath9 here , @xmath10 is the wavefunction in the angle representation and @xmath11 the oam probability amplitude . the @xmath4 radian interval is commonly chosen to be @xmath12 , and we set @xmath13 from here on . it has been experimentally demonstrated that oam of light is conserved under parametric down conversion @xcite . in theoretical studies ,
it has been pointed out that the conservation of oam is related to the phase matching condition for parametric down conversion processes @xcite .
entanglement in oam and azimuthal position is a consequence of this phase matching @xcite .
the conservation of transverse momentum requires that the two - photon wavefunction for the signal ( 1 ) and idler ( 2 ) mode , for a plane wave pump , has to be of the form @xmath14 . using a simplified approach
one can argue that the transverse spatial correlations in the far field originate from the momentum conservation under parametric down conversion .
identifying transverse momentum components in the near field with spatial coordinates in the far field allows us to write the spatial dependence of the wavefunction in position representation as @xmath15 where @xmath16 and @xmath17 are the radial and azimuthal coordinates and @xmath18 is the @xmath4 periodic delta function . from this result
we can expect the azimuthal angles of the photons in the far field to obey @xmath19 , so that the signal and idler photons appear on opposite sides of their respective cones .
the correlation in angular momentum follows on writing eq .
( [ eq : anglecorr ] ) in terms of its angular fourier components : @xmath20 which is an entangled superposition of states with zero total oam .
a more detailed analysis which considers a specific parametric down conversion process shows that more complicated dependences of the wavefunction on the azimuthal angles are also possible @xcite .
the epr paradox describes the apparent violation of the uncertainty principle resulting from measurements on correlated , spatially separated systems .
the original epr argument considers correlations that are strong enough to predict or infer with certainty the value of the observables in one subsystem from measurements on the other , separated , subsystem without disturbing in any way the first subsystem .
the ability to predict with certainty the value of an observable defines , according to epr , an element of reality .
however , non - commuting observables , can not have a simultaneous reality , an expression of which is the uncertainty principle @xcite .
the tension between local elements of reality and quantum complementarity leads to the paradox .
the original epr _ gedankenexperiment _ considers an idealised situation .
the quantum state given by epr on the example of the position and momentum is in the modern language of entanglement a maximally entangled state @xcite , and the measurement of the observables is assumed to be infinitely precise . for oam and angular position this idealised setting would require a parametric down conversion process which creates an entangled photon pair , perfectly correlated in oam and angular position .
an errorless measurement of the oam on the signal photon could then be used to infer the oam of the idler photon , and an errorless measurement of the azimuthal angle on the signal photon would allow us to predict precisely the angle of the idler photon .
as these measurements on the signal photon ` do not disturb the idler photon in any way ' , the predictions would constitute simultaneous elements of reality for the oam and the azimuthal angle of the idler photon .
we stress that the possibility to predict observables of the idler photon with certainty depends on the ability to measure the observables on the signal photon without error .
in particular for a continuous observable a measurement with infinite precision can not be realised experimentally .
a typical experimental setup would allow us to determine whether a continuous variable falls into a previously specified range .
to analyse the possibility of demonstrating an epr paradox experimentally a more realistic situation has to be studied .
this requires the consideration of non - maximal correlations and of measurements with finite precision leading to an error in inferring one observable from a measurement on the other subsystem .
the size of this error determines whether the epr paradox can be demonstrated in the considered experimental setup @xcite .
the inclusion of experimental indeterminacies requires a reformulation of the epr paradox .
in contrast to the original epr setup @xcite , we consider conditional measurements on both subsystems . in reference to the idealised setup in section ( [ ssc : eprpara ] ) , we look at the variance of the oam or the azimuthal position in the idler beam given a specifically set outcome of a measurement in the signal beam . for the oam
this condition will be the measurement of a single @xmath21 value , whereas for the azimuthal position the measurement will be in a range of angles .
for the oam we denote the conditional variance with @xmath22 $ ] , i.e. the variance of @xmath23 in the idler under the condition that a measurement on the signal photon yields an oam of @xmath24 .
the condition for the azimuthal position to fall in a range of angles will be treated as a probability density @xmath25 .
the functional dependence of the probability density is given by @xmath26 and the variable @xmath27 is used to indicate the orientation . for a symmetric @xmath26 , @xmath27 would be the central angle .
the conditional variance for the angle can thus be written as @xmath28 $ ] .
for an actual experiment the error caused by non - maximum correlations is thus included in measured quantities and the theoretical modelling , therefore , concentrates on the description of the error in the measurement .
schemes to measure oam have been theoretically studied and experimentally realised @xcite .
we are therefore mostly concerned with a description of the angle measurement . in the following
we give the criterion for the experimental demonstration of an angular epr paradox as one main result of this work . in the previous section ( [ ssc : eprpara ] )
we pointed out the importance of inference in epr type arguments . to distinguish a measured quantity from an inferred one we label the first with an index @xmath29 and the latter with an index @xmath30 .
the measured quantities are the conditional probability for the oam @xmath31_{\mathsf{m}}$ ] and the conditional probability density for the azimuthal angle @xmath32_{\mathsf{m}}$ ] .
the paradox becomes now apparent if one assumes local realism . from the measured variance @xmath32_{\mathsf{m}}$ ] a minimum variance @xmath33_{\mathsf{i}}$
] can be derived which is still in accordance with the uncertainty relation [ cf .
( [ eq : oamur ] ) ] .
this quantity can be compared to the measured variance @xmath22_{\mathsf{m}}$ ] as the conditioning measurement on the first subsystem ` does not have an instantaneous influence on the second subsystem ' .
an angular epr paradox would then be demonstrated if @xmath34_{\mathsf{m } } \rangle_{m_1 } < \langle \min { \mathrm{var}}[m_2 | p_1(\phi_1;\tau_1)]_{\mathsf{i } } \rangle_{\tau_1}.\ ] ] within the simplified reasoning in section [ ssc : oam ] the correlations in oam are uniform and the correlations in angle isotropic .
this can not be assumed a priori for an experimental test of the criterion eq .
( [ eq : paracrit ] ) . we are therefore considering averaged conditional variances , which take into account that the correlations may vary for different values of @xmath24 and for different orientations of the angle @xmath27 . in the quantity @xmath35_{\mathsf{m } } \rangle_{m_1}$ ] the conditional oam variance @xmath22_{\mathsf{m}}$ ] is averaged over the condition @xmath24 . to find the average @xmath36_{\mathsf{i } } \rangle_{\tau_1}$ ] the minimum conditional variance @xmath33_{\mathsf{i}}$ ] is integrated over @xmath27 on a @xmath4 radian interval with the probability density @xmath37 .
formally the averaging procedure eliminates the dependence of the criterion on particular values of @xmath24 and @xmath27 , but not the dependence on the functional form of @xmath26 .
the dependence on @xmath26 originates from the way in which the condition in eq . ( [ eq : paracrit ] ) is measured and is not directly connected to the correlation in the azimuthal angle . with the usual definition of the variance in terms of probabilities
, we can reformulate the paradox statement : @xmath38_{\mathsf{m } } m_2 ^ 2 - \left ( \sum_{m_2 } p[m_2 | m_1]_{\mathsf{m } } m_2 \right)^2 \right ] \nonumber \\ & & < \int_{-\pi}^\pi d\tau_1 p_1(\phi_1;\tau_1 ) \min \left [ \sum_{m_2 }
p[m_2 | p_1(\phi_1;\tau_1)]_{\mathsf{i } } m_2 ^ 2 \right . \\ & & \qquad \qquad - \left .
\left ( \sum_{m_2 } p[m_2 | p_1(\phi_1;\tau_1)]_{\mathsf{i } } m_2 \right)^2 \right ] \nonumber.\end{aligned}\ ] ] the inferred conditional probability @xmath39_{\mathsf{i}}$ ] will be calculated via a fourier transform from @xmath32_{\mathsf{m}}$ ] and is given by the modulus square of the conditional probability amplitudes @xmath40_{\mathsf{i}}$ ] .
an experimental result which obeys the given inequality [ eq .
( [ eq : paracrit ] ) ] would constitute a demonstration of an angular epr paradox .
the measurement scheme to determine the azimuthal position models experimental techniques employed in recent work @xcite .
a photon is said to have a specific angular probability density if it is detected after passing an angular aperture corresponding to this probability density [ see fig .
[ fig : anglemeas ] ] . with the help of spatial light modulators
smooth aperture functions can be realised .
= a narrow aperture can be used to measure the probability density of the azimuthal position .
the aperture could then be rotated , i.e. the central angle could be varied over a @xmath4 radian range .
detecting the number of photons passing the narrow aperture as a function of the central angle yields eventually a measure for the probability distribution .
the angular apertures will also be used to set the condition in the signal beam .
a fixed aperture can be inserted in the signal beam , and a rotatable aperture can be used to measure the conditional probability density in the idler beam [ see fig .
[ fig : condmeasure ] ] .
= 0.6 we will denote the outcome of such a measurement by @xmath41_{\mathsf{m}}$ ] , i.e. the probability density to detect an idler photon with angle @xmath42 under the condition that the entangled photon in the signal beam passes the aperture @xmath43 oriented at @xmath27 . as the aperture can be described by an aperture function , which we assume translates into a probability density @xmath44 , we will write synonymously @xmath45_{\mathsf{m}}$ ] . the importance of the orientation angle @xmath27 has been stressed in the formulation of the paradox [ cf .
section [ ssc : criterion ] ] . to simplify the notation
we will not write the explicit dependence of @xmath46 on @xmath27 from here on . from the conditional probability density a class of wavefunctions in the angle representation @xmath47 $ ]
is derived , @xmath48 = \sqrt { p[\phi_2 | p_1(\phi_1 ) ] _ { \mathsf{m } } } \exp({\mathrm{i}}\alpha(\phi_2)).\ ] ] here , the phase @xmath49 is undetermined , as the measured probability densities only give the modulus square of the wavefunction .
the wavefunction is then transformed into a conditional oam probability amplitude via a fourier transform [ cf .
( [ eq : fouriertrans ] ) ] : @xmath50_{\mathsf{i } } = \frac{1}{\sqrt{2\pi } } \int_{-\pi}^\pi d\phi_2 \exp({\mathrm{i}}m \phi_2 ) \sqrt{p[\phi_2 | p_1(\phi_1 ) ] _ { \mathsf{m } } } \exp({\mathrm{i}}\alpha(\phi_2)).\ ] ] from the conditional probability amplitudes we can calculate the conditional variance @xmath51_{\mathsf{i}}$ ] by taking the sum over all @xmath23 values @xmath52_{\mathsf{i } } = \min \left [ \frac{\sum_{m_2 }
c[m_2 | p_1(\phi_1)]^2_{\mathsf{i } } l^2 } { \sum_{m_2 } c[m_2 | p_1(\phi_1)]^2_{\mathsf{i } } } - \left [ \frac{\sum_{m_2 } c[m_2 | p_1(\phi_1)]^2_{\mathsf{i } }
l } { \sum_{m_2 } c[m_2 | p_1(\phi_1)]^2_{\mathsf{i } } } \right]^2 \right].\ ] ] this is the inferred minimum variance which can be compared to the measured quantity @xmath53_{\mathsf{m}}$ ] . the phase @xmath49 will be determined by the minimization of the conditional variance @xmath54_{\mathsf{i}}$ ] as detailed in the following section .
calculating the conditional wavefunction from the conditional probability leaves the phase @xmath49 undetermined .
we find that if the variance is calculated for @xmath55 then the minimum variance @xmath56_{\mathsf{i}}$ ] is obtained . to show this we assume the conditional wavefunction to be of the form @xmath57 = a \exp[{\mathrm{i}}\alpha(\phi_2)],\ ] ] where @xmath58{|\psi[\phi_2 | p_1(\phi_1)]|^2}$ ] is a positive , real function , which is periodic in @xmath42 .
applying the orbital angular momentum operator to the wavefunction yields @xmath59 & = & -{\mathrm{i}}\hbar \psi'[\phi_2 | p_1(\phi_1 ) ] \nonumber \\ & = & -{\mathrm{i}}\hbar a'(\phi ) \exp [ { \mathrm{i}}\alpha(\phi_2 ) ] + \hbar \alpha'(\phi_2 ) \exp[{\mathrm{i}}\alpha(\phi_2)],\end{aligned}\ ] ] where the primes denote derivatives with respect to @xmath42 . using the periodicity of @xmath60 the variance of @xmath61
can be evaluated to @xmath62 where we used the fact that @xmath63_{\mathsf{m}}$ ] .
the first integral is the variance of @xmath61 for @xmath55 , while the second and third integral are the variance of @xmath64 : @xmath65_{\alpha = 0 } + \hbar^2 { \mathrm{var}}\alpha'.\ ] ] therefore we obtain the minimum variance if the wavefunction @xmath66 $ ] is real and positive , so that @xmath67 . in the following
we will consider only this case .
in order to measure @xmath68_{\mathsf{m}}$ ] we have to examine the signal photon for the particular oam @xmath24 and determine the oam @xmath23 of the idler photon . a schematic representation of an experimental setup to achieve this is shown in fig .
( [ fig : condexp1 ] ) .
a spatial light modulator is used to produce a hologram which changes the oam in the signal photon by the chosen value @xmath69 to zero @xcite . only beams with @xmath70
have on - axis intensity and can thus be detected behind a pinhole .
the measurement for the idler photon has to distinguish between different values of @xmath23 .
a sorting scheme which is able to determine the oam has been experimentally implemented @xcite .
a coincidence measurement would then yield @xmath22_{\mathsf{m}}$ ] .
= 0.8 experimentally it will not be possible to measure the conditional probability density for a single angle , instead a suitable aperture can be used to test for a range of angles , as described in section [ ssc : anglemeas ] .
analogously to the conditioning aperture , the measured quantity is @xmath71_{\mathsf{m}}$ ] , where the aperture @xmath72 is centred at a particular angle @xmath73 .
the aperture can be described by a probability density @xmath74 , where @xmath75 indicates the central angle .
the measured probability can thus be written as @xmath76_{\mathsf{m } } $ ] . for a very narrow aperture @xmath72
this measurement will give a good estimate of @xmath77_{\mathsf{m}}$ ] @xmath78_{\mathsf{m } } \approx p [ p_2(\phi_2;\tau_2 ) | p_1(\phi_1)]_{\mathsf{m } } \quad \mathrm{for}\:\,\mathrm{suitable}\:\ , p_2(\phi_2).\ ] ] the error made in this approximation is then given by the variance of @xmath42 for the probability density @xmath79 .
the scheme to measure @xmath80_{\mathsf{m}}$ ] is basically shown in fig .
( [ fig : condmeasure ] ) . with the help of a spatial light modulator [ slm ]
any chosen @xmath26 can be set as a condition for the signal . for the idler
the aim is to determine @xmath42 as exactly as possible . to achieve this an slm
could be programmed for a narrow angular aperture @xmath74 centred at @xmath73 , which would then be varied over the @xmath4 radian interval .
eventually this would lead to @xmath81_{\mathsf{m}}$ ] which does not only depend on the condition @xmath26 but also on the chosen analysing aperture @xmath82 .
this is the experimentally measurable estimate of the quantity @xmath83_{\mathsf{m}}$ ] used in the formulation of the angular epr paradox [ cf .
section ( [ ssc : criterion ] ) ] .
obviously there are experimental limitations : a narrow aperture would transmit only little intensity , which would reduce the detection rate in the conditional measurement . also slms have a finite size and resolution , which limits the ability to distinguish between similar apertures
. the interpretation of experimental data would require the inclusion of imperfect correlations originating from the parametric down conversion process and the influence of the optical elements , in particular the angular apertures . in the following we will discuss these aspects briefly .
the possibility to create photon pairs entangled in oam or angular position is based on the conservation of oam under parametric down conversion @xcite , which holds for thin down conversion crystals and in the paraxial limit .
the conservation leads to a perfect anti - correlation in the oam indices so that for a given oam index @xmath84 in the pump , signal and idler obey @xmath85 . in a recently reported down conversion experiment @xcite ,
the spatial correlations of photons entangled in orbital angular momentum have been studied . in this particular experiment signal and idler cone overlap completely and the spatial correlations are such that for a fixed detector position in the signal , the coincidence pattern in the idler shows two distinct spots equally separated in angle from the position exactly opposite the signal detector on the phase matching ring .
the vertex of the separation angle is on the pump axis . in our work
we are concerned with a different angle : the azimuthal position of a photon in a beam is measured from the beam axis , i.e. in a down conversion experiment from the signal and idler axis respectively . for non
degenerate down conversion crystals these two angles are not identical .
the question if they are compatible for the degenerate case is a very interesting one , but it will not be examined in the scope of this work .
therefore , for our theoretical modelling , we make use of the simplified reasoning given in section ( [ ssc : oam ] ) and therefore have an angle correlation @xmath86 in the far field [ cf . eq .
( [ eq : anglecorr ] ) ] .
a recent study on the epr paradox for linear optical momentum and position @xcite included a theoretical prediction of the conditional probabilities .
although this is certainly an interesting additional information for interpreting experimental data , the analysis here does not rely on it , as we would like to end with a criterion that can be applied directly to experimental data .
the effect of imperfect correlations is therefore included in the measured quantities @xmath22_{\mathsf{m}}$ ] and @xmath87_{\mathsf{m}}$ ] .
the advantage of spatial light modulators [ slms ] is that , within the spatial resolution of the slms , angular apertures may be smooth functions of the angle @xmath17 .
rectangular functions , which represent ` cake - slice ' apertures , would result in singular derivatives and hence in an infinite inferred variance @xmath54_{\mathsf{i}}$ ] @xcite .
however , this analysis does not take into account that optical diffraction will have a smoothing effect on the angle probability distribution . to study the influence of even very small smoothing effects we consider in the following theoretical modelling a class of continuously differentiable aperture functions which asymptotically approximate rectangular functions .
on the other hand , aperture functions which differ only slightly may be mapped to the same aperture in the slm , because of the limited resolution .
these aspects have to be discussed more closely in conjunction with specific experimental implementations .
to give a quantitative result we model the measurement of @xmath81 $ ] under the two assumptions that the photon pair is perfectly correlated in angular position and that the angle probability distribution behind the aperture is exactly given by the function describing the aperture .
under these assumptions the conditional probability @xmath81 $ ] is given by the overlap integral of the two probability densities , since the detection probability for an analyzing aperture centred at @xmath73 is given by @xmath88 using the assumption of perfect correlation the probability density will be sharply peaked for @xmath86 and in that sense we may use a @xmath4-periodic @xmath89-function to approximate @xmath90 @xmath91 the idler probability density @xmath92 , may be written as a function centred at @xmath93 but shifted by @xmath75 , which yields @xmath94 . using these results
the detection probability can be rewritten as @xmath95 for the calculation of this integral it is important to use the periodicity of the probability densities if the argument lies outside the @xmath4 radian interval .
the detection probability @xmath96 is the measured conditional probability @xmath80_{\mathsf{m}}$ ] .
varying @xmath75 over the @xmath4 radian interval yields the conditional probability @xmath81_{\mathsf{m}}$ ] . in an experimental test of our criterion
[ eq : paracrit ] ] this quantity would be known from measurements .
we model the conditional angle measurement for different aperture functions . as experiments to validate the uncertainty relation @xcite used apertures which can be described by truncated gaussians , we calculate the effects of these truncated gaussian apertures and also of a set of truncated super gaussians .
the latter allows us to interpolate between gaussian and rectangular apertures . as mentioned in section ( [ ssc : aperslm ] ) rectangular apertures , which could be made from a solid , absorbent material , will lead to an infinite inferred variance @xcite .
all aperture functions discussed in this section are symmetric , i.e. @xmath97 for @xmath98 . the overlap integral in eq .
( [ eq : overlap ] ) can therefore be turned into the convolution of the two probability densities @xmath99(\pi - \tau_2).\ ] ] for simplicity we are modelling the quantity @xmath100_{\mathsf{m}}$ ] with the convolution @xmath101(\phi_2)$ ] as for the symmetric and periodic aperture functions the shift by @xmath102 and the sign of @xmath75 is not relevant . the probability functions describing rectangular apertures can be given in terms of heaviside step functions @xmath103 : @xmath104 according to eq .
( [ eq : convolution ] ) the conditional probability is given by the convolution of the two probability densities @xmath105(\phi_2 ) = \frac{1}{\delta_1 ^ 2 - \delta_2 ^ 2 } \left\ { \begin{array}{ll } \delta_1 - \delta_2 & |\phi_2| < \delta_2 , \\
\delta_1 - \phi_2 & \delta_2 \leq |\phi_2| < \delta_1 , \\ 0 & \delta_1 \leq |\phi_2| \end{array } \right.\ ] ] where we introduced the notation @xmath106 and @xmath107 for @xmath108 . as the convolution is commutative [ @xmath109 we can interchange @xmath110 and @xmath111 for the case @xmath112
. the conditional wavefunction is then given by the positive square root .
graphs of the rectangular apertures and the resulting conditional probability density and wavefunction are shown in fig ( [ fig : rect12 ] ) .
= 0.49 = 0.49 the fourier integral [ eq .
( [ eq : fouriertrans ] ) ] can be calculated analytically by using the fresnel sine and cosine integrals @xmath113 and @xmath114 @xcite : @xmath115 = \frac{1}{\sqrt{\delta_1 - \delta_2 } } \times \bigg [ & \frac{\sin(|m_2| \delta_1)}{|m_2| \sqrt{|m_2| } } \mathcal{c}_2 ( |m_2|(\delta_1-\delta_2 ) ) \\ & -\frac{\cos(|m_2| \delta_1)}{|m_2| \sqrt{|m_2| } } \mathcal{s}_2 ( |m_2|(\delta_1-\delta_2 ) ) \bigg ] .
\end{array}\ ] ] the distribution of conditional probability amplitudes is given in fig .
( [ fig : rect34].a ) for the analytical calculation and a numerical integration .
the numerical results are shown to give an estimate of the accuracy of our integration .
this serves as a reference for aperture functions , where we have not found an analytical solution of the fourier integral .
the conditional variance is shown in fig .
( [ fig : rect34].b ) over a maximum oam index @xmath23 at which the sum in eq .
( [ eq : condvar ] ) is truncated .
the numerical results differ from the analytical points at higher truncation indices because of numerical effects in sampling the @xmath4 radian interval .
since the fresnel integrals tend to @xmath116 for large arguments , the conditional amplitudes @xmath117 $ ] vary with @xmath118 for large @xmath23 , which leads to a logarithmic increase with higher truncation indices and thus to a divergent conditional variance .
the reason for this behaviour is founded in the singular derivative of the conditional wavefunction at @xmath119 .
= 0.49 = 0.49 aperture functions in form of truncated gaussians have been used in the experiment studying the angular uncertainty principle @xcite . in this case
the probability densities @xmath46 and @xmath79 are @xmath120 the graph for gaussian apertures is shown in fig .
( [ fig : gaus12].a ) for the same widths @xmath110 and @xmath111 as in the rectangular case . the resulting conditional probability density and wavefunction
are plotted in fig .
( [ fig : gaus12].b ) .
= 0.49 = 0.49 the conditional probability density @xmath121 $ ] can be analytically calculated using the convolution theorem for the fourier transform @xmath122(\phi_2 ) = \frac{1}{\sqrt{2\pi } } \frac{1}{{\mathrm{erf}}(\pi / w_1){\mathrm{erf}}(\pi / w_2 ) } \nonumber \\ & & \times \exp \left ( -\frac{m_2 ^ 2(w_1 ^ 2 + w_2 ^ 2)}{4 } \right ) \\ & & \times { \mathrm{re}}\left[{\mathrm{erf}}\left ( \frac{\pi}{w_1 } - { \mathrm{i}}\frac{m_2 w_1}{2}\right)\right ] { \mathrm{re}}\left[{\mathrm{erf}}\left ( \frac{\pi}{w_2 } - { \mathrm{i}}\frac{m_2 w_2}{2}\right)\right ] .
\nonumber\end{aligned}\ ] ] an approximate analytical expression for the probability amplitudes can be obtained by treating the gaussians as extended which results in setting the error functions to unity .
this allows us to calculate the @xmath123 $ ] from eq .
( [ eq : gaussconvolution ] ) @xmath124 \approx \left ( \frac{w_1 ^ 2 + w_2 ^ 2}{\pi } \right)^{\frac{1}{4 } } \left ( \frac{1}{{\mathrm{erf}}(\pi / w_1){\mathrm{erf}}(\pi / w_2 ) } \right)^{\frac{1}{2 } } \exp\left ( - \frac{m_2 ^ 2(w_1 ^ 2 + w_2 ^ 2)}{2 } \right).\ ] ] for the chosen widths the agreement with the numerical solution is excellent as can be seen in fig .
( [ fig : gaus34].a ) .
this is , however , a singular case and for different parameter settings the approximation differs more from the numerical solution .
= 0.49 = 0.49 the resulting conditional variance is shown for numerical and analytical calculations .
for the truncated gaussians the variance converges .
the difference between the numerical results and the analytical solution is relatively small , because the contributions from larger values of @xmath23 are very small .
with truncated super gaussian [ tsg ] apertures we can gradually go from the rectangular case to the gaussian .
this is achieved by an additional parameter @xmath125 in our definition of the tsg : @xmath126 where @xmath127 is the complete gamma function and @xmath128 the incomplete gamma function . for values of @xmath129
we have a kurtosis smaller than 3 , i.e. probability densities , which are less peaked than gaussians .
the effect of this parameter on the probability density can be seen in figs .
( [ fig : suga12].a ) and ( [ fig : suga12].b ) .
= 0.49 = 0.49 = 0.49 = 0.49 . for the tsg
all calculations have been done numerically and one can see in fig ( [ fig : suga34].a ) that , even for high values of @xmath125 , the probability distribution , @xmath130|^2 $ ] , differs substantially from the rectangular case .
for the variance , the dependency on the truncation index @xmath23 and @xmath125 is shown in fig .
( [ fig : suga34].b ) . for small values of @xmath125
the variance does not change over the range of truncation indices , for higher values the variance converges more slowly or not at all within the given range . for the highest value of @xmath131 the effect
is more pronounced , but far from the logarithmic increase in the rectangular case .
= 0.49 = 0.49 the theoretical modelling of a conditional angle measurement leads to a conditional variance significantly different from zero .
however , the final result depends not only on the aperture in the signal , which sets the condition , but also on the aperture used to determine the angle .
the angular epr criterion eq .
( [ eq : paracrit ] ) takes the orientation of @xmath25 into account by averaging over the the orientation angle @xmath27 . from the simplified approach in section [ ssc : oam ]
we could assume isotropic correlation for the azimuthal angle , which is expressed in the @xmath18 function in eq .
( [ eq : anglecorr ] ) . within this model
the correlation are therefore the same for every orientation @xmath27 and an averaging is not necessary .
the dependence of @xmath81 $ ] on the aperture functions @xmath79 and @xmath46 reflects the way in which the azimuthal angle is measured . from the presented results ,
it is clear that the broadness of the conditional angle probability density and the oam probability distribution is mostly determined by @xmath26 .
the influence of the analyzing aperture @xmath82 is certainly most controllable for the rectangular case .
but the analysis of tsgs showed that there are values of @xmath125 for which the influence of @xmath82 is comparable to the rectangular case without showing the divergent variance . using these apertures to study the influence of the analyzing aperture on the conditional probability distribution could lead to a more detailed model for the measurement process and thus to a more complete theoretical prediction for the final conditional oam variance . on the other hand , for tsgs
the question remains how well the aperture can be programmed in an slm , in particular as for high values of @xmath125 the features which distinguish it from a rectangular aperture are on a very small scale .
this however might be an interesting aspect in a more detailed analysis of rectangular apertures , which includes optical diffraction , as even small deviations from a perfect rectangular form would give a finite conditional variance .
in this work we have discussed the possibility to demonstrate an angular epr paradox for the conjugate variables of orbital angular momentum [ oam ] and angle .
the paradox is about the apparent violation of an uncertainty relation for incompatible observables measured on correlated , spatially separated subsystems . by using photon pairs entangled in oam and in angle these subsystems
can be realised in an optical experiment .
we have found a testable criterion for an angular epr paradox , which takes experimental indeterminacies into account .
for that we have reformulated the epr paradox using conditional variances , i.e. variances of observables from one subsystem given a preset outcome on the other subsystem . to investigate the feasibility of an experimental demonstration , we have modeled the measurement process under the assumption of perfect angle correlation .
also , angular apertures to set the condition or to determine the angle were assumed to impose their probability characteristics exactly on transmitted photons . under these assumptions
we have studied different classes of aperture functions for the final conditional oam variance .
rectangular functions lead to a divergent conditional oam variance , which does not set any lower bound for the correlations in oam for our criterion .
truncated gaussians result in a quickly converging variance , but the implementation of the apertures with a smooth grayscale transition on an slm will be less accurate than for rectangular apertures with their sharp edge .
truncated super gaussians , which can be varied from the rectangular case to the truncated gaussians , provide an aperture which leads to a convergent variance and still has a controllable influence of the analysing apertures in the idler .
the conditional variances obtained from the theoretical modelling show that an angular epr paradox can be demonstrated . given the current state of experiments we expect an implementation of our criterion to be able to demonstrate the epr paradox for oam and azimuthal position .
we would like to thank roberta zambrini for useful discussions and for the suggestion to use truncated super gaussians .
also , eric yao and miles padgett have been a very helpful in questions regarding the experimental implementation .
this work has received support from the engineering and physical sciences research council [ epsrc ] under the grant gr s03898/01 and the royal society of edinburgh .
franke - arnold , s. , barnett , s. m. , yao , e. , leach , j. , courtial , j. and padgett , m. , 2004 , _ new journal of physics _ , * 6 * , 103 .
einstein , a. , podolsky , b. and rosen , n. , 1935 , _ phys .
_ , * 47 * , 777 .
bohm , d. , 1951 , _ quantum theory _
( englewood cliffs : prentice - hall , inc . ) .
aspect , a. , grangier , p. and roger , g. , 1982 , _ phys .
_ , * 49 * , 91
. weihs , g. , jennewein , t. , simon , c. , weinfurter , h. and zeilinger , a. , 1998 , _ phys .
_ , * 81 * , 5039 .
reid , m. , 1989 , _ phys .
a _ , * 40 * , 913 .
ou , z. y. , pereira , s. f. , kimble h. j. and peng , k. c. , 1992 , _ phys . rev .
_ , * 68 * , 3663 .
howell , j. c. , bennik , r. s. , bentley , s. j. and boyd , r. w. , 2004 , _ phys .
_ , * 92 * , 210403 .
mair , a. , vaziri , a. , weihs , g. and zeilinger , a. , 2001 , _ nature _ , * 412 * , 313 .
franke - arnold , s. , barnett , s. m. , padgett , m. and allen , l. , 2002 , _ phys .
a _ , * 65 * , 033823 .
leach , j. , padgett , m. , barnett , s. m. , franke - arnold , s. and courtial , j. 2002 _ phys .
_ , * 88 * , 257901 .
hofman , h. f. and takeuchi , s. , 2003 , _ phys .
a _ , * 68 * , 032103 .
ghne , o. , 2004 , _ phys .
_ , * 92 * , 117903 .
werner , r. f. and wolf , m. m. , 2001 , _ phys .
_ , * 86 * , 3658 .
molina - terriza , g. , torres , j. p. and torner , l. , 2002 , _ phys .
_ , * 88 * , 013601 .
ekert , a. , 1991 , _ phys .
_ , * 67 * , 661 .
allen , l. , beijersbergen , m. w. , spreeuw , r. j. c. and woerdman , j. p. , 1992
a _ , * 45 * , 8185 .
allen , l. , padgett , m. j. and babiker , m. , 1999 , _ progress in optics _ , * xxxix * , 291 .
allen , l. , barnett , s. m. and padgett , m. j. , 2003 , _ optical angular momentum _ , ( bristol : institute of physics publishing ) .
barnett , s. m. and pegg , d. , 1990 , _ phys .
a _ , * 41 * , 3427 .
barbosa , g. a. and arnaut , h. h. , 2002 , _ phys .
a _ , * 65 * , 053801 .
robertson , h. p. , 1929
_ , * 34 * , 163 .
nielsen , m. a. and chuang , i. l. , 2000 , _ quantum computation and quantum information _
( cambridge : cambridge university press ) .
reid , m. , 1997 , _ quantum and semiclassical optics _ , * 9 * , 489 .
vaziri , a. , weihs , g. and zeilinger , a. , 2002 , _ j. opt .
b _ , * 4 * , s47 . altman , a. r. , kprl , k. g. , corndorf , e. , kumar , p. and barbosa , g. a. , 2004 , arxiv : quant - ph/0409180 v1 .
pegg , d. , barnett , s. m. , zambrini , r. , franke - arnold , s. and padgett , m. j. , 2005 , _ new journal of physics _ , * 7 * , 62 .
gradshteyn , i. s. and ryzhik , i. m. , 2000 , _ tables of integrals , series and products _
, 6th edition ( san diego : academic press ) . | the violation of local uncertainty relations is a valuable tool for detecting entanglement , especially in multi - dimensional systems .
the orbital angular momentum of light provides such a multi - dimensional system .
we study quantum correlations for the conjugate variables of orbital angular momentum and angular position .
we determine an experimentally testable criterion for the demonstration of an angular version of the epr paradox . for the interpretation of future experimental results from our proposed setup ,
we include a model for the indeterminacies inherent to the angular position measurement . for this measurement angular apertures
are used to determine the probability density of the angle .
we show that for a class of aperture functions a demonstration of an angular epr paradox , according to our criterion , is to be expected . |
in his book _ partial differential relations _
@xcite , m. gromov proposed a definition of the blow up operation in the contact category , see exercise ( c ) on page 343 .
this article discusses this definition as well as related constructions .
+ let @xmath0 be a smooth manifold and @xmath1 an embedded submanifold
. the normal bundle of @xmath2 in @xmath0 will be denoted by @xmath3 . recall that it is defined through the short exact sequence of smooth vector bundles over @xmath4 @xmath5 given a complex vector bundle @xmath6 , we denote by @xmath7 the fiberwise projectivization of @xmath8 . + suppose the normal bundle @xmath3 is a complex bundle , then we may produce a manifold @xmath9 , the topological blow up of @xmath0 along @xmath4 .
it is defined as the connected sum @xmath10 of the manifolds @xmath0 and @xmath11 with the reversed orientation along @xmath4 .
let @xmath12 be the zero section of @xmath3 .
the submanifold @xmath4 is embedded in the first factor through @xmath13 and in the second as the section @xmath14 in the category of symplectic manifolds the normal bundle is a complex bundle and the manifold @xmath9 can be endowed with a symplectic structure . in this paper
we address the question for contact manifolds .
+ in the above reference , m. gromov conjectured that a contact blow up
construction exists along a contact submanifold @xmath4 embedded in a contact manifold @xmath0 provided a pair of hypotheses are satisfied .
these are : + * the contact submanifold @xmath15 is a boothby
wang manifold .
see definition [ def : bw ] .
in particular , the reeb vector field associated to @xmath16 has all its orbits periodic with the same period .
let @xmath17 be the quotient space of its orbits and @xmath18 the projection map .
+ * the normal symplectic bundle @xmath3 is isomorphic to the pull back of a symplectic bundle @xmath19 through @xmath20 .
that is , there exists an isomorphism @xmath21 of symplectic bundles .
these two hypotheses allow to give a definition .
nevertheless the contact blow up will not be a contact structure on the topological blow up
@xmath9 of @xmath0 .
we will first illustrate a reason for this in a simple example , see section [ sec : pr ] .
we provide a definition producing a contact structure on a manifold constructed as a different connected sum with @xmath0 .
it has the same geometrical properties as the symplectic blow up .
it is this manifold we would rather call the contact blow up . +
apart from the construction of contact manifolds , the contact blow up construction is relevant for the existence problem of contact structures on @xmath22manifolds .
see @xcite .
+ the content of the paper is organized as follows .
the section [ sec : pr ] provides a brief review of the topological blow up . in section [ sec : bw ] , we introduce the classical boothby
wang construction @xcite .
it will be described with some concrete examples that shall be used later on .
then , three alternative constructions of contact blow up are introduced : + * the contact blow
up for embedded transverse loops , produced as a surgery operation . this had been introduced in the article @xcite , it will be reviewed in section [ sec : srgy ] . *
the contact blow
up defined _ la gromov _ is the content of section [ sec : gromov ] . *
the contact blow up as a contact quotient is described in section [ sec : quotient ] .
+ these three constructions are inspired by the three alternative constructions for the symplectic blow
up : the _ ad hoc _ construction with explicit gluings , the description using frame bundles , found in pages 239 and 243 in @xcite respectively , and the symplectic cut procedure discussed in @xcite .
finally , section [ sec : un ] relates these constructions in the case of transverse loops . +
* acknowledgements .
* we want to acknowledge k. niederkrger for useful discussions .
in particular for asking us to relate the contact cut and the contact blow up .
this project was partially developed during the aim workshop _ contact topology in higher dimensions_. the first and third authors are supported by the spanish national research project mtm201017389 .
second author would like to thank ictp for offering a visiting position that allowed him to develop this article .
in this section we introduce the basic definitions , explain the topological blow up procedure and discuss an example .
a contact structure on a smooth manifold @xmath23 is a maximally non
integrable smooth field @xmath24 of tangent hyperplanes .
a contact manifold @xmath25 is a choice of a contact structure @xmath24 on @xmath0 .
the maximal non integrability can be described in terms of local equations for @xmath24 .
a smooth field @xmath24 of tangent hyperplanes is maximally non
integrable if and only if for any @xmath26 there exist an open subset @xmath27 containing @xmath28 and a @xmath29form @xmath30 such that @xmath31 and @xmath32 .
equivalently , the form @xmath33 is non degenerate when restricted to @xmath34 in case the form @xmath35 can be chosen to be globally defined , i.e. @xmath36 , the contact structure @xmath24 is called coorientable . a contact structure is cooriented if a choice of global contact form has been made .
+ let @xmath37 be a cooriented contact manifold with fixed global contact form @xmath35 , i.e. @xmath36 satisfies @xmath38 , @xmath39 .
a smooth submanifold @xmath40 is called a _ contact submanifold _ if the induced distribution @xmath41 is a contact structure on @xmath4 .
+ the notion of a blow up has its origins in algebraic geometry .
first , we define the concept for a complex vector space .
see @xcite for further details .
[ def : lnbu ] the blow up
@xmath42 of the @xmath43dimensional complex vector space @xmath44 at the origin is the smooth manifold @xmath45,p):p\in l\}\subset{\mathbb{cp}}^n\times{\mathbb{c}}^{n+1}$ ] along with the restriction of the projection onto the second factor @xmath46 .
note that @xmath47 restricted to @xmath48,p):p\in l , p\neq0\}$ ] induces a diffeomorphism onto the image @xmath49 .
the projective space @xmath50 is called the exceptional divisor . the topological blow up of @xmath0 along @xmath4 defined in the previous section coincides with the previous definition if @xmath51 is the origin in @xmath52 .
more generally , from the definition of @xmath9 we can conclude the following [ prop : tpbl ] let @xmath0 be a smooth manifold and @xmath2 a submanifold with complex normal bundle .
there exists a smooth submanifold @xmath53 diffeomorphic to the total space of a projective smooth bundle over @xmath4 such that , as smooth manifolds , @xmath54 . the topological blow
up can be performed along any complex submanifold @xmath4 of a complex manifold @xmath0 . in this case
the blown up manifold @xmath9 inherits a canonical complex structure
. analogously , if @xmath55 is a symplectic manifold and @xmath4 a symplectic submanifold , the topological blow up manifold @xmath9 can also be endowed with a symplectic structure . in the symplectic case
there is no uniqueness , see @xcite . the topological blow
up could also be performed along a contact submanifold of a contact manifold because the normal bundle is symplectic and hence it is also complex .
[ rem : key ] 1 .
suppose the normal bundle @xmath3 splits as a direct sum of isomorphic complex line bundles @xmath56 : @xmath57 .
then there is a second projection map @xmath58 defined as follows . given a point @xmath59 ,
let @xmath60 be non
zero vector in the fiber .
then a point @xmath61 \in \nu_m(s)_p$ ] is mapped to @xmath62)=[l_1/s_p , \cdots , l_r / s_p]$ ] .
it is simple to verify that the map is well defined , i.e. independent of the choice of vector @xmath63 .
the hypothesis above is satisfied in some cases .
for instance , let @xmath4 be the base locus of a projective , resp .
symplectic , lefschetz pencil .
then @xmath4 conforms the hypothesis for @xmath64 . in such case
the fibers of @xmath65 are projective , resp .
this also occurs with contact pencils , see @xcite .
* example * : _ let @xmath66 be a @xmath22dimensional contact manifold and @xmath4 is a @xmath29dimensional contact compact submanifold , i.e. a transverse embedded loop . if we perform a topological blow
up along @xmath4 , the exceptional divisor is @xmath67 .
we are in the hypothesis of the previous remark : @xmath68 is trivial .
therefore we have a projection @xmath69 . in the contact case ,
if we assume that @xmath8 is a contact submanifold , it is not possible to ensure that the fibers of such projection map are contact : there is no contact distribution on @xmath70 whose fibers are all transverse to the contact structure , see @xcite . _
+ in the previous example , the non transversality of the fibers occurs only because we are using the topological blow up as our blown up manifold .
we will further argue from different perspectives that the blown up manifold @xmath9 we should consider in contact topology is not the topological blow up discussed above .
instead , the correct manifold is obtained through a procedure that substitutes @xmath71 by the standard contact sphere @xmath72 , not @xmath70 .
in such a case , the natural projection map @xmath73 is the hopf fibration , whose fibers are transverse to the contact structure . +
in this section we explain the construction of a contact manifold from an integral symplectic manifold as developed in @xcite .
it will be used in understanding the contact structure on the manifold obtained after a contact blow - up . + a symplectic manifold @xmath74 is called integral if the class @xmath75 $ ] lies in the image of the map @xmath76 , i.e. the periods of @xmath77 are integers .
such a form @xmath77 is called integral .
for instance , a khler form on a complex compact manifold is integral if and only if the manifold is a smooth projective algebraic variety . in the definition above a circle
has length @xmath29 .
note that the lift of @xmath75 $ ] to @xmath78 may not be unique if @xmath79 contains torsion elements .
+ given an integral form @xmath80 there exists a hermitian complex line bundle @xmath81 admitting a compatible connection whose curvature is @xmath82 .
see @xcite for the details .
this leads to the following [ def : bw ] let @xmath74 be an integral symplectic manifold .
the boothby
wang manifold @xmath83 is the contact manifold whose total space is the unit circle bundle associated to the line bundle @xmath84 and its contact structure is defined as the restriction of any connection with curvature @xmath85 to the circle bundle .
[ rek : uniqueness_bw ] the contact structure is independent of the choice of connection .
indeed , the space of choices for a connection as above is an affine space modelled on the vector space of flat connections and hence is contractible .
the stability theorem of j. gray applies to ensure the uniqueness up to contactomorphisms of the contact structures . for the case
@xmath86 we will sometimes omit the subindex .
note that the topology of the total space varies with the parameter @xmath87 .
the exact relationship between the topology and the parameter @xmath87 is the content of the following [ lem : cov ] let @xmath74 be a symplectic manifold . then the boothby
wang manifold @xmath88 is a @xmath87covering of @xmath83 .
we fix a hermitian connection on @xmath56 , this induces a hermitian connection on @xmath89 .
define the unitary non linear map between line bundles @xmath90 it preserves the connections on the two bundles .
there exists a unitary connection preserving action of @xmath91 , the cyclic group of order @xmath87 , in @xmath56 given by @xmath92 this action induces the trivial action in @xmath89 and thus becomes the deck transformation group of a covering between the total spaces of the associated circle bundles .
this map is certainly compatible with the contact structures . * examples * : 1 .
let @xmath93 be a lens space , i.e. the orbit space of the action @xmath94 the lens space naturally inherits a contact structure @xmath95 from the standard contact structure of @xmath96 induced by the complex tangencies .
lemma [ lem : cov ] provides a contactomorphism between @xmath97 and @xmath98 + 2 .
consider the @xmath99torus @xmath100 and @xmath101 an integral area form with total area one .
then the boothy
wang manifolds @xmath102 associated to @xmath103 give rise to quotients of the heisenberg group by discrete subgroups @xmath104 and thus provide several examples of contact nilmanifolds different from the 3torus .
+ the construction of the contact blow up will involve the quotient of the product of two boothby wang manifolds . with this in mind
, we proceed to describe the boothby
wang construction when the base sympletic manifold is a product .
we show that the boothby
wang construction and the cartesian product _ commute_. in precise terms , let @xmath105 be the boothby
wang manifold associated to @xmath106 then we have the following : [ thm : bw ] let @xmath107 and @xmath108 be symplectic manifolds and @xmath109 a pair of coprime integers .
consider the product @xmath110 of the boothby
wang manifolds and the action @xmath111 then the space of orbits is a manifold diffeomorphic to @xmath105 .
this space of orbits carries a contact structure induced by a connection with curvature @xmath112 and hence is contactomorphic to @xmath113 let @xmath114 and @xmath115 be the subgroup defined as the image of the embedding @xmath116 let @xmath117 be the @xmath118principal bundle with base space @xmath119 induced by the @xmath120principal bundles @xmath121 and @xmath122 .
our aim is to describe @xmath123 as a bundle over @xmath119 . in general @xmath124 is not a @xmath125principal bundle but it is the case when both @xmath118 and @xmath125 are closed lie groups and @xmath125 is a normal sub - group of @xmath126 actually , they are abelian and since @xmath127 , @xmath123 is also a @xmath128principal bundle over @xmath119 . taking into account the exact group sequence @xmath129 where the second morphism is given by multiplication by @xmath130 ,
we conclude that the space of orbits @xmath123 is a manifold diffeomorphic to @xmath105 .
the claim about the connection and the associated curvature follows from the short exact sequence @xmath131 finally , it follows from remark [ rek : uniqueness_bw ] that the manifolds are , in fact , contactomorphic .
there are a few simple cases worth mentioning .
+ * examples * : 1 .
let @xmath132 and @xmath133 arbitrary .
then neither the topology of the resulting space nor the contact structure depend on @xmath134 .
indeed , @xmath135 is diffeomorphic to @xmath136 analogously , the parameter @xmath137 is vacuous if @xmath138 . in particular
, @xmath139 quotiented by any @xmath140 coprime @xmath120action is diffeomorphic to @xmath120 .
let @xmath141 be symplectic manifolds with the fubini
study form .
then the space @xmath142 is diffeomorphic to @xmath143 regardless of the values @xmath144 , see @xcite for a proof of this fact .
further , the symplectic structure of the associated line bundle depends only on @xmath145 .
note that there is an alternative construction of a contact structure in @xmath143 using an open book decomposition with @xmath146 pages and an even power of a dehn twist as monodromy , however such a procedure may only produce vanishing first chern class and is thus different from @xmath142 if @xmath147 . see @xcite for more .
the previous example can be generalized to construct contact structures on @xmath148 . indeed
the result implies that the total space of @xmath149 is a @xmath150bundle over @xmath151 .
the hopf action is explicit enough for the classifying map to be described as the element @xmath152 consequently the resulting manifold is diffeomorphic @xmath148 if @xmath43 is odd or @xmath87 is even .
we would like to remark that it is not known whether the product of any contact manifold with the @xmath99sphere admits a contact structure .
it will be essential for the contact blow up construction to be able to extend a connection on a submanifold to a global connection , let us now prove that this is possible under suitable conditions : [ lem : exten_bw ] let @xmath4 be a closed submanifold of @xmath153 , possibly with smooth boundary , and @xmath56 the line bundle associated to @xmath77 .
assume that the restriction morphism @xmath154 is surjective and let @xmath155 be a connection over @xmath156 whose curvature is @xmath82 .
then , there is a connection @xmath157 on @xmath56 with curvature @xmath82 such that its restriction to @xmath4 is @xmath155 .
let @xmath158 be a connection on the line bundle @xmath159 with curvature @xmath82 .
denote @xmath160 , then @xmath161 is a closed @xmath29form over @xmath4 . in order to complete our argument we need to extend @xmath162 to a global closed @xmath29form .
+ by hypothesis
the map @xmath163 is a surjection .
therefore there exists a cohomology class @xmath164 $ ] on @xmath165 , such that restricted to @xmath4 coincides with @xmath166 $ ] .
its difference over @xmath4 will be the trivial class on @xmath167 , so @xmath168 , for some smooth function @xmath169 .
we extend @xmath170 to a global smooth function @xmath171 . the form @xmath172 is the required global connection with curvature @xmath82 and extending @xmath155 .
let @xmath173 be a contact manifold .
in this section we recall the blow up construction from section 5 in @xcite .
it is an operation defined in a neighborhood of a transversely embedded loop .
topologically it consists of a surgery along the loop : the interior of @xmath174 is removed and a tubular neighbourhood of the @xmath175sphere @xmath176 is glued along the common boundary @xmath177 .
the sphere @xmath178 whose neighbourhood is attached is called the exceptional divisor .
let us discuss this surgery operation in the contact category .
+ consider the manifold @xmath179 with spherical coordinates @xmath180 .
let @xmath181 be the standard contact form for the contact structure @xmath182 on the sphere @xmath183 .
define the following two contact forms in @xmath184 : @xmath185 fix an integer @xmath186 and consider the diffeomorphism @xmath187 it pulls
back the contact form @xmath188 to @xmath189 $ ] .
+ given a subset @xmath190 , let @xmath191 denote a small neighbourhood of @xmath192 in @xmath0 .
these ingredients suffice to prove the following : [ thm : blow - up ] @xmath193thm .
5.1 in @xcite@xmath194 let @xmath195 be a contact manifold .
let @xmath196 be a smooth transverse loop in @xmath0 .
there exists a manifold @xmath197 satisfying the following conditions : * there exists a contact structure @xmath198 on @xmath197 .
* there exists a codimension@xmath99 contact submanifold @xmath8 inside @xmath197 with trivial normal bundle .
the manifold @xmath199 is contactomorphic to the standard contact sphere @xmath200 .
* the manifolds @xmath201 and @xmath202 are contactomorphic .
the manifold @xmath203 will be called the contact surgery blow up of @xmath0 along @xmath4 .
the contact submanifold @xmath199 is called the exceptional divisor . by gray s stability
, we may assume that a tubular neighbourhood of the embedded loop is contactomorphic to @xmath204 with the contact form @xmath188 as in ( [ eq : std ] ) , for some small radius @xmath205 .
we enlarge this tubular @xmath206neighbourhood using the squeezing technique from @xcite to obtain a radius 2 neighbourhood .
more precisely , we need the following auxiliary lemma : [ lem : ekp ] _ ( _ proposition 1.24 in @xcite _ ) _ [ lem : shrink ] let @xmath207 be a positive integer and @xmath208 a radius .
then the following map is a contactomorphism @xmath209 and it restricts to the identity at @xmath210 .
consider @xmath211 in the lemma above , then we need @xmath87 large enough to satisfy @xmath212 we may therefore assume that the tubular neighbourhood for which the standard equation ( [ eq : std ] ) holds for @xmath188 has radius @xmath64 . in the annulus corresponding to radius @xmath213 use @xmath214 to induce the contact structure given by @xmath215 .
declare @xmath216 to define the contact structure in the radius area @xmath217 $ ] .
it is left to find a strictly increasing function interpolating between @xmath218 and @xmath219 in the middle region .
this can be done , see figure [ fig : h ] . and @xmath220 .
] the process described in the proof can be modified to include the radius squeezing in the gluing map .
it suffices to use @xmath221 as gluing map instead of @xmath214 in the domain .
indeed , denote @xmath222 and consider the contact structures @xmath223 define the map @xmath224 where @xmath225 is the obvious radius in the image .
then the following diagram is commutative in the contact category : @xmath226^{\varphi}\ar@{->}[r]^{\phi_1 } & ( t_2,\xi_0)\ar@{->}[d]^{\psi_k}\\ ( t_{\varepsilon(k)},\xi_l)\ar@{->}[r]^{\phi_l } & ( t_{\varepsilon(k)},\xi_0 ) } \ ] ] where lemma [ lem : ekp ] is performed with parameter @xmath227 .
note that the contactomorphism type of the exceptional divisor is that of the standard sphere , the parameter in the construction allows us to discretely vary the radius of the tubular neighbourhood we are collapsing .
the maps @xmath221 and @xmath228 are smoothly isotopic if and only if @xmath229 is even .
let @xmath230 be a circle coordinate .
consider the morphism @xmath231 if @xmath232 denotes @xmath87times the standard circle action on @xmath183 induced by @xmath233 , then it is clear that @xmath228 is realized as @xmath234 . since @xmath235 and @xmath236 under this identification ,
@xmath237 if and only if @xmath229 is even .
+ it is left to prove that @xmath238 and @xmath214 are not isotopic , for @xmath43 even .
construct two manifolds @xmath239 and @xmath240 by gluing two copies of the manifold @xmath241 respectively using @xmath238 and @xmath214 along the boundary .
these manifolds are not diffeomorphic .
a sphere is a spin manifold and the product formula for characteristic classes implies that so is @xmath242 .
+ the manifold @xmath240 is not spin .
this can be seen by using any section @xmath243 of the twisted bundle @xmath244 , such @xmath243 exists because @xmath245 .
denote by @xmath246 the normal bundle to the section and let @xmath247 be the complex bundle over @xmath151 such that @xmath248 . then @xmath249 .
note that @xmath250 if @xmath43 is odd and @xmath251)=w_2(tx_{|s({\mathbb{s}}^2)})=w_2(\nu(s({\mathbb{s}}^2))=w_2(s^*(\nu(s({\mathbb{s}}^2)\oplus { \mathbb{r}}))= w(e_1).\ ] ] hence @xmath238 and @xmath214 are not isotopic . in particular , the smooth type of the contact blow up manifold will depend on the parity of the positive integer fixed for the construction .
as for the contact type , it follows from theorem 1.2 in @xcite that the maps @xmath228 and @xmath221 are not contact compactly supported isotopic if @xmath252 .
this does not imply that the contact structures are different , but at least there is no local contactomorphism relating the two contact structures .
in this section we develop the contact blow up along a boothby
wang submanifold , as suggested in @xcite.the existence of a minimum radius for the tubular neighbourhood of the submanifold along which we will perform the blow
up will play an important role .
this feature will be revisited in the definition provided in section [ sec : quotient ] .
+ let us review the definition of the symplectic blow up , see @xcite for more details .
let @xmath253 be a symplectic manifold and @xmath4 a symplectic submanifold of codimension @xmath254 .
consider the symplectic normal bundle @xmath255 of @xmath4 in @xmath0 and fix a compatible almost complex structure .
the choice of a compatible almost complex structure for a symplectic form induces a metric and the equality @xmath256 implies that the structure group of @xmath257 can be considered to be @xmath258 . thus @xmath257 is an associated vector bundle of a @xmath258principal bundle @xmath259 .
+ the symplectic blow up of @xmath0 along @xmath4 is obtained by the fiberwise symplectic blow up of @xmath257 . hence we require definition [ def : lnbu ] for the case of symplectic vector spaces .
let @xmath260 be the standard symplectic vector space and @xmath261 the standard fubini
study form on complex projective space .
we will use the following [ def : sybu ] a symplectic blow up of @xmath260 at the origin with radius @xmath262 is a symplectic manifold @xmath263 such that : * @xmath264 is a topological blow up of @xmath265 at the origin . the symplectic form induced on the exceptional divisor @xmath266 is @xmath267 .
* for any @xmath268 , there exists a symplectomorphism @xmath269 * the unitary group @xmath258 acts hamiltonially in @xmath263 .
+ the symplectic blow up of @xmath260 at the origin exist for each @xmath262 .
note that the definition depends on @xmath262 , this parameter does not appear in definition [ def : lnbu ] since any linear homothety at the origin is a complex isomorphism .
let us describe the non linear symplectic blow up of @xmath0 along @xmath4 .
property 3 in the above definition allows us to associate to @xmath117 a bundle @xmath270 over @xmath4 with fiber @xmath271 .
let @xmath272 be a connection in @xmath117 , there are induced coupling forms @xmath35 and @xmath273 , in @xmath257 and @xmath274 respectively , restricting to the symplectic form on each fiber and coinciding away from the radius @xmath275 , see thm . 6.17 in @xcite .
define the forms @xmath276 on the bundles @xmath257 and @xmath277 .
these are symplectic forms close to the zero section and to the exceptional divisor respectively .
+ these forms also coincide away from a neighbourhood of @xmath4 of radius @xmath278 .
let @xmath279 be a neighbourhood of the zero section of the symplectic normal bundle . by the symplectic neighbourhood theorem
there is a neighbourhood @xmath280 of the symplectic submanifold @xmath4 and a symplectomorphism @xmath281 .
thus any fiberwise symplectic blow up on @xmath257 with radius @xmath282 can be glued back to the initial manifold @xmath0 using the symplectomorphism @xmath283 .
the resulting manifold is the symplectic blow up of @xmath0 along @xmath4 with radius @xmath262 .
+ observe that the radius of the tubular neighbourhood of @xmath4 can not be estimated a priori .
therefore the symplectic volume of the exceptional divisor can not be assumed to be arbitrarily large .
this will be an obstruction to develop the gromov s approach in the contact category .
+ * example : * let @xmath284 be a rank@xmath285 symplectic vector bundle over a symplectic manifold @xmath74
. then the total space is symplectic as well .
thus , we are able to blow up the symplectic manifold @xmath284 along its zero section @xmath17 . in case
the symplectic form @xmath77 is of integer class , the symplectic form in the resulting blown up manifold will be of an integer class if the blow up radius is @xmath286 , @xmath287 .
we call this a radius @xmath288 blow up .
+ we now define the contact blow up in terms of the symplectic blow up .
this is the second notion listed in section [ sec : intro ] .
+ let @xmath25 be a contact manifold and @xmath289 a contact submanifold . we assume : * the contact submanifold @xmath4 is contactomorphic to a boothy
wang manifold @xmath290 . + * let @xmath291 be the circle bundle projection .
there exists a symplectic bundle @xmath284 over @xmath17 such that , as symplectic bundles @xmath292 .
the total space of @xmath284 carries a symplectic form @xmath293 in the same cohomology class of @xmath75 $ ] , under the natural identification of @xmath294 with @xmath295 .
as previously explained , there exists a symplectic manifold @xmath296 obtained by blowing up @xmath284 along its zero section @xmath17 .
suppose that the parameter multiplying the class of the exceptional divisor @xmath8 in the symplectic blow
up is a positive integer , i.e. the symplectic form in @xmath297 is integral .
+ the construction of the contact blow
up is based on the following diagram : @xmath298 & \mathbb{s}(v ) \ar@{->}[d ] & \mathbb{s}(\widetilde{v})\supset\mathbb{s}(e)=\mathbb{s}(\widetilde{v})|_e \ar@{->}[d ] \\ & ( s,\xi_s)\cong\mathbb{s}(w ) \ar@{->}_{\pi}[d ] & ( v,\overline{\omega } ) \ar@{->}[ld ] & ( \widetilde{v},{\overline{\omega}_w})\supset e \ar@{->}[l ] \\ & ( w,\omega)}\ ] ] each map is a bundle projection .
it is essential to understand the relation between the contact manifolds @xmath299 and @xmath300 .
this is the content of the following : in the hypotheses above , @xmath301 is a contact submanifold of @xmath302
. there are contactomorphic neighbourhoods @xmath303 and @xmath280 in @xmath302 and @xmath0 respectively .
the choice of symplectic form on @xmath284 implies that there exists a symplectic embedding of @xmath17 in @xmath284 and therefore @xmath301 is contained in @xmath302 as a contact submanifold .
the tubular neighbourhood theorem states that the normal bundle @xmath3 is diffeomorphic to a small neighbourhood of @xmath4 in @xmath0 , but @xmath304 so the same situation applies to @xmath301 in @xmath302 .
the statements now follow from the contact neighbourhood theorem . in consequence , @xmath305 provides a local model .
thus we only need to perform the blow up of @xmath284 along @xmath17 and study whether the boothby
wang structures associated to them allow us to glue back the resulting blown up model to @xmath0 .
this is the content of the following : + [ propo : bundle_blow ] let @xmath306 be a boothby
wang contact submanifold of @xmath307 .
suppose we symplectically blow up
@xmath308 by collapsing a radius @xmath29 neighbourhood .
then , there is a choice of contact form for @xmath309 such that @xmath310 is a contact submanifold of @xmath309 and the complement of an arbitrary small neighbourhood of @xmath310 in @xmath309 is contactomorphic to the complement of some neighbourhood of @xmath301 in @xmath302 . for the sake of a clearer exposition the proof
is explained at the end of this subsection .
+ suppose we can choose a tubular neighbourhood @xmath311 with radius larger than @xmath29 which is contactomorphic to a tubular neighbourhood @xmath312 .
then we have the following the contact blow up of @xmath25 along @xmath289 is the contact manifold @xmath313 obtained by removing the neighbourhood @xmath280 and gluing along its boundary a small neighbourhood of @xmath310 in @xmath309 .
the contact manifold @xmath313 is contactomorphic to @xmath0 away from small neighbourhoods of @xmath310 and @xmath4 respectively .
the _ exceptional divisor _ of the contact blow
up is defined to be @xmath310 , where @xmath8 is the exceptional divisor of the symplectic blow
up over which it is locally modelled .
observe that for the definition to work we need @xmath4 to have a tubular neighbourhood of radius at least @xmath29 inside @xmath0 .
+ * example * : 1 .
the most simple example of contact blow
up is the case of a transverse loop @xmath314 in @xmath66 .
the loop is contactomorphic to @xmath315 and its normal bundle is the pull back of the trivial bundle over the point . thus h1 and h2 are satisfied .
the symplectic model corresponds to the blow up of @xmath316 at the origin , collapsing a neighbourhood of radius @xmath29 , and therefore @xmath317 .
hence , @xmath318 , i.e. the standard contact @xmath319sphere .
this particular case can be seen , at least topologically , as a surgery along a loop .
+ 2 . in the previous example we may symplectically blow up with radius @xmath320 .
the exceptional divisor is then @xmath321 , i.e. the sphere bundle associated to the polarization @xmath322 of @xmath323 , which is the lens space @xmath324 with its standard contact structure .
therefore , even the diffeomorphism type of the blown up contact manifold changes with the blow up radius @xmath320 .
+ note that there is not natural projection map from @xmath310 , the exceptional divisor , to the blow up locus @xmath301 . in the case of a loop in a @xmath22dimensional manifold , the exceptional divisor for a radius @xmath29 blow
up is @xmath72 and the blow up locus is the circle @xmath120 .
this is a difference with respect to the symplectic and algebraic cases where the exceptional divisor is a bundle over the submanifold along which the blow
up is performed .
it is true though that there is a natural projection @xmath325 , but it does not lift to @xmath301 .
the assumption of the integer radius can be fulfilled in certain cases .
for instance in the blow up along a transverse @xmath120 we can use the lemma [ lem : shrink ] .
therefore the construction in this case will have two natural parameters : the integer radius that determines the topology of the exceptional divisor , and the choice of framing in the spirit of the lemma .
the above described construction la gromov does not show in general the appearance of this second positive integer , this is a reason to introduce a third way of defining the blow up highlighting these two choices . to conclude this subsection we prove the assertion that allowed us to glue the boothby
wang construction over the exceptional divisor in the contact blow up construction .
+ _ proof of proposition [ propo : bundle_blow]_. we need to find an appropriate connection on the topological boothby
wang manifold corresponding to @xmath297 .
+ notice from the construction of the symplectic blow up as given in @xcite , we know that given a sufficiently small neighbourhood of @xmath8 in @xmath297 it is possible to choose a symplectic form @xmath293 on @xmath297 such that complement of that neighbourhood in @xmath297 is symplectomorphic to a small neighbourhood of @xmath17 in @xmath284 . furthermore , observe that the exceptional divisor is just the inverse image of @xmath17 contained in @xmath284 as the zero section under the blow up projection @xmath326 .
+ now recall from the definition [ def : bw ] that the contact structure of @xmath309 is determined by the choice of a connection over the associated line bundle whose curvature is @xmath327 .
so let @xmath157 be the connection over @xmath56 that determines the contact structure on @xmath302 , and denote by @xmath328 an arbitrarily small neighbourhood of @xmath17 inside @xmath284 .
from the construction of the symplectic form @xmath293 on @xmath297 we can assume that the map @xmath329 is a symplectomorphism of @xmath330 and @xmath331 .
therefore the connection @xmath332 satisfies the required properties over @xmath331 .
it remains for it to be extended to a connection all over @xmath297 with curvature @xmath327 . by lemma [ lem : exten_bw ]
such an extension is possible provided that the restriction morphism from @xmath333 is surjective .
it is sufficient to show that @xmath334 .
+ indeed , observe that @xmath297 is homotopic to a @xmath335bundle over @xmath17 , with @xmath336 , and hence @xmath337 holds .
note that the manifold @xmath331 is diffeomorphic to @xmath330 and the set @xmath330 is homotopy equivalent to a sphere bundle over @xmath17 with fibers of dimension greater than @xmath99 . from the long exact sequence of homotopy
we conclude that @xmath338 it now follows from lemma [ lem : exten_bw ] that on @xmath309 there is a choice of contact form with the required properties . away from a given small neighbourhood of @xmath310 in @xmath309
a contact form can be chosen such that it induces a contactomorphism from the complement of such a neighbourhood to the complement of some neighbourhood of @xmath17 in @xmath339 this is because we can choose the symplectic form on @xmath297 with the required property . @xmath340
in this section we define the contact blow up of a contact manifold @xmath0 along a boothby
wang contact submanifold @xmath4 using the notion of contact cuts . given a @xmath120action on a manifold @xmath0 , topologically
the cut construction is based on collapsing the boundary of a tubular neighborhood of a given submanifold invariant by the action .
basic knowledge on the contact reduction procedure is assumed in the next few paragraphs , see @xcite . let us recall the construction of a contact cut for a contact @xmath120action as developed by e. lerman : [ thm : contcut ] @xmath193thm . 2.11 in @xcite@xmath194
let @xmath341 be a contact manifold with a @xmath120action preserving @xmath35 and let @xmath342 denote its moment map .
suppose that @xmath120 acts freely on the zero level set @xmath343 .
then the set and @xmath344 for some @xmath345 . ] @xmath346 is naturally a contact manifold .
moreover , the natural embedding of the reduced space @xmath347 into @xmath348 is contact and the complement @xmath349 is contactomorphic to the open subset @xmath350 note that the contact reduction requires the regular value to be @xmath351 , whereas in the symplectic reduction any regular value is licit .
this is so because in the contact reduction it is imposed that the orbits of the isotropy subgroup are tangent to the contact structure , see @xcite .
let @xmath173 be a contact manifold and @xmath352 a codimension@xmath285 contact submanifold .
suppose that @xmath353 for some symplectic manifold @xmath74 , @xmath354 , and @xmath355 as complex bundles over @xmath4 .
we will define the _ contact blow up of @xmath0 along @xmath4_. any isocontact embedding is isocontact if @xmath356 . ] of a contact @xmath319fold in a sphere has trivial normal bundle .
this situation does occur : any closed cooriented @xmath319fold admits an isocontact embedding into the standard contact @xmath357sphere .
a tubular neighbourhood of the contact submanifold @xmath4 is contactomorphic to @xmath358 with the contact structure given by @xmath359 , where @xmath360 is the standard contact form in @xmath361 .
let @xmath362 and consider the @xmath120action @xmath363 @xmath364 this action is generated by the field @xmath365 where @xmath366 are the reeb vector fields associated to @xmath35 and @xmath360 .
the moment map of the above action is @xmath367 @xmath368 the contact cut can only be performed in the pre image of the regular value @xmath369 , it is thus a necessary condition that @xmath370 .
this can always be achieved if @xmath134 is large enough .
let @xmath371 be a contact submanifold of @xmath25 with fixed trivial normal bundle @xmath372 .
let @xmath362 be such that @xmath370 .
the @xmath373contact blow up
@xmath374 of @xmath0 along @xmath4 is defined to be the contact cut of @xmath0 for the moment map associated to the circle action @xmath375 @xmath376 the collapsed region @xmath377 will be called the _ exceptional divisor _ , it is a contact manifold of dimension @xmath378 .
the induced @xmath120action in the level set @xmath379 coincides with the action @xmath380 defined in theorem [ thm : bw ] with @xmath381 and @xmath382 .
thus , the orbit space is @xmath383 notice that both the topology and the contact structure of the exceptional divisor strongly depend on the choice of the parameters @xmath137 and @xmath134 .
consequently , so does @xmath374 .
* example * : 1 . in the case of a contact @xmath22fold , a transverse circle the simplest contact submanifold
is replaced by a ( quotient of a ) standard contact @xmath319sphere , as in section [ sec : srgy ] .
this new construction of the blow up along a transverse loop will be compared with the previous ones in the next section .
+ 2 . consider the contact blow
up along a contact @xmath319sphere @xmath384 .
the topology of the exceptional divisor will depend on the element of the corresponding higher homotopy group , cf .
section [ sec : bw ] .
if in the previous example @xmath25 is a @xmath22dimensional contact manifold , the exceptional divisor of the @xmath385 blow up is contactomorphic to @xmath72 . in higher dimensions , the exceptional divisor of a @xmath385 blow
up along @xmath72 is diffeomorphic to @xmath386 for @xmath387 and @xmath87 even .
we define the contact blow up along a contact submanifold with a general normal bundle
. the construction will clearly coincide with the previous blow up in the case of a trivial normal bundle . in smooth topology the smooth structure of a neighbourhood of a submanifold
is retained by the normal bundle .
the contact geometry nearby a contact submanifold @xmath289 is determined by the normal bundle @xmath257 along with a conformally symplectic structure .
such a structure exists because @xmath257 can be identified with the symplectic orthogonal @xmath388 .
the statement of the contact neighbourhood theorem is as follows : @xmath1932.5.15 in @xcite@xmath194 let @xmath389 and @xmath390 be contact pairs such that @xmath391 is contactomorphic to @xmath392 .
if @xmath393 as conformally symplectic bundles , then there exists a contactomorphism between suitable neighbourhoods of @xmath394 and @xmath395 .
there exist contact submanifolds with non
trivial normal bundle in a closed contact manifold .
let us provide some examples .
+ * examples * : 1 .
let @xmath396 be a cooriented contact manifold and @xmath24 itself be non trivial as an abstract vector bundle .
the contact form provides a contact embedding @xmath397 such that the normal bundle of the contact submanifold @xmath0 is isomorphic to @xmath24 .
let @xmath398 be an integral symplectic manifold and @xmath17 a symplectic submanifold with non
trivial normal bundle .
for instance , @xmath399 . then the normal bundle of the contact submanifold @xmath400 in @xmath307 is also non trivial .
let @xmath173 be a closed cooriented contact manifold .
consider an isocontact embedding @xmath401 see @xcite for the existence of such an embedding .
since the tangent bundle of the spheres are stable , it is simple to give sufficient conditions for the normal bundle to be non trivial , e.g. @xmath0 not spin .
the contact blow up construction has been used in another context . given a complex vector bundle @xmath8 on @xmath0 ,
the contact submanifold @xmath196 is defined as the vanishing set of a section in @xmath402 . then @xmath403)\neq0 $ ] .
this occurs for the base locus of contact lefschetz pencil decompositions of @xmath25 .
see @xcite . in the blow up construction for the trivial normal bundle case
there are two circle actions .
the first one exists on the contact submanifold @xmath4 , since it is a boothby
wang manifold , and it is extended to a local neighbourhood .
the second circle action is the gauge action provided by the complex structure in the conformally symplectic normal bundle .
the latter is still available in the non
trivial normal bundle case , the former can _ a priori _ no longer be extended in a neighbourhood .
+ we hence require a lifting condition for the circle action on @xmath4 : the appropriate set
up is depicted as in the diagram 1 in section [ sec : gromov ] : @xmath298 & \mathbb{s}(v ) \ar@{->}[d ] \\ & ( s,\xi_s)\cong\mathbb{s}_a(w ) \ar@{->}_{\pi}[d ] & ( v,\overline{\omega } ) \ar@{->}[ld ] \\ & ( w,\omega)}\ ] ] where @xmath284 is a symplectic bundle over a symplectic manifold @xmath17 .
assume @xmath404 for simplicity . + in the hypotheses above , the circle action provided by the boothby
wang structure can be naturally extended to a neighbourhood of @xmath4 .
since @xmath17 is a symplectically embedded submanifold of @xmath284 , @xmath400 is a contact submanifold of @xmath307 .
the tubular neighbourhood theorem tells us that the normal bundle @xmath3 is diffeomorphic to a small neighbourhood of @xmath4 in @xmath0 , but after the smooth isomorphism @xmath405 the same situation applies to @xmath400 in @xmath307 .
since the isomorphism holds at the level of symplectic bundles , the contact tubular neighbourhood theorem ensures that there exists a contactomorphism @xmath406 between a contact neighbourhood of the zero section in @xmath407 and a contact neighbourhood of @xmath400 in @xmath307 .
consequently , the circle action in @xmath307 can be carried along @xmath406 to a neighbourhood of @xmath4 .
let us spell out the moment map of the circle action .
we refer to the circle action on the normal bundle induced by its complex structure as the gauge action .
this action is the natural @xmath120action when working with a contact pair @xmath408 .
further , the radius coordinate @xmath409 is a global coordinate regardless of the non triviality of the normal bundle . the remaining action described above
will be referred as the boothby
wang action .
it is the natural action when identifying a neighbourhood of @xmath4 in @xmath0 with a neighbourhood of @xmath400 in @xmath307 via the map @xmath406 in the proof of the lemma .
the moment map of the @xmath120action @xmath410 is @xmath411 .
the moment map of the gauge action is @xmath412 . for the boothby
wang one , the circle action realizes the reeb vector field and thus its moment map is @xmath29 .
we express the @xmath413 coordinate through the contactomorphism @xmath406 as @xmath414 .
since we are using the action @xmath410 the statement follows .
recall that the contact cut can be performed if @xmath351 lies in the image of the moment map .
the same argument using a multiple of the gauge action concludes that we may modify the action in order to ensure this : @xmath406 maps the zero section to @xmath400 and thus the values of @xmath415 form a decreasing sequence in @xmath134 that eventually crosses zero , @xmath416 being bounded below .
the boothby
wang action may as well be arranged to period @xmath137 : the concatenation action is denoted @xmath380 .
we are in position to write the @xmath193contact blow
up@xmath194 let @xmath371 be a contact submanifold of @xmath25 .
let @xmath144 be such that the origin is contained in the image of the moment map @xmath417 for the action @xmath380 .
the @xmath193a , b@xmath194contact blow up @xmath374 of @xmath0 along @xmath4 is defined to be the contact cut of @xmath0 for the action @xmath380 , i.e. @xmath418 .
in this section we relate the three constructions of the contact blow up .
the construction that can be performed in the most general situation is the one involving the contact cut .
it has two degrees of freedom : a pair of positive integers @xmath137 and @xmath134 .
these two parameters relate to previous integers appearing in the first two constructions . indeed , the parameter @xmath419 in the contact surgery blow
up corresponds to @xmath134 . for gromov s
construction , the choice of collapsing radius @xmath420 gives rise , in the case of transverse loops , to the exceptional divisor @xmath421 and it corresponds to the parameter @xmath137 .
it is quite obvious that the diffeomorphism type of the blown up manifolds is the same regardless of the chosen construction as soon as the parameters coincide as just mentioned .
+ let us turn our attention to the contact structure : we restrict ourselves to the case of transverse loops . denote by @xmath422 the surgery contact blow
up defined in section [ sec : srgy ] with parameter @xmath134 .
the contact blow up as defined in section [ sec : gromov ] with radius @xmath137 is denoted by @xmath423 . and
@xmath424 will be the contact
cut blow up as defined in section [ sec : quotient ] , performed with parameters @xmath373 .
let us show that uniqueness holds in this case , more precisely we prove the following [ thm : unique ] let @xmath25 be a contact manifold . performing the blow up along a fixed transverse loop with the three procedures introduced previously , the resulting blown up manifolds @xmath425 , @xmath426 and @xmath427 endowed with the blown up contact structures are contactomorphic .
further , given any pair of integers @xmath373 , the following contactomorphisms hold : @xmath428 the relation between the different constructions is already hinted in section [ sec : srgy ] . since the exceptional contact divisors coincide and the procedure is of a local nature , i.e. the contact manifold is not altered away from a neighbourhood of the embedded transverse loop , the study should focus on the natural annulus contact fibration .
let us review a few facts .
+ a contact fibration is a fibration @xmath429 such that the fibers are contact submanifolds .
we consider contact fibrations over the disk @xmath430 .
the base being contractible , the fibration is trivial and we also assume it to be trivialized . let us introduce the following let @xmath431 be polar coordinates on the disk @xmath432 .
a trivialized contact fibration over the disk @xmath433 is said to be radial if the contact structure admits the following equation @xmath434 where @xmath435 is a smooth function such that @xmath436 .
notice that for the total space of a radial contact fibration to have an induced contact structure it is required that @xmath437 it is convenient to extend the previous definition in order to include the general situation , where lens spaces may appear as exceptional contact divisors : the action in the fiber sphere @xmath96 is generated by an @xmath440rotation along the hopf fiber , whereas the action in the base @xmath432 is the standard @xmath440rotation in the disk .
they preserve respectively the standard contact structure in @xmath96 and the @xmath29form @xmath441 in the disk .
hence , the fibration becomes equivariant if the function @xmath125 is preserved by the action .
+ topologically it is fairly straightforward that the blow up operations we are performing are tantamount to a priori different fillings of the fibration over an annulus to form a manifold lying over the disk this being always considered up to a finite action @xmath439 , for lens spaces fillings .
the transition from @xmath174 to @xmath176 can be understood in these terms : both fibrations over the annulus produced by restricting to @xmath442 are filled in the origin with a circle and @xmath96 respectively . in the transverse loop case
it will be enough to use the following [ lem : unique ] let @xmath284 be a manifold with contact structures @xmath443 and @xmath444 .
assume that there are two smoothly isotopic diffeomorphisms @xmath445 which are contactomorphisms for @xmath443 and @xmath444 respectively .
let the two fibrations be radial contact fibrations with common contact fiber @xmath446 and satisfying that the diffeomorphism @xmath447 is the identity close to the boundary .
then , the contact structures @xmath443 are @xmath444 are isotopic .
+ further , if the fiber is @xmath448 and the contact fibrations are @xmath439equivariant , the contact structures are isotopic through @xmath439equivariant contactomorphisms . this can be reduced to the setup with a fibration @xmath449 with two different radial contact structures @xmath450 such that the hamiltonians @xmath451 and @xmath452 coincide close to the boundary . in that setting
, we just need to construct a path of functions @xmath453 connecting them , relative to the boundary , satisfying the contact equation ( [ eq : contact ] ) and the condition @xmath454 .
but this is possible since the space of such functions is convex .
+ the argument still works in the equivariant case : the only sentence to be added is that the space of equivariant hamiltonians is also convex . thus , to conclude uniqueness we study the contact topology of the different blow up constructions and ensure that the lemma applies .
+ _ proof of theorem [ thm : unique ] _ : let us describe the common model fibration that underlies the three constructions in this case . consider a standard contact neighbourhood @xmath455 of the given fixed loop and the morphism @xmath456 it does generalize the diffeomorphism provided by equation ( [ eq : change ] ) that reflects the case @xmath404 . if @xmath137 is greater than @xmath29 , it becomes a @xmath457 covering .
the covering transformation is provided by @xmath439 acting through : @xmath458 which is free as long as @xmath127 . to understand the change in the contact structure , note that the pull back of the standard contact form @xmath459 is given by @xmath460.\ ] ] denote by @xmath461 the critical radius where the distribution becomes horizontal . in these coordinates , for any fixed small @xmath268 , the projection onto the first two factors @xmath462 provides a radial contact fibration on the annulus and since the function @xmath463 is strictly positive in @xmath464 , it can be extended to the interior of the disk to a @xmath439equivariant radial contact fibration . in order to glue back the model to the manifold
we should quotient the equivariant contact fibration by @xmath439 , this allows us to use the map @xmath465 to insert the model back into the manifold .
+ it is thus left to verify that the three blow up procedures provide examples of such an extension for particular values of @xmath373
. then lemma [ lem : unique ] will apply to provide the uniqueness of the constructions .
note that the contact surgery blow up
construction is by definition a radial contact fibration , with @xmath404 , as shown in section [ sec : srgy ] .
let us study the two remaining cases .
+ to understand the proof in section [ sec : gromov ] , let us proceed backwards and instead of applying the boothby
wang construction , we produce a contact structure and then quotient the resulting contact manifold by the reeb @xmath120action to study whether it is the correct object .
once the coordinate change @xmath465 is performed , the reeb vector field @xmath466 becomes @xmath467 this vector field extends to the interior of the disk fibration and so we may quotient the resulting manifold @xmath241 .
we obtain the blown up symplectic ball @xmath468 as its quotient .
we can further quotient by the free @xmath439action to obtain a non trivial fibration over the disk @xmath432 .
this proves that a suitable choice of connection leads to an equivariant contact fibration .
+ there are other choices of connection though . from the principal bundle point of view , a radial contact fibration over the annulus @xmath469 corresponds to a connection on @xmath470 certainly , after proposition [ propo : bundle_blow ] the contact structure is fixed with the choice of a connection .
note that the space of connections is affine and thus , after gray s stability theorem , the resulting contact structures are contact isotopic for different choices of connections . in conclusion
, this second model also provides an extension of the model fibration .
+ we describe the third procedure also beginning with the resulting contact manifold and giving the pull back of the action .
this contact cut construction is also an equivariant radial contact fibration since the pull back of the vector field generating the @xmath120action , that is @xmath471 is expressed as @xmath472 after the coordinate change .
there , the contact cut is just an equivariant radial contact fibration , see the proof of theorem 2.11 in @xcite .
@xmath340 + using lemma [ lem : unique ] , we can show that the contact blow
up is unique up to the choice of a trivializing chart of the neighbourhood of the transverse loop . in order to prove the uniqueness of the blow up along transverse loops , we would need to study the space of isocontact embeddings of the contact manifold @xmath174 in @xmath0 .
it is probably false that it is connected , which is the requirement needed to ensure the uniqueness of the blow up once the parameters @xmath473 are fixed .
xxxx r. bott , l. tu , differential forms in algebraic topology , _ gtm _ 82 , springer
w. boothby and h. wang , on contact manifolds , _ annals of math .
_ 68 ( 1958 ) , 721734 .
r. casals , d. pancholi , f. presas , almost contact @xmath22folds are contact , arxiv:1203.2166v2 .
y. eliashberg , s. s. kim , l. polterovich , geometry of contact transformations and domains : orderability versus squeezing , _ geom .
_ 10 ( 2006 ) , 16351747
. o. van koert , open books on contact five - manifolds , _ annales de linstitut fourier _ 58 ( 2008 ) , 139157 .
h. geiges , an introduction to contact topology , _ cambridge studies in advanced mathematics _ 109
. h. geiges , constructions of contact manifolds , _ math .
_ 121 ( 1997 ) , 455564 .
e. giroux , structures de contact en dimension trois et bifurcations des feuilletages , _ inventiones math .
_ 141 ( 2000 ) 615689 .
m. gromov , partial differential relations , _ ergeb . math .
_ 9 ( 1986 ) , springer
r. hartshorne , algebraic geometry , _ gtm _ 52 , springer verlag .
e. lerman , contact cuts , _
israel j. math .
_ 124 ( 2001 ) , 7792 .
e. lerman , symplectic cuts , _ math .
_ 2 ( 1995 ) , 247258 .
d. mcduff , d. salamon , introduction to symplectic topology , _ oxford math .
( 1998 ) , oxford .
f. presas , lefschetz type pencils on contact manifolds , _ asian j. math .
_ 6 ( 2002 ) , 277302 .
f. presas , a class of non fillable contact structures , _ geometry & topology _ 11 ( 2007 ) 22032225 .
y. wang , w. ziller , einstein metrics on principal torus bundles , _ j. differential geom .
_ 31 ( 1990 ) , 215248 . | we introduce definitions of contact blow up from several perspectives .
such different approaches to the contact blow up are related .
we prove that the contact topology coincides in the case of blow
ups along transverse embedded loops . |
a blurred image carries less information than a sharp image . due to the confined focal depth , it is hard to capture an image in which all objects / areas of the scene appear quite sharp . only the objects of a scene that are near the focus plane or at the focus plane appear sharp . to handle such situations the images
are acquired using a number of imaging sensors or with multi focus imaging . after taking images , high contrast regions from each of the acquired images
are selected and fused together to create an image that is in focus everywhere .
the fusion process reduces the uncertainty and redundancy from the source images .
there are a number of benefits of using the fusion process including reduced uncertainty , wider temporal and spatial coverage and improved reliability @xcite .
image fusion is basically used due to the limited characteristic of image sensors .
visual sensor networks are crucial in monitoring and surveillance , tracking and object recognition .
now - a - days visual sensor networks provide help in solving various kind of problems related to research .
there are multiple applications of multi - focus imaging such as in macro - photography , wireless image sensor network , focus stacking etc . in the case of wireless image sensor networks
( wisn ) , it is hard to describe the critical situation precisely with a single image . a number of sensors are utilized to receive images of the same scene , and a centralized fusion centre cartels images from various sensors into a single image , the resultant image is amicable for human visual and machine processing @xcite .
then , the blended image is channelized to a higher node .
the simplest image fusion approach is averaging in which two images are merged by taking average of their individual pixels .
the other approaches are : choose maximum absolute or minimum absolute value of pixel intensity in two or more images .
but , these approaches suffer from lower performance and bad visual quality .
further , these methods are integrated with different spatial and frequency domain processing methods to improve the output image quality @xcite .
initial development of fusion takes place in spatial domain because of its simplicity and low computational complexity .
but , use of spatial domain processes result in many undesirable effects such as blurring , low contrast and other spatial degradation .
these undesirable effects imposed by spatial domain algorithms can be handled in frequency domain .
so , the focus is shifting to multi - scale transforms and its various types of applications .
various fusion algorithms for multi - focus images are available using multi - scale transform @xcite .
the most used multi - scale transform is wavelet transform .
there are a number of algorithms for image fusion that use wavelet transform @xcite . in @xcite
a statistical sharpness measure is used on wavelet coefficients to perform adaptive image fusion in the wavelet domain . instead of using dwt to decompose images into the frequency domain , discrete stationary wavelet transform ( dswt )
is used to overcome the lack of translation invariance of the dwt @xcite .
as the wavelets do not represent long edges well in the fused images , multi - focus image fusion is performed by combining both the wavelet and curvelet transforms to improve the quality @xcite .
methods based on multi - scale transforms such as non - subsampled contourlet transform ( nsct ) , shift invariant discrete wavelet transform ( sidwt ) , and discrete wavelet transform ( dwt ) are popular @xcite . however , a majority of the image blending methods based on multi - scale transform are complex and time - consuming , which limit their applications for wireless visual sensor networks equipped with constrained resources .
compliance with the image compression standards provide greater flexibility and portability to image fusion processes and algorithms .
the most widely used image compression standards are jpeg and jpeg2000 . in spite of having greater advantages over jpeg
, jpeg2000 is not used in many in applications because of its complexity . in jpeg
discrete cosine transform ( dct ) is used to compress the images .
so , the dct domain image fusion will be efficient and less time consuming .
if the raw images are fused and compressed at source end ( where images are generated ) , the compressed images take less bandwidth and less space in transmission and storage for further processing . in this paper
we propose a method based on contrast measurement in the dct domain with a reduced computational complexity . the proposed method has equal / better visual and quantitative performance as that of methods proposed in @xcite , @xcite and @xcite .
the remaining paper is arranged as follows : section ii presents the brief discussion on dct domain fusion concept .
section iii outlines the proposed approach of image fusion .
section iv explains the hardware architecture .
section v analyses the experimental results , followed by conclusions in section vi .
this section presents some of the most widely used concepts and algorithms based on dct domain processing . in all of the dct
based fusion methods the maximum or average rule for fusion is applied as per the different contrast measurement criteria .
jinshan tang @xcite proposed two methods namely ; dct+contrast and dct+average . in dct+average
, the dct coefficient of the image is averaged to form a new fused image , but due to the averaging of the dct coefficients the output image gets blurred .
this method is not preferred due to its poor image quality .
the dct+contrast method utilizes the concept of contrast assessment presented by tang et al .
@xcite . in this method ,
each of the blocks is partitioned into fifteen bands of frequency as shown in figure 1 .
the sharpness of each of the coefficients in a frequency band is given as : @xmath2 where @xmath3 is the average amplitude over the spectral band .
both the methods are criticised because of their poor performance on different evaluation criteria @xcite .
haghighat et al .
@xcite proposed another method based on variance in dct domain , namely dct+variance . since variance is a better criterion than mean to estimate the contrast in the image .
this method leads to a better performance than dct+contrast and dct+avarage .
the variance in dct domain is given as : @xmath4 1 1 however , experimental results show that the use of variance results in a lower performance compared to that of other focus measures @xcite .
the variance calculation is computationally complex because it involves floating point multiplication and division .
so , it may not be suitable for resource constrained environments .
the dct+variance method is also criticised due to its computational intensive nature . to overcome the computational intensive nature , phamila et al .
@xcite have proposed another method based on number of high valued ac coefficient in a dct block .
though his method reduces the overall computational complexity and energy requirement , but number of high value ac coefficient is not a good criterion to identify the maximum variance block in an image .
cao et al .
@xcite proposed a method named as dct+sf .
the spatial frequency calculation in dct domain is given as : @xmath5 where _
e(i , j ) _ is the spatial frequency operator in the dct domain .
calculating spatial frequency in dct domain is computationally more complex than dct+variance . due to small size of sensors , various kinds of design constraints
are imposed by wireless images sensor network ( wisn ) on designers such as computational complexity , computational power , communication channels , image compression , and power consumption . for resource constrained environments such as wisn , space application or any remote sensing application , the fusion method should exhibit following properties : 1 ) it should be computationally less complex , 2 ) it should consume less power , 3 ) should be easier to implement on different platforms such as hardware , software or dsp platforms .
in the dct domain coefficient _ d(0,0 ) _ represents the dc value , also know as the mean value of the image block .
the rest of the dct coefficients ( ac coefficients ) represent the high frequency information of the image block . determining the spatial frequency through the ac components of dct
is a computationally intensive task and also not suitable for resource constrained real - time applications such as a wisn .
the coefficients of dct itself represent spatial frequency content in given frequency band .
as dct is the part of the discrete fourier transform ( dft ) , the definition of contrast measurement in dft can also be utilised for dct .
the issue of contrast of complex scenes at different spatial frequencies can be addressed as follows @xcite :
@xmath6 where c represents the sharpness of image in the dft domain , dc is zero frequency content and @xmath7 is the dft coefficient .
since , the dc component in the dft domain has the maximum contribution towards the image brightness , it can be neglected in computation of contrast in dct domain . hence , equation ( [ sch : eq2 ] )
can be modified for the dct domain as : @xmath8 so , instead of computing the spatial frequency , variance or maximum number of high valued ac coefficients in a block using all the transformed ac coefficients , the proposed method adds all of the dct coefficients to measure the contrast of the dct block .
thus computation is drastically reduced and the contribution of individual coefficient is also taken care of . the schematic block diagram of multi - focus image fusion in dct domain is shown in figure 2(a ) . only the images
a and b are considered for simplicity ; but , the procedure can be applied to fuse more than two source images .
the flow chart of the proposed algorithm is shown in figure 2(b ) .
the blending process comprises the following steps : + 1 .
take the source images a and b , divide these into @xmath9 blocks .
denote the blocks of the two images as @xmath10 and @xmath11 .
calculate the dct of the blocks and add the dct coefficients of each of the blocks according to equation ( [ sch : eq3 ] ) and denote the sums as @xmath12 and @xmath13 .
+ 3 . compare @xmath12 and @xmath13 and depending on which of these two is higher , select either a or b. based on the selection prepare the decision map @xmath14 as @xmath15 4.apply consistency verification to mend the visual quality of the output image and use a majority filter to redefine the decision map ; then , construct the dct fused image based on the new decision map as + @xmath16 5.the standardised quantization table is used to quantise the obtained dct coefficients in the jpeg coder and then the output bit stream is entropy coded . + the proposed method compares the absolute amplitude , so , it is named as dct+amp_max in further section .
the block diagram of the proposed multi - focus image fusion ( dct+amp_max ) architecture is shown in figure [ fig : block ] ( a ) .
the proposed architecture uses an @xmath17 dct .
the fused dct coefficients can be further compressed by a jpeg encoder or can be converted back to image by taking inverse dct ( idct ) of the coefficients .
idct architecture is same as the dct architecture with slight changes in coefficients . the block wise description is as follows : 1 1 the presented architecture uses an @xmath17 dct .
it is required to configure memory in such a manner that 8 pixels per clock can be accessed , which is not possible with single memory module .
either we need to choose 4 dual port memory modules or 8 single port memory modules . in this architecture
we have used 8 single port memory modules to get 8 pixels / clock cycle . since only high frequency ac coefficients of the dct
are required to make the decision for fusion , so , an improved dct with 35 bit ( 1 sign bit , 10 integer bit , 24 fractional bit ) precision is synthesised .
the dct architecture is synthesised using two 1d - dct s with an s - ram based memory transposed architecture @xcite .
the dct architecture is shown in figure [ fig : block ] ( b ) .
the idct architecture is same as the dct architecture .
both the architecture working at same frequency .
the presented architecture can be modified easily for other high speed compression applications .
.5 0.5 this block is used to select one out of two @xmath18 dct blocks
. this block sums up all absolute ac coefficients of each dct block and compares the two sums .
accordingly , it generates an output of zero or one .
simultaneously , it also prepares the decision map for consistency verification .
majority filter is applied on decision map and using newly constructed decision map , the dct coefficient are selected according to equation 7 .
the architecture of majority filter and decision block is shown in figure [ fig : architecture1 ] .
the dct coefficient are required for fusion .
the fifo architecture helps to preserves a dct coefficient till the final selection of the coefficients .
the present fifo is synthesised using s - ram based memory . based on output of decision block the data select
block selects one set of the dct coefficients out of two sets of dct coefficients and provides the final fused coefficient .
the dct+amp_max method is applied to a set of reference and non - reference images and its performance is evaluated on a standard pc ( intel core i5 processor , 3.10 ghz , 4 gb ram ) .
extensive evaluation is carried out on different images .
some of the images are shown in figure 5 @xcite . in this section ,
we compare the performance of our method with that of the existing image fusion methods in the dct domain , like dct+ac - max , dct+ac - max+cv , dct+variance , dct+variance+cv and dct+sf , dct+sf+cv . in the first set of experiments ,
the proposed method namely , dct+amp_max is evaluated by fusing 50 pairs of blurred images which are generated using the standard gray scale images shown in figure 5 with an averaging filter .
complementary regions of the images are blurred in every pair of image .
the standard gray scale images are taken as ground truth images .
table [ sch : tb1 ] lists the final universal image quality index ( uiqi ) @xcite , peak signal to noise ratio ( psnr ) , mean square error ( mse ) , mutual information(mi ) and structural similarity index measure ( ssim ) @xcite values of 50 experimental images
. these are fusion evaluation criteria requiring reference images .
the performance of the proposed method is approximately identical to the dct+sf method and it has a better performance compared to that of the dct+variance , and dct+ac - max fusion method . [
cols="^,^,^,^,^,^ " , ] [ sch : tb5 ]
the different design constraints such as power consumption , computational power , image compression , and communication channels for wisn / camera system can be handled by low complexity and high performance algorithms . in terms of quantitative evaluation ,
the performance of the proposed method is approximately identical to most recent dct+sf method at reduced computational complexity .
the method requires only @xmath0 addition for the fusion of @xmath1 size of image .
the hardware implementation of the proposed method consumes less power ( 250 mw at 200 mhz ) .
the fusion approach adopted in the proposed method does not require any floating point arithmetic operations like variance , mean and spatial frequency calculation , which makes it suitable for resource constrained battery powered sensors for energy efficient fusion and subsequent compression .
the proposed architecture is also suitable for high speed applications , portable and hand - held devices because it can process more than 60 frames per second with the resolution of 4k and acquired very less area .
lei tang , feng zhao , zng - gui zhao,the nonsubsampled contourlet transform for image fusion , " _ proceedings of the 2007 international conference on wavelet analysis and pattern recognition _ ,
beijing , china , 2 - 4 nov .
2007 .
l. cao , l. jin , h. tao , g. li , z. zhuang , and y. zhang , multi - focus image fusion based on spatial frequency in discrete cosine transform domain , " _ ieee signal processing letters _ , vol .
220 - 224 , feb . 2015 .
q. shang , y. fan , w. shen , and s. shen , single - port sram - based transpose memory with diagonal data mapping for large size 2-d dct / idct ,
_ ieee transaction on very large scale integration systems _ , vol .
2422 - 2426 , november 2014 .
zhou wang , alan conrad bovik , hamid rahim sheikh and eero p. simoncelli , image quality assessment : from error visibility to structural similarity , " _ ieee transaction on image processing _ , vol .
600 - 612 , 2004 .
haghighat , a. aghagolzadeh , h. seyedarabi , `` a non - reference image fusion metric based on mutual information of image features , '' _ computers and electrical engineering _ , vol .
744 - 756 , sept . 2011 | due to the confined focal length of optical sensors , focusing all objects in a scene with a single sensor is a difficult task . to handle such a situation ,
image fusion methods are used in multi - focus environment .
discrete cosine transform ( dct ) is a widely used image compression transform , image fusion in dct domain is an efficient method .
this paper presents a low complexity approach for multi - focus image fusion and its vlsi implementation using dct .
the proposed method is evaluated using reference/ non - reference fusion measure criteria and the obtained results asserts it s effectiveness .
the proposed method uses only @xmath0 addition to fuse the two @xmath1 image block and consumes only 250 mw power at 200 mhz operating frequency .
the maximum synthesized frequency on fpga is found to be 221 mhz and consumes 42% of fpga resources .
the proposed method consumes very less power and can process 4k resolution images at the rate of 60 frames per second which makes the hardware suitable for handheld portable devices such as camera module and wireless image sensors .
discrete cosine transform , multi - focus image , image fusion , vlsi implementation . |
in december 1998 , in the aftermath of el nino and its companion , la nina , the weather in the roanoke , virginia , area was unusually mild .
weather data have been collected here since 1934 , and record highs and lows for any particular day are known . as part of the daily weather forecasts ,
local tv stations report these record values and compare them to the highest and lowest temperature values of the day .
remarkably , during the _ nine _ days from november 29 to december 7 , 1998 , the previous record highs were _ broken five times and tied once_!@xcite one might wonder , as we did , how frequently such a series of records could possibly occur .
when only few weather data are available , such as in the early years of record keeping , it is obviously quite easy to experience new extremes .
however , as the data sets become larger , the record highs ( and lows ) are pushed to higher ( lower ) values , so that the setting of a new record becomes a far less frequent event .
thus , we began exploring questions such as : how probable was it to set a new record in 1998 , 64 years after the records began
? how do record highs increase with time ? how long do records typically survive before they are broken ? not surprisingly , similar questions have been posed before .
it appears that the earliest studies are due to n. bernoulli who analyzed life expectancies in 1709 @xcite .
later , flood control and structural safety issues were considered , to name just a few of the numerous applications .
beginning in the 1920 s , the mathematical techniques were developed and the study of records , or extremes , became known as `` extreme value statistics,''@xcite which is still an active area of research . in this paper , we provide a pedagogical introduction to some of the most basic results .
we begin in section ii with the simplest model for records and a concise statement of the problem .
there is no attempt to address real records . as a result of complex physical processes ,
the statistics of real weather records is undoubtedly far more intricate than that generated by simple random numbers .
the quotation marks in the title should remind the reader that actual weather records are _ not _ the subject of this article .
section iii is devoted to the full distribution function for the model records , their averages , and standard deviations .
next , we discuss the record lifetimes and derive the associated distribution in section iv .
although we focus on `` record highs , '' a completely analogous line of reasoning can be pursued for `` record lows.''@xcite in the final section , we turn to more general questions and conclude by listing a number of problems which , to the best of our knowledge , are still unsolved . some technical details are provided in the appendix .
our much simplified model for the physical data ( temperatures , water levels , etc . ) is based on a probability density @xmath0 for a real variable @xmath1 .
for example , if we are considering temperatures , @xmath2 might be the probability that the temperature at noon , on a specific day of the year , takes a value between @xmath1 and @xmath3 .
the `` data '' in our simplified model are just a sequence of random numbers : @xmath4 in keeping with the language of weather records , we refer to the ( integer ) label of these numbers as `` time '' ( or `` year '' ) : @xmath5 each of the @xmath1 s is drawn from the _ same _ _ distribution _ @xmath0 .
thus , our data form a set of independent , identically distributed random variables .
this condition is probably the most serious shortcoming when applied to physical reality , where major correlations or variations can be expected .
for example , it precludes `` global warming , '' a situation in which the underlying distribution @xmath0 itself is a function of time . before discussing the sequence and the records ,
let us provide some details about the distribution @xmath0 .
it may be defined over the entire real axis or restricted to an interval , that is , it may have infinite or finite support . for simplicity , we require it to be _ continuous and positive _ , which excludes , for example , dice throwing . as a result , we can ignore the possibility of ties . of course , it must be properly _ normalized _ , that is , @xmath6 . here and in the following ,
any integral limits that are not explicitly specified are to be taken as the appropriate ( upper and/or lower ) bounds of the support .
because the distribution @xmath0 reflects how we model our data set , we refer to it as the `` model distribution , '' or simply `` the model . ''
in particular , we will investigate to what extent our results for record distributions and lifetimes depend on the underlying model .
for later reference , we also introduce a second distribution , simply related to @xmath0 : @xmath7 clearly , @xmath8 is the probability that a random number , drawn from the distribution @xmath0 , will not exceed @xmath9 .
thus , @xmath8 is a monotonic function , varying from 0 to 1 .
the reader familiar with random numbers on a computer will recognize that the inverse of this function , @xmath10 , is just the operation to generate random numbers for an arbitrary density @xmath0 from the uniform distribution on the interval @xmath11 $ ] .
following galambos@xcite , we call @xmath8 the `` common distribution function . ''
returning to the sequence ( [ sequence ] ) , we define the `` record '' @xmath12 as the _ largest _ number in a string of length @xmath13 : @xmath14 where we have included @xmath15 as the ( arbitrary ) initial value . after generating the next new number @xmath16 , we determine whether the record has been broken , according to @xmath17 this procedure is continued until we have obtained a sequence of length @xmath18 .
of course , the records @xmath12 themselves are stochastic quantities .
so , we can define a density for @xmath9 : @xmath19 so that the probability for finding the record to lie between @xmath9 and @xmath20 is just @xmath21 .
our goal is to determine , given the underlying model distribution @xmath0 , this @xmath22 or , at a simpler level , the average record : @xmath23 before delving into the analytic aspects , let us consider computer simulations of these records . to study their statistics , we generate a large number , @xmath24 , of sequences ( modeling , for example , weather data from @xmath24 cities ) : @xmath25 .
based on this ensemble of sequences , we can define the average record as a function of time @xmath26 as an ilustration , following are _ two _ ( @xmath27 ) sequences of _ ten _
( @xmath28 ) random numbers generated by rand ( ) in ms - excel ( quoted to 4 digits for simplicity ) : @xmath29 @xmath30 and the associated sequences of records : @xmath31 @xmath32 the _ average _ record is , therefore , @xmath33 since rand ( ) outputs a random number uniformly distributed between 0 and 1 , it is hardly surpirsing to see that the average record bumps up against 1 here . to get a good grasp of this process , the reader should attempt , say , 100 sequences of @xmath34 on a spreadsheet . on a typical modern pc ,
it takes only `` the blink of an eye '' , literally ! for the computer simulations presented in this article , we used a high quality random generator ( ran2 from _ numerical recipies_@xcite ) and averaged over @xmath35 sequences of length @xmath36 .
this procedure requires just a few minutes running linux on a 450 mhz pentium ii , and produces excellent statistics .
in fact , with @xmath37 sequences , it is possible to generate a reasonable picture of the whole probability distribution for the records
. a theoretical approach to the averages will be our next task .
our goal in this section is to find an analytic form for the probability density @xmath22 .
we will do this recursively , by assuming that we know the value of the record , say @xmath38 , at time @xmath39 .
then , we seek the _ conditional _ probability , @xmath40 , that the record , at time @xmath13 , has the value @xmath9 , provided it had the value @xmath38 at time @xmath39 .
fortunately , it is very easy to write down this quantity !
clearly , it vanishes for @xmath41 , because the new record can not be smaller than the old one .
this leaves us with two possibilities : either the old record stays the same so that @xmath42 , or it is broken , resulting in the new value @xmath43 .
the former is the case if the new random number , @xmath44 , does not exceed @xmath38 .
reviewing eq .
( [ q - def ] ) , we see that this case occurs with probability @xmath8 .
in contrast , if @xmath44 exceeds @xmath38 , then it sets the new record @xmath9 .
this latter case occurs with probability @xmath45 .
summarizing , the conditional probability is @xmath46 the heavyside step function @xmath47 i unity if its argument is positive and zero otherwise .
its derivative is just the dirac delta function @xmath48 , so that the two terms in eq .
( [ conditionalp ] ) can be combined into @xmath49 \ .
\label{condp}\ ] ] because we have exhausted all possibilities for @xmath9 , the conditional probability must be normalized with respect to an integration over @xmath9 . this condition is easily checked : because @xmath50 is a total derivative , its integral is just @xmath51 evaluated at the limits .
so , we have @xmath52 . from the _
conditional _ probability , it is a simple step to arrive at our main target : the record probability density ( the `` record '' distribution ) @xmath22 : @xmath53 substituting the explicit form ( [ condp ] ) for the conditional probability , we obtain a recursion relation for @xmath50 : @xmath54 p(r',t-1 ) \nonumber \\ & = & \frac \partial { \partial r } \ !
\left [ q(r ) \ !
! dr'\,\,\theta ( r - r')\right ] \ !
p(r',t-1 ) \nonumber \\ & = & \frac \partial { \partial r}q(r ) \ !
dr'\,\,p(r',t-1 ) .
\label{p_rec}\end{aligned}\ ] ] clearly , this form is still slightly unwieldy , due to the integration .
let us define the `` barrier '' distribution @xmath55 to be the probability that at time @xmath13 the record is at @xmath9 or lower : @xmath56 so that @xmath57 from eqs . ( [ p_rec ] - [ dqdr ] ) , the function @xmath58 satisfies a very simple recursion equation , namely , @xmath59 equation ( [ q_rec ] ) has a simple interpretation .
recall that @xmath8 is the probability that the next random number is less or equal to @xmath9 , and @xmath55 is the probability that , at time @xmath13 , the record has _ not exceeded _
the product of the two gives the probability that the record remains unbroken after the next time step . the recursion relation ( [ q_rec ] )
is easily solved : @xmath60^t q(r,0)=\left [ q(r)\right]^t \label{q = q^t}\ ] ] because @xmath61 for all @xmath62 .
we can deduce two important properties from this general solution for this barrier distribution @xmath58 . because @xmath8 is a monotonic function ,
so is @xmath63 .
this behavior of @xmath63 may be interpreted as at any given time , it is more difficult to go over a higher barrier ( break a record ) . on the other hand , because @xmath64 for any fixed @xmath65 @xmath58 _ decreases _ with @xmath13 which implies that any given record can be broken , provided we wait long enough ! we emphasize that these `` sensible '' results are completely independent of the details of the underlying distribution @xmath0 . from ( [ q = q^t ] ) , the record distribution follows : @xmath66 ^{t-1}\frac{\partial q(r)}{\partial r } = t\left [ q(r)\right ] .
^{t-1}p(r ) \label{p_simple}\ ] ] once @xmath22 is known , we can compute average values for the records , @xmath67 , through the defintion ( [ averager ] ) , as well as the standard deviations and all higher moments . for later reference
, we provide a simplified representation of the integral in eq.([averager ] ) . substituting ( [ p_simple ] ) into ( [ averager ] )
, we have @xmath68 ^{t-1}p(r)$ ] .
next , recall that @xmath8 is monotonic , so that it can be inverted uniquely to give the function @xmath10 .
now , change the integration variable from @xmath9 to @xmath69 , using @xmath70 .
the result is @xmath71 note that the limits of integration are now _ unique _ , that is , independent of the underlying model ! in the next section , we will illustrate the characteristics of @xmath22 and @xmath67 with two simple , explicitly solvable distributions @xmath72 .
let us illustrate our results by considering two particularly simple cases : a flat distribution @xmath73 and a pure exponential @xmath74 these distributions differ significantly in that the former has an _ upper bound _ for the allowed values of the data . as a result
, the possible record values are also bounded ! in contrast , the latter distribution extends to infinity , setting no limits on the possible records .
studying these two simple distributions will lead us to find rather different , but hopefully generic behavior for bounded versus unbounded models . the flat distribution ( [ p - flat ] ) corresponds to data whose values are equally probable over a given interval .
( [ q = q^t ] ) , we find @xmath75 ( for @xmath76 ) and @xmath60 ^t = r^t \label{q - flat}\ ] ] so that @xmath77 both results are easily interpreted .
the barrier distribution @xmath55 displays explicitly the general properties discussed above : increasing with @xmath9 at fixed @xmath13 and decreasing with @xmath13 at fixed @xmath78 .
in contrast , the record distribution @xmath22 displays a _
maximum in _
@xmath13 for a fixed @xmath9 . for early times , the probability to find the record at @xmath9 is low because this value has not yet been reached , whereas , for late times , it has already been exceeded !
we can also study @xmath22 as a function of @xmath9 for a fixed @xmath13 .
the maximum value always occurs at @xmath79 , increasing linearly with time .
if we apply the normalization condition @xmath80 , the width of the peak must narrow with time . in other words , the late time records are `` piled up '' just below the highest allowed value ( unity in this case ) .
let us investigate how this feature is reflected in the _ average _ record .
( [ averager ] ) , we find easily @xmath81 as expected , @xmath67 increases monotonically as a function of time , reaching its upper bound at @xmath82 : _
t r(t)=1 of course , its _ rate _ of increase must vanish in this limit . in this case
, the asymptotic is @xmath83 .
this behavior is illustrated in fig .
1a , which shows that the exact result , eq .
( [ avr - flat ] ) is in excellent agreement with monte carlo data averaged over @xmath84 sequences with 1000 entries each .
several of these properties are generic in the following sense .
if the underlying distribution @xmath72 has an upper bound , that is , @xmath85 for @xmath86 , then @xmath87 . not surprisingly , the behavior of @xmath72 near @xmath88 will dictate the asymptotics .
for example , if we assume @xmath89 , so that @xmath90 for @xmath1 , then @xmath91 , and we can show that @xmath92 , which is a generalization of the flat case .
we next turn to the purely exponential distribution ( [ p - exp ] ) and its associated @xmath93 . the two characteristic distribution functions are @xmath94^t \label{q - exp}\ ] ] and @xmath95^{t-1}\ .
\label{p - exp}\ ] ] many properties are qualitatively similar to the flat case .
one difference is that the position , @xmath96 , of the maximum in @xmath22 increases with time indefinitely : @xmath97 once again , this behavior is reflected in the average record .
( [ rtoq ] ) and deferring details to appendix a , we obtain @xmath98 which is obviously a monotonically increasing function of time . to exhibit the asymptotic behavior for large times
, we can write eq .
( [ avr - exp ] ) in more compact notation : r(t)= ( t+1 ) - ( 1 ) , where @xmath99 is the digamma function with known asymptotic form.@xcite as a result , @xmath100 where @xmath101 .
note that the _ rate _ of increase also vanishes with time , but with a slower decay , @xmath102 . in fig .
1b , we show the comparison of the asymptotic result ( [ avr - exp - as ] ) with monte carlo data .
the agreement is clearly excellent .
as before , we can consider the behavior of the average record for more general unbounded distributions @xmath0 . here
, the argument rests on a saddle - point approximation for @xmath22 around its maximum value , @xmath96 .
asymptotically , the average can be approximated by the position of the maximum . because @xmath96 becomes very large for late times
, the asymptotics should be controlled by the behavior of @xmath0 for large @xmath1 .
some examples of possible asymptotic behaviors of unbounded @xmath72 s and their associated saddle points @xmath96 are 1 . a power law @xmath103 ( with @xmath104 to ensure that @xmath105 exists ) , resulting in @xmath106 ; 2 .
an exponential @xmath107 , giving @xmath108 ; 3 .
a gaussian @xmath109for which @xmath110 .
we invite the reader to carry out simulations for , say , @xmath111\ ] ] and check the results against the predictions in item 1 .
we would also like to caution the reader that the approach to asymptopia for the gaussian is _ extremely _ slow .
in particular , we find that this regime lies beyond @xmath112 .
next , we turn to a key question in the study of floods or earthquakes .
what is the typical time span between two large events ? at the practical level , how much time do we have to construct dams or to repair levees before the next record - breaking flood ?
of course , our study will not provide an _ exact time span _ until the next disaster ; it will only give an _ estimate _ of how long a record might survive .
we refer to the time span between the setting of a record and its subsequent breaking as its `` lifetime '' ( or `` record time''@xcite ) . in the following ,
we investigate the distribution of these record lifetimes .
specifically , we will consider data sequences with @xmath18 entries .
for each sequence , we will identify the associated records and their lifetimes . by analyzing a large number of data sequences
, we can compile a histogram , @xmath113 , of how likely a record would survive for a time span @xmath114 .
let us introduce a simple representation of all possible histories of records : a tree - like structure .
we begin by generating a data sequence @xmath115 . because we only need to know _ where _ the records occur in this sequence , we can associate this sequence with the following binary string .
if @xmath116 is a new record , replace it by the letter @xmath9 ( `` record '' ) ; otherwise , replace it by @xmath117 ( `` lower '' ) .
note that , had we used a discrete underlying distribution @xmath72 , then we would have to consider the complication of ties . as an example
, the sequence @xmath118 is replaced by @xmath119 . by convention ,
the first entry is always @xmath9 . if a record is established at time @xmath13 and broken at time @xmath120 , then that record s lifetime is defined to be @xmath114 .
clearly , the binary string is much simpler than the original data sequence , but it contains enough information about record lifetimes for us to predict the distribution @xmath113 . the binary strings are easily visualized via a tree structure .
starting from a single vertex ( the `` ancestor '' ) @xmath9 on the first line ( @xmath121 ) , time runs downwards . at time @xmath122
( the second line ) , our string has two possible entries : @xmath9 or @xmath117 . to represent these , we draw two branches from the first line to the second : one to the right ( labeled @xmath9 ) , and one towards the left ( labeled @xmath117 ) .
each of these branches ends in a new vertex .
these branch again , giving us four vertices in total on the third ( @xmath123 ) line . continuing this procedure to the @xmath18th level
, we find @xmath124 vertices there . as an illustration ,
the @xmath125 tree is shown in fig .
clearly , all possible binary strings with 4 elements appear in this tree , each associated with exactly one vertex on the @xmath126 line .
the record lifetimes associated with a given string are now easily identified : following an @xmath117 branch means that the current record survives , while choosing the @xmath9 branch implies that a new record is set .
thus , each vertex can be labeled with the set of record lifetimes @xmath127 leading to it .
2b shows the `` tree of lifetimes '' for the @xmath125 case shown in fig . 2a .
note that the number of entries in these sets varies for different strings .
for example , the far right string in fig .
2a , where every record is broken at the next time step , gives rise to @xmath128 , while the far left string corresponds to @xmath129 : the record is set at @xmath121 and survives the next three time steps .
a simple recursion relation emerges . from a particular entry at time @xmath13 to the two `` daughters '' at time @xmath130 , the set @xmath131 branches into two : @xmath132 and @xmath133 .
moreover , because any distribution @xmath0 will generate the same set of binary strings , it is obvious that the associated lifetimes @xmath134 are completely _ independent _ of the underlying model ! to complete the construction of the histogram , we need to find the _ string _ _ probability _ , that is , the probability that a specific string will appear . at the @xmath18th level , each string is associated with a specific vertex , which is uniquely labeled by the set @xmath135 .
so , let us denote this probability by @xmath136 . at the top level ( @xmath121 ) , this probability is obviously trivial : @xmath137 .
for example , the year record keeping starts , any temperature will be a `` record '' in the second `` year '' ( @xmath122 ) , there are two vertices : @xmath138 and @xmath139 , corresponding to the first record surviving or being broken , respectively . in terms of the original data sequence @xmath140 , these possibilities are given by @xmath141 or @xmath142 .
let us focus first on the former case for which the probability , @xmath143 , is just @xmath144 . at first sight , this expression seems to depend on the model distribution @xmath0 .
however , recalling eq .
( [ rtoq ] ) , we can transform the integration variables to @xmath145 where we have exploited the monotonicity of @xmath146 to replace @xmath147 by @xmath148 .
not only is this integral trivial to compute , it is also _ manifestly independent _ of the underlying model ! in a similar manner , we can compute explicitly that the probability of the latter case , @xmath149 , is also 1/2 .
let us illustrate this process for the next year ( @xmath123 ) .
for the four possible histories , different combinations of @xmath150-functions appear .
explicitly , their associated probabilities are : @xmath151 in fig .
2c , we show all the probabilities in the @xmath125 tree .
note that the sum of all the @xmath152 s at each level is indeed unity ; the probability of having _ any _ history after @xmath18 years must be 1 . in appendix b
, we show that the general result for an arbitrary string of any length @xmath18 is @xmath153 note that the last factor is actually just @xmath18 . another somewhat counter - intuitive result concerns @xmath154 , the probability for the _
last record to survive _ @xmath114 steps ( regardless of what happened earlier ) .
note that the `` last record '' is also the `` highest record , '' since the last record is necessarily larger than all previous records !
now , to say that this record survives @xmath114 steps means that the values in the rest of the string ( @xmath155 of them ) are lower .
therefore , in a string of @xmath18 steps , the last record must have occurred at the @xmath156-th step .
thus , to find @xmath154 , we ask : what is the probability for the highest record to show up at the @xmath157-th step ?
we might guess that the highest record is unlikely to occur near the beginning of the string , because there are many chances for it to be broken later . on the other hand ,
if we use the language of athletics , we might conclude that the record facing the last athlete may be quite high ( given that `` many guys have gone before '' ) and breaking the record may not be easy .
so , perhaps @xmath154 should be peaked in the middle ?
the surprise is that @xmath154 is _ independent of _
@xmath114 ( or equivalently , @xmath157 ) ! in other words , the highestl record _ may occur at any step with equal probability _ !
the proof may be found in appendix c. lastly , we turn to the ( unnormalized ) lifetime distribution , @xmath113 .
we again llustrate by explicit calculations of the @xmath125 case and relegate the general discussion to appendix d. combining the string probabilities with the lifetimes , we obtain : @xmath158 focusing on a particular binary string , we recall that the associated record lifetimes @xmath159 as well as the string probability are entirely model - independent . as a result ,
the lifetime distribution itself is _ universal _ , that is , its form does not depend on the underlying model distribution @xmath0 .
moreover , the result for @xmath160 suggests a surprisingly simple form for the lifetime histogram , namely , @xmath161 the general proof can be found in appendix d. remarkably , @xmath162 is not only universal , but also is independent of @xmath18 , that is , the length of the original data sets !
so , the probability for a record to survive , for example , 5 time steps is the same , no matter how much data we accumulate . phrased differently , this result is not surprising at all : because @xmath113 is also , roughly speaking , the probability for the next huge disaster to strike after @xmath114 time steps , it would be very disturbing if that probability depended on the length of our data sets : that would imply that we could avoid or court disaster by just continuing to take data . clearly , such a result would be nonsensical .
nobody would believe that , by observing the weather , we can change it ! although the lack of an @xmath18-dependence can be understood in this way , the universality with respect to the underlying model @xmath0 is a much stronger statement : as long as _ all _ underlying data arise from the _ same _ distribution , the functional _
form _ of this distribution is completely irrelevant .
flat , exponential or gaussian @xmath72 s all generate identical lifetime histograms . in fig .
3 , we show monte carlo data for the lifetime histograms of flat and exponential @xmath0
. they are indistinguishable , apart from statistical fluctuations !
the agreement with the theoretically expected @xmath163 is again excellent .
we have presented a pedagogical introduction to the statistics of extremes . though we motivated the discussion by reference to weather records , the model studied here is much simplified . generating a sequence of @xmath18 random numbers , all selected from the _ same _ underlying distribution @xmath0 ,
we keep track of the ones which are , say , higher than all the preceding numbers `` record highs . '' studying the statistics of such strings , we identify several universal features , that is , properties which are independent of the underlying distribution @xmath72 .
in particular , we ask how long a record survives before being broken , that is , we focus on the lifetimes of records and compile a histogram .
not only is that histogram universal , but it is also exceedingly simple : records lasting @xmath114 steps occur with a decreasing frequency of @xmath164 .
we also inquire into some details of these strings . as far as records are concerned ,
the sequence of numbers can be reduced to a binary string of @xmath9 s and @xmath117 s . at each step , we only keep track of whether the record is broken @xmath165 or not @xmath166 .
a `` tree '' representing all possible sequences can be drawn which displays whether a record is broken or not at each step .
it is natural to ask , what is the probability that records will be broken at specific steps along the string ?
the answer , given by eq .
( [ p(r ) ] ) or ( [ p_n ] ) , is also universal .
perhaps most surprising is the answer to the question , what is the probability that the overall record will occur at the @xmath114th step ?
the answer is _ independent _ of @xmath114 .
the overall record occurs at any step with _
equal probability_. another interesting question is the following . averaged over many sequences , how does the overall record `` inch up '' with time ?
although the answer is not completely universal , the average behavior falls into one of several classes .
the simplest is due to a @xmath0 which is bounded .
then , not surprisingly , the average just runs into this bound . for unbounded @xmath72 s ,
the asymptotic behavior is mostly dictated by the `` tail '' of the distribution . in this sense
, there is a limited form of universality , that is , properties are independent of the details of the rest of @xmath0 . beyond the average record , we also studied the probability density , @xmath22 , for finding the record at @xmath9 after @xmath13 steps . like the central limit theorem
, there is universality for large values of @xmath13 , provided an appropriate time - dependent rescaling of @xmath167 is included .
however , there are _ three _ limiting distributions , as well as the possibility of _ no _ limiting distribution .
these fascinating properties lie outside the scope of this article .
we refer the interested reader to the book by galambos for example.@xcite looking beyond our simple model for `` weather '' records , we may inquire about a natural generalization a model for `` global warming . '' here , we let the underlying distribution _ drift upwards _ with time . the simplest is a uniform drift , that is , letting the model distribution at time step @xmath13 , @xmath168 , assume the form @xmath169 with @xmath170 .
although numerous results still hold , there are also significant differences . obviously , the average record here must increase linearly . probably the most significant difference between the simple model for `` weather '' records and
this `` global warming '' case is that the limiting distribution for @xmath22 is expected to assume the form @xmath171 , with no time - dependent rescaling .
we are not aware of any general results for such drifting @xmath72 s .
certainly , these expectations are confirmed for a _ flat , drifting _
for large times , we find @xmath172 , with q(r , t)=q^*(r - t ) , where q^*()=^k 1-k1-(k-1 ) .
to arrive at these results , we relied on a generalized version of ( [ q = q^t ] ) : @xmath173 , where @xmath174 .
clearly , these general forms are applicable for _ any _ time - dependent @xmath72 . for example
, we could consider a sequence composed of the _ sum _ of all random numbers generated before .
if the random numbers are distributed such that both positive and negative values are present , we may think of the sequence as the value of a stock , making gains and losses from day to day .
the variations are limited only by the imagination . unlike the simple model presented above
, there are very few known results for the distributions of the record lifetimes in general .
needless to say , exploring the universality classes will be a task both daunting and rewarding . to obtain the desired result ( [ avr - exp ] ) , we start from ( [ rtoq ] ) and the inverse of ( [ q - exp ] ) .
@xmath175 q^{t-1 } & = & t \
\left [ \sum_{n=1}^\infty q^n
/ n\,\right ] q^{t-1 } \nonumber \\ & = & \sum_{n=1}^\infty \frac { t}{n\left ( n+t\right)}=\sum_{n=1}^\infty \left [ \frac { 1}{n}-\frac { 1}{n+t}\right ] \label{avr - exp - ap } \\ & = & \sum_{k=1}^t\frac { 1}{k } \nonumber\end{aligned}\ ] ] those who like a bit more mathematical rigor may consider @xmath176 , which justifies the exchange of the integral and sum . we provide some details for computing the general case @xmath177 .
first , let us introduce an alternate way of labelling the vertices .
instead of the record lifetimes @xmath178 , we keep track of the times @xmath179 when records are broken : that is , @xmath180 denotes the position of the @xmath181th @xmath9 along the string .
so , we may also denote the general string probability by @xmath182 . by convention @xmath183 , while @xmath184 , @xmath185 , etc : @xmath186 where the last line comes from the length of the string : @xmath187 . to obtain @xmath50 for an arbitrary string , we start with p_n ( r_1,r_2, r_k ) = dq_1 dq_n .
to say that the first record @xmath188 associated with @xmath189 has a lifetime of @xmath190 means that the next @xmath191 @xmath9 s ( and their associated @xmath69 s ) are lower .
therefore , the first @xmath191 factors in the integrand are @xmath192 so that the integration over the variables @xmath193 can be performed trivially , with the result @xmath194 .
now , the next record is @xmath195 associated with @xmath196 , so that the next factor in the integrand must be @xmath197 .
there will be no more appearances of @xmath189 in the rest of the integrand .
so , the integral over @xmath189 ( up to @xmath196 ) can be performed , giving @xmath198 .
note that @xmath199 so that this result can be written simply as @xmath200 . in a similar way
, we integrate over the @xmath201 variables @xmath202 up to @xmath196 , arriving at @xmath203 or @xmath204 .
performing the integral over @xmath196 gives @xmath205 .
this process can be carried on until the last integration ( over @xmath206 , up to 1 ) .
the integrand here consists of the result from the integrals over the previous variables @xmath207 as well as the @xmath69 s from the rest of the sequence : @xmath208 . with this additional factor of @xmath209 ,
the last integral provides a factor of @xmath18 .
the final result is @xmath210 rewriting the @xmath211 s in terms of the @xmath212 s , we have p_n ( \ { } ) = , where the last factor is precisely @xmath18 , because the sum of the lifetimes of all records is just the length of the string .
one interesting property of these @xmath50 s follows from the fact that , given a particular string of length @xmath18 , we can `` generate '' two strings of length @xmath213 , by concatenating either an @xmath9 or an @xmath117 at the end . in the former case , @xmath214 stays the same while @xmath215 . in the latter case , the value of @xmath214 increases by 1 . from eq .
( [ p_n ] ) , we obtain the following `` recursion '' relations : @xmath216 not surprisingly , @xmath217 to illustrate , consider the first pair of entries at the 4@xmath218 level in fig .
2 and their relationship to the first entry at the 3@xmath219 level ( @xmath220 ) . from fig 2c , we read off @xmath221 , @xmath222 , and @xmath223 .
these quantities satisfy eqs .
( [ recur1],[recur2 ] ) . [ [ probability - for - strings - where - the - last - record - survived - m - steps ] ] probability for strings where the last record survived @xmath114 steps ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ an easy way to arrive at the lifetime histogram @xmath163 is through the probability for the last record to survive @xmath114 steps , regardless of what happened earlier .
let us define this quantity as @xmath224 .
to be precise , it is @xmath225 where @xmath48 is the kronecker delta ( @xmath48 is unity if its argument vanishes and zero otherwise ) .
be careful with this sum notation : it stands for summing over @xmath226 as well , because @xmath226 is the _ number of records _ in the string .
now , because the `` last '' record is also the overall record , ( that is , all other @xmath9 s are lower ) , it is easy to compute @xmath227 . because we are summing over all possible ways that the records before this step ( @xmath228 ) are broken ,
we may write @xmath229 as remarked in the main text , this result is not intuitively obvious at all . intuition might lead us to an @xmath227 with a maximum in the middle of the string , but eq .
( [ s=1/n ] ) tells us that the distribution is entirely flat .
note that @xmath113 is not a real probability in the sense that @xmath230 .
the normalization condition here is _ \ { } p_n ( \ { } ) = 1 , but the histogram is defined by t_n(m)=_\ { } p_n ( \ { } ) _ i=1^k ( _ i - m ) .
again , we caution the reader that the @xmath231 here involves a sum over @xmath226 as well .
summing over @xmath114 produces @xmath232 .
now , given a string of @xmath18 random numbers , the only way for a record lifetime to be @xmath18 is that the first record survives .
thus , @xmath233 is precisely @xmath234 , so that @xmath235 our goal reduces to proving t_n+1(m)=t_n(m)1m <
there are two types of contributions to @xmath113 .
one is from all the `` interior '' records : @xmath236 that is , @xmath237 with @xmath238 .
the other piece , from the last record alone , is precisely @xmath227 . for @xmath239 ,
let us start with @xmath240 and go from @xmath18 to @xmath213 , by looking at @xmath241 where the @xmath226 s appearing here are still those associated with @xmath240 ( that is , the last record may be labeled by @xmath242 ) . because the sum over @xmath214 and @xmath243 can be performed without the @xmath244 s , eqs .
( [ recur1 ] ) and ( [ recur2 ] ) can be used to obtain the following recursion relation _ \ { } p_n+1 ( \ { } ) _ i=1^k-1 ( _ i - m ) = _ \ { } p_n ( \ { } ) _ i=1^k-1 ( _ i - m ) . but
the quantity ( [ spd ] ) is not the only contribution to @xmath239 , because concatenating an @xmath9 to an @xmath18-string will produce an `` interior '' record for the @xmath245-string .
focusing on a specific @xmath114 , this extra bit is just _
\ { } p_n+1 ( _ 1,_2, ,_k,_k+1=1 ) ( _ k - m ) , which is _ \ { } p_n ( _ 1,_2, ,_k ) ( _ k - m ) = s_n(m ) .
so , we conclude t_n+1(m)=t_n(m)+ . with the help of this equation
, we have @xmath246 but , according to eq .
( [ s=1/n ] ) , s_n+1(m)=s_n(m ) so that t_n+1(m)=t_n(m)+s_n(m)=t_n(m ) .
thus , using ( [ tnn ] ) , we arrive at @xmath247 for any @xmath248 .
see for example , e. j. gumbel , _ the statistics of extremes _ ( columbia university press , new york 1958 ) ; j. galambos , _ the asymptotic theory of extreme order statistics _
( wiley , new york 1978 ) .
gumbel gives a short historical summary in his book . | we present a simple , pedagogical introduction to the statistics of extreme values .
motivated by a string of record high temperatures in december 1998 , we consider the distribution , averages and lifetimes for a simplified model of such `` records .
'' our `` data '' are sequences of independent random numbers all of which are generated from the same probability distribution .
a remarkable universality emerges : a number of results , including the lifetime histogram , are universal , that is , independent of the underlying distribution . |
data from the intermediate energy region of e@xmath5e@xmath6 annihilation , as provided by the jade experiment at petra , allow significant tests of perturbative and non - perturbative aspects of quantum chromodynamics ( qcd ) .
this was stressed in a recently published analysis of jade data @xcite where determinations of the strong coupling constant @xmath0 at @xmath1 = 22 to 44 gev were presented , which are the most precise ones at petra energies . since the analysis is based on event shape observables for which resummed qcd predictions @xcite ( nlla )
were developed long after the petra shutdown , the theoretical uncertainties of previous measurements @xcite could be reduced significantly . by applying lep - established techniques of estimating experimental and theoretical systematics ,
the running of @xmath0 was tested _
consistently _ within a large range of energy scales @xcite . a reliable estimation of the higher order uncertainties of the predictions , individually for each observable , is supported by considering as many observables as possible .
hence , an update of determinations of @xmath0 at petra energies @xcite is performed , now additionally considering new calculations for the @xmath2 parameter @xcite .
furthermore , improved calculations for the jet broadening variables @xmath3 and @xmath4 @xcite are applied .
jade data also provide clearer evidence for non - perturbative effects @xcite , which typically contribute with reciprocal powers of @xmath1 .
they are accounted for by so - called _ power corrections _ which , on general grounds , are found from several theoretical approaches such as renormalons @xcite or the dispersive approach @xcite . in the following , new predictions for power corrections to the differential distributions are tested using jade data and also lep data up to @xmath1 = 183 gev .
the jade detector , one of the experiments at the petra e@xmath5e@xmath6 collider , was designed as a hybrid 4@xmath8-detector to measure charged and neutral particles .
it was operated from 1979 until 1986 at centre - of - mass energies of @xmath1 = 12.0 to 46.7 gev .
this analysis focuses on 20926 events at 35 gev and 6158 events at 44 gev passing the standard jade multihadron selection @xcite .
the @xmath0 analysis is based on the event shape distributions of the thrust @xmath9 , heavy jet mass @xmath10 , the differential 2-jet rate using the durham jet finder , @xmath11 , the total and wide jet broadening @xmath3 and @xmath4 and the @xmath2 parameter ( cf .
@xcite ) . for the latter , no experimental results
have previously been presented at petra energies .
the data are corrected for detector effects and initial state photon radiation using a parton shower plus string fragmention model ( jetset , cf .
@xcite ) and a full detector simulation . as pointed out in @xcite ,
there is good agreement of the monte carlo model with the data , despite the larger hadronization effects at lower energies .
the strong coupling constant @xmath0 is determined by @xmath12 fits of combined nlla @xcite and @xmath13(@xmath14 ) @xcite calculations to the event shape distributions after unfolding for hadronization @xcite .
the previously published measurements of @xmath0 at 35 and 44 gev @xcite are complemented by using new resummed calculations for the @xmath2 parameter @xcite and by applying improved resummed calculations for @xmath3 and @xmath4 @xcite . in the latter case , all single logarithmic terms of the prediction
are now handled more accurately by a proper treatment of the quark recoil against multi - gluon emission . in the case of the @xmath2 parameter
, the combined resummed plus fixed order theory gives an excellent description of the data over the entire range of the distribution , even at the low petra energies @xcite .
the recoil - corrected calculations on the jet broadenings do not yield a significant improvement of the previous fits @xcite .
the excess of the theoretical curve over the data in the 3-jet region is still present , but the @xmath0 results are now shifted by about 3% to higher values .
the values of @xmath0 and the errors obtained at @xmath1 = 35 gev are shown in fig . [ fig - as ] , indicating that the results are consistent with each other . the combination of the individual results of the six observables using the weighted average method @xcite yields as final results @xmath15 various uncertainties emerging from experimental measurement , event selection , hadronization corrections and the choice of the qcd - renormalization scale were regarded as systematic errors of the fits @xcite .
a technique for studying power suppressed contributions to event shapes is the dispersive approach @xcite .
a recent publication @xcite points out that the effect of power corrections on the event shapes mentioned above is simply a _ shift _
@xmath16 of the perturbative spectrum by an amount proportional to @xmath17 and @xmath18 , where @xmath19 = @xmath1 .
the shift depends on two parameters @xmath20 and @xmath21 and an observable - dependent coefficient @xmath22 : @xmath23 @xmath20 and @xmath21 can be expressed in terms of effective coupling moments @xmath24 and @xmath25 @xcite,@xmath26 \\ { \cal p}^{\prime}= a \frac{\mu_i}{q } \hspace*{-1 mm } \left [ \alpha_0^{\prime}(\mu_i)+\alpha_s(q ) + \beta_0\frac{\alpha_s^2}{2\pi } \left(\hspace*{-.5 mm } \ln \frac{q}{\mu_i}+\frac{k}{\beta_0}+2 \hspace*{-.5 mm } \right ) \hspace*{-1 mm } \right]\end{aligned}\]]where @xmath27 contains the so - called milan factor @xmath28 arising from a two - loop calculation of the observable - dependent factor .
the milan factor is expected to be universal @xcite .
@xmath29 is an infrared matching scale @xcite , usually set to 2 gev .
an important feature of the dispersive approach is that the only free parameters @xmath24 and @xmath25 are expected to be approximately _ universal _ @xcite .
the universality of the power corrections is tested using differential data distributions at hadron level measured with jade at @xmath1 = 35 and 44 gev @xcite , and also using data from various lep experiments up to @xmath1 = 183 gev @xcite . both perturbative qcd ( @xmath30+nlla ) and power corrections
are compared by a global simultaneous fit of @xmath0 , @xmath24 and , in the case of the jet broadenings , additionally @xmath25 to the data .
[ fig - global_fits ] shows the results of the fits ( solid lines ) to the data ( points ) . unlike in a similar analysis in @xcite
, each energy has its individual fit range including the maximum peaks in the 2-jet region .
the fitted distributions are in good agreement with the data within the fit ranges . in the case of the jet broadenings , a strong correlation between @xmath24 and @xmath25
is observed , presumably because the pure 1/@xmath19 term @xmath31 containing @xmath25 was found to be small and fairly energy independent above @xmath1 = 35 gev .
a reanalysis of more data at lower energies is needed to resolve both @xmath24 and @xmath25 . for now , this contribution and , hence , the dependence on @xmath25 was neglected .
the corresponding fit results for @xmath0(@xmath32 ) and @xmath24(2 gev ) are presented in table [ tab - fit - results ] , and in fig .
[ fig - con ] with 95% confidence level contours . in the case of @xmath2 parameter and
for comparison thrust @xmath9 , the results for @xmath0 are compatible with the world average value of 0.118 @xcite . furthermore , there is a fair agreement of the corresponding @xmath24 values .
however , @xmath0 from the fits to the jet broadenings yields @xmath0 = 0.0954 @xmath33 0.0017 , @xmath24 = 0.710 @xmath33 0.081 , @xmath25 = -0.707 @xmath33 0.166 and a correlation between @xmath24 and @xmath25 of + 98 % . ]
is far off the world average . here , the universality of the effective coupling moment is not observed .
the presence of possible 1/@xmath34 contributions can be tested by considering lep data and jade data separately in the analysis .
[ fig - con_low+high ] demonstrates that , in the case of the @xmath2 parameter , the jade data reduce significantly the statistical uncertainties of the fit . in the case of @xmath3 ,
a minor indication for 1/@xmath34 terms emerges .
.[tab - fit - results]preliminary results for @xmath0 and @xmath24 .
the errors stated are the statistical errors of the fit . [ cols="^,>,<,>,<,^",options="header " , ] a comparison of the fits based on distributions corrected for hadronization as described in section [ sec - as_update ] and the hadron level fits using the power correction ansatz points out that the transformation from the perturbative to the non - perturbative level is not sufficiently described by a simple shift . clearly , an additional _ squeeze _ of the distributions is needed , i.e. power corrections additionally depending on the value of the observable . in the present fits
, @xmath0 appears as the only parameter inducing a squeeze .
thus , a combination of rather large values for @xmath24 and extremely small values for @xmath0 is needed to fit the hadron level distributions .
this mechanism explains the observed anticorrelation between @xmath0 and @xmath24 as seen in fig.[fig - con ] .
however , a shift is sufficient to describe the impact of hadronization on the _ means _ of the event shape distributions @xcite .
combined resummed and fixed order qcd provides consistent determinations of @xmath0 at petra energies .
recently presented calculations for the @xmath2 parameter , @xmath3 and @xmath4 fit previously published data at @xmath1 = 35 and 44 gev , yielding @xmath35 = 0.123 @xmath36 and 0.123 @xmath37 , respectively , in good agreement with the world average value @xcite .
power corrections to differential event shape distributions are tested using hadron level data between @xmath1 = 35 and 183 gev .
reliable fits are obtained in the case of the @xmath2 parameter and the thrust , where non - perturbative effects are approximately described by the predicted shift of the perturbative spectra .
this does not apply to the case of @xmath3 and @xmath4 , yielding @xmath0 results incompatible with other measurements at the @xmath38 pole . for the present calculations , the universality of the effective coupling moment @xmath24 at the level of 20% is only verified for @xmath2 parameter and thrust
. 9 p.a .
movilla fernndez et al . : eur.phys.j . *
c1 * ( 1998 ) 461 .
s. catani et al . :
* b295 * ( 1992 ) 269 ; s. catani et al .
: nucl.phys . *
b407 * ( 1993 ) 3 ; g. dissertori , m. schmelling : phys.lett . * b361 * ( 1995 ) 167 .
s. bethke : nucl.phys.proc.suppl . 54a ( 1997 ) 314 .
movilla fernndez et al . : pitha 98/21 , hep - ex/9807007 , ichep98 # 646 , vancouver s. catani et al .
: phys.lett . *
b427 * ( 1998 ) 377 .
dokshitzer et al .
: jhep 01 ( 1998 ) 011
. o. biebel : nucl.phys.pr.suppl .
64 ( 1998 ) 22 .
mueller in : p.m. zerwas and h.a .
kastrup ( eds . ) : _ qcd 20 years later _ ,
1 , world scientific , singapore , 1993 . yu.l .
dokshitzer , b.r .
webber : phys.lett .
* b352 * ( 1995 ) 451 .
ellis , d.a .
ross , a.e .
terrano : nucl . phys .
* b178 * ( 1981 ) 421 .
dokshitzer et al . :
jhep 05 ( 1998 ) 003 .
dokshitzer et al .
: nucl.phys . *
b511 * ( 1998 ) 396 . : z.phys . * c59 * ( 1993 ) 1 ; * c72 * ( 1996 ) 191 ; * c75 * ( 1997 ) 193 ; eps - hep97 # 177 , jerusalem ( 1997 ) ; opal - pn 281 ( 1997 ) ; p. pfeifenschneider , theese proceedings .
_ aleph coll .
_ : eps - hep97 # 629 , jerusalem ( 1997 ) .
_ delphi coll .
_ : eps - hep97 # 544 , jerusalem ( 1997 ) .
_ l3 coll .
_ : phys.lett . *
b404 * ( 1997 ) 390 ; * b371 * ( 1996 ) 137 .
dokshitzer , b.r .
webber : phys.lett .
* b404 * ( 1997 ) 321 . | studies of event shape observables at petra and lep energies are presented .
previously published determinations of the strong coupling constant @xmath0 at @xmath1 = 35 and 44 gev are complemented using new resummed qcd calculations for the @xmath2 parameter and improved calculations for the jet broadening variables @xmath3 and @xmath4 which recently became available .
furthermore , recently predicted power corrections to the differential distributions of these observables are investigated . in this study ,
e@xmath5e@xmath6 data between @xmath1 = 35 and 183 gev are considered .
= cmss24 at 20truept = cmss14 at 14truept = cmss10 at 12truept = cmss10 at 10truept = cmss8 at 8truept to 245 mm to * event shapes from jade data + and + studies of power corrections * pedro a. movilla fernndez + iii .
institut , technische hochschule aachen + d-52056 aachen , germany |
10 k. aamodt et al .
( alice collaboration ) , j. instrum . * 3 * , s08002 ( 2008 ) k. nakamura et al .
( particle data group ) , j. phys .
* g37 * , 075021 ( 2010 ) k. oyama et al .
( alice collaboration ) , proc . of the workshop
`` lhc lumi days '' , to be published .
lhcb collaboration , cern - lhcb - conf-2010 - 014 a. pulvirenti et al .
( alice collaboration ) , these proceedings . | low mass vector meson ( @xmath7 ) production provides key information on the hot and dense state of strongly interacting matter produced in high - energy heavy ion collisions . among them , strangeness enhancement can be accessed through the measurement of @xmath0 meson production , while the measurement of the @xmath8 spectral function can be used to reveal in - medium modifications of hadron properties close to the qcd phase boundary .
vector meson production in pp collisions provides a reference for these studies .
moreover , it is interesting by itself , since it can be used to tune particle production models in the unexplored lhc energy range .
the alice experiment at the lhc can access vector mesons produced at forward rapidity through their decays in muon pairs , and at central rapidity in the di - electron decay channel .
the detector is fully described in @xcite . in this paper ,
results from the analysis of the data collected during the 2010 pp run at @xmath2 tev are reported .
the measurement in the dimuon channel was performed using the forward muon spectrometer , that consists of an absorber acting as muon filter , a set of cathod pad chambers ( five stations , each one composed of two chambers ) for the track reconstruction in a dipole field , two stations of two resistive plate chambers for the muon trigger , two absorbers and an iron wall acting as a muon filter . the data sample used for the analysis in the dimuon channel amounts to an integrated luminosity of approximately 85 nb@xmath9 .
since only a fraction of the data contained the full information relevant for the extraction of the integrated luminosity , a subsample corresponding to @xmath10 was used for the measurement of the @xmath0 cross section , while the full sample was used to extract the @xmath11 distribution .
muon pairs were selected asking that each muon track reconstructed in the tracking chambers matches the corresponding tracklet in the trigger stations in the position in the ( x - y ) plane and in the slope in the ( r - z ) plane .
a cut on the muon rapidity @xmath12 was applied in order to remove the tracks close to the acceptance borders .
about 291,000 opposite sign @xmath13 , 197,000 like - sign @xmath14 muon pairs survived these selections .
the combinatorial background was evaluated using the event mixing technique , and normalized to @xmath15 , where @xmath16 , and @xmath17 is the acceptance for a ( @xmath18 ) pair .
the event mixing was checked by comparing the results obtained for like - sign mixed pairs with the real ones .
the shapes of the background calculated with the two methods are identical , while the amount of like - sign pairs estimated with the event mixing differs from the one in the real data by @xmath19 .
we take this value as the uncertainty on the background normalization .
the signal - to - background ratio for @xmath4 gev/@xmath20 is about 1 at the @xmath0 and @xmath1 masses .
alternatively , the combinatorial background contribution to the opposite sign mass spectrum for a given @xmath21 mass bin can be evaluated from the like sign mass spectra using the formula : @xmath22 the two techniques are in good agreement for @xmath23 gev/@xmath20 , while for lower pair transverse momenta both methods fail in describing the background .
the analysis is thus limited to @xmath4 gev@xmath5 . after subtracting the combinatorial background from the opposite sign mass spectrum ,
we obtain the signal mass spectrum shown in fig .
[ fig : phi_all ] ( left ) .
the invariant mass spectrum is fitted with the contributions given by the light meson decays into muons and open charm / beauty contributions .
the free parameters of the fit are the normalizations of the @xmath24 , @xmath25 , @xmath26 and open charm signals .
the other processes ( @xmath27 , @xmath28 , @xmath29 , @xmath30 and open beauty ) are fixed according to the relative branching ratios or cross sections .
the main sources of systematic uncertainty are due to the uncertainty in the background normalization and on the relative normalization of the sources , mainly due to the error on the branching ratios for the @xmath1 and @xmath31 dalitz decays .
the raw number of @xmath0 and @xmath32 resonances obtained from the fit is @xmath33 and @xmath34 .
the @xmath0 production cross section was evaluated in the range @xmath6 , @xmath35 gev/@xmath20 through the formula @xmath36 , where @xmath37 is the measured number of @xmath0 mesons corrected for the efficiency and the acceptance , @xmath38 is the branching ratio in lepton pairs , obtained as a weighted average of the branching ratios in @xmath39 and @xmath40 pairs @xcite , @xmath41 is the number of minimum bias collisions , @xmath42 is the alice minimum bias cross section in pp collisions at @xmath2 tev and @xmath43 is the ratio between the number of single muons in the region @xmath44 , @xmath45 gev/@xmath20 collected with the minimum bias trigger and with the muon trigger .
the minimum bias cross section was measured in a van der meer scan @xcite .
its value is @xmath46 .
the number of minimum bias collisions was corrected , run by run , for the probability of having multiple interactions in a single bunch crossing .
the ratio @xmath43 strongly depends on the data taking conditions and was evaluated run by run .
we obtain @xmath47 .
the systematic error comes from the uncertainty on the background subtraction ( @xmath48 ) , the muon trigger efficiency ( @xmath49 ) , the tracking efficiency ( @xmath50 ) the uncertainty on the @xmath0 branching ratio into dileptons ( @xmath51 ) , on the minimum bias cross section ( @xmath52 ) and on the ratio @xmath43 ( @xmath50 ) . the @xmath3-differential cross section @xmath53 is shown in fig .
[ fig : phi_all ] ( triangles ) . point to point uncorrelated systematic uncertainties are indicated as red boxes .
the fully correlated systematic uncertainty , represented by a blue box on the left side of the plot , amounts to @xmath54 .
a fit to the differential cross section with a power law function , @xmath55^n$ ] , gives @xmath56 gev/@xmath20 and @xmath57 .
the measurements in kaon pairs performed by lhcb in a similar rapidity range ( @xmath58 , open circles ) @xcite and by alice at midrapidity ( @xmath59 , full circles ) @xcite are also plotted , showing that the shapes are similar .
the rescaling of the lhcb cross section to @xmath4 gev@xmath5 and to @xmath60 leads to @xmath61 mb .
there is a @xmath62 difference between the alice and lhcb measurements .
considering the alice statistical error and the part of the systematic uncertainty which are certainly not correlated among the two experiments , the two measurements are in agreement .
the ratio @xmath63 was measured as a function of the transverse momentum , showing a flat trend with an average value of @xmath64 . in order to extract the @xmath1 cross section ,
the @xmath8 and @xmath1 contributions must be disentangled , leaving the @xmath8 normalization as an additional free parameter in the fit to the dimuon mass spectrum .
the result of the fit gives @xmath65 .
the systematic uncertainty was evaluated changing the normalizations of the @xmath66 and @xmath67 according to the uncertainties in their branching ratios , and the background level by @xmath68 , twice the uncertainty in the normalization . from these results ,
it was possible to extract the ratio @xmath69 . this ratio is plotted as a function of @xmath3 in fig .
[ fig : dsigmaomegadydpt ] ( left ) .
the @xmath1 production cross section , calculated from this ratio , is @xmath70 . in fig .
[ fig : dsigmaomegadydpt ] ( right ) the @xmath1 differential cross section is shown .
data are fitted with the power law function , obtaining @xmath71 gev/@xmath20 and @xmath72 . in conclusion ,
the @xmath0 and @xmath1 @xmath3 differential cross sections were measured in pp collisions at @xmath2 tev .
the ratio between the @xmath0 and the @xmath1 cross sections is flat as a function of @xmath3 . in pb - pb collisions work
is in progress to measure @xmath0 production as a function of centrality . |
let @xmath0 be an open finite riemann surface endowed with the poincar ( hyperbolic ) metric .
we will study some properties of holomorphic functions in the riemann surface with uniform growth control . namely we will deal with the banach space @xmath1 of holomorphic functions in @xmath0 such that @xmath2 where @xmath3 is a given subharmonic function that controls the growth of the functions in the space .
the fact that @xmath3 is subharmonic is a natural assumption on the weight that limits the growth since any other growth control given by a weight @xmath4 , @xmath5 can be replaced by an equivalent subharmonic function because @xmath6 is a subharmonic function and @xmath7 with equality of norms , latexmath:[$\sup_s |f|e^{-\psi}= \sup_s we have fixed a metric .
it is then natural to restrict the possible weights @xmath3 , in a way that the functions in @xmath9 oscillate in a controlled way when the points are nearby in the poincar metric .
this is achieved for instance by assuming that @xmath3 has bounded laplacian ( the laplace - beltrami operator with respect to the hyperbolic measure ) .
that is , if in a local coordinate chart the poincar metric is of the form @xmath10 , then we assume that @xmath11 satisfies @xmath12 .
if we want to deal with other weights then it is possible to introduce a natural metric associated to the weight as it is done in the plane in @xcite . in this work
we will only consider the poincar metric and bounded laplacian since it already covers many interesting cases and it is technically simpler .
the problems that we will consider are the following : 1 . the description of the interpolating sequences for @xmath1 : i.e. the sequences @xmath13 such that it is always possible to find an @xmath14 such that @xmath15 for all @xmath16 whenever the data @xmath17 , satisfies the compatibility condition @xmath18 2 . the description of sampling sets for @xmath1 : i.e. the sets @xmath19 such that there is a constant @xmath20 that satisfies @xmath21 in the solution of these problems the poincar distance and the potential theory in the surface play a key role .
this has already been observed by a. schuster and d. varolin in @xcite , where they provide sufficient conditions for a sequence to be interpolating / sampling for functions in a slightly different context where the weighted uniform control of the growth of the functions is replaced by a weighted @xmath22 control .
their condition basically coincides with the description that we reach so our work can be considered as the counterpart of their theorems , although we will give a different proof of their results as well .
we will rely on the well - known case of the disk and some simplifying properties of finite riemann surfaces .
their method of proof looks more promising if one wants to extend the result to riemann surfaces with more complicated topology .
when the surface is a disk , which will be our model situation , the corresponding problems have been solved in @xcite , @xcite and in a different way in @xcite .
of course , the more basic problem of describing the interpolating sequences for bounded holomorphic functions in finite riemann surfaces ( in our notation @xmath23 ) , has been known for a long time , see @xcite ) .
we introduce now some definitions that will be needed to state our results . for any point @xmath24 and any @xmath25
we denote by @xmath26 the domain in the surface @xmath0 that consits of points at hyperbolic distance from @xmath27 less than @xmath28 .
they are topological disks if the center @xmath27 is outside a big compact of @xmath0 , or if @xmath28 is small enough , as we will see in section [ rs ] .
a sequence @xmath29 of points in @xmath0 is hyperbolically separated if there is an @xmath30 such that the domains @xmath31 are pairwise disjoint .
let @xmath32 be the green function associated to the surface @xmath26 with pole at the `` center '' @xmath27 and @xmath33 be the green function associated to the surface @xmath0 .
we define the densities @xmath34 the main result is [ main ] let @xmath0 be a finite riemann surface and let @xmath3 be a subharmonic function with bounded laplacian . 1 . a sequence @xmath13 is an interpolating sequence for @xmath1 if and only if it is hyperbolically separated and @xmath35 .
2 . a set @xmath19 is a sampling set for @xmath1 if and only if it contains an hyperbolically separated sequence @xmath36 such that @xmath37 . in section [ rs ]
we will prove some key properties of finite riemann surfaces .
in particular we need to study the behavior of the hyperbolic metric as we approach the boundary of the surface .
we will also prove some weighted uniform estimates for the inhomogeneous cauchy - riemann equation in the surface , theorem [ dbar ] , that has an interest by itself . in the next section
, we use the tools and lemmas proved in section [ rs ] to reduce the interpolating and sampling problem in @xmath0 to a problem near the boundary that can be reduced to the known case of the disk . finally in section [ lebesgue ]
we show how our results can be extended to other banach spaces of holomorphic functions where the uniform growth is replaced by weighted @xmath38 spaces . a final word on notation .
by @xmath39 we mean that there is a constant @xmath40 independent of the relevant variables such that @xmath41 and by @xmath42 we mean that @xmath39 and @xmath43 .
we start by the definition and then we collect some properties of @xmath0 that follow from the restrictions that we are assuming on the topology of @xmath0 . a finite riemann surface is the interior of a smooth bordered compact riemann surface .
our surface is an open riemann surface and it is in fact an open subset of a compact surface ( the double , see @xcite ) . see figure [ fsurface ] for a typical representation .
observe that the genus is finite and the border of the surface consists of a finite number of smooth closed jordan curves . in most of what follows
the particular case of a smooth finitely connected open set in @xmath44 has all the difficulties of the general case .
the following claim follows from instance from ( * ? ? ?
* prop 7.1 - 7.4 ) [ dbaracotat ] for any @xmath45-form @xmath46 there is a solution @xmath47 to the inhomogeneous cauchy - riemann equation @xmath48 .
moreover since @xmath0 has an essential extension to a compact riemann surface if the data is a smooth form with compact support @xmath49 in @xmath0 then there is a bounded linear solution @xmath50 $ ] with the bound @xmath51 . in this statement and in the following @xmath52
is the poincar length of the @xmath45-form @xmath46 . in the disk we have blaschke factors that are very convenient to divide out zeros of holomorphic functions without changing essentially the norm .
the analogous functions that provide us with the same property in the case of finite riemann surfaces are given by the next proposition : [ blashcke ] there is a constant @xmath53 such that for any point @xmath24 there is a function @xmath54 with @xmath55 in particular @xmath56 is a bounded holomorphic function that vanishes only on the point @xmath27 and for any @xmath30 @xmath57 if @xmath58 .
the obstruction for an harmonic function @xmath47 to have an harmonic conjugate is that for a set of generators @xmath59 of the homology we have @xmath60 , @xmath61 .
if we want @xmath62 for an @xmath63 , we just need that @xmath64 .
being a finite riemann surface there are @xmath65 functions in the algebra of @xmath0 without zeros such that @xmath66 , see ( * ? ? ?
* lemma 1 ) .
thus the function @xmath67 is the logarithm of an holomorphic function @xmath68 .
therefore there is a constant @xmath40 such that any harmonic function @xmath47 in @xmath0 admits an holomorphic function @xmath69 with @xmath70 . take a point @xmath24 and any holomorphic function @xmath71 that vanishes only on @xmath27 .
then @xmath72 is harmonic in @xmath0 and therefore there is a holomorphic function @xmath73 such that @xmath74 .
thus we may define @xmath75 and it has the estimate @xmath76 .
the estimate @xmath77 when @xmath58 holds in finite riemann surfaces , see for instance ( * ? ? ? * theorem 5.5 ) .
the open ends of the riemann surface can be parametrized as follows : the border of the riemann surface @xmath0 is a finite union of smooth closed curves @xmath78 , @xmath79 . near each @xmath78
there is a closed geodesic @xmath80 that is homotopic to @xmath78 .
the subdomain of @xmath0 bounded by @xmath80 and @xmath81 is denoted a `` funnel '' following the terminology of @xcite and @xcite .
we need to be more precise about the hyperbolic metric in the funnel .
there are nice coordinates in the funnel that provide good estimates .
these are given by the collar theorem .
let @xmath82 be the universal holomorphic cover of @xmath0 and let @xmath83 be the deck transformation corresponding to the closed loop @xmath84 . consider the surface @xmath85 .
this an annulus since @xmath86 . if we quotient it by the rest of the deck transformations of the universal cover we get an holomorphic covering map @xmath87 from @xmath88 which is a local isometry ( in @xmath89 and @xmath0 we consider the poincar metric inherited from @xmath82 ) . in fact @xmath90 , where @xmath91 , and @xmath92 maps the unit circle isometrically to @xmath84 .
moreover @xmath92 is an isometric injection of the outer part of the annulus @xmath93 onto the funnel .
these will be called the standard coordinates of the funnel .
see @xcite and @xcite for details .
the poincare metric in the the funnel is explicit in the standard coordinates and it is comparable to the hyperbolic metric on the disk in the coordinate disk @xmath94 when restricted to @xmath95 .
we denote by @xmath96 , @xmath79 the funnels of @xmath0 bounded by @xmath80 and @xmath78 .
[ funnel ] we want to solve the inhomogeneous cauchy - riemann equation on @xmath0 with weighted uniform estimates . in order to get good estimates
it is useful to find functions @xmath97 with precise size control , i.e. , @xmath98 outside a neighborhood of the zero set of @xmath69 . with this function
we can later modify an integral formula to get a bounded solution to the @xmath99-equation when the data has compact support .
the following lemma provides such a function that in other context has been termed a `` multiplier '' : [ multiplier ] let @xmath0 be a finite riemann surface and let @xmath3 be a subharmonic function with bounded laplacian .
then there is a function @xmath69 with hyperbolically separated zero set @xmath100 such that @xmath101 whenever @xmath102 . moreover if we fix any compact @xmath49 in @xmath0 it is possible to find @xmath69 with the above properties and without zeros in @xmath49 . in any of the funnels
@xmath96 we transfer the subharmonic weight @xmath3 to the standard coordinate chart @xmath103 .
we define a weight @xmath104 on the disk @xmath105 in such a way that @xmath104 has bounded invariant laplacian and moreover @xmath106 on the region @xmath103 .
one way to do so is the following : we assume from the very beginning that @xmath3 is smooth ( this is no restriction since otherwise it can be approximated by a smooth function ) .
define @xmath107 where @xmath108 is a cutoff function such that @xmath109 in @xmath110 , @xmath111 in @xmath112 and @xmath113 is taken big enough such that @xmath104 is subharmonic and the invariant laplacian of @xmath104 is bounded above and below .
we are under the hypothesis of the result from @xcite that states that there is an holomorphic function in the disk @xmath114 with separated zero set @xmath115 ( in the hyperbolic metric of the disk ) such that @xmath116 whenever @xmath117 .
since the hyperbolic metric of the disk is comparable to the hyperbolic metric in the funnel , we have found a function @xmath118 with separated zero set such that @xmath119 if @xmath117 .
moreover dividing out @xmath114 by a finite blaschke product we can assume that @xmath114 is zero free in any prefixed compact of the disk .
we consider the `` core '' of @xmath0 to be @xmath120 , where @xmath121 are the outer part of the funnels mapped by @xmath122 .
the values of the @xmath123 are taken so big as to make sure that the compact @xmath49 in the hypothesis of the lemma is contained in the core of @xmath0 .
we adjust the @xmath114 @xmath79 as mentioned before to make sure that they are zero free in the inner part of the funnels @xmath124 .
we finally define @xmath125 in the core of @xmath0 . to patch the different @xmath114 together
we will need to solve a cousin ii problem with bounds .
our data is @xmath114 defined on the inner parts of the funnels mapped by @xmath124 .
the data are bounded above and below in the inner parts of the funnels ( because @xmath3 is bounded above and below in any compact of @xmath0 and @xmath114 have no zeros there ) .
we want to find functions @xmath126 and @xmath127 holomorphic on the core of @xmath0 such that @xmath128 in the inner part of the funnel .
if moreover @xmath129 and @xmath127 are bounded ( above and below ) then the function @xmath69 defined as @xmath130 in each of the funnels @xmath96 and @xmath127 on the core of @xmath0 is holomorphic on @xmath0 and has the desired growth properties . to find the functions @xmath129 observe that since the intersection of the funnel @xmath96 with the core of @xmath0 strictly separates the outer part of the funnel from the inner part of the core we can reduce the cousin ii problem to solving a @xmath99-equation with bounded estimates of the solution on @xmath0 when the data is bounded and with compact support ( the support is in the inner part of the funnels )
this can be achieved by lemma [ dbaracotat ] . with this function
we can then obtain the following result which is interesting by itself : [ dbar ] let @xmath0 be a finite riemann surface and let @xmath3 be a subharmonic function with a bounded laplacian .
there is a constant @xmath20 such that for any @xmath45-form @xmath46 on @xmath0 there is a solution @xmath47 to the inhomogeneous cauchy - riemann equation @xmath131 in @xmath0 with the estimate @xmath132 whenever the right hand is finite .
recall that the notation @xmath133 means the hyperbolic norm of @xmath46 at the point @xmath27 .
let @xmath134 be the form @xmath135 restricted to the funnel @xmath96 .
we take a standard coordinate chart and we may think of @xmath134 as a @xmath45-form defined on the disk @xmath105 and with support in @xmath103 .
consider as in the proof of lemma [ multiplier ] a subharmonic function @xmath104 in the disk with bounded laplacian and such that @xmath106 if @xmath103 . by the results in (
* thm 2 ) there is a solution @xmath136 to the problem @xmath137 in the disk @xmath105 with the estimate @xmath138 observe that the hyperbolic metric of the disk and of the surface @xmath0 in the funnel are equivalent .
we consider @xmath139 , where @xmath140 is a cutoff function with support in @xmath103 and such that @xmath141 if @xmath142 . the function @xmath143 is extended by @xmath144 to the remaining of @xmath0 and it has the estimate @xmath145 . now
@xmath146 coincides with @xmath135 on the outer part of the funnel @xmath96 .
thus the @xmath45-form @xmath147 has compact support in @xmath0 and it satisfies @xmath148 .
the desired solution is then @xmath149 , where @xmath150 is such that @xmath151 .
we must then solve @xmath151 with weighted uniform estimates but with the advantage that @xmath152 has compact support @xmath49 .
let @xmath153 be a solution operator for @xmath154 .
we take the operator @xmath155 given by lemma [ dbaracotat ] the estimate latexmath:[$\sup_s
@xmath69 with @xmath157 and without zeros in @xmath49 as given in lemma [ multiplier ]
. then we define @xmath158 as @xmath159(z ) = f(z ) t[\omega_k / f](z),\ ] ] it solves @xmath160 = \omega_k$ ] with the estimate @xmath161| e^{-\phi } \le c_k \sup_k \langle\omega_k\rangle e^{-\phi}.\ ] ] the solution is thus @xmath162 $ ] .
[ interpolation ] a separated sequence @xmath13 is interpolating for @xmath1 if and only if the sequences @xmath163 are interpolating in @xmath164 .
we only need to prove that we can pass from the local to the global interpolation property .
we split the proof in two steps 1 . from a funnel @xmath96 to global @xmath0 : we need to prove that there are finite sets @xmath165 such that @xmath166 is interpolating globally .
2 . filling up the remainder .
we shall prove that by adding a finite number of points to an interpolating sequence we still get an interpolating sequence .
thus @xmath29 is interpolating if @xmath167 is interpolating .
let @xmath168 be one of the closed curves on the boundary .
take a funnel @xmath169 with outer end curve in @xmath170 and inner end curve in @xmath84 .
the constant of interpolation in the funnel @xmath169 is @xmath171 .
take a cutoff function @xmath172 with support in the funnel such that @xmath173 ( where @xmath40 is the constant in theorem [ dbar ] ) , the support is in a thick annulus of hyperbolic thickness @xmath174 .
we consider a smaller funnel where @xmath175 .
the sequence @xmath29 in this smaller funnel has still at most interpolation constant @xmath49 .
we can interpolate arbitrary values on @xmath29 being small near the inner curve @xmath84 of @xmath169 in the following way .
take some values @xmath176 with norm one .
take a function in the funnel @xmath69 with norm at most @xmath49 that solves the interpolation problem .
we are going to approximate it by a function in @xmath169 that is small near @xmath84 .
cut it off by @xmath172 and correct via the following inhomogeneous cauchy - riemann equation : @xmath177 the function @xmath178 is holomorphic . by using theorem [ dbar ] on it is possible to solve the equation with a solution @xmath47 such that @xmath179 .
the function @xmath180 does not solve the problem directly but it almost does .
we reiterate the procedure ( interpolating the error @xmath181 and with a convergent series we get finally a function @xmath182 such that @xmath183 , @xmath184 and moreover in the inner half of the funnel that we denote by @xmath185 , @xmath186 .
now it is easier to make it global .
take a new cutoff function @xmath108 with support in the funnel @xmath169 and that is one on the outer part of ( i.e. @xmath187 .
then we need to solve @xmath188 with good global estimates in @xmath0 .
these are given by theorem [ dbar ] .
we have solved the interpolation problem when the sequence lies in the funnels . for the general situation
we only need to add a finite number of points .
the existence of `` blaschke''-type factors @xmath189 provided by theorem [ blashcke ] shows that @xmath190 is interpolating if @xmath29 is interpolating ( it is immediate to build functions in the space such that @xmath191 and @xmath192 ) .
for the sampling part we need the following definition [ associats ] given the pair @xmath193 of a finite riemann surface and a subharmonic function defined on it , we associate to it the pairs : @xmath194 of disks @xmath195 and subharmonic functions @xmath104 defined on the disks as follow : if @xmath196 , @xmath79 are the standard charts of the funnels of @xmath0 we define @xmath197 and @xmath104 is any subharmonic function in @xmath195 such that @xmath198 in the region @xmath103 , @xmath199 in @xmath110 and @xmath200 in @xmath201 .
they can be defined similarly as in , but to make sure @xmath202 we may take instead @xmath203 where @xmath4 is any bounded subharmonic function in @xmath195 such that @xmath204 if @xmath201 and @xmath144 elsewhere .
the funnels @xmath96 can be considered funnels of @xmath0 and they are subdomains of @xmath195 too
. we will exploit this double nature in the following theorem [ sampling ] let @xmath0 be a finite riemann surface and let @xmath3 be a subharmonic function with bounded laplacian .
a separated sequence @xmath29 is sampling for @xmath1 if and only if all the sequences in the funnels @xmath205 are sampling sequences for @xmath206 , where @xmath207 are the associated pairs to @xmath0 given by definition [ associats ] .
thus this theorem and proposition [ interpolation ] show that the properties of sampling and interpolation only depend on the behavior of the sequence and the weight near the boundary pieces . to prove theorem [ sampling ] we need some previous results [ uniqueness ] let @xmath0 be a finite riemann surface and let @xmath3 be a subharmonic function with bounded laplacian .
a sequence @xmath13 is a uniqueness sequence for @xmath1 if and only if all the sequences in the funnels @xmath205 are uniqueness sequences for @xmath206 , where @xmath207 are the associated pairs to @xmath0 given by definition [ associats ] .
it is easier to deal by negation .
let @xmath29 be contained in the zero set of a function @xmath14 .
therefore @xmath208 is in the zero set of @xmath209 .
we divide by a finite number of zeros @xmath210 and we obtain a new function @xmath211 without zeros in @xmath212 and such that @xmath213 .
take the disk @xmath195 and consider the cover by two open sets @xmath95 and @xmath201 .
on the first set we have the function @xmath182 and on the second the function @xmath214 . the quotient is bounded above and below in the intersection of the sets .
this defines a bounded cousin ii in the disk @xmath195 problem that can be solved with bounded data .
we get a new function @xmath215 that vanishes in @xmath216 . we can now add the finite number of zeros @xmath210 without harm .
the reciprocal implication follows with the same argument .
the next result is inspired by a result of beurling ( @xcite ) that relates the property of sampling sequence to that of uniqueness for all weak limits of the sequence . in the context of the bernstein space ( in the original work by beurling ) the space
was fixed ( it was @xmath44 , the space of functions was fixed , the bernstein class , and he considered translates and limits of it of the sampling sequence ) . here we need to move and take limits of the sequence ( by zooming on appropriate portions of it ) but we also need to change the support space ( portions of @xmath0 near the funnel that look like the unit disk ) and we will also move the space of functions by changing the weights . we need some definitions : we consider triplets @xmath217 where @xmath218 are disks @xmath219 , @xmath220 are subharmonic functions defined in a neighborhood of @xmath218 and @xmath221 is a finite collection of points in @xmath218 .
we say that @xmath217 converges weakly to @xmath222 ( where @xmath82 is the unit disk , @xmath3 a subharmonic function in @xmath82 and @xmath29 a discrete sequence in @xmath82 ) if the following conditions are fullfilled : * the domains @xmath218 tend to @xmath82 , i.e. : @xmath223 , * the weights @xmath220 tend to the weight @xmath3 in the sense that @xmath224 as measures converges weakly to @xmath225 .
* the sequences @xmath221 converge weakly to @xmath29 , i.e , the measure @xmath226 converges weakly to the measure @xmath227 .
let us fix a point @xmath228 .
if a sequence of points @xmath229 goes to @xmath230 , i.e. @xmath231 , from a point @xmath232 on it will eventually belong to the union of the funnels @xmath233 .
if we take the set of points @xmath234 then @xmath218 is an hyperbolic disk contained in the funnels if @xmath235 is big enough . in each of the @xmath218 we consider the function @xmath236 and @xmath237
thus for any sequence of points @xmath238 with @xmath239 we build a triplet @xmath217 for @xmath235 big enough . let @xmath240 be the set of all triplets @xmath241 which are weak limits of triplets @xmath217 associated to any sequence @xmath238 such that @xmath239 .
the theorem of beurling on our context is [ beurling ] let @xmath0 be riemann surface of finite type and let @xmath3 be a subharmonic function with bounded laplacian .
a separated sequence @xmath29 is sampling for @xmath1 if and only if * the sequence @xmath29 is a uniqueness set for @xmath1 * for any triplet @xmath242 , the sequence @xmath243 is a uniqueness set for @xmath244 .
let us prove that the uniqueness conditions imply that @xmath29 is a sampling sequence .
if it were not , there would be a sequence of functions @xmath245 such that @xmath246 and @xmath247 .
take a sequence of points @xmath238 with @xmath248 .
if @xmath238 are bounded we can take a subsequence of points that we still denote @xmath238 convergent to @xmath249 and by a normal family argument there is a partial of @xmath250 convergent to @xmath251 , such that @xmath191 , @xmath252 and this is not possible .
thus @xmath238 must be unbounded .
then we take the triplets @xmath217 associated to @xmath238 and @xmath253 because @xmath254 and the hyperbolic radius of @xmath218 is @xmath255 . since @xmath220 has bounded laplacian , the mass of @xmath224 restricted to any compact @xmath49 in @xmath82 is bounded , thus we can take a subsequence that converges weakly to a positive measure @xmath256 in @xmath82 which satisfies @xmath257 because all the mesures @xmath224 satisfy this inequalities with uniform constants .
let @xmath3 be such that @xmath258 .
since @xmath259 are all separated with uniform bound , there is a weak limit @xmath243 .
the functions @xmath250 in the disks can be modified by a factor @xmath260 in such a way that @xmath261 satisfies @xmath262 and @xmath263 , if @xmath235 big enough and @xmath264 .
we can add an harmonic function @xmath150 to @xmath265 in such a way that @xmath266 uniformly on compact sets . thus @xmath267 has a partial convergent to @xmath268 , @xmath269 and @xmath270 which was not possible by assumption . in the other direction , we assume that @xmath29 is a sampling sequence for @xmath1 , and @xmath242 .
we want to prove that any @xmath271 that vanishes in @xmath243 is identically @xmath144 .
take a sequence of points @xmath238 that escapes to infinity and @xmath217 the associated triple that converges weakly to @xmath241 . as @xmath272 and @xmath273 uniformly on compact sets
we can take a sequence of radii @xmath274 such that @xmath275 , @xmath276 and @xmath277 .
if @xmath69 vanishes in @xmath243 that means that @xmath69 is very small in @xmath278 .
assume that @xmath279 .
take a cutoff function @xmath280 such that @xmath281 outside @xmath282 , @xmath283 , and @xmath284 .
define @xmath285 , where @xmath286 is the solution estimates by theorem [ dbar ] .
clearly @xmath287 is small in all points of @xmath100 and it has at least norm @xmath214 .
thus we are contradicting the fact that @xmath29 is sampling .
observe that one particular instance of finite riemann surface , where we can apply the result are the disks @xmath195 associated to the funnels with the metric @xmath104 .
the final piece for the proof of theorem [ sampling ] is then [ disc ] if @xmath0 is a finite riemann surface , @xmath3 a subharmonic function with bounded laplacian and @xmath29 is a uniformly separated sequence , then all possible weak limits coincide with the weak limits of the disks associated to the surface , i.e , @xmath288 the proof amounts to the observation that the metric in @xmath195 converges uniformly to the metric in @xmath0 as @xmath289 , and in the definition of weak limits we only consider uniform convergence over compacts .
theorem [ sampling ] follows now immediately from theorem [ beurling ] and lemmas [ uniqueness ] and [ disc ] .
now theorem [ sampling ] and theorem [ interpolation ] show that the property of being a sampling / interpolating sequence are determined by the behavior near the boundary , more precisely in the associated disks . in these disks
there is a precise description of the interpolating and sampling sequences ( see @xcite and @xcite ) that can be transported to the surface .
if we rewrite it we get the density conditions of theorem [ main ] , but the disks are not hyperbolic disks on the surface , they correspond to hyperbolic disks in disks @xmath195 , but since the condition is only relevant near the boundary , then the disks in both metrics look more an more similar
. moreover the difference between the corresponding green functions converge to @xmath144 uniformly as we go to the boundary .
finally , as the sequence is uniformly discrete and the laplacian of the weight is bounded above and bellow , the small difference is absorbed by the fact that the inequalities are strict and this proves theorem [ main ] .
in fact it is possible to replace in the definition of the density , the green function @xmath290 of @xmath26 by the green function @xmath182 of @xmath0 , because as before @xmath291 as @xmath27 approaches the boundary .
we have considered up to now pointwise growth restrictions .
it is possible to obtain from our theorem other results in different banach spaces of holomorphic functions . consider for instance the weighted bergman spaces @xmath292 where @xmath293 is s the hyperbolic area measure in @xmath0 and @xmath294 .
the natural problem in this context is the following : the spaces @xmath299 can be empty if we only ask @xmath3 to be with positive bounded laplacian .
it is then natural to require that the laplacian is strictly bigger than one so that the laplacian plus the curvature of the metric in the manifold is strictly positive and there are functions in the space ( consider the case of the disk @xmath300 for instance ) .
the proof of the theorem is the same mutatis - mutandi as in the @xmath306 setting .
the basic tool that allows us to glue the pieces together is the next theorem which is the generalization of theorem [ dbar ] and it is proved in the same way : [ dbarlp ] let @xmath0 be a finite riemann surface , let @xmath3 be a subharmonic function with a bounded laplacian strictly bigger than one and let @xmath294 .
there is a constant @xmath307 such that for any @xmath45-form @xmath46 on @xmath0 there is a solution @xmath47 to the inhomogeneous cauchy - riemann equation @xmath131 in @xmath0 with the estimate @xmath308 whenever the right hand is finite .
the proof of this result is again the same as in theorem [ dbar ]
. we can separately solve the c - r equation in each funnel using theorem 2 from @xcite .
we glue them together with a c - r equation with data that has compact support that can be solved with the operator .
arne beurling , _ the collected works of arne beurling .
vol . 1 _ , contemporary mathematicians , birkhuser boston inc . ,
boston , ma , 1989 , complex analysis , edited by l. carleson , p. malliavin , j. neuberger and j. wermer .
jozef dodziuk , thea pignataro , burton randol , and dennis sullivan , _ estimating small eigenvalues of riemann surfaces _ , the legacy of sonya kovalevskaya ( cambridge , mass . , and amherst , mass . , 1985 ) ,
64 , amer .
soc . , providence , ri , 1987 , pp . | we provide a description of the interpolating and sampling sequences on a space of holomorphic functions with a uniform growth restriction defined on finite riemann surfaces . |
the electrical control of magnetization dynamics has been a central issue in the field of spintronics@xcite , owing to its possible applications in magnetic memory devices with low power consumption .
a particularly promising mechanism for the electrical control is to utilize the spin hall effect@xcite ( she ) in a normal metal ( nm ) , such as pt or ta , to convert an electric current into a spin current , and subsequently to magnetization dynamics in an adjacent magnet via mechanisms such as spin - transfer torque@xcite ( stt ) . in reverse , the inverse spin hall effect@xcite ( ishe ) can convert the spin current generated by certain means , for instance spin pumping@xcite , into an electric signal .
a particularly intriguing phenomenon that involves both she and ishe is the spin hall magnetoresistance@xcite ( smr ) , in which a charge current in an nm causes a spin accumulation at the edge of the sample due to she , yielding a finite spin current at the interface to a ferromagnet . through ishe
, the spin current gives an electromotive force along the original charge current , effectively changing the magnetoresistance of the nm .
the two major ingredients that determine smr are the spin diffusion@xcite in the nm and the spin current at the nm / ferromagnet interface .
the spin diffusion part has been addressed in detail by chen _
et al . _ for the nm / ferromagnetic insulator ( nm / fmi )
bilayer , such as pt / y@xmath0fe@xmath1o@xmath2 ( pt / yig ) , and fmi / nm / fmi trilayer@xcite .
this approach solves the spin diffusion equation in the presence of she and ishe in a self - consistent manner , where the spin current at the nm / fmi interface serves as a boundary condition .
however , the interface spin current remains an external parameter for which experimental or numerical input is needed@xcite .
on the other hand , a quantum tunneling formalism has emerged recently as an inexpensive tool to calculate the interface spin current from various material properties such as the insulating gap of the fmi and the interface @xmath3 coupling@xcite . the quantum tunneling theory
also successfully explains@xcite the reduced spin pumping spin current when an additional oxide layer is inserted between nm and fmi@xcite .
it is then of fundamental importance to combine the spin diffusion approach with the quantum tunneling formalism for the interface spin current to give a complete theoretical description of the smr , in particular to quantify how various material properties influence the smr . in this article
we provide a minimal formalism that bridges the quantum tunneling formalism to the spin diffusion approach .
we focus on the smr in nm / fmi bilayer realized in pt / yig , and the nm / ferromagnetic metal ( nm / fmm ) bilayer realized in pt / co and ta / co@xcite . the spin diffusion in the nm is assumed to be described by the same formalism of chen _
et al._@xcite , whereas the interface spin current is calculated from the quantum tunneling formalism@xcite . in the nm / fmm bilayer
, we consider an fmm that has long spin diffusion length and a small thickness , such that the spin diffusion effect is negligible and the spin transport is predominately of quantum origin@xcite .
this is presumably adequate for the case of ultrathin co films@xcite , but not for materials with very short spin diffusion length such as permalloy@xcite . within this formalism , the effect of material properties including spin diffusion length of the nm , interface @xmath3 coupling , insulating gap of the fmi , and the thickness of each layer can all be treated on equal footing . in particular , we reveal the signature of quantum interference in smr in nm / fmm bilayer , and discuss the situation in which it can be observed . the structure of the article is arranged in the following manner . in sec .
ii , we detail the quantum tunneling formalism for the interface spin current in the nm / fmi bilayer , and how it is adopted into the spin diffusion approach that describes the nm .
section iii generalizes this recipe to the nm / fmm bilayer , and discuss the observability of the predicted signature of quantum interference in smr .
section iv gives the concluding remark .
we start with the quantum tunneling formalism that calculates the interface spin current in the nm / fmi bilayer , which later serves as the boundary condition for the spin diffusion equation that determines smr .
the quantum tunneling formalism describes the nm / fmi bilayer shown in fig .
[ fig : nmfmi_gr_gi ] ( a ) by the hamiltonian @xmath4 where @xmath5 is the spin voltage of @xmath6 produced by an in - plane charge current @xmath7 , @xmath8 is the fermi energy , @xmath9 is the insulating gap , and @xmath10 is the magnetization .
we choose @xmath11 such that the magnetization has the tendency to align with the conduction electron spin @xmath12 .
the wave function near the interface is @xmath13 where @xmath14 and @xmath15 .
the amplitudes @xmath16 are solved in terms of the incident amplitude @xmath17 by matching wave functions and their first derivative at the interface .
the @xmath18 and @xmath19 regions are assumed to be vacuum or insulating oxides that correspond to infinite potentials such that the wave functions vanish there for simplicity .
we identify the incident flux with @xmath20 where @xmath21 is the density of states per @xmath22 with @xmath23 the fermi wave length .
the spin current inside the fmi at position @xmath24 is calculated from the evanescent wave function @xmath25 \ : .
\label{nmfmi_spin_current_general}\end{aligned}\ ] ] angular momentum conservation@xcite dictates that the interface spin current to be equal to the stt exerts on the magnetization @xmath26 , \label{nmfmi_interface_spin_current}\end{aligned}\ ] ] which defines the field - like @xmath27 and dampling - like @xmath28 spin mixing conductance that in turn can be calculated from the interface spin current@xcite @xmath29 a straight forward calculation yields @xmath30 where @xmath31 is @xmath32 component of pauli matrix , and @xmath33 equation ( [ nmfmi_gr_gi ] ) describes the spin mixing conductance in stt , as well as that in spin pumping since the onsager relation@xcite is satisfied in this approach@xcite .
both @xmath28 and @xmath27 have very weak dependence ( at most few percent ) on the angle of magnetization @xmath34 through @xmath35 , which may be considered as higher order contributions@xcite . in the numerical calculation below we set @xmath36 without loss of generality .
numerical results of the spin mixing conductance @xmath37 are shown in fig .
[ fig : nmfmi_gr_gi ] , plotted as a function the fmi thickness @xmath38 and at different strength of the interface @xmath3 coupling @xmath39 . both @xmath28 and @xmath27
increase with @xmath38 initially and then saturate to a constant as expected , since they originate from the quantum tunneling of conduction electrons that only penetrate into the fmi over a very short distance . at a fmi thickness small compared to fermi wave length @xmath40 , we found that @xmath41 and @xmath42 , therefore the damping - like to field - like ratio is @xmath43 . in most of the parameter space , the torque is dominated by field - like component @xmath44 throughout the whole range of @xmath38 .
only when the magnitude of @xmath3 coupling is large compared to the insulating gap @xmath45 is the torque dominated by the damping - like component @xmath46 , consistent with that found previously@xcite and also in accordance with the result from first principle calculation@xcite .
the magnitude of @xmath37 generally increases with the @xmath3 coupling , yet more dramatically for @xmath28 .
note that @xmath28 and @xmath27 do not depend on the nm thickness in this quantum tunneling approach . and an fmi with thickness @xmath38 .
( b ) the spin mixing conductance @xmath37 versus the fmi thickness @xmath38 , at different values of interface @xmath3 coupling strength @xmath47 .
the insulating gap strength is fixed at @xmath48 .
the absolute units for @xmath37 is @xmath49 which is about @xmath50m@xmath51 depending on the fermi wave length @xmath52 . ]
we adopt the spin diffusion approach of chen _ et al._@xcite to address the effect of the interface spin current in eq .
( [ nmfmi_interface_spin_current ] ) on smr , which is briefly summarized below .
the spin diffusion approach is based on the following assumptions for the spin transport in the nm : ( 1 ) the spin current in nm consists of two parts , one from the spatial gradient of spin voltage and the other the bare spin current caused directly by she , @xmath53 where @xmath54 is spin hall angle , @xmath55 is the conductivity of nm , @xmath56 is applied external electric in @xmath57 direction , and @xmath58 is electron charge .
( 2 ) the spin voltage obeys the spin diffusion equation @xmath59 , where @xmath60 is the spin diffusion length . ( 3 ) spin current vanishes at the edge of nm ( @xmath61 ) , which serves as one boundary condition .
( 4 ) the spin current at the nm / fmi interface is described by eq .
( [ nmfmi_interface_spin_current ] ) , which serves as another boundary condition .
the self - consistent solution satisfying ( 1)@xmath62(4 ) is@xcite @xmath63 , \nonumber \\ \frac{\boldsymbol{j}_{x } \cdot \hat{\boldsymbol{y}}}{j_{\textrm{sh } } } & = & \beta_{x}\sin\theta \big [ \cos \theta \sin \varphi \operatorname{re } \big ( \widetilde{g } \big ) - \cos \varphi \operatorname{im } \big ( \widetilde{g } \big ) \big ] , \nonumber \\ \frac{\boldsymbol{j}_{x } \cdot \hat{\boldsymbol{z}}}{j_{\textrm{sh } } } & = & 1 - \frac{\cosh ( \frac{2x+l_{n}}{2 \lambda})}{\cosh ( \frac{l_{n}}{2 \lambda } ) } - \beta_{x } \sin^{2 } \theta \operatorname{re } \big ( \widetilde{g } \big ) , \label{nmfmi_js}\end{aligned}\ ] ] where @xmath64 and @xmath65 is the bare spin current . here
@xmath66 is a negative parameter ( because we assume the interface @xmath3 coupling @xmath67 ) that bridges our tunneling formalism to the spin diffusion equation , and @xmath68 is the complex spin mixing conductance . through ishe ,
the spin currents in eq .
( [ nmfmi_js ] ) is converted back to a charge current in the longitudinal ( along @xmath69 ) and transverse ( along @xmath70 ) direction @xmath71 the conductivity averaged over the nm layer then follows @xmath72 using @xmath73 , the longitudinal and transverse component of smr read @xmath74 where @xmath75 clearly the fmi thickness @xmath38 affects @xmath76 and @xmath77 only through @xmath78 . to perform numerical calculation of eq .
( [ eq : rho5 ] ) , we make the following assumption on the parameter @xmath79 in eq .
( [ nmfmi_gtilde ] ) that connects the quantum tunneling formalism with the spin diffusion equation .
firstly , @xmath79 contains the density of state per @xmath22 at the fermi surface , which is assumed to be the inverse of fermi energy @xmath80 .
the combined parameter @xmath81 therefore represents the strength of @xmath3 coupling .
other parameters that influence @xmath79 are the spin diffusion length assumed to be @xmath82 nm , the conductivity of the nm film taken to be @xmath83m@xmath84 , and fermi wave length assumed to be roughly equal to the lattice constant @xmath85 nm , all of which are the typical values for commonly used materials such as pt .
these lead to the dimensionless parameter @xmath86 in eq .
( [ nmfmi_gtilde ] ) being expressed in terms of the relative strength of @xmath3 coupling and the spin mixing conductance divided by its unit . in what follows
, we examine the effect of fmi thickness , nm thickness , insulating gap , and interface @xmath3 coupling on smr . on the contrary , the spin hall angle , spin diffusion length , and conductivity
are treated as constants , although in reality they may also depend on the layer thickness or on each other in such thin films@xcite .
the numerical result of smr is shown in fig .
[ fig : nmfmi_smr ] , plotted as a function of the fmi thickness @xmath38 and nm thickness @xmath87 at several values of insulating gap @xmath45 and @xmath3 coupling @xmath39 . as a function of the fmi thickness @xmath38 , both the longitudinal @xmath88 and the transverse @xmath89 component initially increase and then saturate at around @xmath90 , which is expected since conduction electrons only tunnel into the fmi over a short depth , so the interface spin current saturates once the fmi is thicker than this tunneling depth .
the insulating gap @xmath45 obviously affects the tunneling depth , and is particularly influential on the magnitude of longitudinal @xmath88 , as can be seen by comparing plots with different @xmath45 in fig .
[ fig : nmfmi_smr ] .
the magnitude of @xmath88 also generally increases with the @xmath3 coupling @xmath39 , while the transverse component @xmath89 at large @xmath39 displays a nonmonotonic dependence on the fmi thickness . on the other hand , as a function of nm thickness @xmath87 , both smr components increase and peak at around @xmath91 and then decrease monotonically for large @xmath87 .
this can be understood because both @xmath76 and @xmath77 are interface effects that become less significant compared to bulk resistivity @xmath92 when nm thickness increases , and the spin voltage is known to be maximal when the nm thickness is comparable to the spin diffusion length@xcite @xmath91 .
and ( d ) @xmath93 in this system , plotted against the thickness @xmath94 of the fmm and @xmath3 coupling @xmath47 , in units of @xmath95 where @xmath52 is the fermi wave length . ]
we proceed to address the smr in the nm / fmm bilayer , with the assumption that the fmm film is much thinner than its spin diffusion length @xmath96 such that quantum tunneling is the dominant mechanism for spin transport in the fmm , while the spin diffusion inside the fmm can be ignored .
the calculation of the spin current at the nm / fmm interface starts with the model schematically shown in fig .
[ fig : nmfmmfinite_gr_gi ] ( a ) .
the nm and fmm occupy @xmath97 and @xmath98 , respectively .
the nm region is described by eqs .
( [ eq : hamiltonian_nm ] ) and ( [ eq : wave function_nm ] ) , while the fmm layer is described by @xmath99 and the wave function @xmath100 where @xmath101 . the wave functions outside of the bilayer in @xmath102 and @xmath18
are assumed to vanish for simplicity .
the coefficients @xmath103 are again determined by matching wave functions and their first derivative at the interface .
the interface spin current and the spin mixing conductance are calculated from eqs .
( [ nmfmi_spin_current_general ] ) to ( [ mixing_from_js ] ) , with replacing @xmath104 to @xmath105 and @xmath38 to @xmath94 , resulting in @xmath106 \ : , \label{eq : spin_mixing_conductance_nm / fmm}\end{aligned}\ ] ] where @xmath107 with @xmath108 .
apart from a change in magnitude , the pattern of the spin mixing conductance as a function of @xmath3 coupling and the fmm thickness shown in fig .
[ fig : nmfmmfinite_gr_gi ] is almost indistinguishable from that reported in fig . 2 of ref . , which shows clear signals of quantum interference with respect to both @xmath3 coupling and fmm thickness
this similarity is expected , since the only difference between the formalism here and in ref .
is the insulating gap @xmath9 of the substrate or vacuum in the @xmath102 region in fig .
[ fig : nmfmmfinite_gr_gi ] ( a ) , which is assumed to be infinite here for simplicity but finite in ref . .
the insulating gap is spin degenerate and essentially does not influence the spin transport . and transverse @xmath89 component of smr in the nm / fmm bilayer , plotted against the fmm thickness in units of fermi wave length @xmath109 and nm thickness in units of the spin diffusion length @xmath110 , at various strength of @xmath3 coupling @xmath47 . ] to get smr , we use eq .
( [ eq : rho1 ] ) @xmath62 ( [ eq : rho5 ] ) while taking the @xmath111 obtained from eq .
( [ eq : spin_mixing_conductance_nm / fmm ] ) .
the results for the longitudinal @xmath88 and transverse @xmath89 component of smr as functions of fmm thickness @xmath94 and nm thickness @xmath87 are shown in fig .
[ fig : nmfmm_smr ] , for several values of @xmath3 coupling @xmath39 . as a function of nm thickness
, both components reach a maximal at around the spin diffusion length @xmath91 and then decrease monotonically , similar to that reported in fig .
[ fig : nmfmi_smr ] for nm / fmi bilayer and is due to the spin diffusion effect explained in sec . [ section : smr ] . on the other hand , as a function of fmm thickness , both components show clear modulations with an average periodicity that decreases with increasing @xmath3 coupling , a trend similar to that of @xmath28 and @xmath27 shown in fig .
[ fig : nmfmmfinite_gr_gi ] and is attributed to the quantum interference of spin transport .
intuitively , a larger @xmath3 coupling renders a faster precession of conduction electron spin when it travels inside the fmm , hence more modulations appear for a given fmm thickness .
the transverse component of smr is found to be generally one order of magnitude smaller than the longitudinal component .
the experimental detection of the oscillation of smr with respect to fmm thickness @xmath94 shown in fig .
[ fig : nmfmm_smr ] would be a direct proof of our approach .
in a typical nm / fmi set up , however , there are other sources that contribute to the total resistance measured in experiments , therefore it is important to investigate whether there is a situation in which the predicted oscillation of smr can manifest . to explore this possibility
, we use a three - resistor model to characterize the total longitudinal resistance@xcite , which contains the resistor that represents the nm layer ( @xmath112 ) , the fmm layer ( @xmath113 ) , and the interface layer ( @xmath114 ) connected in parallel , each denoted by @xmath115 with @xmath116 . here
@xmath117 is the contribution to the longitudinal resistance in layer @xmath118 that does not depend on the angle of the magnetization , and @xmath119 is the part that depends on the angle which is generally much smaller @xmath120 .
expanding the total longitudinal resistance to leading order in @xmath119 yields @xmath121 each resistance is assumed to satisfy the usual relation to the sample size @xmath122 , where @xmath123 and @xmath124 are the length and the width of the sample , respectively , @xmath125 and @xmath126 are the corresponding resistivity , and @xmath127 is the thickness of layer @xmath118 .
the thickness of the interface @xmath128 is assumed to be intrinsically constant , in contrast to @xmath87 and @xmath129 that can be varied experimentally@xcite .
the percentage change of the total resistance due to the angle of the magnetization is @xmath130 note that the @xmath131 factors are monotonic functions of the layer thickness @xmath132 , and are independent from the angle of the magnetization .
the contribution to the angular dependent part of @xmath133 comes from the anisotropic magnetoresistance ( amr ) which takes the form@xcite @xmath134 since the in - plane charge current @xmath135 runs along @xmath136 as shown in fig .
[ fig : nmfmmfinite_gr_gi ] ( a ) , and we denote @xmath137 as the unit vector along the direction of the magnetization . in addition ,
et al._@xcite showed that the interface resistance has a quadratic dependence on both @xmath138 and @xmath139 , a result of surface spin - orbit scattering .
on the other hand , the smr in the nm has the angular dependence@xcite described by eq .
( [ eq : rho1 ] ) .
these considerations lead to the parametrization of resistivity by @xmath140\;. \nonumber \\ \label{rho_nif}\end{aligned}\ ] ] combinig this with eq .
( [ rtot_percentage ] ) motivates us to propose the following experiment that should isolate the effect of longitudinal smr represented by @xmath141 . from eq .
( [ rho_nif ] ) , we see that @xmath142 and @xmath143 vanish if the magnetization does not have an in - plane component , i.e. , @xmath144 , while @xmath141 remains finite as long as the out - of - plane component is nonzero @xmath145 .
thus we propose to fix the magnetization of the fmm film to be out - of - plane @xmath145 , in which case the percentage change of total longitudinal resistance as a function of fmm thickness takes the form @xmath146 where @xmath92 , @xmath147 , and @xmath76 are those in eqs .
( [ eq : rho1 ] ) and ( [ eq : rho5 ] ) , @xmath148 and @xmath149 are two length scales that can be treated as fitting parameters in experiments .
equation ( [ rtot_sysz0 ] ) indicates that , for the case of only out - of - plane magnetization , the percentage change of magnetoresistance decays with the fmm thickness @xmath129 due to the @xmath150 factor , but also oscillates with @xmath129 due to the @xmath151 factor as quantified in eq .
( [ eq : rho5 ] ) and shown in fig .
[ fig : nmfmm_smr ] . thus varying fmm thickness while keeping its magnetization
out - of - plane may be a proper set up to observe the predicted oscillation of longitudinal smr , provided the fmm thickness remains thinner than its spin relaxation length @xmath152 .
finally , we remark that the convention of labeling coordinate in smr or stt experiments is that the charge current is defined to be along @xmath153 and the direction normal to the film is along @xmath154 .
therefore the coordinate in our tunneling formalism @xmath155 corresponds to @xmath156 in the experimental convention .
in summary , the quantum tunneling formalism for the interface spin current is incorporated into the spin diffusion approach to study the effect of various material properties on smr , in particular the effect of layer thickness , insulating gap , and interface @xmath3 coupling .
the advantage of combining the quantum and diffusive approach is that the effects of all these material properties can be treated on equal footing . for the nm / fmi case
, we reveal an smr that saturates at large fmi thickness since the conduction electrons only tunnels into the fmi over a short distance , whereas the longitudinal and transverse smr display different dependence on the insulating gap and interface @xmath3 coupling . for the nm / fmm case
, we predict that smr may display a pattern of oscillation as increasing fmm thickness due to quantum interference , and propose an experiment to observe it by using fixed out - of - plane magnetization to isolate smr from other contributions .
we anticipate that our minimal model that combines the quantum and diffusive approach may be used to guide the search for suitable materials that optimize the smr , and help to predict novel spin transport effects in ultrathin heterostructures in which quantum effects shall not be overlooked .
the authors acknowledge the fruitful discussions with p. gambardella , f. casanova , and j. mendil .
w. c. and m. s. are grateful for the financial support through a research grant of the swiss national science foundation . | we present a formalism that simultaneously incorporates the effect of quantum tunneling and spin diffusion on spin hall magnetoresistance observed in normal metal / ferromagnetic insulator bilayers ( such as pt / y@xmath0fe@xmath1o@xmath2 ) and normal metal / ferromagnetic metal bilayers ( such as pt / co ) , in which the angle of magnetization influences the magnetoresistance of the normal metal . in the normal metal side
the spin diffusion is known to affect the landscape of the spin accumulation caused by spin hall effect and subsequently the magnetoresistance , while on the ferromagnet side the quantum tunneling effect is detrimental to the interface spin current which also affects the spin accumulation .
the influence of generic material properties such as spin diffusion length , layer thickness , interface coupling , and insulating gap can be quantified in a unified manner , and experiments that reveal the quantum feature of the magnetoresistance are suggested . |
let @xmath0 be a finite dimensional vector spaces ( of dimension @xmath1 )
. of fundamental importance is the basis .
there are various reasons to choose a particular basis , and each basis has advantages and disadvantages .
often it is the action of certain functionals that lends importance to a particular basis .
for example , the lagrange basis is important because it is dual to point evaluation . in general ,
any basis @xmath2 $ ] for @xmath0 has a dual map @xmath3 $ ] for @xmath4 satisfying @xmath5 .
that is , @xmath6 , the identity matrix .
conversely , any basis @xmath7 is dual to some basis @xmath8 of @xmath0 .
it is the dual map @xmath9 that extracts information from the functions that is of particular interest here , which arguably plays a more prominent role than the basis @xmath8 itself .
this process allows one to consider the _ information _ we are trying to capture as our primary goal .
the above concepts are well - known and well - studied . in this paper , we are interested in investigating bases for _ subspaces _ @xmath10 of @xmath0 that are dual to _ subsets _ of the functionals in the map @xmath9 .
for example , suppose that @xmath3 $ ] are linearly independent on the n - dimensional space @xmath0 , and @xmath10 is an m - dimensional subspace of @xmath0 with @xmath11 .
then , given an injective map @xmath12 \rightarrow [ 1:n]$ ] , our questions are : 1 .
is the subset @xmath13 $ ] of @xmath14 linearly independent on @xmath10 ?
2 . if linearly independent , what is the basis @xmath15 for @xmath10 that is dual to @xmath16 in the sense that @xmath17 .
3 . what are the properties of this dual basis . in this paper
we consider the action of subsets of functionals for certain basis on _ subspaces _ of the original space .
this allows us to view how the subspace looks according to information on the whole space . in some sense
, we are addressing the question `` how do you approximate with less ( or not enough ) information ? '' but it is not always possible to construct such bases , because the functionals are not always _ linearly independent _ on the subspace .
for example , the only subset of @xmath18 $ ] that is linearly independent on @xmath19 $ ] is @xmath20 $ ] , not @xmath21 $ ] or @xmath22 $ ] . as we will see in this paper ,
the situation is different in the bernstein basis .
it is the objective of this paper to determine if dual bases exist for certain subspaces of polynomial spaces , and if so to compute and characterize these bases , and then determine their properties . in particular , we show that in the bernstein basis for @xmath23 , any @xmath24-selection of the dual functionals are linearly independent on @xmath25 .
moreover , by choosing a particular symmetric choice of the functionals , we show that the corresponding dual basis converge to the lagrange polynomial basis .
later , we derive certain approximation results concerning quasi - interpolation and a bernstein - type operator .
as defined above , @xmath0 is a vector space of dimension @xmath1 with basis @xmath8 and dual map @xmath9 , and @xmath10 a subspace of @xmath0 of dimension @xmath26 , @xmath27 , with basis @xmath28 . for @xmath29 , we have @xmath30 for some coefficient sequences @xmath31 and @xmath32 . applying the dual basis , we have @xmath33 hence , @xmath34 with @xmath35 .
the matrix @xmath36 embeds the coefficients @xmath37 of @xmath38 in the basis for @xmath10 to it s coefficients in the basis for @xmath0 . by inclusion of @xmath10 into @xmath0
, we can construct the embedding @xmath39 of @xmath10 into @xmath0 . and
since @xmath40 for all @xmath41 , it follows that @xmath42 .
all said , this can be visualized as in the following commutative diagram .
@xmath43^e & & x \\ \rr^{m } \ar[u]^{\phi^m } \ar[rr]^e & & \rr^n \ar[u]_{\phi^n } } \ ] ] moreover , we note that @xmath44 therefore , @xmath45 .
hence , we have the following result concerning the action of @xmath9 on a subspace @xmath10 of @xmath0 . [ prop2 ]
let @xmath46 be an @xmath26-dimensional subspace of a vector space @xmath47 , with basis @xmath28 dual to @xmath48 and @xmath8 dual to @xmath9 .
then , @xmath10 embeds into @xmath0 by the map @xmath49 with @xmath50 , and the bases transform as @xmath51 further , the dual basis for @xmath4 map to the dual basis for @xmath52 by the map @xmath53 since @xmath54 , the functionals @xmath55 are also functionals on @xmath10 .
but since @xmath56 , the functionals in @xmath9 can not be linearly independent on @xmath10
. however , since @xmath57 is a 1 - 1 matrix , @xmath9 does span @xmath52 ( moreover , @xmath58 is a basis for @xmath52 since @xmath59 is invertible ) .
therefore , we can trim @xmath9 to a basis for @xmath52 by removing _ inessential _
vectors ( this is a standard construction , c.f .
this leaves an @xmath26-subvector @xmath16 of @xmath9 that is linearly independent on @xmath10 , where @xmath60 is an injective map @xmath12 \rightarrow [ 1:n]$ ] ( which we call a _ selection map _ ) .
then , with @xmath61 the @xmath62 identity matrix , we have @xmath63 therefore , an @xmath26-subvector @xmath16 of @xmath9 is linearly independent on @xmath10 iff @xmath64 is invertible .
moreover , when @xmath64 is invertible , the basis of @xmath65 that is dual to @xmath16 is @xmath66 , as follows by computing the change of basis @xmath67 : @xmath68 hence , the basis of @xmath10 that is dual to @xmath16 is @xmath69 .
the following summarizes the above statements : [ prop1 ] let @xmath47 be a vector space with basis @xmath8 dual to @xmath9 , and let @xmath46 be a subspace with basis @xmath42 dual to @xmath48 .
let @xmath12 \rightarrow [ 1:n]$ ] denote a selection ( injective ) map . 1 .
@xmath70 is linearly independent on @xmath10 and of length @xmath26 ( hence a basis for @xmath52 ) .
2 . there exist ( at least one ) selections @xmath16 of @xmath9 linearly independent on @xmath10 .
sometimes only one selection is linearly independent , and sometimes all selections are linearly independent ( this is of particular interest later in this paper ) .
3 . @xmath71 4 .
@xmath16 is linearly independent on @xmath10 iff @xmath64 is invertible .
5 . if @xmath16 is linearly independent on @xmath10 , then the basis of @xmath10 that is dual to @xmath16 in the sense that @xmath72 is @xmath73
in this section we state definitions and properties useful for certain constructions that we consider in the subsequent sections .
firstly , by proposition [ prop1 ] , there always exists some @xmath60 such that @xmath16 is linearly independent on @xmath10 , and in this case we have the basis @xmath73 with @xmath74 .
however , this may or may not be true for all @xmath60 .
hence , we give the following definition to make this distinction .
we say the embedding of @xmath46 into @xmath47 is _ complete _ if @xmath75 is linearly independent on @xmath10 for all injective maps ( selections ) @xmath12 \rightarrow [ 1:n]$ ] .
let @xmath76 be the space of polynomials of degree at most @xmath26 , and let @xmath77 be the space of polynomials of degree at most @xmath1 . 1 .
let @xmath28 and @xmath8 be power basis for @xmath78 and @xmath79 .
then , the embedding is not complete .
moreover , @xmath16 is linearly independent on @xmath28 iff @xmath80 $ ] .
2 . let @xmath28 and @xmath8 be bernstein bases for @xmath78 and @xmath79 , respectively .
then , the embedding is complete .
part ( 2 ) is non - trivial , and will be proved later in this paper . for part ( 1 ) , recall that the power basis @xmath81 $ ] dual to @xmath82 $ ] , and @xmath83 is the subspace with @xmath27
. then @xmath84 for @xmath85 , with @xmath61 is the @xmath86 identity matrix .
it follows by proposition [ prop2 ] that @xmath87 on @xmath25 .
let @xmath60 be a selection map .
since @xmath88 is invertible iff @xmath89 , it follows that @xmath16 is linearly independent on @xmath25 iff @xmath89 .
that is , on @xmath25 , @xmath90 = [ \lambda_0^m , \ldots , \lambda_m^m]$ ] .
hence , there is only one choice for the selection @xmath60 of @xmath9 to consider , and this choice is @xmath91 .
hence , the lagrange basis is not complete .
the next properties are characteristic of bernstein bases , which we consider later in this paper . 1 .
let @xmath8 be a basis of finite - dimensional function space .
we say @xmath8 is _ affine on @xmath92 _ if @xmath93 is an affine combination of @xmath37 for all @xmath94 .
that is , @xmath95 for all @xmath96 .
if it holds for all @xmath97 , then we say @xmath8 is _
affine_. 2 .
let @xmath36 be a ( real ) matrix .
we say @xmath36 is _ row affine _ if @xmath98 for all @xmath99 , and _ column affine _ if @xmath100 for all @xmath101 .
[ lem1 ] 1 .
matrix inversion preserves the property row - affine .
that is , the inverse of an invertible row - affine matrix is row - affine .
matrix multiplication preserves the property row - affine .
that is , the product of two row - affine matrices is row - affine .
[ laffine ] for the first result , let @xmath102 be an @xmath103 invertible row - affine matrix .
then @xmath104 for @xmath105 .
since @xmath106 , @xmath107 for @xmath108 .
hence , @xmath109 is row - affine . for the second result , assume @xmath110 with @xmath102 and @xmath111 row - affine of dimensions @xmath112 and @xmath113 , respectively
then , @xmath114 is of dimension @xmath115 , and @xmath116 for @xmath117 . therefore , @xmath114 is row - affine .
[ thm1 ] let @xmath28 and @xmath8 be affine bases of @xmath10 and @xmath0 , respectively , with @xmath10 a subspace of @xmath0 . 1 . if @xmath118 for some matrix @xmath36 , then @xmath36 is row affine .
if , moreover , @xmath64 is invertible for some selection @xmath12 \rightarrow [ 1:n]$ ] , then the dual basis @xmath119 exists and is affine . for ( 1 ) , it follows by affineness of the two bases that @xmath120 let @xmath121 be the functional on @xmath0 such that @xmath122 .
then , @xmath123 and so @xmath124 this establishes ( 1 ) . for ( 2 ) , we recall that @xmath119 is the basis for @xmath10 dual to @xmath16 with @xmath125 , and hence @xmath71 . by part ( 1 ) of this theorem , @xmath64 is row affine , and by lemma [ laffine ] , @xmath126 is row affine as well .
therefore , @xmath127 and so @xmath15 is an affine basis : these results regarding affine bases will concern the bernstein basis , which we investigate in the remaining sections , and in particular we show that dual bases are affine .
we remark here that the same is not true of convexity .
that is , the dual bases constructed are not convex .
indeed , matrix inversion does not preserve convexity , as it does affineness .
morever , as it turns out , certain approximation properties do not actually require convexity .
i.e. , affineness is enough .
hence , our dual bases @xmath15 will enjoy many of the same properties as @xmath128 do , in the bernstein - basis setting , just not convexity . in our construction
we derive dual bases in terms of data maps that are dual on a large space .
however , it is possible to have different data maps that are both dual to the same bases .
for example , in terms of the point - evaluation function @xmath129 and derivative operator @xmath130 , both maps @xmath131 $ ] and @xmath132 $ ] are dual to the power basis @xmath133 $ ] , as can be seen by @xmath134 .
however , even with different data maps , dual bases are the same , as shown next .
this idea is important in constructing approximation operators where some data maps may apply and not others .
more to the point , we will derive dual basis in this paper with respect to dual data maps that involve differentiation , hence do not apply on @xmath135)$ ] . in the last section of this paper ,
we define new data maps that do not involve differentiation to derive certain properties of approximation operators on @xmath135)$ ] .
[ prop3 ] let @xmath9 and @xmath136 be data maps on @xmath0 both dual to the basis @xmath8 .
then both maps are equivalent on @xmath0 , and the dual bases in a subspace @xmath10 are invariant of which dual map is used .
that is , if @xmath15 and @xmath137 are bases for the subspace @xmath10 that are dual to @xmath28 with respect to the selection @xmath60 and data maps @xmath9 and @xmath136 , respectively .
then , @xmath138 . to show that @xmath9 and @xmath136 are equivalent on @xmath0 , let @xmath139 .
then , since both maps are dual to @xmath8 , we have @xmath140 and @xmath141 .
hence , they are equivalent on @xmath0 .
the basis @xmath15 is dual to @xmath75 in the sense that @xmath142 . with this basis
represented @xmath143 , for some embedding , we get @xmath144 therefore , @xmath145 and so @xmath146 , which is depends only on the embedding @xmath36 and not the data map @xmath9 . hence , if @xmath147 is the transformation matrix for @xmath148 , then @xmath149 , and so @xmath138 .
let @xmath23 be the space of polynomials of degree at most @xmath1 and @xmath25 the subspace of polynomials of degree at most @xmath26 , for @xmath27 . in this section
we derive bases for @xmath25 that are dual to subsets of the dual bernstein functionals on @xmath23 .
we begin with some basic formulas involving bernstein polynomials that will be used . *
bernstein basis for @xmath23 : @xmath150 $ ] with @xmath151 , @xmath152 and @xmath153 on @xmath154 $ ] .
that is , it forms a partition of unity on @xmath154 $ ] .
in particular , @xmath155 is a convex combination of @xmath37 for all @xmath156 $ ] .
* degree elevation of bernstein basis ( see @xcite ) : @xmath157 with @xmath36 the @xmath158 matrix with entries @xmath159 for @xmath160 and @xmath161 , with @xmath162 if @xmath163 or @xmath164 .
that is , @xmath165 * dual bernstein functionals : @xmath166 $ ] with @xmath167 therefore , @xmath168 and @xmath169 .
( note that the two forms are equivalent on @xmath23 , but not on all spaces . ) * reduction of dual bernstein basis : on @xmath25 , @xmath170 ( by proposition [ prop2 ] ) .
hence , @xmath171 hence , in the context of this paper , we have the following : let @xmath128 be the bernstein basis for @xmath78 , and let @xmath172 be the bernstein basis for @xmath79 with dual map @xmath9 given above , @xmath173 .
let @xmath36 be the degree elevation matrix .
then , 1
. the embedding @xmath174 of @xmath175 into @xmath176 is complete .
the basis for @xmath78 dual to @xmath16 can be represented @xmath177 , with @xmath178 when @xmath179 . for ( 1 ) , we need to show that @xmath16 is linearly independent on @xmath78 for any selection @xmath180 \rightarrow [ 0:n]$ ]
. the conversion from between the power basis @xmath181 and bernstein basis @xmath172 for @xmath23 can be expressed @xmath182 with @xmath183)$ ] and @xmath184 pascal s ( lower triangular ) matrix @xmath185 = \begin{bmatrix } 1 & 0 & 0 & \cdots & 0 \\ 1 & 1 & 0 & \cdots & 0 \\ 1 & 2 & 1 & \cdots & 0 \\ \vdots & \vdots & \vdots & \ddots & \vdots \\ \binom{n}{0 } & \binom{n}{1 } & \binom{n}{2 } & \cdots & \binom{n}{n } \end{bmatrix}.\ ] ] this follows directly from the identity @xmath186 ( see @xcite , section 2.8 , for a short proof ) .
let @xmath187 $ ] .
then , @xmath188 in @xcite it was shown that the truncated pascal matrix @xmath189 is invertible for any @xmath24 selection @xmath180 \rightarrow [ 0:n]$ ] of the rows of @xmath190 .
therefore , @xmath191 is invertible , and so @xmath16 is linearly independent on @xmath25 . for ( 2 )
, we have @xmath192 with @xmath36 the degree elevation matrix .
therefore , we are exactly in the framework of proposition [ prop1 ] , with @xmath193 and @xmath194 . since the embedding is complete , we have that @xmath64 is invertible for any selection map @xmath60 , therefore , the dual basis of @xmath78 dual to @xmath16 exists for any selection map @xmath60 , and can bre represented @xmath195 . in the case @xmath179 , there is only one selection @xmath196 $ ] , and so @xmath197 .
hence , @xmath198 is the identity matrix , and @xmath199 are therefore dual to @xmath9 .
since , as is well known , @xmath128 and @xmath172 are affine bases , the next result follows directly from theorem [ thm1 ] .
instead , we give a more direct proof below using properties of the degree elevation matrix @xmath36 .
[ thm2 ] for any selection ( injective ) map
@xmath180 \rightarrow [ 0:n]$ ] , @xmath200 is an affine basis of @xmath25 dual to @xmath16 , with @xmath201 .
we note that by the previous lemma @xmath16 is linearly independent on @xmath25 .
therefore , @xmath202 is invertible . by proposition
[ prop1 ] , @xmath203 and @xmath177 is the basis for @xmath25 dual to @xmath16 .
recall that @xmath204 by the chu - vandermonde identity @xmath205 and so @xmath64 is row affine . by lemma [ laffine ]
, @xmath206 is row affine as well .
therefore , @xmath207 and so @xmath15 is an affine basis : the next property is a generalization of the following property of bernstein polynomials : @xmath208 for all @xmath209 , with @xmath210 .
[ thm3 ] let @xmath177 be the dual basis for some selection map @xmath60 . 1 .
@xmath211 2 .
3 . @xmath213 .
recall that the bernstein functions on @xmath79 can be represented @xmath214 in particular , @xmath215 . for @xmath216 ,
the sum of the terms for @xmath217 are zero except when @xmath218 .
hence , @xmath219 therefore , @xmath220 . from this
we get , @xmath221 thereofore , @xmath222 .
hence , @xmath223 implies @xmath224 , and @xmath225
since dual bases @xmath15 are bases for the subspace @xmath10 , and function in @xmath10 can be represented by this basis . in particular , in the berntstein setup above , any polynomial @xmath226 can be represented as @xmath227 for some @xmath228 . and
so , this basis can be used in computation with functions in this polynomial space .
now , since @xmath229 with @xmath146 , we can write @xmath230 .
hence , one can transform the coefficients @xmath37 by the matrix @xmath102 , and then use b - form techniques in computation . in particular , to plot the curves , one can use decasteljau s algortihm on the control polygon with points @xmath231 , for @xmath232 .
this is depicted in figure [ f2 ] .
the control polygon for the coefficients @xmath37 is displayed in solid broken line and the transformed control polygon is dashed .
as shown in the previous section , for bernstein functionals the matrix @xmath233 is invertible for any selection map @xmath60 , and we can therefore find dual bases for any selection map .
moreover , these bases are affine . in this section
we will use this idea to produce a certain `` symmetric '' class of bases , and show that these converge to the lagrange polynomial basis .
we also provide an estimate for the rate of convergence . for @xmath234 ,
let @xmath235 for @xmath232 .
hence , @xmath236 with @xmath237 .
for example : * if @xmath238 then @xmath239 and we get @xmath80 $ ] . *
if @xmath240 and @xmath241 , then @xmath242 and @xmath243 $ ] .
the dual bernstein bases are then @xmath244 with @xmath245 .
the goal in the remainder of this section is to show that symmetric dual bases converge to point evaluation in a certain sense . to get the most general result
, we extend the bernstein functional to allow @xmath246 to be any real number : @xmath247 here , the factorials in the binomial coefficients will involve the gamma function when @xmath209 is a non - integer .
the functionals reduce to the above formulation when @xmath248 is a non - negative integer with @xmath249 . for the following , we use
the _ falling factorial _
notation @xmath250 [ lem3 ] for @xmath251 and @xmath252 , @xmath253 with @xmath254 for non - integer factorials . by the well - known property @xmath255
, we get that @xmath256 therefore , @xmath257 let @xmath251 and @xmath258 for some @xmath26 .
the dual functionals converge to point evaluation in the following sense @xmath259 in particular , with @xmath260 and @xmath261 , @xmath262 let @xmath263 in @xmath25 . then , @xmath264 if @xmath265 and @xmath266 otherwise , and so by the previous lemma we have @xmath267 let @xmath268 and @xmath269 with @xmath270 .
then , @xmath271 with @xmath272 the lagrange basis . in figure [ f1 ]
, we display the dual bases for various degree polynomial spaces and level of refinement to illustrate convergence to the lagrange basis . of degree @xmath273 , @xmath274 and @xmath275 , for @xmath238 ( solid ) and @xmath276 (
dotted ) , respectively .
note that @xmath238 is the bernstein basis , and @xmath277 is close to the lagrange basis .
, width=672 ]
in this section we determine the rate of convergence for this symmetric configuration to lagrange interpolation , which moreover provides an alternate proof or convergence to lagrange interpolation .
recall that @xmath278 is the basis for @xmath25 that is dual to @xmath279 with @xmath280 for @xmath232 , and @xmath281 with @xmath282 .
let @xmath272 be the lagrange basis for @xmath25 dual to point evaluation at @xmath283 for @xmath232 , and let @xmath102 be the matrix such that @xmath284 .
then , we have the following : [ lem4 ] for @xmath285 small : 1 . @xmath286 2 . @xmath287 .
part ( 2 ) follows by a maclaurin s expansion . for ( 1 ) , there are @xmath288 terms . on expanding in powers of @xmath285 , we have @xmath289h + o(h^2 ) \\ & = \alpha^{n-\ell+1 } -\alpha^{n-\ell } \frac{(n+1)n - \ell(\ell-1)}{2 } h + o(h^2 ) \\ & = \alpha^{n-\ell+1 } - \frac12 \alpha^{n-\ell } ( n+\ell)(n+1-\ell)h + o(h^2 ) .
\end{aligned}\ ] ] [ lem5 ] for the symmetric configuration , @xmath290 if @xmath291 or @xmath26 .
otherwise , @xmath292 as @xmath293 , with @xmath294.\ ] ] the lagrange basis is dual to the point evaluation map @xmath295 $ ] , giving @xmath296 . therefore , @xmath297.\ ] ] we also have @xmath298 with @xmath36 the degree elevation matrix . hence , @xmath299 it is easily checked that this vanishes for @xmath291 or @xmath300 ( provided @xmath301 .
hence , we only need to consider @xmath302 . in this case ,
@xmath303 and @xmath304 for @xmath305 , which , since @xmath306 , we can assume as well . factoring out @xmath307 , we can rewrite the second term as @xmath308 by lemma [ lem4 ] , when @xmath309 , we have @xmath310 + o(\frac{1}{k^2}).\end{aligned}\ ] ] likewise , when @xmath311 , @xmath312 \frac{1}{k } + o(\frac{1}{k^2 } ) \\ & = \big(1-\frac{i}{m}\big)^{m - j } + \frac12\big(\frac{m - i}{m}\big)^{m - j } ( m - j ) \big [ \frac{m+j-1}{m } - \frac{m - j-1}{m - i } \big ] \frac{1}{k } + o(\frac{1}{k^2 } ) \\ & = \big(1-\frac{i}{m}\big)^{m - j } \big[1 - \frac12 \frac{(m - j)(2mj - im+i - ij)}{m(m - i ) } \frac{1}{k}\big ] + o(\frac{1}{k^2}).\end{aligned}\ ] ] multiplying these two terms together and by @xmath307 gives us , when @xmath313 , @xmath314 \big[1 - \frac12 \frac{(m - j)(2mj - im+i - ij)}{m(m - i ) } \frac{1}{k}\big ] + o(\frac{1}{k^2 } ) \\ & = b_j^m\big(\frac{i}{m}\big ) -
\frac12 b_j^m\big(\frac{i}{m}\big ) \big [ \frac{(j-1)j(m - i)}{im } + \frac{(m - j)(2mj - im+i - ij)}{m(m - i ) } \big ] \frac{1}{k } + o(\frac{1}{k^2 } ) .
\end{aligned}\ ] ] for @xmath315 , @xmath316 for @xmath317 , @xmath318 putting this all together , we get : @xmath292 with @xmath294.\ ] ] let @xmath319 and @xmath320 with @xmath321 the lagrange basis and @xmath322 with @xmath323 the dual basis for the symmetric configuration given in the previous section . then ,
@xmath324 } = \big({||a||}^2_\infty { ||c||}_\infty \frac{1}{k } + o(\frac{1}{k^2 } ) \big ) { ||\alpha||},\ ] ] and @xmath325 } = { ||a||}_\infty^2 { ||c||}_\infty \frac{1}{k } + o(\frac{1}{k^2}).\ ] ] from @xmath326 , we get @xmath327 since , as we proved in the previous section , that @xmath328 converges to the lagrange basis @xmath272 , we know that @xmath329 . hence
, @xmath330 . we also know that @xmath331 therefore , @xmath332 now , let @xmath333 and @xmath334 for @xmath335 . then , @xmath336 } & = { ||b^m(a - a_k)\alpha||}_{[0,1 ] } \\ & = \max_{t\in [ 0,1 ] } { ||b^m(t ) ( a - a_k)\alpha|| } \\ & \leq \max_{t\in [ 0,1 ] } { ||b^m(t)||}_\infty { ||a - a_k||}_\infty { ||\alpha||}_\infty \\ & = { ||a - a_k||}_\infty { ||\alpha||}_\infty\\ & = \big({||a||}_\infty^2 { ||c||}_\infty \frac{1}{k } + o(\frac{1}{k^2})\big ) { ||\alpha||}_\infty.\end{aligned}\ ] ] in particular , for @xmath337 the standard unit vector with @xmath338 in the i - th slot , we get @xmath325 } = { ||a||}_\infty^2 { ||c||}_\infty \frac{1}{k } + o(\frac{1}{k^2}).\ ] ]
recall that for the symmetric configuration introduced in the previous section the dual basis is represented in terms of the bernstein basis @xmath339 for some @xmath340 matrices @xmath341 .
it seems rather challenging to explicitly characterize all these transformation matrices for arbitrary @xmath26 and @xmath342 , however , for the convenience of the reader we ll list out the first several here . to make things more compact , we define the following notation : @xmath343 , @xmath344 , @xmath345 , and @xmath346 .
= 0mu = 0mu = 0mu @xmath347 a_3,k = 12k^2 & 0 & 0 & 0 -2 ( k1 ) ( 5k1 ) & 6 ( 2k1 ) ( 3k1 ) & -6 ( k1 ) ( 3k1 ) & 2 ( k1 ) ( 2k1 ) 2 ( k1 ) ( 2k1 ) & -6 ( k1 ) ( 3k1 ) & 6 ( 2k1 ) ( 3k1 ) & -2 ( k1 ) ( 5k1 ) 0 & 0 & 0 & 12k^2 .
a_4,k = 72k^3 & 0 & 0 & 0&0 -3 ( k1 ) ( 26k^29k1 ) & 12 ( 2k1 ) ( 3k1 ) ( 4k1 ) & -18 ( k1 ) ( 3k1 ) ( 4k1 ) & 12 ( k1 ) ( 2k1 ) ( 4k1 ) & -3 ( k1 ) ( 2k1 ) ( 3k1 ) 4 ( k1 ) ( 13k^212k2 ) & -32 ( k1 ) ( 2k1 ) ( 4k1 ) & 24 ( 4k1 ) ( 5k^26k2 ) & -32 ( k1 ) ( 2k1 ) ( 4k1 ) & 4 ( k1 ) ( 13k^212k2 ) -3 ( k1 ) ( 2k1 ) ( 3k1 ) & 12 ( k1 ) ( 2k1 ) ( 4k1 ) & -18 ( k1 ) ( 3k1 ) ( 4k1 ) & 12 ( 2k1 ) ( 3k1 ) ( 4k1 ) & -3 ( k1 ) ( 26k^29k1 ) 0 & 0 & 0 & 0 & 72k^3 .
0mu 0mu 0mu a_5,k = 480k^4 & 0 & 0 & 0 + & & 0 & 0 -4 ( k1 ) ( 7k1 ) ( 22k^27k1 ) & 20 ( 2k1 ) ( 3k1 ) ( 4k1 ) ( 5k1 ) & -40 ( k1 ) ( 3k1 ) ( 4k1 ) ( 5k1 ) & 40 ( k1 ) ( 2k1 ) ( 4k1 ) ( 5k1 ) & & -20 ( k1 ) ( 2k1 ) ( 3k1 ) ( 5k1 ) & 4 ( k1 ) ( 2k1 ) ( 3k1 ) ( 4k1 ) 2 ( k1 ) ( 269k^3331k^2109k11 ) & -10 ( k1 ) ( 2k1 ) ( 5k1 ) ( 29k11 ) & 20 ( 5k1 ) ( 59k^3101k^259k11 ) & -20 ( k1 ) ( 2k1 ) ( 5k1 ) ( 23k11 ) & & 10 ( k1 ) ( 5k1 ) ( 37k^242k11 ) & -2 ( k1 ) ( 2k1 ) ( 77k^272k11 )
-2(k1 ) ( 2k1 ) ( 77k^272k11 ) & 10 ( k1 ) ( 5k1 ) ( 37k^242k11 ) & -20 ( k1 ) ( 2k1 ) ( 5k1 ) ( 23k11 ) & 20 ( 5k1 ) ( 59k^3101k^259k11 ) & & -10 ( k1 ) ( 2k1 ) ( 5k1 ) ( 29k11 ) & 2 ( k1 ) ( 269k^3331k^2109k11 ) 4 ( k1 ) ( 2k1 ) ( 3k1 ) ( 4k1 ) & -20 ( k1 ) ( 2k1 ) ( 3k1 ) ( 5k1 ) & 40 ( k1 ) ( 2k1 ) ( 4k1 ) ( 5k1 ) & -40 ( k1 ) ( 3k1 ) ( 4k1 ) ( 5k1 ) & & 20 ( 2k1 ) ( 3k1 ) ( 4k1 ) ( 5k1 ) & -4 ( k1 ) ( 7k1 ) ( 22k^27k1 ) 0 & 0 & 0 & 0 & & 0 & 480k^4 . in particular , for @xmath273 and @xmath348 : @xmath349 for @xmath350 and @xmath351 : @xmath352 for @xmath240 and @xmath353 : @xmath354 @xmath355 for @xmath356 and @xmath357 : @xmath358 @xmath359
in this section we derive some approximation results for our dual basis functions , similar to what is done in @xcite in the multivariate setting . to do so
, we will redefine the bernstein basis and dual functionals over a general interval .
* let @xmath172 with @xmath360 be the bernstein basis over @xmath361 $ ] , with dual functionals @xmath362 hence , @xmath169 .
* let @xmath15 be the basis for @xmath78 dual to @xmath16 for some selection @xmath60 .
that is , @xmath363 .
then , @xmath229 with @xmath146 where @xmath364 .
* let @xmath272 be the lagrange basis at the points @xmath365 $ ] with dual map @xmath366 $ ] .
hence , @xmath367 .
then , for @xmath368 with @xmath369 , is the basis transformation .
let @xmath370 .
the purpose of the map @xmath371 is two - fold .
first , it is dual to @xmath128 , as we show next .
second , it does not involve any derivative evaluations , which makes it suitable for the approximation of functions in @xmath372 $ ] .
@xmath136 is dual to @xmath172 .
@xmath373 ( stability ) let @xmath374 .
then , @xmath375 } \leq { ||a||}_\infty { ||\alpha||}_\infty.\ ] ] for the upper bound : @xmath376 } & = { ||d^m\alpha||}_{[0,1 ] } = { ||b^ma\alpha||}_{[0,1 ] } = { ||\sum_{i=0}^m ( a\alpha)_i b_i^m ||}_{[0,1 ] } \\ & \leq { ||a\alpha||}_\infty { ||\sum_{i=0}^m b_i^m ||}_{[0,1 ] } = { ||a\alpha||}_\infty \leq { ||a||}_\infty { ||\alpha||}_\infty.\end{aligned}\ ] ] for the lower bound , note that @xmath377 and so @xmath378 then , @xmath379}.\end{aligned}\ ] ] since @xmath380 is row - affine , @xmath381 , and so we get @xmath382}.\ ] ] for this we recall from proposition [ prop3 ] that the dual basis is unaffected by which map is used .
hence , we restate this fact in the following lemma : [ lem7 ] suppose that data maps @xmath9 and @xmath148 are both dual to the basis @xmath172 for @xmath79 . then , these data maps are equivalent on @xmath79 , and dual bases @xmath15 of @xmath78 for @xmath11 are identical for both maps . for @xmath383)$ ] ,
@xmath384 } \ { ||m^{-1}||}_\infty.\ ] ] let @xmath385 .
then , @xmath386 , and so @xmath387 } \
\sum_{i=0}^n |c(i , j)| \leq { ||f||}_{[a , b ] } \ { ||m^{-t}||}_1 = { ||f||}_{[a , b ] } \ { ||m^{-1}||}_\infty.\end{aligned}\ ] ] for any selection @xmath60 , let @xmath388 ) \rightarrow \rr : f \mapsto d^m \tilde \lambda^{n}(s)^tf.\ ] ] then , we have following : [ lem6 ] @xmath389 is a linear projector of @xmath390)$ ] onto @xmath78 .
linearity is immediate from @xmath391 and idempotency follows from @xmath392 since @xmath393 .
therefore , @xmath389 is a linear projector .
let @xmath394 $ ] .
then , * @xmath395 } \leq { ||a||}_\infty { ||m_m^{-1}||}_\infty { ||f||}_{[a , b]}$ ] , * @xmath396 } \leq ( 1+{||q_s|| } ) d(f,\$_m)_{[a , b]}$ ] . for @xmath397 $ ] , @xmath398}.\end{aligned}\ ] ]
this establishes the first result . for the second result ,
take an arbitrary @xmath258 .
then , @xmath399 } & \leq { ||f - p||}_{[a , b ] } + \underbrace{{||p - q_sp||}_{[a , b]}}_0 + { ||q_s(p - f)||}_{[a , b ] } \\ & \leq ( 1 + { ||q_s||}_{[a , b ] } ) { ||f - p||}_{[a , b ] } \\ & \leq ( 1 + { ||q_s||}_{[a , b ] } ) d(f,\$_m)_{[a , b]}.\end{aligned}\ ] ]
let @xmath400 be a bernstein - like operator for our dual functions , with @xmath401 .
let @xmath402 } : = \max \ { |f^{(k)}(x)| : x\in[a , b]\},\ ] ] with @xmath403 } : = { ||f||}_{0,[a , b]}$ ] , and @xmath404 , |x - y| \leq h\},\ ] ] the uniform modulus of continuity relative to the interval @xmath361 $ ] .
then , we have the following : @xmath405 } \leq \begin{cases } { ||a||}_\infty \omega(f , b - a ) , & f \in c([a , b ] ) ; \\ ( b - a){||a||}_\infty { ||f'||}_{[a , b ] } , & f \in c^1([a , b ] ) ; \\ \frac12 ( b - a)^2 { ||a||}_\infty { ||f''||}_{[a , b ] } , & f \in c^2([a , b ] ) . \end{cases}\ ] ] by affineness of the basis @xmath406 and convexity of @xmath128 for all @xmath407 $ ] , @xmath408 this gives the first case .
the second case follows from the first case and the estimate @xmath409}.\end{aligned}\ ] ] then , for some @xmath410 between @xmath411 and @xmath209 , @xmath412 d^m_i| \\ & = |f(x ) - \sum_{i=0}^m f(x ) d_i^m + \sum_{i=0}^m f'(x)(\xi_{s(i)}^n - x ) d_i^m + \frac12 \sum_{i=0}^m f''(\eta ) ( \xi_{s(i)}^n - x)^2 d^m_i |
\\ & = |\sum_{i=0}^m f'(x)(\xi_{s(i)}^n - x ) d_i^m + \frac12 \sum_{i=0}^m f''(\eta ) ( \xi_{s(i)}^n - x)^2 d^m_i | \\ & = |f'(x ) \sum_{i=0}^m \xi_{s(i)}^nd_i^m - f'(x ) x + \frac12 \sum_{i=0}^m f''(\eta ) ( \xi_{s(i)}^n - x)^2 d^m_i | \\\end{aligned}\ ] ] by theorem [ thm3 ] ( tranformed to the interval @xmath361 $ ] ) , @xmath413 , and so @xmath414 } ( b - a)^2 \sum_{i=0}^m |d^m_i(x)| \\ & \leq \frac12 { ||f''||}_{[a , b ] } ( b - a)^2 { ||a||}.\end{aligned}\ ] ] | in this paper we study dual bases functions in subspaces .
these are bases which are dual to functionals on larger linear space .
our goal is construct and derive properties of certain bases obtained from the construction , with primary focus on polynomial spaces in b - form .
when they exist , our bases are always affine ( not convex ) , and we define a symmetric configuration that converges to lagrange polynomial bases . because of affineness of our bases , we are able to derive certain approximation theoretic results involving quasi - interpolation and a bernstein - type operator . in a broad sense
, it is the aim of this paper to present a new way to view approximation problems in subspaces . in subsequent work
, we will apply our results to dual bases in subspaces of spline and multivariate polynomial spaces , and apply this to the construction of blended function approximants used for approximation in the sum of certain tensor product spaces .
[ multiblock footnote omitted ] |
a curve on a surface @xmath0 is said to be if it has a continuous and nonvanishing derivative . consider two such curves whose curvatures take values in some given interval , starting in the same direction ( prescribed by a unit vector tangent to @xmath0 ) and ending in another prescribed direction .
it is a natural problem to determine whether one curve can be deformed into the other while keeping end - directions fixed and respecting the curvature bounds . from another viewpoint
, one is asking for a characterization of the connected components of the space of all such curves .
more ambitiously , what is its homotopy or homeomorphism type ?
the answer can be unexpectedly interesting , and it is closely linked to the geometry of @xmath0 . in this article
this question is investigated when @xmath0 is hyperbolic , that is , a ( possibly nonorientable ) smooth surface endowed with a complete riemannian metric of constant negative curvature , say @xmath3 .
it is well known that any such surface can be expressed as the quotient of the hyperbolic plane @xmath4 by a discrete group of isometries .
thus any curve on @xmath0 can be lifted to a curve in @xmath5 having the same curvature ( at least in absolute value , if @xmath0 is nonorientable ) . as will be explained more carefully later ,
this implies that one can obtain a solution of the problem about spaces of curves on @xmath0 if one knows how to solve the corresponding problem in the hyperbolic plane for all pairs of directions .
let @xmath6 ( the unit tangent bundle of @xmath5 ) be two such directions .
let @xmath7 be the set of all smooth regular curves on @xmath5 whose initial ( resp .
final ) unit tangent vectors equal @xmath8 ( resp .
@xmath9 ) and whose curvatures take values inside @xmath10 , furnished with the @xmath11-topology .
there are canonical candidates for the position of connected component of this space , defined as follows .
let @xmath12 denote the universal cover .
fixing a lift @xmath13 of @xmath8 , one obtains a map @xmath14 by looking at the endpoint of the lift to the universal cover , starting at @xmath13 , of curves in @xmath15 .
since @xmath16 is discrete , this yields a decomposition of the former into closed - open subspaces . more concretely ,
let @xmath17 \to \c $ ] be a regular plane curve .
an @xmath18 \to \r $ ] for @xmath19 is a continuous function such that @xmath20 always points in the direction of @xmath21 ; note that there are countably many such functions , which differ by multiples of @xmath22 .
the of @xmath19 is defined as @xmath23 .
because @xmath5 is diffeomorphic to @xmath24 , any regular curve in the former also admits arguments and a total turning .
however , these have no geometric meaning since they depend on the chosen diffeomorphism . in any case , once such a choice has been made , it gives rise to the decomposition @xmath25 where @xmath26 consists of those curves in @xmath15 which have total turning @xmath27 , and @xmath27 runs over all total turnings , viz .
, those for which @xmath9 is parallel to @xmath28 ( regarded as vectors in @xmath24 ) . the closed - open subspaces appearing on the right side of , which will be referred to as the of @xmath29 , are independent of the diffeomorphism : they are in bijective correspondence with the fiber @xmath16 .
if @xmath30 and @xmath31 , so that no restriction is imposed on the curvature , then they are in fact precisely the components of @xmath7 , and each of them is contractible . however , for general curvature bounds they may fail to be connected , contractible , or even nonempty .
the possibilities depend above all upon the relation of @xmath1 to the interval @xmath2 $ ] .
if @xmath1 is contained in @xmath2 $ ] , then at most one of the canonical subspaces of @xmath29 is nonempty .
this subspace is always contractible .
thus @xmath7 itself is either empty or contractible .
if @xmath1 is disjoint from @xmath2 $ ] , then infinitely many of the canonical subspaces are empty , and infinitely many are nonempty .
the latter are all contractible .
if @xmath1 contains @xmath2 $ ] , then none of the canonical subspaces is empty .
we conjecture that most are contractible , but finitely many of them may have the homotopy type of an @xmath32-sphere , for some @xmath33 depending upon the subspace .
this conjecture is partly motivated by results about the corresponding spaces of curves in the euclidean plane obtained in @xcite and @xcite .
finally , if @xmath1 overlaps @xmath2 $ ] , then infinitely many of the canonical subspaces are empty , and infinitely many are nonempty .
nevertheless , as in the previous case , these subspaces may not be connected .
we do not venture a guess at their homotopy type .
after briefly introducing some notation and definitions , [
s : basic ] begins with a discussion of curves of constant curvature @xmath34 in the hyperbolic plane : circles ( @xmath35 ) , horocycles ( @xmath36 ) and elcrics ( @xmath37 ) .
then a transformation is defined which takes a curve and shifts it by a fixed distance along the direction prescribed by its normal unit vectors .
its effect on the regularity and curvature of the original curve is investigated , and applied to reduce the dependence of the topology of @xmath7 to four real parameters , instead of the eight needed to specify @xmath38 and @xmath39 . in
[ s : voidness ] the voidness of the canonical subspaces is discussed .
it is proven that if @xmath1 is contained in @xmath2 $ ] , then the image of any curve in @xmath7 is the graph of a function when seen in an appropriate conformal model of the hyperbolic plane , which we call the mercator model . in particular , at most one of its canonical subspaces is nonempty .
if @xmath1 contains @xmath40 $ ] , then none of the canonical subspaces is empty . in the two remaining cases
, there is a critical value @xmath41 for the total turning such that @xmath42 is nonempty for all @xmath43 and empty for all @xmath44 ( or the reverse , depending on whether @xmath10 contains points to the right or left of @xmath2 $ ] ) .
section [ s : frame ] explains how a curve of constrained curvature may also be regarded as a curve in the group of orientation - preserving isometries of @xmath45 satisfying certain conditions .
this perspective is sometimes useful . in [
s : disjoint ] it is shown that the nonempty canonical subspaces of @xmath7 are all contractible when @xmath1 is disjoint from @xmath2 $ ] .
the idea is to parametrize all curves in such a subspace by the argument of its unit tangent vector when viewed as a curve in the half - plane model , and to take ( euclidean ) convex combinations .
the proof intertwines various euclidean and hyperbolic concepts , and seems to be highly dependent on this particular model . in the mercator model @xmath46 , vertical translations are isometries .
because any curve in @xmath7 may be parametrized as @xmath47 when @xmath1 is contained in @xmath2 $ ] , the bounds on its curvature can be translated into two differential inequalities involving @xmath48 and @xmath49 , but not @xmath50 itself .
this allows one to produce a contraction of the space by working with the associated family of functions @xmath51 , as carried out in
[ s : contained ] .
in [ s : general ] we define spaces of curves with constrained curvature on a general surface @xmath0 , not necessarily hyperbolic nor orientable .
it is shown that if @xmath0 is compact then any such space is nonempty ; this is also true in the euclidean plane , but not in the hyperbolic plane , as mentioned previously .
moreover , riemannian coverings of @xmath0 induce homeomorphisms between spaces of curves on @xmath0 and on the covering space .
the section ends with a brief discussion of spaces of closed curves without basepoint conditions .
several useful constructions , notably one which is essential to the proof of the main result of [ s : contained ] , may create discontinuities of the curvature , and thus lead out of the class of spaces @xmath52 defined in
[ s : general ] . to circumvent this ,
the paraphernalia of @xmath53 functions is used in [ s : discontinuous ] to define another family of spaces , denoted @xmath54 , which are hilbert manifolds .
the curves in these spaces are regular but their curvatures are only defined almost everywhere . the main result of [ s : discontinuous ] states that the natural inclusion @xmath55 is a homotopy equivalence with dense image . sections [ s : general ] and [ s : discontinuous ] are independent of the other ones . in this article
many auxiliary results are stated as exercises .
most of these are straightforward computations or else become trivial once the situation is visualized in the appropriate model .
the more difficult and important ones are accompanied by `` hints '' , sometimes amounting to full sketches of a solution ; these are collected in [ s : hints ] .
the bibliographies of @xcite and @xcite contain more references to works on related problems .
it is assumed that the reader is familiar with the geometry of the hyperbolic plane as discussed , for instance , in chapter 2 of @xcite , the expository article @xcite , chapter 7 of @xcite or chapters 35 of @xcite .
when speaking of the hyperbolic plane with no particular model in mind , we denote it by @xmath5 .
the underlying sets of the ( poincar ) disk , half - plane and hyperboloid models are denoted by : @xmath56 the circle at infinity is denoted by @xmath57 or @xmath58 , the norm of a vector @xmath9 tangent to @xmath5 by @xmath59 , and the riemannian metric by @xmath60 .
when working with @xmath61 or @xmath62 , both @xmath4 and its tangent planes are regarded as subsets of @xmath24 in the disk and half - plane models , we will select the orientation which is induced on the respective underlying sets by the standard orientation of @xmath24 . in the hyperboloid model , a basis @xmath63 of a tangent plane is declared positive if @xmath64 is positively oriented in @xmath65 ; equivalently , the lorentzian vector product @xmath66 points to the interior region bounded by @xmath67 ( i.e. , the one containing the light - cone ) .
given a regular curve @xmath68\to \hh^2 $ ] , its is the map @xmath69\to ut\hh^2,\quad \ta : = \frac{\dot \ga}{\abs{\dot \ga}}.\ ] ] let @xmath70 denote the bundle map ( `` multiplication by @xmath71 '' ) which associates to @xmath72 the unique vector @xmath73 of the same norm as @xmath9 such that @xmath74 is orthogonal and positively oriented .
then the to @xmath19 is given by @xmath75\to ut\hh^2 , \quad \no:= j \circ \ta.\ ] ] assuming that @xmath19 has a second derivative , its is the function @xmath76 \to \r,\quad \ka:=\frac{1}{\abs{\dot\ga}}\big\langle\frac{d\ta}{dt},\no\big\rangle = \frac{1}{\abs{\dot\ga}^2}\big\langle\frac{d\dot\ga}{dt},\no\big\rangle;\ ] ] here @xmath61 denotes covariant differentiation ( along @xmath19 ) .
the hyperboloid model is usually the most convenient one for carrying out computations , since it realizes @xmath5 as a submanifold of the vector space @xmath65 .
for instance , the curvature of a curve @xmath19 on @xmath67 is given by : @xmath77 where @xmath78 denotes the bilinear form on @xmath65 and @xmath79 is the associated quadratic form . [
d : spaces ] let @xmath80 and @xmath81 .
then @xmath7 ( resp .
@xmath82 ) denotes the set of @xmath83 regular curves @xmath84 \to \hh^2 $ ] satisfying : 1 .
@xmath85 and @xmath86 ; 2 .
@xmath87 ( resp .
@xmath88 ) throughout @xmath89 $ ] .
this set is furnished with the @xmath83 topology , for some natural number @xmath90 .
[ d : translation ] let @xmath68\to \hh^2 $ ] be a regular curve and @xmath91 .
the @xmath92\to \hh^2 $ ] of @xmath19 by @xmath93 is the curve given by @xmath94),\ ] ] where @xmath95 is the unit normal to @xmath19 and @xmath96 the ( riemannian ) exponential map . in the hyperboloid model , @xmath97).\ ] ]
[ d : osculating ] two curves @xmath98 \lto{c^2 } s $ ] on a smooth surface @xmath99 will be said to each other at @xmath100 $ ] if one may reparametrize @xmath19 , keeping @xmath101 fixed , so that @xmath102 [ x : osculating ] suppose that @xmath0 is oriented and furnished with a riemannian metric . then @xmath19 and @xmath103 osculate each other at @xmath104 if and only if @xmath105 [ d : circle ] a is a curve consisting of the set of all points lying a fixed distance away from a certain point in @xmath5 .
an is a curve consisting of one component of the set of all points lying a fixed distance away from a certain geodesic in @xmath5 .
a is a curve which meets all geodesics through a certain point of @xmath58 orthogonally .
a is a distance - preserving map @xmath106 ; such a ray is said to @xmath107 .
[ x : constant ] in order to understand the effect of a geometric transformation on a given curve , it is often sufficient to replace the latter by its family of osculating constant - curvature curves . because the hyperbolic plane is homogeneous and isotropic
, one can represent these in a convenient position in one of the models as a means of avoiding calculations .
define an to be either a line or a circle in @xmath24 ; its center is taken to be @xmath108 if it is a line . 1
. in the disk and half - plane models , a circle , horocycle or elcric appears as the intersection with @xmath5 of an euclidean circle which is disjoint , secant or ( internally ) tangent to @xmath58 , respectively .
2 . in the hyperboloid model , a circle , elcric or horocycle appears as a ( planar ) ellipse , hyperbola or parabola , respectively .
3 . sketch `` all '' constant - curvature curves tangent to a given @xmath109 in the various models .
4 . circles , elcrics and horocycles are the orbits of elliptic , hyperbolic and parabolic one - parameter subgroups of isometries , respectively . in particular , they are homogeneous and have constant curvature . 5 . if all points of a circle lie at a distance @xmath110 from a certain point , then its curvature is given by @xmath111 .
thus , all circles have curvature greater than @xmath112 in absolute value .
if all points of an elcric lie at a distance @xmath110 from a certain geodesic , then its curvature is given by @xmath113 .
thus , all elcrics have curvature less than @xmath112 in absolute value . 7 .
the curvature of a horocycle equals @xmath114 .
circles , elcrics , horocycles and their subarcs account for all constant - curvature curves . in view of the sign ambiguity in the preceding formulas
, it is desirable to redefine circles , elcrics and horocycles not as subsets of @xmath5 , but as oriented curves therein .
the @xmath115 of a circle or elcric is now defined so that its curvature @xmath34 is given by @xmath116 respectively .
then @xmath117 is the distance from the circle ( resp .
elcric ) to the point ( resp .
geodesic ) to which it is equidistant . both expressions for the curvature apply to horocycles if these are regarded as circles / elcrics of radius @xmath118 , a convention which we shall adopt .
observe that in all cases the sign of @xmath119 is the same as that of the curvature .
[ x : orientation ] let the circle at infinity be oriented from left to right in @xmath62 and counter - clockwise in @xmath61 . in either of these models ( compare ): 1 .
if an elcric meets the circle at infinity at an angle @xmath120 , then its curvature equals @xmath121 .
2 . if a circle is oriented ( counter-)clockwise , then its curvature is less than @xmath122 ( greater than 1 ) .
3 . extend the preceding assertions to include horocycles .
let @xmath123 and @xmath124 . then there exist exactly two elcrics of curvature @xmath34 through @xmath125 and @xmath126 .
furthermore , in the models @xmath61 or @xmath62 : 1 .
this pair of elcrics is related by inversion in @xmath127 , and their euclidean centers lie along the euclidean perpendicular bisector @xmath128 of @xmath129 .
let @xmath130 and @xmath131 denote the euclidean centers of @xmath57 and the geodesic through the pair @xmath132
any point of @xmath128 except for @xmath130 and @xmath131 is the euclidean center of a unique elcric of curvature in @xmath133 through @xmath125 and @xmath126 .
describe the orientation of this elcric in terms of the position of its center .
2 . using ( a ) , describe how the curvature of this elcric changes as its center moves along @xmath128 .
[ x : normal ] let @xmath134 . 1 .
the normal translation by @xmath93 of a circle of radius @xmath119 is a circle of radius @xmath135 , equidistant to the same point as the original circle .
what happens as @xmath93 goes from 0 to @xmath136 ? 2 .
the normal translation by @xmath93 of an elcric of radius @xmath119 is an elcric of radius @xmath137 , equidistant to the same geodesic as the original elcric .
what happens as @xmath93 goes from 0 to @xmath136 ?
a normal translation of a horocycle is another horocycle , meeting orthogonally the same family of geodesics as the original horocycle .
more concisely , the normal translation of a constant - curvature curve of radius @xmath138 by @xmath91 is a curve of the same type of radius @xmath135 . 1
. a normal translation of an elcric ( resp .
horocycle ) meets @xmath139 in the same points ( resp .
point ) as the original elcric ( resp .
horocycle ) , when represented in one of the models @xmath140 or @xmath62 .
[ x : normal2 ] let @xmath68\to \hh^2 $ ] be smooth and regular , @xmath141}\ka(t)\quad \text{and}\quad \ka_+=\max_{t\in [ 0,1]}\ka(t).\ ] ] assume that @xmath142 $ ] .
then : 1 . the normal translation @xmath143 is regular . in particular
, @xmath144 is regular for all @xmath93 in some open interval containing 0 , and for all @xmath145 in case @xmath146\subs [ -1,1 ] $ ] .
@xmath147 if these are regarded as taking values in @xmath148 . given
@xmath149 $ ] , there exists a unique constant - curvature curve which osculates @xmath19 at @xmath150 .
the @xmath151 of @xmath19 at @xmath150 is defined as the radius of this osculating curve . 1 .
the radii of curvature of @xmath144 and @xmath19 are related by @xmath152 .
the curvature of @xmath153 is given by :
@xmath154 3 . @xmath155 .
the topology of @xmath29 depends in principle upon eight real parameters : three for each of @xmath156 , and two for the curvature bounds .
this number can be halved by a suitable use of normal translations and isometries . in the sequel
, two intervals are said to if they intersect but neither is contained in the other one .
[ p : reduction ] let @xmath157 and @xmath158 be given
. then there exist @xmath159 and @xmath160 such that @xmath161 is canonically homeomorphic to a space of the type listed in table [ ta : reduction ] .
c c c case & @xmath29 homeomorphic to & range of @xmath160 ' '' '' + @xmath1 contained in @xmath2 $ ] & @xmath162 & @xmath163 $ ] ' '' '' + @xmath1 disjoint from @xmath2 $ ] & @xmath164 & @xmath165 ' '' '' + @xmath1 overlaps @xmath2 $ ] & @xmath166 & @xmath167 ' '' '' + @xmath1 contains @xmath2 $ ] & @xmath168 & @xmath169 $ ] ' '' '' + the special role played by the interval @xmath2 $ ] stems from the fact that any normal translation of a horocycle is a horocycle .
the four classes listed in table [ ta : reduction ] are genuinely different in terms of their topological properties , as discussed in the introduction . in the only case not covered by , namely @xmath170 , given @xmath171
, one can find @xmath172 such that @xmath173 is homeomorphic to @xmath174 ; for this it is sufficient to compose all curves in the former with an isometry taking @xmath8 to @xmath175 .
the argument is similar for all four classes , so we consider only the case where @xmath1 overlaps @xmath2 $ ] .
firstly , notice that composition with an orientation - reversing isometry switches the sign of the curvature of a curve .
therefore , by applying a reflection in some geodesic if necessary , it can be assumed that @xmath176 ( instead of @xmath177 ) .
write @xmath178 , @xmath179 and @xmath180
. then normal translation by @xmath181 ( i.e. , the map @xmath182 ) provides a homeomorphism @xmath183 where @xmath184 is obtained by parallel translating @xmath8 by a distance @xmath185 along the ray emanating from @xmath186 and similarly for @xmath187 .
the inverse of this map is simply normal translation by @xmath188 .
all details requiring verification here were already dealt with in . finally , post - composition of curves with the unique orientation - preserving hyperbolic isometry taking @xmath184 to @xmath189 yields a homeomorphism @xmath190 in all cases , the homeomorphism used to prove operates on the curves in a given space by a composition of a normal translation and an isometry .
it is `` canonical '' in the sense that these two transformations , as well as the values of @xmath160 and @xmath172 , are uniquely determined by @xmath191 and @xmath192 .
however , this does not preclude the existence of some more complicated homeomorphism between spaces in the second column of table [ ta : reduction ] .
[ x : reduction ] let @xmath157 and @xmath193 be given . use the argument above to prove : 1 .
the remaining three cases listed in table [ ta : reduction ] .
if @xmath1 contains @xmath2 $ ] , then there exist @xmath194 and @xmath195 such that @xmath196 3 .
if @xmath197 , then there exist @xmath198 and @xmath195 such that @xmath199 note that the hypothesis here is more restrictive than in the first case of the table .
the purpose of this section is to discuss which of the canonical subspaces of @xmath200 are empty . in the sequel @xmath201 ( resp .
@xmath202 ) denotes the space obtained from by omitting the restriction that @xmath203 .
[ l : attainable ] if at least one @xmath204 , then @xmath205 for all @xmath206 . by symmetry
, we may assume that @xmath207 .
further , using normal translation by @xmath208 , we reduce to the case where @xmath209 .
let @xmath210 and @xmath211 be arbitrary . consider the map @xmath212 it is not hard to see that this map is open ; cf . .
we begin by proving that @xmath213 . since sufficiently tight circles are allowed , @xmath214 .
let @xmath215 be an open interval about @xmath216 . for each @xmath217 ,
let @xmath218 denote the elliptic isometry fixing @xmath126 and taking @xmath216 to @xmath50 .
let @xmath219 then the concatenation @xmath220 starts at @xmath216 and ends at @xmath221 . this curve may not be smooth but it may be approximated by a smooth curve in @xmath222 .
this shows that @xmath223 contains the interval @xmath224 about @xmath225 whenever it contains @xmath225 . in turn , this implies that @xmath226 includes @xmath227 .
therefore , if @xmath228 contains any tangent vector @xmath216 based at @xmath126 , it must also contain @xmath229 , by transitivity .
now let @xmath230 in case @xmath231 and @xmath232 otherwise .
if @xmath228 contains tangent vectors at @xmath126 , then it also contains tangent vectors at @xmath233 for any @xmath234 , since @xmath233 can be reached from @xmath126 by the half - circle centered at the midpoint of the geodesic segment @xmath235 of radius @xmath236 .
it follows that @xmath237 .
the converse of is an immediate consequence of the following result ; cf .
also [ c : closed ] .
[ l : unattainable ] let @xmath238 $ ] and @xmath239 .
the elcrics ( or horocycles ) of curvature @xmath240 and @xmath241 tangent to @xmath8 divide @xmath242 into four open regions .
any curve in @xmath243 is confined to one of these regions @xmath244 .
conversely , a point in @xmath244 can be reached by a curve of constant curvature in @xmath245 .
the elcric of curvature @xmath246 tangent to @xmath8 meets @xmath58 at two points ; let @xmath244 denote the open region whose closure contains the one point in @xmath58 towards which @xmath8 is pointing .
let @xmath247 .
comparison of the curvatures of @xmath19 and those of the elcrics bounding @xmath244 shows that @xmath248 for sufficiently small @xmath249 .
suppose for a contradiction that @xmath19 reaches @xmath250 for the first time at @xmath251 $ ] at some point of the elcric @xmath252 of curvature @xmath253 .
( in case @xmath254 , @xmath252 is a horocycle and @xmath255 . )
the function @xmath256 \to \r , \quad \de(t)=d(\ga(t),e)\ ] ] vanishes at @xmath257 and @xmath258 , hence it must attain its global maximum in @xmath259 $ ] at some @xmath260 , say @xmath261 . by , the normal translation of @xmath252 by @xmath262 is another elcric ( horocycle ) , of curvature @xmath263 , which must be tangent to @xmath19 at @xmath264 because @xmath265 . since @xmath266 has a local maximum at @xmath27 , comparison of curvatures yields that @xmath267 , which is impossible .
the last assertion of the lemma holds because constant - curvature curves in @xmath245 foliate @xmath244 .
[ x : rbar ] with the notation of , the image of any curve in @xmath268 is contained in @xmath269 ( cf . ) .
[ c : closed ] the spaces @xmath270 and @xmath271 contain closed curves if and only if at least one @xmath272 . [ c : nontrivial ] let @xmath0 be a hyperbolic surface . then any closed curve
whose curvature is bounded by 1 in absolute value is homotopically non - trivial ( that is , it can not represent the unit element of @xmath273 ) .
indeed , the previous corollary shows that the lift of such a curve to @xmath4 can not be closed
. it will be convenient to introduce another model for @xmath5 , which bears some similarity to mercator world maps .
[ d : mercator ] the underlying set of the @xmath46 is @xmath274 .
its metric is defined by declaring the correspondence @xmath275 to be an isometry . because this map is the composition of the complex exponential with a reflection in the line @xmath276 , @xmath46 is conformal . [
x : compm ] in the mercator model @xmath46 : 1 .
vertical lines @xmath277 , or parallels , represent elcrics of curvature @xmath278 .
horizontal segments , or meridians , are geodesics corresponding in @xmath62 to euclidean half - circles centered at 0 .
vertical translations are hyperbolic isometries .
the riemannian metric @xmath279 is given by @xmath280 3 .
the christoffel symbols are given by @xmath281 the following basic fact will be very important .
[ l : graph ] let @xmath282 $ ] and @xmath8 be the vector @xmath283 based at @xmath284 . then the image of any curve in @xmath285 or @xmath286 is the graph of a function of @xmath216 when represented in @xmath46 .
conversely , if @xmath287 $ ] , then there exist curves in these spaces which are not graphs .
suppose that @xmath288 is not the graph of function of @xmath216 when represented in @xmath46 , i.e. , it is tangent to a parallel @xmath289 at some time , say , @xmath290 .
let @xmath291 be the parallel @xmath292 , which is orthogonal to @xmath19 at @xmath293 by hypothesis ; note that @xmath294 is a geodesic .
let @xmath67 be the meridian through @xmath295 , which is orthogonal to both @xmath291 and @xmath289 .
let @xmath296 be the concatenation of @xmath19 and its reflection in @xmath67 ( with reversed orientation ) .
let @xmath297 be the concatenation of @xmath296 and its reflection in @xmath291 ( again with reversed orientation ) .
then @xmath298 is a closed curve , hence @xmath299 $ ] by [ c : closed ] .
conversely , if @xmath299 $ ] then @xmath300 contains circles , which are closed and hence not graphs .
[ p : unattainable ] consider the decomposition of @xmath7 into canonical subspaces . 1 .
if @xmath1 contains @xmath2 $ ] , then all of its canonical subspaces are nonempty .
if @xmath1 is contained in @xmath2 $ ] , then at most one canonical subspace is nonempty .
3 . if @xmath1 overlaps or is disjoint from @xmath2 $ ] , then
infinitely many of the canonical subspaces are nonempty , and infinitely many are empty .
we split the proof into parts . 1 . by , @xmath301 .
since @xmath302 $ ] , we may concatenate a curve in this space with circles of positive or negative curvature , traversed multiple times , to attain any desired total turning .
, we may assume that @xmath8 is the vector in the statement of .
then the assertion becomes an immediate consequence of the latter .
@xmath303 . because we may concatenate any curve in @xmath304 with a circle ( of positive curvature if @xmath305 and of negative curvature if @xmath306 ) traversed multiple times , @xmath307 for infinitely many values of @xmath27 .
the proof of the remaining assertion , that @xmath308 for infinitely many values of @xmath27 , is worked out in below .
[ d : arcs ] given a regular curve @xmath17 \to \hh^2 $ ] , define maps
@xmath309 \to \ss^1_\infty\ ] ] by letting @xmath310 be the point where the geodesic ray emanating from @xmath311 meets @xmath57 .
the images of @xmath312 shall be denoted by @xmath313 .
[ x : arcs ] let @xmath17 \to \hh^2 $ ] be a regular curve . 1 .
each of @xmath314 is a closed arc , which may be a singleton or all of @xmath57 .
regard @xmath19 as a curve in @xmath62 , and let @xmath108 be the unique point of @xmath57 not on the real line in this model .
then @xmath108 lies in the complement of @xmath315 if and only if @xmath316 is never horizontal .
3 . let @xmath103 be a horocycle of curvature @xmath112 , tangent to @xmath57 at @xmath225 .
then @xmath317 while @xmath318 . what happens if the curvature of @xmath103 is @xmath122 ?
[ x : turnings ] let @xmath156 be fixed . 1 . two curves @xmath319 lie in the same component of this space if and only if the associated maps @xmath320
\to \ss^1_\infty $ ] defined in have the same total turning
. 2 . if @xmath321 everywhere , then @xmath322 is monotone .
similarly , if @xmath323 , then @xmath324 is monotone .
3 . use these facts to show that if @xmath1 overlaps or is disjoint from @xmath2 $ ] and @xmath325 , then then there exists @xmath326 such that @xmath327 is empty for all @xmath328 and nonempty for all @xmath43 .
the group @xmath329 of all orientation - preserving isometries of the hyperbolic plane acts simply transitively on @xmath330 .
therefore , an element @xmath331 of this group is uniquely determined by where it maps a fixed unit tangent vector @xmath332 .
this yields a correspondence between the two sets , viz . , @xmath333 . the @xmath334\to \iso_+(\hh^2)\ ] ] of a regular curve @xmath68\to
\hh^2 $ ] is the image of @xmath335 under this correspondence , and the @xmath336\to \textrm{l}(\iso(\hh^2 ) ) $ ] is its infinitesimal version .
more precisely , @xmath337 is the translation of @xmath338 to the lie algebra @xmath339 , defined by @xmath340).\ ] ] here @xmath341 denotes the derivative ( at the identity ) of left multiplication by @xmath342 .
although @xmath343 depends on the choice of @xmath332 , @xmath337 does not . to be explicit ,
let @xmath344 , @xmath345 be the group of isometries of @xmath65 and @xmath346 be the connected component of the identity , which is isomorphic to @xmath347 .
the corresponding lie algebra is @xmath348 in the hyperboloid model @xmath67 , the identification between @xmath349 and @xmath329 takes the canonical form @xmath350 where @xmath351 denotes the lorentzian vector product in @xmath65 .
( which @xmath352 yields this correspondence ? ) [ x : loid ] let @xmath68\to l $ ] be smooth and regular . 1 .
denoting differentiation with respect to the given parameter ( resp .
arc - length ) by @xmath353 ( resp .
@xmath354 ) : @xmath355 in these formulae @xmath356 are viewed as taking values in @xmath65 , @xmath78 is the lorentzian inner product and @xmath357 the corresponding quadratic form . 2 .
the frame @xmath358\to \so^+_{2,1 } $ ] and logarithmic derivative @xmath359\to \aso_{2,1 } $ ] of @xmath19 are given by @xmath360 similar formulas for the curvature in the disk and half - plane models are much more cumbersome .
however , the frame and logarithmic derivative do admit comparatively simple expressions . for concreteness ,
an element @xmath361 is identified with @xmath362 , that is , with the image under the complex derivative @xmath363 of the tangent vector @xmath364 based at @xmath365 .
similarly , we choose the correspondence between @xmath366 and @xmath367 to be @xmath368 . recall that @xmath366 consists of those mbius transformations of the form @xmath369 [ x : phila ] denote elements of projective groups as matrices in square brackets , the absolute value of a complex number by @xmath370 and , exceptionally , the norm of a vector tangent to @xmath5 by @xmath371 . 1 .
let @xmath68 \to h $ ] be a smooth regular curve . then @xmath372\to \asl_2({\r } )
$ ] is given by @xmath373 and @xmath374\to \psl_2(\r ) $ ] is given by @xmath375 2 .
let @xmath68 \to d $ ] be a smooth regular curve .
then @xmath376 \to \asl_2(\c ) $ ] is given by @xmath377 and @xmath358 \to \iso_+(d)\subs\psl_2(\c ) $ ] is given by @xmath378 a one - parameter group of hyperbolic isometries provides a foliation of @xmath4 by its orbits , which are elcrics by ( d ) .
a regular curve @xmath17\to \hh^2 $ ] if there exist some such group @xmath379 and @xmath380 $ ] such that @xmath381 and @xmath382 are tangent to orbits of @xmath379 .
[ x : grafting ] a regular curve @xmath68\to h $ ] admits hyperbolic grafting if and only if there exist @xmath380 $ ] such that @xmath383 has the form @xmath384 for some @xmath385 and @xmath386 . +
[ x : convex ] in the disk and half - plane models , a curve whose ( hyperbolic ) curvature is everywhere greater than @xmath112 is locally convex from the euclidean viewpoint , i.e. , its total turning is a strictly increasing function . [
p : disjoint ] if @xmath1 is disjoint from @xmath2 $ ] , then each canonical subspace of @xmath29 is either empty or contractible .
we will work in the half - plane model @xmath62 throughout the proof .
points and tangent vectors will thus be regarded as elements of @xmath24 . by
, it may be assumed that @xmath387 and that @xmath8 is parallel to @xmath283 .
let @xmath388 be a fixed valid total turning for curves in @xmath7 , meaning that @xmath389 is parallel to @xmath390 .
given @xmath391 , let @xmath392 \to [ 0,\tau ] $ ] be the unique continuous function such that @xmath393 and @xmath394 $ , } \ ] ] where @xmath370 denotes the usual absolute value of complex numbers . by , @xmath395 is a diffeomorphism of @xmath89 $ ] onto @xmath396 $ ] .
thus it may be used as a parameter for @xmath19 .
now fix @xmath397 and set @xmath398 , both viewed as curves in @xmath62 and parametrized by the argument @xmath399 $ ] , as above .
define @xmath400,~\theta \in [ 0,\tau]).\ ] ] then each @xmath401 \to h $ ] is a smooth curve satisfying @xmath402 and @xmath403 .
moreover , it has the correct total turning @xmath27 , since @xmath404 is always parallel to @xmath405 . by and the definition of osculation , the osculating constant - curvature curves to @xmath19 from the euclidean and hyperbolic viewpoints at any @xmath406 agree . since @xmath407 , they are both equal to a certain circle completely contained in @xmath62 , traversed counterclockwise ; compare ( b ) .
therefore the osculating constant - curvature curve to @xmath408 at @xmath406 is another circle @xmath409 , the corresponding convex combination of the osculating circles @xmath410 to @xmath411 and @xmath412 to @xmath296 at @xmath406 ( see ) .
let @xmath413 denote the @xmath50-coordinate of the euclidean center of @xmath409 and @xmath414 its euclidean radius ( @xmath415 $ ] ) .
more explicitly , @xmath416 the diameter @xmath417 of @xmath409 is given by @xmath418 note that @xmath419 , as this is true for @xmath420 . because @xmath421 by hypothesis , a trivial computation shows that @xmath417 satisfies the same inequalities for all @xmath422 $ ] .
hence the curvature of @xmath408 does indeed take values inside @xmath1 , and @xmath423 defines a contraction of @xmath327 . [ x : disk ] does the proof of also work in the disk model ?
[ x : disjoint ] use the argument in the proof of to show that each component of @xmath424 is contractible if @xmath425 $ ] is disjoint from @xmath2 $ ] . [ c : disjoint ] if @xmath1 is disjoint from @xmath2 $ ] , then @xmath161 is homeomorphic to the disjoint union of countably many copies of the separable hilbert space .
this is an immediate consequence of ( c ) , and .
in this section we will work with the spaces @xmath426 which are introduced in [ s : discontinuous ] .
these are larger than @xmath7 in that they include regular piecewise @xmath427 curves .
our proof of uses such curves and is therefore more natural in the former class .
we recommend that the reader ignore the technical details for the moment and postpone a careful reading of [ s : discontinuous ] .
there is also a way to carry out the proof below in the setting of @xmath7 : whenever a discontinuity of the curvature arises , take an approximation by a smooth curve ( which needs to be constructed ) ; this path is more elementary but very cumbersome .
the reader will notice that similar discussions appear in @xcite , @xcite and @xcite .
recall the mercator model @xmath46 defined in .
suppose that @xmath428 is a smooth regular curve in @xmath46 which can be written in the form @xmath429
\to m,\quad \ga(x ) = ( x , y(x))\ ] ] for some @xmath430 \subs ( 0,\pi ) $ ] ; in other words , the image of @xmath19 is the graph of a function @xmath431 .
a straightforward computation using shows that the curvature of @xmath19 is then given by @xmath432 more important than this expression itself is the observation that it does not involve @xmath50 , only its derivatives ( because vertical translations are isometries of @xmath433 ) .
this can be exploited to express geometric properties of @xmath434 solely in terms of the function @xmath435 , and in particular to prove the following .
[ p : contained ] if @xmath1 is contained in @xmath2 $ ] , then @xmath436 is either empty or contractible , hence homeomorphic to the separable hilbert space . by , it suffices to prove that the banach manifold @xmath437 is either empty or weakly contractible . by
, we can assume that @xmath8 is represented in @xmath46 as the vector @xmath283 based at @xmath438 .
let @xmath439 be a continuous map . by
, it can be assumed that each @xmath440 is smooth . in particular , it is possible to choose @xmath441 and @xmath442 such that the curvatures of all the @xmath443 take values inside @xmath444 . by
, any such curve @xmath19 may be parametrized as in .
the function @xmath445 satisfies the following conditions , whose meaning will be explained below : 1 .
@xmath446 and @xmath447 .
2 . @xmath448 .
3 . @xmath449 for a.e .
@xmath450 $ ] , where @xmath451 \times \r \to \r $ ] is given by : @xmath452,~ x\in [ a , b],~z\in \r).\ ] ] in the present case , @xmath453 is the @xmath216-coordinate of @xmath454 and @xmath455 is the @xmath216-coordinate of @xmath456 .
the real numbers @xmath457 and @xmath458 in ( i ) are the slopes of @xmath8 and @xmath9 regarded as vectors in @xmath24 . condition ( ii ) prescribes the @xmath50-coordinate of @xmath459 .
finally , the inequalities in ( iii ) express the fact that the curvature of @xmath19 takes values in @xmath460 $ ] ; compare .
conversely , suppose that an absolutely continuous function @xmath461 \to \r $ ] satisfies ( i)(iii ) .
if we set @xmath462 and define @xmath19 through , then @xmath463 . thus one can produce a contraction of the original family of curves by constructing a homotopy @xmath464 of the corresponding family of functions @xmath465 , subject to the stated conditions throughout , with @xmath466 independent of @xmath467 .
this is what we shall now do . for each @xmath468 $ ] , let @xmath469 be the solution of the initial value problem @xmath470 where @xmath471 and @xmath472 are as in conditions ( i ) and ( iii )
geometrically , the graph of @xmath279 is an arc of the elcric of curvature @xmath473 with initial unit tangent vector @xmath8 , represented in @xmath46 . in particular , @xmath469 is defined over all of @xmath430 $ ] by .
similarly , for each @xmath468 $ ] , let @xmath474 be the solution of the initial value problem @xmath475 geometrically , the graph of @xmath476 is an arc of the elcric of curvature @xmath473 whose final unit tangent vector is @xmath9 .
although @xmath474 is smooth , it is possible that it is not defined over all of @xmath430 $ ] ; if its maximal domain of definition is @xmath477 $ ] for some @xmath478 , then we extend @xmath476 to a function @xmath430 \to \r \cup \se{\pm \infty } $ ] by setting it equal to @xmath479 over @xmath480 $ ] . using this geometric interpretation
, it follows from that @xmath481 moreover , from the same result one deduces that @xmath482 for each @xmath483 $ ] , define @xmath484\to \r$ ] by @xmath485 notice first that this function does not take on infinite values , since @xmath486 are real functions .
similarly , since three of the functions above ( namely , @xmath487 ) take the value @xmath488 at @xmath489 , and three of them ( namely , @xmath490 ) take the value @xmath458 at @xmath455 , @xmath491 automatically satisfies condition ( i ) .
the reader is asked to show in that it is lipschitz and satisfies ( iii ) as well .
it remains to choose @xmath492 appropriately to ensure that it satisfies condition ( ii ) .
let @xmath493}{f(x ) \leq g_{\ka}(x ) \text { for all $ x \in [ a , b ] $ } } , \\
\ka_- = \max & \set{\ka \in [ \bar \ka_1,\bar\ka_2]}{g_{\ka}(x ) \leq f(x ) \text { for all $ x \in [ a , b ] $ } } , \\
\de=&\set{(\la,\mu)\in [ \ka_-,\ka_+]}{\la \leq \mu } \text{\quad ( cf.~\fref{f : delta})}. \end{aligned}\ ] ] in the @xmath494-plane .
the dashed segments are the intersections of @xmath495 with the lines @xmath496 @xmath497 ) $ ] . ] define @xmath498 to be the area under the graph of @xmath499 : @xmath500 continuous dependence of the solutions of on @xmath34 implies that @xmath501 is a continuous function of @xmath492 .
moreover , from @xmath502 one deduces that @xmath503 for all @xmath504 $ ] . consequently , because the integral of @xmath467 equals @xmath505 , for each @xmath506 $ ] the set @xmath507 is nonempty ( see ) .
in fact , it consists of a unique point @xmath508 . to establish this
, it suffices to show that @xmath509 is a strictly increasing function of both @xmath510 and @xmath511 .
now if @xmath512 then the set of all @xmath450 $ ] for which @xmath513 is nonempty , by and the choice of @xmath514 .
therefore @xmath515 holds over a set of positive measure , while the nonstrict inequality holds everywhere by .
this proves strict monotonicity with respect to @xmath516 ; the argument for @xmath511 is analogous .
continuity of @xmath517 implies continuity of the curve @xmath518 ( which is depicted in bold in ) .
the functions @xmath519\to
\r,\quad f_s = f_{(\la(s),\mu(s))}\ ] ] satisfy all of conditions ( i)(iii ) by construction , and they depend continuously on @xmath467 and @xmath520 .
let @xmath521 ; then @xmath522 by , there is at most one value of @xmath523 $ ] for which the integral of @xmath524 equals @xmath505 .
this implies that @xmath525 , and hence @xmath466 , is independent of @xmath467 .
therefore @xmath526 is indeed a contraction .
[ x : median ] let @xmath527 be the median of @xmath528 \to \r$ ] . 1 .
if each @xmath529 is @xmath530-lipschitz , then so is @xmath527 . in particular
, @xmath527 is absolutely continuous .
if @xmath527 and each @xmath529 are differentiable at @xmath216 , then @xmath531 for some @xmath532 such that @xmath533 .
in particular , if each @xmath529 satisfies the inequalities in ( iii ) ( with @xmath529 in place of @xmath467 ) , then so does @xmath527 .
3 . the function @xmath491 of does not immediately conform to this situation because @xmath534 @xmath535 may take on infinite values .
nevertheless , by subdividing @xmath430 $ ] into at most three subintervals ( where none , one or both of @xmath534 are infinite ) , show that it is lipschitz and satisfies ( iii ) .
in this section we consider spaces of curves with constrained curvature on an arbitrary surface , not necessarily hyperbolic nor orientable .
we study their behavior under covering maps and show that they are always nonempty if @xmath536 is compact ; this should be contrasted with . a is a smooth riemannian 2-manifold . given
a regular curve @xmath537 \to s $ ] , its @xmath538 is the lift of @xmath19 to the unit tangent bundle @xmath539 of @xmath540 : @xmath541\to uts , \quad \ta(t)=\frac{\dot\ga(t)}{\abs{\dot\ga(t)}}.\ ] ] now let an orientation of @xmath542 be fixed .
the to @xmath543 is the map @xmath544\to uts$ ] determined by the requirement that @xmath545 is an orthonormal basis of @xmath546 whose parallel translation to @xmath547 ( along the inverse of @xmath19 ) is positively oriented , for each @xmath548 $ ] .
assuming @xmath543 is twice differentiable , its @xmath549 is given by @xmath550 here @xmath551 denotes covariant differentiation ( along @xmath543 ) .
[ r : alternative ] if @xmath0 is nonorientable , it is more common to define the ( unsigned ) curvature of @xmath17 \to s $ ] by @xmath552 if @xmath0 is orientable , the usual definition coincides with , but @xmath95 is defined by the condition that @xmath553 be positively oriented with respect to a specified orientation of @xmath0 , rather than the parallel translation of an orientation of @xmath542 .
these two definitions are equivalent , since an orientation of @xmath542 determines an orientation of @xmath0 if the latter is orientable .
the definition that we have chosen has the advantage of allowing arbitrary bounds for the curvature of a curve on a nonorientable surface , and in particular the concise formulation of below .
a geometric interpretation for the curvature is the following : let @xmath554\to uts $ ] be any smooth parallel vector field along @xmath19 , and let @xmath18\to \r $ ] be a function measuring the oriented angle from @xmath555 to @xmath556 .
then a trivial computation shows that @xmath557 .
in particular , the @xmath558 of @xmath19 equals @xmath559 . in all that follows ,
the curvature bounds @xmath560 are allowed to take values in @xmath561 , @xmath0 denotes a surface and @xmath562 .
moreover , it is assumed that an orientation of @xmath563 , where @xmath125 is the basepoint of @xmath8 , has been fixed .
[ d : curve ] define @xmath564 to be the set , endowed with the @xmath565 topology ( for some @xmath566 ) of all @xmath565 regular curves @xmath567\to s$ ] such that : 1 . @xmath568 and @xmath569 ; 2 .
@xmath570 for each @xmath571 $ ] . of course , whether @xmath0 is orientable or not , the topological properties ( or even the voidness ) of @xmath572 may be sensitive to the choice of orientation for @xmath563 .
it will follow from ( [ l : c^2 ] ) that the value of @xmath573 is irrelevant in the sense that different choices of @xmath573 yield spaces which are homeomorphic .
because of this , @xmath574 is denoted simply by @xmath575 in this section .
[ x : contractible ] define @xmath576 as in , except that no condition is imposed on @xmath577 , and similarly for @xmath578 and @xmath579 .
then : 1 .
@xmath576 and @xmath580 are contractible .
@xmath579 is homotopy equivalent to @xmath581 .
[ l : covering ] let @xmath582 be a riemannian covering ( a covering map which is also a local isometry ) and @xmath583 .
suppose that @xmath584 preserves the chosen orientation of these tangent planes .
then @xmath585 yields a homeomorphism @xmath586 let @xmath587 and @xmath588 \to \te{s } $ ] be its lift to @xmath589 starting at @xmath590 . since @xmath591 is an isometry , @xmath592 moreover , @xmath593 by hypothesis , hence @xmath594 by continuity . now by , @xmath595 .
therefore @xmath596 for some lift @xmath597 of @xmath9 .
conversely , if @xmath598 , then @xmath599 by the same reasons . since projection and lift ( starting at @xmath590 )
are inverse operations , the asserted homeomorphism holds .
this result is especially useful when @xmath0 is a space form ( e.g. , a hyperbolic surface ) ; for then @xmath0 is the quotient by a discrete group of isometries of the model simply - connected space of the same curvature , which is much more familiar .
the lemma also furnishes a reduction to the orientable case by taking @xmath589 to be the two - sheeted orientation covering of @xmath0 .
[ x : symmetric ] suppose that @xmath1 is symmetric about 0 .
then the conclusion of holds regardless of whether @xmath591 preserves orientation at @xmath600 .
if closed or half - open intervals were used instead in , then substantial differences would only arise in marginal cases .
for instance , in the situation of , if @xmath9 is tangent to @xmath250 , then @xmath7 is empty , while @xmath82 may not be .
the original definition is more convenient to work with since the resulting spaces are banach manifolds ; compare example 1.1 in @xcite .
[ p : compact ] let @xmath540 be a compact connected surface .
then @xmath601 for any choice of @xmath602 and @xmath603 . by passing to the orientation covering if necessary ,
it may be assumed that @xmath0 is oriented .
let @xmath602 be fixed . for @xmath604 , write @xmath605 if @xmath606 .
notice that @xmath607 is transitive : given curves in @xmath572 and @xmath608 , their concatenation starts at @xmath8 and ends at @xmath609 ; its curvature may fail to exist at the point of concatenation , but this can be fixed by a convolution .
let @xmath610 it is clear from the definition that @xmath611 for all @xmath612 , and that @xmath613 and @xmath614 are open subsets of @xmath581 ( cf . ) .
moreover , the family @xmath615 covers @xmath539 .
indeed , given @xmath616 , we can find @xmath612 such that @xmath617 . reversing the orientation of a curve in the latter set
, we establish that @xmath618 , that is , @xmath619 . since @xmath539 is compact , it can be covered by finitely many of the @xmath613 .
let @xmath620 be a minimal cover .
we claim that @xmath621 .
assume that @xmath622 .
if @xmath623 then @xmath624 , and therefore by minimality @xmath625 .
since @xmath626 and @xmath627 for @xmath628 , we deduce that @xmath629 for each @xmath630 .
the open sets @xmath631 are thus nonempty and disjoint . on the other hand ,
every @xmath632 must intersect some @xmath633 with @xmath634 , as @xmath539 is connected .
choose @xmath635 such that @xmath636 .
it is easy to see that @xmath637 must be disjoint from @xmath631 .
thus , if @xmath638 is the interior of @xmath639 , then @xmath640 . we will obtain a contradiction from this . by definition ,
there exist @xmath641 , @xmath642 and @xmath643 .
let @xmath644\to s$ ] be parametrized by arc - length and @xmath645\to ( \ka_1,\ka_2)$ ] denote its curvature .
the tangent bundle @xmath646 has a natural volume form coming from the riemannian structure of @xmath540 . a theorem of liouville states that the geodesic flow preserves volume in @xmath646 .
since @xmath540 is oriented , given @xmath647 we may define a volume - preserving bundle automorphism on @xmath581 by @xmath648 ( where @xmath649 is `` multiplication by @xmath650 '' ) .
let @xmath651 and @xmath652 be the vector fields on @xmath581 corresponding to the geodesic flow and to counterclockwise rotation , respectively . then for any @xmath653 , the vector field @xmath654 defines a volume - preserving flow on @xmath539 ; the projections of its orbits on @xmath540 are curves parametrized by arc - length of constant curvature @xmath655 . by definition ,
the open set @xmath638 is forward - invariant under each of these flows for @xmath656 .
define a map @xmath657 as follows : given @xmath658 , @xmath659 , where @xmath660\to s$ ] is the unique curve in @xmath661 , parametrized by arc - length , whose curvature is @xmath662
. then @xmath663 must be volume - preserving , @xmath664 , but there exists a neighborhood of @xmath665 contained in @xmath631 which is taken by @xmath663 to a neighborhood of @xmath666 contained in @xmath638 , a contradiction .
we conclude that @xmath621 , so that @xmath667 .
furthermore , @xmath668 must have empty interior by the preceding argument .
hence @xmath669 is a dense open set in @xmath539 .
let @xmath603 be given .
since @xmath613 is open , there exists @xmath670 . then @xmath671 . since @xmath672 and @xmath673
, we deduce that @xmath605 ; in other words , @xmath606 . just as in algebraic topology
one considers homotopies with and without basepoint conditions , one can also study the space of all smooth closed curves on @xmath0 with curvature in an interval @xmath10 , but with no restrictions on the initial and final unit tangents .
let this space be denoted by @xmath674 . in some regards this class may seem more fundamental than its basepointed version , the class of spaces @xmath572 considered thus far .
however , in the hirsch - smale theory of immersions , such basepoint conditions arise naturally .
for instance , thm .
c of @xcite states that the space @xmath675 is ( weakly ) homotopy equivalent to the loop space @xmath676 consisting of all loops in @xmath581 based at @xmath8 .
moreover , even if one is interested only in curves , it is often helpful to study @xmath677 by lifting its elements to the universal cover of @xmath0 , and these lifts need not be closed . a further point is provided by the following result .
recall that @xmath581 is diffeomorphic to @xmath678 if @xmath679 or @xmath5 , and to @xmath680 if @xmath681 .
[ l : closed ] let @xmath0 be a simply - connected complete surface of constant curvature and @xmath682 be arbitrary
. then @xmath677 is homeomorphic to @xmath683 .
the group of orientation - preserving isometries of such a surface acts simply transitively on @xmath581 .
given @xmath684 , let @xmath685 denote its unique element mapping @xmath9 to @xmath8 . define @xmath686 this is clearly continuous , and so is its inverse , which is given by @xmath687 .
regard the elements of @xmath677 as maps @xmath688 .
there is a natural projection @xmath689 taking a curve @xmath690 to its unit tangent at @xmath283 .
this is a fiber bundle if @xmath30 and @xmath31 since @xmath0 is locally diffeomorphic to @xmath691 and the group of diffeomorphisms of the latter acts transitively on @xmath692 .
it may be a fibration in certain other special cases , as in the situation of , but in general this can not be guaranteed .
for instance , if @xmath693 is a flat torus , then the homotopy type of the fibers @xmath694 of the map @xmath695 is not locally constant ; this follows from an example in the introduction of @xcite .
we believe that little is known about the topology of @xmath696 beyond what is implied by .
suppose that @xmath697\to s$ ] is a smooth regular curve and , as always , @xmath698 has been oriented .
let @xmath699\to \r^+$ ] denote its speed @xmath700 and @xmath701 its curvature .
then @xmath543 and @xmath702\to uts$ ] satisfy : @xmath703 thus , @xmath543 is uniquely determined by @xmath8 and the functions @xmath704 .
one can define a new class of spaces by relaxing the conditions that @xmath705 and @xmath701 be smooth .
let @xmath706 be the diffeomorphism @xmath707 for each pair @xmath708 , let @xmath709 be the diffeomorphism @xmath710 and , similarly , set @xmath711 notice that all of these functions are monotone increasing , hence so are their inverses .
moreover , if @xmath712 $ ] , then @xmath713 $ ] as well .
this is obvious if @xmath714 is bounded , and if one of @xmath715 is infinite , then it is a consequence of the fact that @xmath716 diverges linearly to @xmath717 with respect to @xmath718 . in what follows , @xmath719 denotes the separable hilbert space @xmath720\times l^2[0,1]$ ] .
a curve @xmath697\to s$ ] is @xmath714- if there exists @xmath721 such that @xmath543 satisfies with @xmath722 when it is not important to keep track of the bounds @xmath715 , we will simply say that @xmath543 is .
the system has a unique solution for any @xmath723 and @xmath658 . to see this , we use coordinate charts for @xmath724 derived from charts for @xmath540 and apply thm
. c.3 on p. 386 of @xcite to the resulting differential equation , noticing that @xmath540 is smooth and @xmath725\subs l^1[0,1]$ ] . furthermore ,
if we assume that @xmath0 is complete , then the solution is defined over all of @xmath726 $ ] . the resulting maps @xmath727\to s$ ] and @xmath728\to ts$ ] are absolutely continuous ( see p. 385 of @xcite ) , and so is @xmath95 . using that @xmath729 and differentiating , we obtain , in addition to , that @xmath730$.}\ ] ] therefore , @xmath731 , @xmath732 and @xmath733 . it is thus natural to call @xmath734 and @xmath701
the and of @xmath543 , even though @xmath735 $ ] .
although @xmath736 is , in general , defined only almost everywhere on @xmath726 $ ] , if we reparametrize @xmath543 by arc - length then it becomes a regular curve , because @xmath737 is continuous .
it is helpful to regard admissible curves simply as regular curves whose curvatures are defined a.e .. [ d : loose ] for @xmath658 , let @xmath738 be the set of all @xmath714-admissible curves @xmath697\to s$ ] with @xmath568 .
if @xmath0 is complete , then this set is identified with @xmath719 via the correspondence @xmath739 , thus furnishing @xmath740 with a trivial hilbert manifold structure .
if @xmath0 is not complete , then @xmath741 is some mysterious open subset of @xmath742 .
however , we still have the following . [
l : mysterious ] for all @xmath682 , @xmath741 is homeomorphic to @xmath742 . by ( a ) , it suffices to prove that this space is weakly contractible .
let @xmath743 be a continuous map , where @xmath744 is a compact space . by a preliminary homotopy
, each @xmath745 may be reparametrized with constant speed .
let @xmath746 and @xmath747 . the curves can be shrunk through the homotopy @xmath748\?t\big ) \quad ( s,\,t\in [ 0,1],~p \in k)\ ] ] so that all @xmath749 have length @xmath511 , which can be chosen smaller than the injectivity radius of @xmath0 at the basepoint of @xmath8 .
then each @xmath750 is determined solely by its curvature , and conversely any @xmath751-function @xmath752 \to \r $ ] determines a unique curve of constant speed @xmath511 in @xmath0 having @xmath8 for its initial unit tangent vector , via . but the space of all such functions is convex . [ l : submersion ] let @xmath0 be a surface and define @xmath753 by @xmath754 . then @xmath755 is a submersion , and consequently an open map .
the proof when @xmath679 is given in @xcite , lemma 1.5 .
the proof in the general case follows by considering riemannian normal coordinates , which are flat at least to second order , in a neighborhood of the basepoint of @xmath577 .
[ d : lurvespace ] define @xmath756 to be the subspace of @xmath738 consisting of all @xmath19 such that @xmath569 .
it follows from ( [ l : submersion ] ) that @xmath756 is a closed submanifold of codimension 3 in @xmath757 , provided it is not empty .
we have already seen in that such spaces may indeed be empty , but this can not occur when @xmath0 is compact ; cf . and .
if @xmath758 and @xmath759 , then we can also consider @xmath543 as a curve in @xmath760 .
however , the topology of the former space is strictly finer ( i.e. , has more open sets ) than the topology induced by the resulting inclusion .
[ l : ltolinclusion ] let @xmath758 , @xmath540 be a surface and @xmath658
. then @xmath761 defines a continuous injection @xmath762 .
the actual curves on @xmath540 corresponding to these pairs are the same , but @xmath763 is not a topological embedding unless @xmath764 and @xmath765
. it may be assumed that @xmath0 is oriented .
the curve @xmath543 corresponding to @xmath766 is obtained as the solution of with @xmath767 the curve @xmath768 corresponding to the right side of in @xmath769 is the solution of with @xmath770 by uniqueness of solutions , @xmath771 . in particular , @xmath763 is injective .
set @xmath772 .
observe that @xmath773 hence , @xmath774 is bounded over @xmath775 .
consequently , there exists @xmath776 such that @xmath777$.}\ ] ] we conclude that @xmath778 is continuous .
suppose now that @xmath779 .
no generality is lost in assuming that @xmath780 .
let @xmath781 define a sequence of @xmath782 functions @xmath783\to \r$ ] by : @xmath784 $ ; } \\ 0 & \text { otherwise . }
\end{cases}\qquad ( n\in \n^+,~t\in [ 0,1]).\ ] ] since @xmath785 is the composite of increasing functions , @xmath786 for any @xmath787 . therefore ,
@xmath788 $ ; } \\
= 0 & \text { otherwise}. \end{cases}\ ] ] hence , @xmath789 as @xmath790 increases , while @xmath791 .
we conclude that @xmath792 is not a topological embedding .
this argument may be modified to prove that @xmath793 is likewise not an embedding for any @xmath616 .
[ l : ctolinclusion ] let @xmath758 and @xmath603 .
then @xmath794 is a continuous injection , but not an embedding , for all @xmath795 .
moreover , the actual curve on @xmath540 corresponding to @xmath796 is @xmath543 itself .
the proof is very similar to that of ( [ l : ltolinclusion ] ) .
the following lemmas contain all the results on infinite - dimensional manifolds that we shall need .
[ l : hilbert ] let @xmath797 be ( infinite - dimensional ) banach manifolds .
then : 1 .
if @xmath797 are weakly homotopy equivalent , then they are in fact homeomorphic ( diffeomorphic if @xmath797 are hilbert manifolds ) .
2 . let @xmath798 and @xmath755 be separable banach spaces .
suppose @xmath799 is a bounded , injective linear map with dense image and @xmath800 is a smooth closed submanifold of finite codimension .
then @xmath801 is a smooth closed submanifold of @xmath755 and @xmath802 is a homotopy equivalence .
part ( a ) follows from thm . 15 in @xcite and cor . 3 in @xcite . part ( b ) is thm . 2 in @xcite .
[ l : net ] let @xmath719 be a separable hilbert space , @xmath803 a dense vector subspace , @xmath804 a submanifold of finite codimension and @xmath805 an open subset of @xmath806 .
then the set inclusion @xmath807 is a weak homotopy equivalence .
we shall prove the lemma when @xmath808 for some submersion @xmath809 , where @xmath638 is an open subset of @xmath719 .
this is sufficient for our purposes and the general assertion can be deduced from this by using a partition of unity subordinate to a suitable cover of @xmath806 .
let @xmath810 be a tubular neighborhood of @xmath805 in @xmath638 such that @xmath811 .
let @xmath812 be a compact simplicial complex and @xmath813 a continuous map .
we shall obtain a continuous @xmath814\times k\to u$ ] such that @xmath815 for every @xmath816 and @xmath817 .
let @xmath818 denote the @xmath763-th vector in the canonical basis for @xmath819 , @xmath820 and let @xmath821 denote the @xmath790-simplex @xmath822 $ ] .
let @xmath823\subs t$ ] be an @xmath790-simplex and @xmath824 $ ] be given by @xmath825 we shall say that @xmath823 $ ] is if @xmath826 is an embedding and @xmath827 . given @xmath828 , let @xmath829 denote the derivative of @xmath830 at @xmath831 and @xmath832 .
define @xmath833 by : @xmath834 notice that @xmath835 ( for @xmath836 and @xmath828 ) .
hence , using compactness of @xmath812 , we can find @xmath837 such that : 1 .
for any @xmath838 , @xmath839\subs t$ ] and it is neat ; 2 . if @xmath838 and @xmath840 for each @xmath763 , then @xmath841\subs t$ ] and it is neat
. let @xmath842 ( @xmath843 ) be the vertices of the triangulation of @xmath812 .
set @xmath844 and @xmath845 for each such @xmath846 , choose @xmath847 with @xmath848 .
let @xmath849,~i=1,\dots , m,~j=0,\dots , n ) .
\end{aligned}\ ] ] for any @xmath850 and @xmath851 , we have @xmath852 since @xmath853 and the @xmath854 are continuous functions , we can suppose that the triangulation of @xmath812 is so fine that @xmath855 for each @xmath851 whenever there exists a simplex having @xmath856 , @xmath857 as two of its vertices .
let @xmath816 lie in some @xmath858-simplex of this triangulation , say , @xmath859 ( where each @xmath860 and @xmath861 ) .
set @xmath862,~j=0,\dots , n).\ ] ] then @xmath863 $ ] is a neat simplex because condition ( ii ) is satisfied ( with @xmath864 ) : @xmath865 the strict inequality coming from and our hypothesis on the triangulation .
define @xmath866 as the unique element of @xmath867 $ ] ( @xmath868 $ ] ) .
observe that for any @xmath816 , @xmath869 @xmath870 $ ] ) and @xmath871 , as it is the convex combination of the @xmath872 . by reducing @xmath837 ( and refining the triangulation of @xmath812 ) if necessary , we can ensure that @xmath873 $ and $ a\in k$.}\ ] ] let @xmath874 be the associated retraction .
complete the definition of @xmath875 by setting : @xmath876$,~$a\in k$).}\ ] ] the existence of @xmath875 shows that @xmath853 is homotopic within @xmath805 to a map whose image is contained in @xmath877 .
taking @xmath878 , we conclude that the set inclusion @xmath807 induces surjective maps @xmath879 for all @xmath880 .
we now establish that the inclusion @xmath807 induces injections on all homotopy groups .
let @xmath880 , @xmath881 be continuous and suppose that the image of @xmath882 is contained in @xmath877 .
let @xmath883 be a close approximation to @xmath663 ; the existence of @xmath884 was proved above .
let @xmath885 and define @xmath886 $ } \\
g_0\big(\tfrac{a}{1-\eps}\big ) & \text{\ \
if\ \ $ \abs{a}\leq 1-\eps$ } \end{cases}\ ] ] notice that we can make @xmath887 as close as desired to @xmath663 by a suitable choice of @xmath884 and @xmath888 .
let @xmath854 be as in .
we claim that there exist continuous functions @xmath889 @xmath890 such that : 1 .
@xmath891 for all @xmath892 ; 2 . for any @xmath892 , @xmath893\subs d\cap t$ ] and
it is neat . to prove this ,
invoke condition ( ii ) above ( with @xmath894 in place of @xmath812 and @xmath663 in place of @xmath853 ) together with denseness of @xmath551 to find constant @xmath895 on open sets which cover @xmath894 , and use a partition of unity . by ( iv ) , for each @xmath892 there exist unique @xmath896 $ ] such that @xmath897 and @xmath898 we obtain thus a continuous map @xmath899 .
since @xmath900 and @xmath901 , we conclude from ( iii ) and uniqueness of the @xmath902 that @xmath903 .
therefore , @xmath904 is a nullhomotopy of @xmath785 in @xmath877 .
[ l : dense ] the subset of all smooth curves in @xmath756 is dense in the latter .
take @xmath905\times l^2[0,1]$ ] , @xmath906\times c^\infty[0,1]$ ] and @xmath805 an open subset of @xmath907 .
then it is a trivial consequence of ( [ l : net ] ) that @xmath908 if @xmath909 .
[ l : smoothie ] let @xmath910 be open , @xmath812 be a compact simplicial complex and @xmath911 a continuous map . then there exists a continuous @xmath912 such that : 1 .
@xmath913 within @xmath914 .
2 . @xmath915 is a smooth curve for all @xmath816 .
3 . all derivatives of @xmath915 with respect to @xmath718 depend continuously on @xmath816 .
thus , the map @xmath916 in induces surjections @xmath917 for all @xmath880 . parts ( i ) and ( ii ) are exactly what was established in the first part of the proof of ( [ l : net ] ) , in the special case where @xmath905\times l^2[0,1]$ ] , @xmath906\times c^\infty[0,1]$ ] , @xmath918 and @xmath919 . the image of the function @xmath920 constructed there is contained in a finite - dimensional vector subspace of @xmath551 , viz . , the one generated by all @xmath921 , so ( iii ) also holds .
[ l : c^2 ] let @xmath0 be complete . then the inclusion @xmath922 is a homotopy equivalence for any @xmath795 . consequently , @xmath923 is homeomorphic to @xmath924 .
let @xmath905\times l^2[0,1]$ ] , let @xmath925\times c^{r-2}[0,1]$ ] ( where @xmath926 $ ] denotes the set of all @xmath927 functions @xmath726\to \r$ ] , with the @xmath927 norm ) and let @xmath928 be set inclusion . setting @xmath929 , we conclude from ( [ l : hilbert](b ) ) that @xmath930 is a homotopy equivalence .
we claim that @xmath931 is homeomorphic to @xmath932 , where the homeomorphism is obtained by associating a pair @xmath933 to the curve @xmath543 obtained by solving , with @xmath705 and @xmath701 as in .
suppose first that @xmath934 .
then @xmath935 ( resp .
@xmath701 ) is a function @xmath726\to \r$ ] of class @xmath936 ( resp .
@xmath937 ) .
hence , so are @xmath938 and @xmath939 , since @xmath830 and @xmath940 are smooth .
moreover , if @xmath941 are close in @xmath565 topology , then @xmath942 is @xmath937-close to @xmath943 and @xmath944 is @xmath936-close to @xmath945 .
conversely , if @xmath946 , then @xmath947 is of class @xmath936 and @xmath948 of class @xmath937
. since all functions on the right side of are of class ( at least ) @xmath937 , the solution @xmath538 to this initial value problem is of class @xmath936 .
moreover , @xmath949 , hence the velocity vector of @xmath543 is seen to be of class @xmath936 .
we conclude that @xmath543 is a curve of class @xmath565 .
further , continuous dependence on the parameters of a differential equation shows that the correspondence @xmath950 is continuous .
since @xmath543 is obtained by integrating @xmath951 , we deduce that the map @xmath952 is likewise continuous . the last assertion of the lemma follows from ( [ l : hilbert](b ) ) .
the first author is partially supported by grants from , and .
the second author gratefully acknowledges the support of and in the form of a post - doctoral fellowship . | let @xmath0 be a hyperbolic surface .
we investigate the topology of the space of all curves on @xmath0 which start and end at given points in given directions , and whose curvatures are constrained to lie in a given interval @xmath1 .
these spaces fall into four qualitatively distinct classes , according to whether @xmath1 contains , overlaps , is disjoint from , or contained in the interval @xmath2 $ ] .
their homotopy type is computed in the latter two cases .
= 1 |
neutron irradiation is certainly not practical for improving the properties of long length commercial superconductors , but a very efficient tool for benchmarking and investigating flux pinning .
it helps understanding vortex physics @xcite and limitations of the current flow @xcite .
efficient pinning centers are introduced , whose size perfectly matches the coherence length of yba@xmath1cu@xmath2o@xmath3 ( ybco ) .
they improve the critical currents , if the density of defects was originally too small , but diminish them , if the total defect concentration becomes too high .
although the resulting defect structure is not perfectly well defined , since only the density of the largest defects , the collision cascades , is known @xcite and many smaller defects of unknown density are produced @xcite , it is clearly established that all defects are randomly distributed and uncorrelated , i.e. , without a preferred orientation .
it is particularly interesting to study the interplay of these defects with the correlated disorder usually present in coated conductors or thin films @xcite . in the present study
, the effect of neutron irratiation on commercial coated conductors was investigated .
similarities and differences in differently produced tapes will be presented in the following in order to provide a comprehensive picture of neutron irradiation of this class of materials .
we do not aim at comparing the properties of the unirradiated samples , since they were supplied at different times and the rapid development during the past few years would not allow a useful comparison between the different production routes .
the different stages of optimisation turned out to be the main source of differences in the radiation response of the investigated tapes .
three different types of commercial ybco tapes were studied .
the first type of samples ( denoted as modrab in the following ) was prepared by metal organic deposition ( mod ) on a rabits template .
the second series of samples ( pldysz ) is based on an ibad ( ion beam assisted deposition ) ysz ( yttrium stabilized zirconia ) template . the superconducting layer was prepared by pulsed laser deposition ( pld ) .
the third series ( mocvdmgo ) had ibad mgo templates with the ybco layer made by metal organic chemical vapor deposition ( mocvd ) .
the modrab samples were irradiated to higher fluences ( @xmath4m@xmath5 and @xmath6m@xmath5 ) than the pldysz and mocvdmgo samples ( @xmath7m@xmath5 ) .
all samples were characterized by four probe transport prior to and after irradiation by fast neutrons .
the transport measurements were made under helium gas flow up to fields of 15 t .
the samples were measured in different fields and temperatures and also rotated in the magnetic field under the maximum lorentz force configuration ( @xmath8 ) . the current was continuously ramped until an abort criterion was reached .
the critical current was defined by the 1@xmath9v@xmath10 criterion .
indium pressed contacts are most suitable for our experiments because of their low resistivity , reliability and removability , which is important for the irradiation process .
the transition temperature ( @xmath11 ) and the irreversibility lines were measured with a constant current of 10ma , while the temperature was slowly decreased at fixed applied fields . @xmath11 and @xmath12 were evaluated at 0.1@xmath9v@xmath10 . only the data presented in figs .
[ fig : crossover ] and [ fig : single_poly ] were measured in a vibrating sample magnetometer ( vsm ) at 10k , while ramping the field with 0.6tmin@xmath13 .
all samples were checked by magnetoscanning @xcite to ensure homogeneous properties .
the samples were irradiated in the central irradiation facility of the triga mark ii reactor in vienna at a power of 250kw ( thermal / fast neutron flux density : @xmath14 ) .
the samples were sealed into a quartz tube to avoid contact between the samples and the cooling water in the reactor .
the quartz tubes were put into aluminium containers , which were filled with water and placed at the correct irradiation position .
the temperature during irradiation did not exceed 60 @xmath15c .
all fluences refer to fast neutrons ( @xmath16mev ) , the corresponding thermal neutron fluences ( @xmath17ev ) were smaller by a factor of 1.27 @xcite .
neutron irradiation creates a wide spectrum of defects in the superconductor by elastic collisions of the neutrons with the atoms . depending on the kinetic energy transferred from the neutron to the primary recoil atom , point defects or larger defect structures appear . in an avalanche like process a huge number of atoms
can be displaced , leading to local melting of the crystal lattice .
the result is a spherical region of amorphous material with a mean diameter of about 3 to 6 nm @xcite . while fast neutrons ( @xmath16mev ) transfer sufficient energy to the primary recoil atom to produce collision cascades , thermal neutrons ( @xmath18ev ) do not lead to defects in the material because their energy is below the displacement threshold .
epithermal neutrons ( in the kev range ) lead to the formation of point defects and point defect clusters .
the density of collision cascades after irradiation to @xmath6m@xmath5 is about@xcite @xmath19m@xmath20 , which corresponds to a mean distance between two cascades of 27 nm , the lattice parameter of the flux line lattice at 3.2 t .
the transition temperature of the modrab samples was 90.6k , of the pldysz 88.5k and of the mocvdmgo 89.1k .
neutron irradiation decreases the transition temperature , @xmath11 , which is ascribed to enhanced impurity scattering caused by the introduced defects .
non - magnetic impurities do not reduce @xmath11 in isotropic s - wave superconductors , but in anisotropic and d - wave superconductors @xcite .
the effect is comparatively small at the chosen fluences .
we find a linear decrease with neutron fluence by around @xmath21km@xmath22 ( 2 k at a fluence of @xmath6m@xmath5 ) , which is around half that found in single crystals @xcite .
the transitions normally do not broaden after irradiation to these neutron fluences .
the irreversibility fields ( @xmath23 ) of the pldysz and mocvdmgo samples are nearly identical and higher than those of the modrab sample .
all irreversibility lines become similar after neutron irradiaton , since only that of the modrab sample is changed significantly.,scaledwidth=50.0% ] the irreversibility lines ( @xmath23 ) of the samples are compared in fig .
[ fig : il ] .
the pldysz and mocvdmgo samples behave almost identically , the irreversibility fields of the modrab samples are lower ( except near @xmath11 ) , but shift close to the values of the other samples after irradiation to 10@xmath24m@xmath5 .
the latter do not change significantly in the accessible field range , since the moderate decrease in @xmath11 , which somewhat decreases @xmath25 at high temperatures , is counterbalanced by a slight steepening of the irreversibility line .
we note that neutron irradiation can not improve the irreversibility line of state - of - the - art coated conductors and that all samples behave similarly after irradiation .
it seems that the modrab samples have a comparatively low density of pinning efficient defects , which is increased by the irradiation , thus enhancing @xmath25 .
the irreversibility line becomes quite insensitive to the defect concentration at high defect densities , which could indicate an upper limit for @xmath25@xcite .
the irradiation has no beneficial effect on the irreversibility line for the other main field orientation ( @xmath26 ) and we observe no significant effect ( modrab ) or a slight degradation ( other samples ) .
the variation of the intrinsic pinning potential is on the length scale of the c - axis lattice parameter ( equivalent to a very high density of pinning centers ) and presumably much more efficient than the cascades due to its correlated two dimensional character .
neutron irradiation enhances the critical current densities in a large field and temperature range , but decreases them at low fields or close to @xmath11 .
the positive effect is larger if the field is oriented perpendicular to the tape ( left panel ) than in the parallel orientation ( right panel).,title="fig : " ] neutron irradiation enhances the critical current densities in a large field and temperature range , but decreases them at low fields or close to @xmath11 . the positive effect is larger if the field is oriented perpendicular to the tape ( left panel ) than in the parallel orientation ( right panel).,title="fig : " ] depending on field , temperature , fluence , and the original defect structure , the irradiation will enhance or decrease the critical currents .
the strongest increase was observed in the sample with the weakest as grown pinning structure ( modrab ) .
the corresponding data at 77k , 64k , and 50k are presented in fig .
[ fig : jc ] for both main field orientations .
the critical currents decrease at low magnetic fields after irradiation , which is discussed in detail in the next section .
@xmath0@xmath27 becomes flatter after the irradiation at low and intermediate fields , which leads to a crossover and an enhancement at intermediate and high fields .
the flattening manifests itself in a smaller exponent @xmath28 of the power - law @xmath29 , which is a reasonable approximation for the field dependence of @xmath0 at intermediate fields .
note that the crossover field between the enhancement and degradation of @xmath0 is generally higher for the field parallel to the ab planes ( right panel in fig .
[ fig : jc ] ) , where an enhancement is found only at high fields and low temperatures .
the currents degrade in the whole field range at 77k .
the ( relative ) enhancement is more pronounced at low temperatures for both orientations , which is in contrast to earlier findings on single crystals @xcite .
this is a consequence of the competition between pinning and thermal energy , which reduces the efficiency of comparatively small defects at high temperatures .
the collision cascades are obviously the largest pinning centers in single crystals , but this does not seem to be the case in coated conductors , where a large number and variety of linear and planar defects exists @xcite , which are more efficient than the cascades at high temperatures .
in addition , neutron irradiation also introduces smaller defects , which are efficient only at low temperatures .
the other types of samples ( pldysz , mocvdmgo ) behave qualitatively similarly but data are available only for a fluence of @xmath7m@xmath5 so far .
a similar reduction of @xmath0 at low fields turns to an enhancement at comparable crossover fields , which hardly depend on the neutron fluence ( see fig . [
fig : crossover ] ) .
the field dependence becomes again weaker at intermediate field , which reduces @xmath28 . only at high fields ,
the hardly changed irreversibility field causes a difference to the modrab samples .
while the relative enhancement diverges for @xmath23 at the originally lower irreversibility fields in the modrab samples , the @xmath0(@xmath30 ) curves merge again near the hardly changed irreversibility fields in the pldsz and mocvdmgo samples . at high temperatures , where the irreversibility field is slightly reduced due to the decrease of @xmath11 , even a second crossover is observed , and @xmath0 is only enhanced at intermediate fields .
the crossover in the field dependence @xmath31 between the unirradiated state and the irradiated state ( fig . [ fig : crossover ] ) , is a feature that is generally observed on conductors grown by physical as well as chemical deposition .
the twofold effect of the irradiation on the current transport indicates that two regimes must be separated : a high field region ( ) which benefits from the additional pinning centres and a low field domain , where a different mechanism causes a @xmath0-depression .
crossover in @xmath31 between the irradiated and the unirradiated state after sequential irradiation ( sample modrab).,scaledwidth=50.0% ] this is explained by considering the specific nature of current transport in a granular medium , i.e. , the critical current density in the grains @xmath32 has to be distinguished from the grain boundary critical current density @xmath33 flowing across the sample .
numerous experiments on bicrystals @xcite showed that , depending on the grain boundary misorientation angle , the current transport is controlled by @xmath33 at low fields , whereas at fields in the tesla range @xmath32 limits the current transport . although it is not fully established that the behaviour of a single well - defined grain boundary is representative for the complex grain boundary network of a coated conductor @xcite , the current limitation by the grains and an increase of @xmath32 after irradiation are straightforward explanations for the observed @xmath0 improvement in high fields .
further evidence for this scenario is provided by neutron irradiation of ybco thin films grown on single - crystalline substrates , which can be regarded as single macroscopic grains and thus serve as a model system for the individual microscopic grains in a coated conductor .
these films show , in agreement with the above assumption of an increase in @xmath32 , enhanced critical current densities in the entire field range ( cf .
[ fig : single_poly]a and reference ) , if the temperature is not too close to the superconducting transition temperature , i.e. , when the ( moderate ) reduction of @xmath11 after irradiation is negligible .
the fact that the @xmath0 decrease occurs at rather low fields , where @xmath33 is dominant in the bicrystal experiments , suggests that the grain boundary network is responsible for this effect .
two different but not mutually exclusive mechanisms are able to explain the reduction of @xmath33 . first , it is important to notice , that the density of the irradiation defects is too low to reduce the cross - section for current transport significantly .
but the defects are mobile during the irradiation and the grain boundaries represent natural barriers for their movement , which causes the defects to accumulate the grain boundaries deteriorate after irradiation .
this effect has , however , only an impact on the overall @xmath0 , if the current transport is limited by the grain boundaries .
hence , the crossover indicates the transition between the low field regime , which is dominated by the grain boundary network , and the high - field regime controlled by the properties of the grains .
comparison of the effect of neutron irradiation on samples grown on single - crystalline ( a ) and poly - crystalline ( b ) substrates .
the critical current density in the single - crystalline sample is enhanced in the entire field range , while the granular sample exhibits a crossover at about the same field as state - of - the - art coated conductors .
the inset shows that the remnant moment of the granular sample decreases when the grain boundary currents are saturated and the flux penetrates the grains .
( data are normalised to the unirradiated state.),title="fig : " ] comparison of the effect of neutron irradiation on samples grown on single - crystalline ( a ) and poly - crystalline ( b ) substrates .
the critical current density in the single - crystalline sample is enhanced in the entire field range , while the granular sample exhibits a crossover at about the same field as state - of - the - art coated conductors .
the inset shows that the remnant moment of the granular sample decreases when the grain boundary currents are saturated and the flux penetrates the grains .
( data are normalised to the unirradiated state.),title="fig : " ] another explanation for the @xmath0 reduction at low fields is based on the field exerted by the intra - granular shielding currents on the grain boundaries : if @xmath32 is larger than @xmath33 , additional flux remains trapped in the grains after a change of the external applied field and the circulating currents in the grain generate an additional field at the grain boundaries .
thus , the enhancement of @xmath32 by the irradiation , which is responsible for the @xmath0 improvement at high fields , increases the grain boundary field and thereby reduces the field dependent @xmath33 at low applied fields .
an experimental test is provided by a method similar to ref . , where the remnant magnetic moment @xmath34 of the sample is assessed after applying successively larger fields @xmath35 .
if @xmath33 is smaller than @xmath32 , flux will initially penetrate along the grain boundaries .
after the grain boundary network has saturated and additional flux is trapped only in the grains , @xmath36 decreases as the intra - grain currents increase the field at the grain boundary .
is not significant , because @xmath37 , where @xmath38 is the radius , which is a few millimetres for the sample and more than a hundred times less for a single grain .
thus , the dominating component of the total magnetic moment stems from the currents flowing across the grain boundaries in all samples expect those with high - angle grain boundaries , which makes a direct assessment of @xmath32 and @xmath33 ( as for example in ref . ) impossible . ]
the expected behaviour of @xmath36 is clearly evident in samples grown at an early stage of optimisation ( fig .
[ fig : single_poly]b ) to the granular substrate and thus carry a rather low @xmath33 @xcite . despite the smaller @xmath0 of this sample ,
the crossover occurs at fields comparable to state - of - the - art coated conductors .
although the two proposed mechanisms are different , they do not exclude each other , but rely on the same argument : a transition between grain boundary to grain controlled current transport leading to a crossover in @xmath31 between the irradiated and the unirradiated state .
it should further be noted that the complexity of the grain boundary network does not allow to draw clear distinctions and the crossover indicates only approximately the transition from one regime to the other .
the irradiation reduces the @xmath0- anisotropy ( sample modrab ) .
the minimum in @xmath0 appears near 9@xmath15 , where the currents are hardly changed by the introduced defects.,scaledwidth=50.0% ] the angular dependence of the critical current in the modrab sample at 64k and 3 t is presented in fig .
[ fig : angamsc ] .
the pronounced peak at 0@xmath15 either comes from intrinsic pinning by the periodic variation of the order parameter along the c - axis , or from correlated defects , such as stacking faults .
there is hardly any trace of a peak at 90@xmath15 in the unirradiated tape .
this indicates that the density of c - axis correlated defects is small .
the random , uncorrelated disorder introduced by the neutrons reduces the @xmath39-anisotropy , @xmath40 , from 2.7 to 1.3 at a fluence of @xmath6m@xmath5 .
the latter is considerably smaller than the @xmath39-anisotropy in a thin film with a claimed ( intrinsic or effective ) anisotropy of only@xcite 1.5 , obtained by a scaling analysis @xmath0(@xmath41 ) @xcite on the basis of the ansiotropic scaling approach by blatter et al .
@xcite such a small intrinsic anisotropy can be excluded in the present sample , since the anisotropy of the irreversibility fields is much higher and not significantly altered by the irradiation .
the shape of @xmath0(@xmath41 ) in the irradiated state is extremely interesting .
the peak at 0@xmath15 is followed by a minimum at about 9@xmath15 and another broad maximum centered at around @xmath42 .
note that this behaviour was induced by the addition of random , uncorrelated disorder and is hardly compatible with the anisotropic scaling approach , although this model assumes such a defect structure , but with weak instead of strong pinning defects .
this is not unexpected , since the scaling of @xmath0 is based on single - vortex weak - collective pinning @xcite which predicts extremely small currents and is obviously not applicable to `` high-@xmath0 '' coated conductors .
the peak at 90@xmath15 indicates the presence of c - axis correlated defects in the mocvdmgo sample .
it disappears after neutron irradiation , the minimum at @xmath43 is amplified.,scaledwidth=50.0% ] the mocvdmgo samples contain c - axis correlated defects , as indicated by the peak at around 90@xmath15 shown in fig .
[ fig : angsp ] .
this peak disappears after irradiation and the currents hardly change between 25@xmath15 and 90@xmath15 at 64k and 4 t , in contradiction to the anisotropic scaling approach .
the minimum , again at about 9@xmath15 , which was present in these samples also before irradition , is amplified . only the intrinsic pinning remains practically unaffected . at higher fields ( not shown in fig .
[ fig : angsp ] ) , a minimum occurs at 90@xmath15 , which is an obvious consequence of the small irreversibility field in this orientation .
the ratio of the irreversiblity fields in the main field orientations is consistent with anisotropic scaling and with expectations for the intrinsic anisotropy ( @xmath44 ) and seems to represent a much more reliable estimate for the intrinsic anisotropy ( the scaling of fields is independent of pinning models ) than the angular dependence of @xmath0 , since this approach does not work for the relevant pinning centers in coated conductors .
commercial state - of - the - art conductors fulfill the requirements of fusion magnets on @xmath0 at 64k , a temperature , where liquid nitrogen ( at low pressure ) can be used as coolant , if the magnetic field is parallel to the ab planes . for the other main field orientation as well as at 77k
the performance is still too low.,scaledwidth=50.0% ] powerful magnets will be needed to confine the plasma of future fusion power plants .
the use of high temperature superconductors instead of nb@xmath2sn would allow operation at higher temperatures and/or fields , which would reduce the cooling requirements and could also result in a more compact and cheaper design .
however , only first concepts for fusion magnets made of coated conductors are being developed at present .
cables with a current capability in the ten ka range are needed and the strands have to be somehow twisted @xcite in order to reduce ac losses . electrical , mechanical , and thermal stabilization of such wires and
quench detection / protection will be demanding issues .
apart from these technological issues , the superconducting properties have to fulfill two basic requirements .
the critical current densities must be comparatively high , since the superconducting volume fraction of a coated conductor is low ( typically 1 - 2% ) and they must not degrade during operation in a hard radiation environment ( fast neutrons ) .
a rough estimate of the required intrinsic critical current density leads to about @xmath45am@xmath5 ( @xmath46am@xmath5 ) at about 12 t ( 14 t ) @xcite , assuming the overall current density of the actual iter ( demo ) cable design for the toroidal field ( tf ) coils or the central solenoid ( cs ) .
these values are certainly achievable with present conductors , but define the operation temperature . cooling with liquid nitrogen would be favorable , but is possible only above 64k ( at low pressure ) . figure [ fig : fusion ] demonstrates that this seems feasible ( although not at 77k ) with present conductors ( mocvdmgo ) if the field is aligned with the ab planes .
the magnetic field in the fusion coils will be close to this orientation , but , depending on the cable and coil design , orthogonal components will occur . in the other limiting case ( @xmath23 )
, the performance currently fails significantly . since the currents drop quickly at small misalignment angles ( figs .
[ fig : angamsc ] and [ fig : angsp ] ) , further conductor development seems mandatory in order to reach the required current densities at these high temperatures .
the field in the poloidal field ( pf ) coils is much smaller ( up to about 6 t in iter and 9 t in demo ) , which is also highlighted in fig .
[ fig : fusion ] .
even these requirements seem to be demanding at 77k or for @xmath23 .
the expected life time fluence ( fast neutrons ) at the main fusion magnets ( pf , cs ) is around @xmath47m@xmath5 and might be even higher in future fusion power plants .
since the currents at 77k already degrade at a fluence of @xmath6m@xmath5 , radiation resistance is an important issue for the application of coated conductors in fusion magnets , at least at elevated temperatures .
the influence of fast neutron irradiation on differently fabricated coated conductors was found to be similar .
the critical currents decrease at low magnetic fields , where inter - granular currents set the upper limit for loss free current flow .
this decrease can be understood either by a slight degradation of the grain boundaries , or by the increasing shielding currents within the grains , which generate an additional field at the grain boundaries .
the irradiation enhances @xmath0 at intermediate and high magnetic fields , where the macroscopic current is limited by pinning , which is improved by the additional pinning centers .
although these pinning centers are uncorrelated , the angular dependence of @xmath0does not obey anisotropic scaling .
this is a direct consequence of the pinning strength of these defects , which can not be considered as weak , as assumed in the scaling approach .
since this should be true for any relevant pinning centers in these `` high-@xmath0 '' conductors , anisotropic scaling based on weak collective single vortex pinning is not applicable .
this work , supported by the european communities under the contract of association between euratom / oeaw was partly carried out within the framework of the european fusion development agreement .
the views and opinions expressed herein do not necessarily reflect those of the european commission .
one of us ( m.c . ) wishes to acknowledge financial support through the nespa project . | various commercial coated conductors were irradiated with fast neutrons in order to introduce randomly distributed , uncorrelated defects which increase the critical current density , @xmath0 , in a wide temperature and field range . the @xmath0-anisotropy is significantly reduced and the angular dependence of @xmath0 does not obey the anisotropic scaling approach . these defects enhance the irreversibility line in not fully optimized tapes , but they do not in state - of - the - art conductors .
neutron irradiation provides a clear distinction between the low field region , where @xmath0 is limited by the grain boundaries , and the high field region , where depinning leads to dissipation . |
the arrow of time is intimately connected to the entropy of the universe .
the second law of thermodynamics inexorably drives us to ever increasing entropy , yet we live in neither a situation of maximal entropy ( a black hole ) nor in a minimal entropy universe .
apparently we thrive in the current `` medium entropy '' universe .
how is this possible ? in this conference , two possible explanations for this homogenous and isotropic universe we live in have been discussed : special initial conditions or eternal inflation combined with anthropic arguments .
in fact , there is a third option : cyclicity . here
the universe oscillates through a series of expansions and contractions . in a successful model of a cyclic nature , the entropy that has been created in each cycle
must yet again be destroyed in order to reset the stage for the next oscillation .
we discuss a `` phantom bounce '' @xcite as a proposal for an oscillating universe in which we postulate violation of the weak energy condition as a mechanism to destroy the ( high entropy ) black holes that are produced during each cycle .
a key advantage of our proposal is that the phantom component of our proposal is testable in astrophysical data soon .
we discuss a scenario in which the universe oscillates through a series of expansions and contractions .
after it finishes its current expanding phase , the universe reaches a state of maximum expansion which we will call `` turnaround '' , and then begins to recollapse .
once it reaches its smallest extent at the `` bounce '' , it will once again begin to expand .
this scenario is distinguished from other proposed cyclic universe scenarios @xcite in that cosmological acceleration due to `` phantom '' energy ( _ i.e. _ , dark energy with a supernegative equation of state , @xmath0 ) @xcite plays a crucial role .
we originally proposed the phantom bounce in 2004 @xcite .
subsequent related work includes @xcite ; see also @xcite .
perturbations in this cosmology were discussed in @xcite .
the idea of an oscillating universe was first proposed in the 1930 s by tolman . over the subsequent decades
, two problems stymied the success of oscillating models .
first , the formation of large scale structure and of black holes during the expanding phase leads to problems during the contracting phase @xcite .
the black holes , which can not disappear due to hawking area theorems , grow ever larger during subsequent cycles . eventually , they occupy the entire horizon volume during the contracting phase so that calculations break down .
( only the smallest black holes can evaporate via hawking radiation . )
the second unsolved problem of oscillating models was the lack of a mechanism for the bounce and turnaround .
the turnaround at the end of the expanding phase might be explained by invoking a closed universe , but the recent evidence for cosmological acceleration removes that possibility . for the observationally
favored density of `` dark energy '' , even a closed universe will expand forever .
thus , cyclic cosmologies appeared to conflict with observations .
our scenario resolves these problems .
our resolution to the black hole overproduction problem is provided by a `` phantom '' component to the universe , which destroys all structures towards the end of the universe s expanding stage .
phantom energy , a proposed explanation for the acceleration of the universe , is characterized by a component @xmath1 with equation of state @xmath2 since the sum of the pressure and energy density is negative , the dominant energy bound of general relativity is violated ; yet recent work explores such models nevertheless .
phantom energy can dominate the universe today and drive the current acceleration .
then it becomes ever more dominant as the universe expands .
with such an unusual equation of state , the hawking area theorems fail , and black holes can disappear @xcite . in `` big rip '' scenarios @xcite , the rapidly accelerating expansion due to this growing phantom component tears apart all bound objects including black holes .
( we speculate about remnants of these black holes below . )
the phantom energy density becomes infinite in finite time @xcite .
the energy density of any field described by equation of state @xmath3 depends on the scale factor @xmath4 as @xmath5 hence , for @xmath6 , @xmath7 grows as the universe expands .
of course , we expect that an epoch of quantum gravity sets in before the energy density becomes infinite .
we therefore arrive at the peculiar notion that quantum gravity governs the behavior of the universe both at the beginning and at the end of the expanding universe ( i.e. , at the smallest and largest values of the scale factor ) . here
we consider an example of the role that high energy density physics may play on both ends of the lifetime of an expanding universe : we consider the idea that large energy densities may cause the universe to bounce when it is small , and to turn around when it is large .
the idea is economical in that it is the _ same physics _ which operates at both bounce and turnaround .
we use modifications to the friedmann equations to provide a mechanism for the bounce and the turnaround that are responsible for the alternating expansion and contraction of the universe .
in particular , we focus on `` braneworld '' scenarios in which our observable universe is a three - dimensional surface situated in extra dimensions . several scenarios for implementing a bounce have been proposed in the literature @xcite . as an example , we focus on the modification to the randall - sundrum @xcite scenario proposed by shtanov and sahni @xcite , which involves a negative brane tension and a timelike extra dimension leading to a modified friedmann equation .
another example is the quantum bounce in loop quantum gravity @xcite .
once the energy density of the universe reaches a critical value , cosmological evolution changes direction : if it has been expanding , it turns around and begins to recontract .
if it has been contracting , it bounces and begins to expand .
we emphasize that the two components we propose here work together : we use a modified friedmann equation as a mechanism for a bounce and turnaround , and we add a phantom component to the universe to destroy black holes . due to the phantom component , the same high energy behavior that produces a bounce at the end of the contracting phase also produces a turnaround at the end of the expanding phase .
in addition , the bounce and turnaround are both nonsingular , unlike the cyclic scenario proposed by steinhardt and turok @xcite , which is complicated by a number of physical singularities related to brane collisions near the bounce @xcite .
this is currently a very controversial topic .
in an oscillating cosmology , what we observe to be `` the big bang '' really is the universe emerging from a bounce .
the universe at this point has its smallest extent ( smallest scale factor @xmath4 ) and largest energy density , somewhere near the planck density .
the universe then expands , its density decreases , and it goes through the classic radiation dominated and matter dominated phases , with the usual primordial nucleosynthesis , microwave background , and formation of large structure .
a period of inflation may or may not be necessary to establish flatness and homogeneity . at a redshift @xmath8 ,
the universe starts to accelerate due to the existence of a vacuum component or quintessence field @xmath1 .
we take a `` phantom '' component with @xmath6 .
the energy density of this component grows rapidly as the universe expands .
any structures produced during the expanding phase , including galaxies and black holes , are torn apart by the extremely rapid expansion provided by the phantom component .
any physics relevant at the high densities near the `` big bang '' again becomes important at the high densities near the end of the expanding phase .
modifications to the friedmann equation become important at high densities , and cause the universe to turn around .
the universe reaches a characteristic maximum density 2@xmath9 ( which might be anywhere in the range from tev to @xmath10 ) , and starts to contract .
as it contracts , at first its energy density decreases ( as the phantom component decreases in importance ) , but then it again increases as matter and radiation become dominant .
eventually it reaches the high values at which the modifications to friedmann equations become important .
once the energy density again reaches the same characteristic scale @xmath11 , the universe stops contracting , bounces , and once again expands . in the standard cosmology , there is no way to avoid a singularity for small radius or scale factor @xmath4 . in the context of extra dimensions ,
however , one can have a bounce at finite @xmath4 so that singularities are avoided .
a nonsingular bounce is obtained if the friedmann equation is modified by the addition of a new negative term on the right hand side : @xmath12,\ ] ] where the function @xmath13 is positive . for a contracting universe to reverse and begin expanding again
, we must have @xmath14 , which results in a condition on @xmath15 , @xmath16 similarly , for an expanding universe , @xmath17 must be negative for the expansion to reverse . a modified friedmann equation of the form of eq .
( [ eq : hubble ] ) can be motivated in the context of braneworld scenarios , where our observable universe is a 3-dimensional surface embedded in extra dimensions .
@xcite showed that einstein s equations in higher dimensions , together with israel boundary conditions on our brane , can give rise to an equation of the form of eq .
( [ eq : hubble ] ) .
different values of energy / momentum in the extra dimensions ( the bulk ) can be responsible for different @xmath13 in eq .
( [ eq : hubble ] ) .
in particular , we focus on `` braneworld '' motivated modifications to the friedmann equation , where the modification to the friedmann equation for the brane bound observer @xcite is @xmath18 where the last term ( @xmath19 is an integration constant ) appears as a form of `` dark radiation '' ( that is constrained like ordinary radiation ) , and @xmath20 corresponds to the metric signature of the extra dimension @xcite .
we will also assume that the bulk cosmological constant is set so that the three - dimensional cosmological constant @xmath21 is negligible .
hence the relevant correction to the friedmann equation is the quadratic term , @xmath22 . for @xmath23 ,
the friedmann equation becomes @xmath24.\ ] ] one way to obtain @xmath23 corresponds to an extra _ timelike _ dimension : models with more than one time coordinate typically suffer from pathologies such as closed timelike curves and non - unitarity .
we use the model in ref .
@xcite to motivate the choice of sign in the friedmann equation , but a more detailed treatment would need to address these other issues to form a fully consistent picture . alternatively , in loop quantum gravity
, there is a quantum bounce that takes place at planck densities in lieu of the singularity in the standard classical friedmann equation @xcite ; if one couples this quantum bounce with a phantom component as in this paper , one would again obtain the same oscillating cosmology as discussed in this paper .
the expansion rate of the universe @xmath25 at @xmath26 ; it is at this scale that the universe bounces and turns around . for this choice of @xmath15 , @xmath27 and
the required condition on @xmath17 is satisfied at both bounce ( @xmath28 ) and turnaround ( @xmath29 ) . on one end of the cycle it goes from contracting to expanding ( this bounce looks to us like the big bang ) , and then at the other end of the cycle it goes from expanding to contracting .
this behavior is illustrated in the figure . in models motivated by the randall - sundrum scenario , the most natural value of the brane tension is @xmath30 , but we treat the problem generally for any value of @xmath31 . at scales above @xmath32 , the validity of eq .
( [ eq : mod2 ] ) breaks down in detail .
however , the approach to @xmath25 and thus the existence of a bounce and turnaround remain sensible . in any case
, we use this braneworld model merely as an example of a correction to the friedmann equation .
other modifications to the friedmann equation might work as well , as long as there is the requisite minus sign in the equation .
black holes pose a serious problem in a standard oscillating universe
. however , the hawking area theorems that guarantee the continued existence of black holes have been constructed in special settings and may not apply here ; e.g. , the same modifications to gravity that give a bounce rather than a singularity in the cosmology may avoid singularities in the black holes .
indeed , when @xmath33 , davies @xcite has shown that the theorem does not hold .
recently , caldwell , _ et al . _
@xcite described the dissolution of bound structures in the `` big rip '' towards the end of a phantom dominated universe .
any black holes formed in an expanding phase of the universe are torn apart before they can create problems during contraction . when are the black holes destroyed ?
we want to be certain that they are torn apart before turnaround . in general relativity ,
the source for a gravitational potential is the volume integral of @xmath34 .
an object of radius @xmath35 and mass @xmath36 is pulled apart when @xmath37 writing @xmath38 during phantom domination and taking @xmath39 for the black hole , we find that black holes are pulled apart when @xmath40 which happens when the energy density of the universe has climbed to a value @xmath41 more massive black holes are destroyed at lower values of @xmath42 , i.e. earlier .
it is the smallest black holes that get shredded last .
we must ensure that the black holes are destroyed before turnaround , so that @xmath43 . as an example
, we can take @xmath44 .
then @xmath45 solar mass black holes , such as those at the centers of galaxies , get pulled apart when @xmath46 , which easily satisfies the above condition .
the most tricky case would be planck mass black holes , which either formed primordially or are relics of larger black holes that hawking radiated .
even these should still be disrupted . from eq .
( [ eq : rhobh ] ) these will be shredded when @xmath47 , before turnaround if the brane tension @xmath48 . however , for gut scale brane tension @xmath49 , only black holes with @xmath50 are disrupted .
fortunately these black holes hawking evaporate in a time @xmath51 where @xmath36 is the black hole mass .
this occurs in only @xmath52 sec for a black hole with @xmath53 .
we also speculate that planck mass remnant black holes that can not disappear ( still containing the singularity ) may be dark matter candidates .
our proposal contains the novel feature that both bounce and turnaround are produced by the same modification to the friedmann equation .
however , it does so at the price of including more than one speculative element : the modified friedmann equation requires a braneworld model to achieve , and the cosmology must be dominated by phantom energy . in many cases a phantom component is difficult to implement from a fundamental standpoint without severe pathologies such as an unstable vacuum ( see , for example , ref .
however , parker and raval @xcite have investigated a cosmological model with zero cosmological constant , but containing the vacuum energy of a simple quantized free scalar field of low mass , and found that it has @xmath54 without any pathologies .
several additional areas also remain to be addressed .
first , as the universe is contracting , those modes of the density fluctuations that we usually throw away as decaying ( in an expanding universe ) are instead growing . hence dangerous structures may form during the contracting phase . at the end of the contracting phase
, there is no phantom energy to wipe out whatever structure is formed . in this sense ,
the initial conditions for structure formation in this picture are set either during the phantom energy dominated epoch near turnaround or by the quantum generation of fluctuations in the collapsing phase @xcite .
black hole formation could still kill the model .
second , it is not obvious that it is possible to create a truly cyclic ( _ i.e. _ perfectly periodic ) cosmology within the context of the `` phantom bounce '' scenario .
the reason for this is entropy production .
we speculate that it may be possible to create quasi - cyclic evolution by redshifting entropy out of the horizon during the period of accelerating expansion .
even more speculatively , we note that the special case of @xmath55 , although disfavored by observation , possesses an intriguing duality between radiation ( @xmath56 ) and phantom energy ( @xmath57 ) . in this case , the behaviors of these components exchange identity under a transformation @xmath58 @xcite , effectively exchanging bounce for turnaround , a symmetry which might be exploited to achieve truly cyclic evolution . on the observational side ,
our key ingredient is testable .
the current expansion of the universe is the subject of much intense investigation .
the universe is apparently accelerating , but the exact nature of the acceleration is not yet known .
the previous value of the equation of state may be discovered over the next decade .
the current uncertainty in the equation of state easily allows for the possibility of a phantom energy ; some @xcite have argued that @xmath33 is an excellent fit to the data .
if upcoming observations discover that such a phantom energy indeed exists , then the community may be forced to conclude that the weak energy condition is violated and will need to rethink many basic assumptions .
phantom energy may be forced upon us , with the helpful consequence of permitting the `` medium '' entropy universe we inhabit .
d. lyth , phys . lett . *
b526 * , 173 ( 2002 ) ; j. martin , p. peter , n. pinto neto and d. j. schwarz , phys .
d * 65 * , 123513 ( 2002 ) ; c. gordon and n. turok , phys .
d * 67 * , 123508 ( 2003 ) ; j. martin and p. peter , phys .
* 92 * , 061301 ( 2004 ) ; t. j. battefeld , s. p. patil and r. brandenberger , arxiv : hep - th/0401010 .
p. binetruy , c. deffayet , and d. langlois , _ nucl.phys .
_ * b565 * , 269 ( 2000 ) ; e.e .
flanagan , s. tye , and i. wasserman , _ phys .
_ * d62 * , 024011 ( 2000 ) ; c. csaki , m. graesser , c. kolda , and j. terning , _ phys .
* b462 * , 34 ( 1999 ) ; j. cline , c. grojean , and g. servant , _ phys .
lett . _ * 83 * , 4245 ( 1999 ) ; l. mersini , _ mod .
phys . lett . _
* a16 * , 1583 ( 2001 ) ; w. goldberger and m. wise , _ phys .
lett . _ * b475 * , 275 ( 2000 ) .
m. p. dabrowski , t. stachowiak and m. szydlowski , phys .
d * 68 * , 103519 ( 2003 ) ; l. p. chimento and r. lazkoz , phys .
lett . * 91 * , 211301 ( 2003 ) ; j. e. lidsey , arxiv : gr - qc/0405055 ; l. p. chimento and r. lazkoz , arxiv : astro - ph/040551 . | an oscillating universe cycles through a series of expansions and contractions . we propose a model in which `` phantom '' energy with a supernegative pressure ( @xmath0 ) grows rapidly and dominates the late - time expanding phase .
the universe s energy density is so large that the effects of quantum gravity are important at both the beginning and the end of each expansion ( or contraction ) .
the bounce can be caused by high energy modifications to the friedmann equation , which make the cosmology nonsingular .
the classic black hole overproduction of oscillating universes is resolved due to their destruction by the phantom energy . |
superpixels are regions of pixels grouped in some perceptually meaningful way , usually following colour or boundary cues .
they are designed to produce a simpler and more compact representation for an image , while keeping its semantic meaning intact .
superpixel segmentation is used most often as an image preprocessing step , with a view towards reducing computational complexity for subsequent processing steps .
the term _ superpixel _ , along with the first notable superpixel algorithm , was introduced by @xcite .
many algorithms followed , using various types of image features , various optimisation strategies and various implementations techniques .
these algorithms have varying specifications and performance requirements . for example , some algorithms aim to find a fixed number of superpixels , others try to find the minimum possible number of superpixels by imposing a colour cohesion requirement , while others place emphasis on matching image boundaries .
sometimes fast processing is required , when for example , the superpixel algorithm is used as a precursor to a tracker . sometimes superpixels are designed not to under - segment the image , when used for example as a means of condensing the image information , to serve as the basis of a labelling problem .
regardless of its design , superpixel segmentation is usually among the first steps in a much longer processing pipeline .
therefore , we believe that , for any superpixel method to be useful , it must satisfy the two following requirements : * it should not decrease the performance of the full processing pipeline ; * it should be fast .
the performance requirement is satisfied in many vision applications by superpixels that are _ compact _ , _ uniform _ and _ follow image edges_. these requirements motivated the * simple iterative clustering algorithm * ( slic ) algorithm of @xcite .
this is simple , efficient and suitable for real - time operation .
still however , the cpu - sequential implementation of slic need 300@xmath1400ms to segment a single 640x480 image . reducing the number of iterations for each clustering pass can make the algorithm faster , at the cost of a decrease in performance . in this work ,
we propose a gpu implementation of the slic algorithm , using the nvidia cuda framework .
our implementation is up to @xmath0 faster than the original cpu implementation of @xcite , making it , to our knowledge , the fastest superpixel method to date .
our full source code with a simple example can be downloaded from https://github.com/carlren/gslicr .
the following sections describe in detail our algorithm and implementation .
[ cols="^,^ " , ] c. ren , v. prisacariu , d. murray , and i. reid , `` star3d : simultaneous tracking and reconstruction of 3d objects using rgb - d data , '' in _ computer vision ( iccv ) , 2013 ieee international conference on _ , pp . 15611568 , dec 2013 .
v. a. prisacariu , o. khler , m. cheng , c. y. ren , j. p. c. valentin , p.
h. s. torr , i. d. reid , and d. w. murray , `` a framework for the volumetric integration of depth images , '' _ corr _ , vol .
abs/1410.0925 , 2014 . | we introduce a parallel gpu implementation of the simple linear iterative clustering ( slic ) superpixel segmentation . using a single graphic card
, our implementation achieves speedups of up to @xmath0 from the standard sequential implementation .
our implementation is fully compatible with the standard sequential implementation and the software is now available online and is open source . |
clinical prediction rule ( cpr ) , pulmonary embolism , topological data analysis , artificial neural network ( ann ) , computer aided detection cad
several statistical and machine learning techniques have been proposed in the literature to deal with output of implicit or explicit rules and good classification performance @xcite .
most available techniques , such as linear discriminant approaches , multilayer perceptrons or support vector machines , are able to achieve a good degree of provisional accuracy but these methods lack accuraccy sufficient for the implementation of computer - aided diagnosis ( cad ) for pulmonary embolism diagnosis .
different studies have been developed for cad system predicting development of pe in patients .
tang et al used data from the shangai xin hua hospital , tourassi et al and patil s used the data collected from the collaborative study of the _ prospective investigation of pulmonary embolism diagnosis ( pioped ) _ and built neural network @xcite . to improve the performance of a cad we built a new system based on topological data analysis and statistical approach for features selection to define the input space for theartificial neural network .
the result of the cad must be interpreted as a new cpr that might be used to assign a probability of an occurrence of pe@xcite .
pulmonary embolism ( pe ) is a relatively common and potentially lethal condition , affecting a proportion between 0.04@xmath0 and 0.09@xmath0 of the general population@xcite , and ranging , in most of the cohorts of outpatients with suspected pe , between 8@xmath0 and 12@xmath0@xcite . among patients who die of pe ,
the largest part is observed in the first few hours from the acute event @xcite . despite diagnostic advances , delays in pe diagnosis
are common and represent an important issue @xcite . as a cause of sudden death
, pe is second only to arrhythmic death . among survivors , recurrent embolism and death
can be prevented with prompt diagnosis and therapy .
current clinical guidelines suggest to perform a first - level , clinical stratification based upon patient s history , clinical findings and , in some cases , physician s judgment . for this purpose , several clinical prediction rulers ( cpr )
have been suggested and validated , and are currently in use in common clinical practice@xcite .
subsequent management of the patient relies mainly on this stratification .
patients at high - risk for pe should immediately undergo a computed tomography pulmonary angiography ( ctpa ) , while patients with intermediate or low pretest probability should be tested for high - sensitive ddimer assay@xcite .
only patients with increased ddimer levels should be further investigated with second level examinations .
ctpa is currently the most widely accepted imaging method recommended to confirm a suspect diagnosis of pe .
however , increasing evidences and its direct and indirect costs suggest a limitation in its use : ctpa has been associated to increased risk of secondary cancer from radiation exposure@xcite and contrast - induced nephropathy@xcite .
another interesting point is that current use of 64-slices detectors has increased the frequency of subsegmental pe diagnoses : this finding , in the absence of a documented deep vein thrombosis ( dvt ) , may cause clinical uncertainty , and lead to unnecessary therapies@xcite .
echocardiography is currently recommended for shocked patients with high suspect of pe . for
haemodinamically stable patients it is currently recommended only to better stratify the prognosis by detecting right ventricle dysfunction ( rvd)@xcite .
biomarkers such as bnp and troponin are used to assess disease severity and associated to right ventricle overload or damage , but are currently not included in diagnostic algorithms .
the extreme aspecificity and the variety of the clinical presentation and symptoms and the slenderness of the clinical signs of pe are cause on a side of the proven underdiagnosis on the other of an excess of negative examinations performed .
the positive ctpa rate is very low - less than 15@xmath0 of the overall number of exams performed for suspected acute pe @xcite .
recent studies have demonstrated the safety of rejecting the diagnosis of pe by the combination of a low clinical probability , assessed by a cpr , and a normal quantitative d - dimer test result , thereby decreasing the need for further diagnostic radiological imaging in up to 30@xmath0 of patients @xcite .
the most widely used cprs are the wells rule ( table ) and the geneva revised score @xcite the wells score has three different risk categories , and more recently it has been simplified assuming two risk categories with a cutoff set to 4 .
perrier et al has been successful in generating a model based solely on objective parameters using the geneva revised score @xcite .
this cpr is easily standardizable and has been validated internally and externally , although less extensively than the wells rule .
both scores appeared to have a comparable predictive value for pe .
regardless of the rule being performed , the proportion of patients with pe is around 10@xmath0 for low probability , 30@xmath0 for moderate probability and 65@xmath0 for high clinical probability category .
it is important to note that these scores have severe limitations .
the wells rule includes the physician s judgment of whether an alternative diagnosis is more likely than pe @xcite this criterion , which carries a major weight in the score , is subjective and can not be standardized .
moreover , it has been suggested that the predictive value of the wells rule is derived primarily from its subjective component @xcite .
the geneve revised score is based on 13 entirely objective variables , requires a blood gas analysis while breathing room air .
interestingly , these parameters have only been evaluated for patients in ed with a clinical performance result not superior to the wells score .
topological data analysis ( tda ) is a method to analyze multidimensional , complex data primarily driven by geometry .
tda is a result of over a decade of research in applications of pure mathematics to practical problems .
the main idea of this approach is that the shape of the data in an abstract multidimensional space drives the analysis by exploring the parallelism of a large number of machine learning algorithms .
the three fundamental concepts of tda are independence of coordinate systems ; insensitivity to deformation ; and compressed representation @xcite .
a typical example of insensitivity to deformation would be writing in a different font as long as the underlying meaning is preserved , while compressed representation refers to approximating a complex shape such that of a circle with a hexagon . using a mathematical concept of lenses @xcite ,
data can be projected onto a subspace suitable for visualization .
the topological features of the subspace are then inspected with traditional statistical approaches such as kolmogorov - smirnov or t - test analysis .
we applied tda to the clinical data of patients suspected of high risk of pe and used shapes of the generated topological networks on page 16 to identify different subgroups of patients and features that statistically significantly explain the observed differences .
we used the tda - derived features as input into artificial neural network analysis .
a complex system may be decomposed into simpler elements , in order to be able to understand it .
also simple elements may be gathered to produce a complex system in line with the approach _
divide et impera_. networks are one approach for achieving this .
there are a large number of different types of networks , but they all are characterized by the following components : * a set of nodes * connections between nodes .
the nodes can be treated as computational units .
they receive inputs , and process them to obtain an output . the connections
determine the information flow between nodes .
the interactions of nodes though the connections lead to a global behavior of the network , which can not be observed in the elements of the network .
this global behavior is said to be emergent , meaning that the abilities of the network supersede the ones of its elements .
networks can be used as very powerful tool because many systems can be assembled into a systems - like network space , applications including proteins , computers , and communities .
one type of network uses nodes as artificial neurons , and an artificial neuron is a computational model inspired by the natural neurons . natural neurons receive signals through synapses located on the dendrites or membrane of the neuron . when the signals received are strong enough ( greater than a certain threshold ) , the neuron is activated and emits a signal though the axon .
this can activate a cascade process .
the complexity of real neurons is highly abstracted when modeling artificial neurons .
these basically consist of inputs , which are multiplied by weights , and then computed by a mathematical function that determines the activation of the neuron .
another function computes the output of the artificial neuron ( sometimes in dependence of a certain threshold ) .
weights can also be negative , so we can say that the signal is inhibited by the negative weight . depending on the weights , the computation of the neuron will be different . by adjusting the weights of an artificial neuron
we can obtain the output we want for specific inputs .
if we scale an ann to hundreds or thousands of neurons , it is both complicated and labor intensive to manually discover all the necessary weights .
however , by identifying algorithms to adjust the weights of the ann in order to obtain the desired output from the network .
this process of adjusting weights is called learning or training @xcite @xcite @xcite.the number of ann types and uses is very high ; hence , there are hundreds of different models considered as anns .
the differences among them can be related to functions , the accepted values , the topology of network , and/or the learning algorithms , etc .
we applied backpropagation algorithm to perform a layered feed - forward anns .
this allows us to organize the artificial neurons in layers to send their signals `` forward '' , and propagate errors backwards .
the network receives inputs by neurons in the input layer , and the output of the network is given by the neurons on an output layer .
there may be one or more intermediate hidden layers .
the backpropagation algorithm uses supervised learning , which means that we provide the algorithm with examples of the inputs and outputs we want the network to compute , and then the error ( difference between actual and expected results ) is calculated .
the benefit of using backpropagation algorithm is that this error is reduced until the ann learns the training data .
the training begins with random weights , and the goal is to adjust them so that the error will be minimal .
further mathematical explanations are detail in@xcite and @xcite . in statistics , the kolmogorov smirnov test ( k s test ) is a nonparametric test for the equality of continuous , one - dimensional probability distributions that can be used to compare a sample with a reference probability distribution ( one - sample k s test ) , or to compare two samples ( two - sample k s test ) . the kolmogorov smirnov statistic quantifies a distance between the empirical distribution function of the sample and the cumulative distribution function of the reference distribution , or between the empirical distribution functions of two samples .
the null distribution of this statistic is calculated under the null hypothesis that the samples are drawn from the same distribution ( in the two - sample case ) or that the sample is drawn from the reference distribution ( in the one - sample case ) . in each case , the distributions considered under the null hypothesis are continuous distributions but are otherwise unrestricted .
the two - sample ks test is one of the most useful and general nonparametric methods for comparing two samples , as it is sensitive to differences in both location and shape of the empirical cumulative distribution functions of the two samples @xcite .
both artificial neural networks and the two scores ( wells and revised geneva ) were evaluated by using receiver - operating characteristic ( roc ) analysis .
medical tests play a vital role in modern medicine not only for confirming the presence of a disease but also to rule out the disease in individual patient .
a test with two outcome categories such as _ test + _ and _ test - _ is known as dichotomous , whereas more than two categories such as positive , indeterminate and negative called polytomous test .
the validity of a dichotomous test compared with the gold standard is determined by _ sensitivity _ and _ _ specificity__@xcite .
receiver operating characteristic ( roc ) _ curve is the plot that displays the full picture of trade - off between the sensitivity ( true positive rate ) and ( 1-specificity ) ( false positive rate ) across a series of inherent validity of a diagnostic test .
this curve is useful in : * evaluating the discriminatory ability of a test to correctly pick up diseased and non - diseased subjects ; * finding optimal cut - off point to least misclassify diseased and non - diseased subjects ; * comparing efficiency of two or more medical tests for assessing the same disease ; * comparing two or more observers measuring the same test total area under roc curve is a single index for measuring the performance a test .
the larger auc - area under curve - , the better is overall performance of the medical test to correctly identify diseased and non - diseased subjects .
equal aucs of two test represent similar overall performance of tests but this does not necessariy mean that both the curves are identical @xcite .
the jaccard similarity coefficient is a statistic used for comparing the similarity and diversity of sample sets .
the jaccard coefficient measures similarity between sample sets , and it is defined as@xcite : @xmath1 where : * _ doctor diagnosis _ -@xmath2 . * _ ann diagnosis _ - @xmath3 .
topologial data analysis was applied on the pe dataset to generate the network in figure ( see ) using ayasdi - iris ( ayasdi , inc , palo alto ) .
we color in red the pathological patients and in blue the healthy group .
iris highlighted two flares and a longer tail in the pathological group . from the comparison of the two flares we selected a sub - set of features characterized with high kolmogorov- smirnov and low p - value ( see table ) .
the information extracted from the comparison of the two flares is that there is a sub - group of patients who is characterized by the recurrence risk to be affected by pe .
iris detected some sub - groups among healthy patients , which present other diseases to be diagnosed properly even though they are not affected by pe . from the patients dataset we random selected 152 patients : 101 pathological and 51 healthy and we studied the discriminatory ability of the system .
tables 5 and 6 show the performance of three classifier : wells score , revised geneva score and the ann model .
any of the three auc was statistically significant different .
the comparative study of three aucs has been made with the software medcalc for windows , version 9.5.0.0 ( medcalc software , mariakerke , belgium)@xcite .
the aucs of revised geneva and wells score obtained from the analysed dataset are directly comparable with the results previously published in literature by other groups@xcite . figure ( see and ) showed the roc curves of three classifiers , they were evaluated by leave - one - out method@xcite .
the difference between the auc of the new classifier both with revised geneva s auc and wells auc was statistically significant at the @xmath4 confidence level , @xmath5 and @xmath6 respectively .
instead , the difference between revised geneva s auc and wells auc is not statistically relevant : @xmath7 .
thus , the network s discriminant power is significantly .
the jaccard coefficient for the ann is @xmath8 as every informatics system also the ann - classifier is affected by the round problem , i.e. the probability value of occurrence of pe in range [ 0.45 ; 0.5 ) is rounded to 0.5 then the patient is classified as pathological . at the state of the art we left to the physician the final judgment over this situations @xcite .
the study derives a new cpr for pulmonary embolism and evaluating patients cohort characteristic with a topological approach .
the new cpr has been obtained training an artificial neural network on the input formed by a set of features selected by iris .
iris extracted new knowledge from the patients dataset by the application of an innovative approach for data analysis : _ topological data analysis_. our results show that the feature selection strategy is beneficial for the performance improvement of an ann trained on the analyzed cohort .
a three - layer neural network can be trained to successfully perform the diagnostic task . in conclusion
a system based on iris and an ann can form the basis of a cad system to assist physicians with the right stratification of patients .
a web - version of the bp - ann has been published and it can be tested at the address _
http:\\cuda.unicam.it_. in the future we will perform a validation of the system both increasing the number of patients in the dataset and using different cohorts , we will perform a comparison study among artificial neural networks and other classification systems .
our dataset allowed us to increase the clinical performance of about 20@xmath0
from a cohort of 1500 patients accepted in the department of intensive and subintesive medicine from 2009 to 2012 at _ ospedali riuniti di ancona _ , a total of 987 patients of average age 70 years were included .
the inclusion criteria for this study were that the patients at least were recorded with wells score , revised geneva score , d - dimer and blood gas @xmath9 , the cutoff of d - dimer was 230ng / ml . for each patients
26 variables were collected ( see table ) . in these patients ,
the diagnosis of pe and the exclusion of pe were made by angio - ct . among these , 793 had pe and 147 did not have pe .
characteristics of the history , objective data from the physical examination and the outcome of d - dimer analysis were tabulated for each patient . to improve the interpretability of our data
the logarithm transformation has been applied to d - dimer and wbc ( white blood count ) . in the present work we analyzed only outpatients .
the patients dataset was processed using topological data analysis with ayasdi iris specifications .
ayasdi iris - ready data files are composed of rows and columns .
the calculations are performed on a row - by - row basis and the results elucidate the columns ( features ) that best define and explain particular groups of rows .
the ayasdi iris data format is a matrix with unique column headers and unique row identifiers .
column headers must uniquely identify the columns .
in contrast , the column that uniquely identifies the rows may be in any column position , it does not have to be the first or the last column .
filter or wrapper methods for features reduction can not be applied to our dataset : the application of these algorithms can be done on a dataset without missing data .
the elimination of rows with null values from our dataset would reduce extremily the number of samples and the covariance matrix is singular ( ill conditioned)@xcite . from the patients dataset we selected only _
d - dimer , revised geneva score and wells score _ for the analysis and variance normalized euclidean as metric function ( equation ) .
the variance normalized euclidean metric is utilized when data are comprised of disparate quantities that are not directly comparable .
the filters were l - infinity centrality , which assigns to each point the distance to the point most distant from it , and _ final diagnosis_. l - infinity centrality , is defined for a data point x to be the maximum distance from x to any other data point in the set .
large values of this function correspond to points that are far from the center of the data set ( equation ) .
it is also useful to use some filters that are not geometric , in that they do depend on a coordinate representation , such as coordinates in principal component or projection pursuit analysis .
the parameters of the configuration were : resolution 60 , gain 3.0x , equalized on ( see figure ) @xcite .
@xmath10 @xmath11 the neural network used in our study had a three - layer , feed - forward architecture and was trained by using the back - propagation algorithm with the sigmoid activation function . according to this learning scheme , the network tries to adjust its weights so that for every training input it can produce the desidered output . during the training phase , the network is presented with pairs of input - output patterns : supervised learning .
it has been shown that this technique minimizes the mean squared error ( mse ) between the decider and the actual network output following an inteative gradient search technique .
the input of the network is a subset of the patient s dataset formed with all the patients but only with selected features by tda ( see table ) .
specifically our network had a hidden layer with 3 nodes , and an output layer with a single decision node .
the network was trained to output 1 if pe was present and 0 if not . in this study
, the network s output was interpreted as the probability of pe being present .
in addition , input data were scaled to -1 and + 1 .
the learning rate was selected to be 0.5 and the momentum coefficient to be 0.9 , the optimal number of iterations ( epochs ) have been found equal to 660 after the execution of 100 trials with the permutations of rows in the patients dataset .
topological data analysis has been performed using the iris software @xcite , properties of ayasdi inc .
iris uses the concepts of metric and lenses .
metric is a distance or similarity measure between point .
lenses are mathematical functions through which you see the data points . different lenses emphasizing different aspects of the dataset and different networks will be generated .
lens executes the division of data points in overlapping bins . through the bins ,
the data are clustered .
each cluster is represented by a node , default node size is proportional to the number of data points in the cluster .
iris connects nodes when corresponding clusters have common data points . during iris exploration
there are two parameters attached to the lens that allow you to control how the data is divided in the bins .
one of these is _ resolution _ which corresponds to the number of bins is put into : * low resolution : coarse view of data ; * high resolution : detailed view and possibly fragmented into singletons .
another lens parameter is _
gain_. gain controls how much overlap there is between bins . increasing the gain increases the number of edges in the output .
oversampling highlights relationship within the data .
finaly the commands _ equalize _ distributes the data points in the bins so that each bin has an equal number of data points .
the ann has been implemented by the rulex software suite @xcite .
the rulex software , developed and commercialized by impara srl , is an integrated suite for extracting knowledge from data through statistical and machine learning techniques .
an intuitive graphical interface allows to easily apply standard and advanced algorithms for analyzing any dataset of interest , providing the solution to classification , regression and clustering problems . to build classifiers
we used a number of graphical components provided by rulex 2.0 .
we utilized visualization and editing components to visualize and export the confusion matrix , the training and validation sets , the results of the classifier , to access statistical data ( e.g. covering , error and relevance ) .
cpr : clinical predictive rule , cad : computer aided detection , ann : artificial neural network , tda : topological data analysis , pe : pulmonary embolism , dvt : deep venous thrombosis , angio - ct : angiography computed tomography , k - s or ks : kolmogorov - smirnov test , roc : receiver operating characteristic , fp : false positive , tn : true negative , fn : false negative , tp : true positive , dd : doctor diagnosis , ad : artificial neural network dianosis , auc : area under curve , wbc : white blood count , mse : mean sqared error ,
damir herman and tanya petrossian are employees of ayasdi , inc . and hold stock in the company .
the authors declare that they have no competing interests .
emanuela merelli , aldo salvi , lorenzo falsetti and cinzia nitti andconceived the project , supervised the study .
lorenzo falsetti wrote the clinical background .
matteo rucco designed the experimental flowchart , performed the computer experiments and wrote the manuscript .
we acknowledge the department of internal and subintensive medicine of ospedali riuniti di ancona - torrette for the patients dataset , the financial support of the future and emerging technologies ( fet ) programme within the seventh framework programme for research of the european commission , under the fet- proactive grant agreement topdrim ( topology - driven methods for multievel complex systems ) , number fp7-ict-31812 , francesco vaccarino and giovanni petri from isi foundation for their introduction to algebraic topology and complex networks .
impara s.r.l . and
ayasdi inc . for their software products .
a special thanks is extendend to damir herman for his support during the tda .
+ [ fig : pplot ] blu lines : parallel plot for pathological patients .
red lines : parallel plot for healthy patients .
green line : difference between mean values of the previous parallel plot ; only d - dimer and wbc exhibit a relevant difference between two classes , but it is not enough to diagnose pulmonary embolism .
+ [ fig : irisnet ] iris highlighted two flares in the pathological group .
that means there are two cluster of patients in the pe population ( red groups ) , also iris detected some sub - groups among healthy patients , which present other diseases to be diagnosed properly even though they are not affected by pe ( blue groups ) .
+ [ fig : ann_roc ] comparison study of roc curves : the auc obtained with the new classifier is greater then the aucs from the other two cprs .
+ [ fig : geneva_roc ] roc curve of revised geneva score .
+ [ fig : wells_roc ] roc curve of wells score . | pulmonary embolism ( pe ) is a common and potentially lethal condition .
most patients die within the first few hours from the event . despite diagnostic advances , delays and underdiagnosis in pe
are common .
moreover , many investigations pursued in the suspect of pe result negative and no more than 10@xmath0 of the pulmonary angio - ct scan performed to confirm pe confirm the suspected diagnosis . to increase the diagnostic performance in pe , current diagnostic work - up of patients with suspected acute pulmonary embolism usually starts with the assessment of clinical pretest probability using plasma d - dimer measurement and clinical prediction rules .
one of the most validated and widely used clinical decision rules are the wells and geneva revised scores .
however , both indices have limitations .
we aimed to develop a new clinical prediction rule ( cpr ) for pe based on a new approach for features selection based on topological concepts and artificial neural network .
filter or wrapper methods for features reduction can not be applied to our dataset : the application of these algorithms can only be performed on datasets without missing data . alternatively , eliminating rows with null values in the dataset would reduce the sample size significantly and result in a covariance matrix that is singular .
instead , we applied topological data analysis ( tda ) to overcome the hurdle of processing datasets with null values missing data .
a topological network was developed using the iris software ( ayasdi , inc .
, palo alto ) .
the pe patient topology identified two flares in the pathological group and hence two distinct clusters of pe patient populations .
additionally , the topological netowrk detected several sub - groups among healthy patients that likely are affected with non - pe diseases . to be diagnosed properly even though they are not affected by pe , in a next study we will introduce also the survival curves for the patients .
tda was further utilized to identify key features which are best associated as diagnostic factors for pe and used this information to define the input space for a back - propagation artificial neural network ( bp - ann ) .
it is shown that the area under curve ( auc ) of bp - ann is greater than the aucs of the scores ( wells and revised geneva ) used among physicians .
the results demonstrate topological data analysis and the bp - ann , when used in combination , can produce better predictive models than wells or revised geneva scores system for the analyzed cohort .
the new cpr can help physicians to predict the probability of pe .
a web - version of the bp - ann has been published and it can be tested at the address _ http:\\cuda.unicam.it _ [ 1995/12/01 ] |
the recent atmospheric , solar , reactor and accelerator neutrino experiments have convinced us that neutrinos are massive particles .
however , the problem of absolute values of their masses is still waiting for a solution .
neutrino oscillations depend on the differences of neutrino masses , not on their absolute values
. apparently three kinds of neutrino experiments have a chance to determine the light neutrino masses : i ) cosmological measurements , ii ) the tritium and rhenium single @xmath0-decay experiments , iii ) neutrinoless double @xmath0-decay experiments .
the measurement of the electron spectrum in @xmath0-decays provides a robust direct determination of the values of neutrino masses . in practice ,
the most sensitive experiments use tritium @xmath0-decay , because it is a super - allowed transition with a low @xmath5-value .
the effect of neutrino masses @xmath6 ( @xmath7=1,2,3 ) can be observed near the end point of the electron spectrum , where @xmath8 .
@xmath9 is the electron kinetic energy .
a low @xmath5-value is important , because the relative number of events occurring in an interval of energy @xmath10 near the endpoint is proportional to @xmath11 .
the current best upper bound on the effective neutrino mass @xmath12 given by , @xmath13 has been obtained in the mainz and troitsk experiments : @xmath14 @xcite .
@xmath15 is the element of neutrino mixing matrix . in the near future
, the karslruhe tritium neutrino mass ( katrin ) experiment will reach a sensitivity of about @xmath16 @xcite . in this experiment the @xmath0-decay of tritium
will be investigated with a spectrometer taking advantage of magnetic adiabatic collimation combined with an electrostatic filter .
calorimetric measurements of the @xmath0-decay of rhenium where all electron energy released in the decay is recorded , appear complementary to those carried out with spectrometers .
the unique first forbidden @xmath0-decay , @xmath17 is particularly promising due to its low transition energy of @xmath18 and the large isotopic abundance of @xmath19 ( @xmath20% ) , which allows the use of absorbers made with natural rhenium .
measurements of the spectra of @xmath19 have been reported by the genova and the milano / como groups ( mibeta and manu experiments ) .
the achieved sensitivity of @xmath21 was limited by statistics @xcite .
the success of rhenium experiments has encouraged the micro - calorimeter community to proceed with a competitive precision search for a neutrino mass .
the ambitious project , called the `` microcalorimeter arrays for a rhenium experiment '' ( mare ) , is planned in two steps , mare i and mare ii .
mare i might reach a statistical sensitivity of 4 ev after 3 years of data taking @xcite .
mare ii is to challenge the katrin goal of 0.2 ev @xcite .
the aim of this paper is to derive the form of the endpoint spectrum of emitted electrons for the @xmath0-decay of @xmath19 , which is needed to extract the effective neutrino mass @xmath12 or to place a limit on this quantity from future mare i and ii experiments .
the ground - state spin - parity is @xmath22 for @xmath19 and @xmath23 for the daughter nucleus @xmath24 , and the decay is associated with @xmath25 ( @xmath26 , @xmath27 ) of the nucleus , _
i.e. _ , classified as first unique forbidden @xmath0-decay .
the emitted electron and neutrino are expected to be , respectively , in @xmath28 and @xmath29 states ( see appendix a ) or vice versa . the emission of higher partial waves is strongly suppressed due to a small energy release in this nuclear transition . or @xmath30 ) for emission of @xmath29 and @xmath28 electrons vs. electron energy @xmath31 for @xmath0-decay of @xmath19 . ]
the differential decay rate is a sum of two contributions associated with emission of the @xmath29 and the @xmath28 state electrons . by considering the finite nuclear size effect the theoretical spectral shape of the @xmath0-decay of @xmath32
is @xmath33 \nonumber\\ & & ~~\times \sqrt{(e_0 - e_e)^2-m^2_{k}}~~ \theta(e_0-e_e - m_k ) \label{eq.sp}\end{aligned}\ ] ] with @xmath34 @xmath35 is the fermi constant and @xmath36 is the element of the cabbibo - kobayashi - maskawa ( ckm ) matrix .
@xmath37 , @xmath31 and @xmath38 are the momentum , energy , and maximal endpoint energy ( in the case of zero neutrino mass ) of the electron , respectively . @xmath39 and @xmath40 in eq .
( [ eq.sp ] ) are relativistic fermi functions and @xmath41 is a theta ( step ) function . @xmath42
denotes an axial - vector coupling constant .
@xmath43 is a coordinate of the @xmath44-th nucleon and @xmath45 is a nuclear radius . the value of nuclear matrix element @xmath46 in eq .
( [ nme ] ) can be determined from the measured half - life of the @xmath0-decay of @xmath19 .
-decay for various values of the effective neutrino mass : @xmath47 and @xmath48 ev . ]
the @xmath0-decay rate of rhenium is a sum of decay rates for emission of @xmath28 and @xmath29 electrons [ see eq . ( [ eq.sp ] ) ] . in fig .
[ fig:1 ] we show the single electron differential decay rate normalized to the particular decay rate as a function of electron energy @xmath31 . the two possibilities offer a different energy distribution of outgoing electrons .
close to the end point , there is a more flat distribution for @xmath29 electrons due to the dependence on squared neutrino momentum @xmath49 [ see eq .
( [ eq.sp ] ) ] .
because of this factor the two particular decay rates depend on the neutrino mass in a different way .
experimentally , it was found that @xmath28-state electrons are predominantly emitted in the @xmath0-decay of @xmath50 @xcite . by performing numerical analysis of partial decay rates associated with emission of the @xmath28 and the @xmath29 electrons ( terms with @xmath51 and @xmath52 in eq .
( [ eq.sp ] ) , respectively ) we conclude that about @xmath53 times more @xmath28-state electrons are emitted when compared with emission of @xmath29-state electrons . the reasons for it are as follows : i ) @xmath54 for @xmath55 ( see appendix b ) , ii ) the maximal momentum of electron ( @xmath56 kev ) is much larger than the maximal momentum of the neutrino ( @xmath57 kev ) .
henceforth , we shall neglect a small contribution to the differential decay rate given by an emission of the @xmath29-state electrons . for a normal hierarchy ( nh ) of neutrino masses with @xmath58 the kurie function of the @xmath0-decay of @xmath50
is given by @xmath59 ^{1/2 } , \label{kpre1}\end{aligned}\ ] ] with @xmath60 and @xmath61 .
the ratio @xmath62 depends only weakly on the electron momentum in the case of the @xmath0-decay of rhenium . with a good accuracy the factor @xmath63
can be considered to be a constant . the current upper limit on neutrino mass from tritium and rhenium @xmath0-decay experiments holds in the degenerate neutrino mass region ( @xmath64 with @xmath65 ) for which @xmath66 .
so far the future rhenium @xmath0-decay experiments will not see any effect due to a smallness of neutrino masses , it is possible to approximate @xmath67 ( k=1,2,3 ) and obtain @xmath68 where @xmath69 . in fig .
[ fig:2 ] we show the kurie plot for @xmath0-decay of @xmath19 versus @xmath70 near the endpoint .
we see that the kurie plot is linear near the endpoint for @xmath71 . however , the linearity of the kurie plot is lost if @xmath12 is not equal to zero .
we note that there is also a possibility of bound - state decay of @xmath19 , where an electron is placed into a bound state above occupied electron shells of the final atom . in ref .
@xcite it was shown that the ratio of decay rates of bound - state and continuum @xmath0-decays of @xmath19 is less than @xmath72 .
for the first unique forbidden @xmath0-decay of @xmath19 to ground state of @xmath24 the theoretical spectral shape is presented .
the decay rate of the process is a sum of particular decay rates associated with emission of @xmath29 and @xmath28 electrons , which depend in a different way on the neutrino mass .
the p - wave emission is dominant over the s - wave .
so , the kurie function , defined by eq .
( [ bre ] ) , is almost linear in the endpoint region .
an observed deviation from the linearity indicates effects of the finite neutrino mass .
the analysis of the the kurie function of the first forbidden @xmath0-decay of @xmath19 show that close to the endpoint it coincides up to a factor to the kurie function of superallowed @xmath0-decay of tritium @xcite .
these findings are important for the planned mare experiment , which will be sensitive to neutrino mass in the sub ev region .
this work was supported in part by the dfg project 436 slk 17/298 and by the vega grant agency under the contract no .
and @xmath73 vs. electron energy @xmath31 for the @xmath0-decay of @xmath19 . ]
we adopt the relativistic electron wave function in a uniform charge distribution in a nucleus , which is expanded in terms of spherical waves : @xmath74 where of particular interest are @xmath29 and @xmath28 states @xcite : @xmath75 \chi_s \\ { \tilde f}_{+2}(r ) [ ( \mathbf{\hat{r}\cdot\hat{p } } ) ( \mathbf{\sigma\cdot\hat{p } } ) - ( \mathbf{\sigma\cdot\hat{r } } ) ] \chi_s \end{array}\right).\nonumber\\\end{aligned}\ ] ] here , @xmath76 is a position vector , @xmath77 , @xmath78 and @xmath79 . by keeping the lowest power in expansion of @xmath80 the radial wave functions
take the form : @xmath81 in approximation up to @xmath82 terms we have @xmath83
relativistic fermi function @xmath84 takes into account the distortion of electron wave function due to electromagnetic interaction of the emitted electron with the atomic nucleus .
it takes the form @xcite : @xmath85 where @xmath86 and @xmath87 following ref .
@xcite it can be written as @xmath88 the constant @xmath89 is given by @xmath90 ^ 2 .
\nonumber\\\end{aligned}\ ] ] the function @xmath91 takes the form @xmath92 we note that @xmath93 the fermi functions @xmath94 and @xmath73 are related to the emission of @xmath29 and @xmath28 electrons , respectively . in fig .
[ fig:3 ] , they are plotted as functions of the kinetic energy of electrons emitted in @xmath0-decay of @xmath95 .
we note that values of @xmath73 are significantly larger than those of @xmath94 .
a quantity of interest is the ratio @xmath96.\end{aligned}\ ] ] as the @xmath5-value of rhenium @xmath0-decay is only @xmath97 kev the above ratio can be to a good accuracy considered to be a constant with the value @xmath98 .
99 e.w . otten , c. weinheimer , rept .
prog . phys . * 71 * , 086201 ( 2008 ) .
katrin collaboration , a. osipowicz _
et al . _ , arxiv : 0109033 [ hep - ex ] ; l. bornschein _
et al . _ ,
a * 752 * , 14 ( 2005 ) ; c. weinheimer , nucl .
. suppl . * 168 * , 5 ( 2007 ) .
mare collaboration , e. andreotti _
et al . _ ,
nucl . instrum . meth .
a * 572 * , 208 ( 2007 ) ; a. nucciotti , arxive : 1012.2290 [ hep - ex ] . c. arnaboldi _ et al . _ ,
lett . * 96 * , 042503 ( 2006 ) .
williams , w.a .
fowler , and s.e .
koonin , apj * 281 * , 363 ( 1984 ) .
f. imkovic , r. dvornick and a. faessler , phys .
c * 77 * , 055502 ( 2008 ) .
m. doi , t. kotani , and e. takasugi , prog .
( supp . ) * 83 * , 1 ( 1985 ) .
m. doi , t. kotani , prog .
phys . * 87 * , 1207 ( 1992 ) . | the planned rhenium @xmath0-decay experiment , called the `` microcalorimeter arrays for a rhenium experiment '' ( mare ) , might probe the absolute mass scale of neutrinos with the same sensitivity as the karlsruhe tritium neutrino mass ( katrin ) experiment , which will take commissioning data in 2011 and will proceed for 5 years . we present the energy distribution of emitted electrons for the first unique forbidden @xmath0-decay of @xmath1 .
it is found that the @xmath2-wave emission of electron dominates over the @xmath3-wave . by assuming mixing of three neutrinos the kurie function for the rhenium @xmath0-decay
is derived .
it is shown that the kurie plot near the endpoint is within a good accuracy linear in the limit of massless neutrinos like the kurie plot of the superallowed @xmath0-decay of @xmath4 . |
gravitationally lensed , multiply imaged radio source time delays are used to constrain the hubble constant .
lensing analyses produce a range of values , some of which are significantly lower than the result of the hst key project ( 72 @xmath0 8 kms@xmath1 mpc@xmath1 ) ; @xcite offers an excellent summary of the current state of the field .
further , although there is a general agreement upon isothermality from a number of differently motivated studies ( e.g. @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) , modelling of a number of systems ( e.g. pg1115 + 080 , b1600 + 434 , he2149 - 2745 , and sbs1520 + 530 ) has shown that these prefer more centrally concentrated , non - isothermal , density profiles @xcite .
the slope of the lensing density profile and the value of hubble constant are heavily degenerate , and the number of constraints from galaxy lens systems small , making it unclear whether the inconsistencies in slope and hubble constant values indicate flaws in the measurement of both or only one of the two quantities .
while some recent statistical studies ( e.g. @xcite ) find values consistent with the consensus value , some unexplained inconsistencies remain among the varied results across the field , and whether these stem primarily from external environmental effects , intrinsic physical attributes of the lensing galaxy or universe , or differences in modeling techniques is still unclear .
a number of factors are known to influence the derivation of the hubble constant from galaxy lens systems . @xcite
have shown that lens models of four - image systems overestimate the hubble constant by up to 15% when neglecting the effect of the contribution from other group members on the potential of the lens galaxy , with an even higher discrepancy for two - image systems . additionally , tidal
stripping and shocking has been shown to induce transient fluctuations in the inner density profile of lens galaxies present in groups , and the resultant profile steepening can also contribute to erroneous hubble constant values if isothermality is assumed @xcite . outside the group environment ,
line - of sight structures can also provide contributions to the lens potential @xcite .
these effects can result in either an over or underestimate of the hubble constant by virtue of a contributed increase or decrease in the density profile slope .
the intrinsic shape of the lensing galaxy has also been suspected to have a direct bearing on hubble constant derivation .
@xcite recently demonstrated the importance of substructures and external perturbations in galaxy lensing ; furthermore , in an investigation with models of 35 galaxy lenses @xcite proposed that shape - modelling degeneracies ( e.g. caused by triaxiality ) could also contribute to changes in the time delays and hence the derived hubble constant values .
triaxiality is of particular interest because the dark matter halos of @xmath2cdm are predicted to be fully triaxial with axis ratios in the mass distribution as low as 0.4 ( e.g. @xcite ; @xcite ) , a prediction which observations have loosely confirmed ( e.g. @xcite ) .
moreover , it has been shown that the triaxiality of cdm halos can effect the overall lensing probabilities and relative number of different image configurations ( double , quadruple , naked cusp lenses ) ( @xcite ; @xcite ) .
triaxial studies in galaxy cluster weak lensing have further revealed that neglecting halo shape can result in significant over and underestimates of cluster concentration and mass @xcite .
it seems apparent then that triaxiality can affect a number of critical observables and derived quantities in both strong and weak lensing and as such can not be ignored .
however , due to the small number of constraints available in galaxy lensing analyses , halo shape is typically addressed only via an elliptical perturbation to the projected spherical lensing potential , ignoring altogether halo structure along the line - of - sight . in this paper
we investigate what errors this often necessary neglect of triaxiality in galaxy lensing analyses introduces in best fit hubble parameter values .
we fit an elliptical isothermal mass model to lensing systems of highly ( as predicted in simulations ) triaxial isothermal galaxy lenses to investigate the maximum discrepancies that can arise from fitting a simplified model to data that in fact originates from a fully triaxial lens .
we ask : to what extent can galaxy triaxiality explain the discrepancies between some lensing - derived values for the hubble constant and those of other methods ?
the outline of the paper is as follows : in 2 we introduce the softened triaxial isothermal model and its lensing properties , 3 goes on to present our analysis method and key results , and 4 draws some conclusions based upon our findings .
a standard @xmath2cdm cosmology with @xmath3 km / s / mpc , @xmath4 , @xmath5 is employed throughout .
to generate a full parameterisation for a softened triaxial isothermal halo we follow the procedure given first in @xcite and implemented for weak lensing in @xcite for a triaxial nfw .
we first generalise the spherical softened isothermal profile to obtain a density profile @xmath6 where @xmath7 is an effective einstein radius , @xmath8 a triaxial core radius , @xmath9 a triaxial radius @xmath10 @xmath11 and @xmath12 the minor : major and intermediate : major axis ratios , respectively , and @xmath13 the angular diameter distance between the lens and source and @xmath14 that between the observer and source .
the mass contained within radius @xmath9 is @xmath15.\label{eq : minr}\ ] ] the virial mass is defined as that mass contained within an ellipsoid containing a mean density 200 times the critical density @xmath16 at the redshift of the halo , @xmath17 ; combining that definition with ( [ eq : minr ] ) gives an expression for @xmath7 as a function of the virial radius @xmath18 : @xmath19}.\ ] ] the full derivation of the lensing properties of a triaxial halo is given by @xcite , and we summarise some of that work here .
the triaxial halo is projected onto the plane of the sky to find its projected elliptical isodensity contours as a function of the halo s axis ratios and orientation angles ( @xmath20 , @xmath21 ) with respect to the the observer s line - of - sight .
the axis ratio @xmath22 of the elliptical density contours is given by @xmath23 the elliptical radius by @xmath24 and the orientation angle @xmath25 on the sky by @xmath26 here @xmath27 where @xmath28 and @xmath29 general expressions for the lensing potential of a softened elliptical isothermal halo are given in @xcite ; we modify them to incorporate the geometry of the halo projection to establish a direct connection between a 3d halo and its lensing potential .
the lensing potential is @xmath30\nonumber\\ & & - \theta_e q s \ln \sqrt{\left(\gamma + s\right)^2 + \left(1 - q^2\right)\mathcal{x}^2}\bigg\},\label{eq : psi}\end{aligned}\ ] ] where @xmath31 @xmath32 @xmath33 @xmath34 where @xmath35 , @xmath36 , and @xmath8 are all angular distances measured in the image plane .
these analytic formulations were confirmed using the independent numerical methods for calculating lensing properties of elliptical density profiles by integrating over those of their spherical counterparts as given in @xcite and applied to a triaxial nfw halo in @xcite , as well as by brute force projection of the triaxial halo .
the time delays and image positions of strongly lensed systems are completely constrained by the light - travel time equation ( e.g. @xcite ) @xmath37\bigg[\frac{d_{l } d_{s}}{d_{ls}}\bigg ] \bigg [ \frac{1}{2 } \left|\vec { \theta}- \vec { \beta}\right|^{2 } - \psi(\vec { \theta})\bigg ] \label{eq : time}\ ] ] and its derivative @xmath38\bigg[\frac{d_{l } d_{s}}{d_{ls}}\bigg]\bigg[(\vec { \theta } - \vec { \beta } ) - \vec { \bigtriangledown}\psi\bigg],\label{eq : dtime}\ ] ] respectively , where @xmath39 and @xmath40 are angular positions in the lens and source planes respectively , @xmath41 is the lensing potential , and @xmath42 is the redshift of the lens .
examining these with some care shows that it is the image positions , and not the time delays between them , that set the normalisation of the model potential : when matching image positions one matches the stationary points of the time - delay surface , thus mandating the model satisfy only the lens equation @xmath43 .
this is observationally independent of the hubble parameter , for while these are all angular values , with @xmath44 , @xmath45 , and the physical distances @xmath46 are not known and can not therefore be used as a constraint on @xmath14 and @xmath47 . once the shape and normalisation of the potential is set by the image position constraints via the first derivative of the time delay surface ,
the normalisation factor @xmath48 is crucially important in setting the time delays between the multiple images , and it is here the hubble parameter can be constrained .
to generically model the impact of unaccounted for triaxial structure in isothermal galaxies we use a markov chain monte carlo ( mcmc ) sampling method to explore the posterior probability distribution obtained when fitting an elliptical power law model to a triaxial isothermal lens .
first , simulated multiple - image lens systems are generated by lensing through a triaxial isothermal halo according to eq .
[ eq : psi ] and the lens equation .
these simulated observations of image positions and time delays are then fit using a general 2d elliptical power law model governed by seven free parameters .
such models can not be constrained using two - image multiple - image systems , as these provide only 5 constraints ( 2 image positions vectors + 1 time delay ) ; 4-image systems fare better , providing 11 constraints ( 4 image position vectors + 3 time delays ) . however , even in 4-image galaxy lens systems , the data is typically able to meaningfully constrain only one of either the slope of the underlying galaxy potential or the hubble parameter , due to an oft - noted strong degeneracy between these two key parameters .
several papers have explored constraints on @xmath49 from multiple image time delays , and degeneracies with parameters describing the lens model , in particular the radial profile .
early work includes @xcite and @xcite .
later , @xcite consider analytic time delays and estimates of @xmath49 from a general family of lensing potentials which allows for angular structure .
their treatment was extended and generalised by @xcite .
@xcite proposes a semianalytic model to understand observed time delays , isolating the important physical properties of the lens models .
for example , building on the treatment of @xcite , he obtains scaling solutions for @xmath49 for general lenses which depend on the mean convergence inside annuli bounded by the multiple images . due to the strong degeneracy , in most of these and other analyses some assumption
is often made regarding the shape of the galaxy density profile and the hubble parameter then obtained under that assumption ( though simultaneous fits have been succesfully carried out in statistical studies , e.g. @xcite ) . most often the galaxy halo is taken to be isothermal with density profile slope @xmath50 , a well - motivated simplification derived from previous studies of galaxy halo profiles ( see eg .
however , for any one galaxy , it is quite unlikely that the slope of the profile will be exactly isothermal . given the strong degeneracy between the hubble parameter and the density profile slope , such systematic errors may be quite important . to avoid having to make such a potentially error - inducing assumption
, we adopt an mcmc fitting method that is capable of exploring the full posterior probability distribution and all its degeneracies .
mcmc methods employ a `` guided '' random walk that returns a sample of points representative of the posterior probability distribution ; the probability of a certain region of parameter space containing the true model is directly proportional to the density of points sampled in that region . from the distribution of sample points
the full posterior probability distribution is obtained , which is easily and direcly marginalized over to obtain fully marginalized most - probable parameter estimates for all parameters .
these most - probable parameter values reflect the full shape of the posterior probability distribution without assumption about the error distribution , and with full knowledge of the prior .
such methods have exploded in popularity recently , and there are several excellent references describing the method in detail , e.g. @xcite , @xcite .
simply put , the sampler functions by stepping through parameter space by taking random steps governed by a transfer function , usually a simple @xmath51-d gaussian , where @xmath51 is the number of parameters of the fitted model .
if the randomly - chosen next step is to a point of higher probability than the current position , the step is taken .
if the next step is to a point of lower probability , the step is taken with probability @xmath52 .
thus , the mcmc sampler spends most of its time in high probability regions , but can move `` downhill '' to regions of lower probability in order to explore the entire space and sample all regions of high probability . crucially
, this method is able to return a true representation of the full posterior probability distribution , regardless of the complexities of that probability distribution , e.g. , tight degeneracies , multiple islands of high probability , or a very flat distribution due to poor constraints from the data .
thus , we are able to fully understand the impacts of fitting 2d models to systems with 3d triaxial structure , without inducing errors by making other limiting assumptions . to implement the method , we must define the posterior probability function , defined in bayesian statistics as @xmath53 where @xmath54 is the likelihood of the data given the model parameters ( the standard likelihood ) , @xmath55 is the prior probability distribution for the model parameters ( e.g. an outside constraint on the slope of the density profile ) , and @xmath56 is a normalising factor called the _ evidence _ , of great value in comparing models of different classes and parameter types , but expensive to calculate and unnecessary for the accurate exploration of the posterior distribution .
the assignment of priors is often a controversial exercise , but in this case it is straightforward to assign weak , physically sensible priors that do not significantly affect our results .
we define the likelihood function as a standard exponential function of the differences between the model image positions and time delays and those of the `` observed '' data : @xmath57\nonumber\\ & \times & \pi_{i=1}^3\exp\left[-\frac{\left(\delta t_i - \delta t_{m , i}\right)^2}{2\sigma_t^2}\right],\end{aligned}\ ] ] and take the measurement errors in the image positions @xmath58 and time delays @xmath59 to respectively be @xmath60 and @xmath61 , reasonable values for strong lensing analyses . in our mcmc sampler
we employ a 7d two - sided gaussian transfer function , and use the covariance matrix of an early run to sample in an optimised basis aligned with the major degeneracies of the posterior ( of which there are several ! ) .
we tune the step sizes of the sampler to achieve an average acceptance rate of 1/3 in each basis direction , run three independent mcmc chains , started at randomly chosen positions in parameter space , for each lens system , and sample the distribution space until the standard var(chain mean)/mean(chain var ) indicator is less than 0.2 , indicating chain convergence .
we utilise the getdist package from the standard cosmomc ( @xcite ) distribution to calculate convergence statistics , parameter contours , and marginalized parameter estimates .
the highly probable region of parameter space is of very small volume compared to the that of the prior parameter space , and we find it is a significant time - saver to first follow an mcmc chain with very large step sizes to find regions of high probability , then restart a chain at this randomly found , but advantageously chosen , starting point with a step size tuned to the size and shape of the posterior distribution .
this does not upset the statistics of the mcmc method ; it is equivalent to throwing away points from the very long burn - in phase that would be required if the step size were kept tuned to the small steps best for exploring the high probability region . in every case
we fit a power - law lensing density profile model to the simulated lensed images and time delays , given by @xcite ( here written with a factor of @xmath62 modification to match a standard elliptical isothermal model when @xmath63 ) , with lensing potential @xmath64^{\eta/2},\ ] ] where @xmath65 is the axis ratio of the isopotential ellipses .
note that here the ellipticity is in the 2d potential , as opposed to our triaxial model in which it is introduced in the 3d isodensity surfaces . at high ellipticities ( @xmath66 ) where the projected axis ratio is small ( @xmath67 ) , near - isothermal elliptical lensing potential models break down , giving unphysical `` dumbbell''-shaped projected mass distributions ( @xcite ) .
however , in the regime of ellipticities we study here ( @xmath68 ) , this problem is insignificant , and we choose this elliptical potential model as representative of those most often fit in lensing analyses and for its analytical simplicity .
the potential is always less elliptical than the underlying mass distribution , such that the axis ratio @xmath65 is always greater than the axis ratio @xmath22 of the isodensity contours .
the underlying density profile steepens with decreasing @xmath69 , such that @xmath70 corresponds to a hubble density profile that goes as @xmath71 .
thus , the seven free parameters are the profile slope parameter @xmath69 , the effective einstein radius @xmath72 , the axis ratio of the model isopotential contours @xmath65 , the orientation angle of those contours @xmath20 , the source position components @xmath73 and @xmath74 , and the hubble parameter @xmath75 .
we begin our analysis with a simple case , similar in method to many typical lensing analyses , in which we set the slope parameter to a range of fixed values and carry out parameter fits , and examine the impacts of triaxiality in that simplified framework .
we then offer an analytic explanation for our numerical results . finally , we examine the problem using the fully generalised power - law model in which the slope is allowed to vary freely , to complete our understanding of the role of triaxiality and to better understand the larger degeneracies that govern the problem of hubble constant determination from strong lensing systems .
to begin to isolate the impacts of triaxiality in the problem of hubble parameter determination from strong galaxy lensing systems , and to allow for direct comparisons to standard methods employed in many analyses ( e.g. @xcite ) , we fix the density profile slope at one of three values and carry out an mcmc exploration of parameter space .
we fix the slope of the underlying halo to either isothermal ( @xmath76 ) , steeper than isothermal ( @xmath77 ) , or shallower than isothermal ( @xmath78 ) , approximating the errors between the true and model slopes sure to sometimes occur if the slope is externally constrained to avoid the strong degeneracy between it and the hubble parameter .
we then fit these models to lens systems of symmetric oblate and prolate triaxial lensing halos at @xmath79 with a 1pc core with axis ratios ( @xmath80 ) and ( @xmath81 , @xmath82 ) , respectively , each containing a mass of @xmath83 m@xmath84 inside the triaxial effective einstein radius @xmath85 .
we fit to each halo type at ten different orientations stepped from @xmath86 to @xmath87 ; at @xmath87 each halo looks maximally elliptical with projected isodensity axis ratio @xmath88 while at @xmath86 the projected profile is circular . when the projected ellipticity is small only 2 images are formed , and we discard these systems because they provide too few constraints to meaningfully study the hubble parameter , leaving 7 prolate systems and 5 oblate systems for study ( there are more 4-image prolate systems because prolate halos are more efficient lenses than oblate , such that their threshold for 4-image systems is at a lower ellipticity : see @xcite ) . at each orientation the halos lens a source at @xmath89 positioned inside the multiple - imaging region at @xmath90 .
the image positions are found via the lens equation using a standard rootfinder , and the time delays between images calculated via the time delay surface ( eq . [ eq : time ] ) ; central images are discarded due to their high level of demagnification , which makes them invisible in observations ( e.g. @xcite ) . the results are plotted in figure [ plot : figure1 ] .
crucially , when the slope of the fitted halo matches that of the underlying lens when they are both isothermal the hubble parameter is accurately returned in every case ! thus , when the slopes are accurately matched in lens modelling , triaxiality induces _ no error _ in estimates of the hubble parameter .
the triaxial cases follow the familiar trends observed when there is a mismatch between the model slope and the true slope of the lens : as e.g. @xcite has previously shown , fitting a model shallower than the true potential leads to underestimates of the hubble parameter , while overestimates are apparent when the fitted model has a steeper slope than the true lensing halo .
the error in the returned hubble parameter is worst when the visible axis ratio is lowest and the lensing potential most elliptical .
figure [ plot : figure2 ] shows the posterior probability distribution when the power law model is fit to a prolate halo oriented at @xmath91 with the slope set to match the true value of the isothermal lensing halo @xmath63 .
the orientation angle of the isopotential contours @xmath20 is omitted for conciseness ; it shows no degeneracies with any other parameter and is very tightly constrained to the true value .
importantly , the only significant degeneracy visible is between the the components of the source position .
thus , when the slope is constrained accurately , a meaningful and accurate value for the hubble parameter can be derived from strong lens systems !
however , the slope must indeed be _
accurately _ constrained ; figure [ plot : figure1 ] shows clearly that a mismatch between the true slope value and the model slope value yields signficantly incorrect values for the hubble parameter . to complete this portion of the analysis
, we seek to further clarify any impact of triaxiality by calculating the best - fitting parameter values for oblate and prolate halos with a range of triaxial minor axis ratios @xmath92 , all oriented at @xmath91 so as to have high projected ellipticities , at the same three fixed values of @xmath69 . the results are plotted in figure [ plot : figure3 ] .
again , when the density profile slope is mismatched to the true underlying value , increased projected ellipticity exacerbates the error induced in the hubble parameter , with the best - fit value of @xmath75 approaching slightly closer to the true value as the minor axis ratio @xmath88 moves closer to 1 .
that this trend occurs both for halos of the same fundamental shape oriented at different angles and for halo of different shape oriented at the same angle , and for both prolate and oblate halos , indicates it is purely associated with the visible ellipticity , with no dependence on the ( very different ) 3d structure of the various halos .
we return to the point that there is essentially no difference in the value of the recovered hubble parameter if the halo is oblate or prolate . as noted
, the geometry of these two cases is quite different . when oriented at @xmath91 a prolate halo is like a rugby ball being looked at edge on its long axis is visible in the plane of the sky and there is relatively little mass along the line of sight .
conversely , the oblate halo is like a pancake looked at edge on , with its one short axis in the plain of the sky and a large amount of mass along the line of sight .
this difference in projected mass is reflected in the potential normalisation @xmath93 . however , the hubble parameter is in no way effected by this significant difference in 3d structure !
the line - of - sight structure of the lensing halo appears unimportant to the hubble parameter problem . to better understand this finding that hubble parameter estimation
is unaffected by the line - of - sight structure of triaxial halos , we look to the case of triaxial isothermal halos fitted with elliptical isothermal models , and more closely examine the expressions for the triaxial isothermal potential and its gradient , and their role in determing h@xmath95 in a strong lensing system .
the potential of the softened triaxial isothermal model was given in eq.[eq : psi ] ; its derivatives are @xmath96\nonumber \\ & -&\frac{\mathcal{y}^2\mathcal{x}q^2}{\gamma\left[\left(\gamma+q^2s\right)^2 - \left(1-q^2\right)\mathcal{y}^2\right]}\bigg\},\label{eq : dpsi1}\\ \frac{\partial \psi}{\partial \theta_y}&=&\frac{\theta_eqq_x}{\sqrt{f}}\bigg\{\frac{1}{\sqrt{1-q^2}}\tanh^{-1}\left(\frac{\sqrt{1-q^2}\mathcal{y}}{\gamma + q^2s}\right)\nonumber \\ & + & \frac{\mathcal{y}(\gamma+q^2s)}{\left(\gamma+q^2s\right)^2 - \left(1-q^2\right)\mathcal{y}^2}\left(1 - \frac{\mathcal{y}^2}{\gamma\left(\gamma+q^2s\right)}\right)\nonumber\\ & -&\frac{\mathcal{y}\left(\mathcal{x}^2 + s\gamma+s^2\right)}{\gamma\left[\left(\gamma+s\right)^2 + \left(1-q^2\right)\mathcal{x}^2\right]}\bigg\}.\label{eq : dpsi } \end{aligned}\ ] ] examining eqs .
[ eq : psi ] , [ eq : dpsi1 ] , and [ eq : dpsi ] we see that , when @xmath8 is small , the triaxial structure of the halo manifests itself only through multiplicative prefactors .
this is not a generic behaviour for all potentials ; it is true of this model because , when @xmath8 is small , all terms carry factors of the same power of @xmath97 , the normalising ratio that carries information about the length of the axis hidden along the line of sight .
all terms in the potential that are important when @xmath8 is small are proportional to @xmath98 or @xmath99 , giving each term an overall normalisation of @xmath97 , and similarly all important terms in its derivative are dimensionless in lengths giving each term a normalisation of @xmath97 . when @xmath8 is large , or were there any terms that carried unequal powers of @xmath97 , this simple multiplicative renormalisation due to triaxial structure would no longer hold .
however , given that there is good evidence that galactic cores are very small ( @xcite ) and that galaxy density profiles are generally isothermal over the central region to which strong lensing is sensitive , this straightforward analysis indicates that triaxiality will not independently lead to any significant change in the shape of the lensing potential and thus creates no significant errors in the measurement of the hubble parameter , as indicated in our simulations .
the multiplicative factors introduced by the triaxial structure of the lensing halo are apparent in the varied values of the normalisation of the potential @xmath100 returned for different orientations of the triaxial halos . to complete our understanding of the problem of hubble parameter estimation from triaxial lens systems
, we now look to the fully general case in which the elliptical power law model is fit with both hubble and slope parameters left free to vary .
we fit this general model to the same family of halos described in section [ sec : fs1 ] .
the complete posterior probability distribution for one of the halos , the prolate halo oriented at @xmath91 such that the ellipticity is approaching the maximum , is shown in figure [ plot : figure4 ] , again absent the well - constrained and independent orientation angle @xmath20 to make the plot clearer .
the well - known degeneracy between the hubble parameter @xmath75 and the slope parameter @xmath69 is clearly visible , but interestingly , is not the only strong degeneracy present .
both the hubble parameter and the density profile slope are also strongly degenerate with the normalisation of the potential @xmath93 , and with the source position * @xmath101*.
the regions of highest likelihood ( indicated by the shading ) are centred on the true parameter values ; however , the most - probable parameter values obtained by marginalizing over all other parameters give consistently high values for the hubble parameter and low values for the profile slope , indicating that while the true parameter values give a slightly higher likelihood value than others , there are other likely parameter value combinations including low values for @xmath69 and high values for @xmath75 that occupy significantly larger volumes in parameter space . were the errors in this problem gaussian , the maximum - likelihood parameter estimates should correspond to the most - probable marginalized parameter value ; however , as seen in figure [ plot : figure4 ] , the posterior probability distribution is highly non - gaussian , leading to the separation of the point of maximum likelihood from the point of maximum probability .
the marginalized most - probable parameter values represent the true best model ; thus , left unbroken , the degeneracies of the problem produce real errors in any lensing result .
to better picture the way these degeneracies function , figure [ plot : figure5 ] plots the posterior probability distribution in the @xmath69-@xmath75 plane , coloured by , respectively , each point s @xmath93 , @xmath65 , @xmath20 , and @xmath73 value .
these clearly show that the degeneracy between the slope and hubble parameter is tightly connected to the normalisation of the potential @xmath93 , the axis ratio @xmath65 of the isopotential contours , and the source position * @xmath101 * , but is completely independent of the orientation angle @xmath20 .
thus , we should not speak just of a @xmath69-@xmath75 degeneracy , but rather of a strong @xmath69-@xmath75-@xmath93-@xmath65-@xmath102 degeneracy !
this demonstrates that 4-image lens systems can not meaningfully constrain the hubble parameter with significance unless something can be done to at least partially break this degeneracy .
constraints on the profile slope or normalisation seem the best hope , as it impossible to learn anything of the source position by methods other than lensing .
the shape of this degeneracy now makes clear the source of the error in hubble parameter and slope returned when the slope is fixed to the wrong value , as in figure [ plot : figure1 ] ; setting the slope is equivalent to drawing a line at a particular value of @xmath69 in the scatter plots of figure [ plot : figure5 ] and taking only the points that fall on that line .
if the line is taken at the wrong place , completely adequate models will be found that will return completely false values for the other model parameters .
figure [ plot : figure6 ] plots the marginalized mean best - fitting and the maximum likelihood values for the hubble parameter @xmath75 , the profile slope parameter @xmath69 , the potential normalisation @xmath93 , and the axis ratio of the isopotential ellipses @xmath65 obtained for the prolate and oblate halos as a function of orientation angle . we find that the most - probable values of @xmath75 and @xmath69 vary inversely , with @xmath75 getting smaller and @xmath69 larger with increasing visible ellipticity .
the potential normalisation @xmath93 decreases significantly for the prolate halos and increases significantly for the oblate halos with increasing ellipticity ; this is as expected , because in the maximum ellipticity orientation prolate halos have minimal mass hidden along the line - of - sight , and oblate halos maximum , and the converse in the circularly - symmetric orientation at @xmath103 .
this result is consistent with the behaviour found in previous analyses that weak lensing mass estimates are strongly effected by unaccounted - for triaxiality @xcite .
the values of the axis ratio of the isopotential contours @xmath65 exhibit the trend expected , decreasing as the visible ellipticity increases from orientation @xmath103 to @xmath104 . to better understand what is happening to the posterior probability distribution as the triaxial halo orientation is changed , in figure [ plot : figure8 ] we plot the 1-@xmath105 ( 68@xmath106 ) confidence contours for @xmath69 , @xmath93 , @xmath65 , and @xmath73 against @xmath75 for prolate and oblate halos at several orientations , again with visible ellipticity increasing with rotation angle .
for both oblate and prolate halos , the @xmath69-@xmath75 contours maintain approximately the same position , but are tighter at higher visible ellipticities .
the least elliptical case alos shows an extention of its contour to lower values of @xmath69 and higher values of @xmath75 .
the contours for the potential normalisation @xmath93-@xmath75 move down to lower normalisations for the more elliptical prolate halos , and up to higher normalisations for more elliptical oblate halos , as expected given the line - of - sight masses of halos of these geometries .
further , they decrease in size as ellipticity increases , as with the @xmath69-@xmath75 contours . for both prolate and oblate halos the contours for isopotential axis ratio @xmath65-@xmath75 move down to lower values as the ellipticity increases , as expected .
the @xmath73-@xmath75 contours behave very similarly to those for @xmath75-@xmath69 , maintaining similar positions in parameter space but changing size . to understand the positioning and size of these contours we consider the nature of the multi - variable degeneracy .
it lies in parameter space such that high values of the axis ratio @xmath65 correspond to high values of the slope parameter @xmath69 , and low values for the hubble parameter @xmath75 and source position @xmath102 .
now , for halos with low amounts of visible ellipticity and thus true values of @xmath65 near 1 , there is less space in parameter space for models at values of @xmath65 higher than the true value , because the posterior distribution is truncated with a hard wall of zero probability at @xmath107 .
thus , the contours will extend proportionally more into the regions of lower than true @xmath65 , corresponding to lower values of @xmath69 and higher values of the hubble parameter @xmath75 and source position @xmath102 .
this trend is exactly what is seen as we move from the low-@xmath65 halos ( rotation angles near @xmath104 ) to the high-@xmath65 halos aligned closer to the circularly symmetric ( @xmath103 ) position .
in addition to their position , the high-@xmath65 halos also produce larger 1@xmath105 contours .
this change in contour size can not be due to the lensing strength : in strong lensing , so long as the same number of images are resolved , the relative strength of the lens and the scale of the image separations does not change the amount of information available to constrain the lens system .
this is born out in the results , in that it is the high-@xmath22 models in both oblate and prolate cases that exhibit larger contours ; in the prolate case these are indeed the stronger lenses with higher projected masses , but in the oblate case they are the weaker lenses , with very low lensing convergences .
instead , this again appears to be the natural product of the truncation of the posterior probability distribution at @xmath107 , which shifts and extends the 1@xmath105 region towards lower values of @xmath65 and @xmath69 and to higher values of @xmath75 .
this effect not an error or a problem but a true representation of the shape of the posterior probability distribution is illustrated and explored in more detail in appendix [ sec : appa ] .
again , to understand the importance of the degree of triaxiality , now in the general case , we carry out the same analysis as above but for prolate and oblate halos of axis ratios @xmath92 , all oriented at @xmath91 .
figure [ plot : figure9 ] plots the marginalized best - fit parameters @xmath75 , @xmath69 , @xmath93 , and @xmath65 as a function of minor triaxial axis ratio @xmath88 for both prolate and oblate halos .
again , increasing the amount of triaxiality does _ not _ significantly change the best - fitting slope or hubble parameter within the errors .
there is a weak trend toward there being larger offsets from the true value at _ smaller _ visible ellipticities , the opposite of that seen in the fixed - slope case .
this trend can be understood as an extension of that seen for this general case as a function of orientation angle ; again , the 1@xmath105 region is shifted towards lower values of @xmath65 and @xmath69 and to higher values of @xmath75 for lower visible ellipticity cases ; again see appendix [ sec : appa ] for further discussion .
these results represent those of the most general analysis , taking wide , flat priors on all parameters ( @xmath108 ; @xmath109 ; @xmath110 ; @xmath111 ; * @xmath101 * and @xmath20 unconstrained ) . however , in most cases , it is possible and indeed preferable to impose a physically realistic prior on the density profile slope , as work on galaxy profiles converges to show that nearly all galaxies are very close to isothermal . to understand the impact on our results of including this physically motivated
prior we impose two different gaussian priors on the slope parameter @xmath69 , centred on the isothermal value @xmath63 , with @xmath112 and @xmath113 .
the marginalized best - fit parameters found under these prior are shown in figure [ plot : figure7 ] .
as expected , the best - fitting parameters migrate towards the true values of the system when the prior is tightened on the slope , as the @xmath114 degeneracy is partially broken in favour of the true values of the lensing model employed in the simulations .
we note a significant offset from the true values in the most - probable marginalized parameters under a flat prior ; the hubble parameter estimates are uniformly high and the slope estimates uniformly low .
is this a product of triaxiality ? examining the results obtained when fitting to halos of different degrees of triaxiality at the same orientation ( as shown in figure [ plot : figure9 ] ) and those obtained when fits are carried out for fixed slope parameters ( as shown in figure [ plot : figure1 ] ) , it is clear the answer is no .
most importantly , the values of @xmath75 and @xmath69 obtained in the general model do not change significantly with the geometry of the halo oblate and prolate halos with the similar visible axis ratios give very similar results .
the visible axis ratio has some impact on parameter estimates due to the changing shape of the posterior probability distribution with respect to the hard wall at @xmath115 , but the line - of - sight structure has no impact .
thus , the offset in parameters can not be due to the 3d shape of the halo .
further , it may seem surprising that the most - probable marginalized parameters under a flat prior do not include the true values within their 1@xmath105 error bars ; however , the reliability of the result is underlined by the steady movement of the best - fitting parameter values to the true underlying values as an ever tighter prior is imposed on the profile slope , culminating in the return to the true value when a @xmath116-function prior is imposed and the slope set to isothermal .
thus , it is not triaxiality , but the large @xmath69-@xmath75-@xmath93-@xmath65-@xmath102 degeneracy , that causes the offset in parameter values .
the shapes of the parameter contours tell us that , with a flat prior , a much larger volume of high - likelihood models reside at high values of @xmath75 , low values of @xmath69 , low values of @xmath93 , and low values of @xmath65 in parameter space .
the effect is strongest for low projected ellipticity halos with values of @xmath65 near 1 due to the hard wall at @xmath107 , as the contours are pushed proportionally even more to low values of @xmath65 and thus lead to even higher best - fitting values of @xmath75 and lower values of @xmath69 .
thus the high values of the hubble parameter are not errors , per se , but the true best estimates of the parameter if the degeneracies in the problem are unbroken .
it is therefore crucial to break the degeneracy in order to obtain meaningful parameter estimates .
both cluster and galaxy halos are predicted to exhibit significant triaxiality ; in this study we have assessed the impact of neglecting this predicted halo shape on hubble constant estimation in galaxy strong lensing analyses , and find it is negligible .
fitting elliptical power law models to triaxial oblate and prolate halos using a flexible mcmc technique shows that , when the density slope of the model matches that of the lensing halo , correct hubble parameter values are always recovered within the errors without bias , a result that we further explore and explain analytically . in the case of mis - matched slopes the presence of projected elliptical structure may slightly increase the error in the recovered hubble parameter values as compared to more spherical cases .
however , prolate and oblate halos of the same projected axis ratio exhibit exactly the same behaviour , in spite of very different line - of - sight geometry , indicating that it is increased projected ellipticity and the resultant shifting of the posterior towards relatively low values of @xmath65 that is significant , not 3d triaxial structure .
thus , we find that halo triaxiality can not contribute to the inconsistencies between some lensing - derived values of h@xmath117 and those of the hst key project
. note that we have considered very triaxial models motivated by ( dark matter ) n - body simulations , which would represent some of the most triaxial galaxies even when baryonic matter ( which plays a crucial role on the scales relevant to strong lensing , see e.g. @xcite ) is included .
the hubble parameter as derived by strong lensing is degenerate not only with the slope of the lensing density profile , but also with the normalisation of the lensing potential , the projected ellipticity of the lens , and the source position . accurately constraining
the profile slope successfully breaks this significant degeneracy .
however , incorrectly constraining the slope , for example taking an average value across a population as true for a given system , will lead to well - fitting models with incorrect hubble parameter values , as the degeneracy is broken at the wrong point .
this problem is best resolved by carrying out statistical studies with many lens systems , in which average values of the profile shape are likely to be correct when applied across a population .
conversely , if the hubble parameter can be considered well constrained , then strong lensing can be successfully used for profile - type determination ( i.e. nfw vs. isothermal ) without requiring the use of complex triaxial models in the fitting process , as our work shows that neither the hubble parameter or density profile slope estimates are biased by neglected triaxiality .
past studies have concluded that certain effects can exacerbate the inconsistencies between lensing - derived hubble parameter values and the hst value ; for example , @xcite showed that neglect of environment can cause an _ overestimate _ in h@xmath118 values , the opposite trend of that required to account for the low values observed in many lensing systems .
thus , the remaining factors proposed and still under investigation that may lead to underestimates of the hubble constant are placed under even greater pressure to explain the discrepancy .
one such promising factor is halo substructure , its effects recently demonstrated by @xcite ; another was highlighted when @xcite showed that , despite findings that in general galaxies strongly tend towards isothermal profiles ( @xcite ) , group interactions many result in profile steepening that would lead to an underestimate of h@xmath118 should an isothermal slope be assumed .
out work shows that this effect may be slightly enhanced by highly elliptical lenses , but the additional impact will be small .
overall , triaxiality can not explain the difficulties encountered in reconciling disparate lensing results for the hubble parameter with those of other techniques ( e.g. @xcite , @xcite , @xcite ) .
however , the monster degeneracies of the general power law model explored in this paper emphasise one reason why achieving consistent lensing results is extremely difficult without a good external constraint on the density profile slope .
large samples of multiply imaged sources with measured time delays , deep optical and infrared imaging and spectroscopic data will soon become available .
future space and ground based facilities such as gaia , lsst , des , and further down the line , dune and the ska , will greatly enhance our knowledge of the physical and environmental properties of lensing galaxies , may generate lensing samples large enough to allow cleaner studies using only galaxies in isolated environments , and will allow statistical studies of the hubble parameter in large numbers of lensing systems ( e.g. , hundreds of systems in des ) in which the strong degeneracies of the general model parameters can be broken , greatly clarifying galaxy lensing s implications for cosmology .
we thank the marshall foundation , the national science foundation , and the cambridge overseas trust ( vlc ) , pparc through a phd studentship ( bmd ) , and the royal society ( ljk ) for supporting this work .
we thank antony lewis for very helpful discussions and our anonymous referee for a very useful review process .
99 bak , j. & statler , t. 2000 , apj , 120 , 110 barkana , r. 1998 , apj , 502 , 531 bett p. , eke v. , frenk c. , jenkins a. , helly j. , navarro j. 2007 , mnras , 376 , 215 corless , v. & king , l. 2007 , mrnas , 380 , 149 debattista , v. , et al . , 2007 , submitted to apj ( astro - ph/0707.0737 ) dobke b. m. , king l. j. , 2006 , a&a , 460 ,
647 dobke b. m. , king l. j. , m. fellhauer , 2007 , mnras , 377 , 1503 gerhard o. , kronawitter a. , saglia r. p. , bender r. , 2001 , aj , 121 , 1936 gorenstein , m.v . , falco , e.e . , shapiro , i.i . , 1988 ,
apj , 327 , 693 jing y. p. , suto y. 2002 , apj , 574 , 538 kassiola , a. & kovner , i. 1993 , apj , 417 , 450 keeton c.r .
2001 , astro - ph/0102341 keeton c. r. , zabludoff a. i. 2004 , apj , 612 , 660 kochanek c. s. 2002 , apj , 578 , 25 kochanek c. s. 2003 , apj , 583 , 49 kochanek c. s. , schechter p. l. , 2004 , ` measuring and modelling the universe ' , carnegie observatories centennial symposia . published by cambridge university press , as part of the carnegie observatories astrophysics series .
edited by w. l. freedman kochanek , c. s. , morgan , n.d . , falco , e. e. , mcleod , b. a. , winn , j. n. , dembicky , j. , ketzeback , b. 2006a , apj , 640 , 47 kochanek , c. s. , schneider , p. , wambsganss , j. 2006b in gravitational lensing : strong , weak , & micro , part 2 , ed .
g. meylan , p. jetzer , p. north ( berlin : springer ) , 91 koopmans l. v. e. , treu t. , fassnacht , c. d. , blandford , r. d. , surpi , g. 2003 , apj , 599 , 70 koopmans l. v. e. , treu t. , bolton a. s. , burles s. , moustakas l. a. , 2006 , apj , 649 , 599 lewis , a. & bridle , s. 2002 , phrevd , 66 , 103511 macci a.v . , dutton a. a. , van den bosch f. c. , moore b. , potter d. , stadel j. , 2007 , mnras , 378 , 55 mackay , d. 2003 , _ information theory , inference , and learning algorithms _ ( cambridge university press : cambridge , uk ) momcheva i. , williams k. , keeton c. , zabludoff a. , 2006 , apj , 641,169 oguri , m. , lee , j. , suto , y. 2003 , apj , 599 , 7 , astro - ph/0306102 oguri m. , keeton c. r. , 2004 , apj , 610 , 663 oguri m. 2007 , apj , 660 , 1 refsdal , s. , surdej , j. , 1994 , rep . prog .
, 57 , 117 rix h .- w . ,
de zeeuw p. t. , cretton n. , van der marel r. p. , carollo c. m. , 1997 , apj , 488 , 702 romanowsky a. j. , kochanek c. s. , 1999 , apj , 516 , 18 rozo , e. , chen , j. , zentner , a. r. ( astro - ph/0710.1683 ) saha p. , coles j. , macci a. v. , williams l. l. r. 2006 , apj , 650 , l17 saha p. , williams l. l. r. , 2006 .
apj , 653 , 936 schneider , p. , ehlers , j. , falco , e.e . , 1992 , _ gravitational lenses _
( springer - verlag : new york ) treu t. , koopmans l. v. e. , 2002 , apj , 575 , 87 wallington s. , narayan r. , 1993 , apj , 403 , 517 witt , h.j . , mao , s. , keeton , c.r . , 2000 , apj 544 , 98 wucknitz , o. , 2002 , mnras , 332 , 951
when fitting the fully general power - law model , we note a trend in which the most - probable parameter values move closer to the true values with increasing projected ellipticity .
we credit this trend to the presence of a hard wall in the prior on the axis ratio @xmath65 at @xmath107 , beyond which the equations governing the lensing potential are undefined . in
projected halo geometries with low ellipticities ( high @xmath65 values ) , this wall limits the volume of parameter space available at values of @xmath65 higher than the true value .
this is not an error ; it reflects the true shape of the posterior probability distribution .
but it does affect the shape and positioning of the resulting isoprobability surfaces , pushing them towards lower values of @xmath65 relative to the true value , and due to the shape of the @xmath69-@xmath75-@xmath93-@xmath65-@xmath102 degeneracy , to higher values of @xmath75 and @xmath102 and lower values of @xmath69 . that this effect depends only on the _ projected _ shape of the potential and not on the 3d geometry
is clear when the contours of prolate halos of one geometry projected at different orientation angles are compared to the contours of prolate halos of different underlying geometries ( a range of minor axis ratios ) all projected at the same orientation .
the @xmath119 plane 1@xmath105 contours for these two cases are shown in figure [ fig : app1 ] . in both cases ,
the projected ellipticity varies from high to low , but the underlying geometry leading to that change is fundamentally different between the two .
however , the exact same trend is visible , with the low ellipticity , high-@xmath65 cases exhibiting larger contours extended toward high values of @xmath75 and low values of @xmath69 . to further confirm that this behaviour is indeed attributable to the necessary imposition of a hard wall of zero probability at @xmath107
, we look for the opposite effect in the high ellipticity , low @xmath65 cases , by artificially imposing a hard wall of zero probability at @xmath120 .
if our explanation is correct , the impositions of these priors should produce a similar but opposite effect in the resulting @xmath119 contours , shifting them to lower values of @xmath75 and higher values of @xmath69 .
the results for both the constant geometry and constant orientation cases are shown in figure [ fig : app2 ] , and indeed the contours behave as expected , with the contours moving more and more to higher values of @xmath69 as the prior is tightened .
thus , the trends seen when lensing through both halos of the same geometries oriented to exhibit different projected ellipticities ( figures [ plot : figure6 ] & [ plot : figure8 ] ) and of different geometries oriented similarly ( figure [ plot : figure1 ] ) that systems with projected axis ratios close to @xmath121 return parameter values further away from the true underlying value of the system than do their more elliptical counterparts is attributable to the real shape of the posterior probability distribution and is a function only of projected ellipticity and independent of 3d structure . | the hubble constant can be constrained using the time delays between multiple images of gravitationally lensed sources . in some notable cases , typical lensing analyses assuming isothermal galaxy density profiles produce low values for the hubble constant , inconsistent with the result of the hst key project ( 72 @xmath0 8 kms@xmath1 mpc@xmath1 ) .
possible systematics in the values of the hubble constant derived from galaxy lensing systems can result from a number of factors , e.g. neglect of environmental effects , assumption of isothermality , or contamination by line - of - sight structures .
one additional potentially important factor is the triaxial structure of the lensing galaxy halo ; most lens models account for halo shape simply by perturbing the projected spherical lensing potential , an approximation that is often necessary but that is inadequate at the levels of triaxiality predicted in the cdm paradigm . to quantify the potential error introduced by this assumption in estimates of the hubble parameter
, we strongly lens a distant galaxy through a sample of triaxial softened isothermal halos and use an mcmc method to constrain the lensing halo profile and the hubble parameter from the resulting multiple image systems .
we explore the major degeneracies between the hubble parameter and several parameters of the lensing model , finding that without a way to accurately break these degeneracies accurate estimates of the hubble parameter are not possible .
crucially , we find that triaxiality does _ not _ significantly bias estimates of the hubble constant , and offer an analytic explanation for this behaviour in the case of isothermal profiles .
neglected triaxial halo shape can not contribute to the low hubble constant values derived in a number of galaxy lens systems .
[ firstpage ] gravitational lensing galaxies : fundamental parameters galaxies : halos cosmology : cosmological parameters cosmology : theory dark matter |
this work was supported by the ` karatheodory ' research fund of university of patras . 00 c. m. cornelius and j. p. dowling , , 4736 ( 1999 ) .
s. y. lin , j. g. fleming , e. chow , j. bur , k. k. choi , and a. goldberg , , r2243 ( 2000 ) . j. g. fleming , s. y. lin , i. el - kady , r. biswas , and k. m. ho , , 52 ( 2002 ) . m. u. pralle , n. moelders , m.p .
mcneal , i. puscasu , a. c. greenwald , j. t. daly , e. a. johnson , t. george , d. s. choi , i. el - kady , and r. biswas , , 4685 ( 2002 ) .
s. y. lin , j. g. fleming , and i. el - kady , , 1909 ( 2003 ) .
i. celanovic , f. o sullivan , m. ilak , j. kassakian , and d. perreault , , 863 ( 2004 ) .
a. narayanaswamy , and g. chen , , 125101 ( 2004 ) .
s. enoch , j. -j .
simon , l. escoubas , z. elalmy , f. lemarquis , p. torchio , and g. albrand , , 261101 ( 2005 ) .
m. florescu , h. lee , a. j. stimpson , and j. dowling , , 033821 ( 2005 ) . h. sai , h. yugami , y. akiyama , y. kanamori , and k. hane , , 1471 ( 2001 ) . s. maruyama , t. kashiwa , h. yugami , and m. esashi , , 1393 ( 2001 ) .
f. marquier , k. joulain , j. p. mulet , r. carminati , and j. j. greffet , opt . commun . * 237 * , 379 ( 2004 ) . m. laroche , f. marquier , r. carminati , and j. j. greffet , opt . commun .
* 250 * , 316 ( 2005 ) .
f. kusunoki , t. kohama , t. hiroshima , s. fukumoto , j. takahara , and t. kobayashi , jpn .
. phys . * 43 * , 5253 ( 2004 ) .
m. zenker , a. heinzel , g. stollwerck , j. ferber , and j. luther , ieee trans .
electron .
devices * 48 * , 367 ( 2001 ) . c. w. hoyt , m. p. hasselbeck , m. sheik - bahae , r. i. epstein , s. greenfield , j. thiede , j. distel , and j. valencia , , 1066 ( 2003 ) .
n. stefanou , v. karathanos and a. modinos , j. phys . : condens .
matter * 4 * , 7389 ( 1992 ) .
n. stefanou , v. yannopapas , and a. modinos , comput .
. commun . * 113 * , 49 ( 1998 ) ; _ ibid _ * 132 * , 189 ( 2000 ) .
s. m. rytov , yu . a. kravtsov , v. i. tatarskii , _ principles of statistical radiophysics , vol . 3 : elements of random fields _ ( springer , berlin , 1989 )
. j. j. greffet and m. nieto - vesperinas , , 2735 ( 1998 ) . c. luo , a. narayanaswamy , g. chen and j. d. joannopoulos , , 213905 ( 2004 ) .
r. b. johnson and r. w. christy , , 4370 ( 1972 ) .
s. norrman , t. andersson , c. g. granqvist , and o. hunderi , phys .
b * 18 * , 674 ( 1978 ) .
f. abels , y. borensztein , and t. lpez - rios , _ festkrperprobleme ( advances in solid state physics ) _
vol 24 ( braunschweig : vieweg ) p.93(1984 ) .
m. m. wind , p. a. bobbert , j. vlieger , and d. bedeaux , physica a * 143 * , 164 ( 1987 ) .
p. a. bobbert and j. vlieger , physica a * 147 * , 115 ( 1987 )
. n. stefanou and a. modinos , j. phys . : condens .
matter * 3 * , 8135 ( 1991 ) .
n. stefanou and a. modinos , j. phys . : condens .
matter * 3 * , 8149 ( 1991 ) . v. yannopapas , a. modinos , and n. stefanou , , 5359 ( 1999 ) .
a. modinos , v. yannopapas , and n. stefanou , , 8099 ( 2000 ) . v. yannopapas , a. modinos , and n. stefanou , opt .
. elec . * 34 * , 227 ( 2002 ) .
k. p. velikov , w. l. vos , a. moroz , and a. van blaaderen , , 075108 ( 2004 ) .
nm ) in air ( @xmath47 - solid line ) , silica ( @xmath48 - dashed line ) and gelatine ( @xmath49 - dotted line ) , with @xmath32 .
we have assumed that the host medium surrounding the spheres also covers the whole space.,width=302 ] | we study the blackbody spectrum from slabs of three - dimensional metallodielectric photonic crystals consisting of gold nanoparticles using an ab initio multiple - scattering method .
the spectra are calculated for different photonic - crystal slab thicknesses , particle radii and hosting materials .
we find in particular that such crystals exhibit a broadband emission spectrum above a specific cutoff frequency with emissivity of about 90% .
the studied photonic crystals can be used as efficient selective emitters and can therefore find application in thermophotovoltaics and sensing .
the main feature of photonic crystals is the ability to tailor the photon density of states and this way control the spontaneous emission of light , aiming at the realization of new optoelectronic devices . in this context
, there has been considerable effort to design and fabricate photonic crystals which allow for control of thermal emission of light , i.e. thermally driven spontaneous emission , promising applications in imaging , sensing and most importantly , in thermophotovoltaics ( tpv ) .
@xcite control of thermal emission can also be achieved by means of microstructured engineering on silicon @xcite or metal surfaces . @xcite depending on the type of application , photonic crystals and structured surfaces can act as narrow- or wide - band , directional or isotropic thermal emitters .
for example , in tpv applications @xcite a quasi - monochromatic emission is preferable whilst in radiation cooling @xcite a broad emission spectrum is desired . in this work
we investigate the emission properties of three - dimensional metallodielectric photonic crystals consisting of gold nanoparticles .
we find , in particular , that the emission spectrum of these crystals can be such that photons are emitted in all directions only when their energies lie above a specific cutoff frequency , with emissivity as large as 90% .
photonic crystals of spherical scatterers have been theoretically studied using multiple scattering theory @xcite which is ideally suited for the calculation of the transmission , reflection and absorption coefficients of an electromagnetic ( em ) wave incident on a composite slab consisting of a number of planes of non - overlapping particles with the same two - dimensional ( 2d ) periodicity . for each plane of particles ,
the method calculates the full multipole expansion of the total multiply scattered wave field and deduces the corresponding transmission and reflection matrices in the plane - wave basis .
the transmission and reflection matrices of the composite slab are evaluated from those of the constituent layers .
having calculated these matrices one can evaluate the transmittance @xmath0 , reflectance @xmath1 , and from those two , the absorbance @xmath2 of the composite slab as functions of the incident photon energy @xmath3 and incident angles @xmath4 and @xmath5 .
transmittance and reflectance are defined as the ratio of the transmitted , respectively the reflected , energy flux to the energy flux associated with the incident wave .
the method applies equally well to non - absorbing systems and to absorbing ones . in terms of speed , convergence and accuracy ,
the multiple scattering method is the best method to treat photonic structures of spherical particles .
the emittance @xmath6 is calculated indirectly by application of kirchoff s law ,
@xcite i.e. from @xmath7 note that recent direct calculations of the thermal emission from photonic crystals have verified the validity of kirchoff s law for the case of photonic crystals
. @xcite in this work we deal with gold spheres with radius of a few nanometres .
traditionally , the gold spheres are treated as plasma spheres whose dielectric function is given by drude s formula @xmath8 where @xmath9 stands for the bulk plasma frequency of the metal and @xmath10 is the relaxation time of the conduction - band electrons .
the dielectric function @xmath11 for bulk gold has been determined experimentally by johnson and christy @xcite over the range from @xmath12 ev to @xmath13 ev .
taking @xmath14 ev and @xmath15 ev @xcite eq . ( [ eq : drude ] ) reproduces satisfactorily the experimentally determined @xmath16 over the whole of the above frequency range , and the experimentally determined @xmath17 at low frequencies .
@xmath18 refers to the collision time of conduction - band electrons in the bulk metal ; due to the finiteness of a nanosphere , the collision time @xmath10 of an electron in a sphere of radius @xmath19 is decreased due to scattering at the boundaries of the sphere .
it is given by @xmath20 where @xmath21 stands for the average velocity of an electron at the fermi surface . for gold @xmath22 ev@xmath23 nm ,
@xcite so that for a sphere of radius @xmath24 nm , using @xmath25 ev , we obtain @xmath26 ev . therefore , a more realistic description of the dielectric function of a gold sphere would be @xcite @xmath27 where @xmath28 is the experimentally determined dielectric function of bulk gold .
we note that @xmath29 as given above , is different from the approximation to the dielectric function of the gold sphere obtained from eq .
( [ eq : drude ] ) with @xmath15 ev and @xmath26 ev only because the imaginary part of the dielectric function is not represented well by eq .
( [ eq : drude ] ) for @xmath30 ev .
taking into account the finite size of a metal nanoparticle for the determination of the actual dielectric function , i.e. eq .
( [ eq : ecorr ] ) , has successfully reproduced experimentally obtained light absorption and scattering spectra of monolayers of gold nanoparticles .
@xcite to begin with , we consider an fcc photonic crystal whose lattice sites are occupied by gold nanospheres of radius @xmath24 nm .
the lattice constant of the crystal is @xmath31 nm corresponding to a volume filling fraction occupied by the spheres , @xmath32 .
the spheres are supposed to be suspended in air .
the crystal is viewed as a succession of planes of spheres ( layers ) parallel to the ( 001 ) surface of fcc . in fig .
[ fig1 ] we show the transmittance , reflectance and absorbance vs energy for light incident normally on a finite slab of the crystal consisting of 128 layers of spheres .
we observe that for energies above @xmath33 ev almost all light ( 92%-98% ) that is incident on the crystal slab is absorbed from the gold spheres .
therefore , from kirchoff s law , i.e. @xmath34 , we can infer that the blackbody radiation is strong above this energy .
indeed , the blackbody radiation intensity of the photonic crystal is given by @xmath35 where @xmath36 is the blackbody radiation intensity ( planck distribution ) @xmath37 @xmath38 is the boltzmann constant and @xmath39 is the speed of light in vacuum . in a metal particle of nanoscale radius , there are two principal sources of light emission/ absorption : the first one is the dipole oscillations due to the surface plasmon resonances and the second is the interband transitions that take place in bulk gold contributing to its dielectric function . in terms of the first source of emission , the crystal can be viewed as a lattice of interacting oscillating dipoles where each one of the dipoles emits light around the surface plasma resonance frequency @xmath40 .
the interaction between the dipoles leads to the creation of a band of resonances around @xmath41 . in two - dimensional systems , this band of resonances
is manifested as a strong absorption/ emission peak @xcite whilst in three - dimensional systems is manifested as two absorption/ emission peaks ( corresponding to two bands of resonances ) with a reflectivity peak ( stemming from a photonic band gap ) lying between them .
@xcite in fig . [ fig1 ]
we can not clearly identify surface plasmon peaks as these must have been submerged into the absorption edge due to interband transitions . in any case , both contributions ( plasmon+interband ) lead to a strong absorption/ emission plateau above 2.2 ev . as the system considered here
can potentially found application in tpv devices we are interested in the spectral hemispherical ( sh ) radiative properties of the photonic crystal .
more specifically , we are interested in the sh emissivity @xmath42 which is the ratio of the sh emissive power of a photonic - crystal slab to the sh emissive power of a perfect blackbody at the same temperature @xmath43 . for a slab infinitely extended in two dimensions : @xmath44 note that @xmath45 in the above equation is the arithmetical average of both polarization modes . in fig .
[ fig2 ] we show the sh emissivity from slabs of the photonic crystal of fig . [ fig1 ] for different slab thicknesses ( number of layers ) .
it is evident that as the number of layers increases the sh emissivity increases accordingly until it reaches a saturation plateau for slabs of 64 layers and above . indeed , for energies above 2.2 ev the emissivity curves for 64 and 128 layers are practically the same .
this fact implies that , if we measure the emission from one of the surfaces of a 128 layers - thick slab , then the emitted radiation must be coming from the 64 outmost layers ( relative to a given surface ) whilst radiation coming from the 64 innermost layers is almost totally absorbed from the 64 outmost layers by the time it reaches the surface of the slab .
the most important finding of fig .
[ fig2 ] is the fact that the photonic crystal behaves more or less as gray body ( emissivity ranging from 88% to 92% ) for energies above 2.2 ev while , at the same time , emits small amounts of radiation for energies below this cutoff .
a similar emission spectrum is observed for the salisbury screen @xcite but with significantly lower values of the sh emissivity .
so far , we have assumed that the gold nanospheres are suspended in air . in order to provide a manufacturable structure
it is necessary to examine the case where the spheres are embedded in a host material of dielectric function @xmath46 . in fig .
[ fig3 ] we show the sh emissivity from a 64-layer slab of an fcc photonic crystal of gold nanospheres with @xmath32 where the spheres are surrounded by air ( @xmath47 ) , silica ( @xmath48 ) and gelatine ( @xmath49 ) .
the introduction of a host material does not change the picture drastically except that it lowers the emission cutoff energy .
this is due to the lowering of the surface plasmon frequency according to @xmath50 when a nanosphere is placed in a host medium @xmath46 , which results in strong thermal emission at lower energies . as a result of the energy shift of the surface plasmon resonance ,
the corresponding emission peak moves away from the interband emission region and a distinct local minimum appears .
this is evident from the emissivity spectra of fig .
[ fig3 ] for silica and gelatine host materials .
the shift of the surface plasmon energy with respect to the host material @xmath46 allows for tuning of the cutoff energy so that it coincides with the band gap of the photodiode of a tpv device .
next we study the effect of the volume filling fraction @xmath51 on the emission properties of the photonic structure under study . in fig .
[ fig4 ] we show the sh emissivity spectra for photonic crystals of different values of @xmath51 .
we have kept the lattice constant the same and changed the sphere radii accordingly .
note that for each value of @xmath51 we obtain a different value of @xmath52 from eq .
( [ eq : tau ] ) .
we observe that the maximum emissivity in the region above the cutoff energy is achieved for @xmath32 .
since our calculations show that the sh reflectivity assumes higher values for @xmath53 , the radiation emitted from the innermost layers is reflected back and reabsorbed before it reaches the surface of the slab .
finally we address the issue of spatial order/ disorder of the photonic crystal under investigation .
as it has been both theoretically @xcite and experimentally @xcite shown , the presence of disorder does not changes practically the absorption/ emission properties of three - dimensional photonic crystals of metal nanoparticles since these quantities mostly depend on single sphere properties such as surface plasmon resonances , and less on the particular spatial arrangement of the spheres .
note that this not true for one- or two - dimensional structures , e.g. a linear chain or a monolayer of nanospheres , where the presence of disorder changes dramatically the absorption/ emission spectra .
@xcite so , since we deal with a three - dimensional structure , a disordered crystal might be easier and cheaper to fabricate . in summary , we have shown that a metallodielectric crystal consisting of gold nanospheres can act as a 90% gray body for energies above a cutoff .
the cutoff frequency can be tuned by appropriate choice of the material hosting the nanospheres .
optimal emissivity is obtained for moderate volume filling fractions and slab thicknesses . |
hadronization is the process through which partons , created in an elementary interaction , turn into the hadrons .
experimentally the hadronization process in free space ( vacuum ) has been studied extensively in @xmath7 annihilation and in semi - inclusive lepton - proton deep inelastic scattering ( dis ) . as a result , the spectra of hadrons produced and their kinematical dependences are rather well known . however
, little is known about the space - time evolution of the process , because the products of this process can only be observed in a detector that is separated from the reaction point by a macroscopic distance .
it is worth to mention that according to theoretical estimates the hadronization process occurs over length scales that vary from less than a femtometer to several tens of femtometers .
the nuclear medium can serve as a detector located directly at the place where microscopic interaction happens .
consequently , leptoproduction of hadrons on atomic nuclei provides a way to investigate the space - time picture of the hadronization process .
the semi - inclusive dis of leptons on nuclear targets is widely used for the study of this process @xcite .
it is most effective to observe at moderate energies of the virtual photon , when the formation time of the hadron is comparable with the nuclear radius .
such possibility provides hermes experiment at desy , which uses electron ( positron ) beam with energy @xmath8 and fixed nuclear targets .
+ the most convenient observable measured experimentally for this process is the nuclear attenuation ratio , which is a ratio of multiplicities on nucleus and deuterium ( per nucleon ) for a given hadron
. we shall denote it as @xmath2 . for a more profound understanding of the hadronization mechanism , it is important to find a variable which allows to present this observable in the most simple functional form . +
usually it is supposed , that @xmath2 is a function of two variables @xmath1 and @xmath0 , which are the energy of photon and the fraction of this energy carried by the final hadron with energy @xmath9 ( @xmath10 ) , respectively also depends on the photon virtuality @xmath11 and on the square of the hadron transverse momentum in respect to the virtual photon direction , @xmath12 .
however , from the experimental data , it is known that @xmath2 is a much sensitive function of @xmath1 and @xmath0 in comparison with @xmath11 and @xmath12 . ] .
+ in our preceding work @xcite we performed a fit for the evidence that the formation time @xmath3 is the best variable for @xmath2 , i.e. that it can be parametrized as a function of a single variable @xmath3 .
three widely known representations for @xmath3 were used for the fit . the experimental data for pions on nitrogen and for identified hadrons on krypton nuclei obtained by the hermes experiment @xcite were used for this fit .
we have demonstrated that the nuclear attenuation ratio can be presented , with good precision , as a function of a single variable @xmath3 instead of a function of two variables @xmath1 and @xmath0 .
moreover , @xmath2 is a linear function of @xmath3 .
we named @xmath3 a scaling variable , because it contains all @xmath1 and @xmath0 dependencies of @xmath2 .
for the fit we have obtained @xmath2 as a function of @xmath3 from experimental data @xcite , where it was measured as a function of @xmath1 with the integration over @xmath0 , and as a function of @xmath0 with the integration over @xmath1 .
this means , that the data were taken at an unequal binning over @xmath1 and @xmath0 . in case of @xmath1-dependence
the detailed bins over the variable @xmath1 were taken , and for each value of @xmath1 the value of @xmath0 averaged over whole range of measured @xmath0 ( @xmath13 ) was taken . in case of @xmath0-dependence
we also performed the above mentioned procedure .
+ recent work of published by the hermes experiment @xcite allows to escape this difficulty , because the data published contains , among others , the so called two - dimensional data , i.e. multiplicity ratio @xmath2 for charged pions produced in a @xmath0 slices as a function of @xmath1 , and in a @xmath1 slices as a function of @xmath0 .
the data were obtained for four nuclear targets and used for a new fit of @xmath2 as a function of @xmath3 .
+ the main aim of this work is to show , that @xmath2 is a function of single variable @xmath3 , rather than a function of two variables @xmath1 and @xmath0 , using the new set of @xmath2 given by the two - dimensional analysis , where the data is split into more regular @xmath1 and @xmath0 bins than in case of traditional presentation in form of @xmath1 - and @xmath0 - dependencies , this will allow to verify the results of our preceding work @xcite in more favourable conditions , i.e. to confirm that in electroproduction of hadrons in nuclear medium we indeed observe scaling , where @xmath3 takes over the role of the scaling variable . + this paper is organized as follows . nuclear attenuation in an absorption model is presented in the next section . in section 3
we discuss the choice of an appropriate form for the variable @xmath3 .
section 4 presents results of the fit .
conclusions are given in section 5 .
the semi - inclusive dis of lepton on nucleus of atomic mass number a is @xmath14 where @xmath15 and @xmath16 denote the initial ( final ) leptons and the hadron observed in the final state .
this process is usually investigated in terms of @xmath2 : @xmath17 experimental data are usually presented at precise values of one variable and average values of another - dependence is studied , we have @xmath18 , where @xmath19 are the average values of @xmath0 for each @xmath1 bin . for @xmath0 - dependence @xmath20,where
@xmath21 are the average values of @xmath1 for each @xmath0 bin . ] . in this work
we adopt a model according to which the origin of the nuclear attenuation is the absorption of the prehadron ( string , dipole ) and final hadron in the nuclear medium . in that case @xmath2 has the following form : @xmath22^{(a-1)}dx}},\end{aligned}\ ] ] where @xmath23 is the impact parameter and @xmath24 the longitudinal coordinate of the dis point .
@xmath25 is the nuclear density function with a normalization condition @xmath26 .
@xmath27 is the probability that neither the prehadron nor the final hadron @xmath16 are absorbed by a nucleon located anywhere in the nucleus .
for @xmath27 we use the one - scale model proposed in ref .
@xcite : @xmath28 where @xmath29 and @xmath30 are the inelastic cross sections for prehadron - nucleon and hadron - nucleon interactions , respectively .
generally speaking @xmath29 is a function of the distance @xmath31 , @xmath1 and @xmath0 . is a function of formation time @xmath3 rather than a function of variables @xmath1 and @xmath0 separately
] however a comparison of simple theoretical models containing @xmath29 as a parameter @xcite , with the experimental data obtained in different kinematical regions , ( in particular in different domains of @xmath1 ) , shows that the approximation @xmath32 leads to a very acceptable agreement with the data .
further , taking into account the qualitative character of this work , we shall use this approximation . in the region of moderate energies ,
@xmath30 are approximately constant for all @xmath16 .
@xmath33 is the probability that at distance @xmath31 from the dis point , the particle is a prehadron and @xmath34 is the probability that the particle is a hadron .
the abovementioned probabilities are related via a condition @xmath35 in analogy with the survival probability for a particle having lifetime @xmath3 in a system where it travels a distance @xmath31 before decaying , @xmath33 can be expressed in the form @xmath36 } , \end{aligned}\ ] ] where @xmath3 is the formation time .
substituting expressions for @xmath33 and @xmath34 in eq.(4 ) one obtains @xmath37 in the framework of our suppositions about @xmath29 and @xmath30 , @xmath38 depends on @xmath1 and @xmath0 only by means of @xmath39 .
+ in more detail , the formation time in the string models can be divided in two parts ( see , for instance , the two scale model presented in refs .
the first part is the constituent formation time @xmath40 , which defines the time elapsed from the moment of the dis untill the production of the first constituent of the final hadron .
the second time interval begins with the production of the first constituent until the second one , which coincides with the yo - yo or final hadron production .
comparison with the experimental data shows that in the second interval , the prehadron - nucleon cross section has values close to the hadron - nucleon cross section @xmath30 . if the difference between these cross - sections is neglected , the model is reduced to one scale model with @xmath41 . substituting @xmath27 in @xmath2 we obtain @xmath42 where @xmath43 is the maximal power of @xmath3 with which we are limited .
although @xmath2 is a polynomial of @xmath3 with maximal power @xmath44 , it is expected that @xmath45 .
the coefficients @xmath46 depend on @xmath47 and nuclear density . for fit
we use three expressions for @xmath2 as first , second and third order polynomials of @xmath3 : @xmath48={a_{11 } + \tau a_{12 } } , \hspace{2.35cm}}\end{aligned}\ ] ] @xmath49={a_{21 } + \tau a_{22 } + \tau^2 a_{23 } } , \hspace{1.1cm}}\end{aligned}\ ] ] @xmath50={a_{31 } + \tau a_{32 } + \tau^2 a_{33 } + \tau^3 a_{34}}}.\end{aligned}\ ] ] in order to get the information on the influence of highest order polynomial forms for @xmath2 , @xmath51 $ ] expression also was checked ( see section 4 ) .
equation ( 8) shows that within our approximation , @xmath2 depends on @xmath1 and @xmath0 only by means of @xmath39 .
this is the reason why we call @xmath3 a scaling variable . in this section
we shall discuss the physical meaning and possible expressions of the formation time @xmath3 .
there are different definitions for the formation time .
we define it as a time scale which is necessary for the prehadron - nucleon cross section to reach the value of the hadron - nucleon cross - section . in the literature
there are three qualitatively different definitions for @xmath3 . in the first extreme case
it is assumed that @xmath52 ( glauber approach ) . in the second extreme case @xmath53 , where @xmath54 is the nuclear radius ( energy loss model @xcite ) . and at last , and in our opinion more realistic definition of the formation time , as a function of @xmath1 and @xmath0 which can change from zero up to values larger than @xmath54 .
experimental data seem to confirm that for moderate values of @xmath1 ( on the order of @xmath55 ) the formation time is comparable with the nuclear size , i.e. the hadronization mostly takes place within the nucleus .
this follows from the comparison of the experimental data for @xmath2 obtained in the region of moderate @xcite and high @xcite energies . at moderate energies
@xmath2 significantly differs from unity and is a sensitive function of @xmath1 and @xmath0 , at high energies @xmath56 and weakly depends on @xmath1 and @xmath0 . for the formation time we shall use expressions which do not contradict the third definition mentioned above
the following expressions are used : + 1 .
formation time for the leading hadron @xcite , which follows from the energy - momentum conservation law @xmath57 where @xmath58 is the string tension ( string constant ) with numerical value @xmath59 .
formation time for the fast hadron , which is composed of characteristic formation time of the hadron @xmath16 in its rest frame @xmath60 and lorentz factor ( see , for instance , ref .
@xcite ) @xmath61 where @xmath9 and @xmath62 are the energy and mass of the hadron @xmath16 , respectively . in ref .
@xcite we have discussed formation time @xmath63 in detail .
in particular we considered the possibility of that @xmath60 being proportional to @xmath62 . in the present paper
we have to deal with hadrons of one type only - charged pions - and looking at interested @xmath1 and @xmath0 dependencies of @xmath3 , because all factors , which do not depend from these variables , can only renormalize coefficients @xmath46 ( see equations ( 9)-(11 ) ) .
the formation time following from the lund string model in ref .
@xcite is following from equation @xmath64 with @xmath65 taken from eq.(4.21 ) of ref .
@xcite . ] @xmath66{\frac{z\nu}{\kappa } } } .\end{aligned}\ ] ] one should note that all three types of formation time have similar behavior with @xmath1 , but different behavior with @xmath0 . at the values of @xmath0 typical for the hermes kinematics ( @xmath67 ) the behavior of @xmath3 defined as in eqs.(12 ) and ( 14 ) with @xmath0
is similar , i.e. they are decreasing with the increase of @xmath0 , while @xmath3 defined as in eq.(13 ) is increasing with the increase of @xmath0 .
the two - dimensional data from @xcite , i.e. the multiplicity ratio @xmath2 for charged pions produced on helium , neon , krypton and xenon nuclei , in a @xmath0 slices as a function of @xmath1 , and in a @xmath1 slices as a function of @xmath0 were used to perform the fit .
= 8.cm = 10.cm the independent fit including 47 experimental points was performed for each nucleus . as it is clear from eq.(2 ) , experimental points corresponding to @xmath1 - dependence @xmath18 , and @xmath0 - dependence @xmath20 enter in fit on equal basis , as a values of function @xmath68 at values of variables @xmath69 equal @xmath70 and @xmath71 , respectively . for the fit @xmath2
has been taken in polynomial forms @xmath72 $ ] [ see eqs.(9)-(11 ) ] , and formation times ( lengths ) as in eqs.(12)-(14 ) .
the results for the reduced @xmath73 denoted as @xmath74 presented in table [ tab : table1 ] .
one can see that for each choice of formation time and for each nucleus , the values of @xmath74 are close for the polynomial approximations @xmath75 $ ] , @xmath76 $ ] , and @xmath77 $ ] , which means that the inclusion in consideration of the higher order polynomials of @xmath3 does not essentially improve the description of the data . + in order to test this , we have also calculated the @xmath51 $ ] polinomial form and obtained the values of @xmath74 close to the ones in case of @xmath77 $ ] .
c c c c c c c c c c c c + & & & + + a&@xmath78&@xmath79 & @xmath80 & @xmath81 & @xmath82 & @xmath80 & @xmath81 & @xmath82 & @xmath80 & @xmath81 & @xmath82 + & 47&@xmath83 & & & & & & & & & + @xmath84&47&@xmath83 & & & & & & & & & + @xmath85&47&@xmath83 & & & & & & & & & + @xmath86&47&@xmath83 & & & & & & & & & + from table [ tab : table1 ] one can see that the fit gives unexpectedly good values for @xmath74 in case of @xmath87 and @xmath88 , for @xmath63 the agreement is much worse . as it is known from experiment @xcite , @xmath18 increases with increasing of @xmath1 , and @xmath20 decreases with increasing of @xmath0 for all nuclei .
our assumption is that these functions indeed present different representations of the same function , which depends _ on variable @xmath3 only_. in turn @xmath3 is a function of @xmath1 and @xmath0 .
now let us discuss the figure . using eqs.(12)-(14 )
, we present @xmath18 and @xmath20 as functions of @xmath3 .
experimental points and results of the fit are presented in fig .
solid points correspond to the @xmath18 obtained from the experimental data for @xmath1-dependence , open points to the @xmath20 from @xmath0-dependence . from the figure
one can easily note that experimental points for @xmath18 and @xmath20 as functions of @xmath3 have the same behavior and approximately coincide when @xmath87 and @xmath88 serve as the variables .
the reason for this is that these variables are approximately proportional to @xmath1 and @xmath89 . in
contrary , variable @xmath63 is proportional to @xmath1 and @xmath0 , and as a consequence @xmath18 and @xmath20 have opposite behavior as functions of @xmath63 .
this means , that without any calculations one can state that @xmath87 and @xmath88 can serve as a scaling variables , but @xmath63 can not . for the sake of convenience
we have renormalized @xmath3 to @xmath90 , where @xmath91 are the maximum values of @xmath3 for each set of data and each choice of the @xmath3 expression .
+ the range of variation of @xmath3 and the numerical values for @xmath91 in all scenarios for instance on krypton are : @xmath92 , @xmath93 , @xmath94 .
presentation of @xmath2 as a function of @xmath24 allows us to place all data in an interval @xmath95 .
this choice does not influence the results of the fit and the values of @xmath2 . on the figure
the linear polynomial is presented @xmath96 with values @xmath4 and @xmath97 = @xmath98 corresponding to the best fit .
solid , dashed and dotted curves represent the @xmath75 $ ] , @xmath76 $ ] , and @xmath77 $ ] polynomial fit , respectively .
one can easily see that the difference between the curves corresponding to @xmath75 $ ] , @xmath76 $ ] , and @xmath77 $ ] is small .
the vertical positions of the experimental points are the same in all scenarios .
the experimental points can be closer together ( or not ) depending on the type of the formation time definition . when looking at the @xmath3 dependencies , the points corresponding to @xmath1-dependence preserve their order in all scenarios .
in case of @xmath0-dependence , the order is the same in the scenario with @xmath63 , but changes to opposite in other scenarios . + as a last remark
, one should note that results of this analysis do not depend from the values of the parameters , in particular from the values of @xmath58 and @xmath99 .
the two - dimensional nuclear attenuation data for charged pions on helium , neon , krypton and xenon nuclei obtained recently by the hermes experiment @xcite were used to perform the fit .
so far it has been supposed that experimentally measured function @xmath18 depends from variable @xmath1 only , and @xmath20 from @xmath0 only . in our preceding work
@xcite , we have assumed that these functions indeed present different representations of the same function , which depends on only one variable @xmath3 . in turn
@xmath3 is a function of @xmath1 and @xmath0 . in this work based on two - dimensional distributions ( more suitable for this investigation ) , the results of our preceding work @xcite are confirmed .
we demonstrate ( see table 1 and figure ) that @xmath18 and @xmath20 as a functions of @xmath3 have the same behavior and approximately coincide , when @xmath3 is used in form of @xmath87 and @xmath88 . in contrary
, they have opposite behavior as a functions of @xmath63 .
this indicates that @xmath87 and @xmath88 can serve as scaling variables , but @xmath63 can not .
we also show that @xmath100 , with a good precision , can be parametrised in a form of a linear polynomial @xmath4 + @xmath5 , where the fitting parameters @xmath4 and @xmath6 do not depend on @xmath1 and @xmath0 .
we conclude that experimentally measured function @xmath2 is a function of single variable @xmath3 , and that @xmath3 scaling follows naturally from the absorption model , but we do not confirm that scaling is a property of an absorption model only
. there could exist other mechanisms which can also lead to the scaling behavior - this is an open question , which requires further investigation .
osborn et al . , phys .
lett . * 40b * ( 1978 ) 1624 j.ashman et al . , z.phys . *
c52 * ( 1991 ) 1 a.airapetian et al . ,
j. * c20 * ( 2001 ) 479 a.airapetian et al .
, phys.lett . *
b577 * ( 2003 ) 37 n.akopov , l.grigoryan , z.akopov , phys.rev.*c76 * ( 2007 ) 065203 ; hep - ph/0703124 ( 2007 ) a.airapetian et al . , nucl.phys .
* b780 * ( 2007 ) 1 a.bialas , acta phys.polon . *
b11 * ( 1980 ) 475 n.akopov , l.grigoryan , z.akopov , eur.phys.j . *
c44 * ( 2005 ) 219 ; hep - ph/0409359 ( 2004 ) x .- n .
wang , x. guo , nucl . phys . , * a696 * ( 2001 ) 788
; e. wang , x .-
wang , phys .
* 89 * ( 2002 ) 162301 b.kopeliovich , phys.lett .
* b243 * ( 1990 ) 141 a.bialas , m.gyulassy , nucl.phys . *
b291 * ( 1987 ) 793 | in this paper it is demonstrated that based on two - dimensional distributions in semi - inclusive deep inelastic scattering ( sidis ) data , obtained recently by the hermes experiment at desy on different nuclei , which contains data for charged pions produced in @xmath0 slices as a function of @xmath1 , and in a @xmath1 slices as a function of @xmath0 , it it is possible to parametrise a ratio of multiplicities on nucleus and deuterium ( per nucleon ) @xmath2 in a form of a function of a single variable @xmath3 , which has the physical meaning of the formation time of hadron .
we call this effect @xmath3 scaling .
@xmath3 is a function of two variables @xmath1 and @xmath0 .
it is also shown that @xmath2 can be presented in a form of a linear polynomial of @xmath3 , @xmath4 + @xmath5 , where parameters @xmath4 and @xmath6 do not depend on @xmath1 and @xmath0 . |
in their book @xcite , m. kulenovi and g. ladas initiated a systematic study of the difference equation @xmath11 for nonnegative real numbers @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 and @xmath5 such that @xmath6 and @xmath7 , and for nonnegative or positive initial conditions @xmath12 , @xmath13 . under these conditions , ( [ eq : 3 - 3 orig . ] ) has a unique positive equilibrium .
one of their main ideas in this undertaking was to make the task more manageable by considering separate cases when one or more of the parameters in ( [ eq : 3 - 3 orig . ] ) is zero .
the need for this strategy is made apparent by cases such as the well known _
lyness equation _
@xcite , @xcite , @xcite . @xmath14 whose dynamics differ significantly from other equations in this class .
there are a total of 42 cases that arise from ( [ eq : 3 - 3 orig . ] ) in the manner just discussed , under the hypotheses @xmath6 and @xmath7 .
the recent publications @xcite , @xcite give a detailed account of the progress up to 2007 in the study of dynamics of the class of equations ( [ eq : 3 - 3 orig . ] ) .
after a sustained effort by many researchers ( for extensive references , see @xcite , @xcite ) , there are some cases that have resisted a complete analysis .
we list them below in normalized form , as presented in @xcite , @xcite .
@xmath15 the dynamics of equation ( [ eq : ladas 2 - 3 ] ) has been settled recently in @xcite , @xcite .
global attractivity of the positive equilibrium of equation ( [ eq : ladas y2k ] ) has been proved recently in @xcite . since eq.([eq : kul ] ) can be reduced to eq.([eq : ladas y2k ] ) through a change of variables @xcite , global behavior of solutions to ( [ eq : kul ] ) is also settled .
equation ( [ eq : kul 1 ] ) is another equation that can be reduced to ( [ eq : ladas y2k ] ) , through the change of variables @xmath16 @xcite .
ladas and co - workers @xcite , @xcite , @xcite , have posed a series of conjectures on these equations .
one of them is the following . * conjecture [ ladas et al . ] * _ for equations ( [ eq : ladas 3 - 2 ] ) and ( [ eq : ladas 3 - 3 ] ) , every solution converges to the positive equilibrium or to a prime period - two solution . _ in this article , we prove this conjecture .
our main results are the following .
[ th : 3 - 3 ] for every choice of positive parameters @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 and @xmath5 , all solutions to the difference equation @xmath17 converge to the positive equilibrium or to a prime period - two solution .
[ th : 3 - 2 orig . ] for every choice of positive parameters @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 and @xmath5 , all solutions to the difference equation @xmath18 converge to the positive equilibrium or to a prime period - two solution . a reduction of the number of parameters of eq.([eq : 3 - 2 orig . ] ) is obtained with the change of variables @xmath19 , which yields the equation @xmath20 where @xmath21 , @xmath22 , and @xmath23 .
the number of parameters of eq.(3 - 3 ) can also be reduced , which we proceed to do next .
consider the following affine change of variables which is helpful to reduce number of parameters and simplify calculations : @xmath24 with ( [ change of coordinates ] ) , eqn.(3 - 3 ) may now be rewritten as @xmath25 where @xmath26 theorems [ th : 3 - 3 ] and [ th : 3 - 2 orig .
] can be reformulated in terms of the parameters @xmath27 , @xmath28 and @xmath29 as follows .
[ th : 3 - 2-l ] let @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 and @xmath5 be positive numbers , and let @xmath27 , @xmath28 , @xmath29 and @xmath30 be given by relations ( [ eq : pqr ] ) .
then every solution to eqn.(3 - 2-l ) converges to the unique equilibrium or to a prime period - two solution .
[ th : 3 - 2 ] let @xmath27 , @xmath28 , @xmath29 be positive numbers .
then every solution to eqn.(3 - 2 ) converges to the unique equilibrium or to a prime period - two solution . in this paper
we prove theorems [ th : 3 - 2-l ] and [ th : 3 - 2 ] ; theorems [ th : 3 - 3 ] and [ th : 3 - 2 orig . ]
follow as an immediate corollary .
the two main differences between eq.(3 - 2-l ) and eq.(3 - 2 ) are the set of initial conditions , and the possibility of having a negative value of @xmath29 in eq.(3 - 2-l ) , while only positive values of @xmath29 are allowed in eq.(3 - 2 ) . nevertheless , for both eq.(3 - 2-l ) and eq.(3 - 2 ) the unique equilibrium has the formula : @xmath31 although it is not possible to prove theorem [ th : 3 - 3 ] as a simple corollary to theorem [ th : 3 - 2 orig . ] ,
the changes of variables leading to theorems [ th : 3 - 2-l ] and [ th : 3 - 2 ] will result in proofs to the former theorems that are greatly simplified .
our main results theorem [ th : 3 - 3 ] and theorem [ th : 3 - 2 orig . ]
imply that when prime period - two solutions to eq.(3 - 3 ) or eq.(3 - 2 ) do not exist , then the unique equilibrium is a global attractor .
we have not treated here certain questions about the global dynamics of eq.(3 - 3 ) and eq.(3 - 2 ) , such as the character of the prime period - two solutions to either equation , or even for more general rational second order equations , when such solutions exist .
this matter will be treated in an upcoming article of the authors @xcite .
this work is organized as follows .
the main results are stated in section [ sec : introduction ] .
results from the literature which are used here are given in section [ sec : results from the literature ] for convenience . in section [ section : existence ] , it is shown that either every solution to eq.(3 - 2-l ) converges to the equilibrium , or there exists an invariant and attracting interval @xmath32 with the property that the function @xmath33 associated with the difference equation is coordinate - wise strictly - monotonic on @xmath34 . in section
[ sec : the equation ( 3 - 2 ) ] , a global convergence result is obtained for eq.(3 - 2 ) over a specific range of parameters and for initial conditions in an invariant compact interval .
theorem [ th : 3 - 2-l ] is proved in section [ sec : proof 3 - 2-l ] , and the proof of theorem [ th : 3 - 2 ] is given in section [ sec : proof 3 - 2 ] .
section [ appendix : computer 1 ] includes computer algebra system code for performing certain calculations that involve polynomials with a large number of terms ( over 365,000 in one case ) .
these computer calculations are used to support certain statements in section [ sec : the equation ( 3 - 2 ) ] .
finally , we refer the reader to @xcite for terminology and definitions that concern difference equations .
the results in this subsection are from the literature , and they are given here for easy reference .
the first result is a reformulation of theorems ( 1.4.5 ) ( 1.4.8 ) in @xcite .
[ th : mm first ] suppose a continuous function @xmath35 ^ 2 \rightarrow [ a , b]$ ] satisfies one of i.iv . : * @xmath33 is nondecreasing in @xmath36 , @xmath37 , and @xmath38 ^ 2 , \quad ( \ , f(m , m)=m\ \&\
f(m , m)=m\ , ) \implies m = m\ ] ] * @xmath33 is nonincreasing in @xmath36 , @xmath37 , and @xmath38 ^ 2 , \quad ( \ , f(m , m)=m\ \&\
f(m , m)=m\ , ) \implies m = m\ ] ] * @xmath33 is nonincreasing in @xmath36 and nondecreasing in @xmath37 , and @xmath38 ^ 2 , \quad ( \ , f(m , m)=m\ \&\
f(m , m)=m\ , ) \implies m = m\ ] ] * @xmath33 is nondecreasing in @xmath36 and nonincreasing in @xmath37 , and @xmath38 ^ 2 , \quad ( \ , f(m , m)=m\ \&\
f(m , m)=m\ , ) \implies m = m\ ] ] then @xmath39 has a unique equilibrium in @xmath40 $ ] , and every solution with initial values in @xmath41 $ ] converges to the equilibrium .
the following result is theorem a.0.8 in @xcite .
[ th : mm 3 - 3 ] suppose a continuous function @xmath35 ^ 3 \rightarrow [ a , b]$ ] is nonincreasing in all variables , and @xmath38 ^ 3 , \quad ( \ , f(m , m , m)=m\ \&\
f(m , m , m)=m\ , ) \implies m = m\ ] ] then @xmath42 has a unique equilibrium in @xmath40 $ ] , and every solution with initial values in @xmath41 $ ] converges to the equilibrium .
[ th : camouzis - ladas ] let @xmath32 be a set of real numbers and let @xmath43 be a function @xmath44 which decreases in @xmath45 and increases in @xmath46 .
then for every solution @xmath47 of the equation @xmath48 the subsequences @xmath49 and @xmath50 of even and odd terms do exactly one of the following : * they are both monotonically increasing .
* they are both monotonically decreasing . *
eventually , one of them is monotonically increasing and the other is monotonically decreasing . theorem [ th : camouzis - ladas ] has this corollary .
if @xmath32 is a compact interval , then every solution of eq.([eq : camouzis - ladas ] ) converges to an equilibrium or to a prime period - two solution .
[ th : kocic - ladas ] assume the following conditions hold : * @xmath51 $ ] .
* @xmath52 is decreasing in @xmath36 and strictly decreasing in @xmath37 .
* @xmath53 is strictly increasing in @xmath36 . *
the equation @xmath54 has a unique positive equilibrium @xmath55 .
then @xmath55 is a global attractor of all positive solutions of eq.([eq : kocic - ladas ] ) .
in this section we prove a proposition which is key for later developments .
we will need the function @xmath56 associated to eq.(3 - 2-l ) . [ prop : invariant and attracting ] at least one of the following statements is true : * every solution to ( 3 - 2-l ) converges to the equilibrium .
* there exist @xmath57 , @xmath58 with @xmath59 s.t .
* * @xmath60 $ ] is an invariant interval for eq.(3 - 2-l ) , i.e. , @xmath61\times [ m^{*},m^ { * } ] ) \subset [ m^{*},m^{*}]$ ] . * * every solution to eq.(3 - 2-l ) eventually enters @xmath60 $ ] . *
* @xmath33 is coordinate - wise strictly monotonic on @xmath60^{2}$ ] .
the next lemma states that the function @xmath62 associated to eq.(3 - 2-l ) is bounded .
[ lemma : f is bounded ] there exist _ positive _ constants @xmath63 and @xmath64 such that @xmath65 and @xmath66 in particular , @xmath67 is invariant for f } f\left([\mathcal{l},\mathcal{u}]\times[\mathcal{l},\mathcal{u}]\right ) \subset [ \mathcal{l},\mathcal{u}]\ ] ] the function @xmath68 associated to eq.(3 - 3 ) is bounded : @xmath69 set @xmath70 and @xmath71 .
the affine change of coordinates ( [ change of coordinates ] ) maps the rectangular region @xmath72 ^ 2 $ ] onto a rectangular region @xmath73 ^ 2 $ ] which satisfies ( [ ineq : bounds for f ] ) and ( [ eq : [ u , l ] is invariant for f ] ) .
[ lemma : p = q ] if @xmath74 , then every solution to eq.(3 - 2-l ) converges to the unique equilibrium .
if @xmath74 then @xmath75 and @xmath76 . thus , depending on the sign of @xmath29 , the function @xmath33 is either nondecreasing in both coordinates , or nonincreasing in both coordinates on @xmath77 . by lemma [ lemma : f is bounded ] , all solutions @xmath78 satisfy @xmath79 $ ] for @xmath80 . a direct algebraic calculation may be used to show that all solutions @xmath81 $ ] of either one of the systems of equations @xmath82 necessarily satisfy @xmath83 . in either case , the hypotheses ( i ) or ( ii ) of theorem [ th : mm first ] are satisfied , and the conclusion of the lemma follows .
we will need the following elementary result , which is given here without proof .
[ lemma : signs of partials ] suppose @xmath84 .
the function @xmath33 has continuous partial derivatives on @xmath85 , and * @xmath86 if and only if @xmath87 , and @xmath88 if and
only if @xmath89 . * @xmath90 if and only if @xmath91 , and @xmath92 if and only if @xmath93 .
we will need to refer to the values @xmath94 and @xmath95 where the partial derivatives of @xmath33 change sign .
if @xmath96 , set @xmath97 for @xmath98 , let @xmath99 ^ 2\ } \quad \mbox{and } \quad \phi(m , m ) : = \max \ { f(x , y ) : ( x , y ) \in [ m , m]^2\}\ ] ] [ lemma : m < phi ] suppose @xmath96 . if @xmath100\subset [ \mathcal{l},\mathcal{u}]$ ] is an invariant interval for eq.(3 - 2-l ) with @xmath101 or @xmath102 , then
@xmath103 or @xmath104 or @xmath105 . by definition of @xmath106 and @xmath107 , @xmath108 and @xmath109 .
suppose @xmath110 the proof will be complete when it is shown that @xmath83 .
there are a total of four cases to consider : ( a ) @xmath111 and @xmath112 , ( b ) @xmath113 and @xmath114 , ( c ) @xmath111 and @xmath114 , and ( d ) @xmath113 and @xmath112 .
we present the proof of case ( a ) only , as the proof of the other cases is similar .
if @xmath111 and @xmath112 , then @xmath115 $ ] and @xmath116 $ ] .
note that @xmath117\times[m , m ] = [ m , m]\times [ m , k_1]\ \bigcup \ [ m , m]\times [ k_1,m].\ ] ] by lemma [ lemma : signs of partials ] , the signs of the partial derivatives of @xmath33 are constant on the interior of each of the sets @xmath100\times [ m , k_1 ] $ ] and @xmath100\times [ k_1,m]$ ] , as shown in the diagram .
( 2.5,2.5 ) ( 0.5,2.5 ) ( 2.5,1.4 ) ( 0.5,0.5 ) ( -0.8,.6)@xmath118 ( 2.8,1.4)@xmath119 ( -0.8,2.4)@xmath120 ( 0.6 , 0.6)(2,2 ) ( 1,0.9)@xmath121 ( 0.6,1.5)(1,0)2 ( 1,1.9)@xmath122 [ fig : diagram 1 ] since @xmath33 is nonincreasing in both @xmath36 and @xmath37 on @xmath123\times[m , k_{1}]$ ] , @xmath124\times[m , k_{1}].\ ] ] similarly , @xmath33 is nondecreasing in @xmath36 and nonincreasing in @xmath37 on @xmath123\times[k_{1 } , m]$ ] , hence @xmath125\times[k_{1 } , m].\ ] ] from ( [ eq : ineq1-new1 ] ) and ( [ eq : ineq2-new1 ] ) one has @xmath126 combine ( [ eq : phi = f(m , m ) ] ) with relation ( [ eq : m = phi ] ) to obtain the system of equations @xmath127 eliminating @xmath128 from system ( [ eq : sys phi phi ] ) gives the cubic in @xmath129 @xmath130 which has the roots @xmath131 only one root in the list ( [ eq : the m roots ] ) is positive , namely @xmath132 substituting into one of the equations of system ( [ eq : sys phi phi ] ) one also obtains @xmath133 , which gives the desired relation @xmath105 .
let @xmath134 , @xmath135 , and for @xmath136 let @xmath137 , @xmath138 . by the definitions of @xmath139 , @xmath140 , @xmath141 and @xmath142 , we have that @xmath143 \subset [ m_\ell , m_\ell]$ ] for @xmath144 .
thus the sequence @xmath145 is nondecreasing and @xmath146 is nonincreasing .
let @xmath147 and @xmath148 .
suppose @xmath96 .
either there exists @xmath149 such that @xmath150 = \emptyset$ ] , or @xmath151 .
arguing by contradiction , suppose @xmath152 and for all @xmath153 , @xmath154 \not = \emptyset$ ] .
since the intervals @xmath155 $ ] are nested and @xmath156 = [ m^*,m^*]$ ] , it follows that @xmath157 \not = \emptyset$ ] . by lemma [ lemma : m
< phi ] , we have @xmath158 continuity of the functions @xmath106 and @xmath107 implies @xmath159 statements ( [ ineq : mstar1 ] ) and ( [ ineq : mstar2 ] ) give a contradiction .
* proof of proposition [ prop : invariant and attracting]*. suppose statement ( a ) is not true . by lemma [ lemma :
p = q ] , one must have @xmath160 .
note that if @xmath161 is a solution to eq.(3 - 2-l ) , then @xmath162 $ ] for @xmath136 .
if @xmath163 , since @xmath164 and @xmath165 we have @xmath166 , but this is statement ( a ) which we are negating . thus @xmath152 , and by lemma [ lemma : m < phi ] there exists @xmath149 such that @xmath150=\emptyset$ ] , so @xmath33 is coordinate - wise monotonic on @xmath167 $ ] .
the set @xmath167 $ ] is invariant , and every solution enters @xmath167 $ ] starting at least with the term with subindex @xmath168 . we have shown that if statement ( a ) is not true , then statement ( b ) is necessarily true .
this completes the proof of the proposition . @xmath169
in this section we restrict our attention to the equation @xmath172 where @xmath173 for @xmath174 , @xmath175 , and @xmath170 , eq.(3 - 2 ) has a unique _ positive _ equilibrium @xmath176 we note that if @xmath177 is an invariant compact interval , then necessarily @xmath178 .
the goal in this section is to prove the following proposition , which will provide an important part of the proofs of theorems [ th : 3 - 3 ] and [ th : 3 - 2 ] .
[ prop : f up down r > 0 on i ] let @xmath27 , @xmath28 and @xmath29 be real numbers such that @xmath179 and let @xmath180\subset ( \frac{q\,r}{p - q } , \frac{p}{q})$ ] be a compact invariant interval for eq.(3 - 2 ) .
then every solution to eq.(3 - 2 ) with @xmath181 $ ] converges to the equilibrium .
proposition [ prop : f up down r > 0 on i ] follows from lemmas [ lemma : f up down ybar p<1 q>1 r>0 ] , [ lemma : f up down ybar sufficient condition ] and [ lemma : ybar is ga when suff cond 2 ] , which are stated and proved next .
[ lemma : f up down ybar p<1 q>1 r>0 ] assume the hypotheses to proposition [ prop : f up down r > 0 on i ] .
if either @xmath182 or @xmath183 , then every solution to eq.(3 - 2 ) with @xmath181 $ ] converges to the equilibrium .
we verify that hypothesis ( iv ) of theorem [ th : mm first ] is true . since @xmath184 for @xmath185 $ ] ,
the function @xmath33 is increasing in @xmath36 and decreasing in @xmath37 for @xmath186 ^ 2 $ ] by lemma [ lemma : signs of partials ] .
let @xmath187 $ ] be such that @xmath188 and @xmath189 we show first that system ( [ eq : mm for p<1 ] ) has no solutions if either @xmath182 or @xmath183 . by eliminating denominators in both equations in ( [ eq : mm for p<1 ] ) , @xmath190 and by subtracting terms in ( [ eq : mm for p<1 b ] ) one obtains @xmath191 since @xmath192 , we have @xmath193 , which implies that for @xmath194 there are no solutions to system ( [ eq : mm for p<1 ] ) which have both coordinates positive .
now assume @xmath195 ; from ( [ eq : mm for p<1 2 ] ) , @xmath196 , and substitute the latter into ( [ eq : mm for p<1 ] ) to see that @xmath197 is a solution to the quadratic equation @xmath198 by a symmetry argument , one has that @xmath199 is also a solution to ( [ eq : mm quadratic p>1 ] ) . by inspection of the coefficients of the polynomial in the left - hand - side of ( [ eq : mm quadratic p>1 ] )
one sees that two positive solutions are possible only when @xmath200 . to get the conclusion of the lemma , note that the fact that ( [ eq : mm for p<1 ] ) has no solutions with @xmath192 is just hypothesis ( iv ) of theorem [ th : mm first ] .
[ lemma : f up down ybar sufficient condition ] assume the hypotheses to proposition [ prop : f up down r > 0 on i ] .
if @xmath201 then every solution to eq.(3 - 2 ) with @xmath181 $ ] converges to the equilibrium . by substituting @xmath202 into @xmath203 we obtain @xmath204 that is , @xmath205 where the @xmath36 has been kept in @xmath206 for bookkeeping purposes .
thus @xmath206 is constant in @xmath36 .
we claim @xmath206 is decreasing in both @xmath37 and @xmath207 . to see that the partial derivative @xmath208 is negative just use @xmath112 and the inequality @xmath209 , which is true by lemma [ lemma : signs of partials ] .
the remaining partial derivative is @xmath210 where @xmath211 we have , @xmath212 since @xmath213 , @xmath214\ ] ] thus we conclude that @xmath215 for @xmath216 $ ] . to complete the proof we verify the hypotheses of theorem [ th : mm 3 - 3 ] .
we claim that the system of equations @xmath217 has no solutions @xmath120 with @xmath192 whenever hypothesis ( [ sufficient condition ] ) holds . by eliminating denominators in both equations in ( [ eq : mm for embed ] ) one obtains @xmath218 and by subtracting terms in ( [ eq : mm for p<1 b embed ] ) one obtains @xmath219 since @xmath192 , we may use the second factor in the left - hand - side term of ( [ eq : mm for p<1 2 embed ] ) to solve for @xmath128 in terms of @xmath129 , which upon substitution into @xmath220 and simplification yields the equation @xmath221 where @xmath222 by hypothesis ( [ sufficient condition ] ) we have @xmath223 , hence @xmath224 , which implies @xmath225 . by direct inspection
one can see that @xmath226 and @xmath227 .
thus ( [ eq : no solutions ] ) has no positive solutions , and we conclude that ( [ eq : mm for embed ] ) has no solutions @xmath228 $ ] with @xmath192 . we have verified the hypotheses of theorem [ th : mm 3 - 3 ] , and the conclusion of the lemma follows .
[ lemma : las ] let @xmath174 , @xmath175 and @xmath170 .
if the positive equilibrium @xmath229 of eq.(3 - 2 ) satisfies @xmath230 , then @xmath229 is locally asymptotically stable ( l.a.s . ) . solving for @xmath29 in @xmath231 gives
@xmath232 then a calculation shows @xmath233 set @xmath234 and @xmath235 .
the equilibrium @xmath229 is locally asymptotically stable if the roots of the characteristic polynomial @xmath236 have modulus less than one @xcite . by the schur - cohn theorem
, @xmath229 is l.a.s . if and only if @xmath237 .
it can be easily verified that @xmath238 if and only if @xmath239 which is true regardless of the allowable parameter values . since @xmath240 by the hypothesis , we have @xmath241 , hence some algebra gives @xmath242 if and only if @xmath243 but ( [ eq : schur conn ] ) is a true statement by formula ( [ eq : ybar formula ] ) .
we conclude @xmath229 is l.a.s .
[ lemma : ybar is ga when suff cond 2 ] assume the hypotheses to proposition [ prop : f up down r > 0 on i ] .
if @xmath244 then every solution to eq.(3 - 2 ) with @xmath181 $ ] converges to the equilibrium .
the proof begins with a change of variable in eq.(3 - 2 ) to produce a transformed equation with normalized coefficients analogous to those in the standard _
normalized lyness equation _
@xcite , @xcite , @xcite @xmath245 we seek to use an argument of proof similar to the one used in @xcite , in which one takes advantage of the existence of _ invariant curves _ of lyness equation to produce a lyapunov - like function for eq.(3 - 2 ) .
set @xmath246 in eq.(3 - 2 ) to obain the equation @xmath247 where @xmath248 we shall denote with @xmath249 the unique equilibrium of eq.([3 - 2-z ] ) . note that @xmath250 it is convenient to parametrize eq.([3 - 2-z ] ) in terms of the equilibrium .
we will use the symbol @xmath45 to represent the equilibrium @xmath249 of eq.([3 - 2-z ] ) . by direct substitution of the equilibrium
@xmath251 into eq.([3 - 2-z ] ) we obtain @xmath252 by ( [ eq : alpha ] ) , @xmath111 iff @xmath253 . using ( [ eq : alpha ] ) to eliminate @xmath29 from eq.([3 - 2-z ] ) gives the following equation for @xmath254 , @xmath255 and @xmath256 , equivalent to eq.([3 - 2-z ] ) : @xmath257 therefore it suffices to prove that all solutions of eq.([eq : y2k2u ] ) converge to the equilibrium @xmath45 .
the following statement is crucial for the proof of the proposition .
[ claim : ybar > 1 ] @xmath258 if and only if @xmath259 . since @xmath260
, we have @xmath261 if and only if @xmath262 , which holds if and only if @xmath263 after an elementary simplification , the latter inequality can be rewritten as @xmath259 . by the hypotheses of the lemma , by claim [ claim : ybar
> 1 ] , and by ( [ eq : new parameters ] ) and ( [ eq : alpha ] ) we have @xmath264 we now introduce a function which is the invariant function for ( [ eq : lyness ] ) with constant @xmath265 ( in this case the the equilibrium of ( [ eq : lyness ] ) is @xmath45 ) : @xmath266 note that @xmath267 for all @xmath268 whenever @xmath261 . by using elementary calculus , one can show that the function @xmath269 has a strict global minimum at @xmath270 @xcite , @xcite , i.e. , @xmath271 we need some elementary properties of the sublevel sets @xmath272 we denote with @xmath273 , @xmath274 the four regions @xmath275 let @xmath276 be the map associated to eq .
( [ eq : y2k2u ] ) ( see @xcite ) .
[ claim : q2q4 ] if @xmath277 , then @xmath278 .
set @xmath279 a calculation yields @xmath280 where @xmath281 by ( [ eq : ineqs u , g , b ] ) , for @xmath282 we have @xmath283 and @xmath284 with @xmath285 , therefore @xmath286 , @xmath287 and @xmath288 .
consequently @xmath289 for @xmath282 . to see that @xmath289 for @xmath290 as well ,
rewrite @xmath291 and @xmath292 as follows : @xmath293 for @xmath290 we have @xmath294 and @xmath285 .
thus @xmath295 , @xmath296 , and @xmath297 , which imply @xmath298 .
[ claim : q1q3 , g > b ] suppose @xmath299 .
if @xmath300 , then @xmath301 .
this proof requires extensive use of a computer algebra system to verify certain inequalities involving rational expressions . here
we give an outline of the steps , and refer the reader to section [ appendix : computer 1 ] for the details .
since @xmath302 , and @xmath303 we may write @xmath304 the expression @xmath305 may be written as a single ratio of polynomials , @xmath306 with @xmath307 .
the next step is to show @xmath308 for @xmath309 .
points @xmath310 in @xmath311 may be written in the form @xmath312 , @xmath313 , where @xmath314 .
substituting @xmath36 , @xmath37 , @xmath45 and @xmath315 in terms of @xmath46 , @xmath316 , @xmath317 and @xmath318 into the expression for @xmath319 one obtains a rational expression @xmath320 with positive denominator .
the numerator @xmath321 has some negative coefficients . at this
points two cases are considered , @xmath322 , and @xmath323 .
these can be written as @xmath324 and @xmath325 for nonnegative @xmath326 .
substitution of each one of the latter expressions in @xmath321 gives a polynomial with positive coefficients .
this proves @xmath327 for @xmath328 .
if now we assume @xmath329 with @xmath285 , we may write @xmath330 the rest of the proof is as in the first case already discussed .
details can be found in section [ appendix : computer 1 ] .
[ claim : q1q3 , g < b ] suppose @xmath331 and @xmath261 . if @xmath300 , then @xmath332 .
the proof is analogous to the proof of claim [ claim : q1q3 , g > b ] .
we provide an outline .
more details can be found in section [ appendix : computer 1 ] . since @xmath333
, we may write @xmath334 with @xmath335 .
also , @xmath336 implies @xmath337 , and @xmath338 for @xmath339 . since @xmath340
we may write @xmath341 for @xmath342 .
the expression @xmath343 may be written as a single ratio of polynomials , @xmath306 with @xmath307 .
the next step is to show @xmath308 for @xmath309 .
this is done in a way similar to the procedure described in in claim [ claim : q1q3 , g > b ] . to complete the proof of the lemma ,
let @xmath344 .
let @xmath345 be the solution to ( [ eq : y2k2u ] ) with initial condition @xmath346 , and let @xmath347 be the corresponding orbit of @xmath348 .
the following argument is essentially the same as the one found in @xcite ; we provided here for convenience .
define @xmath349 note that @xmath350 , which can be shown by applying claims [ claim : q2q4 ] , [ claim : q1q3 , g > b ] and [ claim : q1q3 , g < b ] repeatedly as needed to obtain a nonincreasing subsequence of @xmath351 that is bounded below by @xmath352 .
let @xmath353 be a subsequence convergent to @xmath354 .
therefore there exists @xmath355 such that @xmath356 that is , @xmath357 the set @xmath358 is closed by continuity of @xmath269 .
boundedness of @xmath358 follows from @xmath359 thus @xmath358 is compact , and there exists a convergent subsequence @xmath360 with limit @xmath361 .
note that @xmath362 we claim that @xmath363 . if not , then by claims [ claim : q2q4 ] [ claim : q1q3 , g > b ] and [ claim : q1q3 , g < b ] , @xmath364 let @xmath365 denotes the euclidean norm . by ( [ eq : cont ineq ] ) and continuity ,
there exists @xmath366 such that @xmath367 choose @xmath368 large enough so that @xmath369 but then ( [ eq : by continuity ] ) and ( [ eq : choose l ] ) imply @xmath370 which contradicts the definition ( [ eq : hatc def ] ) of @xmath354 .
we conclude @xmath363 . from this and the definition of convergence of sequences
we have that for every @xmath371 there exists @xmath372 such that @xmath373 .
finally , since @xmath374 we have that for every @xmath371 there exists @xmath372 such that @xmath375 and @xmath376 . since @xmath45 is a locally asymptotically stable equilibrium for eq.([eq : y2k2u ] ) by lemma [ lemma : las ] , it follows that @xmath377 .
this completes the proof of the lemma .
to prove theorem [ th : 3 - 2-l ] it is enough to assume statement ( b ) of proposition [ prop : invariant and attracting ] . also by lemma [ lemma :
p = q ] we may assume @xmath96 without loss of generality . thus we make the following standing assumption , valid throughout the rest of this section for eq.(3 - 2-l ) .
* standing assumption ( sa ) * _ assume @xmath160 and that there exist @xmath57 , @xmath58 with @xmath378 such that for eq.(3 - 2-l ) and its associated function @xmath33 , _ * @xmath60 $ ] is an invariant interval . *
every solution eventually enters @xmath60 $ ] .
* @xmath33 is coordinate - wise strictly monotonic on @xmath60^{2}$ ] .
the function @xmath33 is assumed to be coordinate - wise monotonic on @xmath379 $ ] , and there are four possible cases in which this can happen : ( a ) @xmath33 is increasing in both variables , ( b ) @xmath33 is decreasing in both variables , ( c ) @xmath33 is decreasing in @xmath36 and increasing in @xmath37 , and ( d ) @xmath33 is increasing in @xmath36 and decreasing in @xmath37 .
we present several lemmas before completing the proof of theorem [ th : 3 - 2-l ] . by considering the restriction of the map @xmath348 of eq.(3 - 2-l ) to @xmath379 ^ 2 $ ]
, an application of the schauder fixed point theorem @xcite gives that @xmath379 ^ 2 $ ] contains the fixed point of @xmath348 , namely @xmath380 .
thus we have the following result .
[ lemma : ybar in invariant interval ] @xmath381 $ ] .
[ lemma : f(up , up ) and f(down , down ) ] neither one of the systems of equations @xmath382 have solutions @xmath383 ^ 2 $ ] with @xmath384 . since @xmath385 is the only solution to @xmath386 , it is clear that only @xmath380 satisfies ( s@xmath387 ) .
now let @xmath120 be a solution to ( s@xmath388 )
. from straightforward algebra applied to @xmath389 one arrives at @xmath390 , which implies @xmath83 .
[ lemma : 0<q
< p ] suppose @xmath33 is increasing in @xmath36 and decreasing in @xmath37 for @xmath391 $ ] . then @xmath392 . by the standing assumption ( sa ) , @xmath96 . by lemma [ lemma : signs of partials ] , the coordinate - wise monotonicity hypothesis , and the fact @xmath381 $ ] from lemma [ lemma : ybar in invariant interval ] , we have @xmath393 the inequalities in ( [ eq : p - q sign ] ) can not hold simultaneously unless @xmath394 .
[ lemma : f up down lemma bounds ] if @xmath33 is increasing in @xmath36 and decreasing in @xmath37 for @xmath391 $ ] , then @xmath395 ^ 2 ) \subset ( 1,\frac{p}{q})$ ] . for @xmath391 $ ] ,
the function @xmath396 is well defined and is componentwise strictly monotonic on the set @xmath397 .
then , @xmath398 [ lemma : f up down iff ] let @xmath174 , @xmath175 and @xmath170 . if @xmath33 is increasing in @xmath36 and decreasing in @xmath37 on @xmath379 $ ] , then @xmath399 since @xmath381 $ ] by lemma [ lemma : ybar in invariant interval ] , we have @xmath400 and @xmath401 . by lemma
[ lemma : 0<q
< p ] , @xmath112 , and by lemma [ lemma : signs of partials ] , @xmath402 then , @xmath403 in addition , by lemma [ lemma : signs of partials ] , @xmath404 [ lemma : p - q+r>0 ] suppose @xmath33 is increasing in @xmath36 and decreasing in @xmath37 for @xmath391 $ ] .
if @xmath113 , then @xmath405 .
since @xmath406 for @xmath391 $ ] , and by lemma [ lemma : signs of partials ] , lemma [ lemma : ybar in invariant interval ] and by lemma [ lemma : 0<q < p ] , we have @xmath407 , that is , @xmath408 if the right - hand - side of inequality ( [ eq : ybar bigger than ] ) is nonnegative , then , after squaring both sides of ( [ eq : ybar bigger than ] ) we have @xmath409 further simplification of ( [ eq : ybar bigger than 2 ] ) and the hypothesis @xmath113 yield @xmath410 which , after some elementary algebra , implies @xmath405 .
now assume the right - hand - side of inequality ( [ eq : ybar bigger than ] ) is negative , relation that we may rewrite as @xmath411 if @xmath412 , then @xmath413,which gives the conclusion @xmath414 . if @xmath415 , that is , @xmath416 , then @xmath417 therefore if @xmath418 the conclusion of the lemma follows from this and from ( [ eq : p - q+r > q+r+1 ] ) .
assume @xmath419 from relations ( [ eq : pqr ] ) we have @xmath420 hence assumption ( [ eq : assumption q+r+1<0 ] ) and relation ( [ eq : q+r+1<0 ] ) imply @xmath421 further algebra gives @xmath422 since @xmath423 by ( [ eq : long term ] ) , from inequality ( [ eq : further algebra ] ) we have @xmath424 finally , from ( [ eq : pqr ] ) we have @xmath425 combining ( [ eq : one more ] ) with ( [ eq : yet one more ] ) we obtain @xmath426 .
[ lemma : f up down ybar is ga ] if @xmath113 and @xmath33 is increasing in @xmath36 and decreasing in @xmath37 for @xmath391 $ ] , then every solution converges to the equilibrium . since @xmath414 by lemma [ lemma : p - q+r>0 ]
, we have @xmath427 , which together with lemma [ lemma : f up down lemma bounds ] implies that @xmath428 $ ] is an invariant , attracting compact interval such that @xmath33 is increasing in @xmath36 and decreasing in @xmath37 on @xmath428 ^ 2 $ ] . since @xmath429 ^ 2 ) \subset ( 1,\frac{p}{q})$ ] , we see that every solution to eq.(3 - 2-l ) eventually enters the invariant interval @xmath430 .
the change of variables @xmath431 transforms the equation @xmath432 into the equivalent equation @xmath433 where @xmath434 we claim that for @xmath435 , ( a ) @xmath436 is increasing in @xmath316 , ( b ) @xmath437 is decreasing in @xmath316 , and ( c ) @xmath437 is decreasing in @xmath46 .
indeed , since @xmath112 , @xmath113 , @xmath414 , and @xmath438 we have @xmath439 @xmath440 @xmath441 also , note that eq.([eq : new kl ] ) has a unique equilibrium @xmath249 .
therefore hypotheses ( 1)(4 ) of theorem [ th : kocic - ladas ] are satisfied , so every solution @xmath442 to eq.([eq : new kl ] ) converges to @xmath249 . by reversing the change of variables , one can conclude that every solution to eq.([eq : old kl ] ) converges to the equilibrium .
* proof of theorem [ th : 3 - 2-l ] .
* the four parts of the proof are : * _ @xmath33 is increasing in both @xmath36 and @xmath37 on @xmath379 ^ 2 $ ] : _ by lemma [ lemma : f(up , up ) and f(down , down ) ] the hypotheses of theorem [ th : mm first ] part ( i ) .
is satisfied , hence every solution converges to the equilibrium @xmath229 . *
_ @xmath33 is decreasing in both @xmath36 and @xmath37 on @xmath379 ^ 2 $ ] : _ by lemma [ lemma : f(up , up ) and f(down , down ) ] the hypotheses of theorem [ th : mm first ] part ( ii ) .
is satisfied , hence every solution converges to the equilibrium @xmath229 .
* _ @xmath33 is decreasing in @xmath36 and increasing in @xmath37 on @xmath379 ^ 2 $ ] : _ by the corollary to theorem [ th : camouzis - ladas ] we conclude every solution converges to the unique equilibrium or to a prime period - two solution . * _ @xmath33 is increasing in @xmath36 and decreasing in @xmath37 on @xmath379 ^ 2 $ ] : _ by lemmas [ lemma : signs of partials ] , [ lemma : 0<q
< p ] , and [ lemma : f up down lemma bounds ] , there is no loss of generality in assuming @xmath379 \subset ( k,\frac{p}{q})$ ] , where @xmath443 , which we do .
we consider two subcases . if @xmath111 , then lemma [ lemma : 0<q < p ] , lemma [ lemma : f up down iff ] and proposition [ prop : f up down r > 0 on i ] imply that every solution converges to the unique equilibrium .
if @xmath113 , then lemma [ lemma : f up down ybar is ga ] implies that every solution converges to the unique equilibrium .
this completes the proof of theorem [ th : 3 - 2-l ] .
since theorem [ th : 3 - 2-l ] is just a version of theorem [ th : 3 - 3 ] obtained by an affine change of coordinates , we have also proved theorem [ th : 3 - 3 ] as well .
@xmath169 the first lemma guarantees solutions to eq.(3 - 2 ) to be bounded . [ lemma : 3 - 2 is bounded ] let @xmath174 , @xmath175 and @xmath111 .
there exist positive constants @xmath63 and @xmath64 such that every solution @xmath47 to eq.(3 - 2 ) satisfies @xmath444 $ ] for @xmath445 , and the function @xmath446 satisfies @xmath447\times [ \mathcal{l},\mathcal{u } ] ) \subset [ \mathcal{l},\mathcal{u}]\ ] ] set @xmath448 since @xmath449 then @xmath450 for @xmath451 , , i.e. , @xmath452 from the definition of @xmath64 we have @xmath453 write @xmath454 as @xmath455 , @xmath456 for @xmath457 .
then for @xmath458 , @xmath459 that is , @xmath460 inspection of the proof of proposition [ prop : invariant and attracting ] reveals that , given that we have lemma [ lemma : 3 - 2 is bounded ] , the conclusion of the proposition is true concerning eq.(3 - 2 ) .
the statement is given next .
[ prop : invariant and attracting 3 - 2 ] at least one of the following statements is true : * every solution to eq.(3 - 2 ) converges to the equilibrium .
* there exist @xmath57 , @xmath58 with @xmath461 s.t .
* * @xmath60 $ ] is an invariant interval for eq.(3 - 2 ) , i.e. , @xmath61\times [ m^{*},m^ { * } ] ) \subset [ m^{*},m^{*}]$ ] . * * every solution to eq.(3 - 2 ) eventually enters @xmath60 $ ] . *
* @xmath33 is coordinate - wise strictly monotonic on @xmath60^{2}$ ] .
the proof of theorem [ th : 3 - 2-l ] may be reproduced here in its entirety with the only change being the elimination of the case @xmath113 , which presently does not apply .
everything else in the proof applies to eq.(3 - 2 ) .
the proof of theorem [ th : 3 - 2 ] is complete . | for nonnegative real numbers @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 and @xmath5 such that @xmath6 and @xmath7 , the difference equation @xmath8 has a unique positive equilibrium . a proof is given here for the following statements : theorem 1 .
_ for every choice of positive parameters @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 and @xmath5 , all solutions to the difference equation @xmath9 converge to the positive equilibrium or to a prime period - two solution .
_ theorem 2 .
_ for every choice of positive parameters @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 and @xmath5 , all solutions to the difference equation @xmath10 converge to the positive equilibrium or to a prime period - two solution .
_ difference equation , rational , global behavior , global attractivity , period - two solution . |
the flavor diversity in the electroweak interaction continues to attract theoretical curiosity for new physics beyond the standard model .
pioneer work of ref.@xcite has shown a natural mechanism to suppress the unwanted neutral flavor changing process mediated by the higgs exchange .
recent activities address the general structure of the minimal flavor violation ( mfv ) @xcite .
it is noticed that octet - colored scalars are able to respect mfv .
the general yukawa interaction is given by @xmath0 where @xmath1 refers to the three families of left - handed quark doublets , and @xmath2 and @xmath3 the three families of the up - type and down - type right - handed quark singlets .
the new set of octet - colored scalars @xmath4 and @xmath5 couple to quarks in an analogous fashion of the usual higgs @xmath6 and @xmath7 by the yukawa coupling matrices @xmath8 and @xmath9 with additional proportional coefficients @xmath10 and @xmath11 , as well as the color generator @xmath12 .
all higgs bosons @xmath13 and @xmath14 carry the weak hypercharge @xmath15 , in the convention @xmath16 .
the usual conjugation operation @xmath17 is adopted .
all interactions respect a @xmath18-symmetry , in which down - type fields @xmath7 , @xmath5 , @xmath19 are odd , @xmath20 , but up - type fields @xmath6 , @xmath4 , @xmath21 are even .
the symmetry is used to keep the desirable pattern of interaction that obeys the rule of the minimal flavor violation ( mfv ) .
the complex coefficients @xmath10 and @xmath11 can produce cp violating vertices for the electric dipole moment ( edm ) of the neutron .
however the neutron edm is masked with uncertainty of the hadronic matrix evaluations .
it is interesting to ask the question of any contribution to the electron edm , reached by ongoing and future experiments@xcite .
it turns out that one - loop or two - loop contributions to the electron edm are absent if there is only one octet - colored scalar electroweak doublet .
nonetheless , in this article we show that with two of these octet - colored scalars , the electron edm occurs through an amplitude @xmath22 of the pseudoscalar neutral higgs boson @xmath23 into a pair of octet - colored charged scalars .
then , a cp even amplitude for @xmath24 appear even though @xmath23 is cp - odd . as @xmath23 and one of the photon touch down to an electron line ,
the electron edm emerges via barr - zee@xcite mechanism in an overall two - loop diagram , as shown in fig . 1 . in the past
, the electron edm via barr - zee two - loop mechanism has been exhaustively studied in scenarios of multi - higgs doublets@xcite ( without octet - colored scalars ) or supersymmetric particles@xcite .
our current study uses similar bosonic contribution as in @xcite .
we have found that the present limit of the electron edm has already restricted the parameter space of cp violation of the octet - colored scalars .
as we know that if a cp - odd scalar couples to a conjugated pair of bosons , cp conservation is violated , just like the case in @xmath26 .
we are going to find such a coupling in the mfv scenario with two colorless doublet higgs @xmath6 , @xmath7 , and two octet - colored higgs @xmath4 , @xmath5 . rotating the basis of colorless higgs , @xmath27
we achieve that @xmath28 as @xmath29 .
the relevant vertices , allowed by the exact @xmath30 , are @xmath31 we start with the first term @xmath32 we single out the pseudoscalar higgs @xmath23 from im(@xmath33 ) , or @xmath34 with @xmath35 , @xmath36 @xmath37 including the @xmath38 coupling , we combine results and obtain @xmath39 now we look at the charged components of the color octets .
the light and heavy mass states @xmath40 and @xmath41 are given by the diagonalization of the mass matrix @xmath42 where @xmath43 and @xmath44 are real , and the complex off - diagonal mass squared @xmath45 , @xmath46 we can absorb the phase @xmath47 of latexmath:[$(\eta'-\eta ) \equiv
@xmath25 coupling matrix becomes real and antisymmetric , but the remaining phase @xmath49 in the off - diagonal mass squared is genuine and not removable .
the unitary diagonalization transformation is usually complex , @xmath50 where the @xmath51 unitary matrix is denoted as @xmath52 .
@xmath53 note that diagonal entries arise due to the cp violating phase @xmath54 .
traceless is a consistent check .
we denote @xmath55 .
the relevant diagonal @xmath25 interaction becomes @xmath56 the interaction can produce the electron edm via barr - zee mechanism with the charged octet higgs @xmath57 in the inner loop .
also note that if masses of eigenstates @xmath40 and @xmath41 become equal , the cp violating effect disappears .
other quartic couplings not given in eq.([quartics ] ) do not generate the interaction between @xmath23 and the @xmath57 pair . as the neutral pseudoscalar @xmath23 is not constrained very much from lep data
, we can choose a light value , for example @xmath58 about 100 gev in our numerical analysis .
the four degrees of freedom in the two real @xmath43 and @xmath59 and the complex @xmath60 can be replaced by the other four parameters , two masses @xmath61 and @xmath62 of the light and heavy charged octet higgs , and two angles @xmath63 and @xmath64 .
the cp violating short distance physics induces the effective vertex of the electric dipole @xmath65 of the fermion @xmath66 coupled with the electromagnetic field strength @xmath67 , @xmath68 the two - loop barr - zee diagram also involves the pseudoscalar coupling of @xmath23 to fermions , @xmath69 a \ , \ ] ] where the electron shares similar pattern of the yukawa coupling as the down quark .
evaluating the barr - zee diagram , we obtain @xmath70 \ .\ ] ] eight color channels of @xmath57 have been summed .
the relevant function is defined by @xmath71 which has asymptotic behaviors , @xmath72 the electron edm is evaluated as the difference between two contributions from the light and heavy color octets @xmath40 and @xmath41 .
we illustrate in fig .
2 the edm component due to the light @xmath40 pretending @xmath73 for given @xmath63 and @xmath64 .
the electron edm contribution from the light octet is plotted versus @xmath74 for various @xmath75 gev , assuming @xmath76 .
the current experimental constraint@xcite is shown by the horizontal double - line , which has already imposed a constraint on the parameter space of the model .
predicted electron edm versus @xmath74 for various @xmath77 300 ( solid ) , 500 ( dotted ) , 800 ( dashed ) , 1000 ( dashed - dotted ) gev , with @xmath76 .
the horizontal double - line is the present experimental upper bound .
, width=576 ] similarly , we can obtain the quark edm by rescaling parameters , @xmath78 we can also work out@xcite the associated color dipole moment ( cdm ) @xmath79 of the quark @xmath80 in the effective vertex by replacing color factors and couplings . on the other hand , ref.@xcite , has studied the contribution of the color octet to the gluon dipole@xcite , known as the weinberg operator@xcite , which is subjected to large renormalization suppression@xcite .
due to the sophisticated hadron physics , we only have a qualitative relation between the neutron edm and the cp violating coefficients , mainly based on naive dimensional analysis@xcite . on the contrary ,
the electron edm is well predicted because it is independent of the hadronic structure .
the octet - colored scalars if exist will be copiously produced@xcite by lhc in coming years because of its strong interaction and its larger color representation . when their masses and couplings are determined , the electron edm measurement is the key to disclose its nature of the cp violation .
we show in this paper that the electron edm measurement is sensitive to the cp violating phase in the octet - colored scalars .
xx s. l. glashow and s. weinberg , phys .
d * 15 * , 1958 ( 1977 ) .
a. v. manohar and m. b. wise , phys .
d * 74 * , 035009 ( 2006 ) [ arxiv : hep - ph/0606172 ] .
d. kawall , f. bay , s. bickman , y. jiang and d. demille , aip conf .
proc . * 698 * , 192 ( 2004 ) ; d. kawall , f. bay , s. bickman , y. jiang and d. demille , phys .
lett . * 92 * , 133007 ( 2004 ) [ arxiv : hep - ex/0309079 ] .
s. m. barr and a. zee , phys .
lett . * 65 * , 21 ( 1990 ) [ erratum - ibid .
* 65 * , 2920 ( 1990 ) ] .
d. chang , w.- y. keung and t. c. yuan , phys .
d * 43 * , 14 ( 1991 ) .
r. g. leigh , s. paban and r. m. xu , nucl .
b * 352 * , 45 ( 1991 ) .
d. chang , w.- y. keung and a. pilaftsis , phys .
lett . * 82 * , 900 ( 1999 ) [ erratum - ibid . *
83 * , 3972 ( 1999 ) ] [ arxiv : hep - ph/9811202 ] .
d. chang , w. f. chang and w.- y. keung , phys .
b * 478 * , 239 ( 2000 ) [ arxiv : hep - ph/9910465 ] ; a. pilaftsis , phys .
b * 471 * , 174 ( 1999 ) [ arxiv : hep - ph/9909485 ] ; d. chang , w. f. chang and w.- y. keung , phys .
d * 66 * , 116008 ( 2002 ) [ arxiv : hep - ph/0205084 ] . c. kao and r. m. xu , phys .
b * 296 * , 435 ( 1992 ) .
b. c. regan , e. d. commins , c. j. schmidt and d. demille , phys .
* 88 * ( 2002 ) 071805 .
d. chang , w.- y. keung and t. c. yuan , phys .
b * 251 * , 608 ( 1990 ) .
e. braaten , c. s. li and t. c. yuan , phys .
d * 42 * , 276 ( 1990 ) .
s. weinberg , phys .
lett . * 63 * , 2333 ( 1989 ) .
e. braaten , c. s. li and t. c. yuan , phys . rev . lett .
* 64 * , 1709 ( 1990 ) .
g. boyd , a. k. gupta , s. p. trivedi and m. b. wise , phys .
b * 241 * , 584 ( 1990 ) .
d. chang , w.- y. keung , c. s. li and t. c. yuan , phys .
b * 241 * , 589 ( 1990 ) .
a. manohar and h. georgi , nucl .
b * 234 * , 189 ( 1984 ) .
m. gerbush , t. j. khoo , d. j. phalen , a. pierce and d. tucker - smith , + arxiv:0710.3133 [ hep - ph ]
. m. i. gresham and m. b. wise , phys .
d * 76 * , 075003 ( 2007 ) [ arxiv:0706.0909 [ hep - ph ] ]
. b. a. dobrescu , k. kong and r. mahbubani , arxiv:0709.2378 [ hep - ph ] . | an appended sector of two octet - colored scalars , each an electroweak doublet , is an interesting extension of the simple two higgs doublet model motivated by the minimal flavor violation .
their rich cp violating interaction gives rise to a sizable electron electric dipole moment , besides the quark electric dipole moment via the two - loop contribution of barr - zee mechanism .
* electron electric dipole moment induced by octet - colored scalars * jae ho heo and wai - yee keung physics department , university of illinois at chicago , chicago , il 60607 - 7059 |
the optical emission from low mass x - ray binaries ( lmxbs ) is generally dominated by the reprocessing of x - rays in the accretion disk and/or the mass donor ( see van paradijs & mcclintock 1995 for a review ) .
the time history of the optical emission is thus a convolution of the x - ray intensity history with a function representing the spatial distribution of matter in the system .
the most spectacular examples of this phenomenon are the optical novae accompanying soft x - ray transients , which can cause the optical counterpart to brighten by many magnitudes . of greater interest
are those systems whose x - ray intensity histories are more regularly or sharply modulated .
many lmxbs show periodic optical variability due to orbitally - modulated viewing of the x - ray heated mass donor , providing constraints on the binary inclination and the distribution of reprocessing material in the system ( see van paradijs & mcclintock 1995 and references therein ) .
quasi - periodic oscillations ( qpos ) at various frequencies have been observed in both the x - ray and optical emission from the candidate black hole binary gx 3394 ( motch , ilovaisky , & chevalier 1982 ; motch et al .
1983 ; imamura et al . 1990 ) .
correlated x - ray and optical bursts have been observed from several systems , and the time delay and smearing of the optical burst profiles with respect to the x - ray bursts have been used to probe the location and distribution of matter near the neutron star ( lewin , van paradijs , & taam 1993 ) .
coherent optical pulsations have been detected from the lmxb pulsars her x-1 ( davidsen et al .
1972 ; middleditch & nelson 1981 ) , 4u 162667 ( ilovaisky , motch , & chevalier 1978 ; middleditch et al .
1981 ) , and gx 1 + 4 ( jablonski et al . 1997 ) , and likewise provide an important probe of the binary parameters and the distribution of reprocessing material in the system .
the lmxb 4u 162667/kz tra consists of a 7.66 s pulsar accreting from an extremely low - mass companion in an ultracompact binary with a separation of @xmath4 lt - sec ( middleditch et al .
1981 ; levine et al . 1988 ) .
the 7.66 s x - ray pulsations arise due to anisotropic accretion of matter on the surface of a rotating , highly - magnetized neutron star whose spin and magnetic dipole axes are misaligned .
the spin frequency of the pulsar evolves on short ( @xmath5 yr ) time scales due to accretion torques ( chakrabarty et al .
the optical counterpart has a strong ultraviolet excess ( mcclintock et al .
1977 ) , and optical pulsations are detected with the same frequency as the x - ray pulsations ( ilovaisky et al .
the optical emission is understood as primarily due to reprocessing of the incident x - ray flux by material in the accretion disk and/or the binary companion s surface ( chester 1979 ; mcclintock et al .
the system shows strong , correlated x - ray / optical flares every @xmath6 s that are of undetermined origin ( joss , avni , & rappaport 1978 ; mcclintock et al .
1980 ; li et al .
a @xmath7 hz qpo has been detected in the x - ray emission ( shinoda et al .
1990 ; angelini et al .
1995 ) , and probably arises through an interaction between the pulsar s magnetosphere and the inner edge of the accretion disk ( alpar & shaham 1985 ; lamb et al .
1985 ) .
high - sensitivity timing of the optical pulsations detected an additional weak , persistent pulsation in a lower - frequency sideband of the `` direct '' ( x - ray ) pulse frequency ( middleditch et al .
this sideband was attributed to x - ray reprocessing on the surface of the companion , so that the 0.4 mhz downshift is due to the different effective pulsar rotation rate observed in the companion frame due to the binary orbit .
the observed frequency shift implies a 42-min prograde orbit .
x - ray timing measurements have repeatedly failed to detect periodic pulse arrival time delays or pulse frequency shifts due to a binary orbit , yielding an upper limit of @xmath8 lt - ms for the projected radius of the neutron star orbit ( levine et al .
1988 ; shinoda et al .
1990 ) . for a 42-min orbit ,
the x - ray timing limits imply a mass function @xmath9 , one of the smallest known for any stellar binary .
this paper describes a program to confirm the persistent orbital sideband reported by middleditch et al .
( 1981 ) and also provides the first detailed study of the optical wavelength dependence of x - ray reprocessing in this system . as part of this program
, the first example of an optical qpo in an x - ray pulsar system was discovered . a preliminary account of this work has appeared previously ( chakrabarty 1996 ) .
high speed multicolor optical photometry of 4u 162667/kz tra was obtained on ut 1995 may 2628 ( mjd 4986349865 ) with the 4-m blanco telescope at the cerro tololo inter - american observatory ( ctio ) in la serena , chile .
the observations used the automated single channel aperture photometer ( ascap ) and a dry - ice - cooled varian vpm-159a photomultiplier tube at the @xmath10 ritchey - chretien focus , along with the _ ubvri _ filter set described by graham ( 1982 ) .
the varian phototube , which has an ingaasp photocathode , is sensitive over a wide wavelength range , with a quantum efficiency of 15% at 4000 and 5% at 9000 .
all the observations were made through a 6.6 arcsec circular aperture .
the data were recorded at 1 ms resolution and rebinned to 100 ms resolution for timing analysis .
a log of the long timing observations of 4u 162667 is given in table 1 .
typical target and sky count rates are given in table 2 .
the graham ( 1982 ) photometric standard star e7-s was also observed at various airmasses on each night as a calibration source .
timing stability was maintained using a kinemetrics truetime 468-dc goes satellite synchronized clock .
this clock was synchronized using nist timing signals transmitted via the noaa _ geostationary operational environmental satellite 7 ( goes / west ) _ , located in geostationary orbit at 135@xmath11 west longitude ( beehler & lombardi 1990 ) .
the signals were corrected for the 267 ms mean path delay from the noaa ground station at wallops island , virginia up to _ goes / west _ and down to the ctio receiver ( corresponding to a setting of 57 ms for the propagation delay switch on the clock ) . according to the manufacturer specifications , the clock in this configuration maintains absolute time accuracy within @xmath12 ms of utc(nist ) .
however , an independent check of this accuracy ( e.g. , a cross - check with a global positioning system timing signal or optical pulse timing of a well - established astrophysical time standard like the crab pulsar ) was not available during these observations . some of the long timing observations contained a strong instrumental signal at 60 hz and its higher harmonics in the 1 ms data ( see table 1 ) .
whenever the 60 hz signal was present , it was accompanied by a second strong signal at 1.75 hz and its higher harmonics , and the two signals beat against each other .
these high frequency contaminants appeared intermittently in observations of several different targets .
however , their presence did not affect any of the science analysis for 4u 162667 , all of which was confined to pulse frequencies below 1 hz .
absolute photometry of 4u 162667 on may 26 yielded the following magnitude and colors : @xmath13 , @xmath14 , @xmath15 , and @xmath16 .
the individual magnitudes are given in table 2 .
the @xmath17-band sky background observation was too short to allow determination of a reliable source magnitude , but the presence of 7.66-s @xmath17-band pulsations , in phase with those in the bluer bands ( see below ) , confirms that the source was detected in the @xmath17-band as well . assuming that the @xmath17-band pulsed fraction is also 6% ( as in the other optical bands ; see below ) , the source has @xmath18 . within the uncertainties ,
interstellar reddening can be neglected for this high latitude source , since the total galactic reddening in the direction of 4u 162667 is @xmath19 ( burstein & heiles 1982 ) .
the @xmath20 magnitudes reported here are virtually identical to those measured on 1977 june 12 by mcclintock et al .
( 1977 ) , and they agree within 2@xmath21 with those measured on 1977 august 17 by grindlay ( 1978 ) . however , the present @xmath22-band measurement is 1.8(4 ) magnitudes brighter than previously reported by grindlay ( 1978 ) .
the observed time series were transformed to the solar system barycenter frame using the jet propulsion laboratory de-200 solar system ephemeris ( standish et al .
power spectra of the individual observations in all five bands showed strong 7.66 s pulsations , with a mean barycentric pulse frequency of 130.4398(7 ) mhz .
this is consistent with the x - ray pulsation frequency measured at the midpoint of the ctio observations by the batse all - sky monitor on the _ compton gamma ray observatory _
( chakrabarty et al . 1997 ) . to facilitate a study of the pulse shape and phase relationship of the optical pulsations at different wavelengths ,
a precise pulse phase ephemeris was derived from a subset of the batse observations ( 1995 may 9 1995 june 11 ) contemporaneous with the ctio observations .
based on these data , the barycentric x - ray pulse phase during the ctio observations is well fit by @xmath23 where @xmath24 , @xmath25 hz , @xmath26 hz s@xmath3 , @xmath27 1995 may 27.5 tdb , and pulse phase is defined relative to the intensity minimum of the fundamental harmonic of the 2060 kev batse pulse profile .
the ctio timing observations for each of the five filters were folded according to the timing model in equation ( 1 ) .
the resulting pulse profiles , along with the batse 2060 kev pulse profile , are shown in figure 1 relative to the batse pulse phase .
the mean optical pulse and the batse hard x - ray pulse agree in phase to within 5% ( i.e. , a relative time delay of @xmath28 s ) , consistent with previous measurements ( mcclintock et al .
both the batse hard x - ray pulse and the optical pulse shapes have a dominant sinusoidal component , although the optical pulses clearly have additional harmonic content as well .
the optical pulse shapes do not resemble any of the x - ray pulse shapes in the 213 kev range , all of which have an unusual and distinctive close double - peak structure ( levine et al .
instead , the optical pulse shapes resemble the low - energy ( 12 kev ) and high - energy ( 1560 kev ) pulse shapes .
this is consistent with previous observations by mcclintock et al .
( 1980 ) , who concluded that the 12 kev x - rays were chiefly responsible for the reprocessed optical emission .
the pulsed fraction , independent of wavelength , is approximately 6% ( see table 2 ) , suggesting that the optical emission is dominated by x - ray reprocessing .
figure 2 shows the fourier power spectrum of a 2-hour @xmath29-band observation ( run 15u ) .
the spectrum has been rebinned into uniform logarithmic frequency intervals , the mean noise level due to poisson counting statistics subtracted off , and the power level normalized relative to the double - sided mean source power ( e.g. , miyamoto et al .
in addition to the 130 mhz pulsation , a 0.048 hz qpo was also detected .
this feature was fit with a lorentzian profile ( e.g. , shibazaki & lamb 1987 ) , @xmath30 where @xmath31 is the centroid frequency of the qpo and @xmath32 is its full width at half - maximum ( fwhm ) .
the fractional root - mean - squared ( rms ) amplitude of the qpo was computed by integrating equation ( [ eq - qpo ] ) over the fwhm of the feature and taking the square root , yielding 3.4% for this observation .
the qpo is also detected in the @xmath33 and @xmath22-band observations .
the best - fit parameters for the qpo features are included in table 2 .
these measurements provide the first detection of an optical qpo from an x - ray pulsar .
a similar feature is observed at x - ray wavelengths ( shinoda et al .
1990 ; angelini et al . 1995 ; chakrabarty et al .
1997 , in preparation ) . for comparison ,
the fourier power spectrum of the 26 kev data from a 71 ks observation of 4u 162667 with the proportional counter array on the _ rossi x - ray timing explorer ( rxte ) _ is shown in figure 3 .
this observation was made on 1996 february 11 .
the qpo is clearly visible in the x - ray data with an rms strength of 6% , along with several harmonics of the coherent 7.66 s pulsation .
( the higher harmonics are very strong due to the very sharp 26 kev x - ray pulse shape of 4u 162667 . )
the _ rxte _ data will be discussed in further detail elsewhere ( chakrabarty et al . 1997 , in preparation ) . to study the detailed structure of the optical timing properties close to the pulsation frequency ,
an overresolved power spectrum was constructed by computing the fourier power @xmath34 at fractional bin intervals according to @xmath35 with @xmath36 , where @xmath37 is just the original time series @xmath38 multiplied by complex phase factors @xmath39 .
note that while the @xmath40 shifted power spectrum bins @xmath41 are statistically independent of each other , they are _ not _ independent of the original power spectrum @xmath42 ( although the covariances are easily computed ; see , e.g. , jenkins & watts 1968 ) . generalizing this process , one can construct an @xmath43-times overresolved power spectrum by interleaving the bins from the original power spectrum and @xmath44 frequency - shifted power spectra . operationally , this is equivalent to padding the original @xmath40-point zero - mean time series with @xmath45 zeros and computing the power spectrum of the new @xmath46-point padded time series .
use of an oversampled power spectrum gives more uniform frequency sensitivity at the expense of independent frequency bins , recovering fourier phase information which is normally lost when constructing the power spectrum from the fourier amplitudes .
figure 4 shows a 5-times overresolved power spectrum of 4u 162667 in the immediate vicinity of the main pulsation .
the power spectrum has been normalized relative to the mean noise power . to improve the signal - to - noise ratio ,
the power spectrum shown is the average of power spectra from three separate 1.8-hr observations ( runs 15u , 17b , and 23r ) . in addition to the fundamental pulsation at 130.4398(37 )
mhz , several other features are present . in considering these features , it is important to recall that the peaks due to periodic signals in a finite - length power spectrum are always accompanied by sidelobes .
this can be understood by considering a finite time series as the product of an infinite data stream and a rectangular ( boxcar ) observation window function ( equal to unity during the observation and zero elsewhere ) . by the fourier convolution theorem ,
the fourier transform of this product is equal to the convolution of the fourier transforms of the infinite data stream and the observation window function ( e.g. , press et al .
thus the power at frequency @xmath47 due to a sinusoidal signal with frequency @xmath48 in an ungapped observation of length @xmath49 is @xmath50 ^ 2,\ ] ] where @xmath51 is the power at the signal frequency .
the dotted curve in figure 4 shows the fit of this function to the data ( with the expected mean noise level of unity added ) , assuming that @xmath52 mhz corresponds to the main pulsation peak .
the expected sidelobe structure almost exactly accounts for the two symmetric sidelobes on either side of the main peak .
however , after accounting for the expected sidelobe structure of the main pulsation , a residual power excess is clearly evident at @xmath53 mhz , downshifted 0.395(15 ) mhz from the main pulsation peak .
this is consistent with the 0.4011(21 ) mhz shift for the lower sideband reported by middleditch ( 1981 ) .
the amplitude of the sideband relative to the main pulsation is 25% ( compared to 20% measured by middleditch et al .
1981 ) , corresponding to a sideband pulsed fraction of 1.5% . to evaluate the statistical significance of this excess power , note that the powers in an average of three background - dominated power spectra obey ( within a constant scaling factor ) a @xmath54-distribution with six degrees of freedom . for a power spectrum
normalized such that the mean noise power is unity , the probability of a noise fluctuation in a single bin exceeding a threshold power @xmath55 is @xmath56 thus , the observed excess power of 5.9 has a probability of @xmath57 of being due to a random fluctuation , making it significant at the 4.6@xmath21 level .
accounting for the fact that all bins in the range 0.1290.132 hz were searched reduces the statistical significance to 4.0@xmath21 .
it is especially striking to note that the frequency _ shift _ of the sideband agrees with that reported by middleditch et al .
( 1981 ) despite the fact that the actual value of the main pulsation frequency ( corresponding to the pulsar s spin ) has changed from 130.26 mhz to 130.44 mhz in the 15 years between the two observations ( chakrabarty et al . 1997 ) .
our sideband detection strongly corroborates the 42-min binary period reported by middleditch et al .
unfortunately , the signal - to - noise ratio of this detection is insufficient to permit a more detailed investigation of the binary parameters .
qpos in accretion - powered pulsars are widely believed to arise from interactions between the pulsar magnetosphere and the inner edge of an accretion disk . in the beat frequency model ( alpar & shaham 1985 ; lamb et al .
1985 ) , the qpo arises when inhomogeneous `` clumps '' of material at the inner edge of the disk are captured by the magnetosphere and accreted by the pulsar , producing a broad power spectral feature at the beat frequency @xmath58 between the keplerian frequency at the inner disk edge and the pulsar s spin frequency .
[ an alternative explanation , the keplerian frequency model of van der klis et al .
( 1987 ) , predicts @xmath59 .
however , it can only explain qpos where @xmath60 , and thus is not applicable in 4u 162667 .
] thus , the 0.048 hz x - ray qpo in 4u 162667 bolsters the premise that this is a disk - fed binary .
the discovery of an optical qpo at the same frequency shows that a significant fraction of the optical luminosity from the binary simply mirrors the x - ray emission , and must therefore result from x - ray reprocessing . as noted by middleditch et al .
( 1981 ) , the existence of two distinct , persistent , coherent optical pulsation frequencies suggests that the reprocessing occurs in two separate sites , most likely the surface of the mass donor and the accretion disk .
simultaneous x - ray and optical timing of these pulsations and the qpo may therefore provide a powerful probe of the geometry of the binary and physics of the accretion disk ( e.g. , arons & king 1993 ) .
unfortunately , the batse hard x - ray measurements acquired simultaneously with the optical observations require @xmath61 d to detect the x - ray pulsations , making them useless for a correlative study on short time scales . however , more sensitive x - ray timing missions ( e.g. _ rxte _ ) are well suited to such a task if the observations can be coordinated with ground - based optical observations .
a comparison of figures 2 and 3 gives a flavor of the kind of data sets that can be directly intercompared if measured simultaneously .
ultraviolet timing observations with the _ hubble space telescope _ may also provide an important probe , as the reprocessing in the disk should be more efficient at these wavelengths than in the optical ( arons & king 1993 ; anderson et al .
more generally , time - resolved ultraviolet / optical / infrared photometry may provide an important test for candidate multiwavelength counterparts in time - variable lmxbs whose companions remain unidentified .
while x - ray binaries and cataclysmic variables with a hydrogen - rich mass donor can not reach orbital periods @xmath62 min ( paczynski & sienkiewicz 1981 ) , those with a hydrogen - depleted mass donor can evolve to extraordinarily short ( @xmath6310 min ) orbital periods while maintaining high mass transfer rates ( nelson , rappaport , & joss 1986 ) .
few such systems are known , including the six am cvn cataclysmic variables ( see warner 1995 ) and the x - ray bursters 4u 182030 ( @xmath64 min ) and 4u 191605 ( @xmath65 min ) .
the recently reported 20.6-min ultraviolet modulation in the x - ray burster 4u 1850087 is probably due to the same phenomenon ( homer et al .
1996 ) . with a 42-min binary period ,
4u 162667 is clearly a member of this class as well ( levine et al .
1988 ) .
in addition to the inferred 42-min binary period , there are several other constraints on scenarios for the mass donor in 4u 162667 . the stringent x - ray timing limit ( @xmath66 lt - ms ; levine et al .
1988 ; shinoda et al .
1990 ) on the projected radius of the pulsar orbit implies @xmath67 where @xmath68 is the mass ratio , @xmath69 is the companion mass , @xmath70 is the neutron star mass @xmath71 in units of @xmath72 , and @xmath73 is the binary period in units of 42 min .
the _ a priori _ probability of observing a system within the inclination limit @xmath74 is given by @xmath75 .
a second constraint is the assumption that the mass transfer in an ultracompact binary is driven by angular momentum losses via gravitational radiation ( e.g. , levine et al .
. the resulting mass transfer rate will be ( e.g. , verbunt & van den heuvel 1995 and references therein ) @xmath76 where @xmath43 is the power - law exponent in the companion s mass - radius relation ( @xmath77 ) , and conservative mass transfer via roche lobe overflow has been assumed . since the bolometric x - ray flux from 4u 162667 has been measured ( @xmath78 erg @xmath79 s@xmath3 in the 0.560 kev band ; pravdo et al .
1979 ) , this also constrains the source distance and x - ray luminosity @xmath80 , where @xmath81 km is the neutron star radius and @xmath82 is the distance to the source .
another constraint is a lower limit on @xmath83 from the observed spin history of the pulsar ( chakrabarty et al .
if the pulsar is spinning up , then the maximum possible accretion torque on the neutron star will occur when the magnetospheric radius @xmath84 ( where the magnetic energy density is comparable to the kinetic energy density of the accreting matter ) is close to the corotation radius @xmath85 ( where the magnetic field lines move with the local disk velocity ) .
this limit may be written as @xmath86 , where @xmath87 for a keplerian disk .
however , the torque on the neutron star can be rewritten as @xmath88 where @xmath89 is the pulsar s moment of inertia ( ravenhall & pethick 1994 ) and @xmath90 is the spin frequency derivative a directly measurable quantity . noting that @xmath91 hz s@xmath3 was measured during the spin - up of 4u 162667 , chakrabarty et al .
( 1997 ) thus inferred that @xmath92 yr@xmath3 .
given the measured x - ray flux , the @xmath83 limit requires @xmath93 kpc . in view of the source s high galactic latitude ( @xmath94 ) ,
the resulting distance out of the plane is also a consideration when contemplating mass donors .
however , this need nt be a strong consideration , since it is plausible for neutron star lmxbs to have high velocities due to a supernova `` kick '' , leading to a considerable distance from the plane over the long x - ray lifetime of the source . indeed , the high - latitude ( @xmath95 ) lmxb pulsar her x-1/hz her is @xmath96 kpc distant , over 4 kpc out of the plane ( reynolds et al .
1997 ) .
the possible types of roche - lobe - filling mass donor for 4u 162667 have been discussed previously by levine et al .
( 1988 ) and verbunt , wijers , & burm ( 1990 ) .
i review and update those discussions here .
there are three possible types of roche - lobe - filling donors for a 42-min neutron star binary .
the first possibility is a @xmath97 helium or carbon - oxygen white dwarf . from equation ( [ eq - incl ] )
, the binary inclination for this case must be @xmath98 , with an _ a priori _ observation probability of 16% .
the mass - radius relation for such a star is @xmath99 ( shapiro & teukolsky 1983 ) , so the mass transfer rate from equation ( [ eq - mdot ] ) is @xmath100 yr@xmath3 .
this corresponds to a distance of about 1 kpc .
although the this case has the most favorable inclination constraint of the roche - lobe - filling options ( due to the extremely small @xmath69 ) , the mass transfer rate driven by gravitational radiation alone is an order of magnitude below the minimum mass accretion rate inferred by chakrabarty et al .
( 1997 ) .
the second roche - lobe - filling possibility is a @xmath101 hydrogen - depleted star which is not fully degenerate , with @xmath102 and an _ a priori _ probability of 1% . assuming @xmath103 , equation ( [ eq - mdot ] )
gives @xmath104 yr@xmath3 , corresponding to a distance of 3 kpc .
these are identical to the lower limits on @xmath83 and @xmath82 inferred from the pulsar spin - up measurements , and thus are just consistent with the observations .
finally , a @xmath105 roche - lobe - filling helium - burning star is also a possible donor for a 42-min neutron star binary .
however , the small allowed inclination range ( @xmath106 ) has an _ a priori _ observation probability of only 0.03% , and the expected mass transfer rate ( @xmath107 yr@xmath3 ; savonije , de kool , & van den heuvel 1986 ) would require an unphysically large distance ( 36 kpc ) to be consistent with the observed x - ray flux .
angelini et al .
( 1995 ) have proposed an alternative to a roche - lobe - filling donor .
these authors detected an unusually strong complex of neon emission lines in the x - ray spectrum of 4u 162667 . since neon is a product of helium burning , angelini et al .
( 1995 ) suggest that the mass donor is a helium - burning star which underfills its roche lobe , transferring matter to the neutron star via a radiatively - driven wind .
this possibility is of particular interest given the suggestion that the current spin - down of the pulsar is due to accretion from a retrograde disk ( nelson et al .
1997 ) , since formation of such a disk is much more plausible for wind accretion than for roche lobe overflow .
however , the wind from a low - mass helium - burning star would not provide a high enough mass transfer rate to be consistent with the x - ray data , as is now shown .
a mass of @xmath108 is required for core helium burning ( kippenhahn & weigert 1990 ) .
a @xmath109 helium - burning donor would fill about 25% of its roche lobe ( eggleton 1983 ) and would have @xmath110 and an effective temperature @xmath111 k ( savonije et al . 1986 ) , yielding a blackbody spectrum consistent with or fainter than the optical photometry for distances @xmath112 kpc .
more generally , helium - burning donors with @xmath113 would underfill the roche lobe and be consistent with or fainter than the optical measurements for somewhat larger distances , with @xmath114 and @xmath115 k. assuming a radiatively - driven wind similar to that observed in other hot stars , the wind mass loss rate with these parameters would be @xmath116 yr@xmath3 ( abbott 1982 ) . especially in light of the poor accretion efficiency expected for a wind - fed system ( @xmath117 ; e.g. , frank , king , & raine 1992 ) , this is several orders of magnitude too small to be consistent with the minimum accretion rate , @xmath118 yr@xmath3 .
similar considerations seem to rule out any plausible donor transferring matter via a conventional radiatively - driven wind . on the other hand
, a self - excited wind arising from x - ray heating of the mass donor might be a significant contributor to the mass transfer rate , whether or not the donor fills its roche lobe ( see tavani & london 1993 and references therein ) .
it is instructive to consider whether a stellar spectrum can explain the optical photometry .
the data can in fact be well fit by a single - temperature ( stellar ) blackbody spectrum , @xmath119 where @xmath120 is the flux density from the star per unit frequency @xmath47 , @xmath121 is the stellar radius , @xmath82 is the source distance , @xmath122 erg s is planck s constant , @xmath123 is the velocity of light , @xmath124 erg k@xmath3 is boltzmann s constant , and @xmath125 is the effective temperature of the star .
the best - fit parameters are @xmath126 k and @xmath127 , so that @xmath128 kpc .
a hydrogen - burning star with the correct temperature ( a main sequence or supergiant b star ) is obviously excluded both by the absurd ( @xmath129 kpc ) distance required and the infinitesimal _ a priori _ probability for the necessary binary inclination .
a @xmath130 he - ms star with @xmath131 kpc would have the correct temperature and flux .
however , as noted above , it would severely underfill its roche lobe and so probably could not transfer sufficient mass to the pulsar to explain the observed x - ray flux .
in addition , the implied binary inclination for a 42-min orbit has an a priori probability of @xmath132 0.1% .
however , assuming a typical white dwarf mass - radius relation , @xmath133 ( shapiro & teukolsky 1983 ) , we find @xmath134 kpc for a @xmath97 white dwarf .
as noted above , such a star would just fill its roche lobe , and the expected mass transfer rate would be consistent with the observed x - ray flux . for such a close binary , x - ray heating would certainly dominate the intrinsic luminosity of the degenerate dwarf .
assuming isotropic x - ray emission , we have @xmath135^{1/4 } = \left[\frac{l_x(1-\eta_*)}{4\pi\sigma a^2}\right]^{1/4 } \nonumber \\ & = & 2.1\times 10 ^ 4\ p_{42}^{-1/3}d_{\rm kpc}^{1/2}(1+q)^{-1/6}(1-\eta_*)^{1/4 } { \rm\ \ k } , \end{aligned}\ ] ] where @xmath136 is the binary separation , @xmath137 is the x - ray albedo of the white dwarf , and @xmath78 erg cm@xmath3 s@xmath3 has been used for the x - ray flux ( pravdo et al .
thus , x - ray heating of a low mass wd would naively seem to provide a plausible model for the observed optical emission .
unfortunately , this is an unrealistic model .
the 0.04 hz qpo detected in the x - ray and optical data underscores the presence of an accretion disk ( e.g. , van der klis 1995 ) which must contribute significant optical luminosity . the flux from a geometrically thin , optically thick disk can be written as @xmath139 - 1},\ ] ] where @xmath140 is the disk s surface temperature as a function of mid - plane radius @xmath141 .
given the x - ray timing limits , it is clear that @xmath142 .
the outer disk may be assumed to cut off sharply at the tidal radius of the neutron star ( @xmath143 ; frank et al .
1992 ) , corresponding @xmath144 cm for 4u 162667 .
the inner accretion disk should be disrupted by the pulsar s magnetosphere at roughly the neutron star s corotation radius , @xmath145 cm for a keplerian disk .
( strictly speaking , @xmath85 is an upper limit on @xmath146 , but the calculation is not sensitive to this quantity since most of the disk s optical emission comes from its outer parts . ) for an x - ray heated disk , the disk may be considered as consisting of up to three distinct regions : an innermost region powered primarily by internal viscous dissipation , a middle region of `` shallow '' x - ray heating , and an outer region of `` deep '' x - ray heating ( e.g. , cunningham 1976 ; arons & king 1993 ) .
let us consider each of these regions in turn .
the standard temperature profile for an unirradiated thin accretion disk is set by internal viscous dissipation ( shakura & sunyaev 1973 ; frank et al .
1992 ) @xmath147 where @xmath148 erg @xmath79 k@xmath149 s@xmath3 is the stefan - boltzmann constant , @xmath70 is the neutron star mass in units of 1.4 @xmath150 , @xmath151 is the mass accretion rate in units of @xmath152 yr@xmath3 , and @xmath153 is the mid - plane radius in units of @xmath154 cm . at sufficiently small radii , this internal heating will dominate over x - ray heating in setting the disk temperature . beyond a critical radius ,
however , x - ray heating of the disk surface will modify the temperature profile according to @xmath155 assuming that the disk is irradiated by a central x - ray point source , then we can write @xmath156 as @xmath157^{1/4 } \approx \left[\frac{l_x(1-\eta_d ) } { 4\pi\sigma r^2 } \left(\frac{dh}{dr } - \frac{h}{r}\right)\right]^{1/4 } , \label{eq - tx}\ ] ] where @xmath158 is the x - ray albedo of the disk , @xmath159 is the angle between the normal to the disk surface and the vector from the neutron star , and @xmath160 is the disk s scale height .
thus , we see that the temperature profile due to x - ray heating is sensitive to the functional form of @xmath161 through the approximation for @xmath162 ( which assumes @xmath163 ) . in the shallow x - ray heating region ,
the surface temperature profile is set by x - ray heating but the _ central _ temperature of the disk is still set by internal viscous dissipation .
thus , the disk thickness has the usual value for a standard unirradiated thin disk ( shakura & sunyaev 1973 ; frank et al .
1992 ) , @xmath164 where @xmath165 is an order unity dimensionless parametrization of the kinematic viscosity @xmath166 , @xmath167 is the isothermal sound speed , and @xmath168 is the mean molecular weight in units of the hydrogen atomic mass .
( note that @xmath169 for neutral hydrogen , 1/2 for ionized hydrogen , 4 for neutral helium , 2 for singly - ionized helium , and 4/3 for doubly - ionized helium . )
we thus find @xmath170 shallow x - ray heating will dominate internal viscous heating in setting the disk surface temperature when @xmath171 , which occurs at radii exceeding roughly @xmath172 cm .
if @xmath173 , then x - ray heating will also dominate internal heating in setting the _ central _ temperature of the disk ( lyutyi & sunyaev 1976 ; spruit 1995 ) . here
, @xmath174 is the disk s optical depth , given by @xmath175 for a standard @xmath165-disk ( frank et al .
1992 ) . in this region of deep x - ray heating
, the vertical structure of the disk will not obey the @xmath176 relationship in equation ( [ eq - h98 ] ) , but will instead follow a @xmath177 power law ( e.g. , cunningham 1976 ; vrtilek et al . 1990 ;
arons & king 1993 )
. however , deep x - ray heating will dominate over shallow heating only for radii exceeding roughly @xmath178 cm . in a binary
as compact as 4u 162667 , the disk will terminate before deep heating becomes dominant , so that a combination of internal viscous heating and shallow x - ray heating [ equation ( 12 ) along with equations ( 11 ) and ( 15 ) ] should be valid over the entire disk .
accounting for the presence of an x - ray heated accretion disk , it is clear that an x - ray heated low - mass white dwarf at a distance of 1 kpc is untenable for 4u 162667 , since the expected optical emission from the disk would be considerably brighter than the measured fluxes . instead , it is more appropriate to consider the opposite case of the disk providing most of the optical luminosity by fitting equations ( 10 ) and ( 12 ) along with equations ( 11 ) and ( 15 ) to the measured optical photometry .
the free parameters for the fit are @xmath82 ( or , equivalently , @xmath179 ) and @xmath158 .
as is evident from figure 5 , the x - ray heated disk model can easily fit the data .
however , as shown in figure 6 , the optical data do not provide a strong joint constraint on @xmath82 and @xmath158 .
ultraviolet observations will eventually provide a much stronger constraint , since the uv spectrum from an x - ray heated disk is much more sensitive to the values of @xmath179 and @xmath158 . in the absence of ultraviolet observations
, it should be noted that de jong , van paradijs , & augusteijn ( 1996 ) have concluded that lmxb accretion disks have a very high effective x - ray albedo ( @xmath180 ) , based on observations of 4u 125469 , 4u 175533 , and sco x-1 .
if this is the case , then the optical data constrain the distance to @xmath181 kpc ( 95% confidence ) , corresponding to an x - ray luminosity of @xmath182 erg s@xmath3 .
lcccccl 15u & 1995 may 26 & 0408 & 2.0 & @xmath29 & no & yes ( strong detection ) + 17b & 1995 may 26 & 0638 & 1.8 & @xmath33 & yes & yes ( strong detection ) + 21i & 1995 may 27 & 0018 & 1.7 & @xmath17 & no & no ( clouds ) + 23r & 1995 may 27 & 0308 & 1.9 & @xmath22 & no & yes ( strong detection ) + 25v2 & 1995 may 27 & 0604 & 0.6 & @xmath183 & yes & no ( clouds ) + 29u & 1995 may 28 & 0001 & 0.7 & @xmath29 & yes & no ( clouds ) + 30u & 1995 may 28 & 0046 & 1.0 & @xmath29 & yes & yes + 32v & 1995 may 28 & 0224 & 1.0 & @xmath183 & yes & yes + 35u & 1995 may 28 & 0709 & 1.0 & @xmath29 & yes & yes + 36i & 1995 may 28 & 0813 & 1.0 & @xmath17 & yes & yes ( weak detection ) + lccccc source count rate ( count s@xmath3 ) & 137(2 ) & 122(2 ) & 111(4 ) & 177(3 ) & pulsed count rate ( count s@xmath3 ) & 7.9(3 ) & 6.9(3 ) & 7.0(5 ) & 11.8(5 ) & 25(2 ) sky count rate ( count s@xmath3 ) & 168(2 ) & 182(2 ) & 317(4 ) & 785(3 ) & 5405(30 ) magnitude & 17.50(29 ) & 18.70(24 ) & 18.68(15 ) & 18.68(26 ) & pulsed fraction & 5.8(2)% & 5.7(3)% & 6.3(5)% & 6.7(3)% & qpo fractional rms strength & 3.4% & 2.7% & & 1.9% & qpo centroid frequency ( hz ) & 0.048 & 0.049 & & 0.049 & qpo fwhm ( hz ) & 0.013 & 0.012 & & 0.009 & figure 1 : optical pulse profiles as a function of wavelength for 4u 162667 . also shown is the 2060 kev x - ray pulse profile measured contemporaneously by batse .
two pulses are shown in each panel , and the pulse phase is plotted relative to the intensity minimum of the x - ray pulse . for the @xmath17-band optical and batse x - ray measurements , the intensity of the unpulsed ( dc ) component of the source emission was not measured .
figure 2 : fourier power density spectrum of a 2-hour @xmath29-band observation ( run 15u ) of 4u 162667 , normalized relative to the mean source power .
the dotted line indicates the noise power level due to poisson counting statistics , which has been subtracted off . a 0.048 hz quasi - periodic oscillation with 4.9% fractional rms amplitude
is clearly detected , as is the 7.66-s coherent pulsation due to the neutron star s rotation .
an underlying @xmath184 spectrum is measured on time scales @xmath185 s. figure 3 : fourier power density spectrum of a 71 ks 26 kev _ rxte_/pca observation of 4u 162667 on 1996 feb 11 , normalized relative to the mean x - ray source power .
the dotted line indicates the noise power level due to poisson counting statistics , which has been subtracted off . although this power spectrum can not be directly compared to the optical power spectrum in figure 2 ( since they were not acquired contemporaneously ) , the overall similarities are obvious and bolster the idea that the optical emission is dominated by rapid x - ray reprocessing .
several higher harmonics of the 0.1304 hz coherent pulsation are observed due to the highly non - sinusoidal x - ray pulse shape in the 26 kev band .
figure 4 : a 5@xmath186 over - resolved fourier power spectrum of 4u 162667 in the vicinity of the coherent 7.66-s pulsation , normalized relative to the local noise power .
the solid line is the average of three separate power spectra of 1.8-hr observations .
the dotted line is a model of the main pulsation s expected sidelobe due to the finite length of the power spectrum .
this model is able to explain the symmetric sidelobes on either side of the main pulsation .
a residual sideband downshifted 0.395(15 ) mhz from the main pulsation ( 25% relative amplitude ) is also present , with a statistical significance of 4@xmath21 .
this frequency shift is consistent with a @xmath187 s prograde binary orbit .
figure 5 : optical ( _ ubvri _ ) photometry of 4u 162667 on 1995 may 26 .
the solid curve indicates a typical x - ray heated disk solution .
the particular solution shown assumed @xmath188 , @xmath189 kpc , @xmath190 cm , and @xmath191 cm .
however , a range of values for @xmath158 and @xmath82 provide a good fit to the data ( see figure 6 ) .
figure 6 : distance to 4u 162667 as a function of effective x - ray albedo of the disk , assuming that all the optical flux is due to an x - ray heated accretion disk with @xmath190 cm , and @xmath192 cm .
the shaded region is consistent with the optical photometric data at the 95%-confidence level .
based on observations of other lmxbs , de jong et al .
( 1996 ) conclude that @xmath0 .
if this is correct , then the solutions allowed by the data must lie in the shaded region to the right of the vertical dotted line . | rapid _ ubvri _ photometry of the ultracompact low - mass x - ray binary ( lmxb ) pulsar 4u 162667/kz tra has detected 130.4 mhz ( 7.67 s ) optical pulsations in all five bands .
the optical pulsations , which are at the same frequency as the x - ray pulsations caused by rotation of the highly - magnetized accreting neutron star primary , are understood as reprocessing of the pulsed x - ray emission in the accretion disk or on the surface of the secondary .
the optical pulsed fraction is roughly 6% , independent of wavelength , indicating that the optical emission is dominated by x - ray reprocessing .
a weaker ( 1.5% ) sideband , downshifted 0.395(15 ) mhz from the main optical pulsation , is also present .
this is consistent with a previously reported sideband which was downshifted 0.4011(21 ) mhz from the main pulsation , corroborating the 42-min binary period proposed by middleditch et al .
( 1981 , apj , 244 , 1001 ) . a 0.048 hz optical quasi - periodic oscillation ( qpo ) , corresponding to a previously reported x - ray feature ,
was also detected in some of the observations , with a fractional rms amplitude of 35% .
this is the first measurement of an optical qpo in an x - ray binary pulsar .
i discuss constraints on the nature of the mass donor and show that mass transfer via a radiatively - driven wind is inconsistent with the optical data .
i also review the basic theory of x - ray - heated accretion disks and show that such models provide a good fit to the optical photometry .
if the effective x - ray albedo of lmxb accretion disks is as high as recently reported ( @xmath0 ) , then the optical data imply a distance of @xmath18 kpc and an x - ray luminosity of @xmath2 erg s@xmath3 .
submitted june 1 , 1997 to the astrophysical journal |
in this paper all graphs are finite and may have multiple edges but no loops .
a graph is _ simple _ if it has no multiple edges .
pseudograph _ may have multiple edges and loops . by a _ surface _
we mean a connected compact @xmath0-manifold without boundary .
the nonorientable surface of genus @xmath2 is denoted @xmath3 . by an _
open _ or _ closed disk _ in a surface we mean a subset of the surface homeomorphic to such a subset of @xmath4 .
a _ closed 2-cell embedding _ of a graph is an embedding such that every face is an open disk bounded by a cycle ( no repeated vertices ) in the graph .
a graph must be @xmath0-connected to have a closed @xmath0-cell embedding .
the _ strong embedding conjecture _ due to haggard @xcite says that every @xmath0-connected graph has a closed @xmath0-cell embedding in some surface .
the even stronger _
orientable strong embedding conjecture _ @xcite says that every @xmath0-connected graph has a closed @xmath0-cell embedding in some orientable surface .
the facial walks of every closed @xmath0-cell embedding form a _ cycle double cover _ , a set of cycles in the graph such that each edge is contained in exactly two of these cycles .
therefore , both embedding conjectures imply the well - known _ cycle double cover conjecture _ , which says that every @xmath0-edge - connected graph has a cycle double cover , every spherical embedding of a @xmath0-connected planar graph is an orientable closed @xmath0-cell embedding . for cubic graphs ( but not in general ) we can go from a cycle double cover back to a closed @xmath0-cell embedding ; thus , cubic graphs with cycle double covers ( see for example @xcite ) have closed @xmath0-cell embeddings .
some special classes of graphs are known to have minimum genus embeddings with all faces bounded by cycles ; these are closed @xmath0-cell embeddings .
for example , the complete graph @xmath5 has embeddings with all faces bounded by @xmath6-cycles in a nonorientable surface if @xmath7 , @xmath8 , @xmath6 or @xmath1 mod @xmath9 , and in an orientable surface if @xmath10 , @xmath6 , @xmath1 or @xmath11 mod @xmath12 ( see @xcite ) .
the complete bipartite graph @xmath13 has embeddings with all faces bounded by @xmath1-cycles in a nonorientable surface if @xmath14 is even , and in an orientable surface if @xmath15 is divisible by @xmath1 @xcite . however ,
in general not many graphs are known to have closed @xmath0-cell embeddings . even though the orientable strong embedding conjecture is very strong ,
the study of orientable closed @xmath0-cell embeddings seems to be promising .
one approach is to try to prove the following : [ nsetoose ] if a @xmath0-connected graph has a nonorientable closed @xmath0-cell embedding then it has an orientable closed @xmath0-cell embedding .
while this conjecture appears weaker than the orientable strong embedding conjecture , it is actually equivalent to it , via a result in @xcite based on techniques of little and ringeisen @xcite .
this result says that if a graph @xmath16 has an orientable closed @xmath0-cell embedding , and @xmath17 is a new edge , then @xmath18 has a closed @xmath0-cell embedding in _ some _ surface , which may or may not be orientable .
this paper is a first step towards verifying conjecture [ nsetoose ] .
our goal is to develop techniques to turn nonorientable closed @xmath0-cell embeddings into orientable closed @xmath0-cell embeddings . using techniques of this kind ,
we show that the orientable strong embedding conjecture is true for projective - planar cubic graphs .
the following is the main result of this paper .
[ mainose ] a standard construction then provides a result for general @xmath0-edge - connected projective - planar graphs . given such an embedded graph @xmath16 , we may construct a @xmath0-edge - connected ( hence @xmath0-connected ) cubic projective - planar graph @xmath19 by expanding each vertex @xmath20 of degree @xmath21 to a contractible cycle of length @xmath22 . by theorem [ mainose ]
, @xmath19 has an orientable closed @xmath0-cell embedding , whose oriented faces give a compatibly oriented ( each edge occurs once in each direction ) cycle double cover @xmath23 of @xmath19 . contracting the new cycles to recover @xmath16 , @xmath23 becomes a compatibly oriented cycle double cover @xmath24 of @xmath16 .
( @xmath24 may not correspond to a surface embedding of @xmath16 , but that does not matter . ) some cycles of @xmath23 may become oriented eulerian subgraphs , rather than cycles , of @xmath24 , but those can always be decomposed into oriented cycles , preserving the compatible orientation .
this proves the following .
[ mainocdc ] every @xmath0-edge - connected projective - planar graph has an orientable cycle double cover . in section [ defnot ]
we introduce some notation . in section [ nonori2ori ]
we develop surgeries that convert nonorientable surfaces into orientable surfaces , and use them to prove a special case of our main result . in section [ redn ] we describe some reductions for our problem . in section
[ main ] we arrive at our main result by proving a structural result which shows that either the special case from section [ nonori2ori ] occurs , or some kind of reduction applies .
let @xmath25 denote an embedding of a graph @xmath16 in a surface @xmath26 .
we usually identify the graph and the point - set of its image under the embedding . if @xmath27 , then @xmath28 denotes the closure of @xmath29 in @xmath26 .
face _ is a component of @xmath30 .
the boundary of the face @xmath31 is denoted by @xmath32 .
each component of @xmath32 is traced out by a closed walk in @xmath16 , which we call a _ facial boundary component walk _ of @xmath31 .
@xmath2-cycle face _ is a face with exactly one boundary component , which is a @xmath2-cycle .
an embedding in which every face is an open disk ( @xmath0-cell ) is an _
open @xmath0-cell embedding _ ; then each face has a single boundary component walk , called the _ facial walk_.
if every facial walk of an open @xmath0-cell embedding is in fact a cycle , we have a _ closed @xmath0-cell embedding_. if @xmath26 is not the sphere , then the _ representativity _ of any embedding @xmath25 is defined to be @xmath33 is a noncontractible simple closed curve in @xmath34 .
we say @xmath25 is _ @xmath2-representative _ if @xmath35 .
robertson and vitray @xcite showed that @xmath25 is open @xmath0-cell exactly when @xmath16 is connected and @xmath25 is @xmath8-representative , and closed @xmath0-cell exactly when @xmath16 is @xmath0-connected and @xmath25 is @xmath0-representative .
suppose @xmath36 is a cycle with a given orientation , and @xmath37 and @xmath20 are two vertices on @xmath36 .
denote by @xmath38 the path on @xmath36 from @xmath37 to @xmath20 in the given direction .
if we have a graph embedded on an orientable surface , then all cycles may be oriented in a consistent clockwise direction , and we assume this orientation unless otherwise specified .
if @xmath39 is a path , @xmath40 is defined to be the subpath from @xmath37 to @xmath20 along @xmath39 .
we say two sets ( usually faces ) @xmath31 and @xmath41 _ touch _ if @xmath42 ; we say they _
touch @xmath2 times _ if @xmath43 has @xmath2 components .
a sequence of sets @xmath44 is an _
@xmath45-face chain of length @xmath46 _ if for @xmath47 each @xmath48 is a distinct face of @xmath25 and for @xmath49 , @xmath48 and @xmath50 touch when @xmath51 and do not touch otherwise .
the sets @xmath52 and @xmath53 may be faces of @xmath25 , but need not be ; in this paper they are usually paths .
if @xmath54 and @xmath55 for @xmath47 , we say the face chain _ goes through @xmath56_. a cyclic sequence of distinct faces @xmath57 is called a _
face ring of length @xmath46 _ if ( i ) @xmath58 , and @xmath59 and @xmath60 touch at least twice , or ( ii ) @xmath61 and the sets @xmath62 , @xmath63 , are pairwise disjoint , nonempty when @xmath64 or @xmath65 , and empty otherwise ( subscripts interpreted modulo @xmath46 ) .
we will use face rings only in closed @xmath0-cell embeddings , so we do not need to consider situations in which a face ` touches itself ' , i.e. , face rings of length @xmath8 .
a face ring is _ elementary _ if ( i ) @xmath66 and the two faces touch exactly twice , or ( ii ) @xmath61 and any two faces touch at most once .
a face ring is _ noncontractible _ if @xmath67 contains a noncontractible simple closed curve .
the following observations will be useful for showing that face rings are elementary .
[ 3conn ] if @xmath25 is an embedding of a @xmath6-connected graph , and two faces are contained in some open disk , then the faces touch at most once .
[ 3conn3rep ] if @xmath25 is a @xmath6-representative embedding of a @xmath6-connected graph then any two faces touch at most once .
it is not hard to turn an orientable embedding into a nonorientable embedding , by adding a crosscap in an arbitrary location . on the other hand
, it is not easy in general to construct an orientable embedding from an existing nonorientable embedding of a graph .
the authors of @xcite developed some surgeries which turn embeddings on the projective plane and on the klein bottle into orientable embeddings .
however , the resulting orientable embeddings are not necessarily closed @xmath0-cell embeddings . in this section
we develop some surgeries to convert nonorientable embeddings into orientable embeddings .
we show that under certain conditions this can be applied to convert a closed @xmath0-cell projective - planar embedding into a closed @xmath0-cell orientable embedding
. our overall strategy will be as follows .
suppose @xmath16 is embedded in a nonorientable surface @xmath68 , obtainable by inserting a set @xmath69 of crosscaps in an orientable surface @xmath70 .
we insert an additional set @xmath71 of crosscaps to get an embedding of @xmath16 on a nonorientable surface @xmath72 , in which all facial boundary component walks are cycles .
then we embed in @xmath72 a pseudograph @xmath19 disjoint from @xmath16 , such that cutting along @xmath19 destroys all the crosscaps in @xmath73 . capping any holes due to cutting we
obtain an embedding of @xmath16 on an orientable surface @xmath74 , in which all facial boundary component walks are still cycles . removing any faces
that are not open disks , and capping again , we finish with a closed @xmath0-cell embedding of @xmath16 in an orientable surface @xmath75 . we first define some concepts related to crosscaps , next discuss insertion of crosscaps , then examine cutting to remove nonorientability , and finally apply these ideas to certain projective - planar embeddings .
we regard a _
crosscap _ in a @xmath0-manifold ( with or without boundary ) as just a one - sided simple closed curve in the interior of the manifold ( rather than the usual definition , where a crosscap is a neighborhood of such a curve , homeomorphic to a mbius strip ) .
to _ add a crosscap _ to a @xmath0-manifold we remove a point or a set homeomorphic to a closed disk not intersecting the boundary , locally close the result by adding a boundary component homeomorphic to a circle , then identify antipodal points of this circle to obtain a one - sided simple closed curve .
if we removed a point @xmath76 or closed disk @xmath77 , we call this _ inserting a crosscap at @xmath76 or @xmath77_. if @xmath76 is an interior point of an edge @xmath17 of an embedded graph , we call this _ inserting a crosscap on @xmath17_. in figures we represent a crosscap by a circle ( representing the added boundary component ) with an x inside it . to recover the original @xmath0-manifold ( up to homeomorphism )
we may _ collapse _ the crosscap by identifying all its points into a single point ( the result is homeomorphic to what we obtain by cutting out a mbius - strip neighborhood of the crosscap and capping the resulting hole with a disk ) .
we define the following surgeries for inserting crosscaps . these , or closely related operations , have been used in earlier papers such as @xcite , and generalize ideas used by haggard @xcite .
[ inscc ] * let @xmath59 and @xmath60 be two distinct face occurrences at a vertex @xmath20 .
( they may be different occurrences of the same face . ) then @xmath59 and @xmath60 partition the edges incident with @xmath20 into two intervals , @xmath78 and @xmath79 .
choose a closed disk @xmath77 close to @xmath20 such that all edges of @xmath78 cross it once , and no edges of @xmath79 intersect it .
add a crosscap at @xmath77 , re - embedding the parts of the edges of @xmath78 between @xmath77 and @xmath20 so that their order around @xmath20 is reversed .
we call this _ inserting a crosscap between @xmath59 and @xmath60 near @xmath20_. it does not matter whether we insert the crosscap across @xmath78 or @xmath79 : using the other one just amounts to pulling @xmath20 through the crosscap , which does not change the embedding . if one of @xmath78 or @xmath79 is a single edge , then this is equivalent to inserting a crosscap on that edge . *
if @xmath59 and @xmath60 are distinct faces and @xmath80 is a path @xmath39 ( possibly a single vertex ) then the effect of inserting a crosscap between @xmath59 and @xmath60 at any vertex of @xmath39 , or on any edge of @xmath39 , is the same , so we just talk about _ inserting a crosscap between @xmath59 and @xmath60_. * suppose @xmath81 is an elementary face ring of length @xmath82 .
interpreting subscripts modulo @xmath83 , insert a crosscap between @xmath48 and @xmath84 for @xmath85 .
we call this _ inserting crosscaps along @xmath86_. ( this can also be defined for elementary face rings of length @xmath0 , or for suitable face chains , although we do not need this here . ) while the above definition allows us some freedom in exactly how we place the crosscaps , in practice it may be convenient to make a definite choice about the location of the crosscaps , to help in tracing boundary component walks in the new embedding .
now we introduce our cutting operation .
our approach is fairly general , although we will need only a special case of it for our projective - planar results .
we first need to examine what happens if we cut through an individual crosscap on a surface .
suppose @xmath77 is a neighborhood of a crosscap @xmath87 , which we think of as a circle @xmath88 with antipodal points identified .
there is a consistent local orientation @xmath89 on @xmath90 .
cut along a curve @xmath91 which crosses @xmath87 at a single point @xmath76 .
if antipodal points of @xmath88 were not identified , we would cut @xmath77 into two pieces @xmath92 and @xmath93 . because we do identify antipodal points , @xmath94 and @xmath95 remain joined along the cut crosscap @xmath96 , a line segment joining two copies of @xmath76 .
@xmath94 and @xmath95 are joined with a twist so we now have a local orientation @xmath97 everywhere in the neighborhood of the cut crosscap @xmath96 , including on @xmath96 itself , which reverses relative to @xmath89 when we cross @xmath96 .
see figure [ fig - inscc ] : at left is a planned cut ; in the middle the cut crosscap is shown as two copies of the line segment @xmath96 that are to be identified ; and at right we see the result of the identification . as mentioned earlier
, we consider a nonorientable surface built up by adding crosscaps to an orientable surface ; we then cut through a pseudograph @xmath19 intersecting those crosscaps to remove the nonorientability .
[ oriset1 ] suppose @xmath98 is a @xmath0-face - colorable embedding of a pseudograph @xmath19 on an orientable surface @xmath26 .
suppose we choose a finite set of points @xmath39 so that each point of @xmath39 is an interior point of an edge of @xmath19 .
if we insert a crosscap @xmath99 at every @xmath100 , cut the resulting surface along all edges of @xmath19 , and use disks to cap the boundary components of the resulting @xmath0-manifold with boundary , the result is a finite union of pairwise disjoint orientable surfaces .
it suffices to show that the @xmath0-manifold with boundary @xmath101 obtained by cutting along @xmath19 is orientable ; then by capping we obtain a union of orientable surfaces .
consider a fixed global orientation @xmath89 ( local choice of clockwise direction ) for @xmath26 .
when we add the crosscaps , @xmath89 gives a consistent local orientation on a neighborhood of each crosscap @xmath99 , excluding @xmath99 itself .
let @xmath102 and @xmath103 be the two color classes of faces of @xmath98 .
define an orientation @xmath97 on @xmath101 that is equal to @xmath89 on each face in @xmath102 and opposite to @xmath89 on each face in @xmath103 .
when we pass between two faces of @xmath98 in @xmath101 , we pass between a face in @xmath102 and a face in @xmath103 via a cut crosscap .
as previously discussed , crossing the cut crosscap @xmath104 reverses the orientation , in agreement with @xmath97 .
therefore @xmath97 provides a consistent global orientation for @xmath101 . in lemma [ oriset1 ] ,
@xmath19 need not be connected .
also , we may allow @xmath19 to contain free loops , edges incident with no vertex , which map to simple closed curves in an embedding ; we can always insert vertices to turn these into ordinary loops . instead of starting with the pseudograph @xmath19 and inserting the crosscaps , we may start with the crosscaps and add @xmath19 . therefore , the following is equivalent to lemma [ oriset1 ] .
[ oriset2 ] suppose @xmath105 is a nonorientable surface containing disjoint crosscaps @xmath106 , @xmath107 , @xmath108 , @xmath109 .
suppose @xmath110 is an embedding on @xmath105 of a pseudograph @xmath19 , such that each @xmath111 contains exactly one point of @xmath19 , an interior point of an edge that crosses @xmath111 .
suppose that when we collapse every @xmath111 , @xmath112 , we get a @xmath0-face - colorable embedding of @xmath19 in an orientable surface @xmath26 .
then if we cut @xmath105 along all edges of @xmath19 and use disks to cap the boundary components of the resulting @xmath0-manifold with boundary , the result , @xmath113 , is a finite union of pairwise disjoint orientable surfaces . in operation
[ oriset2 ] , the condition that each crosscap @xmath111 be crossed by exactly one edge of @xmath19 , implying that the pseudograph after collapsing every @xmath111 is still @xmath19 , is important .
if this is not the case , then the embedding obtained by cutting along @xmath19 may not be orientable , even if collapsing the crosscaps yields a @xmath0-face - colorable orientable embedding of some graph .
for example , suppose we add four crosscaps @xmath114 to a sphere to get @xmath115 , and take @xmath19 to consist of two disjoint loops @xmath116 and @xmath117 , where @xmath116 and @xmath117 both cross @xmath106 and @xmath107 , @xmath118 is crossed only by @xmath116 and @xmath119 is crossed only by @xmath117 .
when we collapse the crosscaps we get a @xmath0-face - colorable embedding of a graph on a sphere , but if we cut @xmath105 along @xmath116 and @xmath117 the result is not orientable . with some extra conditions it is possible to allow more than one edge of @xmath19 to cross a crosscap @xmath111 , but we will not pursue the details here .
the pseudograph @xmath19 in operation [ oriset2 ] indicates where to cut .
the graph we wish to embed is a different graph , @xmath16 .
we assume that @xmath16 also has an embedding @xmath120 in @xmath105 , disjoint from the embedding @xmath110 of @xmath19 .
note that @xmath16 may cross some or all of the crosscaps @xmath111 and still be disjoint from @xmath19 ; there may be several edges of @xmath16 crossing each crosscap , and an edge of @xmath16 may cross several crosscaps . in this context
vertex - splitting in @xmath19 preserving @xmath0-face - colorability allows us to assume , if we wish , that @xmath19 is a union of vertex - disjoint cycles or free loops .
if @xmath16 is connected , then when we cut along @xmath19 and cap , we get an embedding @xmath121 of @xmath16 in an orientable surface , one of the connected components of @xmath113 .
we do not disturb the order of the edges around any of the vertices of @xmath16 , so @xmath121 has the same set of facial boundary component walks as @xmath120 , although the faces may not be open disks .
now we combine operations [ inscc ] and [ oriset2 ] to prove an easy case of theorem [ mainose ] , which applies to graphs with arbitrary vertex degrees , not just cubic graphs .
[ old4.2 ] let @xmath25 be a closed @xmath0-cell embedding of a @xmath0-connected graph @xmath16 in the projective plane .
suppose @xmath25 contains a noncontractible elementary face ring @xmath122 which has odd length @xmath123 .
then @xmath16 has a closed @xmath0-cell embedding in some orientable surface .
represent the projective plane in the standard way as a disk with antipodal boundary points identified ; the boundary represents a crosscap added to the sphere , which we call the _ outer crosscap _ @xmath124 . by homotopic shifting we may ensure that the closure of exactly one face of @xmath86 intersects @xmath124 , and @xmath124 cuts that face into exactly two nonempty pieces . see figure [ fig - ofr ] .
insert crosscaps @xmath106 , @xmath107 , @xmath108 , @xmath125 along @xmath86 as in operation [ inscc](c ) , to obtain a new embedding @xmath120 of @xmath16 in @xmath126 .
all faces of @xmath86 turn into a new face @xmath41 in @xmath120 whose boundary has two components @xmath127 and @xmath128 .
all other facial walks in @xmath25 remain unchanged in @xmath120 . because faces in @xmath86 are disjoint except that consecutive faces intersect in a single component , each of @xmath127 and @xmath128 is a cycle .
thus , all facial boundary component walks of @xmath129 are cycles . by following @xmath86
, we may construct a simple closed curve @xmath19 that passes through each face of @xmath86 and each inserted crosscap exactly once , and is disjoint from @xmath16 .
@xmath19 crosses each of the @xmath130 crosscaps @xmath124 , @xmath106 , @xmath108 , @xmath125 .
if we consider @xmath19 as an embedded pseudograph with one vertex ( any point of @xmath131 ) and one loop , then collapsing all crosscaps leaves @xmath19 as a simple closed curve in the sphere , which is a @xmath0-face - colorable embedding of @xmath19 in an orientable surface
. therefore , by operation [ oriset2 ] , cutting along @xmath19 and capping yields an orientable surface , in which @xmath16 is embedded with all facial boundary components cycles . removing any faces that are not open disks and capping any resulting holes with disks ,
we obtain a closed @xmath0-cell embedding of @xmath16 in an orientable surface , as required .
theorem [ old4.2 ] does not work with elementary face rings of even length because then we get a face with a single boundary component that self - intersects , instead of two components @xmath127 and @xmath128 . in terms of our original strategy , here @xmath70 is the sphere , @xmath132 , @xmath68 is the projective plane , @xmath133 , @xmath72 is @xmath126 , and @xmath74 and @xmath75 are the orientable surfaces in the last two sentences of the proof .
the proof of theorem [ old4.2 ] uses only a very simple version of our strategy , and there is an alternative way to obtain the final embedding in this proof , namely the surgery of fiedler et al .
* lemma a ) .
however , we have examples of noncubic @xmath0-connected projective - planar graphs where we can find orientable closed @xmath0-cell embeddings using more complicated applications of operations [ inscc ] and [ oriset2 ] .
also , our approach continues the development of a coherent set of tools for constructing closed @xmath0-cell embeddings , begun in papers such as @xcite .
in this section we describe reductions that allow us to restrict our attention to a smaller class of projective - planar cubic graphs .
most of these reductions are standard , but the reduction we give for @xmath1-cycles is new .
we state our results in a general setting and then apply them to projective - planar cubic graphs .
we also prove a technical result which will be used to deal with certain ` planar @xmath1-edge - cuts ' .
let @xmath134 be the class of @xmath0-connected cubic graphs , and let @xmath135 be a subclass of @xmath134 closed under taking minors in @xmath134 ( i.e. , if @xmath136 and @xmath137 is a minor of @xmath138 , then @xmath139 also ) . in constructing orientable closed @xmath0-cell embeddings for graphs in @xmath135 , standard reductions used for cycle double covers ( see @xcite ) can be applied to deal with any @xmath0-edge - cut , or nontrivial @xmath6-edge - cut ( a @xmath6-edge - cut that is not just the edges incident with a single vertex ) . since we can exclude @xmath0-edge - cuts
, we can also exclude multiple edges . a @xmath0-connected cubic graph with no @xmath0-edge - cuts or nontrivial @xmath6-edge - cuts
is said to be _
cyclically-@xmath1-edge - connected _ , so we may state the following .
[ minproperties ] given @xmath135 as above , an element @xmath16 of @xmath135 having no orientable closed @xmath0-cell embedding ( i.e. , no orientable cycle double cover ) and , subject to that , fewest edges is simple and cyclically @xmath1-edge - connected .
since @xmath16 has no nontrivial @xmath6-edge - cuts , and since @xmath140 has an orientable embedding , it follows that @xmath16 as in lemma [ minproperties ] has no triangles ( @xmath6-cycles ) . for the cycle
double cover conjecture there are also reductions to exclude @xmath1-cycles and in fact cycles of length up to @xmath141 , due to goddyn @xcite ( see also @xcite ) and huck @xcite
. however , these reductions do not work if we add the condition of orientability .
here we show that there is at least a @xmath1-cycle reduction for the orientable cycle double cover conjecture .
[ no4cycle ] given @xmath135 as above , an element @xmath16 of @xmath135 having no orientable closed @xmath0-cell embedding ( i.e. , no orientable cycle double cover ) and , subject to that , fewest edges has no @xmath1-cycle . by lemma [ minproperties ]
@xmath16 is simple , cyclically-@xmath1-edge - connected , and has no triangles .
since @xmath142 has a toroidal closed @xmath0-cell embedding ( in which there are three faces , all hamilton cycles ) , we know that @xmath143 .
assume that @xmath16 contains a @xmath1-cycle @xmath144 .
since @xmath16 has no triangles , each @xmath145 has a neighbour @xmath146 .
since @xmath143 , cyclic-@xmath1-edge - connectivity implies that @xmath147 , @xmath148 , @xmath149 and @xmath150 are distinct , and moreover that @xmath151 is @xmath0-connected .
hence @xmath152 , so since @xmath153 has fewer edges than @xmath16 it has an orientable closed @xmath0-cell embedding @xmath120 .
suppose the faces of @xmath120 containing @xmath154 are @xmath59 and @xmath60 , where @xmath155 traversed clockwise uses directed edge @xmath156 , and @xmath157 uses @xmath154 .
suppose the faces containing @xmath158 are @xmath127 and @xmath128 , where @xmath159 uses directed edge @xmath158 when traversed clockwise , and @xmath160 uses @xmath161 . in @xmath120 subdivide @xmath154 to obtain the path @xmath162 , and @xmath158 to obtain the path @xmath163 .
if @xmath164 we may add the edges @xmath165 and @xmath166 inside @xmath59 to obtain an orientable closed @xmath0-cell embedding of @xmath16 .
a similar argument applies if @xmath167 .
thus , @xmath168 and @xmath169 .
if in addition @xmath170 then we add a handle from @xmath59 to @xmath127 and add edges @xmath165 , @xmath166 along the handle , to obtain an orientable embedding of @xmath16 . in the new embedding the faces @xmath59 and @xmath127 are replaced by @xmath171 and @xmath172 , where @xmath173 and @xmath174 .
now @xmath175 is a cycle because @xmath168 , and @xmath176 is a cycle because @xmath177 , so this is an orientable closed @xmath0-cell embedding of @xmath16 .
therefore , @xmath178 , and similarly @xmath179 .
since @xmath180 and @xmath179 we may add the edge @xmath165 inside @xmath59 and add the edge @xmath166 inside @xmath60 , to obtain an orientable closed @xmath0-cell embedding of @xmath16 .
since @xmath16 has an orientable closed @xmath0-cell embedding in all cases , our assumption was false , and @xmath16 has no @xmath1-cycle , as required .
we may apply our results when @xmath135 is the class of projective - planar cubic graphs .
[ minproperties2 ] let @xmath16 be a @xmath0-connected projective - planar cubic graph that has fewest edges subject to having no orientable closed @xmath0-cell embedding .
then @xmath16 is simple , cyclically-@xmath1-edge - connected , and has no @xmath6- or @xmath1-cycles .
now we introduce the technical result for dealing with certain ` planar @xmath1-edge - cuts . '
it is easier to prove it first in a dual form , for near - triangulations . a _
near - triangulation _ is a graph embedded in the plane so that every face is a triangle , except possibly for the outer face , which is a cycle .
a _ separating cycle _ in an embedded graph is a cycle whose removal disconnects the surface into two components , each of which contains at least one vertex of the graph .
if @xmath16 is a connected graph embedded in the plane , @xmath181 represents its outer walk .
an _ interior _
vertex of @xmath16 is a vertex not on @xmath181 .
an _ interior path _ in @xmath16 is a path none of whose internal vertices lie on @xmath181 ( although one or both ends may lie on @xmath181 ) . if @xmath182 is a cycle of @xmath16 , then @xmath183 or @xmath184 represents the embedded subgraph of @xmath16 on and inside @xmath36 .
if @xmath16 is understood we just write @xmath185 or @xmath186 .
@xmath187 represents the set of neighbors of vertex @xmath20 in @xmath16 , and @xmath188 $ ] represents @xmath189 .
if @xmath19 is a subgraph of @xmath16 , then a _ chord _ of @xmath19 is an edge that is not in @xmath19 but whose two ends are in @xmath19 . if @xmath190 is a cutvertex of @xmath19 , then a chord of @xmath19 is _ @xmath190-jumping _ if its ends lie in different components of @xmath191 . [ a0 ] in a near - triangulation ,
a minimal cutset separating two given nonadjacent vertices induces either a chordless separating cycle , or a chordless interior path with ends on the outer cycle .
[ c0 ] suppose @xmath16 is a near - triangulation with outer cycle @xmath192 and @xmath190 , @xmath193 are vertices of @xmath192 with @xmath194
. then either @xmath16 has an interior @xmath195-path , or @xmath196 has an @xmath190-jumping chord .
[ old5.4 ] let @xmath16 be a simple graph ( no loops or multiple edges ) embedded in the plane so that all faces are triangles except the outer face , which is a @xmath1-cycle @xmath197 in that clockwise order ; there is at least one interior vertex ; and there are no separating triangles .
* then there is a chordless interior @xmath198-path in @xmath16 .
* moreover , if all chordless interior @xmath199-paths in @xmath16 have length of the same parity ( all even , or all odd ) , then @xmath16 has an interior vertex of degree @xmath1 .
nontrivial examples as in ( ii ) above do occur .
at left in figure [ fig-4c ] is an example where all chordless interior @xmath198-paths have even length ; it has several interior vertices of degree @xmath1 . for ( i ) ,
@xmath200 because there are no separating triangles , and hence by observation [ c0 ] there is an interior @xmath201-path .
a shortest such path is chordless .
we prove ( ii ) by contradiction .
assume it does not hold , and @xmath16 is a counterexample with fewest vertices .
so , all interior @xmath199-paths have length of the same parity , but there is no interior vertex of degree @xmath1 . if @xmath202 , @xmath16 is a wheel with central vertex of degree @xmath1 , so we must have @xmath203 . since @xmath16 is a counterexample and there are no separating triangles , every interior vertex has degree at least @xmath204 . since there are no separating triangles , @xmath199 , @xmath205 .
[ b0 ] if @xmath206 is a separating @xmath1-cycle then there are chordless interior @xmath207-paths in @xmath185 whose lengths have different parities . if not , then by minimality of @xmath16 there is an interior vertex of @xmath185 of degree @xmath1 in @xmath185 . but
this is also an interior vertex of @xmath16 of degree @xmath1 in @xmath16 , a contradiction .
[ b1 ] there is no interior vertex adjacent to both @xmath208 and @xmath209 , or both @xmath210 and @xmath211 .
suppose there is an interior vertex @xmath20 with @xmath212 .
since @xmath203 , at least one of the subgraphs @xmath213 or @xmath214 has an interior vertex .
but this subgraph contradicts claim [ b0 ] .
if there is an interior vertex @xmath20 adjacent to both @xmath210 and @xmath211 then at least one of @xmath215 and @xmath216 has an interior vertex ; without loss of generality assume it is @xmath153 .
there is a chordless interior @xmath217-path @xmath218 in @xmath219 , by ( i ) if @xmath219 has an interior vertex , and because @xmath200 so that @xmath220 if @xmath219 has no interior vertex . for every chordless interior @xmath221-path @xmath222 in @xmath153
, @xmath223 is a chordless interior @xmath198-path in @xmath16 , so all such paths @xmath222 have length of the same parity , contradicting claim [ b0 ] .
let the neighbors of @xmath208 in anticlockwise order be @xmath224 . by claim
[ b1 ] , @xmath225 .
as all interior faces are triangles and there are no separating triangles , @xmath226 is a chordless path in @xmath16 . by claim [ b1 ] ,
no vertex of @xmath227 is adjacent to @xmath209 .
[ b2 ] every @xmath1-cycle containing @xmath208 is @xmath228 , or has the form @xmath229 , or has the form @xmath230 where @xmath190 is an interior vertex of @xmath231 .
every @xmath1-cycle containing @xmath208 has the form @xmath232 with @xmath233 .
if @xmath234 and @xmath235 then by claim [ b1 ] , @xmath36 is @xmath228 .
so , without loss of generality , we may assume that @xmath236 .
then , also by claim [ b1 ] , @xmath237 .
if @xmath190 is on @xmath238 then since @xmath238 is chordless , @xmath36 must be @xmath239 , as specified .
so we may assume that @xmath190 is an interior vertex of @xmath231 .
if @xmath51 then @xmath36 is as specified , so suppose @xmath240 .
let @xmath241 and @xmath242 . since @xmath238 is chordless , observation [ c0 ] implies that @xmath219 has an interior @xmath243-path .
let @xmath218 be a shortest , hence chordless , such path .
for any chordless interior @xmath244-path @xmath222 in @xmath153 , @xmath245 is a chordless interior @xmath198-path in @xmath16 , so all such paths @xmath222 have length of the same parity .
since @xmath153 has an interior vertex , @xmath246 , this contradicts claim [ b0 ] .
the following are immediate consequences of claim [ b2 ] .
[ b2a ] there is no interior vertex @xmath190 not adjacent to @xmath208 but adjacent to @xmath247 and @xmath248 with @xmath249 .
[ b2b ] there is no separating @xmath1-cycle in @xmath16 of the form @xmath250 or @xmath251 , or ( by symmetry ) @xmath252 or @xmath253 .
since @xmath147 has degree at least @xmath204 , there exist @xmath254 so that the neighbors of @xmath147 in clockwise order are @xmath255
. since @xmath238 is chordless , neither @xmath256 nor @xmath257 is on @xmath238 , and by claim [ b1 ] , neither @xmath256 nor @xmath257 is @xmath209 .
there are triangular faces @xmath258 and @xmath259 .
since there are no separating triangles , @xmath260 . by claim [ b2b ] , @xmath261 .
see the right side of figure [ fig-4c ] .
let @xmath262 , and let @xmath263 . since @xmath238 is chordless , since @xmath264 , and
since @xmath154 and @xmath265 are edges of @xmath16 but not @xmath19 , @xmath266 has no @xmath256-jumping chord in @xmath19 .
[ w1v3path ] there is an interior @xmath267-path in @xmath19 avoiding all neighbors ( in @xmath16 ) of @xmath208 , @xmath147 , @xmath148 and @xmath257 , except @xmath256 itself .
let @xmath268 \cup n_h[u_2 ] \cup n_h[w_2 ] ) - \{w_1\}$ ] . since @xmath269 , we know that @xmath270 .
it suffices to show that there is a @xmath267-path in @xmath271 .
let @xmath272 - \{w_1\})$ ] , and @xmath273 - \{w_1\ } ) \cup n_h[u_2 ] \cup \{u_4 , u_5 , \ldots , u_{k+1}\}$ ] . then @xmath274 .
suppose @xmath275 . since @xmath266 has no @xmath256-jumping chord in @xmath19 , @xmath276 and hence @xmath277 .
but then either @xmath278 or @xmath279 is a separating triangle in @xmath16 , a contradiction .
hence @xmath280 .
assume there is no @xmath281-path in @xmath271 .
then there is a minimal cutset contained in @xmath29 separating @xmath256 and @xmath209 in @xmath19 , which by observation [ a0 ] induces a chordless path @xmath56 starting on @xmath282 and ending on @xmath283 .
since @xmath266 has no @xmath256-jumping chord in @xmath19 , there are no vertices of @xmath284 on @xmath282 , and no vertices of @xmath285 on @xmath286 .
hence , the first vertex of @xmath56 belongs to @xmath285 and the last to @xmath284 .
let @xmath287 denote the last vertex of @xmath56 that belongs to @xmath285 , and @xmath288 its immediate successor , which must belong to @xmath284 . since @xmath266 has no @xmath256-jumping chord in @xmath19 , we can not have both @xmath289 .
suppose @xmath290 , so @xmath291 or @xmath147 .
then @xmath292 . if @xmath293 then @xmath294 , a contradiction .
if @xmath291 and @xmath295 then @xmath296 contradicts claim [ b2a ] .
thus , @xmath291 and @xmath297 .
since @xmath266 has no @xmath256-jumping chord in @xmath19 , by observation [ c0 ] there is an interior @xmath298-path in @xmath299 .
let @xmath218 be a shortest , hence chordless , such path .
since @xmath300 has an interior vertex @xmath256 , by claim [ b0 ] it has interior @xmath301-paths @xmath302 and @xmath303 whose lengths have different parities .
but then @xmath304 and @xmath305 are chordless interior @xmath198-paths whose lengths have different parities , a contradiction
. therefore , @xmath306 .
we consider three possibilities for @xmath288 .
cases ( 2 ) and ( 3 ) are shown at right in figure [ fig-4c ] .
\(1 ) suppose that @xmath307 , so that @xmath288 is a vertex of @xmath308 .
since @xmath309 , we either get a separating triangle in @xmath16 if @xmath310 or @xmath148 , or violate claim [ b2a ] if @xmath311 , @xmath312 .
\(2 ) suppose that @xmath313 .
let @xmath314 and let @xmath315 . if @xmath287 is adjacent to a vertex @xmath316 of @xmath283 we get a separating triangle if @xmath317 or @xmath148 , or violate claim [ b2a ] otherwise .
also , @xmath318 , and @xmath266 has no @xmath256-jumping chord in @xmath19 .
therefore , @xmath319 has no @xmath288-jumping chord in @xmath320 .
if there is an interior @xmath321-path in @xmath320 that avoids @xmath322 $ ] , then we may take @xmath218 to be a shortest , hence chordless , such path .
then for any chordless interior @xmath301-path @xmath222 in @xmath323 , @xmath324 is a chordless interior @xmath201-path in @xmath16 , so all such paths @xmath222 have length of the same parity .
since @xmath153 has an interior vertex , @xmath256 , this contradicts claim [ b0 ] . therefore , @xmath325 ) - \{x_2\}$ ] separates @xmath288 and @xmath209 in @xmath320 .
let @xmath326 $ ] and @xmath327 , so that @xmath328 . because @xmath319 has no @xmath288-jumping chord in @xmath320 , @xmath329 . applying observation [ a0 ] to a minimal cutset contained in @xmath330 separating @xmath288 and @xmath209 , which induces a chordless path @xmath331 which starts on a vertex of @xmath332 and ends on a vertex of @xmath333 , we see that @xmath331 has an edge @xmath334 with @xmath335 , @xmath336 .
since @xmath319 has no @xmath288-jumping chord in @xmath320 , we can not have both @xmath337 , so @xmath338 .
but then if @xmath339 or @xmath148 we have a separating triangle @xmath340 , and otherwise @xmath341 violates claim [ b2a ] .
\(3 ) suppose that @xmath342 .
after modifying @xmath319 to be @xmath343 , @xmath153 to be @xmath344 , and @xmath333 to be @xmath345 , the proof is almost identical to ( 2 ) above .
so our assumption was incorrect , and there is a @xmath267-path in @xmath346 .
let @xmath347 be a @xmath281-path as in claim [ w1v3path ] that is as short as possible ; then @xmath347 is chordless . as @xmath347 contains no neighbor of @xmath208 , @xmath147 , @xmath148 or @xmath257 except @xmath256 , the paths @xmath348 and @xmath349 are chordless interior @xmath198-paths whose lengths have different parities , giving the final contradiction that proves ( ii ) .
[ z1 ] let @xmath16 be a cyclically-@xmath1-edge - connected cubic graph embedded in a surface , and let @xmath36 be a cycle of @xmath16 with a fixed orientation bounding a closed disk @xmath350 in the surface .
suppose that precisely four edges touch @xmath36 from outside of @xmath350 , at vertices @xmath147 ,
@xmath148 , @xmath149 , @xmath150 in that order around @xmath36 , and let @xmath351 ( taking @xmath352 ) .
if all @xmath353-face chains through @xmath350 have length of the same parity , then there is a @xmath1-cycle face contained in @xmath350 .
let @xmath19 be the intersection graph of the sets @xmath354 and the closures of all faces inside @xmath350 , and for each @xmath355 let @xmath356 denote the face outside @xmath350 intersecting @xmath350 along @xmath357 .
essentially @xmath19 is a subgraph of the dual of @xmath16 , but we replace each face @xmath356 by @xmath358 to make sure that there are four distinct vertices representing @xmath127 , @xmath128 , @xmath359 and @xmath360 , even if some of these faces are actually the same . since @xmath16 is cubic , @xmath19 is a near - triangulation , with outer @xmath1-cycle @xmath361 .
since @xmath16 is cyclically-@xmath1-edge - connected , @xmath19 is simple and has no separating triangles
. apply proposition [ old5.4 ] ( ii ) to @xmath19 .
we are now ready to prove our main theorem .
the main step is a structural result , theorem [ structure ] .
[ ppfacering ] suppose @xmath25 is a projective - planar embedding of a @xmath6-connected cubic graph , with @xmath362
. then for any noncontractible simple closed curve @xmath91 with @xmath363 , the faces traversed by @xmath91 , in their order along @xmath91 , form a noncontractible elementary face ring .
label the faces along @xmath91 as @xmath59 , @xmath60 , @xmath108 , @xmath364 . interpreting subscripts modulo @xmath83 , clearly @xmath48 and @xmath84 touch for each @xmath355 .
if @xmath48 touches @xmath50 for some @xmath365 , @xmath355 or @xmath65 , then we can find a noncontractible closed curve intersecting @xmath16 in fewer points than @xmath91 , a contradiction .
thus , we have a noncontractible face ring .
if @xmath82 then the face ring is elementary by observation [ 3conn3rep ] , so assume that @xmath366 .
if @xmath80 has three or more components , then we can find a contractible simple closed curve lying in @xmath367 cutting @xmath16 at exactly two vertices that form a cutset in @xmath16 , contradicting @xmath6-connectivity .
thus , the face ring is elementary .
[ structure ] if @xmath16 is a cyclically-@xmath1-edge - connected cubic graph with a @xmath0-representative embedding @xmath25 in the projective plane , then the embedding has a @xmath1-cycle face or a noncontractible elementary face ring of odd length or both .
assume for a contradiction that @xmath16 has neither a @xmath1-cycle face nor a noncontractible elementary face ring of odd length .
by lemma [ ppfacering ] , @xmath25 has a noncontractible elementary face ring of length @xmath368 , so @xmath369 for some @xmath370 .
let this face ring be @xmath371 .
subscripts @xmath355 for @xmath48 are to be interpreted modulo @xmath372 . since @xmath16 is cubic and @xmath6-connected , if @xmath373 then each component of @xmath80 is a single edge .
similarly , if @xmath374 then each @xmath375 has one component which is a single edge .
therefore , @xmath376 is a closed mbius strip bounded by a cycle @xmath377 .
@xmath377 contains distinct vertices @xmath378 in that order such that @xmath379 is the edge @xmath380 .
let @xmath381 ( subscripts modulo @xmath382 ) , so that the boundary of @xmath48 is @xmath383 .
subscripts @xmath355 for @xmath145 , @xmath384 and related objects are to be interpreted modulo @xmath382 .
removing the interior of @xmath385 from the projective plane leaves a closed disk @xmath350 , which is the union of the closures of the faces not in @xmath86 , and which contains all vertices of @xmath16 .
we assume that @xmath377 goes around @xmath350 clockwise , and all cycles contained in @xmath350 will also be oriented clockwise .
suppose @xmath386 , so that @xmath387 . every @xmath388-face chain through @xmath350 extends to an noncontractible elementary face ring using observation [ 3conn ] , and is therefore of even length .
but then by corollary [ z1 ] there is a @xmath1-cycle face contained in @xmath350 , contradicting the fact that @xmath16 has no @xmath1-cycles .
thus , @xmath389 , so that @xmath374 . by observation [ 3conn3rep ] , every face ring is elementary .
thus , every noncontractible face ring has even length . for each @xmath355 , @xmath390 , let @xmath391 be the set of faces in @xmath350 whose closures intersect @xmath384 , and @xmath392 . if @xmath393 , let @xmath394 denote the distance between @xmath355 and @xmath395 in the cyclic sequence @xmath396 .
we proceed by proving a sequence of claims .
note that we implicitly use the fact that @xmath397 .
in particular , all face rings we contruct are valid if @xmath397 .
[ d1 ] since @xmath16 is cubic and @xmath6-connected and since @xmath398 , for any two faces @xmath31 and @xmath41 , @xmath399 is either empty or a single edge .
[ d2 ] if @xmath400 then @xmath401 .
suppose not .
without loss of generality we may assume that @xmath402 and @xmath403 .
let @xmath404 .
we may construct a noncontractible simple closed curve @xmath91 passing through @xmath405 and intersecting @xmath16 at @xmath406 points , a contradiction .
[ d3 ] for every @xmath407 , @xmath408 has exactly one component , which is a subpath of @xmath377 with at least one edge .
any component of @xmath408 can not be a single vertex because @xmath16 has no vertices of degree @xmath1 or more , so it is a subpath of @xmath377 with at least one edge .
suppose @xmath408 has two distinct components , @xmath103 and @xmath409 .
suppose @xmath103 intersects @xmath384 and @xmath409 intersects @xmath410 where we choose @xmath355 and @xmath395 so that @xmath394 is as small as possible . by claim [ d2 ] ,
if @xmath412 then @xmath16 has a @xmath0-edge - cut , and if @xmath413 then @xmath16 has a nontrivial @xmath6-edge - cut , contradicting the fact that @xmath16 is cyclically-@xmath1-edge - connected .
suppose that @xmath414 .
without loss of generality assume that @xmath415 and @xmath416 . using claim [ d1 ] ,
suppose that @xmath417 is the edge @xmath418 where @xmath419 occur on that order along @xmath420 , and @xmath421 is the edge @xmath422 where @xmath423 occur in that order along @xmath424 . since @xmath16 is cubic , @xmath425 , @xmath426 , @xmath427 and @xmath428 .
let @xmath429 be the closed disk bounded by the cycle @xmath430 .
let @xmath431 and @xmath432 .
then for every @xmath433-face chain @xmath434 of length @xmath2 through @xmath429 , we have a noncontractible face ring @xmath435 , which must have even length , so that @xmath2 is always even .
therefore , by corollary [ z1 ] there is a @xmath1-cycle face contained in @xmath429 , contradicting the fact that @xmath16 has no @xmath1-cycles . by claim [ d3 ] ,
the elements of @xmath436 can be cyclically ordered along @xmath377 according to their intersection with @xmath377 . within each @xmath391
we may linear order the elements of @xmath391 along @xmath384 as @xmath437 , where @xmath438 contains @xmath145 and @xmath439 contains @xmath440 . by claim [ d1 ] , @xmath441 is the length of @xmath384 .
possibly @xmath442 .
note that @xmath443 .
the following is immediate . [ d3a ] if @xmath444 and @xmath445 then @xmath446 and @xmath447 do not touch .
[ d4 ] we do not have @xmath448 for any @xmath355 . if @xmath448 then @xmath449 , violating claim [ d2 ] .
[ d5a ] since @xmath16 has no nontrivial @xmath6-edge - cut , if @xmath450 then @xmath446 and @xmath451 do not touch .
[ d5b ] if @xmath452 and @xmath453 then @xmath446 and @xmath454 do not touch . without loss of generality
assume that @xmath402 .
suppose that @xmath455 , @xmath453 , and @xmath456 and @xmath457 touch .
let @xmath83 be the largest @xmath83 so that @xmath458 ; by claim [ d4 ] , @xmath459 or @xmath1 .
let @xmath460 and @xmath461 .
then @xmath341 and @xmath462 are internal vertices of @xmath420 and @xmath463 , respectively .
let @xmath464 be the first vertex of @xmath465 encountered when travelling clockwise along @xmath466 from @xmath341 .
let @xmath429 be the closed disk bounded by the cycle @xmath467 .
let @xmath431 and @xmath468 .
then for every @xmath433-face chain @xmath469 of length @xmath470 through @xmath429 , by claims [ d3a ] and [ d5a ] we have a noncontractible face ring @xmath471 , which must have even length , so that @xmath470 always has the same parity . therefore , by corollary [ z1 ] there is a @xmath1-cycle face contained in @xmath429 , contradicting the fact that @xmath16 has no @xmath1-cycles .
[ d5c ] if @xmath472 then no face in @xmath391 touches a face in @xmath473 .
this is similar to the proof of claim [ d2 ] .
[ d6 ] if @xmath474 and @xmath475 then @xmath441 is odd .
equivalently , if @xmath441 is even then @xmath476 or @xmath477 . without loss of generality assume that @xmath478 .
if @xmath479 and @xmath480 then , by claims [ d3a ] and [ d5a ] , @xmath481 is a noncontractible face ring , which must have even length , so that @xmath482 must be odd .
[ d7 ] if @xmath474 and @xmath483 , or @xmath476 and @xmath484 , then @xmath441 is even . without loss of generality
assume that @xmath485 , where @xmath479 and @xmath486 . by claim [ d4 ] ,
thus , and by claims [ d3a ] and [ d5a ] , @xmath488 is a noncontractible face ring , which must have even length , so that @xmath482 must be even .
[ d8 ] if @xmath489 then @xmath441 is odd and @xmath490 . without loss of generality assume that @xmath491 , and @xmath492 . by claim [ d4 ] , @xmath479 and @xmath493 .
thus , and by claims [ d3a ] and [ d5a ] , @xmath494 is a noncontractible face ring , which must have even length , so that @xmath495 must be odd . by claim [ d4 ] , @xmath496 .
[ d9 ] if @xmath441 is even then either @xmath497 and @xmath498 is even , or @xmath499 and @xmath500 is even . without loss of generality
assume that @xmath491 . by claim [ d6 ] , @xmath501 or @xmath502 :
again without loss of generality assume that @xmath501 . by claim [ d4 ] , @xmath479 , and by claim [ d8 ] ,
if @xmath493 , then @xmath504 is odd by claim [ d6 ] , but then @xmath505 is a noncontractible face ring ( by claims [ d3a ] , [ d5a ] and [ d5b ] ) of odd length , a contradiction .
therefore @xmath506 , and from claim [ d7 ] , @xmath504 is even .
[ d10 ] if @xmath507 then either @xmath498 and @xmath508 are even and @xmath509 , or @xmath498 is an odd number at least @xmath6 and @xmath510
. without loss of generality assume that @xmath511 .
if @xmath482 is even then by claim [ d9 ] we know that @xmath495 is even and @xmath502 . if @xmath482 is odd then by claim [ d7 ] @xmath512 .
[ d11 ] we can not have @xmath441 , @xmath498 , @xmath508 , @xmath513 all greater than @xmath8 . without loss of generality assume that @xmath402 .
suppose that @xmath514 . by claim [ d6 ] , @xmath482 and @xmath495
are odd . by claims
[ d3a ] , [ d5a ] and [ d5b ] , @xmath515 is a noncontractible face ring of odd length , a contradiction .
suppose now that no @xmath441 is even . by claim [ d11 ] , some @xmath441 is equal to @xmath8 , say @xmath516 .
since no @xmath441 is even , claim [ d10 ] implies that @xmath482 , @xmath504 , @xmath517 , @xmath108 , @xmath518 are all odd numbers at least @xmath6 , while @xmath519 . in particular , @xmath520 , which means that @xmath59 is a @xmath1-cycle face , a contradiction .
therefore , some @xmath441 is even , and by claim [ d9 ] either @xmath521 or @xmath483 . without loss of generality
we may assume that @xmath495 is even and @xmath501 .
by claim [ d4 ] , @xmath479 , and by claim [ d10 ] , @xmath504 is even and @xmath522 , so that , by claim [ d4 ] again , @xmath523 . by claims
[ d3a ] , [ d5a ] and [ d5b ] , @xmath524 is a noncontractible face ring of odd length , a contradiction , if @xmath525 . therefore , @xmath526 . to satisfy claims [ d10 ] and [ d11 ] , and since some @xmath441 is even , we must have the following situation , or one rotationally equivalent to it : @xmath501 , @xmath495 even , @xmath504 even , @xmath522 , @xmath517 even , @xmath527 even , @xmath528 , and @xmath529 odd and at least @xmath6 . by lemma [ ppfacering ] , @xmath530 is a noncontractible face ring .
the union of the closures of these faces is a closed mbius strip with boundary cycle @xmath531 , which we may divide into @xmath532 subpaths @xmath533 in the same way that @xmath377 is divided into subpaths @xmath384 .
we see that @xmath534 and we may orient @xmath531 and label its subpaths so that @xmath535 , @xmath536 , and the orientation of @xmath531 agrees with that of @xmath377 on these two subpaths . if @xmath537 is the length of @xmath533 we see that @xmath538 is even , @xmath539 is unknown , @xmath540 is odd , @xmath541 is even , @xmath542 , @xmath543 , @xmath544 is odd , and @xmath545 . by claim [ d9 ] , @xmath546
so that @xmath547 . by symmetry ,
@xmath548 also . in the same way
, @xmath549 is a noncontractible face ring , and we get a closed mbius strip whose boundary cycle is divided into @xmath532 subpaths of lengths @xmath550 , @xmath551 , where we see that @xmath552 , @xmath553 , @xmath554 , @xmath555 is odd , @xmath556 is unknown , @xmath557 is odd , @xmath558 , and @xmath559 . by claim [ d9 ] ,
@xmath560 so that @xmath561 .
now all even numbers @xmath441 ( @xmath495 , @xmath504 , @xmath517 , @xmath527 ) have been shown to equal @xmath0 . by the same reasoning ,
all even numbers @xmath537 must be @xmath0 .
in particular , @xmath562 , so that @xmath563 .
since @xmath563 and @xmath564 we may write @xmath565 . for each @xmath566 , let @xmath567 denote the neighbor of @xmath37 to which @xmath37 is joined by an edge that is not an edge of @xmath377 . by claim [ d3 ] ,
no such @xmath567 is a vertex of @xmath377 .
if @xmath568 does not touch @xmath569 then @xmath570 is a noncontractible face ring of length @xmath204 , which is odd , a contradiction .
so @xmath568 touches @xmath569 .
write @xmath571 where @xmath288 precedes @xmath462 in the clockwise order around @xmath572 . by claim [ d3 ] , @xmath288 and @xmath462 are not vertices of @xmath377 .
if @xmath573 , @xmath574 and @xmath462 are not all equal , then @xmath575 is a nontrivial @xmath6-edge - cut , a contradiction .
therefore , @xmath576 is a vertex shared by @xmath569 , @xmath577 , and @xmath568 .
similarly , there is a vertex @xmath578 shared by @xmath569 , @xmath579 , and @xmath580 , with neighbor @xmath581 , where neither @xmath287 nor @xmath341 is on @xmath377 . since @xmath569 is not a @xmath6- or @xmath1-cycle face , @xmath582 and @xmath583 , so that @xmath584 and @xmath585 . since @xmath586 is not a @xmath1-cycle face , @xmath587 .
since the neighbors of @xmath341 are @xmath287 , @xmath588 and @xmath589 , @xmath341 is adjacent to neither @xmath590 nor @xmath591 ; similarly , @xmath462 is adjacent to neither @xmath590 nor @xmath591 .
thus , @xmath592 is an independent set and @xmath593 do not belong to this set .
now there is a contractible simple closed curve intersecting the embedding at precisely four points , one interior point of each of the four edges @xmath594 , @xmath595 , @xmath596 , @xmath597 , in that order .
this curve bounds an open disk @xmath77 containing @xmath287 , @xmath288 , @xmath598 and @xmath599 .
let @xmath19 be the subgraph of @xmath16 induced by @xmath600 .
since @xmath16 is cyclically-@xmath1-edge - connected , @xmath19 is either a single edge , or is a @xmath0-connected graph embedded in a closed disk @xmath429 bounded by a cycle @xmath601 through distinct vertices @xmath287 , @xmath288 , @xmath598 , @xmath599 in that order ( and possibly containing other vertices ) . if @xmath19 is a single edge , then either @xmath602 , @xmath603 , @xmath604 , and there is a noncontractible simple closed curve through @xmath605 , @xmath568 and @xmath580 that intersects @xmath16 at only @xmath6 points ; or @xmath606 , @xmath607 , @xmath604 , and there is a noncontractible simple closed curve through @xmath59 , @xmath569 and @xmath586 that intersects @xmath16 at only @xmath6 points . in either case
we have a contradiction , so @xmath19 is @xmath0-connected and we have @xmath429 and @xmath601 as described above .
let @xmath608 and @xmath609 .
then for every @xmath433-face chain @xmath610 of length @xmath470 through @xmath429 , @xmath611 is a noncontractible face ring , which must have even length , so that @xmath470 is always even .
therefore , by corollary [ z1 ] there is a @xmath1-cycle face contained in @xmath429 , a contradiction .
since every possibility leads to a contradiction , our original assumption must be wrong , and @xmath25 does have either a @xmath1-cycle face or a noncontractible elementary face ring of odd length .
now we prove the main result , which we restate .
suppose @xmath16 is a @xmath0-connected cubic graph with a projective - planar embedding @xmath25 but with no orientable closed @xmath0-cell embedding and that , subject to these conditions , @xmath16 has a minimum number of vertices . by corollary [ minproperties2 ]
, @xmath16 is simple , cyclically-@xmath1-edge - connected ( hence @xmath6-connected ) , and has no @xmath6- or @xmath1-cycles . if @xmath612 then @xmath16 can be embedded in the plane @xcite , and hence has a spherical closed @xmath0-cell embedding .
otherwise , by theorem [ structure ] , since @xmath16 has no @xmath1-cycle , @xmath25 must have a noncontractible elementary face ring of odd length
. then by theorem [ old4.2 ] , @xmath16 has an orientable closed @xmath0-cell embedding , as required .
we would like to strengthen theorem [ mainose ] to say that we have a @xmath2-face - colorable orientable closed @xmath0-cell embedding for some fixed small @xmath2
. then we could improve corollary [ mainocdc ] to say that we have an orientable @xmath2-cycle double cover ( the cycles can be colored using at most @xmath2 colors so that each edge is contained in two cycles of different colors ) .
however , this seems difficult .
the obstacle is our @xmath1-cycle reduction : we see no obvious way to avoid increasing the number of face colors needed when @xmath168 and @xmath169 in the proof of lemma [ no4cycle ] .
the authors thank john ratcliffe for useful discussions regarding the surgery operation discussed in lemma [ oriset1 ] .
brian alspach , luis goddyn and cun quan zhang , graphs with the circuit cover property , ( 1994 ) 131154 . j. r. fiedler , j. p. huneke , r. b. richter and n. robertson , computing the orientable genus of projective graphs , ( 1995 ) 297308 .
luis a. goddyn , `` cycle covers of graphs , '' ph.d .
dissertation , university of waterloo , canada , 1988 .
roland hggkvist and klas markstrm , cycle double covers and spanning minors i , ( 2006 ) 183206 .
andreas huck , reducible configurations for the cycle double cover conjecture , proc .
5th twente workshop on graphs and combinatorial optimization ( enschede , 1997 ) , _ discrete appl .
_ * 99 * ( 2000 ) 7190 .
f. jaeger , a survey of the cycle double cover conjecture , in `` cycles in graphs , '' ed .
b. r. alspach and c. d. godsil , _ ann . discrete math . _ * 27 * , north - holland , amsterdam , 1985 , 112 .
charles h. c. little and richard d. ringeisen , on the strong graph embedding conjecture , proc .
9th southeastern conf . on combinatorics , graph theory and computing ( boca raton , florida , 1978 ) , _
* 21 * ( 1978 ) 479487 .
neil robertson and robin thomas , on the orientable genus of graphs embedded in the klein bottle , ( 1991 ) 407419 . n. robertson and r. vitray , representativity of surface embeddings , in `` paths , flows , and vlsi - layout , '' ed .
b. korte et al . , _ algorithms and combinatorics _ * 9 * , springer - verlag , berlin , 1990 , 293328 .
xiaoya zha , the closed @xmath0-cell embeddings of @xmath0-connected doubly toroidal graphs , ( 1995 ) 259271 .
cun - quan zhang , `` integer flows and cycle covers of graphs , '' , marcel dekker , new york , 1997 . | in a _ closed @xmath0-cell embedding _ of a graph each face is homeomorphic to an open disk and is bounded by a cycle in the graph .
the orientable strong embedding conjecture says that every @xmath0-connected graph has a closed @xmath0-cell embedding in some orientable surface .
this implies both the cycle double cover conjecture and the strong embedding conjecture . in this paper
we prove that every @xmath0-connected projective - planar cubic graph has a closed @xmath0-cell embedding in some orientable surface .
the three main ingredients of the proof are ( 1 ) a surgical method to convert nonorientable embeddings into orientable embeddings ; ( 2 ) a reduction for @xmath1-cycles for orientable closed @xmath0-cell embeddings , or orientable cycle double covers , of cubic graphs ; and ( 3 ) a structural result for projective - planar embeddings of cubic graphs .
we deduce that every @xmath0-edge - connected projective - planar graph ( not necessarily cubic ) has an orientable cycle double cover .
epsf.sty amssym.def |
solar prominences ( a.k.a . filaments ) are relatively cool structures embedded in the million - degree corona at heights well above the chromosphere . above the solar limb , prominences appear as bright structures against the dark background , but when viewed as filaments on the solar disk they can be brighter or darker than their surroundings , depending on the bandpass used to observe them .
magnetic fields are thought to play an important role in supporting the prominence plasma against gravity , and in insulating it from the surrounding hot corona .
most quiescent prominences exhibit intricate filamentary structures that evolve continually due to plasma flows and heating / cooling processes . in some cases the threads appear to be mostly horizontal , while in other cases they are clearly radially oriented ( `` hedgerow '' prominences ) .
figure [ fig : proms ] shows several examples of prominences observed in h@xmath2 at the big bear solar observatory ( bbso ) and the dutch open telescope ( dot ) .
the examples in figs . [
fig : proms]a and [ fig : proms]b show mainly vertical threads , while the prominence in fig .
[ fig : proms]c shows horizontal threads .
off - limb observations in he ii 304 indicate that there are higher altitude parts that are optically thin in h@xmath2 and therefore not visible on the disk ( or at least have not been clearly identified in disk observations ) .
figure [ fig : proms]d shows a prominence above the solar limb as observed in he ii 304 with the secchi / euvi instrument on the stereo spacecraft .
the upper parts of the prominence consist of vertical threads with an intricate fine - scale structure .
movie sequences of quiescent and erupting prominences can be found at the stereo website .
prominence plasma is highly dynamic , exhibiting horizontal and vertical motions of order 1070 km s@xmath3 @xcite .
recent high - resolution observations of filaments on the solar disk indicate that they consist of a collection of very thin threads with widths of about 200 km , at the limit of resolution of present - day telescopes @xcite .
individual threads have lifetimes of only a few minutes , but the filament as a whole can live for many days .
it seems likely that these thin threads are aligned with the local magnetic field .
high - resolution images of prominences above the limb have been obtained with the solar optical telescope ( sot ) onboard hinode .
for example , @xcite observed horizontal threads in a prominence near an active region , and studied the oscillatory motions of these threads .
@xcite observed a hedgerow prominence consisting of many thin vertical threads , and they used multi - wavelength observations to estimate the amount of absorption and `` emissivity blocking '' in the prominence and surrounding cavity .
@xcite observed another hedgerow prominence and found that the prominence sheet is structured by both bright quasi - vertical threads and dark inclusions .
the bright structures are downflow streams with velocity of about 10 km s@xmath3 , and the dark inclusions are highly dynamic upflows with velocity of about 20 km s@xmath3 .
the downflow velocities are much less than the free - fall speed , indicating that the plasma is somehow being supported against gravity .
@xcite proposed that the dark plumes contain relatively hot plasma that is driven upward by buoyancy .
@xcite observed a persistent flow of h@xmath2 emitting plasma into a prominence from one side , leading to the formation of vertical threads .
they suggested that the vertical threads are stacks of plasma supported against gravity by the sagging of initially horizontal magnetic field lines .
direct measurements of the prominence magnetic field can be obtained using spectro - polarimetry . a comprehensive effort to measure prominence magnetic fields
was conducted in the 1970 s and early 1980 s using the facilities at pic du midi ( france ) and sacramento peak observatory ( usa ) .
this work showed that ( 1 ) the magnetic field in quiescent prominences has a strength of 315 g ; ( 2 ) the field is mostly _ horizontal _ and makes an angle of about @xmath4 with respect to the long axis of the prominence ; ( 3 ) the field strength increases slightly with height , indicating the presence of dipped field lines ; ( 4 ) most prominences have _ inverse _ polarity , i.e. , the component of magnetic field perpendicular to the prominence axis has a direction opposite to that of the potential field .
these earlier data likely included a variety of quiescent prominences , some with predominantly horizontal threads , others with more vertical threads .
in more recent work , @xcite reported full - stokes observations of a limb prominence in he i 5876 ( he @xmath5 ) , and derived magnetic field strengths of 3045 g , somewhat larger than the values reported in earlier studies . @xcite
published the first vector - field map of a prominence with a spatial resolution of a few arcseconds .
they found that the average magnetic field in this prominence is mostly horizontal with a strength of about 20 g and with the magnetic vector pointing @xmath6 to @xmath7 off the prominence axis , consistent with the earlier studies .
however , the map also shows clearly organized patches where the magnetic field is significantly stronger than average , up to 80 g .
it is unclear how these patches are related to the fine threads seen at higher spatial resolution .
recently , @xcite observed he i 10830 in a polar crown prominence above the limb , and found evidence for fields of about 30 g that are oriented only @xmath8 from the vertical direction .
how is the plasma in hedgerow prominences supported against gravity ?
many authors have suggested that quiescent prominences are embedded in large - scale _ flux ropes _ that lie horizontally above the polarity inversion line .
the prominence plasma is thought to be located near the _ dips _ of the helical field lines .
the magnetic field near the dips may be deformed by the weight of the prominence plasma .
others have suggested that the magnetic field in hedgerow prominences is vertical along the observed threads , and that the plasma is supported by mhd waves .
however , relatively high frequencies and wave amplitudes are required , and it is unclear why such waves would not lead to strong heating of the prominence plasma . @xcite and @xcite showed that the prominence plasma can be supported by the pressure of a coronal plasma lower down along an inclined field line ; however , this mechanism only works for nearly horizontal field lines . furthermore , hedgerow prominences are located in coronal cavities where the plasma pressure is very low . therefore , it seems unlikely that the vertical threads seen in hedgerow prominences can be supported by coronal plasma pressure on nearly vertical field lines . in this paper
we propose that hedgerow prominences are embedded in magnetic fields with a complex `` tangled '' structure .
such tangled fields have many dips in the field lines where the weight of the prominence plasma can be counteracted by upward magnetic forces .
our purpose is to demonstrate that such tangled fields provide a viable mechanism for prominence support in hedgerow prominences .
@xcite recently invoked tangled fields in the interpretation of spectropolarimetric observations of an active region filament .
while such filaments are quite different from the hedgerow prominences considered here , this work shows that tangled fields have important effects on the measurement of prominence magnetic fields .
such effects will not be considered in this paper .
the paper is organized as follows . in section 2
we propose that hedgerow prominences are supported by tangled magnetic fields , and we discuss how such fields may be formed . in section 3 we present a simple model for the dynamics of plasma along the tangled field lines , and we show that weak shock waves naturally occur in such plasmas . in section 4
we develop a magnetostatic model of tangled fields based on the _ linear force - free field _
approximation , and in section 5 we study the response of such fields to gravitational forces . in section 6 we simulate the distribution of plasma in a cylindrical prominence thread . in section 7 we discuss the formation of vertical threads by rayleigh - taylor instability .
the results of the investigation are summarized and discussed in section 8 .
the spectro - polarimetric observations of prominences described in section [ sect : intro ] are consistent with the idea that quiescent prominences are embedded in coronal flux ropes that lie horizontally above the polarity inversion line ( pil )
. figure [ fig : cartoon2]a shows a vertical cross - section through such a flux rope .
the magnetic field also has a component into the image plane , so the field lines are helices , and the plasma is assumed to be located at the dips of these helical windings . a dip is defined as a point where the magnetic field is locally horizontal and curved upward . as indicated in the figure , the magnetic field may be deformed by the weight of the prominence plasma , creating v - shaped dips .
the magnetic field near such dips is well described by the kippenhahn - schlter model , and several authors have developed local magnetostatic models of the fine structures observed in quiescent prominences ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
however , recent observations of `` dark plumes '' and rotational motions @xcite within prominences remind us again that prominences have complex internal motions , and it is not clear how such motions can be explained in terms of a single large flux rope
. perhaps the magnetic structure of hedgerow prominences is more complicated than that predicted by the flux rope model ( figure [ fig : cartoon2]a ) . in this paper
we propose an alternative model , which is illustrated in figure [ fig : cartoon2]b .
following @xcite , we suggest that hedgerow prominences are formed in current sheets that overlie certain sections of the pil on the quiet sun . unlike those previous authors we suggest that the current sheet extends only to limited height ( @xmath9 mm ) , and may extend only a limited distance along the pil .
furthermore , we propose that _ tangled _ magnetic fields are present within these current sheets .
a tangled field is defined as a magnetic structure in which the field lines are woven into an intricate fabric , and individual field lines follow nearly random paths .
we suggest that the field is tangled on a spatial scale of 0.11 mm , comparable to the pressure scale height @xmath10 of the prominence plasma ( @xmath11 mm ) .
the prominence plasma is assumed to be located at the many dips of the tangled field lines .
the tangled field is confined horizontally by the vertical fields on either side of the sheet , and vertically by the weight of the prominence plasma .
a key feature of a tangled field is that the plasma and field are in magnetostatic equilibrium , i.e. , the lorentz force is balanced by the gas pressure gradients and gravity .
therefore , a tangled field is quite different from `` turbulent '' magnetic fields in which large - amplitude alfvn waves are present ( e.g. , the solar wind ) . in a tangled field
the magnetic perturbations do not propagate along the field lines . in this paper
we examine the basic properties of tangled fields , and we investigate their ability to support the prominence plasma .
we suggest that the tangled field may be formed as a result of magnetic reconnection , not the twisting or stressing of field lines .
quiescent prominences are located above polarity inversion `` lines '' that are often more like wide bands of mixed polarity separating regions with dominantly positive and negative polarity . in these mixed - polarity zones ,
magnetic flux elements move about randomly and opposite polarity elements may cancel each other .
new magnetic bipoles frequently emerge from below the photosphere .
these processes causes the `` recycling '' of the photospheric flux about once every 2 to 20 hours , and the coronal flux is recycled even faster @xcite .
it is likely that the interactions between these flux elements produce a complex , non - potential magnetic field in the low corona . within
this environment magnetic reconnection is likely to occur frequently at many different sites in the corona above the inversion zone .
each reconnection event may produce a bundle of twisted field , and the twisted fields from different events may collect into larger conglomerates to form a tangled field . the tangled field may rise to larger heights ( as a result of its natural buoyancy ) , and may collect into a thick sheet that is sandwiched between smoother fields , as illustrated in figure [ fig : cartoon2]b .
the observed prominence consists of plasma that is trapped within this sheet .
new tangled field is continually injected into the sheet from below , producing vertical motions within the sheet .
we suggest that the `` dark plumes '' observed by @xcite may be a manifestation of such vertical motions of the tangled field .
the spatial distribution of plasma within the prominence is determined in part by the dynamics of plasma along the tangled field lines .
figure [ fig : flow_cartoon ] shows the contorted ( but generally downward ) path of an individual field line in the tangled field .
note that there are several `` dips '' where the field line is horizontal and curved _ upward _ , and `` peaks '' where the field is horizontal and curved _ downward_. tracing the field line upward from a dip , one always reaches a peak where the field line again turns downward .
therefore , the question arises whether the plasma collected in the dips would remain in these dips or be siphoned out of the dips via the peaks of the field lines . to answer this question , we consider a simple model for the motion of the prominence plasma along the magnetic field . for simplicity
we assume that the flow takes place in a thin tube surrounding the selected field line ( i.e. , the divergence of neighboring field lines is neglected ) , and the cross - sectional area of this tube is taken to be constant .
we assume a steady flow is established along the tube .
let @xmath12 and @xmath13 be the plasma velocity and density as functions of position @xmath14 along the tube , then conservation of mass requires @xmath15 = constant .
the equation of motion of the plasma is @xmath16 where @xmath17 is the plasma pressure , @xmath18 is the height above the photosphere , and @xmath19 is the acceleration of gravity . the equation of state is written in the form @xmath20 , where @xmath21 and @xmath22 are constants ( we use @xmath23 to describe non - adiabatic processes ) . eliminating @xmath17 and @xmath13 from equation ( [ eq : flow1 ] ) , we obtain the following equation for the parallel flow velocity : @xmath24 where @xmath25 is the sound speed ( @xmath26 ) .
the above equation has a critical point where the flow velocity equals the sound speed ( @xmath27 ) .
therefore , a transition from subsonic to supersonic flow can occur only at points where the rhs of this equation vanishes , @xmath28 .
these sonic points are located at the peaks of the field lines where matter can be siphoned out of one dip and deposited into another dip at lower height .
the resulting flow pattern is indicated in figure [ fig : flow_cartoon ] .
as the supersonic flow approaches the next dip , it must slow down to subsonic speeds , which can only happen in a shock .
therefore , the tube has a series of subsonic and supersonic flows separated by shocks and sonic points .
the role of these shocks is to dissipate the gravitational energy that is released by the falling matter .
the position and strength of the shocks can be computed if the height @xmath18 of the flow tube is known .
neighboring peaks are generally not at the same height .
therefore , each section between neighboring peaks is approximated as a large - amplitude sinusoidal perturbation superposed on a generally downward path : @xmath29 where @xmath14 is the position along the flow tube , @xmath30 is the distance between neighboring peaks ( as measured along the flow tube ) , @xmath31 is the amplitude of the perturbation in height , @xmath32 is a phase angle , and @xmath33 is the background slope .
the phase angle is chosen such that the peaks in the flow tube ( where @xmath28 ) are located at @xmath34 and @xmath35 , then the slope is given by @xmath36 the sonic points will then be located at @xmath34 and @xmath35 . figure [ fig : flow_shock]a shows the height @xmath18 for @xmath37 and @xmath38 rad , so that @xmath39 .
let @xmath40 , @xmath41 and @xmath42 be the pressure , density and sound speed at the sonic points , then @xmath43 , and the sound speed can be written as @xmath44^{(\gamma-1)/2 } = c_0 [ v(s)/c_0 ] ^{-(\gamma-1)/2 } .\ ] ] inserting this expression into equation ( [ eq : flow2 ] ) , we obtain @xmath45 where @xmath46 , and @xmath47 is the pressure scale height at the sonic points .
equation ( [ eq : flow3 ] ) can be integrated as follows : @xmath48 + \frac{1}{\gamma-1 } \left [ u^{1-\gamma}(s ) - 1 \right ] = - \frac{z(s)-z(s_0 ) } { \gamma h_{p,0 } } , \label{eq : flow4}\ ] ] where @xmath49 is the position of a sonic point .
equation ( [ eq : flow4 ] ) can be solved for @xmath50 by newton - raphson iteration .
when @xmath51 , there is an analytic solution for @xmath50 in terms of the lambert @xmath52 function ( see * ? ?
* ) ; in the present paper we assume @xmath53 .
the supersonic solution @xmath54 is computed with @xmath55 , and the subsonic solution @xmath56 is computed with @xmath57 .
the dashed curves in figure [ fig : flow_shock]b show @xmath54 and @xmath56 , and figure [ fig : flow_shock]c shows the corresponding sound speeds @xmath58 and @xmath59 . here
we assumed a scale height @xmath60 , equal to the amplitude of the field - line distortions .
the mach number of the flow is given by @xmath61^{(\gamma+1)/2 } .\ ] ] the shock is located at the point @xmath62 where the mach number @xmath63 before the shock and the mach number @xmath64 after the shock satisfy the following relationship : @xmath65 which follows from the rankine - hugionot conditions for parallel shocks @xcite . therefore , the actual flow velocity @xmath12 between the two sonic points is given by the full curve in figure [ fig : flow_shock]b , and the sound speed @xmath25 is given by the full curve in figure [ fig : flow_shock]c .
the plasma density @xmath13 along the flow tube is determined by mass conservation ( @xmath15 = constant ) , and is plotted in figure [ fig : flow_shock]d . note that there is a strong peak in the density at the dip in the field line , @xmath66 .
the dashed curve in figure [ fig : flow_shock]d shows the density profile that would exist if the plasma were in hydrostatic equilibrium ( he ) : @xmath67 , \label{eq : rho}\ ] ] where @xmath68 and @xmath10 are the density and pressure scale height at the dip , @xmath69 .
the deviations from hydrostatic equilibrium are significant only in those regions where the flow velocity is comparable to the sound speed .
we define an average flow velocity @xmath70 by @xmath71 for the case shown in figure [ fig : flow_shock ] we find @xmath72 , so the average flow speed is less than the sound speed .
therefore , the contorted shape of the flow tube significantly reduces the flow velocity compared to the supersonic free fall that would occur in a straight vertical tube .
the cooler parts of the prominence are thought to have a temperature @xmath73 k. assuming a hydrogen ionization fraction of 10% , a helium abundance of 10 % and @xmath53 , the sound speed @xmath74 km s@xmath3 , and we predict an average flow velocity @xmath75 km s@xmath3 .
the vertical component of this velocity is @xmath76 km s@xmath3 , less than the observed vertical velocities in prominence threads ( 510 km s@xmath3 ) .
note that the predicted velocity is relative to the _ pattern _ of the tangled field , therefore , if the tangled field expands in the vertical direction it will push the prominence plasma upward .
we speculate that the observed upward motions in hedgerow prominences ( e.g. , * ? ? ?
* ) are due to such large - scale changes in the tangled field .
in this section simple models for tangled fields are developed .
a volume @xmath77 in the corona is considered , and the plasma inside this volume is assumed to be in magnetostatic equilibrium , @xmath78 , where @xmath79 is the plasma pressure , @xmath80 is the density , @xmath81 is the acceleration of gravity , and @xmath82 is the lorentz force .
all quantities are functions of position @xmath83 within the volume .
the lorentz force is given by @xmath84 where @xmath85 is the electric current density and @xmath86 is the magnetic field . using tensor notation , equation ( [ eq : f ] )
can also be written as @xmath87 , where @xmath88 is the magnetic stress tensor , a special case of maxwell s stress tensor @xcite : @xmath89 the first term describes magnetic pressure , and the second term describes magnetic tension . in a tangled field both pressure and tension forces are important .
if gravity and plasma pressure gradients are neglected , then @xmath90 , so the magnetic field @xmath91 must satisfy the force - free condition : @xmath92 where @xmath93 may in general be a function of position . in the special case that @xmath2 is constant throughout the volume , equation ( [ eq : fff ] ) becomes a linear equation for @xmath94 , and the solutions are called linear force free fields ( lfff ) .
@xcite has shown that in a closed magnetic system with a prescribed magnetic helicity ( @xmath95 , where @xmath96 is the vector potential ) the lowest - energy state is a lfff .
therefore , in this paper only lfffs are considered , and @xmath2 is treated as a free parameter .
we find that lfffs can be tangled .
the typical length scale of the tangled field is given by the inverse of the @xmath2 parameter , @xmath97 . in the following
we first consider the case that @xmath98 is small compared to the domain size @xmath99 in _ any _ direction , and then consider the boundary effects .
section [ sect : cylinder ] describes tangled fields in a cylindrical domain . in the absense of boundary conditions , the solution of equation ( [ eq : fff ] )
can be written as a superposition of planar modes : @xmath100 , \label{eq : sum}\ ] ] where @xmath101 is the number of modes , @xmath102 is the wave vector ( @xmath103 ) , @xmath104 is the mode amplitude , @xmath105 is a phase angle , and @xmath106 $ ] are unit vectors that are mutually orthogonal and form a right - handed basis system : @xmath107 here @xmath108 and @xmath109 are the direction angles of the wave vector relative to the cartesian reference frame .
figure [ fig : volume ] shows an example of a field with @xmath110 modes , an isotropic distribution of direction angles @xmath111 , and randomly selected phase angles @xmath105 .
the starting points of the field lines are randomly selected from the central part of the box , and the field lines are traced until they reach the box walls .
note that individual field lines follow random paths , and that different field lines are tangled together .
now consider an ensemble @xmath112 of fields with different ( randomly distributed ) phase angles @xmath105 , but with a fixed number of modes @xmath101 , and with fixed mode amplitudes @xmath104 and direction angles @xmath113 .
the phase angles are assumed to be uniformly distributed in the range @xmath114 $ ] , and angles from different modes @xmath115 and @xmath116 are assumed to be uncorrelated .
then the ensemble average of the magnetic field vanishes , @xmath117 where @xmath118 denote the average over phase angles @xmath105 . also , the average of the tensor @xmath119 is given by @xmath120 @xmath121 where @xmath122 is the unit tensor ( @xmath123 ) .
expression ( [ eq : tens1 ] ) is independent of position @xmath83 , so the magnetic field is statistically _ homogeneous _ , but is not necessarily isotropic .
we study the statistical properties of field lines in models with different values of the mode number @xmath101 . for each @xmath101
we construct a series of models ( @xmath124 ) with different phase angles , but with constant values of the mode amplitudes @xmath104 and direction angles @xmath125 .
the mode amplitudes are taken to be the same for all modes ( @xmath126 ) .
for each realization @xmath127 of the phase angles @xmath105 we trace out the field line that starts at the origin ( @xmath128 ) , and we measure the square of the radial distance @xmath129 as function of position @xmath14 along the field line : @xmath130 we then average this quantity over @xmath131 realizations of the phase angles to obtain the mean square distance @xmath132 . for @xmath133 both mutually orthogonal directions @xmath125 and randomly chosen directions are considered . in both cases
we find that the field lines follow long helical paths , and @xmath134 increases quadratically with @xmath14 .
therefore , for @xmath133 the field lines do not behave randomly . for @xmath135
only randomly chosen directions are considered . for @xmath136
some of the field lines are long helices , while others have more random paths , and for @xmath137 all field lines seem random , however , in both cases @xmath132 is not well fit by a power law .
true random walk behavior of the field lines , as indicated by a _
linear _ dependence of @xmath138 on distance @xmath14 , is found only when the number of modes is increased to @xmath139 . in the limit of large @xmath101 , @xmath140 .
we now consider a larger ensemble @xmath141 in which not only the phase angles @xmath105 but also the mode amplitudes @xmath104 and direction angles @xmath113 are allowed to vary . from now on @xmath142 will denote an average over this larger ensemble .
the direction angles are assumed to have an isotropic distribution , i.e. , the angle @xmath109 is uniformly distributed in the range @xmath143 $ ] , and @xmath144 is uniform in the range @xmath145 $ ] , so that @xmath146 .
furthermore , the mode amplitudes @xmath104 are assumed to be uncorrelated with the direction angles , and @xmath147 , where @xmath148 is a constant .
then @xmath149 , so the mean magnetic field vanishes .
further averaging of equation ( [ eq : tens1 ] ) shows that @xmath150 is an isotropic tensor : @xmath151 it follows that @xmath152 , so @xmath148 equals the r.m.s .
value of the total field strength .
the ensemble average of the magnetic stress tensor , equation ( [ eq : t ] ) , is given by @xmath153 which is also isotropic .
note that the diagonal components of @xmath154 are negative , so the effects of magnetic pressure dominate over the effects of magnetic tension .
therefore , the isotropic tangled field has a positive magnetic pressure , @xmath155 .
the average energy density is @xmath156 .
the relationship @xmath157 is similar to that for a relativistic gas ( e.g. , * ? ? ?
the tangled field must be confined within a certain volume ( e.g. , a current sheet , see figure [ fig : cartoon2]b ) , and the confinement must be effective for a period much longer than the alfven travel time across the volume .
what are the conditions for such confinement ? to answer this question
we must consider the boundary region between a tangled field and a smooth field .
the tangled field is assumed to be characterized by a high value of @xmath158 , and the smooth field presumably has a much lower value of @xmath159 . to last a long time
, the magnetic field near the boundary must be approximately in equilibrium ( non - linear force - free field ) . the force - free condition ( [ eq : fff ] )
implies that @xmath2 is constant along field lines , so there can not be many field lines that pass from the smooth region to the tangled region .
therefore , one important condition for the survival of the tangled field is that the two regions are nearly disconnected from each other magnetically .
another requirement is that the two regions are approximately in pressure balance . to show that these conditions can be satisfied
, we now consider a simple model for the boundary region .
the interface between the tangled and smooth fields is approximated by a plane surface , here taken to be the plane @xmath160 in cartesian coordinates .
the above - mentioned condition on the lack of connectivity between the smooth and the tangled fields requires @xmath161 at @xmath160 .
the tangled field in @xmath162 is assumed to be a lfff with a specified value of @xmath2 .
the solution of the lfff equation is again written as a superposition of planar modes .
however , in the present case the modes are grouped into pairs with closely related wave vectors @xmath163 and @xmath164 , and with the same amplitude @xmath104 and phase @xmath105 : @xmath165 .
\label{eq : sum2}\end{aligned}\ ] ] here @xmath101 is the total number of modes , @xmath166 and @xmath167 are defined in equations ( [ eq : e1 ] ) and ( [ eq : e2 ] ) , @xmath168 is the modified wave vector , and the unit vectors @xmath169 and @xmath170 are defined by @xmath171 note that @xmath170 differs from @xmath172 only in the sign of the @xmath173-component , whereas @xmath169 has the sign of the @xmath174- and @xmath175-components reversed .
therefore , the unit vectors @xmath176 $ ] again form a right - handed basis system . the magnetic field at the boundary ( @xmath177 is given by @xmath178 @xmath179 , \label{eq : sum3}\ ] ] which satisfies @xmath180
therefore , it is possible to construct a tangled field that is disconnected from its surroundings .
we now consider the statistical average of the tensor @xmath181 at @xmath160 .
averaging over phase angles @xmath105 , we obtain @xmath182 @xmath183 and further averaging over mode amplitudes and direction angles yields @xmath184 here we assume an isotropic distribution of direction angles , and we use @xmath185 , where @xmath148 is the r.m.s .
field strength in the interior of the tangled field ( see section [ sect : volume ] ) .
note that at the boundary @xmath186 , while in the interior @xmath152 , so the r.m.s .
field strength at the boundary is reduced by a factor @xmath187 compared to that in the interior .
the magnetic pressure at @xmath160 is given by @xmath188 where @xmath189 is the average pressure in the interior of the tangled field [ see equation ( [ eq : stress ] ) ] .
equation ( [ eq : pext ] ) shows that it is possible to maintain pressure balance between the tangled field and its surroundings . here
an infinitely long cylinder with radius @xmath190 is considered .
we adopt a cylindrical coordinate system @xmath191 , and we assume that the radial component of magnetic field vanishes at the cylinder wall , @xmath192 . in the appendix
we analyze the eigenmodes of the lfff equation in the domain @xmath193 subject to the above boundary condition .
we find that this eigenvalue problem has a discrete set of modes , and the number of modes depends on the dimensionless parameter @xmath194 .
figure [ fig : tube ] shows the resulting magnetic fields for @xmath195 , 4.5 and 6.0 . in the first case only the axisymmetric ( lundquist ) mode is present , so the field lines are helical . assuming the cylinder axis is vertical , there are no dips in the field lines .
if cool plasma were to be injected into such a structure , it would spiral down along the field lines and quickly reach supersonic speeds .
in contrast , for @xmath196 and @xmath197 there are multiple modes of the lfff , and the random superposition of these modes creates a tangled field with many dips where prominence plasma can be supported . the field - line dips ( i.e. , sites where @xmath198 and @xmath199 ) are indicated by _ magenta _
dots in the middle and right panels of figure [ fig : tube ] .
we will return to this model in section [ sect : thread ] .
the above models for a tangled field are purely force - free and do not have any magnetic forces to support the prominence plasma against gravity . to include such effects , we now consider the `` elastic '' properties of the tangled field , i.e. , its response to external forces .
specifically , the weight of the prominence causes the tangled field to be compressed in the vertical direction , resulting in a radially outward force on the plasma . also , shearing motions may occur within the tangled field as dense plasma moves downward and less dense `` plumes '' move upward ( e.g. , *
this results in shear deformation of the tangled field and associated magnetic stresses that counteract the plasma flows . in the following both of these effects
are considered in some detail .
we first consider the effects of gravity on a layer of tangled magnetic field . the magnetostatic equation ( @xmath200 )
can not be solved analytically for a tangled field , so we make the following approximation : @xmath201 = \left [ { \bf b } ( { \bf r } ) - \frac{1}{\alpha h_b } \hat{\bf z } \times { \bf b } ( { \bf r } ) \right ] e^{-z / h_b } .
\label{eq : compress}\ ] ] here @xmath91 is the lfff given by equation ( [ eq : sum ] ) , and @xmath202 is the magnetic scale height of the modified field ( we assume @xmath203 ) .
note that @xmath204 as required .
this modified field @xmath205 is no longer force - free , but has the following statistical properties : @xmath206 e^{-2z / h_b } , \label{eq : tens2b } \end{aligned}\ ] ] where @xmath207 .
therefore , the magnitude of the modified field drops off exponentially with height @xmath175 .
let @xmath208 be the magnetic stress tensor of the modified field .
taking its statistical average , we find for the nonzero components of the stress tensor : @xmath209 note that for @xmath210 the stress tensor is nearly isotropic .
the net force on the plasma is given by @xmath211 , and the average force follows from equations ( [ eq : mxx ] ) and ( [ eq : mzz ] ) : @xmath212 note that the average force acts in the positive @xmath175 direction , i.e. , the magnetic force counteracts the force of gravity . in effect , the plasma is being supported by the magnetic pressure of the tangled field .
the tangled field acts like a hot gas that has a significant pressure but no mass .
the _ average _ density of the plasma that can be supported by the tangled field is given by @xmath213 where @xmath19 is the acceleration of gravity .
the horizontal components of lorentz force , @xmath214 and @xmath215 , do not vanish for any particular realization of the tangled field , and can not be written as the gradients of a scalar pressure @xmath79 .
the reason is that expression ( [ eq : compress ] ) is not an exact solution of the magnetostatic equilibrium equation .
however , equation ( [ eq : fxyp ] ) shows that the horizontal forces vanish when averaged over the fluctuations of the isotropic tangled field .
therefore , expression ( [ eq : compress ] ) is thought to give a good approximation for the effects of gravity on the tangled field .
we now apply the above model to the vertical threads observed in hedgerow prominences .
to explain the observed heights of such prominences , we require that the magnetic scale height @xmath202 is at least 100 mm .
the size @xmath98 of the magnetic tangles is assumed to be in the range 0.11 mm , so @xmath216 . for @xmath217 g and @xmath218 mm , we find @xmath219 @xmath220 , which corresponds to an average ( total ) hydrogen density @xmath221 @xmath222 .
this is only about 0.05 times the density @xmath223 @xmath220 or @xmath224 @xmath1 typically observed in hedgerow prominences .
this comparison shows that the pressure of the tangled field _ inside _ a prominence thread is not sufficient to support the weight of the prominence plasma . to support the plasma with tangled fields , we need to take into account the magnetic coupling between the vertical thread and its surroudings .
such coupling is neglected in the above plane - parallel model .
we now assume that the tangled field pervades not only the observed vertical threads but also their local surroundings .
the density in the surroundings is less than that in the threads , so the force of gravity is also much lower .
this difference in gravitational forces leads to vertical motions ( downflows in the dense threads , upflows in the tenuous surroundings ) that create magnetic stresses in the tangled field .
the magnetic coupling between the prominence and its surroundings causes the weight of the dense prominence to be distributed over a wider area . in effect , the prominence plasma is being supported by the radial gradient of the magnetic pressure of the tangled field over this larger area . in the following we estimate the magnetic stresses and vertical displacements resulting from these forces .
the tangled field is modeled either as a vertical slab with half - width @xmath190 , or as a vertical cylinder with radius @xmath190 .
the prominence is located at the center of this slab or cylinder , and has a half - width or radius @xmath225 .
then the average density in the tangled field region is given by @xmath226 where @xmath41 is the density of the prominence , and we neglect the mass of the surroundings .
the exponent @xmath227 for the slab model or @xmath228 for the cylindrical model .
as discussed in section [ sect : compress ] , observations of hedgerow prominences indicate @xmath229 @xmath220 , and to explain the observed heights of such prominences with @xmath217 g , we require @xmath230 . according to equation ( [ eq : rhoa2 ] ) ,
this implies @xmath231 for the slab model , or @xmath232 for the cylindrical model .
the observed threads have widths down to about 500 km , which corresponds to @xmath233 km .
therefore , the magnetic coupling by the tangled field must extend to a surrounding distance of at least 5 mm for the slab model , or 1.1 mm for the cylindrical model .
more generally , equations ( [ eq : rhoa1 ] ) and ( [ eq : rhoa2 ] ) yield the following expression for the magnetic scale height of tangled field : @xmath234 therefore , the maximum height of the prominence depends strongly on the magnetic field strength @xmath148 .
according to the present model , magnetic stress builds up in the tangled field as a result of the difference in gravitational forces between the thread and its surroundings .
can the field support such shear stress ? to answer this question we examine the effect of vertical displacements on the tangled field . for simplicity we neglect the mean vertical force given by equation ( [ eq : fzp ] ) , and we focus on _ relative _ displacements .
let @xmath235 be the new position of a fluid parcel originally located at position @xmath236 . in the limit of a perfectly conducting plasma , the deformed field @xmath237 at the new position @xmath235
is given by @xmath238 where @xmath91 is the original field , and @xmath239 is the jacobian of the transformation .
we assume @xmath240 which yields @xmath241 where @xmath242 is the vertical displacement .
the original field @xmath91 is assumed to be a realization of the isotropic tangled field given by equation ( [ eq : sum ] ) . using equation ( [ eq : tens2 ] ) ,
we obtain @xmath243 , \label{eq : tens3 } \end{aligned}\ ] ] where @xmath244 this yields the following expressions for the off - diagonal components of the stress tensor : @xmath245 the lorentz force is given by @xmath246 , and since @xmath247 is independent of @xmath175 , the average vertical force is given by @xmath248 where @xmath249 is the density perturbation ( @xmath250 ) . using this equation
, we can determine the vertical displacement @xmath242 for a given density variation @xmath251 .
in the following subsections we solve the above equation for the slab and cylinder models .
we first consider a slab with infinite extent in the @xmath252 , @xmath253 and @xmath254 directions . the coordinate @xmath173 perpendicular to the slab
is in the range @xmath255 , where @xmath190 is the half - width of the sheet in which the tangled field is embedded ( see figure [ fig : cartoon2]b ) .
then the density perturbation is given by @xmath256 & \mbox{for } ~ |x| < r_0 , \\
- \rho_0 r_0/r & \mbox{otherwise } .
\label{eq : drho1 } \end{array } \right.\ ] ] inserting this expression into equation ( [ eq : fzpp ] ) and solving for the vertical displacement , we obtain @xmath257 ( x^2 - r_0 ^ 2 ) & \mbox{for } ~ |x| < r_0 , \\ c ( r_0/r ) [ ( r - r_0)^2 - ( r-|x|)^2 ] & \mbox{otherwise } , \end{array } \right.\ ] ] where @xmath258 .
here we applied no - stress boundary conditions ( @xmath259 ) at @xmath260 .
note that @xmath261 and its derivative are continuous at the edges of the prominence ( @xmath262 ) .
the relative displacement across the tangled field is given by @xmath263 figure [ fig : shear]a shows the function @xmath261 for @xmath264 mm , @xmath265 mm , @xmath217 g and @xmath266 @xmath220 , so that @xmath267 mm .
note that a relatively small deformation of the tangled field ( @xmath268 ) is sufficient to redistribute the gravitational forces over the full width of the tangled field .
however , @xmath269 is larger than the pressure scale height of the prominence plasma ( @xmath11 mm )
. therefore , the deformation of the tangled field in the neighborhood of the prominence may have a significant effect on the distribution of the prominence plasma .
this issue will be further discussed in section [ sect : thread ] . for comparison of the tangled field slab model with the flux rope model ( figure [ fig : cartoon2 ] ) ,
we define the average _ sag angle _
@xmath270 of the prominence relative to its surroundings : @xmath271 where @xmath148 is the r.m.s .
field strength of the tangled field , and we assumed @xmath272 .
this expression is similar to that derived for the kippenhahn - schlter model : @xmath273 where @xmath274 is the magnetic field through the mid - plane of the prominence .
equation ( [ eq : ks ] ) describes the angle of the field lines in the flux rope model shown in figure [ fig : cartoon2]a .
therefore , the flux - rope and tangled field models are similar in their ability to explain the magnetic support of the prominence plasma , provided the half - width @xmath190 of the tangled field region is similar to the radius @xmath190 of the flux rope . we now consider a cylindrical model for a prominence thread with @xmath275 the distance from the ( vertical ) thread axis .
the density perturbation is given by @xmath276 & \mbox{for } ~ r < r_0 , \\
- \rho_0 ( r_0/r)^2 & \mbox{for } ~ r_0 < r < r .
\label{eq : drho2 } \end{array } \right.\ ] ] the vertical displacement is obtained by solving equation ( [ eq : fzpp ] ) , which yields @xmath277 & \mbox{for } ~
r < r_0 , \\
h_0 [ 2 r^2 \ln ( r / r_0 ) -r^2 + r_0
^ 2 ] /(r^2-r_0 ^ 2 ) & \mbox{for } ~ r_0 < r < r , \end{array } \right .
\label{eq : hr}\ ] ] where @xmath278 and we applied no - stress boundary conditions at @xmath279 .
then the total displacement across the tangled field is @xmath280 figure [ fig : shear]b shows the vertical displacement @xmath281 for @xmath282 mm , @xmath283 mm , @xmath284 g and @xmath285 @xmath220 , which yields @xmath286 mm and @xmath287 mm . in this case
@xmath269 is significantly larger than the pressure scale height ( @xmath11 mm ) , mainly because of the lower field strength compared to the case shown in figure [ fig : shear]a . in section [
sect : thread ] we consider the effect of such deformation on the field - line dips , and on the spatial distribution of the prominence plasma .
the above analyses only provide an rough estimate for the density of prominence plasma that can be supported by the tangled field .
the actual density distribution @xmath288 is likely to be much more complex for several reasons .
first , plasma will tend to collect at the dips of the field lines , so the density will vary on the spatial scale @xmath98 of the tangled field and on the scale of @xmath10 ; this effect will be considered in more detail in section [ sect : thread ] .
second , the density will vary with time because there are flows along the field lines ( see section [ sect : flows ] ) and these flows are likely to be non - steady .
also , the magnetic structure is not fixed and will continually evolve as dipped field lines are distorted by the weight of the prominence plasma . to predict the actual density will require numerical simulations of the interaction of tangled fields with prominence plasma , which is beyond the scope of the present paper .
according to the present theory , hedgerow prominences are supported by the pressure of a tangled magnetic field , which acts like a tenuous gas and is naturally buoyant .
it is well known that a tenuous medium supporting a dense medium is subject to rayleigh - taylor ( rt ) instability .
therefore , we suggest that the observed vertical threads may be a consequence of rt instability acting on the tangled field and the plasma contained within it . as cool plasma collects in certain regions of the tangled field , the weight of the plasma deforms the surrounding field , which causes even more plasma to flow into these regions .
a detailed analysis of the formation of prominence threads by rt instability is complicated by the fact that we presently do not understand how a tangled field responds to shear deformation . in section [
sect : shear ] we estimated the relative vertical displacement @xmath269 of the prominence plasma assuming no reconnection occurs during the deformation of the magnetic field by gravity forces [ see equations ( [ eq : dh1 ] ) and ( [ eq : dh2 ] ) ] . in this case
the tangled field behaves as an `` elastic '' medium with magnetic forces proportional to the displacement .
however , it is not clear that this approximation is valid .
high - resolution observations of prominences indicate that the dense threads move downward relative to their more tenuous surroundings with velocities of the order of 10 - 30 @xmath289 .
if the threads and their surroundings are indeed coupled via tangled fields , these relative motions imply that the field is continually being stretched in the vertical direction .
therefore , the shear stress continually increases with time , unless there is internal reconnection that causes the shear stress to be reduced .
we speculate that tangled fields have a tendency to relax to the lfff via internal reconnection .
a similar relaxation processes occurs in the reversed field pinch and other laboratory plasma physics devices .
therefore , the long - term evolution of prominence threads likely involves small - scale reconnection within the tangled field .
the tangled field may behave more like a `` plastic '' medium that is irreversibly deformed when subjected to shear stress .
such plasticity makes it possible to understand how the dense threads can move downward relative their the surroundings at a small but constant speed .
these flows significantly deform the tangled field , but the field is nevertheless able to support the plasma against gravity . a detailed analysis of reconnection in tangled fields and its effect on the prominence plasma is beyond the scope of the present paper .
the observed vertical structures likely reflect the non - linear development of the rt instability in hedgerow prominences . to establish a vertical column of mass resembling
a prominence thread will likely require vertical motions over a significant height range ( tens of mm ) .
starting from a homogeneous density distribution , it may take several hours for the threads to form by rt instability .
we now construct a model for the density distribution in a fully formed ( vertical ) prominence thread supported by a tangled field .
it is assumed that the rt instability has produced a vertical thread that is clearly separated from the rest of the prominence plasma .
therefore , only a single thread and its local surroundings are considered , and the tangled field is assumed to be contained in a vertical cylinder with radius @xmath282 mm . as discussed in section [ sect : flows ] , there will in general be mass flows along the tangled field lines , but for the purpose of the present model we neglect such flows and we assume that the plasma is in hydrostatic equilibrium along the field lines . to construct the density model , we first compute a particular realization of the lfff with @xmath290 ( see appendix for details ) . to account for the weight of the prominence plasma , this field
is further deformed as described by equation ( [ eq : hr ] ) .
the deformation parameters are @xmath291 mm and @xmath292 mm , which yields @xmath293 mm , somewhat larger than the values used in figure [ fig : shear]b .
as shown in section [ sect : compress ] , the weight of the prominence plasma causes the strength of the tangled field to decrease with height [ see equation ( [ eq : compress ] ) ] , but for simplicity this gradient is neglected here .
the pressure scale height is assumed to be constant , @xmath294 mm , which corresponds to a temperature of about 8000 k , typical for h@xmath2 emitting plasma in prominences .
we introduce cartesian coordinates @xmath295 with the @xmath175 axis along the cylinder axis ; the @xmath173 and @xmath174 coordinates are in the range @xmath296 $ ] , and the height @xmath175 is in the range @xmath297 $ ] .
the density @xmath298 in this volume is computed on a grid with @xmath299 grid points , using the following method .
we randomly select a large number of points within the cylinder and trace out the field lines that pass through these points . for each field line
we plot the height @xmath18 as function of position @xmath14 along the field line , and we find the peaks and dips in the field line . for each dip
we find the two neighboring peaks ( @xmath300 and @xmath301 ) and we determine the depth @xmath302 of the valley between these peaks .
we then iteratively remove shallow dips with @xmath303 by concatenating neighboring sections .
figure [ fig : dips ] shows an example for one particular field line ; the remaining dips with @xmath304 are indicated by squares .
we then compute the density @xmath13 in each interval , assuming hydrostatic equilibrium along the field line [ see equation ( [ eq : rho ] ) ] .
since we are interested only in _ relative _ densities , we set @xmath305 , the same for all dips on all field lines . finally , we distribute the density @xmath13 onto the 3d grid by finding the grid points that lie closest to the path of the field line . this process is repeated for 8000 field lines to obtain the density @xmath298 throughout the 3d grid .
figure [ fig : mass_plot ] shows the resulting density distribution .
the three panels show different projections obtained by integration in the @xmath173 , @xmath174 and @xmath175 directions , respectively [ for example , figure [ fig : mass_plot]a shows @xmath306 .
note that the plasma is concentrated in the central part of the cylinder ; this is due to the deformation of the magnetic field described by the vertical displacement @xmath281 , which changes the distribution of the field - line dips compared to that in the lfff .
the plasma is highly inhomogenous ( @xmath307 ) , and is distributed in sheets corresponding to surfaces of field - line dips . in some regions there
are multiple sheets along the line of sight ( los ) .
this is consistent with observations of the h i lyman lines in solar prominences , which indicate multiple threads along the los .
the above model is quasi - static and does not take into account the expected mass flows along the field lines ( section [ sect : flows ] ) , nor the dynamical processes by which the threads are formed ( section [ sect : rt ] ) .
constructing a more realistic 3d dynamical model of a prominence thread supported by tangled fields is beyond the scope of the present paper .
however , our main conclusion that the density distribution within the thread is highly inhomogeneous is likely to be valid also in the dynamical case .
we propose that hedgerow prominences are supported by magnetic fields that are `` tangled '' on a spatial scale of 1 mm or less .
a key feature of the model is that the plasma is approximately in magnetostatic balance , therefore , the model is different from earlier models in which the plasma is supported by mhd - wave pressure . in the present case
the perturbations of the magnetic field do not propagate along the field lines .
the tangled field is located within a large - scale current sheet standing vertically above the pil ( figure [ fig : cartoon2]b ) , and is not magnetically connected to the photosphere on either side of the pil .
such tangled fields may be formed by flux emergence followed by magnetic reconnection in the low corona . in this paper
we use a variety of methods to explore the interactions of prominence plasma with tangled fields . in section [
sect : flows ] a simple model for the downward flow of plasma along the distorted field lines was developed .
it was shown that such flows naturally develop standing shocks where the gravitational energy of the plasma is converted into heat ; this may be important for understanding the heating of prominence plasmas .
the average flow velocity is less than the sound speed , indicating that the tangled field is able to support the prominence plasma against gravity . in section [
sect : lfff ] linear force - free models of tangled fields were constructed .
tangled fields can be described as a superposition of planar modes .
we studied the statistical properties of such fields , and found random - walk behavior of the field lines when the number of modes is sufficiently large ( @xmath308 ) . to produce tangling of the field lines on a scale of 1 mm or less , as required for our model of hedgerow prominences
, we need @xmath309 @xmath310 , much larger than the values typically found in measurements of photospheric vector fields .
therefore , according to the present model , hedgerow prominences are embedded in magnetic fields with high magnetic helicity density .
we also considered the conditions for confinement of a tangled field by the surrounding smooth fields , and showed that despite the high helicity density tangled fields can be in pressure balance with their surroundings .
in section [ sect : forces ] the `` elastic '' properties of a tangled field were considered , i.e. , their linear response to gravitational forces assuming ideal mhd .
we found that the weight of the prominence plasma can be supported by the nearly isotropic magnetic pressure of the tangled field .
the tangled field pervades not only the observed vertical threads , but also their local surroundings .
the magnetic coupling between the threads and their surroundings is quite strong : vertical displacements of only 0.52 mm are sufficient to counteract the shear stress resulting from the different gravitational forces . in effect
, the weight of a dense thread is distributed over an area that is larger than the cross - sectional area of the thread . as discussed in section [ sect : shear ] , the observed densities in prominence threads ( @xmath311 @xmath1 ) can be supported by tangled fields with field strengths in the range 315 g. in section [ sect : rt ] we proposed that the observed vertical structures in hedgerow prominences are a consequence of rayleigh - taylor ( rt ) instability acting on the tangled field and the plasma contained within it
. the tangled field acts like a hot , tenuous gas and is naturally buoyant ; its support of the dense prominence plasma is likely to be unstable to flows that separate the gas into dense and less - dense columns .
the observed vertical structures likely reflect the non - linear development of the rt instability .
a detailed analysis of this instability is complicated by the fact that it requires internal reconnection to occur within the tangled field , and it is unclear how rapidly such reconnection can proceed .
therefore , it is difficult to predict the vertical velocities in prominence threads .
clearly , more advanced numerical simulations of the interaction of tangled fields with prominence plasma are needed .
finally , in section [ sect : thread ] we simulated the density distribution in a prominence thread , using a cylindrical model for the tangled field and its deformation by gravitational forces .
the results indicate that the threads have an intricate fine - scale structure .
multiple structures are superposed along the los , consistent with observations of the h i lyman lines @xcite .
while our model does not take into account any dynamical processes , the main conclusion that the density distribution is highly inhomogeneous is likely to be valid also in the dynamical case .
the images shown in figures [ fig : proms]a and [ fig : proms]b were obtained at the big bear solar observatory , which is operated by the new jersey institute of technology . the image shown in figure [ fig : proms]c
was obtained at the dutch open telescope , which is operated by utrecht university in the observatorio del roque de los muchachos of the instituto de astrofsica de canarias , la palma , spain .
according to the model presented in this paper , the vertical threads in hedgerow prominences are supported by tangled magnetic fields that pervade the dense threads and their local surroundings . in this section
we construct a cylindrical model of this tangled field .
we use a cylindrical coordinate system @xmath191 with @xmath275 the distance from the cylinder axis ( @xmath193 where @xmath190 is the cylinder radius ) , @xmath312 the azimuthal angle , and @xmath175 the height along the axis .
we assume that the radial component of the field vanishes at the cylinder wall , @xmath192 , so the field lines are confined to the interior of the cylinder .
it follows that the axial magnetic flux must be constant along the cylinder : @xmath313 in cylindrical coordinates the force - free condition reads : @xmath314 we again take @xmath2 to be constant ( lfff ) .
the general solution of the above equations with the boundary condition @xmath192 can be written as a superposition of discrete eigenmodes enumerated by an index @xmath115 : @xmath315 where @xmath316 is the mode amplitude ; the functions @xmath317 , @xmath318 and @xmath319 describe the radial dependence of each mode ; @xmath320 is the axial wavenumber ; @xmath321 is the azimuthal wavenumber ( non - negative integer ) ; and @xmath322 is the phase .
here @xmath227 refers to a fully symmetric mode with @xmath323 ( also known as the lundquist mode ) , while @xmath324 refers to modes that have either a @xmath312 or @xmath175 dependence .
the latter modes only exist under certain conditions ( see below ) . inserting expressions ( [ eq : br ] ) , ( [ eq : bp ] ) and ( [ eq : bz ] ) into equations ( [ eq : f1 ] ) , ( [ eq : f2 ] ) and ( [ eq : f3 ] )
, we find that the radial dependencies can be expressed in terms of bessel functions : @xmath325 where @xmath326 is the radial wavenumber , @xmath327 is the bessel function of order @xmath321 , and @xmath328 is its derivative .
the boundary condition at the cylindrical wall requires @xmath329 for all modes . introducing a parameter @xmath108 in the range @xmath330 $ ] such that @xmath331 and @xmath332
, we obtain the following equation for @xmath108 : @xmath333 where @xmath194 .
the roots of this equation can be found numerically for any given values of @xmath334 and @xmath127 .
depending on the value of @xmath334 , the equation may have one or more solutions : 1 .
there always exists at least one solution , the axisymmetric mode ( @xmath335 ) with @xmath336 . in this case @xmath337 and @xmath338 , so this mode is invariant with respect to translation along the @xmath175 axis ( lundquist mode ) .
2 . there may be additional axisymmetric modes that are not invariant to translation . in the case @xmath339 and @xmath340 , equation ( [ eq : bc ] ) yields @xmath341 , where @xmath342 is a root of @xmath343 ( @xmath344 , @xmath345 , @xmath346 , etc . ) .
solutions exist only when @xmath347 ; for @xmath334 in the range @xmath348 $ ] there exists one such solution with @xmath349 ; for @xmath334 in the range @xmath350 $ ] there exist two solution , etc .
since @xmath351 is symmetric with respect to @xmath352 , each solution @xmath108 also has a second solution @xmath353 , but the magnetic structure of these solutions is the same , so we do not count it as a separate mode .
3 . for higher values of @xmath334 there exist non - axisymmetric solutions ( @xmath354 ) .
the first modes with @xmath355 occur at @xmath356 ; these `` kink '' modes are apparently more easily excited than the axisymmetric modes with @xmath357 .
the first modes with @xmath358 occur at @xmath359 . for a given value of @xmath334
, we systematically find all roots of equation ( [ eq : bc ] ) , starting with @xmath360 and then increasing @xmath127 until no more roots are found .
figure [ fig : modes ] shows @xmath108 as function of @xmath334 .
note that the number of modes @xmath101 increases with @xmath334 .
we now compute the axial flux and magnetic energy density . inserting equations ( [ eq : bz ] ) and
( [ eq : fz ] ) into equation ( [ eq : phi1 ] ) , we find that only the lundquist mode contributes to the axial magnetic flux : @xmath361 where we used @xmath362 for axisymmetric modes . the mean magnetic energy density is defined by @xmath363 where @xmath142 denotes an average over @xmath312 and @xmath175 .
let @xmath364 and @xmath365 denote two _ different _ modes , then averages of cross products such as @xmath366 @xmath367 vanish .
therefore , the magnetic energy can be written as a sum over individual modes , @xmath368 , where @xmath369 r ~dr , \label{eq : e1 } \\
e_n & = & \frac{\tilde{b}_n^2}{8 \pi r^2 } \int_0^r \left [ f_{r , n}^2 ( r ) + f_{\phi , n}^2 ( r ) + f_{z , n}^2 ( r ) \right ] r ~dr ~~~~~ \mbox{for } ~ n > 1 .
\label{eq : en}\end{aligned}\ ] ] for simplicity we consider _ equipartition _ tangled fields in which the various modes @xmath115 of the lfff have equal magnetic energy .
this implies @xmath370 , the same for all modes ( including the lundquist mode ) , which provides a relationship between the mode amplitudes @xmath316 .
the phase angles @xmath322 of the non - axisymmetric modes are assigned random values in the range @xmath143 $ ] .
this results in a tangled magnetic field @xmath91 described by equations ( [ eq : br ] ) , ( [ eq : bp ] ) and ( [ eq : bz ] )
. bommier , v. , & leroy , j .-
1998 , in _ new perspectives on solar prominences _ , iau colloq .
167 , eds .
d. webb , d. rust , b. schmieder .
series , vol .
150 ( astron .
soc . of the pacific , san francisco ) , p. 434
leroy , j. l. 1989 , in dynamics and structure of quiescent solar prominences ; proc . of workshop ,
palma de mallorca , spain , nov .
1987 , ed .
e. r. priest ( dordrecht , kluwer academic publishers ) , p. 77 - 113 .
lin , y. , martin , s. f. , engvold , o. 2008 , in _ subsurface and atmospheric influences of solar activity _ , ed .
r. howe , r.w .
komm , k.s .
balasubramaniam , g.j.d .
series , vol .
383 ( astron . soc . of the pacific , san francisco ) , p. 235
lpez ariste , a. , aulanier , g. 2007 , in _ the physics of chromospheric plasmas _ , ed . by p. heinzel , i. dorotovic , r.j .
series , vol .
368 ( astron . soc . of the pacific , san francisco ) , p. 291
malherbe , j .-
1989 , in _ dynamics and structure of quiescent solar prominences _ , proc . of workshop ,
palma de mallorca , spain , nov .
1987 , ed .
e.r . priest
( dordrecht , kluwer academic publishers ) , p. 115 paletou , f. 2008 , in _
sf2a-2008 : proc . of the annual meeting of the french society of astronomy and astrophysics _ , ed . c. charbonnel , f. combes , r. samadi , p. 559 .
available online at http://proc.sf2a.asso.fr | solar prominences are an important tool for studying the structure and evolution of the coronal magnetic field . here
we consider so - called `` hedgerow '' prominences , which consist of thin vertical threads .
we explore the possibility that such prominences are supported by tangled magnetic fields .
a variety of different approaches are used .
first , the dynamics of plasma within a tangled field is considered .
we find that the contorted shape of the flux tubes significantly reduces the flow velocity compared to the supersonic free fall that would occur in a straight vertical tube .
second , linear force - free models of tangled fields are developed , and the elastic response of such fields to gravitational forces is considered . we demonstrate that the prominence plasma can be supported by the magnetic pressure of a tangled field that pervades not only the observed dense threads but also their local surroundings .
tangled fields with field strengths of about 10 g are able to support prominence threads with observed hydrogen density of the order of @xmath0 @xmath1 . finally , we suggest that the observed vertical threads are the result of rayleigh - taylor instability
. simulations of the density distribution within a prominence thread indicate that the peak density is much larger than the average density .
we conclude that tangled fields provide a viable mechanism for magnetic support of hedgerow prominences . |
the tremendous progress of understanding of perturbative superstring theory as well as its duality symmetries in various spacetime backgrounds @xcite has led to a well substantiated proposal of m theory , the unifying theory of all superstring theories @xcite-@xcite .
the new objects contained in m - theory which are solitonic gravitational back - reactions of various d - branes , are the m2 and m5-branes .
these objects were expected naturally from the eleven dimensional ( 11d ) supergravity and , in this framework , they are considered as fundamental as the strings are for the various 10d supergravities .
the basic obstacle in understanding these objects as fundamentals , lies in the absence of a coupling constant and the consequent problem of the definition of their self - interactions .
an interesting proposal to define the self - interactions of the branes is to use their euclidean instantons to interpolate between vacua ( asymptotic states ) with different number of branes @xcite .
the simplest ones could be instantons interpolating between states of one and two branes .
the study of the simplest such instantons is already a difficult problem but one hopes that finding explicit solutions and trying to understand their moduli space would be and interesting approach .
the classification of all such instantons and the geometry of their moduli space is probably beyond the present day capabilities .
almost three decades ago the problem of determining the corresponding self - duality equations for membranes has been solved in the case of @xmath10 dimensional embedding spacetime in ref @xcite and in the case of @xmath11 in ref @xcite . in the latter , using results from the study of nahm equations @xcite for yang - mills ( ym ) monopoles it was shown that the self - duality equations form an integrable model and its lax pair and conserved quantities were determined
. an important work by ward @xcite clarified further the situation reducing the self - duality equations to laplace equation in 3d flat space , for the time function of the surface . this function is a level set function -as it is called in morse theory- and the problem of determining topology changing membrane instantons is reduced into the search for non - trivial ( multi - valued ) time functions .
recent progress in this direction has been made in the works @xcite .
the classification of self - duality ( sd ) equations in various dimensions has shown that the possible sd equations are determined by the existence of cross products of vectors which in turn is related to the four division algebras @xcite . apart from the trivial cases of branes of dimension @xmath12 embedded in @xmath13 space dimensions , the interesting cases concern the @xmath2 brane ( membrane ) and the @xmath0 brane . in references @xcite and @xcite the study of membrane instanton equations in higher dimensions has provided some insight in the difficulty of the problem and certain explicit solutions have been obtained ( see also references @xcite-@xcite ) . in the present work we extend the study to the case of 3-branes in @xmath14 spatial dimensions using the octonionic cross product for three 8-dimensional vectors . we obtain a convenient form of a four complex set for the sd equations which enable us to reduce the sd equations in six dimensions .
the sd equations imply the euclidean second order equations for the 3-branes and satisfy automatically the gauss law of volume preserving diffeomorphisms symmetry of the theory . in the case of external fluxes in the theory such as the one coming from @xmath15-wave supersymmetric gravitational backgrounds , it is possible to redefine the time and to reduce the sd equations to the case of flat background , and find explicit solutions for the 3-sphere and the 3-torus . in the case of the sphere the instantons are periodic in euclidean time going from a finite radius to infinity and back in finite time , while in the case of the three torus the periodic solution interpolates between finite radii .
the layout of this paper is as follows . in section 2
we review the hamiltonian formalism and the equations of motion ( eom ) for 3-branes in the light cone gauge and in flat spacetime , as well as , the corresponding gauss law .
we discuss the symmetry in this gauge which is the volume preserving diffeomorphisms of the 3-brane and which gives the possibility to describe the 3-brane as an incompressible fluid . in sections 3 and 4
, we derive the first order self - duality equations in eight dimensions and its various lower dimensional reductions and we present them in a very useful and suggestive complex form in four - dimensional complex space . we show that these equation imply the second order equations and the gauss law in euclidean signature spacetime . in section 5 , assuming factorization of time , we solve analytically the 3-brane self - duality equations for the case of spherical ( @xmath16 ) and toroidal ( @xmath17 ) topology of the brane .
finally , in section 6 we present our conclusions and discuss the application of our methods to the issue of topology change of three branes , a problem which is relevant to cosmological brane models .
in this section we briefly present the hamiltonian system in terms of the nambu 3-brackets . for the @xmath16 brane in @xmath18 minkowski spacetime
the hamiltonian in the light - cone gauge is = d_3 ( _ i^2+\{x_i , x_j , x_k}^2 ) [ ham ] where the indices run the @xmath19 transverse dimensions .
the corresponding eom are _
i = \{\{x_i ,
x_j , x_k } , x_j , x_k } , and the gauss law takes the form @xmath20 the volume element is d_3 = _ 2 ^2_3 d_1d_2d_3 , and the nabu 3-bracket for @xmath16 is defined as follows \{x_i , x_j , x_k } = ^ _ x_i _ x_j _ x_k for @xmath16 there are four functions , namely the polar coordinates of the unit four - vectors , e_1=_1_2_3 , e_2=_1_2_3 , e_3=_2_3 , e_4=_3 [ r4v ] they satisfy the @xmath21 relations \{e_a , e_b , e_c } = ^abcde_d , [ so4 ] where the indices take the values @xmath22 .
these equations are instrumental for the factorization of the time from the internal coordinates of the 3-brane .
it is possible to write down explicitly the infinite dimensional algebra of the volume preserving diffeomorphisms @xmath16 , \{y _ , y _ , y _ } & = & f^_y _ , using as a basis the hyper - spherical harmonics y_()&=&y_nlm(_1,_2 , _ 3 ) , = ( nlm ) , m =- l , , l , l=0,1,2 , ,
the structure constants @xmath23 can be expressed in terms of the 6-@xmath24 symbols of @xmath25 since @xmath26 .
when the brane has the toroidal topology @xmath17 , the global symmetries are @xmath27 and three translations along the cycles of the torus .
the basis of functions on @xmath17 is taken to be e_n = e^in,^3 , n = ( n_1,n_2,n_3 ) this basis defines an infinite dimensional volume preserving diffeomorphism group of the torus through the nabu - lie 3-algebra \ { e_n_1 , e_n_2 , e_n_3}= -i det ( n_1,n_2,n_3 ) e_n_1+n_2+n_3 it is important to notice that @xmath28 the above algebra remains invariant under the transformations @xmath29 .
thus , for this 3-dimensional brane there is an important discrete symmetry , namely the 3-dimensional modular group @xmath30 which could have implications in the quantum mechanical spectrum of this object .
the eom for the 3-torus are formally the same as in @xmath16 but it is much more convenient to complexify the eight transverse coordinates to four complex ones as we shall see later in the next section .
it is known that @xmath12-fold , @xmath31-dimensional vector cross products are defined only in the following cases p=1,2,3 , & & n = p+1 + p=2&&n=7 + p=3&&n=7 + p=3&&n=8.[crpro ] in the above cases , for any set of @xmath12 vectors , @xmath32 , their @xmath12-fold cross product , which we denote by @xmath33 , is a linear map which satisfies the following properties _ p(v_1,v_2 , , v_p)v_k & = & 0 , k=1,2,3 , , p , while the norm of the cross product satisfies the important property , |_p(v_1,v_2 , , v_p)|^2&= & det(v_iv_j ) [ ncp ] in the present work we will focus on the last case @xmath34 of ( [ crpro ] ) since the other cases can be obtained by appropriate reductions to lower dimensions .
now , we proceed to derive the light - cone self - duality equations for 3-branes living in eight transverse flat dimensions .
we note in passing that it is possible after toroidal compactification to obtain interesting equations for charged 3-branes in lower dimensions .
it is easy to observe that the potential energy term of the hamiltonian ( [ ham ] ) can be rewritten in terms of the determinant of the induced metric @xmath35 the cross product of three 8-dimensional vectors of case ( [ crpro ] ) is explicitly given as @xcite _ i(v_1,v_2 , v_3 ) : = _ ijklv_1^jv_2^kv_3^l , = 1,2,8 [ ncp3 ] where the definition of @xmath36 and conventions used in the paper are given in the appendix . setting @xmath37 where @xmath38 and @xmath39 , we find that the potential energy of the 3-brane is the norm squared of the above defined cross product of its tangent vectors ( see eq ( [ ncp ] ) ) . another way to see that is to use directly the identity ( [ phiphi ] ) given in the appendix for the tensor @xmath36 .
it is obvious now that the hamiltonian ( [ ham ] ) of the 3-brane , can be written as = d_3 ( _ i^2+\{x_i , x_j , x_k}^2 ) d_3 ( _ k - i _ k)(_k+i _
k ) , [ ham2 ] where @xmath40 and @xmath41 for vacuum configurations , @xmath42 , and in euclidean time we find the self - duality equations _
i = _ ijkl\{x_j , x_k , x_l } [ sd8 ]
there is a natural generalization of the self - duality equations in 8 dimensions in terms of the symbol @xmath36 , _
i = _ ijkl\{x_j , x_k , x_l}. [ sd88 ] where now the indices @xmath43 run for 1 to 8 . for convenience ,
we introduce the shorthand notation x_ijk\{x_i , x_j , x_k}[notijk ] where the indices run from 1 to 8 .
we know however that the same equation exists in seven dimensions that is , @xmath44 , because of the existence of the cross product @xmath45 given in section 3 .
then , employing the definition of @xmath36 given in the appendix , we find that eq ( [ sd88 ] ) implies the following eight non - linear ( first order ) differential equations _ 1&= & x_823+x_865+x_847+x_735+x_762+x_524+x_346 + _ 2&= & x_183+x_684+x_857+x_743+x_167+x_635+x_154 + _ 3&= & x_812+x_485+x_867+x_472+x_751+x_265+x_416 + _ 4&= & x_862+x_835+x_871+x_567+x_732+x_512+x_136[sda ] + _ 5&= & x_816+x_845+x_872+x_647+x_371+x_623+x_146 + _ 6&= & x_851+x_824+x_873+x_457+x_172+x_325+x_314 + _ 7&= & x_814+x_836+x_825+x_546+x_531+x_432+x_612 + _ 8&= & x_321+x_615+x_642+x_534+x_174+x_376+x_275 we notice that the self - duality equations of 3-branes in seven dimensions mentioned before , are simply obtained from the above system , by eliminating the last equation and the terms containing the index 8 on the right - hand side of the remaining seven equations . next , we proceed to the complexification of the general system ( [ sda ] ) by choosing specific pairing of the 8 real coordinates , as follows z_1= x_1+i x_4 , z_2= x_2+i x_5 , z_3= x_3+i x_6 , z_4= x_8+i x_7 [ x2z ] after some elaborate manipulation of the original system of equations , we arrive at the following simple form _
1&=&-12 \{z_1,z_k,|z_k}+ \{|z_2,|z_3,|z_4}[ce1 ] + _ 2&=&-12 \{z_2,z_k,|z_k}+ \{|z_3,|z_4,|z_1}[ce2 ] + _
3&=&-12 \{z_3,z_k,|z_k}+ \{|z_4,|z_1,|z_2}[ce3 ] + _
4&=&-12 \{z_4,z_k,|z_k}+ \{|z_1,|z_2,|z_3}[ce4 ] we observe that the equations ( [ ce1]-[ce4 ] ) have an @xmath46 symmetry acting on the complex vector @xmath47 .
it is possible from the four complex equations to obtain various interesting reductions to lower dimensions .
for example , if we demand that @xmath48 are real ( i.e. , @xmath49 , or if @xmath50 are all pure imaginary , that @xmath51 , then we obtain the self - duality equation of three branes in four real dimensions _
i=_ijkl\{y_j , y_k , y_l } , where @xmath52 represent the non - zero coordinates of @xmath50 for each of the above two cases .
a more detailed study of toroidal compactifications and double dimension reduction to lower dimensional branes , as well as their relation to extended continuous toda systems , will be discussed in a forthcoming work @xcite .
we now study the solutions of the four complex non - linear differential equations ( [ ce1]-[ce4 ] ) . this system can be considerably simplified if we seek solutions with time factorization z_a= _
a(t ) f_a(_1,_2,_3).[ansatz ] in the subsequent analysis , we work out the cases of spherical and toroidal topology . for spherical topology of the closed 3-brane , @xmath16 , we choose the functions @xmath53 to be @xmath54 , where @xmath55 are the polar coordinates of unit 4-vectors ( [ r4v ] ) in @xmath56 . in this factorization ,
the first term of the right - hand side of the complex equations ( [ ce1]-[ce4 ] ) is proportional to @xmath57 which is identically zero .
the simplified equations now read _ 1&=&|_2|_3|_4[cz1 ] + _ 2&=&|_3|_4|_1[cz2 ] + _ 3&=&|_4|_1|_2[cz3 ] + _ 4&=&|_1|_2|_3[cz4 ] we notice that the equations are invariant under a global @xmath58 symmetry @xmath59 with @xmath60 . in the following , we consider the conjugate equations and form the four combinations @xmath61 for @xmath62 . assuming polar form _
k = r_k(t ) e^i_k(t ) , [ polarf ] we find r^2_k_k&=&q , where @xmath63 is a conserved quantity q =- r_1r_2r_3r_4 , = _
i=1 ^ 4_i furthermore , subtraction for two different values of @xmath64 gives three new constants of motion |_k|^2-|_l|^2 = c_kl , which are the analogue of euler s equations for the rigid body . substituting ( [ polarf ] ) in the sums ,
@xmath65 we observe = 2 r_1r_2r_3r_4 for @xmath66 , using the conservation laws we write this equation as follows s= [ wei4 ] ( left ) and @xmath67 ( right ) , title="fig:",scaledwidth=46.5% ] ( left ) and @xmath67 ( right ) , title="fig:",scaledwidth=46.5% ] this equation can be solved using standard elliptic functions .
depending on the initial conditions , we divide the solutions into two classes : those with @xmath68 where there is no rotation , and those with @xmath69 where we have rotating instantons . in figure [ fig:1 ]
we plot the ` effective potential ' ( i.e. the square of the right - hand side of the differential equation ) for @xmath68 and @xmath69 . in the first case ( left plot )
we show two admissible cases .
the upper curve stands for four distinct real roots ( radii ) , and the second curve corresponds to a double root , i.e. , when two radii are equal .
curves below this one do not cross the real axis and correspond to complex radii which is not admissible . in the second plot ,
we fix the value of @xmath70 and draw three characteristic cases .
first , we consider the case @xmath68 and the possibility of static solutions , @xmath71 . if one of the radii is zero ( @xmath72 or @xmath73 ) , then we obtain static solutions living in the other six dimensions .
in general , we obtain a solution in terms of the elliptic integral of the first kind @xmath74 .
in particular , the positivity of all roots requires @xmath75 . assuming @xmath76 while taking @xmath77 at the initial time @xmath78 , integration of ( [ wei4 ] ) gives t - t_0 = ( i k ( ) - f(^-1()| ) ) [ times ] in order to find @xmath79 we need the inverse of the elliptic integral of the first kind @xmath80 which is given by the jacobi amplitude @xmath81 .
furthermore , in order to absorb unimportant factors , we redefine time @xmath82 and make the definitions k&= & + & = & =( am(t , k ) ) , [ sigt ] ( also written in the literature as @xmath83 ) and finally solve ( [ sigt ] ) for @xmath79 . in figure [ fig : y ] ( left side ) we plot time as a function of @xmath84 for @xmath85 and three sets of @xmath86 values . in all cases , the radius goes to infinity at finite time . for the sake of simplicity , we exemplify the above for the case of a double root @xmath87 , i.e , when two of the radii are equal . then _
a^s= ( 2- ^-1 ( ) ) if we define @xmath88 , a new time @xmath89 , and @xmath90 we obtain s(t ) a= this takes its minimum value for @xmath91 and grows rapidly to infinity in finite time , at the zeros of the denominator .
a few representative cases are plotted in the right side of figure [ fig : y ] . for @xmath68 and a double root @xmath92 for the values @xmath93 in the case of spherical topology .
, title="fig:",scaledwidth=46.5% ] for @xmath68 and a double root @xmath92 for the values @xmath93 in the case of spherical topology .
, title="fig:",scaledwidth=46.5% ] the case @xmath69 can be separated into two classes : first , pure uniform rotational motion with time independent radii ( euler tops ) , and second , when we have both rotation and bounce motion . in the second class
we consider three different cases according to the initial conditions for the radii and the value of @xmath63 . without loss of generality we can order the initial values of the radii in degreasing magnitude , that is @xmath94 , which implies @xmath76 .
the solution can be written in terms of the elliptic integral of the first kind , once we use the four roots @xmath95 which are algebraic functions of @xmath96 and @xmath63 . from the geometry of the potential plot
, we see that when @xmath68 , @xmath97 . when @xmath69 but smaller than a critical value , @xmath98 , the root @xmath99 is negative , @xmath100 and @xmath101 increase while @xmath102 decreases .
when @xmath103 , two roots become equal , @xmath104 , and the potential has a local maximum @xmath105 . for the case of toroidal topology , we define z_a = _
a(t ) e ^n_a [ taz ] the four vectors @xmath106 are not completely arbitrary .
when we substitute ( [ taz ] ) into the system of the four complex equations ( [ cz1]-[cz4 ] ) we find that the above ansatz provides a solution only if n_1 + n_2 + n_3 + n_4 = [ ncond ] then we get _ 1&=&+in |_2|_3|_4[tz1 ] + _ 2&=&-i n |_3|_4|_1[tz2 ] + _ 3&=&+i |_4|_1|_2[tz3 ] + _ 4&=&-i |_1|_2|_3,[tz4 ] where @xmath31 stands for the determinant n = det ( n_j , n_k , n_l ) . from these and their complex conjugates
we readily get , |_1|^2= |_3|^2&=&+i n ( |_1|_2|_3 |_4-_1_2 _ 3 _ 4 ) + addition of the appropriate equations gives |_1|^2-|_3|^2&= & c_13[ts1 ] + and find that q_1=q_3 = q ( |j+j ) , q_2=q_4=-q=- ( |j+j)[currents ] where @xmath107 and @xmath108 its complex conjugate .
we observe now that @xmath109 , which implies @xmath110 so that the charge @xmath63 is conserved .
writing @xmath111 , we find that @xmath112 . furthermore , using the redefinitions @xmath113 and @xmath114 , while taking @xmath66 , we obtain the differential equation & = & , with @xmath115 .
this is exactly the same equation as in the case of spherical topology , but notice that @xmath116 are bounded form equations ( [ ts1]-[ts4 ] ) .
we note that with the same method it is also possible to solve the four - dimensional 3-brane self - duality equations presented before , where we shall end up with the real system in the place of equations ( [ cz1]-[cz4 ] ) . in this case , the system of equations is the elegant integral system of refs @xcite .
the solution for the case of toroidal topology is expressed for both cases @xmath68 and @xmath69 by the same elliptic integral as in the spherical topology , but from the conservation laws we find the interesting result that the motion is oscillatory and bounded and so , the 4-dimensional manifold is compact .
this is contrary to the case of spherical topology where we have always non - compact world - volumes .
since this euclidean world volumes are calibrations of the ( @xmath117)-dimensional embedding spacetime , our solution is among the first examples of compact non - associative ( octonionic ) calibrations .
indeed , ordering the four radii in decreasing order @xmath118 , inspection of the equations ( [ ts1]-[ts4 ] ) shows that the constants @xmath119 must satisfy @xmath120 with @xmath121 positive while the variable @xmath122 is bounded between @xmath123 and @xmath124 , where now @xmath125 . in figure
[ fig : tt ] , the potential is plotted for three characteristic values of @xmath63 .
the time evolution of the four radii , is also shown in figure [ tor4 ] . as already emphasized , all four radii are bounded and interpolate between a minimum and a maximum value . for @xmath126 and three characteristic @xmath63 values , @xmath68 , @xmath127 and the critical value @xmath103 for toroidal topology.,scaledwidth=46.5% ] and the sets @xmath128 .
notice that all of the radii are bounded.,scaledwidth=40.0% ]
the present work provides a method for solutions of the self - duality equations for 3-branes in higher dimensions .
the factorization of time exploits the finite sub - algebras of the volume preserving diffeomorphisms and reduces the sd equations to well known integrable systems with explicit solutions in terms of the standard elliptic functions .
the new result is that the 3-sphere instanton interpolates between flat space - time at infinity and a finite radii 3-sphere .
this is similar to the einstein - rosen wormhole solutions of general relativity .
the minimization of the euclidean 4-volume in 8-dimensions , which is the origin of the sd equations , classifies their solutions as calibrations of the background geometry . for the sphere
this calibration is non - compact while for the 3-torus the calibration is compact @xcite .
it is possible to solve with the same method sd equations in maximally supersymmetric geometry backgrounds such as @xmath15-waves with fluxes in @xmath117 spacetime dimensions by redefining the time variable @xcite , a trick which has also a physical implication in the interpretation of the time as the group renormalization scale and the sd equations as flow equations between various geometries
. it would be interesting to apply similar ideas with those of ref @xcite to transform the nonlinear three - brane sd equations into a linear laplace equation problem for @xmath117 space time dimensions to study possible topology changes @xcite .
another interesting direction is to abandon the factorization of time and try to solve the three - brane sd equations in @xmath117 dimensions requiring axial symmetry and proceed in analogy with the study of real time bps states of 3-branes .
the authors would like to thank cern theory division for their kind hospitality and the stimulating atmosphere during which the main part of this work was realized .
egf would like to thank also the theory department of ens in paris for their kind hospitality and stimulating atmosphere during which the last parts of this paper were finished .
the structure constants of the octonionic multiplications @xmath129 and its dual @xmath130 , which measures the non - associativity of octonions , are given by ( for a textbook , see for example @xcite ) _
ijk&=&\ { when two indices are summed , a useful multiplication rule connecting the two symbols is _
ijk_lmk=_il_jm-_im_jl+_ijlm [ psiphi ] the corresponding identity for the @xmath36 symbols is written as follows @xcite _ abcl_ijkl & = & ( _ ai_bj-_aj_bi)_ck + & + & ( _ bi_cj-_bj_ci)_ak + & + & ( _ ci_aj-_cj_ai)_bk + & + & _ abij_ck+ _ bcij_ak+ _ caij_bk + & + & _ abjk_ci+ _ bcjk_ai+ _ cajk_bi + & + & _ abki_cj+ _ bcki_aj+ _ caki_bj [ phiphi ] m. j. duff , p. s. howe , t. inami and k. s. stelle , phys . lett .
b * 191 * ( 1987 ) 70 .
doi:10.1016/0370 - 2693(87)91323 - 2 t. kugo and p. k. townsend , nucl .
b * 221 * ( 1983 ) 357 .
doi:10.1016/0550 - 3213(83)90584 - 9 e. witten , nucl .
b * 443 * ( 1995 ) 85 doi:10.1016/0550 - 3213(95)00158-o [ hep - th/9503124 ] .
m. j. duff and j. x. lu , phys .
b * 273 * ( 1991 ) 409 .
doi:10.1016/0370 - 2693(91)90290 - 7 m. cvetic , h. lu and c. n. pope , nucl .
b * 644 * ( 2002 ) 65 doi:10.1016/s0550 - 3213(02)00792 - 7 [ hep - th/0203229 ]
. m. j. duff , int .
j. mod .
a * 11 * ( 1996 ) 5623 [ subnucl . ser .
* 34 * ( 1997 ) 324 ] [ nucl .
* 52 * ( 1997 ) no.1 - 2 , 314 ] doi:10.1142/s0217751x96002583 [ hep - th/9608117 ] .
t. banks , w. fischler , s. h. shenker and l. susskind , phys . rev .
d * 55 * ( 1997 ) 5112 doi:10.1103/physrevd.55.5112 [ hep - th/9610043 ] .
w. taylor , rev .
phys . * 73 * ( 2001 ) 419 doi:10.1103/revmodphys.73.419 [ hep - th/0101126 ] .
s. kovacs , y. sato and h. shimada , jhep * 1602 * ( 2016 ) 050 doi:10.1007/jhep02(2016)050 [ arxiv:1508.03367 [ hep - th ] ] .
d. berenstein , e. dzienkowski and r. lashof - regas , jhep * 1508 * ( 2015 ) 134 doi:10.1007/jhep08(2015)134 [ arxiv:1506.01722 [ hep - th ] ] .
e. g. floratos and a. kehagias , phys .
b * 427 * ( 1998 ) 283 doi:10.1016/s0370 - 2693(98)00340 - 2 [ hep - th/9802107 ] .
e. g. floratos and g. k. leontaris , nucl .
b * 512 * ( 1998 ) 445 doi:10.1016/s0550 - 3213(97)00775-x [ hep - th/9710064 ] .
m. yamazaki , phys .
b * 670 * ( 2008 ) 215 doi:10.1016/j.physletb.2008.11.001 [ arxiv:0809.1650 [ hep - th ] ] .
e. corrigan , c. devchand , d. b. fairlie and j. nuyts , nucl .
b * 214 * ( 1983 ) 452 .
doi:10.1016/0550 - 3213(83)90244 - 4 r. ivanov , hamiltonian formulation and integrability of a complex symmetric nonlinear system , phys
a350 , 232 - 235 ( 2006 ) joyce , dominic , compact manifolds with special holonomy , oxford university press on demand , 2000 .
+ joyce , dominic , the exceptional holonomy groups and calibrated geometry , math/0406011 _ lectures given at a conference in gokova , turkey , may 2004_. | we study the octonionic selfduality equations for @xmath0-branes in the light cone gauge and we construct explicitly , instanton solutions for spherical and toroidal topologies in various flat spacetime dimensions @xmath1 , extending previous results for @xmath2 membranes . assuming factorization of time we reduce the self - duality equations to integrable systems and we determine explicitly periodic , in euclidean time , solutions in terms of the elliptic functions .
these solutions describe 4d associative and non - associative calibrations in @xmath3 dimensions .
it turns out that for spherical topology the calibration is non compact while for the toroidal topology is compact .
we discuss possible applications of our results to the problem of 3-brane topology change and its implications for a non - perturbative definition of the 3-brane interactions .
cern - th-2017 - 045 + + emmanuel floratos @xmath4 and george k. leontaris@xmath5 + @xmath6 _ institute of nuclear physics , nrcs demokritos , _ _
athens , greece _
@xmath7 _ department of physics , university of athens , _ _ athens , greece _
@xmath8 _ theory department , cern , _ _
ch-1211 , geneva 23 , switzerland _
@xmath9 _ physics department , theory division , ioannina university , _ _ gr-45110 ioannina , greece _ |
why is it so plausible that galactic mergers and tidal interactions were more frequent in the past ?
several theoretical reasons come to mind : * hierarchical clustering , in which small objects are progressively incorporated into larger structures @xcite , is common to many accounts of galaxy formation . in the `` core - halo '' picture @xcite , clustering of dark matter creates galaxy halos which subsequently accumulate cores of baryons , forming visible galaxies . *
tidal encounters generate short - lived features ; a population of binary galaxies with highly eccentric orbits is required to explain the peculiar galaxies observed today @xcite . if these binaries have a flat distribution of binding energies , their merger rate has declined with time as @xmath2 , and the @xmath3 or so merging galaxies in the ngc catalog are but the most recent additions to a population of about @xmath4 remnants @xcite . * the cdm model @xcite provides a concrete example of galaxy formation in which merging of dark halos is easily calculated and clearly important @xcite .
observations , though not always reaching the redshift range emphasized in this meeting , also imply rapid merging at high redshift : * various counting strategies indicate that the pair density grows like @xmath5 , where @xmath6 @xcite . *
peculiar morphology becomes more common with increasing redshift @xcite .
for example , the fraction of irregular galaxies in the cfrs survey increases from about @xmath3% at @xmath7 to a third at @xmath8 @xcite .
thus , both theory and observation support the notion that there was `` a great deal of merging of sizable bits and pieces ( including quite a few lesser galaxies ) early in the career of every major galaxy '' @xcite .
but the _ nature _ of these early mergers is not so clear ; were the objects involved dominated by dark matter , by gas , or by stars ? and can we learn anything about early mergers by studying present - epoch examples ?
merging is hard to prove at redshifts @xmath9 ; cosmological dimming renders tidal tails nearly invisible , while bandshifting effects complicate interpretation of the observations @xcite . but circumstantial evidence implicates merging in various high-@xmath10 objects .
the most extensive and unbiased sample of high - redshift galaxies are the `` lyman - break '' objects at @xmath11 , which have rest - frame uv luminosities consistent with star formation rates of @xmath12 @xcite .
the actual rates could be several times higher , since much of the uv emitted by young stars may be absorbed by dust ( eg .
spectra show gas outflows with velocities of @xmath13 @xcite , atypical of quiescent galaxies but fairly normal for starburst systems .
heavily obscured high-@xmath10 starbursts have been detected at sub - mm wavelengths @xcite .
these have ir spectral energy distributions similar to ultra - luminous starburst galaxies like arp 220 and appear to be forming stars at rates of @xmath14 . at low redshifts ,
luminous starbursts are often triggered by mergers of gas - rich galaxies @xcite .
the gas in such systems is highly concentrated ; h@xmath15 surface densities of @xmath16 to @xmath17 are typical of nearby starbursts @xcite , and similar surface densities are indicated in high-@xmath10 starbursts @xcite . in the potential of an axisymmetric galaxy ,
gas becomes `` hung up '' in a disk several kpc in radius ( frenk , these proceedings ) instead of flowing inward .
violently changing potentials in merging galaxies enable gas to shed its angular momentum and collapse to as little as @xmath18% of its initial radius @xcite .
but models based on mergers of low-@xmath10 disk galaxies may not apply to high - redshift starbursts @xcite .
first , bar instabilities in isolated galaxies can drive rapid gas inflows without external triggers @xcite .
second , disks forming at higher redshifts are more compact @xcite and thus may already have the surface densities associated with starbursts .
third , the starbursts in lyman - break galaxies occur on scales of several kpc ( weedman , these proceedings ) , whereas inflows concentrate gas into much smaller regions .
nonetheless , these objects also have irregular morphologies suggestive of mergers , and deep hdf images reveal faint asymmetric features which may be due to tidal interactions @xcite .
mergers seem to be the `` best bet '' for high-@xmath10 starbursts , but something more than naive extrapolation from low-@xmath10 is needed to test this conjecture . at low redshifts ,
powerful radio sources are often associated with merger remnants ; some @xmath19% exhibit tails , fans , shells , or other signatures of recent collisions @xcite .
but at redshifts @xmath20 the most striking morphological feature of powerful radio sources is a near - ubiquitous alignment between the radio lobes and continuum optical emission @xcite .
this `` alignment effect '' seems at odds with the merger morphologies seen at low redshift ; one explanation invokes jet - induced star formation ( eg .
@xcite ) .
recent observations suggest the alignment effect is compatible with mergers @xcite .
strong polarization is found in several @xmath21 radio galaxies , implying that the aligned emission is scattered light from an obscured agn ( eg .
@xcite ) ; in several cases there is good evidence that dust is the primary scattering agent @xcite .
hst imaging of the radio galaxy 0406244 at @xmath22 reveals a double nucleus and what appear to be tidal debris illuminated by an agn @xcite . from a theoretical perspective , merging may even be _ necessary _ to form powerful radio sources .
the most plausible engines for such galaxies are rapidly spinning black holes ( blandford , these proceedings ) . accretion from a disk ca nt spin up a black hole unless the accretion phase lasts @xmath23 ; on the other hand , two black holes of comparable mass can coalesce to produce a rapidly - spinning hole @xcite .
evidence that low - redshift quasars frequently occur in interacting systems has been accumulating for two decades @xcite .
early claims that quasars have close companions are supported by recent studies out to redshifts @xmath24 @xcite .
even more telling are the tidal tails and other signs of violent interactions in nearby cases @xcite .
the very nature of these interactions makes their detection difficult at higher redshifts tidal tails and other signs are hidden by cosmological dimming and quasar glare . nor does the low-@xmath10 evidence preclude the possibility that high - redshift quasars may have nothing to do with mergers .
however , the peak in quasar activity at @xmath25 to @xmath26 seems to broadly coincide with other indications of extensive merging activity reviewed above . given the observational difficulties ,
a compelling case that this high-@xmath10 activity is driven by mergers probably awaits a theory for the formation of supermassive black holes .
complementing the data gathered by looking back to high redshift is information gleaned by `` archeological '' studies of objects at @xmath27 .
the oldest components of the milky way provide evidence that mergers of small galaxies played an important role @xcite : 1 .
a `` second parameter '' which may not be age @xcite is required to account for variations in the stellar content of globular clusters . 2 .
this second parameter is correlated with orbital direction ; clusters with retrograde orbits have oosterhoff class i variables @xcite .
3 . halo stars with @xmath28 \sim -1 $ ] have a large range of @xmath29 $ ] values @xcite .
the outer halo exhibits retrograde rotation with respect to the rest of the galaxy @xcite .
the halo is not completely well - mixed , as indicated by observations of star streams and moving groups @xcite .
items 13 indicate that different parts of the halo have different enrichment histories , items 2 & 4 imply that some part of the halo fell in on a retrograde orbit , and item 5 is direct evidence for the gradual dissolution of fragments after merging .
halo accretion is clearly an ongoing process , as shown by the discovery of the sgr i dwarf galaxy @xcite and by observations of high - latitude a stars @xcite .
but two different arguments suggest that the _ bulk _ of the halo fell into place long ago .
first , halo stars are old .
the halo as a whole shows a well - defined turn - off at @xmath30 , corresponding to ages @xmath31 ; only @xmath32% of the stars appear younger @xcite . to be sure ,
this does not rule out recent accretions of objects containing only old stars , but most dwarf galaxies in the local group contain intermediate - age stars as well . thus ,
unless the accreted galaxies were unlike those we observe today , most fell in more than @xmath33 ago .
second , galactic disks are dynamically fragile ; accretion of satellite galaxies can easily ruin a stellar disk .
analytic estimates limit the mass accreted by the milky way to less than @xmath34% in the past @xmath35 @xcite .
n - body experiments show less disk heating than the analytic work predicts ; dark halos absorb much of the damage , and disks may tilt as well as thicken @xcite . still , accretion events of any size increase the disk s vertical dispersion , @xmath36 .
significant structure is seen in the @xmath37age relation ; most striking is the jump from @xmath38 to @xmath39 which marks the transition to the @xmath40-old thick disk @xcite . in sum , the milky way last suffered a significant merger at least @xmath41 ago ;
relics of this event include the outer stellar halo and possibly the thick disk .
presumably , the milky way s dark halo was largely in place at this time , since a major merger would have disrupted even the thick disk .
galaxy clusters are old in two distinct respects : first , cluster galaxies probably collapsed early ; second , dynamical processes run faster in proportion to @xmath42 .
thus clusters should contain remnants of many high - redshift mergers .
archeological evidence from nearby clusters provides important clues to these mergers .
after some controversy , it s generally accepted that elliptical galaxies can be formed by fairly _
recent _ mergers of disk galaxies . support for
this position includes : * studies of proto - elliptical merger remnants like ngc 7252 @xcite and models of disk galaxy mergers reproducing such objects @xcite .
* h@xmath43 line strengths in some ellipticals indicating recent star formation @xcite .
* `` fine structures '' in elliptical galaxies correlating with residuals in luminosity color and luminosity line strength relations @xcite .
these results enable us to trace the gradual assimilation of recent merger remnants into the larger population of field ellipticals .
but such evidence is not available for cluster ellipticals , which seem to be a more homogeneous population ( eg .
studies of the fundamental plane out to @xmath44 indicate that cluster ellipticals evolve passively and probably formed the bulk of their stars at @xmath45 @xcite .
thus cluster ellipticals are unlikely to show the signs which betray aging merger remnants in the field .
counter - rotating or otherwise decoupled `` cores '' are probably the clearest signs that cluster ellipticals were formed by ancient mergers @xcite .
high - resolution imaging shows that kinematically distinct nuclear components are usually _ disks _ @xcite .
such disks typically have high metal abundances @xcite and low velocity dispersions @xcite .
these properties indicate that they formed dissipationally during major mergers @xcite ; merger simulations producing counter - rotating nuclear gas disks back up this hypothesis @xcite .
the nature of the mergers which formed cluster ellipticals is unknown ; often invoked are highly dissipative encounters of gaseous fragments . but
the existence of counter - rotating disks indicates that the penultimate participants ca nt have been very numerous or very gassy .
if many small objects coalesced , the law of averages would make counter - rotation extremely rare . and
counter - rotation is unlikely to arise in essentially gaseous mergers since gas flows ca nt interpenetrate .
once formed , kinematically distinct disks would be easily disrupted by dissipationless mergers @xcite .
thus observations of such structures in cluster galaxies imply that few mergers occur once a cluster has virialized . this is entirely plausible on dynamical grounds since encounters at speeds higher than about twice a galaxy s internal velocity dispersion do nt result in mergers @xcite . in elliptical galaxies ,
@xmath46-process elements are more abundant with respect to fe than they are in the disk of the milky way @xcite .
this may constrain the timescale for star formation in ellipticals , since @xmath46-process elements are produced in sn ii , which explode on a short timescale , while fe is also produced in sn ia , which explode after @xmath47 .
indeed , @xmath48 \simeq 0.5 $ ] for the nuclear disks in cluster ellipticals @xcite .
high @xmath46-process abundances indicate that sn ia played little role in enriching these galaxies ; on the face of it , they also imply that cluster ellipticals formed on timescales @xmath49 ( eg .
@xcite ) .
high abundances of @xmath46-process elements with respect to fe are also seen in x - ray observations of the diffuse gas in galaxy clusters ( eg . @xcite , but see @xcite ) .
the large amounts of metals in cluster gas require remarkably high sn rates which may not be possible with a salpeter imf @xcite .
these results undermine the argument that high @xmath50 ratios imply short enrichment timescales , since abundances in the cluster gas presumably represent integrated metal production over @xmath40 .
the abundance patterns of cluster ellipticals are clearly inconsistent with mergers of present - day spirals , but do not preclude mergers of moderately gas - rich galaxies containing substantial stellar disks .
young star clusters are observed in star - forming galaxies like the lmc @xcite and in intense starburst galaxies @xcite .
these clusters have half - light radii of less than @xmath51 , masses of @xmath52 to @xmath53 , and metal abundances comparable to their parent starbursts .
their luminosity functions follow power laws with slopes of @xmath54 to @xmath55 , intriguingly close to the mass function of giant molecular clouds @xcite .
however , it s not entirely clear that cluster luminosity is a good indicator of mass since some range of cluster ages is usually present .
evidence is accumulating that the globular cluster systems of field ellipticals are partly due to cluster formation in merger - induced starbursts : * ongoing and recent mergers ( eg . , ngc 4038/9 , ngc 7252 , ngc 3921 ) have populations of blue luminous clusters with ages of less than @xmath56 @xcite . * older remnants ( eg . , ngc 3610 ) have redder and fainter clusters with ages of a few gyr @xcite .
* predicted specific frequencies is defined as the number of globular clusters divided by the galaxy luminosity in units of @xmath57 . ] in merger remnants increase to @xmath58 or @xmath26 over @xmath40 as the stellar populations fade @xcite . *
globulars in elliptical galaxies have bimodal color ( metalicity ) distributions .
these findings imply that metal - rich star clusters form during mergers and are gradually assimilated into existing globular cluster populations @xcite .
however , the large populations of metal-_poor _ globulars found in cluster ellipticals are _ not _ consistent with mergers of field spirals @xcite ; predicted specific frequencies of metal - poor clusters are @xmath59 , while in fact @xmath60 .
this problem is even worse for cluster systems in cd galaxies , which have @xmath61 ; obviously , no amount of merging between metal - rich systems will produce metal - poor clusters !
the question of high-@xmath62 in cluster ellipticals boils down to this : fewer stars , or more globulars ?
one way to get fewer stars is to merge galaxies _ after _ their metal - poor globulars have formed but before they build up substantial disks .
for example , the milky way as it was @xmath40 ago could serve as a building - block for cluster ellipticals ; the halo of our galaxy , considered alone , has @xmath60 .
however , mergers of milky way halos ( or dwarf elliptical galaxies @xcite ) still fall short of the high @xmath63 values of cd galaxies . another way to end up with fewer stars is to eject most of the gas after the initial epoch of cluster formation ; the problem here is that the ejection efficiency must be _ higher _ in cd galaxies , which have the deeper potential wells and should be better at retaining gas @xcite . alternately , the production of globular clusters may have been more efficient in high - redshift starbursts . even at low-@xmath10 , about @xmath64% of the uv emitted by starbursts comes from knots identified with young clusters @xcite ; if all these clusters survive , the specific frequency for a pure starburst population is @xmath65 .
moreover , these clusters are concentrated where the surface densities are highest ; it s likely that net yields of star clusters increase rapidly with increasing surface density .
if so , then globular cluster systems reflect the starburst histories of their parent galaxies : large populations of metal - poor globulars are due to efficient cluster production in early starbursts , while predominantly metal - rich systems ( eg . , ngc 5846 ) formed in more recent starbursts .
metallicity distributions for cluster systems support this idea ; giant elliptical galaxies have a range of distributions with multiple peaks between @xmath66 \simeq -1.2 $ ] and @xmath67 @xcite .
such variety seems hard to explain in a picture where internal events determine the timing of cluster formation ( eg . ,
@xcite ) ; on the other hand , it s easy to imagine that different distributions result from the different merging histories of individual galaxies .
circumstantial evidence suggests that merging played an important role in galactic evolution long before the present epoch .
the key points of the argument can be summed up as follows : 1 .
starbursts and agn are signposts of high - redshift mergers ; the high incidence of such objects at @xmath68 to @xmath34 reflect frequent merging of juvenile galaxies . 2 .
the bulk of the milky way s halo merged more than @xmath33 ago as part of this activity .
cluster ellipticals merged before @xmath68 ; their immediate progenitors were few and only moderately gassy .
the metal - rich globular cluster systems of these ellipticals are relics of their final mergers . finally , direct observations of high - redshift events are complemented by archeological investigation of nearby systems . both approaches
are needed to discover what happened at redshifts @xmath0 to @xmath1 .
i thank alex stephens and hector velzquez for communicating results in advance of publication .
i also thank jun makino and the university of tokyo for hospitality while i prepared this article .
this research made use of nasa s astrophysics data system abstract service .
travel to the conference was covered by air miles accumulated while following the grateful dead .
hughes , d.h . , serjeant , s. , dunlop , j. , rowan - robinson , m. , blain , a. , mann , r.g . ,
ivison , r. , peacock , j. , efstathiou , a. , gear , w. , oliver , s. , lawrence , a. , longair , m. , goldschmidt , p. , & jenness , t. 1998 , nature , 394 , 241 | theoretical considerations and observational data support the idea that mergers were more frequent in the past . at redshifts
@xmath0 to @xmath1 , violent interactions and mergers may be implicated by observations of lyman - break galaxies , sub - mm starbursts , and active galactic nuclei .
most stars in cluster ellipticals probably formed at such redshifts , as did most of the halo and globular clusters of the milky way ; these events may all be connected with mergers .
but what _ kind _ of galaxies merged at high redshifts , and are present - epoch mergers useful guides to these early collisions ?
i will approach these questions by describing ideas for the formation of the milky way , elliptical galaxies , and systems of globular clusters . |
the minimal supersymmetric standard model ( mssm ) @xcite implies the existence of five physical higgs bosons @xmath11 ,
@xmath1 , @xmath2 , and @xmath12 @xcite . for the search of these particles a precise knowledge of all possible decay modes
is necessary .
the higgs boson decays to supersymmetric ( susy ) particles could be very important if they are kinematically allowed .
this can be the case for the charged higgs boson @xmath0 , and the neutral higgs bosons @xmath1 and @xmath2 .
if all susy particles are very heavy , the @xmath0 decays dominantly into @xmath13 ; the decays @xmath14 and/or @xmath15 are dominant below the @xmath13 threshold @xcite .
if all decay modes into susy particles are kinematically forbidden the @xmath1 and @xmath2 decay dominantly into a fermion pair of the third generation . in the case that the susy particles are relatively light , the branching ratios of the @xmath0 , @xmath1 and @xmath2 decays at tree level were studied in @xcite and @xcite . the susy
qcd corrections in @xmath16 were calculated in the on shell scheme for the processes @xmath17 in @xcite , @xmath18 in @xcite , and for the decays of all higgs particles into squark pairs in @xcite , including squark mixing and a proper renormalization of the mixing angle @xmath19 @xcite . + within this work we will discuss the branching ratios of the higgs decays including all susy qcd corrections in @xmath16 .
we will see that the decay modes into susy particles ( squarks of the third generation , charginos and neutralinos ) become more important when the qcd corrections are taken into account .
we first review some tree level results @xcite .
the squark mass matrix in the basis ( @xmath20 , @xmath21 ) , with @xmath22 or @xmath23 , is given by @xcite @xmath24 where @xmath25 and @xmath26 the mass eigenstates @xmath27 ( with @xmath28 ) are related to the su(2)@xmath29 eigenstates @xmath30 by @xmath31 .
level decay width of @xmath32 is then given by @xmath33 for @xmath34 @xmath35 denotes the neutral higgs bosons ( i. e. @xmath36 , @xmath37 , @xmath38 ) and @xmath39 . for @xmath40 one
has @xmath41 and @xmath42 , @xmath43 , @xmath44 , and @xmath45 .
the expressions for the couplings @xmath46 are given in @xcite .
the decay widths of @xmath0 and @xmath1 into squarks can be large in the case of large squark mixing .
the decay width of @xmath2 into @xmath47 is directly proportional to @xmath48 with @xmath49 and @xmath50 .
starting from the threshold these widths are steeply increasing with increasing @xmath51 up to a maximum and then decreasing . for large @xmath51
they become proportional to @xmath52 . + the decay widths into quarks are given by @xmath53\ , , \label{eberl : h+quarks } \end{aligned}\ ] ] with @xmath54 , @xmath55 , @xmath56 , @xmath57 , @xmath58 being the @xmath11 @xmath1 mixing angle , and the yukawa couplings @xmath59 and @xmath60 : @xmath61 for large @xmath5 @xmath62 can also become large and , therefore , the sbottom and bottom modes become important .
the formulae for the decay widths into charginos and neutralinos have essentially the same structure as eq .
( [ eberl : h+quarks ] ) with the appropriate masses and couplings ( without the color factor 3 ) , see e. g. @xcite . if the mass of the decaying higgs particle is large , the decay widths into fermions ( quarks , charginos / neutralinos ,
) become proportional to @xmath51 .
+ in the chargino / neutralino sector one has quite generally the following behaviour : + @xmath63 : @xmath64 @xmath65 is gaugino
like @xmath66 @xmath67 is small , + @xmath68 : @xmath64 @xmath65 is higgsino
like @xmath66 @xmath67 is large .
+ we have calculated the widths of all important modes of @xmath0 , @xmath1 and @xmath2 decays : + [ cols="<,>,^ , < " , ] formulae for these widths are found e. g. in ref.@xcite . in principle , also the decays @xmath69 and @xmath70 ( @xmath71 and @xmath72 ) could contribute via their gauge couplings . as the squarks of the first two generations are supposed to be heavy ,
these decays will be strongly phase - space suppressed .
even if they were kinematically allowed , they would have a rate at most comparable to that of @xmath73 , @xmath1 @xmath74 @xmath75 and @xmath76 ( see fig .
[ eberl : figbranch ] below ) .
we have neglected loop induced decay modes ( such as @xmath77 , @xmath78 , and @xmath79 ) and three - body decay modes @xcite .
in the following branching ratios including susy qcd corrections in @xmath16 will be shown .
for further details concerning the theoretical calculation of these corrections we refer to @xcite .
+ we have chosen \{@xmath80 , @xmath81 , @xmath82 , @xmath83 , @xmath84 , @xmath5 , @xmath85 , @xmath86 } as the basic input parameters of the mssm , taking @xmath87 , @xmath88 and @xmath89 . here
@xmath83 ( @xmath90 ) is the su(2 ) ( u(1 ) ) gaugino mass , @xmath91 , and ( @xmath92 , @xmath93 ) are the mass matrix parameters of the slepton sector @xcite .
we have taken @xmath94 gev , @xmath95 gev , @xmath96 gev , @xmath97 gev , @xmath98 , @xmath99 , and @xmath100 for the @xmath35 decay .
we have used @xmath101 , with @xmath102 , and the number of quark flavors @xmath103 for @xmath104 ( for @xmath105 ) .
+ we have considered two scenarios : we have implemented the new higgs mass bound from aleph @xcite , @xmath107 gev for @xmath108 .
this leads for the parameter set i to @xmath109 gev and for the parameter set ii to @xmath110 gev . + for the parameter set i we have ( in gev units ) + @xmath111 = ( 96 , 255 , 100 , 130 , 465 , 73 , 137 ) , + and for the parameter set ii + @xmath111 = ( 100 , 257 , 113 , 137 , 820 , 92 , 108 ) . + fig .
[ eberl : figqcd ] ( a ) shows the tree - level and the susy
qcd corrected decay widths @xmath112 and @xmath113 and fig .
[ eberl : figqcd ] ( b ) shows the tree
level and the susy
qcd corrected decay widths @xmath114 and @xmath115 using the parameter set i. the modes into bottom quarks and sbottoms are very small compared to the top and stop modes in fig .
[ eberl : figqcd ] ( b ) and therefore not shown .
[ eberl : figbranch ] a c show susy qcd corrected branching ratios larger than 1% for the parameter set i and fig .
[ eberl : figbranch ] d
f for the parameter set ii .
all susy modes are summed up , e. g. in the @xmath1 decay into @xmath116 .
+ in most cases , the susy
qcd corrections to the higgs decays into quarks are negative and into squarks positive .
therefore , the branching ratios for decays into squarks are enhanced by including the susy
qcd corrections .
+ * acknowledgements * + this work was supported by the `` fonds zur frderung der wissenschaftlichen forschung '' of austria , project no .
p10843phy .
99 h. e. haber and g. l. kane , phys .
( 1985 ) 75 .
j. f. gunion , h. e. haber , g. l. kane , and s. dawson , the higgs hunter s guide , addison - wesley ( 1990 ) j. f. gunion and h. e. haber , nucl .
( 1986 ) 1 ; ( 1993 ) 567 ( e ) .
z. kunszt and f. zwirner , nucl .
( 1992 ) 3 .
a. bartl , k. hidaka , y. kizukuri , t. kon and w. majerotto , phys .
( 1993 ) 360 .
a. bartl , h. eberl , k. hidaka , t. kon , w. majerotto , and y. yamada , phys .
( 1996 ) 538 .
a. bartl , h. eberl , k. hidaka , t. kon , w. majerotto , and y. yamada , phys .
( 1996 ) 167 , and references therein .
r. a. jimnez and j. sol , phys .
( 1996 ) 53 .
j. a. coarasa , r. a. jimnez , and j. sol , phys .
( 1996 ) 312 .
a. bartl , h. eberl , k. hidaka , t. kon , w. majerotto , and y. yamada , hep
ph/9701398 , to be published in phys .
b , and references therein .
h. eberl , a. bartl , and w. majerotto , nucl .
( 1996 ) 481 .
a. djouadi , j. kalinowski and p. m. zerwas , z. phys .
( 1996 ) 435 .
j. f. gunion , g. l. kane and j. wudka , nucl .
( 1988 ) 231 ; + m. c. peyranre , h. e. haber and p. irulegui , phys .
( 1991 ) 191 .
+ a. mndez and a. pomarol , nucl .
( 1991 ) 369 .
aleph collaboration , preprint cern
ppe/97071 , subm .
to phys .
lett . b. | we present a detailed study of the decays of the higgs bosons @xmath0 , @xmath1 , and @xmath2 within the minimal supersymmetric standard model including susy qcd corrections .
we find that the supersymmetric modes @xmath3 ( @xmath4 , and for large @xmath5 @xmath6 ) can dominate the @xmath0 ( @xmath1 , @xmath2 ) decays in a wide range of the model parameters due to the large yukawa couplings and mixings of @xmath7 and @xmath8 .
compared to the conventional modes @xmath9 , and @xmath10 , the supersymmetric modes can have an important impact on the higgs boson searches at future colliders .
= -1 cm = 13.9 cm 0.0 in 0.0 in _ _
hep - ph/9709253 * higgs particle decays in supersymmetry * + |
stars with masses @xmath4 formed in the early universe may survive to the present day .
such old low - mass stars should typically have very low metallicites because they were formed from a gas that had not been significantly enriched with heavy elements .
they carry invaluable information on the early chemical evolution of the galaxy @xcite .
recently , @xcite discovered a low - mass and extremely metal - deficient star in the halo of the galaxy .
the star has extremely small abundances of carbon and nitrogen .
the upper limit of corresponding metallicity of the star is estimated to be @xmath5 @xcite .
furthermore , the ratio of the abundances of carbon and oxygen derived from the spectra of the star suggests that the parent cloud from which the star was born had been chemically enriched by core - collapse supernovae @xcite .
it is generally thought that metal - free primordial stars are predominantly massive ( _ e.g. _ , * ? ? ?
* ) although not extremely massive @xcite .
this is largely because metal - free gas lacks efficient coolants other than hydrogen molecules ; a primordial gas cloud is thermally stable throughout its pre - stellar collapse @xcite .
theoretical studies suggest that a trace amount of heavy elements provide efficient cooling in low - metallicity gas , which may then enable the formation of low - mass stars ( see * ? ? ? * for a review ) .
there is another argument that the existence of dust grains in low - metallicity gas is crucial for the low - mass star formation @xcite .
although it is still unclear whether or not gas metallicity is the key quantity to determine the characteristic stellar mass , and whether or not there is `` critical metallicity '' for low - mass star formation , there must be a mechanism for the formation of low - mass stars in extremely low - metallicity gas ] .
sn explosions of the first generation stars are thought to have enriched the early universe with the first heavy elements .
three - dimensional simulations of early sn explosions are performed by , _
e.g. _ , @xcite and @xcite .
these studies show heavy elements are effectively dispersed into the inter - galactic medium .
@xcite argue that ultraviolet radiation from a massive progenitor star modifies the density structure of the ambient gas before sn explosion .
overall , an early sn explosion can trigger the formation of low - mass and low - metallicity stars .
@xcite study the evolution of sn remnants with metallicities of @xmath6 and @xmath7 .
the sn shell becomes gravitationally unstable for a certain range of parameters such as explosion energy , ambient gas density , and metallicity .
they show that the typical mass scale of the fragments is @xmath8@xmath9 and argue that a low - mass star forms after the further evolution of the fragment .
@xcite semi - analytically calculate the evolution of sn shells before and after the shell fragmentation in a primordial gas .
they show that the jeans mass of the fragment is below the solar mass when the fragment becomes optically thick . however , it remains uncertain whether the gas cloud further fragment or not . in this paper , we study sn - triggered star formation , using hydrodynamic simulations .
we first run one - dimensional hydrodynamic simulations to follow the thermal evolution of an early sn remnant .
we here assume the ambient gas around the sn has nearly the primordial composition with a trace amount of heavy elements , @xmath10 and @xmath11 , presumably ejected by earlier generation stars .
we determine whether or not the shocked sn shell is gravitationally unstable to fragment by using linear analysis of density perturbations .
we show that gravitational instability is triggered in the shell for a wide range of the supernova explosion energies , @xmath12@xmath13 erg , and for large ambient gas density , @xmath14 .
finally , we study the subsequent evolution of the collapsing fragments , using a one - zone code .
while the gas cloud continues collapsing gravitationally , it cools rapidly by dust thermal emission and eventually fragments to sub - solar mass clumps .
we perform one - dimensional hydrodynamic calculations to follow the dynamics of an sn remnant until the shocked gas shell undergoes gravitational instability ( hereafter , we call this phase _ ante - fragmentation phase _ ) .
we use the code of @xcite and @xcite which employs the second - order lagrangian finite - difference scheme in spherically symmetric geometries .
the basic equations are presented in @xmath152 of @xcite .
we implement the relevant radiative cooling processes to the hydrodynamical code as follows . in the high temperature regime ( @xmath16 ) , assuming that the gas is in collisional ionization equilibrium , we adopt the cooling function of @xcite .
the cooling functions for @xmath17 are derived by interpolating the cooling functions for @xmath18 and @xmath19 .
the cooling rate due to inverse compton scattering is also included . in the lower temperature regime ( @xmath20 ) , we solve the non - equilibrium chemical reactions and radiative processes of the following fifteen species : @xmath21 , h , h@xmath22 , h@xmath23 , d , d@xmath22 , d@xmath23 , he , he@xmath22 , he@xmath23 , h@xmath24 , h@xmath25 , hd , hd@xmath22 , and heh@xmath22 .
the molecular cooling rates are taken from @xcite for h@xmath24 , @xcite and from @xcite for hd .
we also calculate the fine - structure transition line cooling of heavy elements including c , o , si , and fe .
these species are assumed to be in the form of , , , and , respectively .
the fraction of each species in the gas is derived by multiplying the solar abundance by metallicity ( @xmath26 ) .
the solar abundances of c , o , si , and fe relative to hydrogen are @xmath27 , @xmath28 , @xmath29 , and @xmath30 , respectively @xcite throughout the one - dimensional calculation .
also , the atomic level transition rates are taken from @xcite .
after the supernova shell becomes gravitationally unstable ( see [ subsec : criterion ] below ) , fragments in a shell are expected to collapse further to trigger star formation .
we perform one - zone calculations to follow the thermal and chemical evolution of one of such collapsing gas clouds .
we assume that the collapse occurs approximately in a spherically symmetric manner .
then , the time evolution of the gas density , @xmath31 , is described as @xmath32 in this equation , @xmath33 denotes the free - fall time as @xmath34 where @xmath35 is the newtonian gravitational constant .
the evolution of specific internal energy , @xmath36 , is @xmath37 where @xmath38 denotes the pressure and @xmath39 and @xmath40 denote heating and cooling rate per unit volume , respectively .
the temperature of the cloud , @xmath41 , is related to the internal energy as @xmath42 where @xmath43 is the mean molecular weight , @xmath44 is the mass of hydrogen nuclei . in equation ( [ eq : u2 t ] ) , @xmath45 is the adiabatic index which is calculated by taking average over all species as _
e.g. _ , @xcite . in our one - zone model , the low - metallicity gas
is assumed to consist of the following chemical species : @xmath21 , h , h@xmath22 , h@xmath23 , d , d@xmath22 , d@xmath23 , he , he@xmath22 , he@xmath23 , c , c@xmath22 , o , o@xmath22 , h@xmath24 , h@xmath46 , h@xmath47 , hd , hd@xmath22 , hd@xmath23 , oh , h@xmath24o , h@xmath24o@xmath22 , o@xmath24 , o@xmath25 , co , co@xmath22 , co@xmath24 , and co@xmath25 .
we assume that 72 % of carbon and 46 % of oxygen are condensed to dust grains as in the local interstellar matter @xcite .
we include atomic cooling by and but neglect the contributions from less abundant and because these species do not have effect on the thermal evolution of the gas cloud @xcite .
the adiabatic compressional heating and the chemical heating by h@xmath24 formation are the dominant heating processes .
we also consider the dust cooling in the calculations .
the reaction rates and corresponding cooling rates are taken from @xcite .
we use the sobolev length approximation for optically thick cooling as in @xcite and @xcite .
we set metallicities of the circumstellar medium @xmath48 conservatively and @xmath5 , the upper bound of the metallicity of the star @xcite observed .
we assume that the progenitor star is hosted by an early small dark matter halo with mass @xmath49 at a redshift of @xmath50 @xcite .
the gravitational potential of the dark matter is fixed as in @xcite .
we work with the @xmath40cdm model with the matter density @xmath51 , the baryon density @xmath52 , the cosmological constant @xmath53 , the hubble constant @xmath54 , in units of @xmath55 , and the absolute density fluctuation @xmath56 consistent with wmap 7 data @xcite .
the density of the circumstellar medium at the time when sn explosions occur depends on a number of factors such as the mass of the hosting halo @xcite and the ultraviolet luminosity of the progenitor star @xcite .
thus , for simplicity , we assume that the surrounding gas density is initially uniform for our one - dimensional hydrodynamic simulations and the initial density is set @xmath571 , 10 , 100 , and 1000 @xmath58 .
we assume that the surrounding gas with primordial composition is pre - enriched with a trace amount of heavy elements .
thus , through this paper , the hydrogen mass fraction is @xmath59 , the helium mass fraction is @xmath60 @xcite , and the deuterium number fraction is @xmath61 @xcite .
we set the initial gas temperature @xmath62 , and initial ionization fraction of hydrogen is uniformly @xmath63 .
the fraction of hydrogen molecule is @xmath64 at the beginning k by radiative cooling even if we start the simulation with the temperature of @xmath65 k. ] .
the radial velocity is set @xmath66 .
we test three explosion energies , @xmath67 @xmath68 , @xmath69 , and @xmath70 erg .
we assume a point - source explosion and at the time @xmath71 , the corresponding thermal energy is deposited in the innermost cell .
figure [ fig : prof_l ] shows the structures of the shell for @xmath72 at the time ( 1 ) @xmath73 , ( 2 ) @xmath74 , ( 3 ) @xmath75 , and ( 4 ) @xmath76 yr .
figure [ fig : nt1d_l ] illustrates the evolution of the gas density and temperature for a fluid element that is densest when the shell becomes gravitationally unstable and that is initially located at 20.8 pc from the center .
the evolution of the sn shell structure is described as follows . in the initial sedov - taylor phase ,
the sn remnant expands adiabatically as long as the radiative cooling time , @xmath77 , is longer than the shell expansion time scale , @xmath78 @xcite .
then , radiative cooling becomes effective and the gas just behind the shock starts to lose its thermal energy . at @xmath79 yr after the sn explosion , the cold shell is pushed by the large pressure of the inner hot gas , forming a very dense shell ( the pressure - driven snowplough phase ) . during the pressure - driven phase ,
the fluid element plotted in figure [ fig : nt1d_l ] enters the shock front , increasing its density and temperature rapidly .
the inner hot region still impacts the cold shell , generating multiple weak shocks behind the shell .
we see this as oscillatory features from epoch ( 3 ) to epoch ( 4 ) in figure [ fig : nt1d_l ] . finally , at epoch ( 4 ) , the shell is decelerated sufficiently to be gravitationally unstable as we shall discuss in [ subsec : criterion ] . figure [ fig : prof_s ] shows the structures of the shell for @xmath80 at the time ( 1 ) @xmath81 , ( 2 ) @xmath82 , ( 3 ) @xmath83 , and ( 4 ) @xmath84 yr .
figure [ fig : nt1d_s ] illustrates the evolution of the gas density and temperature for a fluid element initially located at 34.3 pc from the center . in this model ,
the shell is in the sedov - taylor phase around epoch ( 1 ) and then enters the pressure - driven snowplough phase at @xmath85 yr . in this case with low ambient gas density , the transition from the sedov - taylor phase to the pressure - driven phase occurs later than the model with high ambient gas density .
when the shell sweeps up denser ambient medium , radiative cooling of the shell , which is generally proportional to the square of gas density , becomes larger , and the enhanced radiative cooling makes the shell evolve faster . the shell continues to expand even after the pressure inside the shell declines due to radiative cooling , conserving the momentum ( the momentum - conserving snowplough phase ) . in our calculation , the shell enters the momentum - conserving phase by epoch ( 3 ) .
the shell thickness increases because of the pressure in the shell .
the internal energy of the shell remains higher by less effective radiative cooling . around epoch
( 4 ) , the ellapsed time reaches the dynamical timescale for the dark matter halo hosting the progenitor star ( @xmath86 yr for @xmath87 halo at @xmath88 ) .
there are several criteria proposed for the instability of an expanding and decelerating gas shell .
@xcite derive approximate conditions for the collapse of an expanding and decelerating shell by linear analysis of density perturbations under the thin - shell approximation .
then , @xcite and @xcite include the effect of the internal structure of the shell to the result by @xcite . for a shell with average column density over the all solid angle , @xmath89 , shell radius , @xmath90 , and velocity @xmath91 ,
the angular wavenumber of the perturbations that grow most rapidly is @xmath92 and the growth rate of the perturbations with wavenumber @xmath93 is @xmath94 ^{1/2}. \label{eq : o_peak}\ ] ] in equations ( [ eq : eta_peak ] ) and ( [ eq : o_peak ] ) , @xmath95 is the effective sound speed as @xmath96 ^{1/2 } , \label{eq : c_eff}\ ] ] where @xmath97 is the peak density in the shell and @xmath98 is the sound speed for an isothermal shell with temperature @xmath41 .
the numerical factor @xmath99 is determined as 0.39 by the detailed linear analysis for a shell with its outer boundary defined by the shock front and with its inner boundary defined by the stellar wind from a central star @xcite .
note that replacing @xmath95 with @xmath100 and @xmath89 with @xmath101 , where @xmath102 is the average pre - shell mass density , in equations ( [ eq : eta_peak ] ) and ( [ eq : o_peak ] ) recovers the result of @xcite .
in equation ( [ eq : o_peak ] ) , the first and second terms in the right hand side represent the effect of the shell expansion and the last term represents the effect of the shell internal pressure and self - gravity .
if the peak growth rate , @xmath103 , is positive , perturbations with the corresponding wavenumber have a growing solution , which means the sn - driven shell becomes gravitationally unstable and fragments with scales corresponding to @xmath93 start to grow in the shell .
it is obvious that fast expansion ( large @xmath91 ) tends to smooth out perturbations .
for the gas to become gravitationally unstable , its expansion velocity must be sufficiently small such that @xmath104 in this section , we discuss whether the gravitational instability can be triggered or not .
we apply the criterion by @xcite to our calculations because the internal structure of a shell is taken into account in their analysis . in order to investigate the stability of the shell ,
we calculate the time evolution of the peak growth rate , @xmath103 [ equation ( [ eq : o_peak ] ) ] by using the radius of the shock front , the shock velocity , volume - weighted mean density , mass - weighted mean temperature , and mean molecular weight in the shell . in the snowplough phase ,
the shell is defined by the outer boundary ( shock front ) and the inner boundary .
the outer boundary is determined as position where the density is 1.1 @xmath105 .
the displacement of the inner boundary is defined as the point where density recovers the pre - shell density , @xmath105 , inside the shock radius .
after the shell becomes gravitationally unstable , the shell is assumed to be separated into fragments .
if the time for fragmentation is comparable to or larger than the dynamical timescale of the hosting halo , @xmath106 , we do not count such cases as successful fragmentation of the sn shell in the cosmological context .
after the dynamical timescale , it is likely that the halo itself collapses or is incorporated into a large halo , thereby the structure of the entire system would be much different .
applying the fragmentation condition , we find that for the sn model of @xmath107 , the compressed and decelerated shell becomes gravitationally unstable at @xmath108 yr .
by contrast , for @xmath109 , the peak growth rate @xmath103 becomes positive at @xmath110 yr when the shell is about to fall back ( @xmath111 ) by the gravitational potential of the host halo . in this model , since the time @xmath112 is comparable to the dynamical timescale , it is uncertain whether the shell fragments or collapses into the center of the halo .
table [ tab : frag ] shows the physical properties of the shell at the time @xmath113 for sn models we study .
the results are shown only for @xmath114 because they are almost exactly the same as the results of the @xmath115 models .
it is because the fine - structure cooling of metals such as c and o has little effect on the evolution of the sn remnant in such an extremely metal - poor regime until the shell becomes gravitationally unstable .
also , the time @xmath113 is insensitive to the sn explosion energy . for a radiative remnant , the shock radius increases only as @xmath116 in the pressure - driven phase @xcite .
the time @xmath113 depends on the ambient gas density . as the ambient gas density is larger , denser and colder shell is formed as we discuss in [ subsubsec : s ] and the growth rate of the perturbations on the shell , @xmath117 becomes positive earlier .
the jeans mass ( typical mass ) of the fragment is shown in the last column of table [ tab : frag ] .
the mass decreases as ambient gas density increases because the density of the shell swept up by the blast wave becomes larger .
figure [ fig : ox ] shows the results of all the models that we investigated in a ( @xmath118 ) plane .
these results are essentially the same for the respective model with @xmath119 .
the circles indicate models where both our criteria are met . for models with ambient density @xmath120 , the time for the gravitational instability , @xmath113 , is comparable to @xmath106 .
these models are marked by triangles .
we can not determine from our 1d calculations whether the shell fragments before the host halo evolves significantly .
it would be needed to run a cosmological simulation that incorporate the hierarchical growth of dark halos , in order to follow the evolution of such systems properly .
rrrrrrrrrr 1 & 1.0 & 10.23 & 2.9e+00 & 568.6 & 115.9 & 22.84 & 3.0 & 3.0e+05 & 4.8e+05 + & 10.0 & 10.84 & 6.7e+00 & 466.8 & 196.6 & 12.89 & 8.7 & 1.3e+06 & 2.4e+05
+ & 30.0 & 10.35 & 1.1e+01 & 419.7 & 250.9 & 9.17 & 14.4 & 2.6e+06 & 1.6e+05 + 10 & 1.0 & 3.39 & 5.9e+01 & 300.2 & 45.6 & 3.53 & 6.9 & 1.7e+05 & 4.1e+04 + & 10.0 & 2.99 & 1.7e+02 & 239.8 & 75.2 & 1.72 & 21.5 & 6.7e+05 & 1.7e+04 + & 30.0 & 2.63 & 2.4e+02 & 215.2 & 93.9 & 1.42 & 36.5 & 1.3e+06 & 1.2e+04 + 100 & 1.0 & 1.06 & 1.1e+03 & 191.6 & 17.9 & 0.68 & 14.2 & 9.3e+04 & 4.9e+03 + & 10.0 & 0.83 & 2.4e+03 & 182.8 & 28.5 & 0.43 & 38.5 & 3.5e+05 & 3.0e+03 + & 30.0 & 0.72 & 3.8e+03 & 186.4 & 35.2 & 0.33 & 58.4 & 6.4e+05 & 2.5e+03 + 1000 & 1.0 & 0.32 & 1.4e+04 & 202.5 & 6.9 & 0.19 & 19.5 & 5.1e+04 & 1.5e+03 + & 10.0 & 0.25 & 3.1e+04 & 206.5 & 10.7 & 0.12 & 48.8 & 1.8e+05 & 1.0e+03 + & 30.0 & 0.22 & 4.0e+04 & 223.0 & 13.2 & 0.12 & 69.8 & 3.4e+05 & 1.0e+03 we investigate the subsequent evolution of the fragment for the models with large ambient gas density @xmath121 where the sn - driven shell becomes gravitationally unstable well before @xmath122 yr . in these models , the time , @xmath113 when sn shells fragment is comparable to the free - fall timescale , @xmath33 , of the shell at the time @xmath113 .
the fragments are expected to contract gravitationally on the free - fall time scale ( equation [ one - zone : drhodt ] ) .
curves in figure [ fig : nt ] denote the evolutionary trajectories of the core of each fragment calculated by our one - zone code for the metallicities @xmath123 and @xmath119 when sn explosion energy and ambient density are @xmath124 . the initial density , temperature and the chemical species abundances are taken from the result of our one - dimensional calculation .
we use the result at the time when the gravitational instability is triggered in the shell .
the temperature stays between a few hundred to one thousand kelvins while the density increases over many orders of magnitudes . the compressional heating is largely balanced by radiative cooling until the hydrogen atomic number density increases to @xmath125 .
there are , however , two slight declines as clearly seen in figure [ fig : nt ] .
the first decline appears at @xmath126 where cooling by hd molecular lines is efficient .
the second decline appears at @xmath127 where cooling by dust thermal emission is efficient . in the latter regime ,
the gas cloud core becomes further unstable and is likely to fragment into multiple clumps @xcite .
the gravitational instabilities occur when the dust cooling timescale is shorter than the free - fall timescale .
the gas cooling by dust is due to heat transfer from gas particles to cold dust grains ( see * ? ? ?
then , essentially the gas pressure determines the juncture for efficient dust cooling as @xmath128 where @xmath129 and @xmath130 denote the mean radius and mass density of an individual dust grain , @xmath131 is dust mass fraction , and @xmath132 is the accommodation coefficient for heat transfer between gas and dust . from the typical values of @xmath133 , @xmath134 @xcite , @xmath135 , and @xmath136 ,
the region on the @xmath137-@xmath41 plane where dust cooling becomes efficient is @xmath138 the triangles in figure [ fig : nt ] denotes the points where the gas clouds satisfy the condition given by equation ( [ dustdom : criterion ] ) .
the corresponding jeans mass at the point is @xmath139 for @xmath140 and @xmath141 for @xmath142 .
the jeans mass further decreases as the gas temperature rapidly decreases by dust cooling , as can be seen in figure [ fig : nt ] .
the rapid dust cooling causes deformation of the cloud core into a filamentary structure .
@xcite show , using three - dimensional simulations , that an initially perturbed gas cloud is significantly elongated to filamentary shape and further fragments into small pieces when dust cooling operates for the metallicities @xmath143 .
the deformation instability occurs when the deformation parameter , defined as the effective ratio of specific heats , @xmath144 , is less than a critical value @xmath145 @xcite . in our calculations , @xmath146 declines rapidly below the critical value when the dust cooling becomes efficient .
the deformation parameter is 0.5 at a minimum when @xmath147 for @xmath148 .
the value is common with all explosion energies we set . for @xmath149 ,
the minimum value of @xmath146 is 0.30.4 at @xmath150 .
the contracting gas clouds are strongly unstable to deformation in these regimes . when the instability occurs , regardless of the sn explosion energies and metallicities , the jeans mass is 0.010.1 @xmath151 which is the characteristic mass of a fragment ( protostar ) .
the ambient gas quickly accretes onto the protostars and their mass likely become larger @xcite
. however , the subsequent evolution of the protostars will be highly dynamical . according to @xcite
, some of the protostars can attain velocities above the escape velocity of the parent gas cloud through dynamical interactions among them .
they would then escape from the gas cloud before the protostars become massive .
one natural outcome would then be the formation of low - mass stars .
we have studied the evolution of an early sn remnant with a metallicity of @xmath152 .
the supernova shell satisfies the criterion for gravitational instability at @xmath153@xmath154 yr , to form multiple fragments with the characteristic jeans mass of @xmath155@xmath9 for a wide range of the sn explosion energies , @xmath12@xmath13 erg , and large ambient gas density , @xmath156 .
we have then followed the thermal evolution of one of the fragments using one - zone calculations .
the gas cloud core evolves roughly isothermally until the density increases to @xmath157 , where efficient cooling by dust thermal emission brings the gas temperature from @xmath8 k to @xmath158 k rapidly .
the unstable core is expected to deform and fragment again into multiple clumps ( protostars ) with masses of @xmath159@xmath160 regardless the sn explosion energies and metallicities .
if some of the protostars escape from the host gas cloud before growing in mass over @xmath161 , they would live long to the present day .
we argue that the shell fragmentation of an early sn remnant is a viable scenario for the formation of low - mass and extremely low - metallicity stars .
there have been several theoretical studies which suggest the formation of low - mass stars in low - metallicity gas using numerical simulations . for metallicites of @xmath162
, small mass clumps are formed from a rotating gas cloud @xcite and in turbulent gas @xcite in many conditions these authors examined .
our hydrodynamic calculations are based on sn explosions of an early generation of stars .
the end point of the hydrodynamic simulation of the sn remnant , when the shocked shell becomes gravitationally unstable , provides physically motivated initial conditions for our one - zone calculations .
therefore , we are able to propose , for the first time , a consistent model of how low - metallicity stars are formed in the early universe .
we note that there are still limitations in the one - dimensional and one - zone simulations .
non - spherical structures of an sn remnant might have substantial effects on the shell fragmentation . @xcite and @xcite
show that shell instabilities might enhance density perturbations around an sn remnant to form dense clumps in early epochs of the remnant evolution .
dense gas clumps could form even earlier than we have shown in this paper .
clearly , full three - dimensional calculations are needed to study in detail the evolution of an sn shell .
our calculations may provide basic configurations for such studies on the formation of low - mass stars in low - metallicity gas .
currently standard models of galaxy formation are based on the hierarchical growth of cold dark matter halos .
recent high - resolution @xmath163-body simulations show that tens of thousands of early small halos assemble a milky way size halo @xcite .
although stars formed in the earliest halos are likely populated in the central bulge region , there should be a substantial fraction of the remnants of early halos orbiting in the halo of the galaxy @xcite .
@xcite may have found one of such `` fossils '' .
future surveys of metal - poor stars will provide the relative number fraction of extremely metal - poor stars , which will then constrain the overall formation efficiency of low - mass stars in the scenario we propose here .
the authors thank k. omukai , m. n. machida , k. iwasaki , t. tsuribe , and colleagues at ipmu for fruitful discussions and helpful advice .
we are grateful to an anonymous referee of apj , who suggested us to derive the metallicity dependence of the critical pressure for the dust cooling .
part of the simulations were performed at the center for computational astrophysics at the national astronomical observatory of japan .
this work is supported in part by the grants - in - aid for young scientists ( s : 20674003 , b : 21740139 ) by the japan society for the promotion of science . n.y .
acknowledges financial support from world premier international research center initiative ( wpi initiative ) , mext , japan .
beers , t. c. , & christlieb , n. 2005 , , 43 , 531 bromm , v. , yoshida , n. , & hernquist , l. 2003 , , 596 , l135 bromm , v. , yoshida , n. , hernquist , l. , & mckee , c. f. 2009 , , 459 , 49 bromm , v. , & yoshida , n. 2011 , , 49 , 373 iocco , f. , mangano , g. , miele , g. , pisanti , o. , & serpico , p. d. 2009 , , 472 , 1 iwasaki , k. , & tsuribe , t. 2008 , , 60 , 125 iwasaki , k. , inutsuka , s. i. , & tsuribe , t. 2011a , , 733 , 17 iwasaki , k. , inutsuka , s. i. , & tsuribe , t. 2011b , , 733 , 16 lai , d. 2000 , , 540 , 946 larson , d. , dunkley , j. , hinshaw , g. , et al .
2011 , , 192 , 16 larson , r. b. 1969 , , 145 , 271 lipovka , a. , nunez - lopez , r. , & avila - reese , v. 2005 , , 361 , 850 love , s. g. , joswiak , d. j. , & brownlee , d. e. 1994 , , 111 , 227 omukai , k. 2000 , , 534 , 809 omukai , k. , & yoshii , y. 2003 , , 599 , 746 omukai , k. , tsuribe , t. , schneider , r. , & ferrara , a. 2005 , , 626 , 627 omukai , k. , & nishi , r. 1998 , , 508 , 141 omukai , k. , hosokawa , t. , & yoshida , n. 2010 , , 722 , 1793 ostriker , j. p. , & mckee , c. f. 1988 , reviews of modern physics , 60 , 1 santoro , f. , & shull , j. m. 2006 , , 643 , 26 spitzer , l. , jr .
1978 , , 72 , 349 schneider , r. , ferrara , a. , salvaterra , r. , omukai , k. , & bromm , v. 2003 , , 422 , 869 schneider , r. , omukai , k. , bianchi , s. , & valiante , r. 2012 , , 419 , 1566 sedov , l. i. 1958 , reviews of modern physics , 30 , 1077 shu , f. h. , lizano , s. , galli , d. , cant , j. , & laughlin , g. 2002 , , 580 , 969 stahler , s. w. , & palla , f. 2005 , the formation of stars , by steven w. stahler , francesco palla , pp . 865 .
isbn 3 - 527 - 40559 - 3 .
wiley - vch , january 2005 . ,
sutherland , r. s. , & dopita , m. a. 1993 , , 88 , 253 whalen , d. , abel , t. , & norman , m. l. 2004 , , 610 , 14 whalen , d. , van veelen , b. , oshea , b. w. , & norman , m. l. 2008 , , 682 , 49 wang , h. , mo , h. j. , jing , y. p. , yang , x. , & wang , y. 2011 , , 413 , 1973 whalen , d. , hueckstaedt , r. m. , & mcconkie , t. o. 2010 , , 712 , 101 | we study the formation of low - mass and extremely metal - poor stars in the early universe .
our study is motivated by the recent discovery of a low - mass ( @xmath0 ) and extremely metal - poor ( @xmath1 ) star in the galactic halo by @xcite .
we propose a model that early supernova ( sn ) explosions trigger the formation of low - mass stars via shell fragmentation .
we first perform one - dimensional hydrodynamic simulations of the evolution of an early sn remnant .
we show that the shocked shell undergoes efficient radiative cooling and then becomes gravitationally unstable to fragment and collapse in about a million years .
we then follow the thermal evolution of the collapsing fragments using a one - zone code .
our one - zone calculation treats chemistry and radiative cooling self - consistently in low - metallicity gas .
the collapsing gas cloud evolves roughly isothermally , until it cools rapidly by dust continuum emission at the density @xmath2@xmath3 .
the cloud core then becomes unstable and fragments again .
we argue that early sne can trigger the formation of low - mass stars in the extremely metal - poor environment as @xcite discovered recently . |
polaritons - hybrid quasiparticles originating from the strong coupling between light modes and matter excitations - represent an important research area , both for fundamental and practical reasons : their bosonic character may lead to bose - einstein condensation @xcite , and the larger coherence length of the light modes may enhance the carrier mobility in organic semiconductors @xcite . while originating primarily in quantum - optical settings such as josephson junctions and arrays of coupled cavities @xcite , polaritonic excitations are now also studied in molecular systems @xcite .
alternatively , trapped ions can be used to study the properties of strongly correlated systems under well - controlled conditions and with manageable , slower time scales than solid - state systems @xcite .
polaritonic systems can be modeled in ion trap experiments by coupling the electronic and vibrational degrees of freedom of the ions confined by a harmonic potential .
the atomic excitations are encoded into the ions electronic states , and the local vibrational modes of the ions , described by quantum mechanical harmonic oscillators , take on the role of the light modes .
the coulomb repulsion between the ions leads to couplings of the local phonons .
finally , a suitably chosen external laser field realizes jaynes - cummings type interactions@xcite between each ion s electronic state and its local phonon mode @xcite .
the full system is then described by a jaynes - cummings - hubbard model @xcite , in which the total number of electronic and vibrational excitations is conserved . for filling factor one ( @xmath0 excitations on @xmath0 sites ) , the ground state varies , with the detuning of the external laser field , from a phonon superfluid phase , in which vibrational excitations are delocalized over the entire chain , to an atomic mott insulator phase , in which electronic excitations reside on individual ions .
while all excitation energy is stored in the vibrational degree of freedom , in the former case , it is fully absorbed by the electronic degree of freedom , in the latter . in the intermediate regime
, the system exhibits a polaritonic phase of coupled atom - phonon excitations .
when the filling factor is varied , the model allows for further phases , such as a polariton glass @xcite . in this work , we devise an interaction scheme which induces coherent transitions between these two phases .
this interaction will allow us to design pulse sequences @xcite for multidimensional nonlinear spectroscopy of the polariton dynamics .
in contrast to conventional spectroscopic techniques , our interaction does not change the number of excitations in the system , but rather switches between electronic or phononic character of the excitations . in a phase - coherent three - pulse sequence
, the system evolves through both the superfluid and the insulator quantum phases , as well as a coherent superposition of both .
we will see that nonlinear spectroscopy is particularly well suited to identify the relevant excitation processes , as well as the dominant decoherence mechanisms via the observed two - dimensional lineshapes . to ease scalability towards large system sizes ,
in which quantum phase transitions become most pronounced , we renounce of single - ion addressability in our spectroscopic protocol , which , however , could be exploited to add spatial resolution to the obtained nonlinear spectra @xcite .
the dynamics of polaritonic systems can be simulated by subjecting a collection of trapped ions to appropriate laser fields @xcite , taking into account both the electronic and motional states of all the ions .
as mentioned before , the relevant electronic states of the trapped ions are modeled as non - interacting spin-@xmath1 systems @xcite , while the motion is determined by their common trap potential , which , to a good approximation , is described by a three - dimensional harmonic oscillator @xcite .
the analogy to polaritonic systems is achieved by identifying the phonons in the ion trap with the photons of coupled cavities . in the remainder of this article
we refer to the actual physical degree of freedom , i.e. , the phonons , rather than the mathematically equivalent photons in the case of an optical polaritonic system . near their motional ground state
, the interplay between the potential and the strong coulomb repulsion of the ions determines the chain geometry .
the equilibrium positions of the ions are determined by the ratio of the trap frequencies @xmath2 along the different spatial directions .
a linear chain arrangement is achieved when , e.g. @xmath3 , and allows for convenient laser addressing of individual ions , with typical inter - ion spacings of @xmath4 and typical trap frequencies on the order of @xmath5mhz and @xmath6mhz . in the following
, we will consider such a linear chain of laser - cooled ions , which allows us to assume that the ions displacement from their equilibrium positions is much smaller than their separation .
employing a second - order expansion@xcite in terms of the displacement along the @xmath7-direction leads to the hamiltonian ( we set @xmath8 ) @xmath9 where @xmath10 creates a local phonon along the @xmath7-direction at the ion @xmath11 , and @xmath12 is the spin-@xmath1 ladder operator at site @xmath11 .
the electronic states of all the ions are described by the same transition frequency @xmath13 .
due to large spatial separation between adjacent ions , direct transfer of electronic excitations between the ions may be neglected
. however , the coulomb repulsion of the ions affects the motional degrees of freedom in eq .
( [ eq.h0 ] ) : it is responsible for both , modulations of the local oscillator frequencies @xmath14 and the phonon - phonon couplings @xmath15 , where @xmath16 is the ratio between the confining trap frequencies in axial @xmath17 and radial @xmath18 direction , and @xmath19 the ions equilibrium positions in units of @xmath20 .
@xcite an essential tool for the control of the quantum state of trapped ions are coherent laser interactions @xcite .
the interaction hamiltonian , generated by a continuous - wave laser , may be written as @xmath21 and describes the interaction of the @xmath11-th ion ( at the position @xmath22 ) with a laser beam of frequency @xmath23 , wavevector @xmath24 , and rabi frequency @xmath25 . for laser - cooled ions of low kinetic energy , the trap potential is well approximated by a static harmonic potential @xcite , and we may express the ion position in terms of the harmonic ladder operators .
for example , when @xmath24 coincides with one of the motional axes , e.g. , the @xmath7-direction , in the case of a single ion , we obtain @xmath26 , with the bosonic ladder operators as defined above , and the lamb - dicke parameter @xmath27 , which depends on the frequency @xmath28 of the harmonic motion . in general , for arbitrary numbers of ions , tuning the frequency @xmath23 allows to engineer effective couplings between the electronic and vibrational degrees of freedom , as well as between the electronic states of different ions @xcite . to see this , it is important to notice that the electronic spectrum of the ions is modulated by vibrational sidebands [ as can be seen in eq .
( [ eq.h0 ] ) ] , which originate in the ions motion around their equilibrium positions . in analogy to stokes and anti - stokes lines in raman spectroscopy @xcite ,
these resonances are found when the laser detuning coincides with multiples of the frequency of the harmonic motion . using tunable narrow - band lasers ,
specific sidebands can be resonantly selected .
this effectively creates coupled electronic and vibrational excitations : driving the first red ( blue ) sideband leads to laser - induced transitions which create an electronic excitation while a phonon is removed ( created ) .
hence , the dynamics of , e.g. , the red sideband is described by a jaynes - cummings hamiltonian which induces couplings of the type @xmath29 .
a more detailed treatment that includes the contribution of various harmonic modes is described in appendix [ appendixa ] . in this paper
, we consider the scenario considered in ref .
to couple motional and electronic degrees of freedom , and thereby create polaritonic excitations , one irradiates the entire ion chain with a travelling - wave laser , oriented along the radial @xmath7-direction , whose frequency is chosen as @xmath30 , i.e. , the laser is detuned from the radial red sideband transition by @xmath31 .
@xcite in the lamb - dicke limit ( c.f .
appendix [ appendixa ] ) , and after performing the rotating wave approximation , the dynamics of @xmath0 ions is effectively described in an appropriate rotating frame by the hamiltonian @xcite @xmath32 where , as before , @xmath10 creates a local phonon along the @xmath7-direction at site @xmath11 , and @xmath12 is the spin-@xmath1 ladder operator of ion @xmath11 .
the coupling between spin and phonon degrees of freedom is denoted with an effective rabi frequency @xmath33 .
the detuning @xmath31 from the red sideband transition determines the effective local spin s energy splitting .
the validity of the above hamiltonian is limited to the case where all parameters @xmath34 , @xmath31 , and @xmath33 are small @xcite , for the following reasons : the ratio of squared trap frequencies @xmath34 must be small to justify the quadratic expansion of the coulomb interaction term which lead to the motional hamiltonian ( [ eq.h0 ] ) .
the detuning @xmath31 must be significantly smaller than the trap frequency @xmath28 to ensure the validity of the rotating wave approximation which was employed to obtain eq .
( [ eq.h ] ) .
finally , when the laser - induced atom - phonon coupling becomes too large , multiple phonons can be created or destroyed in one electronic transition , thus higher - order terms would have to be included in the interaction term in eq .
( [ eq.h ] ) . ) for @xmath35 ions as a function of the ratio between detuning and interaction strength @xmath36 .
the eigenstates are indicated , their subscripts denote the number of spin excitations , and the superscripts label the individual states within a spin excitation submanifold . , scaledwidth=99.0% ] we reiterate that the coupling between spins and phonons is of polaritonic ( rather than polaronic ) nature , since it leads to an exchange of atomic and phononic excitations : one easily verifies that the total number of excitations @xmath37 is a constant of motion of the hamiltonian ( [ eq.h ] ) .
@xcite due to the interaction between phononic and electronic degrees of freedom , an initial spin excitation localized on a specific ion will be transferred to the phononic subspace , spread due to the coulomb interaction , and transfer back to the electronic subspace .
our present analysis aims to understand the dynamics in a given @xmath0-excitation manifold of the system .
in particular , we restrict to the case where the number of excitations equals the number of ions , i.e. , for a filling factor one . in this case , the ground state of the system s
@xmath0-excitation manifold for @xmath38 is given by the uncorrelated _ atomic insulator _
( ati ) state @xmath39 denotes the excited electronic state , and @xmath40 the phononic vacuum .
conversely , for @xmath41 , the lowest - energy state is characterized by excitations of the motional eigenmode of lowest energy @xcite , created by the operator @xmath42 .
this state is denoted the _ phonon superfluid _ ( phsf ) state , @xmath43 , the system is described by a polaritonic superfluid phase @xcite .
this phase is characterized by a delocalized ground state of polaritonic character , i.e. , containing both atomic and phononic excitations @xcite .
let us now also briefly discuss the excited states for the simplest nontrivial example : the two - excitation subspace in a two - ion chain .
the structure of the eight - dimensional spectral manifold is depicted in fig .
[ fig.levelscheme]b ) , as a function of the ratio @xmath44 . if @xmath45 , the spin hamiltonian @xmath46 dominates over all other terms in the hamiltonian . in the case of the two - ion chain considered here ,
this hamiltonian has three eigenvalues @xmath47 , @xmath31 , and @xmath48 ( which are determined by the number of excited spins ) , and consequently we identify three submanifolds of states , which differ in the number of spin excitations : the submanifold @xmath49 , with the ground state @xmath50 , encompasses all three states with close to zero spin excitations .
the first excited - state submanifold contains four states @xmath51 with a single spin excitation .
finally , both spins are excited in the highest excited state @xmath52 . since our analysis restricts to the two - excitation subspace ,
the respective spin excitations are complemented by phonon excitations among the two eigenmodes to reach a total of two excitations . in the other limit , when @xmath53 , the same three submanifolds can again be identified , however , their energies are reversed : the state @xmath54 now constitutes the ground state manifold , and the zero - spin submanifold @xmath49 yields the highest - excited states . in between these two limits ,
when @xmath55 , the eigenstates lose their clear character , and in the thermodynamic limit , one expects an abundance of avoided crossings which accompany the quantum phase transition and point to quantum chaotic behavior @xcite .
nonlinear spectroscopy provides powerful tools to probe the dynamics and decay processes of complex interacting quantum systems @xcite .
a general theoretical framework in the context of trapped - ion systems was recently developed @xcite , which allows for the identification of a variety of phenomena ranging from decoherence - free subspaces @xcite to phonon currents in a nonequilibrium steady - state @xcite and structural phase transitions in ion crystals @xcite .
the two essential ingredients are sequences of short , perturbative pulses separated by tunable time - delays , and a mechanism that allows to post - select the phase signature of particular excitation sequences . in large ensembles ,
the latter can be achieved via phase matching , which is not possible for trapped - ion systems , due to the limited amount of scattered photons per incoming probing pulse @xcite .
a feasible alternative is presented by phase - cycling which was developed to probe nuclear magnetic resonance systems with nonlinear spectroscopy @xcite . in the following
, we shall employ these ingredients to investigate the dynamics induced by eq .
( [ eq.h ] ) .
for the present simulation study , we consider the interaction with a global laser field , which illuminates all the ions in the trap with equal intensity and is tuned resonantly to the motional sideband of the collective mode @xmath56 .
this allows to realize an interaction of the form@xcite ( see appendix [ appendixa ] ) @xmath57 where @xmath58 creates collective excitations among the spins , where the coefficients @xmath59 decompose the collective mode @xmath56 into local contributions as @xmath60 .
it is possible to switch between the atomic insulator and the phonon superfluid states by inducing @xmath0 interaction events of the above form , when @xmath56 describes the vibrational eigenmode of lowest energy .
these interactions conserve the number of total excitations in the system : each spin excitation is accompanied by a phonon annihilation , and vice versa .
phase cycling with respect to the phase @xmath61 can distinguish between the different contributions to the total signal .
this technique further renders weak , perturbative pulses sufficient to obtain multidimensional spectra combined with a precise time resolution , and allows to post - select the evolution of specific coherences . , and evolves freely during @xmath62 .
two excitation pulses then transfer the chain into the atomic insulator phase @xmath63 .
post - selecting the phase @xmath64 allows to separate the excitation process into two steps , and to study the evolution of the coherence @xmath65 during @xmath66 .
the readout signal is created at the end of the sequence by fluorescence detection at ion @xmath67.,scaledwidth=40.0% ] based on an extension of the single quantum coherence signal @xcite , we design the following excitation scheme ( see fig .
[ fig.diagram ] ) : 1 .
the system is initialized in the phonon superfluid state @xmath68 .
2 . during the first time interval , @xmath62
, the system evolves in the phonon superfluid state ( [ eq.psi_phsf ] ) .
this evolution is described by the liouville space superoperator @xmath69 $ ] , where the liouvilian @xmath70 contains the coherent evolution according to the hamiltonian ( [ eq.h ] ) , as well as the influence of possible noise sources .
some of the dominant noise sources of trapped - ion experiments are mentioned in the course of our simulations in the next section .
we obtain @xmath71 .
the first excitation pulse populates higher - excited states of the system . by phase cycling with respect to the phase signature @xmath72 , where @xmath73 denotes the phase of the @xmath74-th pulse
[ see eq .
( [ eq.h_i ] ) ] , it is possible to post - select those processes , in which the excitation pulse transfers the system into a superposition state of phonon superfluid and atomic insulator .
thus , the contributing terms are described by @xmath75 , where @xmath76 represents the liouville space superoperators associated with an excitation acting only on the left ( l ) side of the density matrix .
similarly , we introduce @xmath77 the decay of this superposition can be observed during the second time interval @xmath66 , and is described by @xmath78 4 . the phase cycling procedure described above post - selects pathways , in which the second excitation transfers the system fully into a population of the atomic insulator state , whose dynamics is monitored during @xmath79 .
we obtain @xmath80 5 .
finally , the signal is obtained by measuring the spin population of one of the ions ( characterized by the ion s index @xmath67 ) via fluorescence @xcite .
the measured signal is given by the nonlinear correlation function@xcite @xmath81 where @xmath82 denotes the initially prepared superfluid state and @xmath83 the trace . to efficiently analyze this multidimensional signal , we will fourier transform two of the time delays , @xmath84 or @xmath85
let us discuss the conditions and implications of each of the above steps in further detail .
the initialization of the phonon superfluid state can be achieved via a combination of ground - state laser cooling , optical pumping , and well - defined pulses on blue sideband transitions @xcite .
if the system parameters are chosen such that the initial state @xmath68 actually represents an eigenstate , there is no time evolution during @xmath62 .
the presence of small contributions of other excited states will become visible in the resulting two - dimensional spectrum .
such contributions would also open up the possibility of removing spin excitations with the first pulse . for @xmath86 ,
the state @xmath68 is very close to an eigenstate of the system : when @xmath87 , it constitutes the ground state , when @xmath88 the highest excited state , with small corrections due to the finite spin - phonon coupling @xmath33 . in this case
, we can neglect contributions from other diagrams , in which , for instance , an initial spin excitation creates additional pathways .
these need to be taken into account only when @xmath89 , which shall not be our concern in this paper .
finally , the measurement of spin populations is easily realized by fluorescence detection @xcite . since the spatial decomposition of the addressed eigenmode is contained in the spin excitation operator @xmath90 , the signal can be dependent on @xmath67 , i.e. , we obtain spatial information about the relevant vibrational mode only with local readout , without resorting to local excitations .
single - site addressing is in principle possible and is able to enhance the information content compared to conventional nonlinear spectroscopic signals @xcite .
whereas the spatially resolved readout of the fluorescence light is routinely realized via high - resolution cameras , even for large @xmath0 , the addressing of individual ions with focussed lasers becomes technically demanding for increasing system sizes @xcite .
since the macroscopic properties of the quantum phases and the corresponding phase transition is most interesting in this large-@xmath0 limit , we avoid the usage of focussed excitations to ease scalability . in the present analysis
, we have considered the impulsive limit , which implies that the length of the applied pulses is much shorter than the characteristic time scale of the relevant system dynamics .
the dynamics of the effective polaritonic hamiltonian ( [ eq.h ] ) is determined by the motional frequencies , as well as by the detuning of the optical field .
experimentally this typically leads to relevant time scales on the order of hundreds of microseconds @xcite .
moreover , the evolution can be slowed down artificially by adiabatically changing the ratio of trap frequencies .
these time scales allow for the applications of pulses that , on the one hand , are short compared to the system dynamics , and , on the other hand , are long enough to spectrally resolve individual transitions .
whereas often the information contained in single quantum coherence signals is also available in technically less demanding ramsey interferometric measurement sequences @xcite , this is not the case for the present simulation study . in a weak - field approximation ,
phase - cycling enables us to isolate those events where exactly @xmath0 excitations are exchanged with the interaction pulses of which no clear analogue exists in the context of a ramsey experiment .
( where we omit the site index @xmath67 in ( [ eq.s23 ] ) , since the fluorescence signal is independent of @xmath67 for @xmath35 ) of the pulse sequence ( [ eq.signal-definition ] ) in a two - ion chain implementing the hamiltonian ( [ eq.h ] ) , with @xmath92mhz , @xmath93khz , @xmath94khz , @xmath95khz .
the panels a d zoom into specific resonance structures to highlight the two - dimensional lineshapes . ]
we now present simulations of the three - pulse sequence ( [ eq.signal-definition ] ) with the level scheme in fig .
[ fig.levelscheme ] for the ratio @xmath96 at @xmath35 . the initially prepared state @xmath68 has strong overlap with the manifold ground state @xmath50 .
small corrections from the first single spin excitation submanifold become visible through oscillations during @xmath62 . after two interactions with the first pulse , the strongest contribution to the generated state @xmath63 is given by the highest excited state @xmath52 .
in the subsequent time interval @xmath66 , the decay of the coherence @xmath97 can be observed in the two - dimensional spectrum .
[ fig.f2f3 ] displays the two - dimensional spectrum @xmath98 ( where we omit the site index @xmath67 in ( [ eq.signal-definition]-[eq.s23 ] ) , since the fluorescence signal is independent of @xmath67 for @xmath35 ) .
our simulations were carried out using the ` qutip ` package @xcite .
the parameters used in our simulation are @xmath92mhz , @xmath93khz , @xmath94khz , @xmath95khz and reflect those of a recently reported experiment @xcite . to obtain realistic results
, we further include the effect of the dominant error source collective dephasing of the qubits due to fluctuations in the magnetic field that determines the common electronic resonance frequency @xcite , described by the dephasing operator @xmath99 with decay constant @xmath100khz .
another relevant noise source is caused by fluctuations of the laser intensity of the driving fields@xcite , which we do not consider here . in the following ,
we analyze the position and two - dimensional lineshapes of cross - peaks , which are direct traces of the character ( localized or delocalized ) of the involved states .
we identify four strong resonances in fig .
[ fig.f2f3 ] , which are labelled a , b , c , d .
all peaks can be uniquely attributed to a specific excitation pathway by extracting the corresponding resonance frequencies along @xmath101 and @xmath102 .
resonances a and c at @xmath103khz originate from the coherence @xmath104 during the @xmath66-evolution .
hence , it is those two peaks that trace the evolution of the macroscopic " coherent superposition of the two distinct phases @xmath105 . the stronger resonance a at @xmath106 signals the evolution of the population @xmath107 during @xmath79 . the states @xmath108 and @xmath109 are populated during @xmath79 due to the weak residual coupling of the spin - phonon interaction .
two less pronounced resonances at @xmath110 khz reveal the coherent evolution of superposition states that involve the coherences @xmath111 .
furthermore , the two - dimensional lineshapes of the resonances can be used as a tool to identify the nature of the underlying states . the ability to analyze the two - dimensional lineshapes@xcite represents another distinct advantage of multi - dimensional spectroscopy , and becomes particularly useful in the present context .
the influence of the collective dephasing process increases quadratically with the difference of occupation numbers that contribute to the superposition @xcite , i.e. , we have @xmath112 , and therefore is most pronounced for the decay of electronic coherences between the ground state manifold and the highest excited state .
when both states contain exactly the same number of spin excitations , they evolve in a decoherence - free subspace @xcite . here , however , the non - zero coupling between spins and phonons perturbs the symmetry of the subspace and will always lead to finite linewidths . hence , both resonances a and c are broadened strongly along @xmath101 ( where the discrepancy of spin excitations is two ) , and much less along @xmath102 ( where the states differ by no more than a single spin excitation ) . as discussed before , the resonance a represents the population @xmath113 during @xmath79 , i.e.
, it evolves in a decoherence - free subspace .
in contrast , the resonance c is also broadened along @xmath102 , but due to the smaller difference in the spin population between @xmath114 and @xmath115 , the broadening is less pronounced compared to the strong broadening along @xmath101 .
resonances b and d at @xmath116 khz pertain to processes in which the two interactions with the first pulse only create a single spin excitation again , due to the residual spin - phonon coupling and the system therefore evolves in a coherence between one of the two single - spin states @xmath108 and @xmath109 and the initially prepared ground state @xmath50 . at resonance b ,
the second pulse then creates two spin excitations on the _ bra _ side of the density matrix , such that the system evolves in the coherence @xmath117 during @xmath79 .
consequently , the difference of spin excitations remains unchanged during both propagation intervals , and the resonances are broadened equally along both @xmath101 and @xmath102 .
finally , the resonances in the close - up d represent a population during @xmath79 , such that they are only broadened along @xmath101 just like resonance a. , see eq .
( [ eq.s^13 ] ) , of the pulse sequence ( [ eq.signal-definition ] ) in a two - ion chain , with the hamiltonian ( [ eq.h ] ) and parameters @xmath92mhz , @xmath93khz , @xmath119khz , @xmath95khz . ]
we finally analyze the signal @xmath120 , eq .
( [ eq.s^13 ] ) , for the case of negative detuning @xmath121 khz . here , the two - spin state @xmath115 forms the ground state , and the zero - spin manifold represents the highest - excited states [ see fig . [ fig.levelscheme]b ) ] .
the signal @xmath118 correlates the frequencies of the two free evolution periods when the system is fully described by populations .
consequently , the main resonance can be found at zero frequency , with a broadening solely due to the finite propagation time .
the signal measures the deviations from the @xmath122 limit , as the four resonances that can be found at nonzero frequencies all stem from deviations of the ground state @xmath115 from @xmath63 , and of @xmath123 from @xmath124 .
the resonances at @xmath125 khz yield the dominant correction to @xmath63 , and the resonance pairs at @xmath126 khz those to @xmath124 in the simulated realization of the polaritonic hamiltonian .
we have proposed a nonlinear measurement scheme to probe the nonequilibrium dynamics of polaritonic quasiparticles in trapped ion chains .
suitably chosen laser pulses can drive the system between an atomic insulator and a superfluid phase , and create superposition states of the two phases . on top of the identification of the contributing eigenstate frequencies ,
the analysis of the two - dimensional lineshapes in the measurement setup allows for a clear identification and characterization of the states , since the dominant noise source is sensitive to the number of spin excitations present in the system .
hence , the broadening of the resonance along specific axes indicates the discrepancy of electronic excitations between the two coherently superposed states .
the variation of the third time delay , which was kept fixed in our present study , would further allow to monitor transport and relaxation processes within the @xmath0-excitation submanifold .
the measurement scheme proposed in this work does not require individual laser access to the ions and is therefore easily scalable towards larger @xmath0 .
local information on the excitations can still be achieved by spatially resolved collection of the fluorescence light , which is standard practice in ion trap experiments .
the distinct couplings of the ions to different collective modes are then reflected in the locally collected measurement signals .
the system we studied a collection of two - level atoms that couple to harmonic motional modes could also be interesting for experimental studies of the role of the vibrational backbone for exciton transport , which is believed to represent an important mechanism for biomolecular transport processes @xcite .
to this end , one would model the direct exciton couplings , which is lacking in the polariton model , by adding effective spin - spin interactions @xcite . in the context of biomolecules , an electron - phonon coupling that preserves the number of electronic excitations , i.e. , a polaronic coupling
, would be more appropriate @xcite than the polaritonic coupling considered in the present study .
m. g. and f. s. thank the german national academic foundation for support .
in this appendix , we derive eq . ( [ eq.h_i ] ) .
our derivation closely follows standard approaches in the trapped - ion literature , which can be found , for instance in refs .
its starting point is the hamiltonian @xmath127 which describes the interaction of the ion chain with the external laser field of frequency @xmath23 , wavevector @xmath128 which points along the @xmath7-axis , and phase @xmath61 .
we assume spatially homogeneous illumination of all ions , whose optical resonance frequency is given by @xmath13 .
note that due to the large separation of the ions ( see section [ sec.introions ] ) , the spatial dependence of the laser field needs to be taken into account .
we first diagonalize the vibrational hamiltonian , @xmath129 and express the motional dependence as a function of the lamb - dicke parameters @xmath130 as @xcite @xmath131 where the spatial orientation of the laser is adjusted such that only the vibrational modes along one direction are addressed @xcite .
the excitation frequency is set on resonance with the red sideband corresponding to the vibrational eigenmode @xmath132 , i.e. , @xmath133 .
next we employ a rotating wave approximation and change to the interaction picture , employing the transformation @xmath134 t\right),\end{aligned}\ ] ] resulting in @xmath135 we shall consider the lamb - dicke limit @xcite , @xmath136 , which is reached when the amplitude of the ion s motion is much less than @xmath137 . in this case
we may approximate the exponential in eq .
( [ eq.h_1 t ] ) by its taylor series to linear order , @xmath138 retaining only non - oscillatory terms in eq .
( [ eq.h_1 t ] ) , we then arrive at @xmath139 for @xmath140 we finally obtain eq .
( [ eq.h_i ] ) by identifying @xmath141 and @xmath142 . in the present of case of a two - ion chain ,
we have two delocalized eigenmodes , which are given by the operators @xmath144 the center - of - mass ( high - energy ) mode , and @xmath145 the breathing ( low - energy ) mode .
hence , for @xmath35 the coefficients in eq .
( [ eq.h_i ] ) are @xmath146 .
h. deng , h. haug and y. yamamoto , http://link.aps.org/doi/10.1103/revmodphys.82.1489[rev .
phys . * 82 * , 1489 ( 2010 ) . ]
plumhof , t. stferle , l. mai , u. scherf and r.f .
mahrt , http://www.nature.com/nmat/journal/v13/n3/full/nmat3825.html[nature mater .
* 13 * , 247 ( 2014 ) . ]
et al . _ ,
http://arxiv.org/abs/1409.1900[arxiv : 1409.1900 [ cond - mat ] ] m. j. hartmann , f. g. s. l. brando , and m. b. plenio , nature phys .
* 2 * , 849 ( 2006 ) .
et al . _ ,
, http://link.aps.org/doi/10.1103/physreva.80.060301[phys .
a * 80 * , 060301 ( 2009 ) . ]
k. toyoda _
et al . _ , http://link.aps.org/doi/10.1103/physrevlett.111.160501[phys
* 111 * , 160501 ( 2013 ) . ]
m. gessner , f. schlawin , h. hffner , s. mukamel , and a. buchleitner , new j. phys .
* 16 * , 092001 ( 2014 ) . | we devise a phase - coherent three - pulse protocol to probe the polariton dynamics in a trapped - ion quantum simulation .
in contrast to conventional nonlinear signals , the presented scheme does not change the number of excitations in the system , allowing for the investigation of the dynamics within an @xmath0-excitation manifold . in the particular case of a filling factor one ( @xmath0 excitations in an @xmath0-ion chain ) ,
the proposed interaction induces coherent transitions between a delocalized phonon superfluid and a localized atomic insulator phase .
numerical simulations of a two - ion chain demonstrate that the resulting two - dimensional spectra allow for the unambiguous identification of the distinct phases , and the two - dimensional lineshapes efficiently characterize the relevant decoherence mechanism . |
the aim of this talk is to show that , when in a medium there is a finite charge density ( i.e. , a finite chemical potential ) , a new set of lepton pair - producing processes actually arises .
we then calculate a new contribution to the 3-loop photon self - energy .
the various cuts of this self - energy contain higher loop contributions to the usual processes of @xmath0 , @xmath1 , @xmath2 , and an entirely new channel : @xmath3 .
we calculate the contribution of this new reaction to the differential production rate of back - to - back dileptons .
it is finally shown that , within reasonable values of parameters , this process may become larger than the differential rate from the bare tree level @xmath0 . at zero temperature , and at finite temperature and zero charge density ( note :
henceforth , a finite density will imply a finite charge density ) , diagrams in qed that contain a fermion loop with an odd number of photon vertices ( e.g. fig .
[ furry ] ) are cancelled by an equal and opposite contribution coming from the same diagram with fermion lines running in the opposite direction ( furry s theorem @xcite ) .
this statement can also be generalized to qcd for processes with two gluons and an odd number of photon vertices .
a physical perspective is obtained by noting that all these diagrams are are encountered in the perturbative evaluation of green s functions with an odd number of gauge field operators . at zero ( finite ) temperature ,
in the well defined case of qed we observe quantities like @xmath4 ( @xmath5 $ ] ) under the action of the charge conjugation operator @xmath6 . in qed we know
that @xmath7 . in the case of the vacuum @xmath8
, we note that @xmath9 , as the vacuum is uncharged . as a result where @xmath13 and @xmath14 is a chemical potential . here , however , @xmath15 , where @xmath16 is a state in the ensemble with the same number of antiparticles as there are particles in @xmath17 and vice - versa .
if @xmath18 i.e. , the ensemble average displays zero density then inserting the operator @xmath19 as before , we get and furry s theorem will now break down .
one may say that the medium , being charged , manifestly breaks charge conjugation invariance and these green s functions are thus finite , and will lead to the appearance of new processes in a perturbative expansion .
the appearance of processes that can be related to symmetry - breaking in a medium has been noted elsewhere @xcite .
let us , now , focus our attention on the diagrams of fig .
[ 2vert ] for the case of two gluons and a photon attached to a quark loop ( the analysis is the same even for qed i.e. , for three photons connected to an electron loop ) . such a process does not exist at zero temperature , or even at finite temperature and zero density . at finite density
this leads to a new source of dilepton or photon production @xmath28 . in order to obtain the full matrix element of a process containing the above as a sub - diagram
one must coherently sum contributions from both diagrams which have fermion number running in opposite directions .
the amplitude for @xmath29 is : @xmath30 \frac{d^{3}q}{(2\pi)^{3 } } tr [ \gamma^{\mu } \gamma^{\beta } \gamma^{\rho } \gamma^{\delta } \gamma^{\nu } \gamma^{\alpha } ] \frac{(q+p - k)_{\alpha } q_{\beta } ( q+p)_{\delta } } { ( q+p - k)^{2 } q^{2 } ( q+p)^{2 } } , \nonumber \ ] ] @xmath31 \frac{d^{3}q}{(2\pi)^{3 } } tr [ \gamma^{\nu } \gamma^{\delta } \gamma^{\rho } \gamma^{\beta } \gamma^{\mu } \gamma^{\alpha } ] \frac{(q+k - p)_{\alpha } q_{\beta } ( q - p)_{\delta } } { ( q+k - p)^{2 } q^{2 } ( q - p)^{2}}. \label{2g1p } \ ] ] again , the extension of furry s theorem to finite temperature does not hold at finite density : as , if we set @xmath32 , we note that @xmath33 , and , as a result , @xmath34 . of course ,
if we now let the chemical potential go to zero ( @xmath35 ) , we note that for the transformation @xmath32 , we obtain @xmath36 , and , thus , @xmath37 .
the analysis for fermion loops with larger number of vertices is essentially the same . to calculate the contribution made by the diagram of fig .
[ 2vert ] , to the dilepton spectrum emanating from a quark gluon plasma , we calculate the imaginary part of the photon self - energy containing the above diagram as an effective vertex ( fig [ self1e ] , see reference @xcite for details ) . where , the @xmath39 s represent bare gluon propagators , and the @xmath40 s represent the effective vertices .
we calculate in the limit of photon three momentum @xmath41 .
the imaginary part of the considered self - energy contains various cuts .
we concentrate , solely , on the cut that represents the process of gluon - gluon to @xmath42(see fig.([self1e ] ) ) .
this is a process , which , to our knowledge , has not been discussed before .
the other possible cuts represent finite density contributions to other known processes of dilepton production . where @xmath45 is the electromagnetic coupling constant .
the rate of production of a hard lepton pair with total momentum @xmath41 , at one - loop order in the photon self - energy ( i.e. , the born term ) , is given for three flavours as in the above equation , the first term on the _ r. h. s. _ is the contribution from the up and down quark sector ; and the second part is the contribution from the strange sector . in a realistic plasma ,
the net charge and baryon imbalance is caused by the valence quarks brought in by the incoming charged baryon - rich nuclei .
the baryon and charge imbalance is , thus , manifested solely in the up and down quark sector ; hence , the chemical potential influences only the distribution function of the up and down quarks .
the strange and anti - strange quarks are produced in equal numbers in the plasma ; resulting in a vanishing strange quark chemical potential .
thus , dilepton production , from the channel indicated by fig.([self1e ] ) , will only receive contributions from the up and down quark flavours . the initial temperatures of the plasma , formed at rhic and lhc ,
have been predicted to lie in the range from 300 - 800 mev @xcite . for this calculation
we use estimates of @xmath47(fig .
[ mu ] ) and @xmath48(fig .
[ nu ] ) . to evaluate the effect of a finite chemical potential
, we perform the calculation with two extreme values of chemical potential @xmath49 ( left plots ) and @xmath50(right plots ) @xcite .
the calculation , is performed for three massless flavours of quarks . in this case , the strong coupling constant is ( see @xcite ) the differential rate for the production of dileptons with an invariant mass from 0.5 to @xmath52 is presented . on purpose , we avoid regions where the gluons become very soft . in the plots ,
the dashed line is the rate from tree level @xmath53 ( eq . ( [ qqbar ] ) ) ; the solid line is that from the process @xmath54 .
we note that in both cases the gluon - gluon process dominates at low energy and dies out at higher energy leaving the @xmath53 process dominant at higher energy .
as the reader may have noted , in the above calculation , @xmath55 , thus , we are not unequivocally in the perturbative regime . at asymptotically high temperatures ( @xmath56 ) , however , @xmath57 , in this limit one may make the hard thermal loop ( htl ) approximation @xcite ( note : only the main results will be quoted here
; the details will appear elsewhere @xcite ) .
at very high temperature , contributions from loop diagrams are dominated by loop momenta @xmath58 of the order of the temperature @xmath11 .
if the momentum flowing in the external legs is much smaller ( of the order of @xmath59 , i.e. , @xmath60 ) , then loop corrections are of the same order in @xmath61 as the bare diagrams , and , thus , must be resummed into the tree amplitudes .
@xmath64 \frac{\hat{q}_{s_{1},\alpha } \hat{q}_{-s_{2},\beta } \hat{q}_{-s_{3},\delta } } { p^{0 } - s_{2}e_{2 } + s_{3}e_{3 } } \nonumber \\ & & \bigg [ s_{2 } \frac{s_{1 } ( \tilde{n}(e_{3}-s_{3}\mu ) - \tilde{n}(e_{3}+s_{3}\mu ) ) + s_{3 } ( \tilde{n}(e_{1}+s_{1}\mu ) - \tilde{n}(e_{1}-s_{1}\mu ) ) } { k^{0 } + s_{1}e_{1 } + s_{3}e_{3 } } \nonumber \\ & - & s_{3 } \frac{s_{1}(\tilde{n}(e_{2}-s_{2}\mu ) - \tilde{n}(e_{2}+s_{2}\mu ) ) + s_{2}(\tilde{n}(e_{1}+s_{1}\mu ) - \tilde{n}(e_{1}-s_{1}\mu ) ) } { k^{0}-p^{0 } + s_{1}e_{1 } + s_{2}e_{2 } } \bigg ] , \end{aligned}\ ] ] where , @xmath65 , and @xmath66 .
note that if @xmath14 is set to zero this amplitude vanishes identically .
the conventional htl term ( i.e. , terms proportional to @xmath67 ) from triangle graphs such as these would come from the terms with @xmath68 and @xmath69 .
the term proportional to @xmath67 is @xmath70 \bigg \ { \frac{\hat{q}_{+,\alpha } \hat{q}_{+,\beta } \hat{q}_{+,\delta } } { p^{0 } - \vec{p } \cdot \hat{q } } \nonumber \\ & & \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\ ! \bigg [ \frac { \frac{\partial \tilde{n}(e_{3}+\mu ) } { \partial e_{3 } } ( -\vec{k}\cdot\hat{q } ) - \frac{\partial \tilde{n}(e_{3}-\mu ) } { \partial e_{3}}(-\vec{k}\cdot\hat{q } ) } { k^{0 } - \vec{k}\cdot\hat{q } } + \frac { \frac{\partial \tilde{n}(e_{2}-\mu ) } { \partial e_{2 } } ( ( \vec{p}-\vec{k})\cdot\hat{q } ) - \frac{\partial \tilde{n}(e_{2}+\mu ) } { \partial e_{2}}((\vec{p}-\vec{k } ) \cdot \hat{q } ) } { k^{0}-p^{0 } + ( \vec{p}- \vec{k})\cdot\hat{q } } \bigg ] + \nonumber \\ & & \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\ ! \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\
! \frac{\hat{q}_{-,\alpha } \hat{q}_{-,\beta } \hat{q}_{-,\delta } } { p^{0 } + \vec{p}\cdot\hat{q } } \bigg [ \frac { \frac{\partial \tilde{n}(e_{3}-\mu ) } { \partial e_{3 } } ( -\vec{k}\cdot\hat{q } ) - \frac{\partial \tilde{n}(e_{3}+\mu ) } { \partial e_{3}}(-\vec{k}\cdot\hat{q } ) } { k^{0 } + \vec{k}\cdot\hat{q } } + \frac { \frac{\partial \tilde{n}(e_{2}+\mu ) } { \partial e_{2 } } ( ( \vec{p}-\vec{k})\cdot\hat{q } ) - \frac{\partial \tilde{n}(e_{2}-\mu ) } { \partial e_{2}}((\vec{p}-\vec{k } ) \cdot \hat{q } ) } { k^{0 } - p^{0 } - ( \vec{p } - \vec{k } ) \cdot \hat{q } } \bigg ] \bigg\}. \nonumber\end{aligned}\ ] ] note that if we change @xmath71 in the last two terms then @xmath72 and the whole term becomes identically zero .
thus there is no term proportional to @xmath67 in this diagram .
the next term is of order @xmath73 and is nonzero @xcite .
however , as is well known , the htl approximation of the @xmath74 vertex is of the order of @xmath67 @xcite .
thus , in the high temperature approximation , the @xmath75 htl vertex is suppressed as compared to the @xmath74 htl - resummed vertex . as a result the dilepton production rate from resummed two - gluon - fusion process will also be suppressed compared to the rate from the resummed born term , in this limit . on summing up all the htl contributions to the self energy of the gluon we get two dispersion relations for the transverse and longitudinal modes of the gluon .
a plot of the two modes @xcite is as shown in fig .
( [ dsprsn ] ) .
the upper branch is the dispersion relation for the transverse quasi - particle excitation , the lower one is the longitudinal excitation .
the straight line , corresponding to free gluons is shown for reference .
note that there is no minima in either branch ( except at @xmath76 ) , unlike in the case of quarks @xcite .
hence , we will not see any sharp van hove peaks @xcite in the resulting dilepton spectra emanating from this process .
one may thus expect that the high temperature ( and , perhaps , as a result , the very low invariant mass ) dilepton production spectra emanating from in - medium gluon - gluon fusion will be suppressed compared to that from in - medium quark anti - quark fusion .
however , as we have noted in the previous section the intermediate invariant mass rate from bare gluon - gluon fusion is comparable and may be larger than that from bare quark anti - quark fusion .
the entire calculation above is at full chemical and thermal equilibrium . in the early part of an ultrarelativistic heavy ion collision ,
the gluon number has been predicted to be much higher than at full chemical equilibrium .
in such a scenario , the contributions to dilepton spectrum from processes such as those presented here will probably out - shine those from other channels .
the authors wish to thank y. aghababaie , a. bourque , s. das gupta , f. gelis , s. jeon , d. kharzeev , c. s. lam and g. d. mahlon for helpful discussions .
a.m. acknowledges the generous support provided to him by mcgill university through the alexander mcfee fellowship , the hydro - quebec fellowship and the neil croll award . this work was supported in part by the natural sciences and engineering research council of canada and by _ le
fonds pour la formation de chercheurs et laide la recherche du qubec .
_ see for example , s. a. chin , ann .
108 , 301 ( 1977 ) ; h. a. weldon , phys .
b , * 274 * , 133 ( 1992 ) ; g. wolf , b. friman and m. soyeur , nucl . phys .
* a640 * , 129 ( 1998 ) ; o. teodorescu , a.k .
dutt - mazumder and c. gale , phys .
c * 63 * , 034903 ( 2001 ) . | we investigate the effects of a charge asymmetry on the spectrum of dileptons radiating from a quark gluon plasma .
we demonstrate the existence of a new set of processes in this regime . the dilepton production rate from the corresponding diagrams is shown to be as important as that obtained from the born - term quark - antiquark annihilation .
address = department of physics , mcgill university , montreal , qc , canada h3a 2t8 , [email protected] , address = department of physics , mcgill university , montreal , qc , canada h3a 2t8 , [email protected] , |
in many experiments with radioactive atoms , elements are produced as ions and then neutralized .
the most important requirement for a neutralization system is the fast release of neutral atoms , with respect to their radioactive decay time . in particular , in our francium experiment at infn s national laboratories in legnaro , italy @xcite , we produce fr ions that have to be neutralized before accumulation in a magneto - optical trap .
the system must be very efficient , as a large sample of fr atoms is requested for nuclear decay , atomic parity violation and permanent electric dipole studies .
our apparatus follows the scheme of the stony brook experiment @xcite , where many measurements on fr have been performed so far .
francium isotopes are produced at the tandem accelerator facility in legnaro via the fusion - evaporation reaction @xmath3 with an energy of the primary beam around 100 mev , we are able to produce fr isotopes in the mass number range 208 - 211 .
these isotopes have a lifetime that is long enough for laser cooling and trapping .
the system is at the moment optimized for maximum production of @xmath4fr , which has a half - life of 191 s. francium atoms are ionized at the surface of the gold target and transported through an electrostatic beamline at an energy of 3 kev towards a spectroscopic cell designed for laser cooling .
the ionic beam is focused on a 99.96% pure yttrium foil , placed inside the cell on the opposite side with respect to the entrance aperture . in order to be able to detect the trap
, we have to optimize all the processes , from production to laser trapping . with only a few days of beamtime per year , it is not convenient to test the setup with francium : we decided to use stable rubidium , always available in high quantities . to this purpose
a rb dispenser was placed in the reaction chamber near the target : rb atoms that arrive on the gold surface are ionized and injected in the electrostatic line , following the same path as fr .
a sketch of the apparatus is shown in fig .
[ fg : expsetup ] .
yttrium has a work function @xmath5 ev , that is lower than fr and rb ionization potentials , respectively @xmath6 ev and @xmath7 ev .
the probability of release in neutral form at the y surface is close to unity , according to the saha - langmuir equation @xmath8 where @xmath9 and @xmath10 are the number of desorbed ions and neutral atoms , @xmath11 and @xmath12 are the statistical weights due to degeneracy of the ion and atom ground states respectively ( @xmath13 for alkalis ) , and @xmath14 is boltzmann s constant .
this ratio is @xmath15 at 1000 k. in spite of the fact that high temperature enhances diffusion , we fixed an upper limit of 1050 k in order to preserve the cell coating @xcite and hence not to compromise the trapping efficiency .
this is a limitation because melting point of y is @xmath16=1799 k , and because the total release fraction grows with the ratio @xmath17 @xcite . in this context ,
our goal was to check whether the efficiency of the neutralizer was high enough at a chosen temperature .
so far the release efficiency of radioactive atoms embedded in the neutralizer has been measured by monitoring their nuclear decay @xcite .
we used atomic laser spectroscopy to observe the time evolution of the released neutrals and deduce the diffusion time .
we also took advantage of the extraordinary sensitivity given by our magneto - optical trap , which allows us to observe as few as 50 atoms .
if an ion is implanted at a given depth inside the y foil , then it has to diffuse towards the surface to be released in atomic form . in order to achieve a good release efficiency , the time necessary to complete this process has to be small compared with the radioactive lifetime of the investigated atom .
since the transverse dimension of the ionic beam is much larger than the penetration depth , we can use the one - dimensional diffusion equation for the concentration @xmath18 of the diffusing species in the y foil at the distance @xmath19 from the surface : @xmath20 @xmath21 is the radioactive decay rate ( @xmath22 for rb ) , @xmath23 the diffusion coefficient and @xmath24 is the incoming current of ions impinging on the y foil . note that
@xmath24 is not directly observable : it has to be deduced from the measurement of its integral @xmath25 ( the total current ) on the neutralizer volume and from the presumed implantation distribution , discussed in sections [ sz : pulse ] and [ sz : cont ] .
empirically , we expect that the dependence of @xmath23 on the temperature follows the arrhenius law @xmath26 @xmath27 is the activation energy and @xmath28 is the theoretical asymptotic value of @xmath23 for large temperatures . the characteristic diffusion time is given by @xmath29 where @xmath30 is the mean implantation depth of ions in the y foil . in the case of @xmath31 ions all
implanted at the same distance @xmath30 from the surface at the time @xmath32 , i.e. @xmath33 , the diffusion equation with @xmath34 has an analytical solution : the flux of neutral atoms released in the cell at the time @xmath35 is given by @xcite @xmath36 for an implantation distribution with a finite extent , the function can be different , specially for @xmath37 ; in a rough approximation , this difference can be taken into account with a correction factor , which is expected to be of the order of unity in the case that @xmath38 @xcite .
for instance , melconian _ et al .
_ take the case of a distribution modelled by a gaussian times a linear term , with a characteristic implantation depth @xmath30 @xcite . in this case
, the function becomes @xmath39 if a constant total ionic current @xmath40 is sent to the neutralizer starting at the time @xmath32 , for stable isotopes ( @xmath22 ) we see that the neutralized released current takes the form @xmath41 .
\label{eq : flcw}\ ] ] in the case of the implantation distribution considered by melconian _ et al .
_ , we have @xmath42 .\ ] ] we will see in the following that we can operate in two modes .
if we let ions accumulate in cold yttrium for a certain time and then we turn on the neutralizer ( pulsed mode ) , all the implanted ions are released according to eqs .
( [ eq : fl0 ] ) or ( [ melc ] ) . instead , if we turn on the ionic beam with the neutralizer already working ( continuous mode ) , we are in the case described by eqs .
( [ eq : flcw ] ) or ( [ melc2 ] ) . in our model of the release process
, we neglect the contribution of desorption to the release time .
we can reasonably assume that , due also to the accuracy of such a model , discussed in section [ sz : contpuls ] , desorption contribution is well included in our uncertainty .
in any case , as also discussed in ref .
@xcite , the diffusion coefficients derived from our data analysis can be properly considered as upper limits .
after being released , atoms can be either directly detected in the vapor phase ( rubidium ) , or they can be first collected in a small volume ( about 1 mm@xmath43 ) in order to enhance our sensitivity ( rubidium and francium ) .
the confinement is provided by a magneto - optical trap in the standard configuration : six trapping and repumping laser beams respectively tuned on the d2 and d1 lines , and a constant gradient magnetic field provided by two anti - helmholtz coils .
the total number of atoms in the cell , when the six laser beams and the magnetic field are present , is the sum of two components : the number of trapped atoms @xmath44 and the number of atoms in the vapor phase @xmath45 which is supposed to be at thermal equilibrium with the cell walls .
the time evolution of @xmath44 and @xmath45 is well described by the coupled rate equations @xcite @xmath46 where @xmath47 is the rate related to the loading process ( the product @xmath48 represents the number of atoms per unit time loaded from the vapor to the mot ) , @xmath49 is the rate describing the collisional loss of atoms from the mot ( @xmath50 represents the number of trapped atoms per unit time which go back from trap to vapor phase due to collisions with background gas in the cell ) and @xmath51 is the rate characterizing the loss of atoms from the vapor phase ( @xmath52 represents the number of atoms per unit time definitively lost to the vacuum system or by chemisorption to the cell walls ) .
@xmath53 is the current of atoms released by the neutralizer and suitable for trapping .
we did not report the quadratic terms due to collisions between trapped atoms , which are negligible .
we can find the quasi - stationary state solution @xmath54 of eqs.([eq : rate ] ) under the condition @xmath55 ( slowly varying incoming current hypothesis ) : @xmath56 in practice , provided that @xmath53 varies with a time scale much longer than @xmath57 , we obtain that the detected number of trapped atoms and the total number of atoms in the vapor phase are proportional to the released current at any time .
therefore , the measurement of the trap signal or of the fluorescence signal directly gives all the information about diffusion inside the neutralizer .
this constitutes an innovative way of measuring the diffusion coefficient , with respect to other experiments that measure the release efficiency by detecting the radioactivity from the residual ions in the neutralizer .
at room temperature the rb vapor density is high enough to be directly detected through resonant excitation and looking at the atomic fluorescence . in order to minimize changes in the apparatus we made measurements by only switching off the trapping magnetic field , but keeping in place the six laser beams .
the laser frequency is scanned across the whole profile of the d2 rb line to search for the maximum signal .
fluorescence light is in this case collected on a si photodiode .
we tried different spectroscopic schemes and we obtained the best signal - to - noise ratio with the one described above . in the case of francium ,
it is not possible to directly observe the atomic vapor fluorescence , because the signals are extremely weak .
the solution is given by the mot , that allows to accumulate fr ( and also rb ) atoms in a small volume at low temperature .
in fact the signal increases by about two orders of magnitude because of doppler broadening elimination and , at low densities , far from saturated vapor pressures , by several orders of magnitude because of higher densities . as already stressed , we are able to detect traps of only a few tens of atoms .
atomic fluorescence from trapped atoms is detected by a ccd cooled camera ( fig .
[ fg : expsetup ] ) .
a dedicated software performs background subtraction , to reduce the contribution of scattered light to the noise , and real time correction for laser intensity fluctuations @xcite .
the detector was accurately calibrated and the software directly gives the number of atoms in the cold cloud .
when we turn on the ionic beam and send it on a hot neutralizer set at a constant temperature , we expect the spectroscopic signals to be well described by eqs .
( [ eq : flcw ] ) and ( [ melc2 ] ) ( fig . [ contrb ] ) .
the main advantage of this method is that the temperature of yttrium is kept constant during the measurement , and the vacuum conditions for the mot are very stable .
we analyzed rb data taken in the continuous regime for several temperatures .
the comparison of the results for the fits with eq .
( [ eq : flcw ] ) and with eq .
( [ melc2 ] ) showed an interesting feature , which allows us to estimate the impact of the implantation distribution on our results : while the two resulting curves are almost indistinguishable , assuming a sharp implantation distribution yields diffusion times that are systematically lower by a factor 0.7 with respect to the results obtained in the hypothesis of an extended initial distribution such as the one used by melconian _
et al._. this means that the accuracy for the measured diffusion coefficient is anyway limited at the 30% level by the theoretical model , also for previous experiments @xcite . )
( with an additional offset ) was fitted to the experimental data . ] in order to enhance the signals , it is possible to operate in a pulsed mode .
we let the ions accumulate in cold yttrium : at room temperature , the diffusion coefficient is very small , and the ions stay embedded in the neutralizer .
when the neutralizer is turned on , the atoms are released all together in a time scale of the order of @xmath58 , according to eq .
( [ eq : fl0 ] ) or ( [ melc ] ) . in practice , for both rb and fr we usually kept the ionic current on for 600 s ( which corresponds to more than three @xmath4fr radioactive half - lives ) .
this method gave excellent results to estimate the neutralizer release efficiency , and to determine the best operating temperature for our neutralizer .
most measurements have been done in the pulsed mode , because of the higher signal - to - noise ratio .
we obtained a set of data for each neutralizer operating temperature : the main goal was to optimize the yttrium temperature in order to obtain a good release efficiency . the function @xmath59
, corresponding to the flux @xmath60 ( eq . ( [ melc ] ) ) , was fitted to each acquired data set , with the free parameters @xmath61 , @xmath62 and @xmath63 ( fig .
[ fg : tempineutr1 ] ) .
the offset @xmath61 is due to the presence of rb impurities in yttrium : even with fresh neutralizers , never exposed to rb current , we detect rb vapor in the cell coming from the heated yttrium foil .
note that @xmath64 is not left as a free parameter , because it is totally correlated with the other parameters @xmath62 and @xmath63 : for any value @xmath65 , we can find the parameters @xmath66 and @xmath67 , such that @xmath68 .
therefore , @xmath64 is manually set to the time at which the signal begins rising , just after turning on the neutralizer .
since the neutralizer does not reach the operating temperature instantaneously , @xmath64 is affected by an error corresponding to the signal rising time .
the initial part of the curve critically depends on the way the neutralizer is turned on , and on the initial distribution of ions inside yttrium ( cf .
( [ eq : fl0 ] ) and ( [ melc ] ) ) .
therefore , we decided to consider data after a time @xmath69 , corresponding to a signal which is equal to 3/4 of the maximum signal for the whole curve ( fig .
[ fg : tempineutr1 ] ) . ) , at a temperature of 960 k. ] ) ) . ]
release time data are consistent with arrhenius law for the diffusion coefficient @xmath23 .
our experimental results are then fitted accordingly ( fig .
[ fg : tempineutr2 ] ) : @xmath70}\ ] ] where @xmath71 is the diffusion time at 1000 k ; fit parameters are both the time @xmath71 and the activation energy @xmath27 .
the choice of such a fit function lies in the fact that @xmath71 better describes the diffusion time in the temperature range of our measurements , while @xmath72 would be affected by a larger error , due to the extrapolation to infinite temperature .
results of fit are @xmath73 s and @xmath74 ev .
we used the trim code @xcite to evaluate the mean implantation depth for rb in y at 3 kev , which gives a value of 4.9 nm . from this data
we can then deduce the diffusion coefficient at 1000 k : @xmath75 we also performed measurements on rb using the mot detection method , to prepare the fr measurements and to test our diffusion model also for the mot signal evlution .
experimental data and fit curve for rb mot in pulsed regime are shown in fig .
( [ fg : motrb ] ) .
a good agreement was found between data and model , and also with the results obtained from vapor fluorescence detection described above . the fit procedure used in this case
is discussed in section [ sz : frmeas ] .
the diffusion time of francium in yttrium has been measured using the magneto - optical trap .
we operate in the pulsed mode in order to get higher signals : we use the same method as for rb , with detection of trapped atoms . with a collisional rate @xmath49 of about 0.5 s@xmath76
, we checked that the condition @xmath77 is verified for all the acquired data .
as compared to vapor fluorescence data , mot fluorescence is quite difficult to analyze in the pulsed mode , to deduce the diffusion time .
when we turn on the neutralizer , francium atoms come out of yttrium along with many other impurities already contained in the neutralizer ( as rb ) : vacuum suddenly deteriorates , and then slowly comes back to the equilibrium value . since the coefficient @xmath49 in eq .
( [ eq : ststsol ] ) strongly depends on vacuum quality , in the first part of the curve ( @xmath78 ) @xmath49 depends on time and it is not easy to interpret the data .
however , retaining only data with @xmath79 can not be a solution , because in this regime @xmath62 and @xmath63 are totally correlated ( it is not possible to distinguish them in the fitting procedure ) .
our compromise was to consider data after a time @xmath69 that corresponds to a signal which is equal to half the maximum signal of the whole curve ( fig .
[ fg : trapfit ] ) . for these data vacuum effects
are satisfactorily reduced and the order of magnitude is reasonable .
it was not always easy to reach convergence in fits : in practice , we kept @xmath62 and @xmath63 as free parameters , and we set the offset and @xmath64 to a reasonable value for each fit . in some cases ,
@xmath80 and @xmath62 were too correlated to obtain a good convergence : we had to repeat the fitting procedure with @xmath62 set by hand to realistic values .
the error for the obtained parameters includes several contributions : statistical error from the fit and uncertainties of @xmath61 and @xmath64 .
we saw in section [ sz : rbmeas ] that the methodology was checked with rb atoms .
the agreement between mot and vapor data demonstrates the validity of our measurements and of the fitting procedure .
we expect to be able to improve our results .
because the presented measurements were initially intended to test the final release efficiency , in terms of number of atoms @xmath31 rather than release time @xmath63 , we preferred to begin with the pulsed method , which gives higher signals and higher signal - to - noise ratios .
nevertheless , in our limited beam - time we managed to acquire data in the continuous regime for one neutralizer temperature : we remind that in this regime we expect no vacuum - related problems .
the analysis for the obtained curve gave very promising results , in agreement with the pulsed method . ) with @xmath81 .
the exposure time of the ccd camera was set to 1 s. ] in fig .
[ fg : timefit ] the results for the fitted parameter @xmath58 are reported as a function of temperature .
we performed release time measurements using also the @xmath82 decay rate signal , and we obtained similar results , at least as order of magnitude , thus confirming the consistency of our method .
fit results according to eq.([eq : taud ] ) are : @xmath83 an activation energy of @xmath84 ev was measured for @xmath85k in yttrium in ref .
@xcite , so our results for both fr and rb are consistent with literature .
we used the trim code @xcite to evaluate the mean implantation depth for fr in y at 3 kev , which gives a value of 5.1 nm .
the measured diffusion coefficient at 1000 k is then @xmath86 this is to our knowledge the first determination of diffusion time of francium ions in yttrium and the first time that diffusion parameters are derived from the study of the time evolution of the signal from a magneto - optical trap . according to ref .
@xcite it is possible to define a parameter @xmath87 , which takes into account the loss of atoms due to nuclear decay during the diffusion towards surface .
this is the so - called `` release fraction '' which gives the percentage of radioactive ions neutralized before the decay and then suitable for trapping .
in other words , one could interpret @xmath88 as a sort of neutralization efficiency . from the diffusion time measurement
we determine the release fraction as a function of temperature as shown in fig .
[ fg : relfract ] . and
@xmath71 given from our fit .
dashed curves give the uncertainty interval . ] note that our measurements of the diffusion coefficients for @xmath4fr allow us to predict the release fraction for @xmath89fr , which we also produce and trap , since we can reasonably suppose that diffusion features are identical for the two isotopes with negligible mass difference . because the apparatus was not optimized for the production of this isotope
, we did not perform systematic measurements on @xmath89fr .
nevertheless , our prediction of @xmath90 provides an estimation of the optimal conditions for the @xmath89fr neutralization : due to shorter lifetime of about 72 s , we expect @xmath89fr neutralization to be more efficient at higher temperatures with respect to @xmath4fr ( fig .
[ fg : relfract ] ) .
analog considerations can be applied to @xmath91fr : due to very similar radioactive lifetime , the curve calculated for @xmath91fr is indistinguishable from the @xmath4fr one .
we measured for the first time the diffusion parameters of francium and rubidium ions in yttrium .
the methodology that we developed takes advantage of laser spectroscopy techniques , and of the very high sensitivity offered by laser cooling and trapping .
since we directly measure diffusion times , our method can be considered complementary of release efficiency measurements by nuclear decay .
another advantage is the possibility to measure diffusion coefficients also for stable elements .
+ experimental data are consistent with the diffusive model considered , demonstrating the validity and the potential useful application of our method .
the diffusion coefficient is determined together with an activation energy consistent with previous measurements . for fr
, a release fraction of more than 80% is obtained for the 210 isotope at 1050 k , while for the short - lived 209 isotope we predict a release fraction around 70% at the same temperature . | we report the first measurement of the diffusion coefficients of francium and rubidium ions implanted in a yttrium foil .
we developed a methodology , based on laser spectroscopy , which can be applied to radioactive and stable species , and allows us to directly take record of the diffusion time .
+ francium isotopes are produced via fusion - evaporation nuclear reaction of a 100 mev @xmath0o beam on a au target at the tandem xtu accelerator facility in legnaro , italy .
francium is ionized at the gold - vacuum interface and fr@xmath1 ions are then transported with a 3 kev electrostatic beamline to a cell for neutralization and capture in a magneto - optical trap ( mot ) .
a rb@xmath1 beam is also available , which follows the same path as fr@xmath1 ions .
the accelerated ions are focused and implanted in a 25 @xmath2 m thick yttrium foil for neutralization : after diffusion to the surface , they are released as neutrals , since the y work function is lower than the alkali ionization energies .
the time evolution of the mot and the vapor fluorescence signals are used to determine diffusion times of fr and rb in y as a function of temperature . |
the phenomenon of coherent population trapping ( cpt ) , reviewed by arimondo in ref .
@xcite , and the phenomenon of electromagnetically induced transparency ( eit ) , reviewed in refs .
@xcite and @xcite , are two sides of the same coin .
both phenomena have been studied for a long time in three - level systems cpt in lambda type ; and eit in all the three kinds namely lambda type , ladder type , and vee type .
the physics underlying the two phenomena are related , but are also distinct in many ways .
the aim of this article is to highlight the differences between the two , partly to eliminate the confusion existing in the literature .
some of this confusion arises due to the use of the term eit for what are really cpt processes .
cpt was first observed as a decrease in the fluorescence emission from na atoms in a vapor cell @xcite .
it was understood to arise from the fact that the atoms were being optically pumped into a dark non - absorbing state by the excitation beams .
once pumped , atoms were trapped in that state because there was no way to spontaneously decay from there hence the name * coherent population trapping*. the dark state was created by destructive interference between excitation pathways from two stable levels to a common excited level .
the three levels form the required @xmath3-type system , as shown in fig .
[ lambda ] .
it is customary to call the laser driving the @xmath4 transition as the * probe * laser , and the one driving the @xmath5 transition as the * control * laser .
both transitions are electric - dipole ( e1 ) allowed transitions ; and thus the transition @xmath6 is e1 forbidden .
the detunings are labeled by @xmath7 s , the powers by the respective rabi frequencies @xmath8 s , and the spontaneous decay rates by @xmath9 s . @xmath10 and @xmath11 ( i.e. for the e1 allowed transitions ) are about 110 mhz , while @xmath12 for the forbidden transition is close to 0 . in reality
, there is decoherence between the levels due to inter - atomic collisions , collisions with walls , spontaneous scattering through the upper level , etc .
it is important to note that the two lasers are required to be * phase coherent * so that the dark superposition state can be formed .
eit works on a slightly different principle .
the phenomenon is usually studied in the regime where the control laser is * strong * and the probe laser is * weak * , i.e. @xmath13 . under these conditions , the control laser shifts the energy levels of the atom away from line center through the ac stark effect , which can also be understood as the creation of new * dressed states * of the coupled atom - photon system @xcite .
the shift is equal to the rabi frequency of the control laser .
the absorption of the probe laser splits into a classic autler - townes doublet with a splitting equal to the control rabi frequency , and shows enhanced transparency at line center a transparency induced by the control laser .
hence the name * electromagnetically induced transparency*. quantum interference also plays a role in eit , but mainly when the control rabi frequency is of the order of the excited - state linewidth . the resulting ( destructive ) interference between the transition amplitudes from the ground state to the control - induced dressed states is particularly important in lambda systems , as discussed in ref .
the interference then leads to the probe absorption being identically zero at line center .
dressed - state interference is not so important in vee and ladder systems @xcite .
in addition , since the probe photon is not involved in the interference , the probe beam in eit experiments can be ( and is ) generally derived from a * phase independent * laser .
the above analysis shows that both eit and cpt are two - photon processes , because the transitions involve two different levels .
however , in cpt both the probe and control photons are equally important .
by contrast , in eit the transparency is entirely created by the strong control laser , while the weak probe laser only plays the role of measuring this transparency . with these basic differences in mind ,
let us now contrast the two phenomena in terms of various experimental parameters . 1
. * scan axis . *
an important difference between eit and cpt is the scan axis . in cpt
, this is the relative detuning between the two beams .
if the experiment is done using magnetic sublevels of a degenerate transition , then the frequency difference between the two beams is only of the order of few mhz .
the required phase coherence can then be achieved using an acoustic - optic modulator ( aom ) , and the scan axis is the frequency of the aom driver .
+ the cpt phenomenon can also be used to access the clock transition in alkali atoms like rb and cs . in this case , the probe and control beams differ in frequency by the ground hyperfine interval , which is of order few ghz and hence too large to be achieved with an aom .
therefore , it is done in one of the two following ways : a. taking two independent lasers and beating their outputs on a sufficiently fast photodiode , as has been done in ref .
@xcite . in this case , the scan axis is the frequency of the beat signal .
b. taking one laser and feeding its output through an electro - optic modulator ( eom ) .
the eom driver is set to half the clock frequency because it produces two sidebands spaced apart by twice the eom frequency .
the scan axis is the frequency of the eom driver .
+ on the other hand , the scan axis in eit is the frequency of the phase - independent probe laser .
if the transition is in the optical regime , the frequency is of order @xmath14 hz , or 6 to 9 orders - of - magnitude larger than that for cpt .
2 . * subnatural resonance the relevant natural + linewidth @xmath9 . *
the discussion of the scan axis brings us to the question of what defines whether the feature is subnatural or not . in cpt experiments ,
the scanning device is at the difference frequency between the two lower levels .
the relevant natural linewidth is therefore the linewidth of transition between these two levels , which as mentioned before is close to zero because the transition is electric - dipole forbidden .
the linewidth of the upper level does not enter the picture .
as an example , we consider the observation of a resonance linewidth of 42 hz for coherent dark resonances in a cs vapor cell filled with buffer gas @xcite .
the resonance is on the ground hyperfine splitting of @xmath15cs the clock transition at 9.1 ghz used in the si definition of the second .
the experiments were done on the @xmath16 line of cs , but the 5 mhz linewidth of the upper state does not enter the picture .
nowhere in the above work do the authors call their resonance * subnatural * , because it is well known that the natural linewidth of the clock transition is well below 1 hz
. similarly , the phase - locked lasers used in the above study have a linewidth of order 1 mhz .
this does not prevent them from seeing a 42 hz feature because they are looking for a beat signal between two phase - coherent lasers , which can be much narrower than the linewidth of the laser .
this shows that the cpt phenomenon can be used for precision spectroscopy on the ground levels @xcite .
+ on the other hand , we have shown that the narrow features of eit are really * subnatural * and can be used for high - resolution spectroscopy of the upper level @xcite .
this is because the scan axis is the optical frequency of the probe laser , and the relevant natural linewidth is the linewidth of the upper level . in the case of the rb @xmath16 line used in that work ,
the upper level had a linewidth of 6 mhz .
therefore , any feature that is narrower than 6 mhz would be called subnatural . the narrowest feature that we have reported is 0.85 mhz or @xmath17 @xcite .
this corresponds to a @xmath18 of @xmath19 , which is better than the @xmath18 of @xmath20 for the cpt experiment on the cs clock transition in ref .
* fluorescence versus absorption .
* one important consequence of the fact that cpt results from the creation of a dark state is that there is a concomitant decrease in the fluorescence from the cell it becomes * dark*. recall that the first observation of cpt was the appearance of a dark region in the fluorescence from a na vapor cell @xcite . in that study ,
an inhomogeneous magnetic field was applied along the axis of the cell , so that the dark state was created only in a small region . as a result ,
the cpt phenomenon was seen as a dark line in a bright cell . by contrast , in eit , the strong control laser is always being absorbed ( and atoms are fluorescing ) , and the induced transparency is seen only in the absorption signal of the weak probe laser .
therefore , at least to first order , there is no change in the fluorescence whether the probe laser is on or off .
* effect of buffer gas . *
because cpt is a ground - state coherence phenomenon , any technique that increases the coherence time will give a narrower linewidth .
one of the most common methods is to use a buffer gas in the vapor cell typically a few torrs of a gas like ne or n@xmath1 .
for example , the cpt experiment on the clock transition in ref .
@xcite was done in a cs vapor cell filled with n@xmath1 buffer gas . on the other hand ,
the use of such cells for eit experiments actually kills the signal , because the buffer gas considerably broadens the normal probe absorption signal due to collisions and swamps any modification due to the control .
power in the two beams*. as mentioned before , both beams play an equally important role in cpt , while eit is studied in the regime where the control is strong and the probe is weak .
this leads to an important difference
there is not much point in increasing the control power beyond the probe power in cpt .
the role of control power is to increase the linewidth due to increased decoherence through the upper level*power broadening*which increases the spontaneous scattering rate , and the narrowest feature is obtained when the two powers are equal .
* effect of detuning . *
the upper level in cpt causes decoherence ( and hence broadens the resonance ) by inducing real transitions from the lower levels .
this transition rate ( and hence linewidth ) can be reduced by detuning from the level .
but the reduced transition probability also results in a smaller signal strength and signal - to - noise ratio ( snr ) .
note that the cpt resonance still occurs at the same * relative * detuning between the two beams , which is the point at which the two - photon raman resonance condition is satisfied .
on the other hand , detuning the control laser from the upper level in eit causes the resonance to shift within the absorption profile of the probe laser .
the resonance again appears where the probe detuning matches the control detuning , but the detuned * lineshape * is very different .
many of the above points will be validated below through experiments on the @xmath16 line of @xmath2rb .
the relevant energy levels of the @xmath2rb d@xmath1 line ( @xmath21 transition ) at 780 nm used for the eit and cpt experiments are shown in fig .
[ rubidium_87_levels ] .
the required @xmath22-type system is formed using magnetic sublevels of the @xmath0 transition along with opposite circular polarizations for the control and probe beams .
the natural linewidth of the upper state is 6 mhz , and the saturation intensity the intensity at which the transition gets power broadened by a factor of @xmath23 is given by ( page 209 of ref .
@xcite ) @xmath24 where @xmath25 nm is the wavelength of the transition and @xmath26 ns is the lifetime of the excited state @xcite .
the value for this level is 1.67 mw/@xmath27 .
the resultant scheme is shown in fig .
[ magneticlevels ] .
the figure also shows that applying a magnetic field is equivalent to detuning the beams from resonance .
thus , the effect of detuning can be studied by applying a magnetic field .
the experimental setup for the cpt experiment is shown schematically in fig .
[ schema](a ) .
the required phase coherence for the control and probe beams is obtained by deriving both beams from a single laser , and using aoms to produce the required frequency shift .
the laser consists of a grating stabilized diode laser system , as described in ref .
the linewidth of the laser after stabilization is 1 mhz .
the output beam is elliptic , and has a size of 3 mm @xmath28 4 mm .
since magnetic sublevels are used , the two beams are required to be at the same frequency when no field is present .
this is achieved by using two aoms in the path of the control beam
one aom with a downshift of 180 mhz , and the other compensating for this shift by having a double - passed upshift of 90 mhz .
the double passing ensures that the direction of the beam does not change when it is scanned .
the frequency of the aoms is set using function generators .
\(a ) ( b ) the power in the beams is controlled using @xmath29 waveplates in front of respective pbss .
the two beams have orthogonal linear polarizations , and are mixed using a pbs . their polarizations are changed to circular using a @xmath30 waveplate in front of the vapor cell .
the vapor cell is cylindrical with dimensions of 25 mm diameter @xmath31 50 mm length .
the cell is inside a three - layer magnetic shield that reduces stray fields to @xmath32 mg .
two kinds of cells are used
one that is pure and contains both isotopes in the natural abundances , and the second that contains only @xmath2rb with 20 torr of ne as buffer gas .
the laser is locked to the @xmath0 transition using the saturated absorption spectroscopy ( sas ) signal from another vapor cell .
this ensures that the ( unshifted ) probe beam is locked to the @xmath33 transition .
the control beam is scanned around the @xmath34 transition by scanning the frequency of the double - passed aom .
the polarizations after the cell are made linear using a second @xmath30 waveplate , and the beams are separated using another pbs .
the probe beam alone is detected using a photodiode therefore , the photodiode signal is proportional to probe transmission .
since the sas signal used for locking corresponds to absorption by zero - velocity atoms , detecting the non - scanning probe beam allows us to have a flat doppler - free background for the cpt signal .
the schematic for the eit experiments is shown in fig .
[ schema](b ) .
this is essentially the same as that used in the cpt experiments , except that the two beams are derived from independent laser systems . in this case , the control laser is locked to the @xmath35 transition using an sas signal , which ensures that the control beam is resonant with the @xmath36 transition .
the probe laser is scanned around the same @xmath0 transition by varying the angle of the grating , which means that the eit spectrum appears on a doppler pedestal .
the absorption through a normal vapor cell is about 5% .
therefore , the photodiode signal is multiplied suitably to get this degree of absorption off resonance .
the absorption through a buffer - gas filled cell is about a factor of 2 lower .
hence the photodiode signal is multiplied by a suitably smaller number .
as mentioned above , laser locking to the relevant hyperfine transition was done using an sas signal .
the sas spectrometer consists of a low - power probe beam ( @xmath37 20 w ) , whose absorption is saturated by a strong pump beam ( @xmath38 2 mw ) which is counter - propagating with respect to the probe . probe transmission then shows doppler - free peaks at the location of various hyperfine transitions ( page 309 of ref .
this is because only zero - velocity atoms are resonant with both beams . in order to make the spectrum appear on a flat background
, the signal from a second identical probe beam is subtracted from the main one .
the resulting spectra for both kinds of cells are shown in fig .
the spectrum obtained in a pure cell shows six peaks three corresponding to real hyperfine transitions , and three spurious `` crossover '' resonances .
the crossover resonances appear exactly halfway between two real peaks , and occur because of interaction with non - zero - velocity atoms in the vapor cell .
the spectrum obtained in a buffer cell , taken under identical conditions , shows a complete lack of any peaks .
this shows unequivocally that such a cell can not be used for eit experiments .
however , as we will see later , they are quite useful for cpt experiments , because collisions with a buffer gas increase the coherence time and result in a significant reduction of the linewidth . the cpt resonance obtained in a pure cell as a function of the relative ( raman ) detuning between the beams is shown in fig .
[ cptpure ] .
the probe transmission signal shows a peak when the detuning is zero the point at which the dark state is formed .
the linewidth of the resonance is 15 khz , which is not * subnatural * because the natural linewidth between the lower levels is close to 0 .
the observed linewidth is actually determined by non - radiative decoherence processes such as background collisions , spontaneous scattering through the upper level , and transit time across the beams .
the 1 mhz linewidth of the laser does not prevent the observation of such a narrow linewidth because the path length difference between the probe and control beams when they reach the atoms is negligible ; certainly orders - of - magnitude less than the coherence length of the laser .
the signal level away from resonance ( background level ) is flat , which is because the probe beam is locked and hence only addresses zero - velocity atoms .
the eit resonance as a function of probe detuning is shown in fig .
[ eitpure ] .
the resonance also appears as a peak in the probe transmission spectrum , but there are important differences from the cpt case .
first , the linewidth is 3.9 mhz , which is * subnatural * given the 6 mhz linewidth of the excited state . secondly , the background level away from resonance has a doppler profile because the probe beam is being scanned . in cpt experiments ,
the control beam causes decoherence through the upper level by increasing the scattering rate .
the scattering rate is given by @xmath39 the above equation shows that the scattering rate increases initially and then reaches a saturation value as the intensity is increased .
therefore , the linewidth will show a similar behavior and reach a saturation value when the control intensity is high .
the results are shown in fig .
[ powervar](a ) .
the solid line is the fit to eq . , and shows that the measured data follows the expected behavior .
however , the saturation intensity obtained from the fit is about 0.5 mw/@xmath27 , which neither corresponds to that of the upper state nor to that determined by the decoherence rate between the lower levels , but is some combination of the two . on the other hand
, the strong control beam in eit experiments causes power broadening of the upper level . for a control intensity of @xmath40 ,
the power - broadened linewidth is given by @xmath41 where @xmath42 is the natural linewidth and @xmath43 is the saturation intensity for the transition .
thus , as the control power increased , the linewidth of the upper level , and consequently the linewidth of the eit resonance , increases .
these results are shown in fig .
[ powervar](b ) .
the solid line is a fit to the above equation , which yields a value of @xmath44 mw/@xmath27 .
this value is close to the 1.67 mw/@xmath27 saturation intensity of the upper state , which is as expected because the dressed states created by the control laser involve the upper state . as seen from the magnetic sublevel structure shown in fig .
[ magneticlevels ] , the two beams get detuned in the presence of a magnetic field . for cpt , the scattering rate given by eq . in the presence of detuning @xmath45 changes to @xmath46
this shows that the scattering rate decreases as the detuning is increased hence , the linewidth of the cpt resonance will also decrease .
the results are shown in fig .
[ cpt_detuning ] .
the solid line , which is a fit to the above equation , shows that the reduced scattering rate describes the data well . on the other hand ,
the lineshape of the detuned eit resonance becomes asymmetric and non - lorentzian .
this is because the detuned eit resonance moves inside the absorption profile .
the lineshape of the resonance , shown in fig .
[ eit_detuning ] , confirm this explanation .
it is incorrect to define a linewidth for the asymmetric lineshape , and hence it is not done .
the presence of a buffer gas in the vapor cell results in an increase in the coherence time of the lower levels .
this is advantageous for cpt experiments because it results in a narrower linewidth .
but , as mentioned before , such cells give no eit signal .
therefore , only cpt results are shown in fig .
[ cptbuffer ] .
the linewidth of the cpt resonance decreases from 15 khz in a pure cell to 8 khz in a buffer cell .
as mentioned before , the absorption through such a cell off resonance is about 3% .
hence the photodiode signal is adjusted to reflect this .
we have verified that cpt resonances can be seen in the fluorescence signal from the cell , while eit is not .
but the detection ( with a photomultiplier tube ) is complicated by the fact that the cell is inside a multilayer magnetic shield .
in addition , the powers in the beams have to be increased to get a good signal , which results in a large increase in the resonance linewidth .
hence the curve is not shown .
just like two phase - coherent beams can put the atom in a dark state , a similar arrangement can be used to create a bright state .
the result is enhanced absorption at line center , exactly opposite to the reduced absorption seen in cpt .
the first observation of this was in the closed @xmath47 transition in @xmath48 @xcite .
the authors called the phenomenon electromagnetically induced absorption ( eia)to highlight the fact that there was enhanced absorption on resonance .
however , we feel that a more appropriate term would be cbs standing for * * c**oherent - trapping in a * * b**right * * s**tate while the term eia should be reserved for enhanced absorption of a weak probe laser in the presence of two or more strong control beams in multilevel systems , as demonstrated by us in refs . @xcite and others in refs
. @xcite .
the conditions for observing cbs on a @xmath49 transition are : a. it is a closed transition .
b. @xmath50 . c. @xmath51 , so that there are multiple magnetic sublevels in the ground state .
all these conditions are met for the @xmath52 transition in @xmath53 .
a detailed study of cbs results will be presented in a future publication .
the experimentally observed line shapes for both cpt and eit can be obtained from a density matrix analysis of the three - level lambda system . in terms of the various density matrix elements ,
the time evolution is given by the following set of equations : @xmath54 there are only 6 equations for the 9 elements because the off - diagonal elements are complex conjugates of each other .
the equations are solved in steady state , i.e. @xmath55 .
parameters corresponding to the experimental situation are : @xmath56 khz , @xmath57 mhz , @xmath58 khz , @xmath59 mhz , and @xmath60 mhz . for cpt
, the quantity of interest is the upper state occupation , given by @xmath61 .
the above equations are solved numerically , so that all orders of rabi frequency are taken into account .
the spectrum is doppler averaged over the maxwell - boltzmann velocity distribution corresponding to room temperature vapor . the experimental cpt spectrum in fig . [ cptpure ]
is obtained with equal control and probe powers of 30 w , which corresponds to a maximum intensity at the center of the gaussian beam(s ) of 0.64 mw/@xmath27 .
this is reduced by a factor of 10 so that the spectrum does not change significantly when the upper state linewidth @xmath62 is changed , which is required because the dark state does not involve the upper state .
the other thing needed in the calculation is the decoherence rate between the lower levels .
this is taken to be 1 khz , adjusted so that the theoretical linewidth is comparable to the experimental one .
this value is reasonable for the wall - collision rate in our cell . the resulting spectrum calculated with above parameters is shown in fig . [ cpttheory ] . for eit
, the absorption of the probe beam is determined by the imaginary part of the coherence between levels @xmath63 and @xmath64 , and is therefore given by @xmath65 . since eit is studied in the weak probe regime , the density matrix equations are solved to first order in the probe rabi frequency .
the steady state solution is given by : @xmath66 for the experimental spectrum shown in fig .
[ eitpure ] , the control power is 10 mw .
this gives a maximum intensity at the center of the gaussian beam of 212.3 mw/@xmath27 .
taking 50% of this value as the average intensity over the entire beam , we get an intensity of 106.1 mw/@xmath27 .
this corresponds to a rabi frequency of @xmath67 . the resulting spectrum calculated with value of @xmath68 is shown in fig . [ eittheory ] .
there are two things to note about the calculated spectrum : a. the doppler averaging results in significant linewidth reduction , as demonstrated by us in ref .
in fact , for this value of rabi frequency , the autler - townes doublet peaks are separated by @xmath69 , which results in an eit resonance that is neither subnatural nor lorentzian .
this is evident from the inset shown in the figure .
b. the observed spectrum is a convolution of the eit spectrum with the linewidth of the probe laser .
this means that 1 mhz should be added to the calculated linewidth , since this is not accounted for in the calculation . with this correction
, the theoretical and experimental linewidths agree quite well .
to conclude , we have highlighted the differences between the related phenomena of coherent population trapping , which requires two phase - coherent beams of roughly equal power , and electromagnetically induced transparency , which requires independent lasers with large power in the control beam and negligible power in the probe beam .
the literature is full of cases where the term eit is used for experiments that are of the cpt kind .
perhaps this is because of the pizazz associated with the phrase eit .
but an immediate clue as to the kind of experiment can be had from the number of lasers used
one laser implies cpt while two lasers imply eit .
the two phase - coherent beams from the same laser required for cpt are then produced by modulation , either acousto - optic or electro - optic .
cpt experiments gain from the use of vapor cells filled with buffer gas because this increases the ground - state coherence time and hence narrows the linewidth of the resonance , whereas eit can not be seen in such a cell . under the right conditions ,
the control laser in eit leads to subnatural resonances for probe absorption .
we have recently shown that the use of a control beam with a laguerre - gaussian profile instead of the usual gaussian leads to a further narrowing with an unprecedented linewidth of @xmath70 being obtained @xcite .
this work was supported by the department of science and technology , india .
s k acknowledges financial support from inspire fellowship , department of science and technology , india . the authors thank s raghuveer for help with the manuscript preparation , and e. arimondo for helpful discussions and a critical reading of the manuscript .
vn conceived the study and wrote the initial draft of the paper ; sk and mpk did the experiments ; sk and vb did the theoretical simulation ; all authors reviewed the manuscript for intellectual content , and read and approved the final version .
r. f. gutterres , c. amiot , a. fioretti , c. gabbanini , m. mazzoni , and o. dulieu .
determination of the @xmath71 @xmath72 state dipole matrix element and radiative lifetime from the photoassociation spectroscopy of the @xmath73 @xmath74 long - range state .
, 66:024502 , 2002 . | we discuss the differences between two well - studied and related phenomena
coherent population trapping ( cpt ) and electromagnetically induced transparency ( eit ) .
many differences between the two such as the effect of power in the beams , detuning of the beams from resonance , and the use of vapor cells filled with buffer gas are demonstrated experimentally .
the experiments are done using magnetic sublevels of the @xmath0 transition in the d@xmath1 line of @xmath2rb . + * keywords * : electromagnetically induced transparency ; coherent population trapping ; coherent control ; quantum optics . |
let @xmath1 denote a sequence of real orthonormal polynomials with respect to the weight function @xmath2 on the interval @xmath3 ( see e.g. @xcite ) , i.e. @xmath4 the distribution of these polynomials along the orthogonality interval can be complementarily measured by means of the spreading properties of the normalized - to - unity density function @xmath5=p_n^2 ( x ) \omega(x ) , \label{eq : rakhmanov_density}\ ] ] which is called rakhmanov s density of the polynomial @xmath6 , to honor the pioneering work @xcite of this mathematician who has shown that this density governs the asymptotic @xmath7 behaviour of the ratio @xmath8 .
physically , this probability density characterizes the stationary states of a large class of quantum - mechanical potentials @xcite . beyond the variance ,
the spreading of the orthogonal polynomials is best analyzed by the information - theoretic properties of their associated rakhmanov probability densities ; namely , the fisher information @xcite , the renyi entropy @xcite and the shannon entropy @xcite .
the information - theoretic knowledge of the orthogonal polynomials up to 2001 is reviewed in ref @xcite , where the quantum - mechanical motivation and some physical applications are also given , and up today in @xcite , where emphasis is made on asymptotics .
therein , it is pointed out that the study of the information - theoretic measures of orthogonal polynomials was initiated in the nineties with the asymptotic computation of the renyi and shannon entropies of the classical orthogonal polynomials @xcite . up until now , however , the explicit expressions of these two spreading measures are not known save for the shannon measure for some particular subclasses of the jacobi polynomials ; namely the chebyshev polynomials of first and second type @xcite and the gegenbauer polynomials @xmath9 with integer parameter @xmath10 @xcite . on the other hand ,
the fisher information for all classical orthogonal polynomials on a real interval ( hermite , laguerre , jacobi ) has been calculated in a closed form in 2005 @xcite .
this has allowed us to find , more recently , the cramer - rao information plane ( i.e. the plane defined by both the fisher information and the variance ) for these systems @xcite .
this paper has two main purposes .
first , to introduce a more appropriate set of spreading measures ( to be called spreading lengths of renyi , shannon and fisher types ) in the field of orthogonal polynomials following the ideas of m.j.w .
hall @xcite .
the spreading lengths are direct measures of position spread of the variable @xmath11 , as the root - mean - square or standard deviation @xmath12 .
so they have the following properties : same units as @xmath11 , invariance under translations and reflections , linear scaling with @xmath11 and vanishing in the limit as @xmath13 approaches a delta function .
the second purpose is to compute all these lengths for the hermite polynomials . in this sense
we first compute their moments around the origin and , as well , the standard deviation of these objects .
then we compute the entropic or frequency moments @xcite of these polynomials by means of the bell polynomials , so useful in combinatorics @xcite .
this achievement allows us to obtain the renyi lengths of hermite polynomials by means of an efficient algorithm based on the known properties of bell s polynomials . with respect to the shannon length ,
since its explicit expression can not be determined , we obtain its asymptotics and we derive a family of sharp upper bounds by use of an information - theoretic - based optimization procedure .
moreover , the fisher information is explicitly calculated .
finally , these information - theoretic quantities are applied to the computation of the spreading lengths of the main quantum - mechanical prototype of physical systems : the harmonic oscillator .
the structure of the paper is the following . in section [ sec2 ]
we describe the spreading lengths for the orthogonal polynomials @xmath14 and discuss their properties . in section [ sec3 ] the computational methodology to obtain the entropic moments and the associated renyi lengths of the hermite polynomials is developed by use of the bell polynomials .
in addition , the moments with respect to the origin and the fisher length are explicitly given .
furthermore , the asymptotics of the hermite polynomials is obtained and a family of sharp bounds are variationally determined . in section [ sec4 ]
the spreading lengths of hermite polynomials is numerically discussed and a linear correlation with the standard deviation is computationally discovered . in section [ sec5 ] the previous results
are applied to the quantum - mechanical harmonic oscillator .
finally , some open problems about the asymptotics of entropic moments are posed , our conclusions are given and an appendix with the computational approach of the power of an arbitrary polynomial by use of the bell polynomials is included .
here we shall refer to the hypergeometric - type polynomials , i.e. the polynomials @xmath14 defined by the orthogonality relation given by eq .
( [ eq : orthonormality ] ) .
it is well - known that they can be reduced to one of the three classical families of hermite , jacobi and laguerre by appropriate linear changes of the variable @xcite .
the spreading of these polynomials along their orthogonality interval @xmath15 can be measured by means of ( i ) the moments around a particular point of the orthogonality interval ; usually it is chosen the origin ( moments around the origin : @xmath16 ) or the centroid @xmath17 ( central moments or moments around the centroid : @xmath18 ) @xcite and ( ii ) the frequency moments @xmath19:=\langle \left [ \rho_n(x)\right ] ^k\rangle$ ] @xcite of the associated rakhmanov density @xmath13 given by eq .
( [ eq : rakhmanov_density ] ) .
the symbol @xmath20 denotes the expectation value of @xmath21 with respect to the density @xmath13 as @xmath22 these two classes of ordinary and frequency moments are complementary spreading measures of global type . while the ordinary moments measures the distribution of the probability with respect to a particular point of the support interval @xmath23 , the frequency moments measure the extent to which the probability is in fact distributed .
so it happens that the latter moments are , at times , much better probability estimators than the ordinary ones @xcite .
moreover , they are fairly efficient in the range where the ordinary moments are fairly inefficient @xcite ; see also the brief , recent summary about these quantities done in ref . @xcite .
therein we learn that the frequency moments are also called `` entropic moments '' because they are closely connected to the renyi entropies @xcite @xmath24:=\frac{1}{1-q}\ln \langle [ \rho_n(x)]^{q-1}\rangle;\quad q>0;\quad q\neq 1 , \label{eq : def_renyi}\ ] ] and the tsallis entropies @xcite @xmath25:=\frac{1}{q-1}\left [ 1-\langle [ \rho_n(x)]^{q-1}\rangle\right];\quad q>0 ; \quad q\neq 1 , \label{eq : def_tsallis}\ ] ] as well as to other information - theoretic quantities , such as the brukner - zeilinger entropy @xcite and the linear entropy @xcite . among the first class of moments we highlight the familiar root - mean - square or standard deviation @xmath26 given by @xmath27 because it is a direct measure of spreading @xcite in the sense of having the same units as the variable , and has various interesting properties : invariance under translations and reflections , linear scaling ( i.e. , @xmath28 for @xmath29 ) and vanishing as the density approaches a dirac delta density ( i.e. in the limit that @xmath11 tends towards a given definite value ) . from the second class of moments we fix our attention on the renyi lengths @xcite defined by @xmath30\equiv \exp\left(r_q[\rho_n]\right ) = \langle \left[\rho_n(x)\right]^{q-1}\rangle^{-\frac{1}{q-1}}=\left\{\int_{\delta}dx[\rho_n(x)]^{q }
\right\}^{-\frac{1}{q-1}},\quad q>0 , \quad q\neq 1 , \label{eq : def_renyi_length}\ ] ] which are not only direct spreading measures but also have the three above - mentioned properties of the standard deviation . among them
we highlight the second - order renyi length ( also called onicescu information @xcite , heller length @xcite , disequilibrium @xcite and inverse participation ratio in other contexts @xcite ) @xmath31:= l_2^r[\rho_n]=\langle \rho_n(x)\rangle^{-1}=\left\ { \int_{\delta } dx [ \rho_n(x)]^2\right\}^{-1},\ ] ] and , above all , the shannon length @xcite @xmath32:= \lim_{q\to 1 } l_q^r[\rho_n]=\exp\left ( s[\rho_n]\right ) \equiv
\exp \left[-\int_{\delta } dx \rho_n(x ) \ln\rho_n(x)\right],\ ] ] where @xmath33 $ ] is the shannon information entropy @xcite .
there exists a third class of spreading measures for the classical orthogonal polynomials , which are qualitatively different in the sense that , in contrast to the previous ones , they have a local character .
indeed they are functionals of the derivative of the associated rakhmanov density @xmath13 , so that they are very sensitive to local rearrangements of the variable .
the most distinctive measure of this class is the so - called fisher length @xcite defined by @xmath34}}\equiv \left < \left [ \frac{d}{dx}\ln\rho_n(x)\right]^2\right>^{-\frac{1}{2}}=\left\{\int_{\delta}dx\frac{[\rho_n'(x)]^2}{\rho_n(x)}\right\}^{-\frac{1}{2 } } , \label{eq : def_deltax}\ ] ] where @xmath35 $ ] denotes the fisher information of the classical orthogonal polynomials @xcite .
this quantity measures the pointwise concentration of the probability along the orthogonality interval , and quantifies the gradient content of the rakhmanov density providing ( i ) a quantitative estimation of the oscillatory character of the density and of the polynomials and ( ii ) the bias to particular points of the interval , so that it measures the degree of local disorder .
it is worthy to remark that the fisher length , as the heisenberg and renyi lengths , is a direct spreading measure and has the three properties of translation and reflection invariance , linear scaling and vanishing when the density tends to a delta density .
in addition , the fisher length is finite for all distributions @xcite .
moreover , the direct spreading measures just mentioned have an uncertainty / certainty property : see refs .
@xcite for the heisenberg , shannon and renyi cases , and @xcite for the fisher case
. finally , they fulfill the inequalities @xcite @xmath36\leq ( 2\pi e)^{\frac{1}{2}}(\delta x)_n , \label{eq : delta_inequalities}\ ] ] where the equality is reached if and only if @xmath13 is a gaussian density .
in this section we calculate the moments - with - respect - to - the - origin @xmath37 , @xmath38 , the entropic moments @xmath39\equiv \langle[\rho_n(x)]^q\rangle$ ] , of integer order @xmath40 , and the renyi , shannon and fisher spreading lengths of the rakhmanov density of the orthonormal hermite polynomials @xmath41 given by @xmath42 with @xmath43 where the last factor vanishes when @xmath44 and @xmath45 have opposite parities .
these polynomials are known to fulfil the orthonormality relation ( [ eq : orthonormality ] ) with the weight function @xmath46 on the whole real line @xcite .
thus , according to eq .
( [ eq : rakhmanov_density ] ) , the expression @xmath47(x)=e^{-x^2 } \tilde{h}_n^2(x ) \label{eq : rho_n_hermite}\ ] ] gives the rakhmanov density of the orthonormal hermite polynomial @xmath41 .
the moments of the hermite polynomials @xmath41 are defined by the moments of its associated rakhmanov density ( [ eq : rakhmanov_density ] ) ; that is @xmath48 the use of ( [ eq : def_h_alternative ] ) , ( [ eq : orthonormality ] ) and ( [ eq : def_xk ] ) gives @xmath49 which can be also obtained from some tabulated special integrals @xcite . in the particular cases
@xmath50 and @xmath51 we obtain the first few moments around the origin : @xmath52 so that the second central moment of the density @xmath53 describes the standard deviation .
the @xmath40th - order frequency or entropic moment of the hermite polynomials @xmath41 is defined by the corresponding quantity of its associated rakhmanov density given by eq .
( [ eq : rakhmanov_density ] ) ; that is , @xmath54=\left\langle \left\{\rho_n\right\}^{q-1}\right\rangle=\int_{-\infty}^\infty \left\{\rho_n(x)\right\}^q dx = \int_{-\infty}^\infty e^{-qx^2}\{\tilde{h}_n(x)\}^{2q}dx;\ ; q\ge 1 \label{eq : def_wq_hermite}\ ] ] the evaluation of this quantity in a closed form is not at all a trivial task despite numerous efforts published in the literature @xcite save for some special cases . here
we will do it by means of the bell polynomials which play a relevant role in combinatorics @xcite .
we start from the explicit expression ( [ eq : def_h_alternative ] ) for the hermite polynomial @xmath41 ; then , taking into account the appendix below we have that its @xmath55th - power can be written down as @xmath56^p=\left[\sum_{k=0}^n c_kx^k\right]^p = \sum_{k=0}^{np } \frac{p!}{(k+p)!}b_{k+p , p}(c_0,2!c_1,\ldots,(k+1)!c_k ) x^k , \label{eq : hnp_bell_expanssion}\ ] ] with @xmath57 for @xmath58 , and the remaining coefficients are given by eq .
( [ eq : def_c_l ] ) .
moreover , the bell polynomials are given by @xmath59 where the sum runs over all partitions @xmath60 such that @xmath61 the substitution of the expression ( [ eq : hnp_bell_expanssion ] ) with @xmath62 into eq .
( [ eq : def_wq_hermite ] ) yields the value @xmath63&=&\sum_{k=0}^{2nq } \frac{(2q)!}{(k+2q)!}b_{k+2q,2q}(c_0,2!c_1,\ldots,(k+1)!c_k ) \int_{-\infty}^\infty e^{-qx^2}x^kdx\nonumber\\ & = & \sum_{j=0}^{nq}\frac{\gamma\left(j+\frac12\right)}{q^{j+\frac12}}\frac{(2q)!}{(2j+2q ) ! } b_{2j+2q,2q}(c_0,2!c_1,\ldots,(2j+1)!c_{2j } ) , \label{eq : wqrhon_bell}\end{aligned}\ ] ] where the parameters @xmath64 are given by eq .
( [ eq : def_c_l ] ) , keeping in mind that @xmath57 for every @xmath58 , so that the only non - vanishing terms correspond to those with @xmath65 so that @xmath66 for every @xmath58 . in the particular cases
@xmath67 and 2 , we obtain the value @xmath68=\int_{-\infty}^\infty \rho_n(x)dx=1\ ] ] for the normalization of @xmath13 , as one would expect , and @xmath69&=&\int_{-\infty}^\infty \left\{\rho_n(x)\right\}^2dx=\int_{-\infty}^\infty e^{-2x^2 } ( \tilde{h}_n(x))^4dx\nonumber\\ & = & \sum_{j=0}^{2n}\frac{\gamma\left(j+\frac12\right)}{2^{j+\frac12}}\frac{4!}{(2j+4 ) ! } b_{2j+4,4}(c_0,2!c_1,\ldots,(2j+1)!c_{2j } ) \label{eq : w_2_n}\end{aligned}\ ] ] for the second - order entropic moment . as well , the entropic moments of arbitrary order @xmath40 for the hermite polynomials of lowest degree ( e.g. , @xmath70 and @xmath51 ) have the values @xmath71=\sqrt{\frac{\pi^{1-q}}{q}},\qquad w_q[\rho_1]= \frac{2^q}{\pi^\frac{q}{2}q^{q+\frac12 } } \gamma\left(q+\frac{1}{2}\right ) , \qquad w_q[\rho_2]=\frac{\pi^\frac{1-q}{2}2^q(2q)!}{q^{2q+\frac12}}\mathcal{l}_{2q}^{(-2q-\frac12)}\left(-\frac{q}{2}\right),\ ] ] where @xmath72 denotes a laguerre polynomial . then , we can now calculate all the information - theoretic measures of the hermite polynomials which are based on their entropic moments @xmath39 $ ] just calculated , such as the renyi and tsallis entropies given by eqs .
( [ eq : def_renyi ] ) and ( [ eq : def_tsallis ] ) , respectively , and the brukner - zeilinger and linear entropy which are closely related to the first - order entropic moment @xmath73 $ ] given by eq .
( [ eq : w_2_n ] ) . particularly relevant
are the values @xmath30=\left\{w_q[\rho_n]\right\}^{-\frac{1}{q-1}}=\left(\sum_{j=0}^{nq}\frac{\gamma\left(j+\frac12\right)}{q^{j+\frac12}}\frac{(2q)!}{(2j+2q ) ! } b_{2j+2q,2q}(c_0,2!c_1,\ldots,(2j+1)!c_{2j})\right)^{-\frac{1}{q-1}};\;q=2,3,\ldots \label{eq : lqrhon}\ ] ] for the renyi lengths ( [ eq : def_renyi_length ] ) of hermite polynomials , and @xmath31=\left\{w_2[\rho_n]\right\}^{-1}=\left(\sum_{j=0}^{2n}\frac{\gamma\left(j+\frac12\right)}{2^{j+\frac12}}\frac{4!}{(2j+4 ) ! } b_{2j+4,4}(c_0,2!c_1,\ldots,(2j+1)!c_{2j})\right)^{-1},\ ] ] for the heller length of hermite polynomials .
let us just explicitly write down the values @xmath74=\pi^\frac12 q^\frac{1}{2(q-1)}\ ] ] @xmath75=\left(\frac{\pi^\frac12 q}{2}\right)^\frac{q}{q-1 } q^\frac{1}{2(q-1 ) } \left(\gamma\left(q+\frac12\right)\right)^{-\frac{1}{q-1}}\ ] ] @xmath76=\pi^\frac12 2^\frac{q}{1-q } q^\frac{4q+1}{2q-2}\left((2q)!\mathcal{l}_{2q}^{(-2q-\frac12)}\left(-\frac{q}{2}\right)\right)^{-\frac{1}{q-1}}\ ] ] for the renyi lengths of the first few hermite polynomials with lowest degrees , and the values @xmath77=\sqrt{2\pi},\qquad
l[\rho_1]=\frac43\sqrt{2\pi},\qquad l[\rho_2]=\frac{64}{41}\sqrt{2\pi}\ ] ] for the onicescu - heller length of these polynomials . here
we determine the asymptotics of the shannon length @xmath78 $ ] of the hermite polynomials @xmath41 and its relation with its standard deviation .
moreover we use an information - theoretic - based optimization procedure to find sharp upper bounds to @xmath78 $ ] and we discuss their behavior in a numerical way .
the shannon length or exponential entropy of the hermite polynomials @xmath41 is defined by @xmath32=\exp\left\{s[\rho_n]\right\ } , \label{eq : nrhosrho}\ ] ] where @xmath79=-\int_{-\infty}^\infty e^{-x^2 } \tilde{h}_n^2(x)\ln\left[e^{-x^2}\tilde{h}_n^2(x)\right]dx \label{eq : shonnonrhon}\ ] ] is the shannon entropy of the hermite polynomial of degree @xmath45 .
simple algebraic operations yield to @xmath79=n+\frac12-\int_{-\infty}^\infty e^{-x^2 } \tilde{h}_n^2(x)\ln\tilde{h}_n^2(x)dx \label{eq : shannon_h}\ ] ] the logarithmic integral involved in this expression has not yet been determined in spite of serious attempts for various authors @xcite .
nevertheless these authors have found its value for large @xmath45 by use of the strong asymptotics of hermite polynomials ; it is given as @xmath80 then , from ( [ eq : shannon_h ] ) and ( [ eq : shannon_hlnh ] ) one has the asymptotical value @xmath79=\ln\sqrt{2n}+\ln\pi-1+o(1)=\ln\frac{\pi\sqrt{2n}}{e}+o(1)\ ] ] for the shannon entropy , and @xmath32\simeq \frac{\pi\sqrt{2n}}{e};\;n\gg 1 \label{eq : n_h}\ ] ] for the shannon length of the hermite polynomial @xmath41 . from eqs .
( [ eq : delta_x ] ) and ( [ eq : n_h ] ) one finds the following relation @xmath32\simeq \frac{\pi\sqrt{2}}{e } ( \delta x)_n;\;n\gg 1 \label{eq : nrho_delta}\ ] ] between the asymptotical values of the shannon length and the standard deviation of the hermite polynomial @xmath81 @xcite .
remark that this relation fulfils the general inequality ( [ eq : delta_inequalities ] ) which mutually relate the shannon length and the standard deviation . since the evaluation of the shannon @xmath82 given by eq .
( [ eq : shannon_h ] ) is not yet possible for a generic degree , it seems natural to try to find sharp bounds to this quantity .
here we do that by use of the non - negativity of the kullback - leibler entropy of the two arbitrary probability densities @xmath83 and @xmath21 defined by @xmath84=\int_{-\infty}^\infty \rho(x)\ln\frac{\rho(x)}{f(x)}dx.\ ] ] indeed , for @xmath85 , the rakhmanov density ( [ eq : rakhmanov_density ] ) associated to the hermite polynomials , the non - negativity of the corresponding kullback - leibler functional yields the following upper bound to @xmath82 : @xmath79\le -\int_{-\infty}^\infty \rho_n(x)\ln f(x)dx , \label{eq : s_le_rholnf}\ ] ] then we make the choice @xmath86 duly normalized to unity , as prior density , the optimization of the upper bound ( [ eq : s_le_rholnf ] ) for the choice ( [ eq : def_f ] ) with respect to the parameter @xmath87 yields the following optimal bound @xmath79\le \ln\left[a_k\langle x^k\rangle_n^\frac{1}{k}\right],\;k=2,4,\ldots\ ] ] with the constant @xmath88 finally , taking into account the expectation values @xmath89 of the rakhmanov density @xmath90 of the hermite polynomial @xmath41 given by eq . ( [ eq : xk ] ) , we have that @xmath91\le \ln\left[\frac{(ek)^\frac{1}{k}}{k}\gamma\left(\frac{1}{k}\right ) \left[\frac{k!}{\gamma\left(\frac{k}{2}+1\right)}\,_2f_1\left(\left . \begin{array}{c } -n,-\frac{k}{2}\\ 1 \end{array } \right| 2 \right)\right]^\frac{1}{k}\right],\end{aligned}\ ] ] and @xmath92\le \left[\frac{(ek)^\frac{1}{k}}{k}\gamma\left(\frac{1}{k}\right ) \left[\frac{k!}{\gamma\left(\frac{k}{2}+1\right)}\,_2f_1\left(\left . \begin{array}{c } -n,-\frac{k}{2}\\ 1 \end{array } \right| 2 \right)\right]^\frac{1}{k}\right]\equiv c_{k , n } , \label{eq : def_ckn}\end{aligned}\ ] ] as optimal bounds for the shannon entropy and shannon length of @xmath41 , respectively . in table
[ tab : ckn ] we give the value @xmath93 of @xmath94 which provides the best upper bound @xmath95 for the first few values of @xmath45 .
.values of @xmath94 which provides the best upper bound @xmath95 for the first few values of @xmath45 .
[ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] finally , let us point out that the fisher information of the hermite polynomials @xmath41 is given @xcite by @xmath96=\int_{-\infty}^{+\infty } \left\{\frac{d}{dx}\rho_n(x)\right\}^2\frac{dx}{\rho_n(x)}=4n+2 .
\label{eq : fisher_h}\ ] ] thus , the fisher length of these polynomials is @xmath97}}=\frac{1}{\sqrt{4n+2 } } , \label{eq : fisher_length_rhon}\ ] ] whose values lie down within the interval @xmath98 $ ] . the comparison of this expression with eq .
( [ eq : delta_x ] ) allows us to point out that not only the general relation ( [ eq : delta_inequalities ] ) between the fisher length and the standard deviation is fulfilled , but also that @xmath99 .
in this section we examine and discuss some computational issues relative to the spreading lengths of hermite polynomials @xmath100 and their mutual relationships . in contrast with the polynomials orthogonal on a segment of the real axis for which an efficient algorithm based on the three - term recurrence relation has been recently discovered for the computation of the shannon entropy by v. buyarov et al .
@xcite , up until now there is no explicit formula or at least a stable numerical algorithm for the computation of the renyi and shannon lengths of unbounded orthogonal polynomials , such as hermite polynomials , for any reasonable @xmath101 .
a naive numerical evaluation of these hermite functionals by means of quadratures is not often convenient except for the lowest - order polynomials since the increasing number of integrable singularities spoils any attempt to achieve a reasonable accuracy for arbitrary @xmath45 .
we propose an analytical , error - free and easily programmable computing approach for the entropic moments @xmath39 $ ] and the renyi spreading lengths @xmath102 $ ] of the orthonormal polynomials @xmath41 , which are given by eqs .
( [ eq : wqrhon_bell ] ) and ( [ eq : lqrhon ] ) , respectively , in terms of the combinatorial multivariable bell polynomials defined by eq .
( [ eq : bell_def ] ) .
this approach requires the knowledge of the expansion coefficients @xmath103 given by eq .
( [ eq : def_c_l ] ) and the determination of the partitions @xmath104 given by eq .
( [ eq : partitions ] ) . in figures [ fig:1 ] and [ fig:2 ]
we have shown the renyi lengths @xmath102 $ ] with @xmath105 and @xmath106 , and the shannon length @xmath78 $ ] of the orthonormal hermite polynomials in terms of the degree @xmath45 within the range @xmath107 $ ] , respectively .
we observe that both shannon and renyi lengths with fixed @xmath40 have an increasing parabolic dependence on the degree @xmath45 of the polynomials .
moreover , this behaviour is such that for fixed @xmath45 the renyi length decreases when @xmath40 is increasing . as well , in figure [ fig:2 ] we plot the optimal upper bound @xmath108 for the shannon length @xmath78 $ ] according to eq .
( [ eq : def_ckn ] ) .
$ ] with @xmath109 ( @xmath110 ) , @xmath111 ( @xmath112 ) , @xmath113 ( @xmath114 ) , and @xmath106 ( @xmath115 ) , in terms of the degree @xmath45.,width=377 ] $ ] ( @xmath114 ) and its optimal upper bound @xmath108 ( @xmath112 ) in terms of the degree @xmath45.,width=377 ] finally , for completeness , we have numerically studied the relation of the shannon length @xmath78 $ ] and the onicescu - heller length @xmath116 $ ] of these polynomials with the standard deviation @xmath26 within the range @xmath107 $ ] .
the corresponding results are shown in figures [ fig:3 ] and [ fig:4 ] , respectively .
we observe that when @xmath45 varies from @xmath117 to @xmath118 the two spreading lengths have a quasilinear behaviour in terms of the standard deviation .
moreover , we found the fit @xmath32=1.723 ( \delta x)_n+2.00 \label{eq : fitn}\ ] ] for the shannon length , with a correlation coefficient @xmath119 , and @xmath31=1.204(\delta x)_n+2.92,\ ] ] for the onicescu - heller length or second - order renyi length with a correlation coefficient @xmath120 .
in fact , the global dependence of @xmath78 $ ] on @xmath26 is slightly concave , being only asymptotically linear according to the rigorous expression ( [ eq : nrho_delta ] ) , i.e. @xmath78\simeq \frac{\pi\sqrt{2}}{e}(\delta x)_n\simeq 1.63445 ( \delta x)_n$ ] for @xmath121 , which is in accordance with ( [ eq : fitn ] ) .
it is worthwhile remarking ( i ) the @xmath122 behaviour with @xmath123 of the global spreading lengths ( standard deviation , renyi and shannon lengths ) and ( ii ) the @xmath124-law which is followed by the ( local ) fisher length . this difference may be associated to the gradient - functional form ( [ eq : def_deltax ] ) of the fisher length , indicating a locality property in the sense that it is very sensitive to the oscillatory character of the polynomials .
$ ] with @xmath109 ( @xmath110 ) , @xmath111 ( @xmath112 ) , @xmath113 ( @xmath114 ) , and @xmath106 ( @xmath115 ) , in terms of the standard deviation @xmath26 for @xmath107$].,width=377 ] $ ] in terms of the standard deviation @xmath26 for @xmath107$].,width=377 ]
the states of the one - dimensional harmonic oscillator , with a potential @xmath125 , are determined by the following probability density @xmath126 where @xmath100 are the orthogonal hermite polynomials @xcite . since @xmath127
, this density can be expressed in terms of the rakhmanov density of the orthonormal hermite polynomials @xmath41 defined by eq .
( [ eq : rho_n_hermite ] ) as follows : @xmath128 taking this into account , we obtain for the harmonic oscillator the following expressions @xmath129 @xmath130=\lambda^{\frac{q-1}{2}}w_q[\rho_n];\,q=0,1,2,\ldots\ ] ] for the ordinary and entropic moments , and @xmath131=-\ln\sqrt{\lambda}+s[\rho_n],\ ] ] @xmath132=\lambda f[\rho_n],\ ] ] for the shannon and fisher information measures in terms of the corresponding quantities of the hermite polynomials , whose values are given by eqs .
( [ eq : xk ] ) , ( [ eq : wqrhon_bell ] ) , ( [ eq : shannon_h ] ) and ( [ eq : fisher_h ] ) , respectively .
then , it is straightforward to find that all the uncertainty measures of the harmonic oscillator ( namely , standard deviation and renyi , shannon and fisher lengths ) are equal , save for a multiplication factor @xmath133 , to the corresponding spreading lengths of the hermite polynomials which control their wavefunctions .
this is to say that @xmath134=\lambda^{-\frac12 } l_q^r[\rho_n]\ ] ] @xmath135=\lambda^{-\frac12}n[\rho_n],\quad ( \delta x)_n^{\rm ho}=\lambda^{-\frac12}(\delta x)_n,\ ] ] whose values can be obtained by keeping in mind eqs .
( [ eq : delta_x ] ) , ( [ eq : lqrhon])-([eq : shannon_h ] ) , and ( [ eq : fisher_length_rhon ] ) respectively .
here we want to pose the following unsolved problems : to find the asymptotics of the frequency or entropic moments @xmath39 $ ] defined by @xmath54=\int_{-\infty}^\infty \left[\omega_h(x ) \tilde{h}_n^2(x)\right]^qdx = \int_{-\infty}^\infty e^{-qx^2}\left[\tilde{h}_n(x)\right]^{2q}dx,\,q\ge 1 \label{eq : wqrhon_2}\ ] ] in the two following cases : @xmath136 up until now the only known result is due to alexander aptekarev et al @xcite who have shown that these quantities , which are weighted @xmath137-norms of the orthonormal hermite polynomials @xmath41 as @xmath54=\parallel\tilde{h}_n(x)\parallel_{2q}^{2q},\ ] ] behave asymptotically ( @xmath138 ) as @xmath54=\left(\frac{2}{\pi}\right)^q\frac{\gamma\left(q+\frac12\right)\gamma\left(1-\frac{q}{2}\right)}{\gamma(q+1)\gamma\left(\frac32-\frac{q}{2}\right)}(2n+1)^\frac{1-q}{2}(1+o(1))\ ] ] when @xmath139 $ ] . to extend and generalize this result in the sense mentioned above , it might be useful for the interested reader to keep in mind the two following related results : * lars larsson - cohn @xcite studied the @xmath140-norms of the monic hermite polynomials @xmath141 , orthogonal with respect to the gaussian weight @xmath142 , defined by @xmath143 keeping in mind that @xmath144 , he found the following asymptotical @xmath145 result @xmath146 with the values @xmath147 + however , the asymptotics of the entropic moments @xmath148 $ ] of the hermite polynomials @xmath141 defined by @xmath149=\int_{-\infty}^\infty \left[\omega_g(x ) h_n^2(x)\right]^q dx = \left(\frac{1}{2\pi}\right)^\frac{q}{2}\int_{-\infty}^\infty e^{-\frac{q}{2}x^2}\left[h_n(x)\right]^{2q } dx \label{eq : wqhn_larsson}\ ] ] has not yet been found . *
ruth azor et al @xcite have used combinatorial techniques for the asymptotics ( @xmath138 ) of the following functionals of the hermite polynomials @xmath100 orthogonal with respect to @xmath150 : @xmath151:=\int_{-\infty}^\infty e^{-x^2}[h_n(x)]^k dx , k\in\mathbb{n}.\ ] ] they were only able to solve it in the particular non - trivial case @xmath152 , where @xmath153=\frac{3}{4n}\sqrt{\frac{3}{\pi}}\frac{6^{2n}}{(n!)^2}\left\{1-\frac{1}{4n}+\frac{3}{16n^2}+o(n^{-3})\right\}.\ ] ] besides , they have also considered the asymptotics ( @xmath154 ) of @xmath155 $ ] as well as that of the quantities @xmath156:=\sqrt{\frac{2}{\pi}}\int_{-\infty}^\infty e^{-2x^2 } \left[\tilde{h}_n(x)\right]^kdx,\ ] ] finding the following results @xmath151\sim \left[1+(-1)^{kn}\right]\left(2^{kn-1}\pi\right)^\frac12 \left(\frac{kn}{e}\right)^\frac{kn}{2 } e^{-\frac{n-1}{2}},\;\text{for } k\gg 1\ ] ] @xmath156\sim\frac{1+(-1)^{kn}}{\sqrt{2}}\left(\frac{kn}{e}\right)^{\frac{kn}{2}}e^{-(n-1)},\;\text{for } k\gg 1\ ] ] the solution of the problems ( [ eq : wqrhon_2])-([eq : problems ] ) and/or ( [ eq : wqhn_larsson ] ) , and the full determination of the asymptotics of @xmath155 $ ] and @xmath157 $ ] for ( @xmath138 , @xmath158 ) , either by means of an approximation - theoretic methodology @xcite or by use of combinatorial techniques @xcite , is of great relevance from both mathematical and applied points of view from obvious reasons , keeping in mind the physico - mathematical meaning of these quantities as previously mentioned .
it is worthwhile remarking that , in particular , the quantity @xmath39 $ ] is not only the @xmath159th - power of the @xmath137-norm of the hermite polynomials but it also represents the entropic moment of order @xmath40 of the harmonic oscillator and their isospectral physical systems , which fully determines the renyi and tsallis entropies of these objects .
in this paper we have introduced new direct measures of orthogonal - polynomials spreading other than the root - mean - square or standard deviation @xmath12 which have an information - theoretic origin ; namely , the renyi , shannon and fisher length .
they share with @xmath12 various interesting properties : same units as the variable , invariance under translations and reflections , linear scaling , vanishing in the limit that the variable tends towards a given definite value , and global character .
in contrast with @xmath12 , they do not depend on any particular point of the orthogonality interval what allow them to be considered as proper spreading lengths .
the renyi and shannon lengths are powerlike and logarithmic functionals of the polynomial @xmath6 while fisher length is a gradient functional of it , what allows one to state that the former lengths are global measures and the latter one has a locality property . for the renyi length of the orthonormal hermite polynomials @xmath41
we have developed a computational methodology which is based on the combinatorial multivariable bell polynomials whose arguments are controlled by the expansion coefficients of @xmath41 . for the shannon length of these polynomials , since it can not be calculated explicitly , we give ( i ) its asymptotics and the relation with the standard deviation , and ( ii ) sharp upper bounds by means of an information - theoretic - based optimization procedure .
moreover , it is computationally found that the second - order renyi length ( also called onicescu - heller length ) and the shannon length linearly depend on the standard deviation of the polynomials . on the other hand ,
the fisher length is analytically shown to have a reciprocal behaviour with @xmath12 mainly because of its local character .
finally , the previous results have been applied to the quantum - mechanical harmonic oscillator and some open problems related to the asymptotics of the frequency or entropic moments of hermite polynomials are posed .
their solution would allow not only to complete this work but also to extend considerably the related findings of various authors in connection to the @xmath140-norms of hermite polynomials @xcite .
consider the polynomial @xmath160 with degree @xmath45 , @xmath161 its @xmath55th - power is @xmath162^p=\left(\sum_{k=0}^n c_k x^k\right)^p = \sum_{\pi(p)}\frac{p!}{j_0 !
j_1!\cdots j_n!}(c_0x^0)^{j_0 } ( c_1x^1)^{j_1}\cdots(c_nx^n)^{j_n},\ ] ] where the sum is defined over all the partitions @xmath163 such that @xmath164 .
the previous expression of @xmath165^p$ ] can be written as @xmath162^p=\sum_{\pi(p)}\frac{p!}{j_0 ! j_1!\cdots j_n ! } c_0^{j_0 } c_1^{j_1}\cdots c_n^{j_n } x^{j_1 + 2j_2+\cdots+nj_n } = \sum_{k=0}^{np } a_{k , p}(c_0,\ldots , c_n ) x^k,\ ] ] where @xmath166 where the sum is defined over all the partitions @xmath167 such that @xmath168 the bell polynomials are defined as @xmath169 where the sum is defined over all the partitions @xmath60 such that @xmath170 the relation between the coefficients @xmath171 and the bell polynomials can be established as @xmath172 where @xmath57 for @xmath58 . from equation ( 3l ) from @xcite
we obtain that @xmath173 then , @xmath174 where , again , @xmath57 for @xmath58 .
we are very grateful for partial support to junta de andaluca ( under grants fqm-1735 and fqm-2445 ) and ministerio de educacin y ciencia under project fis2008 - 2380 .
the authors belong to the andalusian research group fqm-0207 .
r. a. fisher , theory of statistical estimation , proc .
cambridge phil .
( 1925 ) 700725 , reprinted in collected papers of r.a .
fisher , edited by j.h .
bennet ( university of adelaide press , south australia ) , 1972 , 1540 .
j. s. dehesa , w. van assche , r. j. yez , information entropy of classical orthogonal polynomials and their application to the harmonic oscillator and coulomb potentials , methods appl .
4 ( 1997 ) 91110 .
e. romera , j. c. angulo , j. s. dehesa , reconstruction of a density from its entropic moments , in : r. l. fry ( ed . ) , proc .
first int .
workshop on bayesian inference and maximum entropy methods in science and engineering , american institute of physics , new york , 2002 , p. 449 .
j. s. dehesa , r. gonzlez - frez , p. snchez - moreno , the fisher - information - based uncertainty relation , cramer - rao inequality and kinetic energy for the d - dimensional central problem , j. phys .
40 ( 2007 ) 18451856 . | the renyi , shannon and fisher spreading lengths of the classical or hypergeometric orthogonal polynomials , which are quantifiers of their distribution all over the orthogonality interval , are defined and investigated .
these information - theoretic measures of the associated rakhmanov probability density , which are direct measures of the polynomial spreading in the sense of having the same units as the variable , share interesting properties : invariance under translations and reflections , linear scaling and vanishing in the limit that the variable tends towards a given definite value .
the expressions of the renyi and fisher lengths for the hermite polynomials are computed in terms of the polynomial degree .
the combinatorial multivariable bell polynomials , which are shown to characterize the finite power of an arbitrary polynomial , play a relevant role for the computation of these information - theoretic lengths .
indeed these polynomials allow us to design an error - free computing approach for the entropic moments ( weighted @xmath0-norms ) of hermite polynomials and subsequently for the renyi and tsallis entropies , as well as for the renyi spreading lengths .
sharp bounds for the shannon length of these polynomials are also given by means of an information - theoretic - based optimization procedure .
moreover , it is computationally proved the existence of a linear correlation between the shannon length ( as well as the second - order renyi length ) and the standard deviation . finally , the application to the most popular quantum - mechanical prototype system , the harmonic oscillator , is discussed and some relevant asymptotical open issues related to the entropic moments mentioned previously are posed .
orthogonal polynomials , hermite polynomials , spreading lengths , computation of information measures , shannon entropy , renyi entropy , fisher information , bell polynomials .
89.70.cf 33c45 , 94a17 , 62b10 , 65c60 |
magneto - hydrodynamic ( mhd ) turbulence in the presence of a mean - field @xmath3 has a tendency to become two - dimensional ( 2d ) .
this tendency was early recognized by inspection of the fourier energy spectra in direct numerical simulations ( dns ) .
the energy distribution is indeed anisotropic , residing in wavevector mostly perpendicular to the mean magnetic field @xcite .
ideally one would like to quantify the two - dimensionalization as a scaling relation between parallel and perpendicular wavenumbers having the same energy density , @xmath4 .
if @xmath5 the anisotropy is scale independent , and the aspect ratio @xmath6 of the isocontour of the fourier spectrum does not change with scales .
if instead @xmath7 the aspect ratio increases with wavenumber , that is , the spectrum becomes more and more 2d at smaller and smaller scales . in dns ,
the parallel spectral extent is generally very short , due the limited achievable reynolds numbers , and the parallel spectrum rarely shows a power - law , rendering the distinction between scale - dependent and scale - independent anisotropy difficult . in contrast , the two - point correlation in real space expressed by ii - order structure functions , @xmath8 , shows in general nicer power - law scaling in the parallel direction , allowing one to quantify the scale - by - scale anisotropy .
in analogy with fourier spectra , the scaling relation involves parallel and perpendicular increments that have the same energy @xmath9 , also known as eddy anisotropy .
the ii - order structure function anisotropy has been widely studied in dns of incompressible mhd turbulence . using a local mean - field to identify parallel and perpendicular increments ,
one finds a scale - dependent anisotropy @xcite : the anisotropy grows at smaller and smaller scales , suggesting a complete 2d state at small enough scales .
furthermore , the anisotropy is controlled by @xmath10 where @xmath11 indicates the root - mean - square amplitude of turbulent fluctuations , the stronger @xmath3 the stronger the anisotropy @xcite .
the anisotropy is also stronger in regions of stronger magnetic field @xcite .
similarly , the anisotropy in the fourier spectrum increases for the stronger @xmath3 @xcite .
however , employing structure function to measure the anisotropy with respect to the global mean - field returns a scale - independent anisotropy @xcite , implying that the two - dimensionalization does not increase at smaller scales but reaches an asymptotic value .
a similar dichotomy exists in solar wind measurements , in which one can compute the two - point correlation in time from time series of data collected in - situ by spacecraft and then adopt the taylor hypothesis to obtain spatial increments .
the increments are thus taken along the radial direction , but the anisotropy with respect to the magnetic field is recovered thanks to its variable direction with respect to the radial .
as in dns , the structure function @xmath8 is found to be more energetic along perpendicular increments than along parallel increments .
again , a local mean - field analysis yields a scale - dependent anisotropy @xcite , while a global mean - field analysis indicates a scale - independent anisotropy @xcite .
several authors @xcite showed that for strong turbulence the scale - dependent anisotropy is smoothed out in a global mean - field analysis , even in the presence of a strong mean - field . on the other hand
, @xcite noted that since the local mean - field is a random variable , the local ii - order structure functions involve higher - order statistics and can not be the real space equivalent of the power spectra .
however for small enough @xmath10 the global and local measures are expected to coincide . in this work
we investigate the process of two - dimensionalization of mhd turbulence , focusing on the anisotropy measured in the global frame . in this frame
, one can obtain a dynamical equation ( labelled khypp equation ) that relates ii - order and iii - order structure functions @xcite , extending to incompressible mhd the von karman - howart - monin equation for incompressible hydrodynamic turbulence . according to the khypp equation , for stationary and homogenous turbulence in the inertial range ,
the divergence of the iii - order structure function @xmath0 is proportional to the cascade rate of turbulence @xmath12 .
the divergence is negative , implying that the cascade is achieved by removing positive correlations , and thus increasing the amplitude of @xmath8 ( flattening its power - law index ) .
thus , by characterizing the anisotropy of @xmath0 one can get insight into the process of two - dimensionalization of mhd turbulence .
previous studies on the iii - order structure functions in dns of the mhd equations were limited to 2d @xcite thus leaving out the issue of anisotropy . in a recent work , @xcite reported for the first time the vector @xmath0 measured in an experiment of rotating hydrodynamic turbulence .
as rotation was increased , the anisotropy of the ii - order structure functions also increased .
they found that the two - dimensionalization can be associated with the tilting of the vector @xmath0 toward the plane orthogonal to the rotation axis and that the tilting begins at small scales and then propagates to larger and larger scales . in the present work ,
we carry out a similar analysis on data from three - dimensional ( 3d ) dns of incompressible mhd turbulence by computing for the first time the 3d iii - order structure functions , in the presence or absence of a mean - field .
we find that the degree of two - dimensionalization as measured by @xmath8 is associated to the relative excitation of pseudo and shear alfvn polarizations for stationary turbulence with mean field @xmath3 .
we also analyze the full khypp equation in developed decaying isotropic turbulence , offering a description of the cascade in real space based on structure functions ( sf ) , analogous to the usual kolmogorov cascade in fourier space .
note that while the latter is based on the assumption of locality , the khypp equation is free from this assumption ( requiring only homogeneity ) , thus representing a more general description of the cascade process in mhd turbulence .
the iii - order structure function @xmath0 has also been computed in solar wind data to obtain the cascade rate of solar wind turbulence @xcite .
these rates are consistent with the heating rate estimated from proton temperature gradients @xcite , suggesting that turbulence may supply the heating required to sustain the non - adiabatic expansion of the solar wind . however
, applying the khypp equation to the solar wind is a bit problematic since solar wind turbulence is neither stationary nor homogenous ( e.g. @xcite ) . to obtain the cascade rate one should compute the divergence of @xmath0 , which is quite difficult with single - spacecraft data , since
increments are taken only along one direction at time ( see however @xcite for an integral form that exploits the four cluster spacecraft with the minimal assumption of axisymmetry ) .
the cascade rate can thus be retrieved only assuming a form for the unknown anisotropy of @xmath0 .
although some theoretical predictions exist ( e.g. @xcite ) , two methods are commonly employed in observations that assume respectively isotropy or an anisotropic model based on the geometrical slab - plus-2d turbulence that was introduced by @xcite to describe the two - point correlation function of solar wind turbulence .
we will exploit the data from our dns to test the two methods employed in solar wind data against known anisotropic iii - order structure functions and estimate the possible systematic errors .
the plan of the paper is as follows . in section [ sec2 ]
we give a brief introduction to the khypp equation , while in section [ sec3 ] we describe the method employed to compute structure functions ( sf ) of ii - order and iii - order .
the results are presented in section [ sec4 ] , where we first describe the anisotropy of ii - order structure functions of the simulations considered .
the rest of the section is dedicated to the iii - order sf .
we consider first a simulation of decaying turbulence without mean magnetic field , allowing us to test the soundness of our analysis method and to verify that the time - dependent khypp equation holds in developed decaying turbulence .
then , we analyze two simulations of turbulence with mean - field that have a different strength of anisotropy . finally , in section [ sec5 ] , we test on runs with @xmath13 the two methods employed to measure the cascade rate in the solar wind .
we conclude with a discussion on the results and on the application to the solar wind turbulence .
the von karman - howart - yaglom , politano - pouquet equation ( khypp ) for non - stationary , anisotropic , and incompressible mhd @xcite can be obtained from the original mhd equations written in term of the elssser variables @xmath14 , by subtracting the mhd equations evaluated at different positions @xmath15 and @xmath16 and by averaging in the volume . under the assumptions of incompressible , homogeneous turbulence one obtains @xmath17 where we have defined the two - point correlation , @xmath18 , and @xmath19 stands for the volume average .
these equations describe the evolution of the ii - order structure function for each elssser variable , @xmath20 .
the divergence term in the left hand side is the iii - order structure function , @xmath21 , involving products of @xmath22 , and @xmath23 , which we name yaglom flux in the following . on the right hand side ( rhs ) , @xmath24 and @xmath25 represent pressure terms and sweeping terms ( responsible for the alfvn effect ) respectively , both vanishing for globally homogeneous turbulence @xcite .
the remaining terms represent dissipation .
the first one involves the laplacian with respect to the increments @xmath26 , it vanishes for vanishing viscosity ( for simplicity we assumed equal viscosity and resistivity @xmath27 ) .
the second one , @xmath28 , is the dissipation rate of the elssser energies ( @xmath29 ) . in the former , the derivatives of the primitive fields @xmath30 do not commutate with the averaging operation , and
the dissipation rate remains finite for vanishing viscosity .
summing the contributions of both elssser fields , one finally gets an expression for the total energy and cascade : @xmath31 where @xmath32 , @xmath33 , and @xmath34 .
note that because of homogeneity in the above equations - all the variables depend only on the vector separation , @xmath35 .
this equation is valid for decaying turbulence and describes the classical scenario of a turbulent flow in which the dissipation of energy is achieved through a cascade of energy toward smaller scales , where fluctuations are finally damped by viscosity . in forced turbulence
one should add on the right hand side the forcing terms ( @xmath36 ) that inject energy ( usually ) at large scales .
for stationary turbulence ( @xmath37 ) forced at large scales ( @xmath38 only at large scales ) the injection , cascade , and dissipation all occur at the same rate . at high reynolds number one
expects their respective ranges to be well separated , in analogy to the kolmogorov cascade in fourier space , so one has that : i ) at large scales the second and last terms in equation are negligible and @xmath39 , the forcing balances the dissipation rate , ii ) at small scales the second term is negligible , @xmath40 , and the damping rate is equal to the dissipation rate , and finally iii ) at intermediate scales the last term is negligible , yielding @xmath41 that is , the cascade rate , which is equal to the dissipation rate , is given by the divergence of the yaglom flux .
note that equation can be used as a definition of the inertial range as being the ensemble of scales for which the equation is approximately satisfied .
the definition should hold for quasi - stationary forced turbulence and for developed decaying turbulence .
as we will see , for developed decaying turbulence , the time - dependent term is non - negligible only at large scales where @xmath42 ( the dissipation is balanced by the decay of the ii - order sf at large scales ) . we can now give a more physical interpretation of the cascade process by rewriting equation in terms of the autocorrelation function , @xmath43 , with @xmath44
the autocorrelation functions are related to structure functions by @xmath45 and using @xmath46 , the khypp equation becomes : @xmath47 showing that @xmath0 is a flux of negative correlations .
a permanent flux of negative correlations toward small scales is equivalent to constantly building new small - scales gradients .
for instance , the formation of 2d quasi - perpendicular turbulence will be revealed by a yaglom flux bringing negative correlation at small perpendicular scales , hence the vector @xmath0 must be quasi - uniform , parallel to the @xmath48 axis , and pointing toward the parallel @xmath49 axis . coming back to equation , for turbulence in stationary state or decaying turbulence ,
the cascade is a constant at all inertial - range scales ( @xmath50 ) .
thus , the inertial range anisotropy can not appear as different cascade rates in the parallel and perpendicular directions .
the anisotropy instead will show up in the shape of the domain of @xmath35 for which @xmath51 .
to illustrate such an anisotropy , one can assume some particular symmetry of the flux @xmath0 to characterize the cascade , with the additional advantage of obtaining a direct relation to the cascade rate , so as to avoid computing the divergence of @xmath0 .
the simplest assumption is that of _ isotropic turbulence _ for which @xmath0 depends only on the scalar increment @xmath26 .
rewriting the divergence in spherical coordinates , and assuming stationary conditions and vanishing viscosity , equation in the inertial range becomes the isotropic khypp equation , yielding @xmath52 in which the yaglom flux @xmath53 is projected along the increment .
this form is often used in solar wind studies , since although one does not have access to the full divergence in in - situ data , the cascade rate can be obtained directly from the projected yaglom flux .
the inertial range occupies a volume that is a sector of a sphere , it can be defined as isotropic since it has the same extent and location on parallel and perpendicular increments .
a strongly anisotropic case case is that of _ 2d turbulence _
, obtained when the yaglom flux in the inertial range depends only on the in - plane increments , @xmath54 : @xmath55 where @xmath56 are the in - plane components of @xmath0 and @xmath57 denotes derivatives with respect to the in - plane increments .
the yaglom flux can have out - of - plane components , but the cascade rate is determined only by the in - plane components .
note that a yaglom flux having only in - plane components in the inertial range is a sufficient condition to have a 2d cascade .
assuming isotropy of the in - plane increments , that is , a dependence only on the scalar separation @xmath48 , one obtains again a direct relation between iii - order sf and the cascade rate : @xmath58 with @xmath59 indicating the projection of @xmath60 on the radial direction in cylindrical coordinates .
note that the inertial - range domain is not confined to the 2d plane even in this anisotropic turbulence .
as we will see , the anisotropy of the inertial - range domain shows up in its different extent and location along parallel and perpendicular increments . for completeness ,
we consider finally the case of _ 1d turbulence _
, when the iii - order sf depends only on one coordinate , say , @xmath49 .
one obtains the cascade rate as @xmath61 note that the geometrical model of _ slab turbulence _ does not correspond to the 1d turbulence , since in the former , fluctuations are assumed to be perpendicular but to depend only on parallel wavevectors ( and hence parallel separations ) .
the slab geometrical configuration would indeed have a vanishing divergence .
.runs and parameters for simulations . [ cols="<,^,^,^,^,^,^,^,<",options="header " , ] [ table1 ]
we consider three high - resolution simulations of mhd turbulence whose parameters are listed in table [ table1 ] .
run a is a developed decaying simulation of incompressible mhd turbulence without mean - field , representing
_ isotropic turbulence_. run b is a simulation of weakly compressible mhd turbulence , where @xmath62 is the sound speed , while @xmath63 . ] , with mean - field @xmath2 and anisotropic forcing .
the forcing is applied only to components perpendicular to the mean - field and to wavevectors mainly perpendicular to the mean - field .
thus we are dealing with _ strong anisotropic turbulence _ of fluctuations with _ shear - alfvn _ polarizations , a configuration akin to reduced mhd . for fluctuations with mainly perpendicular wavevectors ,
pseudo polarizations have components parallel to the mean field , while shear polarization have components perpendicular to the mean field . in this limit ,
the former are absent in the reduced mhd formalism . ] .
finally run c is a simulation of incompressible mhd , again with mean - field @xmath2 , but with a forcing which is isotropic in both components and wavevectors . in the latter
the forcing is actually a freezing of the modes @xmath64 that maintains an equal amount of _ pseudo - alfvn and shear - alfvn polarizations _ at large scales , along with equipartition between magnetic and kinetic energy and between elssser energies .
this simulation can be classified as a case of _ weak anisotropic turbulence _ in terms of the strength parameter @xmath65 ( see table [ table1 ] ) , although it does not have the properties of classical weak turbulence @xcite .
indeed , the 3d spectrum has a relatively strong excitation in the parallel direction , resulting in a peculiar anisotropy @xmath66 , with an isotropic spectral index @xmath67 in all directions ( corresponding to a 1d spectrum with slope @xmath68 ) and all the anisotropy appearing as a power anisotropy at large scales @xmath69
. we will not discuss the properties and the origin of such a spectrum that can be found in @xcite ) ; what is mostly relevant for the present analysis is that run c has a different 3d anisotropy compared to run b , although in both runs energy resides mainly in perpendicular wavevectors .
we will use three measures to characterize the simulations : ii - order structure functions computed in the frame defined by the local mean - field ( local @xmath8 ) , ii and iii order structure functions , respectively @xmath8 and @xmath0 , computed in an absolute frame attached to the @xmath70-axis , which is the direction of the global mean - field when it is present . for all structure functions , computation is made calculating increments in real space .
local @xmath8 are obtained following method i of @xcite , i.e. the local field at scale @xmath35 is defined as @xmath71 $ ] .
note that the measure of anisotropy , i.e. the ratio @xmath72 of local @xmath8 , is not unambiguously defined . in a turbulent medium fluctuations
have a wide range of excited scales and the definition of the mean field depends on both scale and position .
thus , higher - order statistics may be introduced in the local @xmath8 to a different extent , depending on the averaging procedure .
however , the two - point average employed in this work is a good working definition , at least in simulations of homogeneous turbulence since it was shown to yield the same results as line average or volume averages , provided that the averaging scale is smaller than the correlation length @xcite .
the computation of local @xmath8 is made for the whole range of available increments @xmath35 but the average is made on a subset of grid points ( typically @xmath73 ) , which was checked to be a sufficient statistics for the anisotropy to converge .
on the other hand the iii - order structure functions @xmath0 are signed quantities and their computation requires large statistics to converge . thus given the number of grid points @xmath74 in a given direction , we compute @xmath0 either on a smaller range of increments ( @xmath75 with typically @xmath76 , @xmath77 is the grid size ) , or on a coarser grid of increments ( @xmath78 , with typically @xmath76 ) , but still on all the grid points , so the averages are made on a sample of @xmath79 data .
the dissipation rate ( @xmath80 ) entering the khypp equation is obtained directly from the 3d fourier spectra of magnetic and velocity field ( @xmath81 and @xmath82 respectively ) , @xmath83 the terms appearing in equation are evaluated for four consecutive snapshots separated by approximately a nonlinear time ( for the time derivative @xmath84 we use a simple first - order scheme ) .
all quantities entering the khypp equation are normalized to the dissipation rate @xmath80 and then averaged over the four snapshots .
the 3d data are finally reduced for purposes of representation and analysis by performing an isotropization ( averaging over polar and azimuthal angle @xmath85 and @xmath86 in spherical coordinates ) , @xmath87 or an axisymmetrization ( averaging over the azimuthal angle @xmath86 in cylindrical coordinate with axis along the @xmath88 ) , @xmath89 in the following we drop the subscripts @xmath90 and @xmath91 and eventually mention explicitly the average procedure used for representation .
we measured the anisotropy of ii - order sf in two frames . in the global frame the increments @xmath49 and @xmath92 are taken parallel and perpendicular to a fixed direction @xmath70 , which is the direction of the mean - field @xmath93 when it is present . in the local frame
the parallel and perpendicular directions are relative to the scale - dependent mean - field direction @xmath94 ( see section [ sec2 ] ) .
the measure of the anisotropy is obtained by identifying the couples of increments ( @xmath95 ) at which the parallel and the perpendicular sf have the same value : @xmath96 , i.e. the function @xmath97 measures the aspect ratio of isolevels of the sf at different scales , also known as eddy shape .
scale - independent anisotropy results in a linear relation @xmath98 , that is , an aspect ratio @xmath99 that does not change with scale .
conversely , scale - dependent anisotropy results in a deviation from the linear scaling @xmath100 , with @xmath7 : the aspect ratio of sf , @xmath99 increases at smaller scales , that is , eddies become more and more elongated in the parallel direction . in figure [ fig1 ]
we plot @xmath97 for the two measures of anisotropy , local ( gray squares ) and global ( black diamonds ) , for the three runs listed in table [ table1 ] .
the dashed line , with slope @xmath101 , is a reference for scale - independent anisotropy .
the dotted line is a reference for the scale - dependent anisotropy predicted by the critical balance relation @xcite : @xmath102 the insets display the same plots compensated by the critical balance anisotropy in order to better appreciate the scaling relation @xmath97 . in the local frame , all runs have a scale - dependent anisotropy extending to a wide range of scales .
the function @xmath97 has a slope flatter than @xmath103 , thus the anisotropy grows with decreasing scales and eddies are more and more elongated in the parallel direction .
the scaling law actually follows the critical balance anisotropy equation in the range where @xmath104 , extending for about a decade for runs a , b , and c in the intervals @xmath105 , @xmath106 , and @xmath107 respectively .
this critical balance anisotropy is quite robust since in all runs the anisotropic relation holds rather well even outside the above - mentioned range of scales where @xmath108 and @xmath109 have a clear power - law scaling .
note that run c is a weak turbulence simulation , and it is not obvious that it should have a critical balance anisotropy ( see @xcite for an explanation based on a heuristic model of anisotropic turbulence ) .
consider now the anisotropy measured in the global frame ( black diamonds in figure [ fig1 ] ) , which will be more relevant for the following analysis , since it is related to the iii - order sf by the khypp equation . as expected , run a is perfectly isotropic , the aspect ratio is unity at all scales .
the anisotropy of strong turbulence , run b , becomes scale - independent ( it has a slope equal to one ) for scales @xmath110 , approaching a constant ratio @xmath111 at small scales . for weak turbulence ,
run c , the anisotropy becomes scale - independent only at very small scales ( @xmath112 ) , with an aspect ratio @xmath113 that is smaller compared to strong turbulence .
the vertical bars in the plot bound the inertial range as identified from equation .
is constant .
we measure the slope of @xmath114 in logarithmic scales along the parallel and perpendicular directions and define inertial range scales as those ones having a slope @xmath115 ( see figure [ fig3]c and figure [ fig4]c respectively ) .
this is the procedure used for run b and c. we anticipate that for run a the divergence of the iii - order sf is not a nice straight horizontal line ( figure [ fig2]c ) , so in this case the inertial range is defined by the scales at which @xmath116 is about twice larger than all other terms in the khypp equation ( corresponding to a slope @xmath117 ) . as a partial cross - check , the chosen thresholds return a dissipative scale that matches the scale at which @xmath118 has a local maximum , which is the standard procedure for the identification of the dissipative scale via ii - oder sf . ] while run b has a scale - independent anisotropy that develops in the inertial range , in run c scale - independent anisotropy is attained only at dissipative scales .
thus , at inertial - range scales the anisotropy is mostly scale dependent in our weak turbulence simulation ( run c ) : it follows the critical balance scaling @xmath119 even when measured in the global frame , contrary to expectations @xcite .
we consider first the isotropic case ( run a ) for which @xmath8 is isotropic , so we expect also to find an isotropic iii - order sf . in figure
[ fig3 ] , panel ( a ) , we plot the yaglom flux @xmath0 ( iii - order sf ) , averaged along the polar angle of cylindrical coordinates with axis @xmath120 , and normalized by the scalar increment @xmath26 .
we consider relatively small scales ( the largest scale is @xmath121 ) to highlight inertial range features , as will be clearer below . the yaglom flux is almost radial at large scales and becomes remarkably radial at smaller scales @xmath122 . the length of the arrows increases toward the origin , indicating that the intensity of the cascade increases when approaching the inertial range ( the gray area ) while keeping the same ( radial ) direction .
note also that the arrow length is constant on circles of given @xmath26 , meaning that @xmath123 as expected for isotropic turbulence , equation . in panel
( b ) we plot the isocontours of the divergence of the yaglom flux , normalized by the dissipation rate .
the isocontours are roughly isotropic at large scales and become perfectly isotropic at small scales ( @xmath124 ) . in a small interval of scales around @xmath125 ,
the divergence @xmath126 has a maximum reaching the value @xmath127 .
thus , the dissipation rate is approximately equal to the cascade rate , and these scales can be identified as the inertial range of turbulence . however , the divergence is not strictly a constant , as expected for the inertial range .
note that although the latter is very short , it is uniformly distributed among scales .
finally , in panel ( c ) we plot in logarithmic scales , after isotropization and normalization by @xmath128 , all the terms appearing in the khypp equation , equation , namely , the divergence of the yaglom flux @xmath126 ( thick solid line ) , the dissipative term @xmath129 ( red triple - dot - dashed line ) , and the time - dependent term @xmath84 ( dotted line ) . the dissipative term dominates at small scales , while the time - dependent term ( decay ) dominates at large scales . the cascade term , @xmath126 , is larger than the other terms for scales @xmath130 ( the gray area in panel ( a ) ) .
these two extrema can be identified with the injection scale and the dissipative scale , respectively .
it is worth noting that the dissipation scale defined in this way coincides with the estimate based on ii - order sf in figure [ fig1 ] .
from this logarithmic plot it is clear that the inertial range is quite small in this decaying simulation because @xmath126 is much larger than other terms only in a small interval centered at @xmath125 , where it is roughly horizontal and equal to 0.9 ( it should be equal to one in an ideally infinite inertial range ) . in the same plot we also traced , as a blue long - dashed line ,
the sum of the three terms just discussed , that should amount at all scales to the dissipation rate @xmath128 ( thick solid horizontal line ) for good energy conservation .
the sum is an almost horizontal straight line , only a factor @xmath131 higher than the dissipation rate for scales @xmath132 .
this confirms that the statistics is large enough to ensure convergence and that the conservation of energy holds with sufficient accuracy except at very small scales where a small numerical dissipation probably kicks in : the injection at large scale ( including the decay ) , the cascade in the inertial range , and the dissipation at small scales all occur at the same rate .
to summarize , the analysis of the isotropic turbulence confirms the theoretical expectations : inside the inertial range the yaglom flux is radially directed and its magnitude scales linearly with the scalar increment @xmath26 .
the divergence of the yaglom flux is approximately constant in the inertial range , and it is uniformly distributed among scales ( isotropic ) .
note , however , that the extent of the inertial range is very limited due to the relatively small reynolds number that prevents the formation of a large range of scales where the divergence of the yaglom flux is the dominant term . with the current resolution ( @xmath133 )
it is at best half a decade , indicating that this is the minimal resolution required for studies of decaying turbulence ( although hyperviscosity would probably alleviate the problem ) .
this is relevant for solar wind studies @xcite , in which expansion induces an additional decay of magnetic and kinetic energy on top of the decay due to the turbulent dissipation .
let s now turn to the anisotropic run b , which is a simulation of strong turbulence with guide field and forced at large scales on components perpendicular @xmath3 . in figure [ fig3 ]
, panel ( a ) we plot again the iii - order sf , i.e. the yaglom flux @xmath0 , averaged over the polar angle in cylindrical coordinates with axis along increments parallel to the mean magnetic field . at variance with the isotropic case ,
the arrows are colored according to the angle @xmath134 formed with respect to the perpendicular direction ( black color stands for @xmath135 ) , and their length is normalized to the perpendicular increment @xmath136 .
the yaglom flux is remarkably vertical in the inertial range ( the gray area ) , and it is proportional to the perpendicular increments @xmath137 ( the arrow length is uniform in the whole inertial range , after normalization ) .
this suggest that turbulence is undergoing a purely 2d cascade , equation . in panel ( b )
the normalized divergence of the yaglom flux is constant over a large interval of parallel and perpendicular increments , with a value close to 1 ( the light - green area at level @xmath138 ) .
however it is not uniformly distributed among scales , the isocontour of constant divergence extends to smaller perpendicular scales , suggesting that the cascade is not removing positive correlation from the parallel direction .
this is consistent with figure [ fig1]b that shows a clear anisotropy of @xmath8 in favor of a two - dimensionalization in the perpendicular plane .
this can be better appreciated in panel ( c ) , where we plot cuts along the parallel and perpendicular directions of the divergence of the yaglom flux in logarithmic scales .
there is a very nice constant divergence in the perpendicular direction , with a value close to the dissipation rate ( the horizontal thick solid line at level 1 ) , covering about one decade in the range @xmath139 . in the parallel direction ,
an approximate inertial range is also found , having a smaller extent and shifted to larger scales @xmath140 .
we recall that from @xmath126 ( panels ( b ) and ( c ) ) one can identify the location of the inertial range and its distribution among parallel and perpendicular scales . on the other hand ,
the simple dependence of the yaglom flux allows us to identify the cascade as a 2d cascade with @xmath141 we anticipate that @xmath142 have only perpendicular components because there is a dominance of shear - alfvn polarizations .
indeed , these polarizations have @xmath143 lying in the perpendicular plane and , as a consequence , the cascade is 2d .
this simple picture of 2d cascade does not hold anymore as soon as pseudo polarizations , which have out - of - plane components , are energetically important .
we finally consider the case of weak turbulence with guide field ( run c ) , which is forced isotropically at small wavevectors @xmath144 by imposing at all times ( freezing ) the corresponding fourier modes of the fluctuating fields @xmath145 ( note that their @xmath146 components have also equal energy ) . in fig.[fig4 ] , panel ( a ) one can see immediately that the yaglom flux is oblique , with an angle @xmath134 that changes with scales ( we normalize arrow length by the perpendicular increment @xmath147 as in figure [ fig3 ] ) .
this means that the iii - order sf has a parallel component and a non - negligible dependence on the parallel increments , thus contributing to the cascade rate through the divergence of the yaglom flux .
such contribution seems to decrease at small parallel scales , where the yaglom flux becomes vertical , hinting to a milder two - dimensionalization of this weak turbulence cascade .
isocontours of @xmath126 are plotted in panel ( b ) , with the usual normalization by @xmath128 .
the inertial range can be identified with the light - orange area at level @xmath148 , extending to a wide range of parallel and perpendicular scales .
its distribution is non - uniform and more complex than that of strong turbulence ( figure [ fig3]b ) , reflecting the dependence of the iii - order sf from @xmath48 and @xmath49 .
panel ( c ) shows cuts of the divergence of the yaglom flux along the parallel and perpendicular directions .
although the divergence is not exactly constant , the inertial range covers more than one decade in the perpendicular direction , @xmath149 , yielding a cascade rate that is slightly higher than the dissipation rate @xmath128 . on the other hand ,
the cut in the parallel direction is less flat , making more questionable the identification of an inertial range in this direction .
the approximate parallel inertial range is shorter and located at larger scales , @xmath150 .
decomposing fluctuations in pseudo and shear alfvn polarizations proves useful to analyze in some more detail the relation between the yaglom flux and the cascade rate in run c , which is a simulation of incompressible mhd .
the decomposition is made in fourier space , where pseudo alfvn polarizations and shear alfvn polarizations are oriented along the unitary vectors ( e.g. @xcite ) , @xmath151^{1/2}},\;\;\;\ ; { { \boldsymbol{\xi_{ps}}}}= \frac{{{\boldsymbol{b_0}}}-({{\boldsymbol{k}}}\cdot{{\boldsymbol{b_0}}}){{\boldsymbol{k } } } } { [ 1-({{\boldsymbol{k}}}\cdot{{\boldsymbol{b_0}}})^2]^{1/2}}. \label{eq : vers}\end{aligned}\ ] ] this decomposition is completely equivalent to the decomposition into toroidal and poloidal components of the magnetic fluctuations . in incompressible
mhd the same decomposition applies to the velocity field , which is also solenoidal .
shear alfvn polarizations are the proper alfvn modes in full mhd .
their component is perpendicular to both the mean field @xmath93 and the wavevector @xmath152 , being incompressible .
the pseudo alfvn polarizations are the incompressible limit of slow modes in mhd .
their component lies in the plane identified by the mean field and the wavevector , and it is again perpendicular to the wavevector . for fluctuations with strong anisotropic spectra ( @xmath153 ) ,
shear polarizations have wavevectors and components lying in the plane perpendicular to @xmath3 , thus they represent the 2d modes in the slab - plus-2d decomposition introduced by @xcite . instead , pseudo polarizations have wavevectors in the plane perpendicular to @xmath3 but components along @xmath3 .
this polarization , which is absent in reduced mhd , is instead present in 2.5d configurations with out - of - plane mean field , and should not be confused with the slab component that has wavevectors parallel to @xmath3 and fluctuations perpendicular to it ( and is thus included in reduced mhd ) .
after decomposing fluctuations in pseudo and shear alfvn polarizations , we go back to the real space and compute separately all the contributions to the khypp equation , equation .
note that for strictly parallel wavevectors the pseudo - shear decomposition degenerates , so we remove such modes ( the slab component ) in the following analysis to avoid arbitrary partition of energy into the shear and pseudo polarizations . using parseval s theorem @xmath154 , the decomposed khypp equation can be written as : @xmath155 where we summed the @xmath156 species and dropped @xmath156 superscripts , i.e. @xmath157 , @xmath158 $ ] , etc ... the iii - order sf on the rhs is split into three lines containing respectively ( 1 ) the strain of the shear polarizations on the shear and pseudo energies , ( 2 ) the strain of the pseudo polarizations on the shear and pseudo energies , and ( 3 ) the mixed terms accounting for the exchange of energy during the cascade between the shear and pseudo polarizations .
we anticipate that the mixed terms are negligible in the inertial range , thus we split the above equation into a system of equations for @xmath159 and @xmath160 , with their own cascade rate @xmath161 and @xmath162 , @xmath163 note that each equation contains only quadratic terms of a given alfvn polarization ( shear or pseudo ) even in the iii - order sf . on the rhs of equations - ,
the second terms represent an active contribution to the cascade of pseudo alfvn polarizations : if it is vanishing or negligible the pseudo alfvn polarizations can be said passive .
this separation will appear justified by the analysis of each term , which is presented in figure [ fig5 ] in the same format of figs .
[ fig3 ] , [ fig4 ] , i.e. with the the yaglom flux normalized by the perpendicular increment , @xmath164 , and the divergence normalized by the cascade rate , @xmath165 .
we consider separately the two contributions in the rhs terms of the equation for the pseudo energies , equation . in the first column
we have the iii - order sf accounting for the strain of shear polarizations acting on the pseudo energy , @xmath166 ( the first term in the rhs ) .
the yaglom flux ( upper panel ) is perpendicular , as in run b , since the vectorial part of @xmath167 is made of only shear polarizations , which have only perpendicular components .
its magnitude depends mostly on @xmath48 as can be guessed by the length of the arrows , with a small dependence on @xmath49 at large parallel scales .
this dependence is better appreciated on the map of @xmath168 ( middle panel ) that has slightly inclined isocontours instead of horizontal isocontours . the inertial range identified with the green area at level
@xmath169 is non uniformly distributed in the @xmath170 plane , occupying preferentially perpendicular scales @xmath171 independently of parallel scale . in the bottom panel
one can better locate the inertial range which is much more extended toward smaller scales in the perpendicular than in the parallel direction .
in the second column , we analyze the iii - order sf accounting for the strain of pseudo polarizations on the pseudo energy , @xmath172 , the second term in the rhs of equation .
note that the yaglom flux is now horizontal ( excepts at small parallel scales @xmath173 ) , since for a spectrum ( or ii - order sf ) with energy residing mainly in perpendicular wavevectors , pseudo polarizations have a dominant parallel component .
the contribution to the divergence ( middle panel ) is almost complementary to the previous term , being larger at large parallel and perpendicular scales while it is negligible in the inertial range .
this is better seen in the bottom panel , in which it is also shown clearly that this iii - order sf does not form an inertial range on its own , but it is responsible for a non - constant energy flux from large to small scales in both parallel and perpendicular directions .
one would be tempted to model the two terms , @xmath167 and @xmath174 , as 2d+1d components respectively , according to the orientation of the yaglom flux . however ,
such a decomposition does not work since the candidate for the 1d component ( @xmath174 ) has a strong dependence on @xmath48 ( second column , top panel ) , and its divergence does not yield any identifiable inertial range ( second column , middle and bottom panels ) .
finally in the third column we analyze the sum of the two terms just discussed , @xmath175 , the whole rhs of equation .
the yaglom flux is now oblique , with dependence on both @xmath92 and @xmath49 , the latter being mostly inherited from the strain of pseudo polarizations ( @xmath174 ) , indicating a smaller degree of two - dimensionalization for turbulence of pseudo alfvn polarizations , compared to the strong turbulence of run b. in the middle panel the inertial range extends to parallel and perpendicular scales in a way somewhat similar to run b , although the yaglom flux is quite different . comparing the bottom panels in columns one , two , and three
, one can see that the strain of both shear and pseudo polarizations contribute to the formation of the wide inertial range in the perpendicular direction @xmath149 , as well as to the shorter inertial range found at larger scales in the parallel direction @xmath176 ( their ranges coincide with those one obtained without separating shear and pseudo polarizations ) .
more precisely , shear polarizations contribute with a constant cascade at small scales ( @xmath177 ) , while pseudo polarizations control the injection of energy into the cascade at large scales ( @xmath178 ) .
we do not repeat the analysis of the strain of shear and pseudo polarizations on shear energy , equation , since it yields qualitatively similar results .
the only noticeable difference is that the yaglom flux @xmath179 is radially directed instead of being horizontal .
we plot in the fourth column the whole contribution to the cascade of shear energy , @xmath180 . comparing the third and fourth columns
, one can see that ( i ) shear and pseudo energy cascade qualitatively in a similar way ( top panels ) , although the yaglom flux for shear polarizations , @xmath181 , is more perpendicular ( more 2d ) ; ( ii ) the cascade rate is slightly stronger for shear energy than for pseudo energy ( middle panels ) ; and ( iii ) a clear inertial range is found for both shear and pseudo energies ( bottom panel ) following the decomposition of equations - . in the fourth column , bottom panel , we also plot the sum of the mixing terms , equation , along the parallel and perpendicular increments ( gray lines ) .
they are negligible compared to all other terms in the inertial range , indicating that the shear and pseudo polarizations cascade without exchanging their energy , as already pointed out by @xcite for decaying simulations . however , at large scales @xmath182 , they are of the same order of the term accounting for the strain exerted by pseudo polarizations ( ii column , bottom panel in figure [ fig5 ] ) , suggesting an exchange of energy between the shear and pseudo polarizations .
to summarize , in run c the freezing of large scales ( forcing ) causes an exchange of energy between pseudo and shear polarization at large scales . the two polarizations cascade in a similar manner ( same rate , same inertial range ) under the action ( strain ) of both the shear and pseudo polarizations .
the flux toward small scales that is triggered by pseudo polarizations is smaller in magnitude and not constant ( it is not a proper inertial range on its own ) , but it affects considerably the total cascade rate at large inertial - range scales ( @xmath178 ) . on the other hand , the shear polarizations control the constant energy flux at small inertial - range scales ( @xmath177 ) . thus according to the separation made in equations - ,
the two cascades are not totally independent . indeed ,
whatever the polarization considered ( @xmath159 or @xmath160 ) , the cascade is triggered by pseudo polarizations at the interface between injection scales and inertial range scales , while it is sustained by shear polarizations at smaller inertial range scales .
the presence and importance of the pseudo polarizations is revealed by the yaglom flux , the more pseudo polarizations are energetic , the more @xmath0 becomes oblique .
in measuring the cascade rate of solar wind turbulence through iii - order sf , one needs to make assumptions on the unknown anisotropy of @xmath0 . despite the limitations due to the imposed symmetries , this brings the advantage of increasing the statistic , on the one hand , and of avoiding the computation of the divergence of @xmath0 on the other hand .
two common assumptions usually employed to reduce in - situ data will be tested against our anisotropic dns ( run b and c ) in order to highlight possible systematic errors on the measure of the cascade rate in the solar wind .
the simplest assumption is that of isotropic turbulence , equation , whose expression is repeated here for convenience , @xmath183 and involves projecting @xmath0 along the direction of increments ( @xmath184 ) .
as discussed in section [ sec2 ] , the cascade rate is directly obtained from the iii - order sf without any need to compute derivatives .
this `` isotropic '' method has been mainly employed for fast polar wind @xcite on ulysses measurements .
admitting some form of anisotropy , the minimal assumption is that of axisymmetry . to avoid computing derivatives along the two increments , one again resorts on simplified geometrical models , as the 2d+1d turbulence employed by @xcite on wind and ace data in the ecliptic solar wind .
this method assumes that the iii - order sf has parallel and perpendicular components that depend only on the parallel and perpendicular increments , respectively ( the cascade is independent in the two directions ) .
the total cascade rate is obtained by combining the two independent equations for ( isotropic ) 2d - perpendicular and for 1d - parallel cascades ( equation and equation respectively ) .
such a `` hybrid '' cascade rate reads @xmath185 the isotropic and hybrid cascades are obtained by first computing the 3d iii - order sf , and then by applying the corresponding average ( isotropy or axisymmetry ) .
similarly , the `` true '' cascade rate is obtained from equation : @xmath186 we now compare in figure [ fig6 ] the true cascade rate ( thick solid line ) with the isotropic ( thin solid line ) and the hybrid ( thin dashed line ) cascade rate for run b and c ( top and bottom panels , respectively ) .
all cascade rates are normalized by the dissipation rate @xmath80 obtained directly from spectra , equation .
each panel represents a cut in the @xmath170 plane taken along a fixed direction that forms an angle @xmath85 with the mean - field direction .
the isotropic cascade rate is strongly angle dependent , independently of the run considered .
it returns increasing cascade rates for increasing angles , as can be expected since both runs b and c have a dominant perpendicular cascade .
it overestimates the true cascade rate at large angles ( @xmath187 ) while it underestimates the true cascade rate by a factor @xmath188 even at @xmath189 .
the hybrid method instead performs very well : it does not vary with angle @xmath85 and yields the correct cascade rate at oblique angles @xmath190 .
in the nearly parallel direction , both the hybrid and isotropic methods strongly underestimate the true cascade rate ( by a factor @xmath191 ) or they completely fail in yielding any cascade rate .
indeed , neither of the two methods is able to account for the dependence of the parallel component of the iii - order sf on the perpendicular increments , @xmath192 , which is instead fundamental in our strongly anisotropic runs ( e.g. figure [ fig2]a ) . indeed
, the 1d cascade entering the hybrid method is basically negligible at all angles ( red dashed line in fig [ fig6 ] ) .
this implies that the cascade is with a good approximation a 2d cascade in both our anisotropic runs . on the other hand , a linear scaling for the 1d cascade
is found in the solar wind ( e.g. @xcite ) , suggesting that the iii - order sf of solar wind turbulence is less anisotropic than that one of runs b and c.
we studied the anisotropy of mhd turbulence with or without guide field ( @xmath3 ) by means of structure functions ( sf ) of ii and iii order , in three direct numerical simulations of mhd turbulence at moderate and high resolution ( see table [ table1 ] ) .
we consider isotropic decaying turbulence ( run a ) , forced strong turbulence with mean - field ( run b ) , and forced weak turbulence with mean - field ( run c ) .
we used ii - order sf ( @xmath8 ) to characterize the anisotropy with respect to the local mean - field and to the global mean - field .
the anisotropy with respect to the local mean - field is scale dependent in all runs ( with or without mean - field ) and follows the critical balance prediction @xmath119 .
this confirms previous theoretical and numerical findings @xcite , in the local frame the anisotropy grows at smaller and smaller scales .
when instead we used as a reference the global mean - field , we found isotropy for mhd turbulence without mean - field ( run a ) and strong two - dimensionalization for turbulence with mean field , with run b being more anisotropic than run c ( the small - scale aspect ratio is @xmath193 and @xmath194 respectively ) .
surprisingly , for run c , we found a scale - dependent anisotropy consistent with critical balance even when @xmath8 is computed in the frame attached to the mean field @xmath3 , at variance with previous analysis in dns @xcite and in solar wind measurements @xcite which reported scale - independent anisotropy .
a possible explanation is that run c has a strong magnetic field ( @xmath195 ) and so the local and global frames almost coincide . in favor of this explanation ,
the anisotropy becomes scale independent at smaller scales .
however , there are other `` non - standard '' aspects of run c that could be related to scale - dependent anisotropy in the global frame : the special 3d spectrum that has an isotropic spectral slope but anisotropic energy levels in parallel and perpendicular direction @xcite , the strong excitation of pseudo alfvn polarizations ( see below ) , and the isotropic forcing ( freezing ) that maintains a magnetic excess at large scales @xcite
. these three aspects could be all related to one another and are under investigation .
we then analyzed the vectorial iii - order sf ( or yaglom flux , @xmath0 ) , which is related to @xmath8 computed in the global frame by a generalization to incompressible mhd @xcite of the von karman - howart - yaglom relation in hydrodynamics ( khypp equation in the following ) .
the iii - order sf , @xmath0 , is expected to be anisotropic for mhd turbulence with mean - field , but it has never been reported in dns or experiments ( except for rotating hydrodynamic turbulence experiments , @xcite ) . despite some anisotropic models exist @xcite , they have never been verified in dns .
we tested the khypp equation in isotropic decaying turbulence ( run a ) and found that the yaglom flux is almost purely radial , its divergence is isotropic , and the maximum of @xmath126 is of the order of the dissipation rate ( only a factor 0.8 smaller ) , confirming theoretical expectations @xcite .
however , the condition defining the inertial range , @xmath50 , is only fairly satisfied in our decaying turbulence that has an inertial range of only half a decade despite the high resolution ( @xmath133 ) .
more interestingly , we showed that the familiar concept of kolmogorov cascade in fourier space also applies in the real space with sf .
note that no assumption on locality of nonlinear interactions is needed to obtain the khypp equation . despite this
, we found that injection , cascade , and dissipation occur at the same rate and that the cascade is achieved via a fairly constant nonlinear transfer in the inertial range .
in particular , by comparing the divergence of the yaglom flux with the other terms appearing in the khypp equation ( dissipation term and decaying term ) , one can determine the dissipation scale and the injection scale of turbulence .
this has important applications in solar wind turbulence , since , due to expansion , it is an intrinsically decaying turbulence for which the injection scale is not easily identified @xcite . for strong turbulence with mean - field
, we found that @xmath0 has only perpendicular components and depends only on perpendicular increments ; this is a 2d turbulence as defined in equation .
the corresponding inertial range extends to small perpendicular scales , while it is shorter and located at larger scales in the parallel direction : the anisotropy of @xmath0 emerges as a non - uniform distribution of the scales having @xmath50 .
a similar non - uniform distribution is found in the forced , weak turbulence case ( run c ) , with an important difference concerning the orientation of @xmath0 . the latter is no more strictly perpendicular , but oblique . in particular , the yaglom flux is almost radial at large parallel scales in the inertial range and becomes more and more perpendicular at small parallel scales .
we interpret this behavior as a weaker two - dimensionalization of turbulence , since the oblique orientation of @xmath0 implies a dependence on parallel and perpendicular scales , in contrast to the purely 2d case in which @xmath196 . by decomposing fluctuations into shear alfvn and pseudo alfvn polarizations in run c
we also found that energetically important pseudo polarizations are responsible for the non - perpendicular iii - order sf .
the pseudo / shear decomposition allows us to prove that the two polarizations cascade independently , that is , without exchanging their energy , in the inertial range .
one can thus write separate khypp equations for pseudo and shear energies .
however , strictly speaking , the two cascades are not independent , since in each of them the strain of pseudo and shear polarizations enters the expression for the yaglom flux . in particular the strain exerted by pseudo polarizations is associated with a non - constant cascade of energy at the large inertial - range scales ; thus pseudo polarizations control the injection of energy in run c ( although becoming passive at smaller inertial - range scales , e.g. @xcite ) . in @xcite , it is shown how the equipartition of kinetic and magnetic energy at the large frozen scales is associated to the weaker anisotropy of the 3d spectrum of run c , having the property of a 3d iroshnikov - kraichnan turbulence . here
we argue that kinetic / magnetic equipartition at the largest scales induces an exchange of energy between pseudo and shear polarizations , thus allowing pseudo polarizations to control the cascade at the interface between injection and inertial - range scales .
finally , on the anisotropic runs b and c we tested two methods that are commonly applied to obtain the cascade rate in the solar wind .
the methods both rely on assumptions of the anisotropy of the iii - order sf , allowing one to obtain the cascade rate directly from @xmath0 ( i.e. without computing the divergence ) .
the first method assumes that @xmath0 is isotropic .
the second method ( hybrid method ) assumes axisymmetry and models the anisotropy as a geometric superposition of a 1d ( parallel ) cascade and an isotropic 2d ( perpendicular ) cascade .
we found that the isotropic method is strongly angle dependent , yielding correct cascade rates only at large angles between the direction of increments and the magnetic field , translating to angles between the magnetic field direction and the solar wind direction @xmath197 ( needless to say , the isotropic method works very well for run a ) .
for fast polar wind , emanating from stable coronal holes , turbulence is expected to have a strong anisotropy @xcite , comparable to our run b and c. we thus suggest that the angle dependence of the isotropic method is at the origin of the relatively small number of intervals ( @xmath198 ) in which the linear scaling @xmath199 is found in polar solar wind in - situ data @xcite .
in contrast , the hybrid method performs very well on our anisotropic runs , yielding an angle - independent cascade rate for @xmath200 .
both methods fails to obtain any cascade rate when increments are parallel to the mean - field .
this is due to the dependence @xmath192 that is excluded in both assumptions but present in all our runs .
this , together with the good performance of the hybrid method , indicates that the anisotropic runs b and c are fairly well described by a 2d cascade model , despite their different degree of anisotropy .
note that in the solar wind a 1d cascade , actually a linear scaling @xmath201 , is indeed measured with the hybrid method , indicating that solar wind turbulence in the ecliptic has a different and probably much weaker anisotropy compared to our dns .
we conclude by noting that the ratio @xmath202 employed in our simulation is lower than the actual value found in the solar wind , which is @xmath203 at scale of few hours .
the small value of fluctuations amplitude was chosen to highlight the effects of anisotropy and makes the comparison of our results with solar wind turbulence at large scales a bit weak , being more meaningful for scales shorter than a minute . in a future work
we plan to apply a similar analysis to simulations with larger ratio @xmath10 also dropping the assumption of axisymmetry .
in fact , cluster observations @xcite showed that the solar wind turbulence is not axis - symmetric ( but see @xcite for an explanation in term of an observational bias ) .
dns of mhd turbulence in the framework of the expanding box model @xcite also show that expansion causes non axis - symmetric 3d spectra @xcite .
note that expansion is also responsible for a differential decay of fluctuations , thus cross helicity and magnetic / kinetic imbalance of fluctuations also enter the khypp equation in the expanding solar wind @xcite .
another interesting direction is to explore the role of velocity shears on the khypp equation @xcite , which have been invoked to be the main driver for the turbulence cascade in both the ecliptic @xcite and the polar solar wind @xcite .
again , numerical simulations with the expanding box model seem to be a very promising tool to understand the combined effect of shear and expansion ( e.g. @xcite ) in shaping the cascade of solar wind turbulence .
_ acknowledgments _ this project has received funding from the european union s seventh framework programme for research , technological development and demonstration under grant agreement no 284515 ( shock ) .
website : project-shock.eu/home/,av acknowledges the interuniversity attraction poles programme initiated by the belgian science policy office ( iap p7/08 charm ) . ph acknowledges grant p209/12/2023 of the grant agency of the czech republic .
computational resources were provided by cineca ( grant 2014 hp10clf0zb and hp10cnmqx2 m ) . | the measure of the third - order structure function , @xmath0 , is employed in the solar wind to compute the cascade rate of turbulence . in the absence of a mean field @xmath1
, @xmath0 is expected to be isotropic ( radial ) and independent of the direction of increments , so its measure yields directly the cascade rate . for turbulence with mean field , as in the solar wind , @xmath0 is expected to become more two dimensional ( 2d ) , that is , to have larger perpendicular components , losing the above simple symmetry . to get the cascade rate one should compute the flux of @xmath0 , which is not feasible with single - spacecraft data , thus measurements rely on assumptions about the unknown symmetry .
we use direct numerical simulations ( dns ) of magneto - hydrodynamic ( mhd ) turbulence to characterize the anisotropy of @xmath0 .
we find that for strong guide field @xmath2 the degree of two - dimensionalization depends on the relative importance of shear and pseudo polarizations ( the two components of an alfvn mode in incompressible mhd ) .
the anisotropy also shows up in the inertial range .
the more @xmath0 is 2d , the more the inertial range extent differs along parallel and perpendicular directions .
we finally test the two methods employed in observations and find that the so - obtained cascade rate may depend on the angle between @xmath3 and the direction of increments . both methods
yield a vanishing cascade rate along the parallel direction , contrary to observations , suggesting a weaker anisotropy of solar wind turbulence compared to our dns
. this could be due to a weaker mean field and/or to solar wind expansion . |
Subsets and Splits
Data Mining Abstracts
This query searches for and retrieves 2000 documents containing the phrase 'data mining' in the abstract, providing limited but specific document samples.