article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the formation of amorphous solids as glasses and granular media , i.e the _ glass and jamming transitions _ , are still unsettled and fascinating questions in condensed matter physics despite all the works that have been devoted to this subject .
these phenomena occur in a variety of systems which , even if microscopically very different , share common features . among the others we recall supercooled liquids , colloidal suspensions and non - thermal jamming systems ( e.g. vibrated granular materials ) @xcite .
basic _ glassy properties _ include a dramatic slowing down of dynamics when a proper external parameter is tuned ( e.g. temperature is lowered for liquids ) and the occurrence of a complicated relaxation : non exponential and spatially heterogeneous @xcite .
when relaxation times become longer than experimental scales , equilibrium can no more be achieved and the systems freeze into an amorphous phase .
even the basic issues in understanding these phenomena remain unsolved . in particular
, it is not settled whether the dynamical arrest is due to the proximity of a phase transition and whether this putative phase transition is a static or purely dynamical one .
however experiments make it clear that , if an _ ideal glass transition _ occurs , it should have an unconventional behavior with mixed first and second order features . on the one hand , the divergence of relaxation times and the fact that both entropy and internal energy seem continuous ( for molecular liquids ) is indicative of a second order transition . on the other hand
, there is a discontinuous order parameter : the infinite time limit of the fourier transform of the density
density correlation ( analogous to edwards - anderson parameter for spin glasses ) has a finite jump at the transition .
this corresponds to the fact that the modulation of the microscopic density profile of the glass does not appear continuously from the flat liquid profile ( we generally refer for all systems to glass and liquid phase meaning the regime before and after the freezing into an amorphous solid ) . besides these mixed first / second order properties , another unconventional feature , compared to usual phase transitions , concerns the scaling of relaxation times .
different functional forms have been proposed in the literature .
a very popular and successful representation is the vogel - fulcher law .
this suggests that the _ logarithm _ of the relaxation time diverges as the inverse of the distance from the transition , i.e. @xmath3 , for molecular liquids .
finally , one of the most puzzling features is the absence of any experimental evidence of a static diverging correlation length . in particular ,
the dramatic slowing down of dynamics does not seem to be due to an increasing long range order as for e.g. ferromagnetic transition : typical glass configurations are not very different from instantaneous configurations of the liquid .
an enormous amount of theoretical approaches have been proposed in the last fifty years to describe these physical phenomena . however , if one takes the _ a priori assumption _ that a real glass or jamming phase transition takes place at a finite temperature @xmath4 ( at a density less than the close - packed one for the jamming transition ) then the number of scenarii reduces drastically .
leaving aside very phenomenological ones , there remain only two : either a very subtle static transition occurs ( random first order scenario @xcite ) or the transition is purely dynamical .
the first possibility invokes @xcite a thermodynamic transition similar to the one occurring for the so called discontinuous spin glass models , e.g. rem and p - spin models @xcite .
this is a one step replica symmetry breaking transition with a discontinuous ( edwards - anderson ) order parameter but no discontinuity for energy and entropy .
this approach has been successful in explaining and predicting many physical phenomena related to the glass transition @xcite .
although mean field models upon which it is based are very well understood by now , some of the finite dimensional predictions are still semi - phenomenological and need further theoretical works to be put on a firm and solid basis , see @xcite for recent works in this direction .
the possibility of a purely dynamical glass transition have been mainly investigated through the so called kinetically constrained models ( kcm ) ( @xcite@xcite and references therein )
. these are stochastic lattice gases based on the ansatz that glass or jamming transitions are due to effective geometrical constraints on the rearrangements of the atoms or molecules generated close to the transition .
static correlations beyond those present in dense liquids are assumed to play no role .
recently , in a joint work with d.s.fisher @xcite , we have shown that indeed some kcms display a purely dynamical transition on finite dimensional lattices with the basic features of the previously described glass - jamming transition .
note that these two scenarii are not necessarily in contrast because it might well be that , despite similar behaviors , the glass - jamming transition in molecular liquids , colloids and granular media are of different origin .
the thermodynamic scenario might apply to the glass transition of molecular liquids whereas jamming transitions might correspond to purely dynamic phase transitions .
here we give an extended explanation of the result in @xcite and we review some of our previous works ( and some others when needed ) on kcm .
we focus in particular on the different behaviors and tools needed to deal with different choices of the stochastic dynamics .
this paper is not intended to be an extensive review on kcms : for all the results previous to our works we refer to the reviews @xcite and references therein , while some more recent results for kinetically constrained spin models can be found in @xcite .
the outline of the paper is the following . in section [ kcm ]
we introduce the models distinguishing those with glauber ( kcsm ) and kawasaki ( kclg ) dynamics and further dividing both classes into cooperative and non cooperative models . in section [ kinds ]
we review the different types of dynamical arrest that occur and motivate the analysis of kcms from a theoretical point of view . in section [ noncoop ]
we present tools and results for non cooperative models . in section [ bethe ]
we describe the jamming transition that cooperative kcms display on bethe lattices . in section [ knight ]
we discuss the class of cooperative models ( knight models ) introduced in @xcite proving that they display an ideal glass transition .
in particular , we show that this transition is related to a new type of percolation transition , which we call _ jamming percolation_. in this section we also explain which are the tools one needs in general to study cooperative kcms , using knight models as an example
. finally , we present our conclusions in section [ conclu ] .
kcms are stochastic lattice gases with hard core exclusion , that is on each site there can be one or zero particle .
in other words , a configuration on a lattice @xmath5 is given by the set of two - valued occupation variables on each site @xmath6 : @xmath7 .
these represent occupied and empty sites ( particles and vacancies ) , respectively .
the dynamics is given by a continuous time markov process which consists of a sequence of jumps for models with conservative ( kawasaki ) dynamics and birth / death for models with non conservative ( glauber ) dynamics .
the former are also known as kinetically constrained lattice gases ( kclg ) , the latter as kinetically constrained spin models ( kcsm ) or facilitated spin models ( @xmath7 are interpreted as spin up and spin down ) .
for all the models we consider , dynamics satisfies detailed balance w.r.t .
bernoulli product measure , @xmath8 , at density @xmath9 .
thus , there are no static interactions beyond hard core and an equilibrium transition can not occur . however , in order for a move to be allowed , it is not enough to verify the hard core constraint .
indeed the jump or birth / death rates are non zero only if the configuration satisfies some additional local constraints , hence the name _ kinetically constrained_. these mimic the geometric constraints on the possible rearrangements in physical systems , which could be at the root of the dynamical arrest @xcite .
for example , in a highly dense liquid , numerical simulations @xcite have shown that a molecule is caged by its neighbors and it can not move on a substantial distance unless the neighbors move and the `` cage is opened '' .
experimental evidence of such _ cage effect _ for colloidal suspensions and granular materials have indeed been detected @xcite .
numerical simulations show that , for proper choices of the constraints , kcm display glassy features including stretched exponential relaxation , super - arrhenius slowing down and dynamical heterogeneities@xcite .
therefore , despite their simplified and discrete character they could capture the key ingredients of glassy dynamics , at least on proper time scales .
indeed , several works have been recently devoted to understanding the mechanism which induces these glassy properties and evaluating the typical time / length scales involved .
kcsms are endowed with a non conservative dynamics : each site changes from occupied to empty and from empty to occupied with rate @xmath10 and @xmath11 , respectively .
the value of @xmath12 , which encodes the kinetic constraints , does not depend on @xmath13 .
thus detailed balance is satisfied with bernoulli product measure , @xmath8 , at density @xmath9 . +
kcsms can be divided into two classes : non - cooperative and cooperative ones . for the former it is ( for the latter it is not ) possible to construct an allowed path which completely empties _ any _
configuration provided a proper finite cluster of vacancies is present somewhere .
we call this cluster a _ defect_. among non cooperative models we recall the fredrickson - andersen @xcite one spin facilitated ( fa@xmath14 ) . for
fa@xmath14 a move in @xmath15 is allowed only if at least one of the nearest neighbors is empty : @xmath16 if @xmath17 , @xmath18 otherwise .
this has recently received a renewed attention , especially since it has been proposed as a model for strong glasses @xcite .
it is easy to check that the presence of a single vacancy in this model allows one to empty the whole lattice .
+ among cooperative models we recall fa@xmath19 on an hyper - cubic lattice of dimension @xmath20 with @xmath21 @xcite .
here the constraint requires that at least @xmath19 of the surrounding sites are empty in order for the birth / death rate to be non zero .
as can be directly checked , for all these models it is not possible to devise a finite seed of vacancies which allows emptying the whole lattice .
consider , e.g. , the case @xmath22 , @xmath23 ( with periodic boundary conditions ) and focus on a configuration which contains two adjacent rows which are completely filled .
these particles can never be erased , even if the rest of the lattice is completely empty : it does not exist a finite defect which can destroy them , thus the model is cooperative .
the restriction on @xmath19 comes from the fact that all cases @xmath24 are very special because at any value of @xmath9 there exist finite sets of forever blocked particles , namely a fraction of the system is frozen at all densities .
although they might be interesting in themselves , they do not seem suitable to describe the slow dynamics close to glass - jamming transition because they are non - ergodic at any temperature / density .
+ a different class of cooperative models is the one we introduced in @xcite , which we call knight models . on a square lattice kinetic constraints are the followings : a move on @xmath15 can occur only if [ the two north - east _ or _ the two south - west fourth nearest neighbors are empty ] _ and _ [ the two north - west _ or _ the two south - east forth nearest neighbors are empty ] . the definition of the north - east , south - east , north - west and south - west neighbors and an example of the constraints is given in fig.[rule ] . in formulas @xmath25 where @xmath26\ ] ] @xmath27\ ] ] with @xmath28 , @xmath29 , @xmath30 , @xmath31 , @xmath32 , @xmath33 , @xmath34 and @xmath35 . + [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] finally , we recall the one - dimensional east model . in this case
the constraint requires a vacancy on the right nearest neighbors , that is @xmath36 .
note that on a finite lattice the presence of a single vacancy on the rightmost site allows to empty the whole lattice .
however the model does not belong to the above defined non - cooperative class , since the vacancy should occur in a specific position .
indeed , it is usually classified as cooperative : due to the direct nature of constraints the relaxation involves the cooperative rearrangements of large regions as @xmath37 @xcite .
kclg are endowed with a conservative dynamics .
a particle in @xmath15 attempts at a fixed rate to jump to a random nearby empty site @xmath38 and the move occurs with rate @xmath39 . here
@xmath40 does not depend on the configuration on @xmath15 and @xmath38 . since dynamics preserves the number of particles , detailed balance on finite volume is satisfied w.r.t .
the measure which is uniform on configurations with fixed particle number . on the other hand , on infinite volume
detailed balance holds w.r.t any bernoulli product measure @xmath8 .
again , we can classify kclg as non - cooperative and cooperative models .
for the former it is ( for the latter it is not ) possible to construct a finite group of vacancies such that for _ any _ configuration it can be moved all over the lattice and any jump of a particle to a neighboring empty site can be performed when the particle is adjacent to the empty cluster .
we will call this mobile cluster which facilitates jumps a _ macrovacancy_. as can be immediately checked the model in which there are no further kinetic constraints besides hard core , the so called normal lattice gas or symmetric simple exclusion process ( ssep ) , is non cooperative and the minimal macrovacancies are simple vacancies . among cooperative models we recall kob andersen model ( ka ) @xcite on a cubic lattice . here
a particle can jump to a neighboring site only if both in the initial and final position at least @xmath41 of its nearest neighbors are empty , where @xmath42 .
analogously , one can define ka models on hyper - cubic lattices of dimension @xmath20 and for different values of the parameter @xmath41 , with @xmath43 .
the restrictions on @xmath41 are due to the fact that @xmath44 corresponds to ssep while a model with @xmath45 has a finite fraction of particles that is frozen at any density ( cf .
fa models with @xmath46 ) .
again , one can directly check that all these models are cooperatives according to the definition above .
consider , e.g. ka in @xmath47 with @xmath42 and focus on a configuration containing a completely filled slab which spans the lattice and has a two by two transverse section .
any finite cluster of vacancies can never overcome this filled structure , therefore it is not possible to construct a finite macrovacancy that moves everywhere in the lattice .
another possible choice of cooperative kclg are conservative knight models @xcite defined as follows .
a move from @xmath15 to @xmath38 occurs only if the configuration satisfies the requirement needed to allow the move in @xmath15 _ and _ in @xmath38 for the non conservative knight model defined in previous section ( plus @xmath38 should be empty ) . on the other hand , an example of a non cooperative kclg
is provided by ka model with @xmath48 on a triangular lattice .
indeed , two neighboring vacancies form the required mobile macrovacancy as will be further explained in section [ noncoop ] .
let us take the point of view that glass and jamming transitions observed in experiments are not just cross - overs but they do correspond to a real thermodynamic or dynamic phase transition . for simplicity we will focus on lattice models assuming that the presence of the underlying lattice does not change qualitatively the physics . as explained in the introduction , from the experiments it is clear that _ ideal glass - jamming transitions _
( if they exist ) have peculiar features compared to standard first or second order phase transitions .
in particular , they are characterized by a diverging timescale although no growing static correlation lengthscale has ever been found , contrary to second order phase transition where the relaxation timescale diverges because long - range thermodynamic order sets in .
this strongly suggests that ideal glass - jamming transitions are purely dynamic phase transitions , i.e. they are characterized by an ergodicity breaking without any singularity in the thermodynamics .
is it possible ?
this question has been addressed in the mathematical physical literature ( see @xcite for a detailed discussion ) focusing on interacting hard core particle systems on finite dimensional lattices .
if particles interact via a short - range potential and the rates for exchanging a particle with a nearest neighbor vacancy are always strictly positive then it has been proved that in two dimensions the ensemble of the gibbs measures , @xmath49 , coincides with the ensemble of stationary ( under the dynamical evolution ) measures . in higher dimensions
it has been proved that @xmath49 coincides with all reversible stationary measures , where reversible measure means ( roughly speaking ) that it verifies detailed balance .
it is not known if there exist stationary non reversible measures .
it seems unlikely because they would lead to circulating probability currents without any net injection of energy inside the systems , see @xcite . in any case
they are certainly not relevant for the problem of the glass transition .
thus , as far as glass - jamming transitions are concerned , one can safely assume ( as it has been proved in two dimensions ) that the ensemble of gibbs measures coincides with the ensemble of stationary measures .
therefore , if a dynamical transition takes place , this would lead to more than one stationary measure and as a consequence to more than one gibbs measure : dynamical transitions has to be accompanied by thermodynamic transition .
+ would this mean that the ideal glass - jamming transitions must be only of thermodynamic origin and no pure dynamical transition can take place ? this is puzzling , because the absence of any indication of growing static correlation lengths seems to be at odds with the very existence of a thermodynamic transition inducing a diverging relaxation time .
if one does not want to let down the assumption that experimental glass - jamming transitions correspond to real phase transitions there remains mainly two possibilities .
a thermodynamic transition indeed takes place but it is of a completely new type and such that is not visible in any given @xmath50point correlation function . a good candidate for that
is the random first order theory @xcite .
otherwise , some hypothesis that lead to the previous conclusion ( no purely dynamical transition ) must be violated .
the only one that is reasonable to violate on physical grounds is the fact that the rates for exchanging a particle with a nearest neighbor vacancy have to be strictly positive .
consider for example hard spheres with brownian or monte carlo dynamics ( that is a reasonably good model for colloidal or even granular systems displaying glass - jamming transitions ) . if the density is high enough some moves can have rate zero .
this can have a dramatic consequence on the dynamics because now the configuration space can be broken into subsets that are not connected by any dynamical move when density is large enough , see fig.[coins ] for a trivial example .
formally , the markov chain associated to the dynamical evolution becomes reducible contrary to the case of the lattice gas with strictly positive rates .
+ as a conclusion the only possibility of having a purely dynamic transition seems to boil down to having zero ( i.e. degenerate ) rates depending on the configuration around a site .
this encodes the physical effect explained previously , see also fig .
[ coins ] , that is certainly present for systems of hard objects as colloids and granular media .
the dynamical transition that may be induced by degenerate rates is a reducible - irreducible phase transition . at low density ( high temperature )
almost any equilibrium configuration is contained in the same ergodic component . instead , at high density ( low temperature )
, the configuration space sampled with the equilibrium measure is fractured in many different components ( and no one of them covers almost all the space ) .
kcms are the simplest model that take this phenomenon into account . because of their simplicity
they allow to study in great detail the issue of glass - jamming transition as purely dynamical transition .
indeed for kcms ( see section [ kcm ] ) some rates are zero , hence their name _ kinetically constrained _ , and this implies that the configuration space on finite lattices is generically reducible .
in particular , there exist configurations which can not be connected one to the other using allowed moves .
for example , for fa model with @xmath23 and @xmath22 ( see section [ kcm1 ] ) , a configuration which is empty with the exception of two adjacent rows which are completely filled can never be connected to any configuration which does not contain this slab .
this implies that on any finite volume @xmath5 the process is not ergodic and @xmath8s are not the unique invariant measures .
the crucial issue is whether ergodicity is restored in the thermodynamic limit and whether this depends upon the value of @xmath9 .
two final remarks are in order .
first , as shown in @xcite , thanks to the product form of the equilibrium measure , for kcms establishing ergodicity at density @xmath9 corresponds to proving that there exists an irreducible set of configurations on @xmath51 such that @xmath52 is irreducible w.r.t .
the chosen dynamics on @xmath53 with periodic boundary conditions and @xmath54 . ]
( i.e. configurations connected one to the others by allowed paths ) which has unit probability w.r.t .
@xmath8 in the thermodynamic limit .
another way to put it is that the only possible way of breaking ergodicity is by the reducible - irreducible transition discussed above .
the second remark is that this type of transition is very different from the critical slowing down related to second order phase transitions . in those cases
the ergodicity breaking is induced by the thermodynamic limit ( the number of degrees of freedom goes to infinity and so the relaxation time may go to infinity ) . instead , for the type of dynamical transition
advocated above , finite size systems are typically non ergodic and the crucial question is whether in the thermodynamic limit the ergodicity is restored and up to which density . for non - cooperative models
it is immediate to conclude that ergodicity holds at any density @xmath57 .
indeed , the probability of finding at least a defect or a macrovacancy for an equilibrium configuration goes to one in infinite volume .
starting from this macrovacancy , by definition of non - cooperative kcsm , one can empty all the lattice .
therefore , two typical equilibrium configurations are almost surely connected by a path ( with strictly positive rates ) in the configuration space .
the situation is similar for non - cooperative kclg . in this case configurations with a macrovacancy
can be connected by a path which subsequently performs all nearby particle exchanges by previously moving the macrovacancy near the interested sites .
+ the case of cooperative kcms is much more involved and indeed it can lead to an ergodicity breaking transition . in general , as discussed previously , in order to prove ergodicity one should construct an irreducible set and prove that it has unit probability w.r.t . @xmath8 in the thermodynamic limit . for kcsms ( with periodic boundary conditions ) the relevant irreducible set is composed by configurations that can be completely emptied by allowed moves , as it can be found easily analyzing these systems at low density .
this component can be identified by the following deterministic procedure : subsequently remove all particles for which the constraint is verified until reaching a completely empty configuration or one in which there is a backbone of mutually blocked particles .
note that this backbone is uniquely determined by the initial configuration : it does not depend on the chosen order to erase particles .
if the backbone is empty , we say that the configuration is _ internally spanned _ and , by definition , it belongs to the relevant irreducible component .
therefore , ergodicity breaking takes place if and only if equilibrium configurations contain an infinite backbone of blocked sites in the thermodynamic limit .
note that , as discussed previously , we will consider only choices of the constraints such that a blocked cluster has necessarily to be infinite .
thus , the problem of the existence of an ergodicity breaking transition for cooperative kcsms can be reformulated as a percolation transition for the final configuration of the above cellular automata . for fa models
the latter coincides with bootstrap percolation procedure @xcite or k - core percolation @xcite : a particle is removed if it has less than @xmath41 neighbors .
the results in @xcite establish that bootstrap percolation converges to a completely empty lattice for all @xmath57 , namely no percolation transition occurs .
therefore , it is immediate to conclude that fa@xmath19 on any hyper - cubic @xmath20-dimensional lattice does not display an ergodicity breaking transition at any @xmath57 for all @xmath19 and @xmath20 . on the contrary
, for the cellular automata corresponding to knight models , a _ jamming percolation _ transition ( as we called it ) occurs at a critical density @xmath58 as proved in @xcite .
therefore , knight models in infinite volume are ergodic for @xmath59 and non ergodic for @xmath60 . in section [ knight ] we will explain the mechanism which induces the percolation transition for the blocked structures of knight models and discuss its character .
+ for cooperative kclg , the prove of ergodicity is more involved .
for example , for ka model on square lattices with @xmath48 , we have shown that the irreducible component which has unit probability in the thermodynamic limit is the one composed by configurations which can be connected by an allowed path to a configuration which have a frame of vacancies on the boundary @xcite . establishing that this set has unit probability in the thermodynamic limit is more involved than in the corresponding kcsm ( i.e. fa model ) .
this is due to the fact that there do not exist a deterministic bootstrap - like procedure which allows to establish whether a configuration belongs or not to the irreducible set .
finally note that , apart from ergodicity breaking , other types of transition might take place in lattice models for glass - jamming transitions : ( 1 ) a change from an exponential to a _ stretched exponential relaxation _ for density - density correlation in kcsm or its fourier transform for kclg , ( 2 ) a _ diffusive / subdiffusive transition _ in the behavior of a probe particle and ( 3 ) a _ break down of the conventional hydrodynamic limit_. in the case of cooperative kcsm numerical simulations @xcite seems to indicate that connected correlation functions , e.g. @xmath61 , decay in time as stretched exponential , @xmath62 $ ] , with a stretched exponent @xmath63 that is less than one and density dependent ( @xmath63 decreases as @xmath9 is increased ) .
analytical works based on mode coupling approximations @xcite support these results .
the standard explanation of this behavior is dynamic heterogeneity , which has indeed been experimentally detected @xcite and consists in having different relaxation times in different regions of the system .
the superposition of these exponential relaxations would then lead to an _ effective _ stretched exponential behavior .
although this is likely what happens for kcsms and real liquids , it has been recently proved in @xcite that relaxation is exponential at very large times for all cooperative and non - cooperative models ( in the ergodic regime ) considered so far .
more precisely , relaxation is exponential on times of the order of the inverse of the spectral gap of the liouvillian operator generating the dynamics ( that is the slowest relaxation time over all one time quantities ) and this is proved to be finite at any @xmath9 in the ergodic regime .
the discrepancy between rigorous and numerical results is likely due to the different time regimes that are investigated : a correlation function can be well approximated as a stretched exponential for times @xmath64 of the order of its relaxation time , which may of course be shorter than the slowest relaxation time given by the inverse of the spectral gap . ]
@xmath65 and decay as an exponential when @xmath66 .
an interesting question is whether the effective numerical behavior coming from fit on times of the order of @xmath65 can be made precise : in the scaling limit @xmath67 , with @xmath68 of the order of one , do correlation functions converge to stretched exponentials ? .
another transition different from the ergodicity breaking is a diffusive/ subdiffusive transition for the motion of tagged particles .
the result that the self - diffusion coefficient , @xmath69 , is positive at any density less than one , which is known to hold for ssep can not be immediately extended to kclg due to the degeneracy of jump rates .
in particular for ka models , from numerical simulations @xcite it had been conjectured that @xmath70 only below a finite critical density .
however , in @xcite it has been proved @xmath70 at any density @xmath57 for non - cooperative models and the result have been extended to cooperative models ( in the ergodic regime ) in @xcite . finally , a third possibility is a breakdown of the conventional hydrodynamic limit that usually holds on long length and time scales @xcite . for kclg ,
if the constraint is released , the macroscopic density profile evolves via the diffusion equation @xmath71 where @xmath72 for the normal lattice gas .
again , due to the presence of constraints , one can not apply the techniques which have been developed to establish this hydrodynamic limit for stochastic lattice gases @xcite .
furthermore , for some cooperative kclg , it has been conjectured @xcite that the macroscopic diffusion coefficient @xmath73 would vanish at high density , leading to a sub - diffusive evolution of density profiles . in @xcite
it is proved that for a class of non cooperative models hydrodynamic limit holds with @xmath73 vanishing ( as power law of @xmath74 ) only for @xmath75 .
extending these results to cooperative models would require establishing the scaling of relaxation times on finite lattices with the size of the lattice , which has not yet been settled for cooperative kclg .
in section [ kinds ] we explained that , thanks to the presence of the macrovacancies ( for kclg ) and defects ( for kcsm ) , in the thermodynamic limit non cooperative models are ergodic at any @xmath57 as it occurs for the corresponding models without kinetic constraints . in this section
we shall explain that the slow dynamics for non cooperative models can be understood in terms of motion of macrovacancies or defects , i.e. as renormalized lattice gases .
indeed these mobile regions , when @xmath75 , essentially perform independent random walks and substantial relaxation takes place only when they pass by .
so the high density dynamical behavior is encoded in their properties ( size , density , timescale for motion , ) . the analysis and the physical scenario is very similar for non - cooperative kcsm and kclg . in the following
we shall focus on kclg which are in a sense richer because , apart from the relation timescale , one can study directly the self - diffusion coefficient of tagged particles , @xmath76,bulk diffusion coefficients , @xmath73 , etc .
let us first focus on the scaling with density of @xmath76 .
consider for example ka model with @xmath48 on a triangular lattice ( see section [ kcm ] ) .
a particle in @xmath15 can be moved to the neighbouring empty site @xmath77 thanks to the fact that the other empty site ( @xmath78 ) is nearest neighbors of the particle both in its initial and final position . ]
the triangular lattice @xmath5 , represented in fig .
[ tria1 ] , is the union of sites in a square lattice @xmath79 and in its dual @xmath80 which is obtained by displacing @xmath79 of @xmath81 with @xmath82 , @xmath83 .
two sites @xmath84 are nearest neighbors if @xmath85 or @xmath86 or @xmath87 .
dynamics can be reformulated in terms of vacancy motion : a vacancy can move to a neighboring occupied site only if it has at least one empty neighbor both in its initial and final position .
since two neighboring sites of the triangular lattice share a common third neighbor , it is immediate to see that any of the two vacancies of a neighboring couple can move to the common third neighbor .
therefore a couple of neighboring vacancies is a finite size freely mobile cluster .
furthermore it is possible to move any given particle into an empty nearest neighbor provided the couple of vacancies is nearby .
for example , if we want to move a particle from @xmath15 to @xmath77 it is sufficient to put the macrovacancy in @xmath88 , as shown in fig.[tria1 ] .
thus , a couple of neighboring vacancies is a macrovacancy according to the definition in section [ kcm ] . a simple heuristic argument based on the independent motion of these macrovacancies leads to the correct high density dependence of @xmath76 .
let us focus on dimensions larger than two ( for lower dimension the reasoning changes due to recurrent properties of random walks ) .
call @xmath89 the density of macrovacancies and @xmath90 the timescale on which they move .
the self - diffusion coefficient is expected to be proportional to the inverse of the time @xmath91 on which each particle moves of one step . on the timescale
@xmath90 the number of particles that have jumped is of the order @xmath92 where v is the total number of sites .
thus we find @xmath93 as a consequence , at density close to one , we get @xmath94 . since for the @xmath48 ka on a triangular lattice macrovacancies are formed by two neighboring vacancies , in the limit @xmath95 we get @xmath96 and @xmath97 .
hence , @xmath98 .
the same arguments leads to @xmath99 for ssep , where macrovacancies are single vacancies .
the result for ssep has been rigorously proved in @xcite by establishing upper and lower bounds . for ka model on a triangular lattice , as we have shown in @xcite ( see also @xcite for a different choice of non cooperative constraints ) , it is also possible to turn the heuristic argument into a proof deriving upper and lower bounds @xmath100 with @xmath101 and @xmath102 independent from @xmath9 .
the representation of non - cooperative kcm as renormalized lattice gases turns out to be useful also to obtain these rigorous results .
let us recall the basic strategy since it allows to further understand the role of macrovacancies .
the idea is to start from the proof for ssep in @xcite and to modify it by letting macrovacancies play the role of vacancies .
the procedure is the following : first construct a suitable auxiliary model that is easy to analyze and then , using a variational formula for @xmath76 , show that @xmath103 where @xmath104 is a function determined explicitly ( that is strictly positive for @xmath57 ) .
the technique for the upper bound is more classical @xcite and the lower bound is more important because it shows that , despite the dynamics slows down , particles still diffuse on large distances and timescales .
therefore , we will just focus on the lower bound in the following . in @xcite the auxiliary model
is constructed in this way : the tracer has a neighboring vacancy at time zero and the only possible moves are jumps of the tracer and exchanges of the occupation variables in @xmath38 and @xmath105 , where both @xmath38 and @xmath105 are nearest neighbors of the tracer .
note that the latter moves are not allowed for ssep but one can reconstruct them trough a path of allowed moves thanks to the presence of the vacancy . for the auxiliary model
it is immediate to prove that @xmath106 and is bounded from above and below by density independent constants .
indeed , a move of the tracer from @xmath15 to @xmath107 can always occur via two steps : since the tracer has always at least one neighboring vacancy , this can be brought ( in one move ) in @xmath108 and then the jump of the tracer from @xmath15 to @xmath108 can occur . using a variational formula for @xmath76 @xcite and comparing the dynamics of the normal lattice gas and
the auxiliary model leads to @xmath109 .
the term @xmath110 comes from the requirement of having at least a vacancy at time zero near the tracer ; @xmath111 is a density independent constant which accounts for the maximal length of the path needed to exchange the occupation variable in @xmath15 and @xmath105 trough neighboring jumps ( the only moves allowed for ssep ) . for ka on the triangular lattice
one defines the following auxiliary process .
we require a macrovancancy near the tracer in the initial configuration and we allow only moves which either displace the tracer to a nearest neighbor or move the macrovacancy into another couple of sites which are also near the tracer ( for a precise definition of the auxiliary model see @xcite ) .
again , the latter move is not allowed in ka but can be reconstructed by a finite path of allowed moves . as before , it is easy to show that @xmath106 and is independent from @xmath9 . comparing ka and auxiliary dynamics we get @xmath112 , where @xmath113 comes from the requirement of having a macrovacancy .
note that this procedure is generalizable to all non cooperative models : if the macrovacancy is formed by @xmath114 empty sites it leads to @xmath115 .
therefore , at least for the motion of the tracer , non cooperative kclg are simply renormalized ssep : the typical diffusion time of the tagged particle ( i.e. the inverse of @xmath76 ) goes here as the inverse of the density of macrovacancies , instead of vacancies .
the strategy of comparing dynamics with the one of a faster unconstrained model and reconstructing the moves of the latter by a proper path of allowed moves can be also used to derive upper bounds for the density and size dependence of the relaxation time , @xmath65 , for non - cooperative kclg and kcsm ( see @xcite and @xcite , respectively ) .
the paths are always constructed using the fact that , if there is a macrovancancy ( defect ) , we can move it everywhere and facilitate any nearest neighbor jump ( birth / death ) of particles . therefore comparison with the unconstrained model in general gives an extra factor @xmath116 ( the cost of creating the mobile region ) times a term related to the length of the path ( which can bring a dependence both on the lattice size , @xmath117 , and on @xmath9 ) .
note that in principle also an entropy term should be accounted ( for the bounds on @xmath65 and @xmath76 ) , counting all the possible configurations which have to pass trough the same bottleneck @xcite . however , this is just a constant factor for non cooperative models ( more precisely it is the entropy on a number of sites proportional to the size of the macrovacancy ) and can not change the scaling in @xmath74 .
+ from results in @xcite , one has that for non - cooperative models the dependence of @xmath65 on the lattice size @xmath117 is the same as for models without constraints : @xmath118 for kclg , @xmath65 bounded by a finite constant uniformly in @xmath117 for kcsm ( where @xmath65 is the inverse of the spectral gap of the lioviullian operator , see @xcite ) . on the other hand ,
since relaxation occurs trough the macrovacancies which become rarer at higher density , the density dependence is different .
for example , for ka on a triangular lattice with particle sources at the boundary , we find @xmath119 , to be compared with @xmath120 for ssep .
bethe lattices ( or in the mathematical literature random c - regular graphs ) are often used in the physics literature as an approximation of finite dimensional , e.g. hyper - cubic , lattices . because locally they have a cayley tree structure with connectivity @xmath121 , their analysis is greatly simplified and one can often obtain exact results on the thermodynamics and the dynamics of the model embedded on them . for proper choices of the constraints ,
kcms display a jamming transition on bethe lattice @xcite and , thus , provide an almost solvable example of jamming transition . in the following we present a summary of the results that have been obtained focusing on a simple case : the f - facilitated fredrickson - andersen model on a bethe lattice with connectivity @xmath122 ( we consider @xmath123 in order to avoid any finite blocked cluster ) .
the results remain qualitatively the same for particle models but they are more tedious to derive @xcite . as discussed previously , a jamming ( and ergodic ) transition for fa models will take place if and only if at a certain density an infinite cluster of blocked ( jammed ) particles appears .
all the particles inside this cluster must have at least @xmath19 blocked neighbors in order to be blocked themselves .
thus , this infinite cluster is the spanning cluster of the so called @xmath124core percolation @xcite , ( or bootstrap percolation @xcite ) a problem that has received a lot of attention recently @xcite . for a given site
the bethe lattice local structure is cayley - tree like with @xmath122 branches going up from each node and one going down . using this crucial feature
it is easy to write a self - consistent equation on the probability @xmath125 that a site is occupied by a blocked particle ( belongs to the @xmath124core ) because it has more than @xmath126 neighboring particles above it which are blocked ( without taking advantage of vacancies below)@xcite : @xmath127 except for @xmath128 for which the transition is continuous , the solution of this equation leads to @xmath129 for @xmath130 and @xmath131 with @xmath132 , see the example for @xmath133 and @xmath23 in fig .
[ figp ] . ) as a function of @xmath134 for @xmath133 and @xmath23.,width=340 ] the number of blocked particles or the number of sites belonging to the @xmath124core are polynomial functions of @xmath125 that can be derived easily and that have the same behavior of @xmath125 .
they are at the same time discontinuous as in first order phase transitions and singular because of the square root , as in second order phase transitions .
the mechanism behind this behavior has been understood in detail @xcite .
in particular the singular square root behavior is due to the extreme fragility of the infinite spanning jammed cluster at the transition and to the existence of a related diverging lengthscale . in fact , close to the transition , a given jammed particle is connected to clusters of jammed particles that are only marginally blocked , i.e. such that removing only one of their blocked neighbors is enough to unblock them .
these clusters form the so called _ corona _ of the infinite jammed cluster ( or @xmath124core spanning cluster ) .
the size distribution of corona clusters can be computed analytically @xcite and it has been found that their average size diverges at the jamming ( @xmath124core percolation ) transition as @xmath135 coming from the jammed phase .
this explains the singular square root behavior found previously@xcite : roughly speaking , decreasing the temperature from @xmath136 to , say , @xmath137 ( where @xmath138 ) one unblocks first a number of particles proportional to @xmath139 .
unblocking a particle unblocks also all the marginally blocked ( corona ) particles attached to it .
this leads to a `` domino effect '' such that the effective number of unblocked particles is @xmath140 , i.e. the net change in the fraction of blocked site is @xmath141 . since corona clusters ,
i.e. marginally stable clusters , percolate at the transition one expects an associated diverging lengthscale .
this is indeed the case , see @xcite : coming from the jammed phase there is lengthscale that diverges as @xmath142 .
the analysis of the jamming transition coming from the unjammed phase is much more difficult and one has to resort to numerical simulations . in @xcite the persistence , @xmath143 where @xmath144 equals one
if the site @xmath145 has remained in the same state ( empty or occupied ) from time @xmath146 to time @xmath64 and zero otherwise , and the occupation variable correlation function , @xmath147 have been measured starting from equilibrium initial conditions . , @xmath133 case for different temperatures approaching the critical temperature @xmath148.,width=340 ] we plot in fig .
[ figk3f2 ] the persistence as a function of time for the @xmath23 , @xmath133 case .
different curves corresponds to different densities approaching the critical point .
is @xmath149 . ]
the correlation has a similar behavior @xcite .
the existence of the plateau is a direct consequence of the discontinuous character of the jamming transition : the fraction of blocked particles is strictly positive at the transition coming from the jammed phase instead it is zero coming from the unjammed one .
this translates into a dynamical behavior such that @xmath150 and @xmath151 and equal to the fraction of blocked sites .
associated to the non - commutation of these two limits there is a diverging time scale , the time on which @xmath152 equals , say , half of their plateau value .
the numerics indicates that it follows a power law divergence @xmath153 with @xmath154 .
finally , it has been found a diverging dynamical lengthscale associated with the divergence of the timescale @xcite ( let us recall that no static correlation can be found because the equilibrium measure is uncorrelated from site to site ) . in order to unveil
this length one has to focus on the fluctuations of the persistence or the correlation@xcite .
these are encoded in the dynamical susceptibility @xmath155 ^{2}\rangle / t\ ] ] @xmath156 develops a peak ( see fig.[fig_chi ] ) for @xmath157 that diverges for @xmath158 . vs time @xmath64 at temperature @xmath159
system size @xmath160.,width=340 ] this divergence is related to the fact that close to @xmath4 more and more particles have to evolve in a correlated way in order to make the system relax .
note that , exactly as for ferromagnets the divergence of the fluctuations of the magnetization is a signature of a diverging lengthscale , so it is the divergence of the peak of @xmath156 at the jamming transition .
however , the precise characterization of this dynamic lengthscale and its relationship with the results on the jammed phase have still to be worked out .
let us summarize the features of the jamming transition found on the bethe lattice .
it has a _ first order _
character because the fraction of jammed particles and the long - time limit of correlation and persistence functions ( called in spin - glasses the edwards - anderson parameter ) are discontinuous at the transition . at the same time
it is characterized , as _ second order _ phase transitions , by a diverging timescale and diverging lengthscales .
although the detailed connection between time and lengthscales ( from the jammed and unjammed phase ) has still to be worked out from numerics it seems clear that all these quantities diverge as power law in @xmath161 .
it is interesting to note that qualitatively these features are also shared by mean - field disordered systems , as p - spins models , close to their dynamical transition and the mode coupling theory of the glass transition @xcite .
the following section is devoted to the analysis of jamming transitions in finite dimensional models . as we will see some of the features found on the bethe lattice persist , for example the mixed ( first and second order ) character of the transition .
instead others , as the divergence of the relaxation times , change and actually become closer to what expected for glass - jamming transitions : now the logarithm of @xmath162 diverges as a power law in @xmath161 , i.e. in a vogel - fulcher like form .
in this section we give an extended explanation of the results obtained in collaboration with d.s.fisher @xcite for the knight models defined in section [ kcm ] . as already mentioned an _
ergodicity breaking transition _ occurs at @xmath58 : above @xmath1 a finite fraction of the system is frozen .
this is due to an underlying percolation transition for the clusters of mutually blocked particles , which we called _
jamming percolation_. let us give a brief summary of the main results letting for the next sections their derivation . as discussed in section 3 the dynamical transition of the knight models can be studied using the following cellular automata @xcite .
start from an initial configuration sampled with equilibrium measure , @xmath8 , then empty an occupied site if the constraint is satisfied .
then continue the procedure until no more particles can be removed .
this final configuration is either completely empty or one that contains a percolating cluster of particles which do not satisfy the constraint .
let us call @xmath163 the density of this final configuration .
we will show that at the critical density of site directed percolation on a square lattice , @xmath164 , a discontinuous percolation transition occurs : @xmath165 for @xmath166 and @xmath167 for @xmath168 . since the final backbone for this cellular automata contains all the particles that are frozen under the stochastic evolution of knights , we conclude that below @xmath169 knight model is ergodic and ergodicity is broken above .
furthermore , the fraction of the system which is frozen coincides with @xmath163 and has a finite jump at the transition . + as for the fa on a bethe lattice , we find that the dynamical transition transition has a _ first order _ character with a discontinuous edwards - anderson parameter and , at the same time , is characterized by diverging time and lengthscales , as second order like transitions . + the diverging lengthscale can be determined studying finite size effects .
consider the model on a finite lattice @xmath53 with periodic boundary conditions and evaluate the probability @xmath170 that it can be completely emptied by allowed moves , i.e. the probability that @xmath53 is _ internally spanned_. since at any fixed density @xmath166 the final backbone of the cellular automata is empty , @xmath170 goes to one in the thermodynamic limit . on the other hand ,
if the limit @xmath171 is taken first , @xmath170 goes to zero .
the relevant length @xmath172 is the crossover length which separates the two regimes : @xmath173 for @xmath174 and @xmath175 for @xmath176 . in other words
, @xmath172 is the dynamical length which corresponds to the typical size of blocked clusters ( i.e. the incipient percolating clusters for jamming percolation ) . as we will explain , by analytical arguments we find @xcite that @xmath177 .
here @xmath178 where @xmath179 is the critical exponent of the parallel directed percolation ( dp ) correlation length , @xmath180 , and @xmath181 is the exponent relating parallel and transverse dp correlation lengths , @xmath182 .
we also rigorously proved , are given for granted in physical literature .
indeed , they have been verified both by numerical simulations and analytical works trough renormalization technique ( see @xcite for a review ) . however , a rigorous mathematical proof of these results for dp is still lacking . our bounds for @xmath183 are rigorous modulo the assumption @xmath184 , @xmath185 . ] the upper and lower bounds @xmath186 @xcite .
these establish that @xmath183 diverges faster than power law and the lower bound coincides with the value we expect for @xmath183 . + due to this diverging length , the relaxation time @xmath65 should diverge as @xmath187 with @xmath188 . indeed , relaxation should require the diffusion of regions of density @xmath183 ( see @xcite for similar results for fa model ) . in @xcite
the rigorous bound @xmath189 is proved , where @xmath65 is the inverse of the spectral gap of the lioviullian operator @xmath190 ( i.e. the worst relaxation time on all one time quantities ) .
therefore knight models exhibit a vogel - fulcher like relaxation : the log of times diverge as power law when the critical density is approached .
this is different from the findings for fa model on bethe lattices , were @xmath65 diverges as power law ( see section [ bethe ] ) .
since in the knights model there are no static interactions between the particles it is clear that their glass - jamming transition is purely dynamical and not related to any thermodynamic transition . in the following
we shall explain the arguments leading to the proof of an ergodicity breaking transition at @xmath169 and of its mixed first / second order character ( we refer to @xcite for rigorous proofs ) .
these will unveil the mechanism which induces these remarkable properties which are typical of an ideal glass transition .
also , it will be clear why we choose the specific form of knight constraints and how one can modify or generalize them to higher dimension keeping a transition with the same character .
let us prove that ergodicity is broken above @xmath169 , namely a finite fraction of sites is frozen .
this is equivalent to showing that the origin belongs with finite probability to a percolating cluster of frozen sites . here , as in @xcite
, we call a site _ frozen _ if it can not be unblocked even by first emptying with allowed moves an arbitrarily large number of sites .
[ ] [ ] a ) [ ] [ ] b ) [ ] [ ] c ) consider the sublattice which is obtained from the square lattice by erasing all sites except the origin , its north - east and south - west neighbors , the north - east and south - west neighbors of the latter and so on as in fig .
[ lattice ] a ) , b ) . then construct a directed graph on this sublattice by drawing arrows connecting each site to its north - east neighbors as in fig .
[ lattice ] c ) .
notice that this corresponds simply to a two dimensional square lattice ( only rotated and squeezed ) .
also , since there are no static correlations in the equilibrium measure , if the configuration on the original lattice is chosen with @xmath8 , the same holds for the configuration on the sublattice .
therefore the results for site directed percolation @xcite imply that , if @xmath191 , the origin belongs with finite probability to an infinite directed ( i.e. following the direction of the arrows ) percolating cluster of occupied sites .
it is now easy to check that if we restore the whole lattice and consider knight dynamics , all sites belonging to the cluster which percolates on the sublattice are frozen : _ a finite fraction of the system is frozen above @xmath169_. indeed , each of these sites has at least one occupied ne neighbor and at least one occupied sw neighbour , therefore it is blocked along ne - sw diagonal . in the following , we call _ ne - sw cluster _ a directed occupied path on the sublattice obtained with the procedure above .
the definition is analogous for the _ nw - se cluster _
, where the sublattice is the one constructed erasing all sites but the origin , its nw and se neighbours and so on .
note that in order to establish the existence of frozen clusters we have not used the possibility of blocking along the nw - se diagonal . on the other hand , this should be taken into account if we want to show that a given configuration does not contain frozen sites as we will do in the following section .
let us prove that the system is ergodic in the thermodynamic limit for @xmath166 , which corresponds to showing that the fraction of frozen sites is zero .
note that if the rule contained blocking only along one of the two diagonals , ergodicity would follow immediately from the fact that occupied directed paths do no percolate below @xmath169 . however , for the constraints we have chosen , a percolating directed path implies a frozen cluster but the converse is not true .
indeed , due to the fact that a site can be blocked along either the ne - sw or the nw - se diagonal or both , a ne - sw cluster can be blocked either if it spans the lattice or if it is finite but both its ends are blocked by a t - junction with nw - se percolating paths ( see fig .
[ block1 ] a ) ) . by using such t - junctions
it is also possible to construct frozen clusters which do not contain neither a spanning ne - sw nor a spanning nw - se cluster : all ne - sw ( and nw - se ) clusters are finite and are blocked at both ends by t junctions by finite nw - se ( ne - sw ) ones ( see figure [ block1 ] b ) ) . as we will show in detail , these t - junctions are crucial to make the behavior of jamming percolation very different from site directed percolation although the two transitions share the same critical density .
[ ] [ ] a ) [ ] [ ] b ) let us now show how the proof or ergodicity below @xmath169 works .
the strategy consists in constructing a set of configurations , @xmath52 , on @xmath53 and show that @xmath52 are internally spanned and @xmath54 .
this , as explained in section [ kinds ] , concludes the proof of ergodicity for @xmath166 and ( together with results in previous section ) establish that ergodicity breaking occurs at @xmath169 .
consider a configuration within which there is an empty square of size @xmath192 and focus on the sufficient conditions to empty the next shell , i.e. to construct an allowed path which empties the @xmath193 square .
the initial vacancies guarantee that we can empty a centered segment of length @xmath194 external to each side .
consider for example site @xmath15 in fig .
[ emptying ] a ) , both its sw and se couples are inside the empty square : the constraint is satisfied on both diagonals and @xmath15 can be emptied .
[ ] [ ] x [ ] [ ] n [ ] [ ] a ) [ ] [ ] b ) [ ] [ ] n [ ] [ ] i(n/4 ) square .
site @xmath15 can be emptied since its se and sw neighbours are all inside the empty square .
b ) the triangular structure of height @xmath195 which can be emptied above each side .
@xmath196 denotes the @xmath197 sites on each corner which are not guaranteed to be emptied ( unless a proper condition is fulfilled by the configuration outside the empty @xmath192 nucleus).,title="fig : " ] square .
site @xmath15 can be emptied since its se and sw neighbours are all inside the empty square .
b ) the triangular structure of height @xmath195 which can be emptied above each side .
@xmath196 denotes the @xmath197 sites on each corner which are not guaranteed to be emptied ( unless a proper condition is fulfilled by the configuration outside the empty @xmath192 nucleus).,title="fig : " ] it is immediate to check that , even if the rest of the lattice is occupied , this procedure can be continued until emptying the triangular structure above each side depicted in fig . [ emptying ]
this is formed by centered lines whose length decreases of four sites at each step , therefore its overall height is @xmath195 where @xmath198 stands for the integer part of @xmath15 . in order to empty the @xmath199 square ,
five sites remain to be emptied on each corner : those indicated by question marks in fig.[emptying ] b ) .
consider for example one of these site on the top left corner , indicated as @xmath38 in fig.[necessary ] .
the constraint along the nw - se diagonal is verified , since both its se neighbors are inside the empty square . instead ,
since neither the ne nor the sw couples are contained in the empty region , @xmath38 can be blocked along ne - sw diagonal and in this case we can not empty it . however , it is possible to derive a necessary condition for @xmath38 to be frozen , which gives in turn a sufficient condition to construct an allowed ( i.e. verifying the constraint at each elementary move ) path in the configuration space which allows to empty @xmath38 .
focus on the square of size @xmath200 centered on the top left corner and containing half of the triangular structure associated to the up and left sides , i.e. the square inside the dashed region in fig .
[ emptying]b ) .
consider the sublattice constructed from @xmath38 inside this square by erasing all sites except its north - east and south - west neighbors and so on ( as we did for the origin in figure [ lattice ] ) .
it can be directly checked that @xmath38 can be frozen only if it belongs to a ne - sw path , i.e. a directed path on this sublattice , which spans the square ( see fig . [ necessary ] ) .
[ ] [ ] [ ] [ ] @xmath38
b ) and show the necessary condition for @xmath38 to be frozen : it should belong to a ne - sw cluster spanning the dashed square . otherwise
: if one of its end is free it can be unblocked from there ; if they are both blocked by t - junctions with nw - se clusters these terminate inside the empty region and can be unblocked from there . ]
this is due to the fact that any occupied cluster along the nw - se diagonal can be unfrozen if it terminates inside the empty region ( @xmath192 square plus triangular region ) .
therefore , a sufficient condition to construct an allowed path to empty @xmath38 ( and the other remaining four sites at the top left corner ) is that in the dashed square region there is not a percolating ne - sw cluster .
since the size of this region is proportional to @xmath201 and @xmath166 , such a cluster occurs with a very small probability , @xmath202 for large @xmath201 with @xmath203 the parallel correlation length for dp @xcite . as a consequence
the probability of emptying @xmath38 increases at least exponentially fast to one when the size of the empty nucleus , @xmath201 , is increased .
the same argument can be applied to the other four sites indicated by question marks and to those on the other three corners of the lattice , concluding that the probability @xmath204 to go from the empty @xmath192 to the empty @xmath193 nucleus is bounded from below by @xmath205 .
thanks to the exponential increase towards one , the probability that this procedure can be continued up to infinite size stays strictly positive because it is bounded from below by @xmath206 . finally , using the @xmath207 different positions for the initial empty nucleus
, we conclude that the probability for @xmath53 to be internally spanned converges to one in the thermodynamic limit at any @xmath166 .
this is due to the fact that , in order to prevent the expansion of a large empty nucleus , we should require long dp clusters , which are highly improbable below @xmath169 .
results in sections [ noergo ] and [ ergo ] prove that an ergodicity breaking transition occurs for knights at @xmath169 , due to the percolation transition of blocked structures . in the present and the following section
we show that this transition , which we call _ jamming percolation _ , have features which are qualitatively different from those of dp and any conventional percolation transition .
indeed it is discontinuous , i.e. the density of the frozen clusters have a finite jump at the transition ( the critical clusters are compact rather than fractal ) and their typical size increases faster than any power law when @xmath208 . in order to prove discontinuity we construct a set of configurations which have finite probability on the infinite lattice at @xmath169 and show that the origin is frozen for all these configurations .
we will make use of the blocked structures containing t - junctions among ne - sw and nw - se clusters which have been introduced in section [ ergo ] .
consider a configuration in which the origin belongs to a ne - sw path of length @xmath209 : this occurs with probability @xmath210 . now
focus on the infinite sequence of pairs of rectangles of increasing size @xmath211 with @xmath212 , @xmath213 and intersecting as in fig.[disc ] .
[ ] [ ] a ) [ ] [ ] [ ] [ ] b ) [ ] [ ] c ) [ ] [ ] @xmath214 [ ] [ ] @xmath215 [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] describer in the text .
b ) dotted non straight line stand for ne - sw ( nw - se ) clusters spanning the rectangles with long side in te ne - sw ( nw - se ) direction .
c ) frozen structure containing the origin , title="fig : " ] describer in the text .
b ) dotted non straight line stand for ne - sw ( nw - se ) clusters spanning the rectangles with long side in te ne - sw ( nw - se ) direction .
c ) frozen structure containing the origin , title="fig : " ] describer in the text .
b ) dotted non straight line stand for ne - sw ( nw - se ) clusters spanning the rectangles with long side in te ne - sw ( nw - se ) direction .
c ) frozen structure containing the origin , title="fig : " ] if _ each _ of these rectangles with long side along the ne - sw ( nw - se ) diagonal contains a ne - sw ( nw - se ) percolating path ( dotted lines in fig .
[ disc ] b ) ) , the origin is frozen .
this can be directly checked : an infinite backbone of mutually blocked particles constituted by pieces of these paths connected by t - junctions ( continuous line in fig .
[ disc ] c ) ) occurs .
therefore the probability that the origin is frozen , @xmath216 , is bounded from below by @xmath217 where @xmath218 is the probability that a rectangle of size @xmath219 with short side in the transverse direction is spanned by a dp cluster .
this probability converges to one exponentially fast in @xmath220 due to the anisotropy of critical clusters for directed percolation .
recall that there are a parallel and a transverse length for dp with different exponents , i.e. a cluster of parallel length @xmath221 has typically transverse length @xmath222 .
let us divide the @xmath219 rectangle into @xmath223 slices of size @xmath224 .
for each slice the probability of having a dp cluster along the parallel direction at @xmath169 is order unity .
thus , the probability of _ not _ having a dp cluster in each of the slice is @xmath225 $ ] . from this result and above inequality for @xmath114
, it follows immediately @xmath226 .
therefore the infinite cluster of jamming percolation is `` compact '' , i.e. of dimension @xmath20 at the transition .
note that to obtain discontinuity two ingredients of the constraints were crucial : the existence of two transverse blocking directions each with an underlying percolation transition ; the anisotropy of these transitions .
indeed , anisotropy is necessary to have the increase towards one of the probability that above rectangles are spanned when their size is increased . in turn , this is necessary to get a finite probability for the construction which freezes the origin . in this section
we explain the arguments in @xcite leading to the result @xmath227 for @xmath166 , where @xmath228 . as already explained , @xmath172 is the crossover length dividing the regime in which the probability for a finite region to be internally spanned goes to zero or to one when @xmath171 , i.e. the length below which finite size effects are important . for jamming percolation
it corresponds to the diverging size of the incipient spanning clusters and for knight models to the typical size of the region that has to be rearranged to unblock a given site .
let us sketch separately the arguments leading to @xmath229 and to @xmath230 with @xmath231 two positive constants . to establish the lower bound we construct a set of configurations containing a frozen backbone and we show that the probability of this set goes to one when @xmath171 and @xmath232 with @xmath233 .
again , we make use of the t - junctions described in section [ ergo ] .
consider the set of ne - sw and nw - se paths of length @xmath234 intersecting as in fig .
[ lower ] .
[ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] as can be directly checked , this structure can be emptied only starting from its border since each finite directed path terminates on t - junctions with paths in the opposite direction . therefore ,
if the construction is continued up to the border and we consider periodic boundary conditions , all sites in the structure are frozen .
furthermore , a similar frozen backbone exists also if one or more of these paths is displaced inside an adjacent rectangular region of size @xmath235 , as shown in fig .
[ lower ] .
therefore the probability that the regions is not internally spanned , @xmath236 , is bounded from below by the probability that _ each _ of these rectangles contains at least one path connecting its short sides .
this leads to @xmath237 where @xmath238 is the number of the rectangles contained in the structure and @xmath239 is the probability of not having a dp path in a region @xmath235 as long as @xmath240 ( see previous section ) .
the inequality ( [ ineqxx ] ) leads to @xmath241 for @xmath242 , therefore @xmath243 . on the other hand , in order to establish the upper bound , we show that for @xmath244 there is typically an nucleus of vacancies which can be expanded until emptying the whole lattice . from the results in section [ ergo ]
it is easy to see that the probability to expand an empty nucleus to infinity is dominated by the probability of expanding it up to @xmath245 .
indeed , above this size the probability of an event which prevents expansion is exponentially suppressed .
therefore , considering the @xmath246 possible position for a regions that it is guaranteed to be emptyable up to size @xmath203 , we get that the probability that @xmath53 is internally spanned is roughly bounded as @xmath247 for @xmath248 , where @xmath249 is the probability that a small empty nucleus can be expanded until size @xmath203 . in the emptying procedure described in section [ ergo ] we subsequently increase from @xmath201 to @xmath250 the size of the empty nucleus .
if we require instead that on a region of size @xmath251 around each corner there is not a spanning dp cluster , we can expand directly to @xmath203 .
the probability of the latter event gives @xmath252 ( see @xcite ) .
therefore , we get the desired upper bound on @xmath253 that has the same scaling as the lower bound . in the following we present numerical results on the percolation of blocked structures that support our theoretical findings . starting from an initial configuration
we sequentially pick away all particles that are not blocked , so that the final configuration contains the backbones of forever blocked particles .
averaged over @xmath254 initial configurations ( from left to right respectively ) .
inset : histograms of the number of initial configurations ( y axis ) leading to a fraction of sites @xmath255 ( x - axis ) for @xmath256 , @xmath254 initial configurations and initial density @xmath257 ( when increasing the initial density the right peak moves to the right).,title="fig : " ] in fig .
[ pb ] we show the probability that a site belongs to the infinite spanning cluster for lattice sizes @xmath258 averaged over @xmath254 samples .
though the transition takes place at the critical density of dp @xmath259 , as we have proved , finite size effects appear already at @xmath260 .
indeed , the probability that there _ exists _ a frozen cluster is substantial for @xmath9 fifteen percent below @xmath1 even in our largest systems , ( @xmath261 ) : it is thus hard to study the asymptotic critical behavior ( see @xcite for an analogous problem in the context of bootstrap percolation ) .
but in a slightly different model one can get closer to the transition @xcite : these data are consistent with the predicted @xmath262 with @xmath263 , but the small range of @xmath264 available makes the uncertainties in @xmath265 large .
another theoretical finding confirmed by numerics is the first order character of the transition . indeed
the histograms of the density of blocked structures , @xmath255 , clearly show a well defined two peak shape as in usual first order transitions ( see inset of fig .
the peak at a non zero density is clearly distinct from the one at zero density and moves to the right by increasing the density of particles in the initial configuration , thus showing that asymptotically the transition is first order .
furthermore there are no finite size effects on the peak position but only on its weight .
thus , the continuous curves @xmath266 in fig .
[ pb ] are the average of two curves : the trivial one , @xmath267 , corresponding to the left peak and the dotted one corresponding to the right peak . by changing the system
size the weights of the two peaks change and so the continuous averaged @xmath268 shifts to the right . finally , let s focus on the predictions for the dynamics .
first , as already discussed , an ergodicity breaking transition occurs at @xmath269 .
furthermore , the first - order character of the percolation of blocked structures implies a discontinuous jump of @xmath270 .
this is the plateau of the correlation function , see fig.[corr ] and is the analog of the edwards - anderson parameter in spin - glasses .
[ ] [ ] @xmath271 [ ] [ ] @xmath64 in fact for @xmath59 , ergodicity implies that @xmath272 since the system always relaxes to equilibrium . instead for @xmath273 , by separating the contribution from sites which are occupied and blocked forever and the remaining sites , one can show that @xmath274 @xcite , where @xmath275 is a strictly positive for all @xmath57 and @xmath216 , as defined previously , is the probability the the origin is frozen .
therefore , @xmath276 , for @xmath277 .
furthermore the relaxation timescale , @xmath65 , diverges also very fast at the transition as discussed previously . as a check
, we performed numerical simulations of standard montecarlo dynamics .
the results for @xmath278 are plotted in fig .
[ corr ] for system size @xmath279 .
the curves clearly shows a developing ( discontinuous ) plateau which becomes infinite after a certain ( size dependent ) density .
however , the initial curves until @xmath280 have no finite size effects .
this shows that the developing of a discontinuous plateau and the increasing of the timescale start far from the critical density , in agreement with an essential singularity for the laws for @xmath281 ( for a power law divergence @xmath65 should diverge much closer to @xmath2 ) .
note that , because of the very severe finite size effects , one needs much larger system sizes ( and much larger timescales ) in order to measure successfully the exponent of the essential singularity .
we leave that for future study .
as discussed in this work cooperative kcms display a remarkable physical behavior . in particular , knights model undergoes a purely dynamical phase transition due to a jamming percolation : a giant blocked cluster appears at @xmath58 .
this new type of transition has features that are very different from usual ( first and second order ) phase transitions and that are similar to the ones indeed expected for glass - jamming transitions .
in particular the fraction of jammed particles is discontinuous ( as for first order phase transition ) although time and lengthscales diverge ( as for second order phase transitions ) .
furthermore the relaxation time diverges with a vogel - fulcher like form , i.e. much faster than a power law .
the extension and the universality of our results are fundamental open questions . in three dimensions , two natural generalizations of our jamming percolation
exist : one composed of dp clusters which should slow down as a double exponential of @xmath282 and the other of directed sheet - like structures which will have exponential slowing down like we have found in 2d .
the key ingredients are kinetic constraints that enable huge jammed clusters to form out of small objects without these becoming much more common or much larger .
a very important issue is whether the mechanism that we devise to create a jamming percolation , which is based on interacting dp clusters , is the only possible one or there are others that may lead to a very similar dynamical transition , see @xcite . for the future , the connection between our results and the jamming transition found for continuum particle systems @xcite needs exploring .
one should analyze the effects of constraint - violating processes occurring with a very low rate : these are certainly present for molecular liquids undergoing a glass transition .
similarly , it would be very important to generalize our results to systems of particles in the continuum .
finally , it would be very interesting to compare the geometrical and statistical properties of jamming percolation clusters to the ones that can be measured in experiments in colloidal and granular systems .
+ * acknowledgments * + many of the results that we presented were obtained with d.s.fisher .
we are very happy to thank him for the long - standing collaboration on glass - jamming transitions .
we thank also m. sellitto for having investigated with us the jamming transition on the bethe lattice summarized in section v. we are grateful to l. berthier for helpful discussions and collaboration on related subjects .
gb is partially supported by the european community s human potential program contracts hprn - ct-2002 - 00307 ( dyglagemem ) . 99 p.g.de benedetti and f.h.stillinger _ nature _ * 410 * , 267 ( 2001 ) .
er.weeks et al .
_ science _ * 287 * , 627 ( 2000 ) ; v.trappe et al .
_ nature _ * 411 * , ( 2001 ) 722 .
g.marty , o.dauchot _ phys.rev.lett _ * 94 * 015701 ( 2005 ) m.a.ediger _ ann.rev.phys.chem . _ * 51 * , 99 ( 2000 ) ; e.vidal-russel , n.e.israeloff _ nature _ * 408 * , 695 ( 2000 ) ; l.a.deschenes , d.a.vande bout _ science _ * 292 * 255 ( 2001 ) ; l.berthier et al .
_ science _ * 310 * 1797 ( 2005 ) ; c.bennemann et al . _
nature _ , * 399 * , 246 ( 1999);o.dauchot , g.marty , g.biroli , _
phys.rev.lett _ * 95 * , 265701 ( 2005 ) .
t.r.kirkpatrick , d.thirumalai , _
phys.rev.lett _ * 58 * , ( 1987 ) 2091 ; t.r.kirkpatrick , d.thirumalai , p.g.wolynes _ phys.rev.a _ * 40 * , ( 1989 ) 1045 .
bouchaud j - p , cugliandolo l f , kurchan j and mzard m , in _ spin - glasses and random fields _ , edited by young a p ( world scientific , 1997 ) .
x. xia and p.g .
wolynes , pnas * 97 * 2990 ( 2000 ) and refs therein . j .-
bouchaud and g. biroli , j. chem . phys . *
121 * , 7347 ( 2004 ) .
s.franz _ europhys.lett . _ * 73 * 492 ( 2006 ) m. dzero , j. schmalian , p. g. wolynes , cond - mat/0502011 .
m.a.moore cond - mat/060241 j. jackle j. phys .
matter * 14 * , ( 2002 ) 1423 ; f. ritort , p. sollich _ adv . in phys .
_ * 52 * ( 2003 ) , 219 .
g.h.fredrickson , h.c.andersen _ phys.rev.lett . _ * 53 * , 1244 ( 1984 ) ; _ j.chem.phys . _
* 84 * , 5822 ( 1985 ) c.toninelli,g.biroli , d.s.fisher_phys.rev.lett_ * 96 * , 035702 ( 2006 ) .
j.jackle , s.eisinger _ z.phys.b : cond.mat._ * 84 * , 115 ( 1991 ) w. kob and h.c .
andersen , _
phys.rev.e _ * 48 * ( 1993 ) 4364 .
p.sollich , m.evans _
phys.rev.lett _ * 83 * 3238 ( 1999 ) d.aldous ,
p.diaconis _ j.stat.phys .
_ * 107 * 845 ( 2002 ) j.p.garrahan , d.chandler _ phys . rev.lett_
* 89 * 035704 ( 2002 ) s.whitelam , l.berthier , j.p.garrahan , _ phys . rev.lett_ * 92 * , 185705 ( 2004 ) ; _ phys.rev.e _ * 71 * , 026128 ( 2005 ) r.jack , p.mayer , p.sollich cond - mat/0601529 n.cancrini , f.martinelli , c.roberto , c.toninelli cond - mat/0603745 l.berthier , j.p.garrahan _ j.chem.phys . _ * 119 * , 4367 ( 2003 ) j.p .
garrahan , d. chandler , proc .
* 100 * , 9710 ( 2003 ) .
j.chem.phys _ * 95 * , 544 ( 1991 ) g.h.fredrickson , s.a.brawer _ j.chem.phys _ * 84 * 3351 ( 1986 ) m.einax , m.schulz _ j.chem.phys _ * 115 * 2282 ( 2001 ) i.s.graham , l.pich , m.grant ,
_ j.phys.cond.matt_ , * 5 * , 6491 ( 1993 ) ; _
phys.rev.e_ , * 55 * , 2132 ( 1997 ) p.harrowell , _
phys.rev.e _ * 48 * , 4359 ( 1993 ) g.h.fredrickson _ ann.n.y.acad.sci_ , * 484 * , 185 ( 1986 ) s.butler , p.harrowell , _ j.chem.phys . _ * 95 * , 4466 ( 1991 ) c. toninelli , g. biroli and d.s .
fisher in preparation .
l. berthier , g. biroli , c. toninelli in preparation .
j.reiter , f.mauch , j.jackle , _ physica a _ , * 184 * , 493 ( 1992 ) c.toninelli,g.biroli , d.s.fisher _ phys.rev.lett . _
* 92 * 185504 ( 2004 ) ; _ j.stat.phys .
_ * 120 * , 167 ( 2005 ) c.toninelli , g.biroli cond - mat/0512335 w. kob , in `` slow relaxations and nonequilibrium dynamics in condensed matter '' , vol .
session lxxvii of les houches summer school ( ed .
barrat , m. feigelman and j. kurchan ) , published by edp sciences and springer .
e.r.weeks , d.a.weitz _ phys.rev.lett .
_ * 89 * 095704 ( 2002 ) ; _ chem.physics _ * 284 * , 361 ( 2002 ) l.bertini , c.toninelli _ j.stat.phys._ , * 117 * , 549 - 580 ( 2004 ) c.toninelli , g.biroli _ j.stat.phys._ , * 117 * , 27 - 54 ( 2004 ) h.spohn , _ large scale dynamics of interacting particles _ berlin springer ( 1991 ) m. aizenmann , j.l .
lebowitz , _
j.phys.a _ * 21 * ( 1988 ) 3801 .
pittel b , spencer j and wormald n ,
_ j. comb .
b _ * 67 * ( 1996 ) 111 .
r.h.schonmann , _ ann . of probab._,*20 * ( 1992 ) , 174 - 193 c. kipnis , c. landim , _ scaling limits of interacting particle systems _ , grundlheren der mathematischen wissenschaften 320 , springer - verlag , berlin , new york , 1999 .
c.landim , c.toninelli in preparation h.spohn _ j.stat.phys .
_ * 59 * ( 19990 ) 1227 ; _ physica a _ * 163 * , ( 1990 ) 134 .
j.kurchan , l.peliti , m.sellitto _ europhysics lett . _
* 39 * ( 1997 ) 365 m. sellitto , g. biroli and c. toninelli .
* 69 * ( 4 ) , 496 ( 2005 ) .
chalupa j , leath p l and reich r , _ j. phys .
c : solid state phys . _
* 12 * ( 1979 ) l31 .
s. n. dorogovtsev , a. v. goltsev , and j. f. f. mendes phys .
lett . * 96 * , 040601 ( 2006 ) .
schwarz , a.j .
liu , l.q .
chayes , europhys . lett .
* 73 * , 560 ( 2006 ) .
abdelaziz amraoui , andrea montanari , tom richardson , rudiger urbanke , 42nd allerton conference on communication , control and computing , cs.it/0410019 .
a. dembo and a. montanari in preparation .
adv . in phys
_ * 49 * 815 ( 2000 ) franz s and parisi g , _ j. phys . :
cond . matter _
* 12 * ( 2000 ) 6335 p.de gregorio , a.lawlor , p.bradley , k.a.dawson phys.rev.lett .
* 93 * ( 2004 ) 025501 . | we present a detailed physical analysis of the dynamical glass - jamming transition which occurs for the so called knight models recently introduced and analyzed in a joint work with d.s.fisher@xcite .
furthermore , we review some of our previous works on kinetically constrained models .
the knights models correspond to a new class of kinetically constrained models which provide the first example of finite dimensional models with an _ ideal glass - jamming transition_. this is due to the underlying percolation transition of particles which are mutually blocked by the constraints .
this _ jamming percolation _ has unconventional features : it is discontinuous ( i.e. the percolating cluster is compact at the transition ) and the typical size of the clusters diverges faster than any power law when @xmath0 .
these properties give rise for knight models to an ergodicity breaking transition at @xmath1 : at and above @xmath2 a finite fraction of the system is frozen . in turn , this finite jump in the density of frozen sites leads to a two step relaxation for dynamic correlations in the unjammed phase , analogous to that of glass forming liquids .
also , due to the faster than power law divergence of the dynamical correlation length , relaxation times diverge in a way similar to the vogel - fulcher law .
* keywords : * kinetically constrained lattice gases , glassy dynamics , bootstrap percolation , jamming transition . |
the crab nebula @xmath0-ray emission has been established as a standard candle for the @xmath0-ray astronomy ( e.g. meyer et al .
this conclusion based mainly on the observations at energy ranges between @xmath1 mev and a few gev ( satellite telescopes ) and between @xmath1 gev and @xmath2 tev ( cherenkov telescopes ) .
the theoretical interpretation of the emission from the crab nebula suggests that this may not be exactly the case . the @xmath0-ray spectrum is widely interpreted in the two component radiation model in which the lower energy emission is due to the synchrotron process and the higher energy emission results as a consequence of the inverse compton ( ic ) process .
it has been argued that the ends of the synchrotron and ic components can flicker due to the non - stationary acceleration of leptons at the pulsar wind shock .
first evidences of the variability of the end of synchrotron component has been reported based on the analysis of the egret data ( de jager et al .
1996 , ramanamurthy et al .
1995 ) . also situation has not been clear at the highest observer energies ( above @xmath2 tev ) , where some measurements seemed to be contradictory ( aharonian et al .
2004 , aharonian et al .
2006 ) .
the spectrum from the crab nebula at energies 100 - 400 mev is very steep , with the differential spectral index close to 4 ( abdo et al .
it cuts off at about @xmath1 mev , i.e. at clearly larger energies than previously reported @xmath3 mev cut - off derived from the egret measurements ( kuiper et al .
2001 ) . on the other hand ,
the @xmath0-ray spectrum at gev energies is very flat linking correctly with the tev spectrum measured by the cherenkov telescopes ( e.g. albert et al .
unexpectedly , the agile telescope has recently reported an enhanced @xmath0-ray emission above @xmath4 mev from the crab nebula by a factor of 2 - 3 during september 19 - 21 , 2010 in respect to steady emission ( tavani et al .
2010 , 2011 ) .
this observation has been confirmed by the fermi - lat telescope which observed enhanced @xmath0-ray emission during the interval september 18 - 22 , 2010 ( buehler et al .
2010 , abdo et al .
the @xmath0-ray flux reached the value of @xmath5 ph .
s@xmath7 above 100 mev .
the flaring component has a differential spectral index @xmath8 and is coincident with the crab nebula .
the flare consists from three separated enhancements lasting 6 - 12 hours ( balbo et al .
the lack of pulsed emission suggests that the flare is probably related to the crab nebula rather than the pulsar ( hays et al .
2010 ) .
the enhanced @xmath0-ray emission at about 1 tev has been also reported by the argo - ybj air shower array ( aielli et al .
this flaring component extends up to september 27th 2010 , i.e. it lasts a few days longer than the 100 mev emission reported by the satellite telescopes .
the flux measured during the time interval september 17 - 22nd , 2010 was about 3 - 4 time higher than usual .
unfortunately , this enhanced emission has not been confirmed by the observations of the magic ( mariotti et al .
2010 ) and veritas collaborations ( ong et al .
2010 ) . the increased optical emission about 3 arcsec east of the crab pulsar and from the wisp north - west of the pulsar has been reported by the hubble space telescope ( caraveo et al .
this corresponds to the brightening of this same region in the x - rays as reported by the chandra observations ( tennant et al .
however , other x - ray telescopes have not reported any significant change in the crab nebula spectrum and morphology ( sakamoto et al .
2010 , shaposhnikov et al .
moreover , no any glitch of the crab pulsar has been noted during the last 60 days before the @xmath0-ray flare ( espinoza et al .
2010 ) . in this paper
we investigate the hypothesis that the variable @xmath0-ray emission above @xmath4 mev is due to the synchrotron radiation from electrons which are accelerated near the pulsar wind shock region to different maximum energies .
we expect that variable gev @xmath0-ray emission should be accompanied by the variable @xmath0-ray emission above a few tev from the inverse compton process .
discovery of such variable emission by the present cherenkov telescopes ( hess , magic , veritas ) and the future cherenkov telescope array ( cta ) will allow to constrain the physical processes at the acceleration site and investigate the connections between the pulsar and the nebula .
note that recently komissarov & lyutikov ( 2010 ) proposed the localization of the @xmath0-ray flare emission region with the compact inner knot , based on their recent results on relativistic mhd simulations of the crab nebula .
the variability time scale of the @xmath0-ray emission on the level of days can be understood assuming that the emission comes only from a part of the pulsar wind shock .
therefore , either only a small region of the axially symmetric pulsar wind shock is excited or the emission region is moving relativistically towards the observer .
the axial symmetry of the pulsar wind ( with the axis parallel to the rotational axis of the pulsar ) is consistent with the observations of the jet - torus morphology of the crab nebula ( e.g. hester 2008 ) and with the mhd simulations of the pulsar wind structure ( komissarov & lyubarsky 2003 ) .
the first hypothesis seems unlikely since the wisp regions have already extensions which are larger than the light travel distance scale corresponding to 3 - 4 day time scale variability .
here we consider the second hypothesis in which the @xmath0-ray emission region is related to the shock region which still moves with substantial lorentz factor , @xmath9 , in the outward direction from the pulsar .
this might be the region in the pulsar wind which is in the process of being decelerated at the pulsar wind shock region .
it is expected that in this region efficient reconnection process of the pulsar wind magnetic field can occur . as a result
, good conditions are produced for acceleration of particles to the highest possible energies .
the first wisps in the crab nebula appear at the distance of @xmath10 cm from the pulsar ( caraveo et al .
in fact , recent optical observations of the wisp confirm its variability which may be related to the observed @xmath0-ray variability by the agile and fermi .
the observed day time scale @xmath0-ray variability allows to constrain @xmath9 .
we assume that emission extends along the axially symmetric shock structure as observed in the crab nebula and expected from the mhd simulations ( komissarov & lyubarsky 2003 ) . the turbulence which arrive to the shock is relatively thin , i.e. it is thinner than the light crossing time scale multiplied by the lorentz factor of the emission region .
the distant observer is able to detect emission only from a part of the shock with the opening angle @xmath11 ( see fig . 1 ) .
based on such scenario we can estimate @xmath12 on , @xmath13 produced radiation is collimated within the angle @xmath11 for the lorentz factor of the shock of the order of , @xmath14 provided that the variability time scale of the @xmath0-ray emission is @xmath15 . in the case of the crab nebula shock with dimension @xmath16 cm ,
the minimum lorentz factor of the emission region should be of the order of @xmath17 where @xmath18 days .
we conclude that for the sub - day time scale variability ( as observed in @xmath0-rays by the fermi - lat telescope , balbo et al .
2010 ) , the lorentz factor of the emission region in the shock has to be at least mildly relativistic .
we suggest that the variable @xmath0-ray emission observed from the crab nebula comes from a part of the pulsar wind which is just behind the shock and still moves with relativistic velocities .
5.2truecm we constrain the physical parameters of the emission region in the fluid frame ( in which the electric field vanishes ) based on the observations .
this reference frame moves towards the observer with the lorentz factor @xmath19 .
we consider relativistic electrons with characteristic lorentz factors @xmath20 in the fluid reference frame which are immersed in the perpendicular magnetic field with the strength in the fluid frame @xmath21 .
the synchrotron spectrum from the crab nebula extends up to a hundred mev , showing a break at , @xmath22100 mev , during the quiescent stage ( abdo et al .
this break can be related to a break in the electron spectrum according to , @xmath23 where the critical magnetic field is @xmath24 g , and @xmath25 is the electron rest mass . applying the above value for the break in the synchrotron spectrum , @xmath26
, we get the constraint on the product of the magnetic field strength within the emission region and the lorentz factor of radiating electrons , @xmath27 on the other hand , the @xmath0-ray flare observed by the agile and the fermi telescopes lasted for a few days and show a clear variability on a sub - day time scale , @xmath28 ( balbo et al .
the duration of the flare has to be comparable ( or longer ) to the cooling time scale of electrons on the synchrotron process in the fluid frame , @xmath29 . by comparing these two time scales ,
we get the lower limit on , @xmath30 these two conditions allows us to estimate the magnetic field strength at the acceleration region in the fluid frame on , @xmath31 the magnetic field strength in the observer s frame is , @xmath32 . for the variability time scale observed by the fermi ( of the order of 0.5 day , balbo et al .
2010 ) , we obtain the lower limit on @xmath33 g , assuming that @xmath34 .
the magnetic field as seen from the observer reference frame is estimated on @xmath35 g. on the other hand , a simple extrapolation of the magnetic field from the crab pulsar surface up to the location of the first optical wisps in the crab nebula allows to estimate their magnetic field strength .
we assume a dipole magnetic field structure in the pulsar inner magnetosphere below the light cylinder radius , @xmath36 , and a toroidal structure in the pulsar wind , @xmath37 , where the surface magnetic field of the crab pulsar @xmath38 g , the light cylinder radius @xmath39 cm , @xmath40 g is the magnetic field strength at the light cylinder , the crab pulsar rotational period is @xmath41 ms , and @xmath42 is the velocity of light .
the distance of the optical wisp from the pulsar is @xmath43 cm , estimated from the angular distance of 3 arcsec ( caraveo et al .
2010 ) , and for the distance to the crab nebula of 1650 pc .
this wisp has been recently reported as showing optical variability .
then , the magnetic field strength at the distance of the wisp is @xmath44 g. this estimate of the magnetic field fits well to the above estimate based on the variability time scale of the @xmath0-ray emission . in any way , this lower limit is significantly larger than the magnetic field strength within the whole volume of the crab nebula estimated on @xmath45 g. such low values of the magnetic field has been already predicted based on the spherically symmetric mhd model developed by kennel & coroniti ( 1984 ) .
they have been successfully applied in modelling of the multiwavelength non - thermal emission from the crab nebula in terms of the popular synchrotron self - compton model ( see e.g. , de jager & harding 1992 , atoyan & aharonian 1995 , or recently meyer , horns & zechlin 2010 ) .
the discrepancy between the magnetic field values estimated for the inner wisp and the whole crab nebula can be naturally understood in terms of the recent axially symmetric mhd model discussed by komissarov & lyutikov ( 2010 ) . in this model
the observed features in the inner part of the crab nebula ( i.e. the inner knot and possibly also the wisp regions ) are created by parts of the pulsar wind shock which lays significantly closer to the pulsar than the equatorial part of the pulsar wind shock .
therefore , the magnetic field strength at the considered by us the wisp region can be much stronger in respect to the whole nebula .
the wisp region is quite extended but for the @xmath0-ray flaring might be responsible only a part of this region which is more directed towards the observer ( see fig 1 ) . the possible variability of the pulsar wind termination shock ( discovered in mhd simulations by camus et al . 2009 ) can be responsible for the change of conditions in the acceleration region directed towards the observer at the earth .
in general , the variable emission at the gev energies observed from the crab nebula might be due to the change in the magnetic field strength at the acceleration site or the change of the plasma conditions which determines the acceleration process .
its influence on the acceleration of particles is usually described by the so called acceleration parameter .
we limit our considerations only to the second case since it is difficult to imagine situation in which the magnetic field , which source lays in the pulsar , can change significantly on a time scale of a few days in the pulsar wind .
the problem is how electrons can be accelerated to considered energies in the classical acceleration processes in such a relatively strong magnetic field . in the case of the shock acceleration scenario or acceleration in a turbulent region ,
the maximum lorentz factors of electrons are limited by the synchrotron energy losses to @xmath46 , where @xmath47 is the acceleration coefficient , and the magnetic field strength at the acceleration site is defined in the shock reference frame .
it is scaled with @xmath48 g. this limit has been obtained from the comparison of the energy gains of electrons from the shock acceleration mechanism , @xmath49 , with their synchrotron energy losses , @xmath50 , where @xmath51 is the energy of electrons , @xmath52 is the larmor radius of electrons , and @xmath53 is the thomson cross section .
the maximum energies of synchrotron photons produced by electrons with the lorentz factors , @xmath54 , are independent on the magnetic field strength .
they can be estimated from @xmath55 mev ( see eq .
note that the maximum energies of synchrotron photons produced by electrons accelerated in the shock scenario can not exceed @xmath1 mev since @xmath56 .
this is also true in the case of relativistic shocks if considered in the shock reference frame ( kirk & reville 2010 ) .
however , in the case of relativistic shocks these maximum energies observed in the rest frame can be enlarged by the doppler factor of the shock as shown in the analytical calculations by achterberg et al .
therefore , detection of synchrotron gev emission from the crab nebula requires that the emission region moves towards the observer with the doppler factor of the order of @xmath5710 , consistent with the lower limit estimated in sect . 2 .
on the other hand , monte carlo simulations of the acceleration of particles by the first order fermi shocks propagating in a realistically modelled turbulent magnetic fields show that the process may not be efficient and the maximum energies of accelerated particles can be lower than expected in previous modellings ( niemiec & ostrowski 2004 , niemiec , ostrowski & pohl 2006 ) .
another possibility , originally considered by coroniti ( 1990 ) and michel ( 1994 ) , is the acceleration of electrons in the reconnection regions of the magnetic field in the pulsar wind before it reaches the shock region ( e.g. lyubarsky 1996 , lyubarsky & kirk 2001 , kirk & skjaeraasen 2003 ) . in the case of acceleration in the reconnection regions ( along the magnetic field lines ) ,
electron energy does not need to be limited by the synchrotron energy losses ( kirk 2004 ) .
electrons are injected into the magnetic field of the pulsar wind zone producing synchrotron @xmath0-rays with energies overcoming the synchrotron energy loss limit on the shock acceleration process mentioned above . due to the observed shape of the spectrum from the crab nebula , we assume that electrons reach the equilibrium spectrum which can be described by a differential power law spectrum with the characteristic cut - off at @xmath58 , @xmath59 , where @xmath60 is the spectral index considered in the range 3.0 - 3.6 in order to be consistent with the observed synchrotron spectrum at energies below the break .
electrons with such equilibrium spectrum produce synchrotron photons up to gev energies and also tev @xmath0-rays by scattering variety of soft radiation fields inside or close to their acceleration site . between them
the best defined is the microwave background radiation ( mbr ) .
the level of the soft synchrotron emission produced by low energy electrons fulfilling the whole crab nebula can not be so well defined in the region of the acceleration of electrons close to the pulsar wind shock .
the reported variability of a hundred mev - gev @xmath0-ray emission from the crab nebula seems to be naturally interpreted as a result of the non - stationary acceleration of the end of the electron spectrum in the region when the pulsar wind interacts with the nebula .
we propose that such emission variable on a short time scale likely comes from the region just behind the pulsar wind shock which still moves mildly relativistically .
these electrons should produce synchrotron emission which sometimes can extend up to gev energies . on the other hand , these same electrons can produce also variable multi - tev @xmath0-rays as a result of the inverse compton scattering of the mbr and very low energy ( in radio range ) synchrotron radiation .
12.5truecm we calculate the example spectra from the synchrotron process for different parameters of the electron spectra which potentially could be responsible for the observed variability of @xmath0-ray emission reported by the agile and the fermi telescopes . for these same parameters
we also calculate the ic @xmath0-ray spectra in order to show how the end of ic spectrum should vary at energies above 1 tev . in the calculations we apply the full formulae from blumenthal &
gould ( 1970 ) .
only the well defined soft photon target have been taken into account when calculating the ic spectra , i.e. the mbr and synchrotron radiation from the nebula . in the case of the synchrotron radio emission
, we applied the observations of the crab nebula in the ghz energies ( baars & hartsuijker 1972 ) .
the differential photon spectrum of this emission is well described by a single power law with spectral index 1.26 .
this synchrotron differential photon density has been obtained by simple averaging over the whole volume of the nebula with the radius of 2 pc .
we do not consider other possible low energy radiation fields such as the infrared radiation or the microwave radiation ( see .
e.g. atoyan & aharonian 1996 ) .
the photon densities of these other radiation fields is much more difficult to define preciously in the region of the pulsar wind shock .
moreover their contribution to the total ic @xmath0-ray spectrum at multi tev region seems to be less important in respect to the comptonization of the mbr and the radio synchrotron photons ( e.g. atoyan & aharonian 1996 ) . at first
, we model the steady synchrotron spectrum and the flare synchrotron spectrum by changing the maximum energies of accelerated electrons and keeping constant the magnetic field strength at the acceleration region which is fixed on @xmath61 g. for this magnetic field strength electrons have to have lorentz factors @xmath62 , in order to be consistent with the observations of the cut - off in the synchrotron spectrum during the quiescent state ( see fig .
we also calculate the @xmath0-ray spectra expected from electrons accelerated to maximum energies which might correspond to the emission during the recently observed flare , applying @xmath63 , and the supposed super - quiescent stage for which @xmath64 ( fig .
electrons with the lorentz factors corresponding to the quiescent stage scatter the mbr almost in the klein - nishina ( kn ) regime since the energy of the mbr from the peak of the planck spectrum is equal to @xmath65 . on the other hand , scattering of synchrotron radio emission
is regulated to the border between t - kn regimes , due to the power law spectrum of the synchrotron photons with the spectral index @xmath66 .
the synchrotron and ic spectra are calculated for the two spectral indexes of electron spectrum , 3.0 and 3.6 ( see fig .
2a , b , respectively ) .
as expected the @xmath0-ray emission produced by these electrons in the gev and tev energy ranges show clear correlations . however , the end of the tev @xmath0-ray spectrum from scattering of mbr and synchrotron radio emission vary on a lower level than the end of the synchrotron spectrum due to the ic scattering in ( or close to ) the klein - nishina regime .
the change in the tev @xmath0-ray spectrum between the quiescent and flare states is substantial at energies above @xmath2 tev .
moreover , the ic component , due to the scattering of the radio emission , is steeper and start to dominate at lower energies than the ic component from scattering of the mbr .
the calculation of the level of the ic component from scattering of radio emission is less reliable due to the uncertain determination of the photon density of the radio synchrotron photons ( produced in the outer nebula ) in the region close to the pulsar shock . in fig .
2c , we also show the @xmath0-ray spectra in the case of a very strong magnetic field at the emission region , equal to @xmath67 g , and a few values of the lorentz factor of electrons at the break of their spectrum .
such strong magnetic field might appear as a result of the compression due to the decelerating pulsar wind .
the lorentz factors of electrons , producing radiation in the quiescent stage , should be of the order of @xmath68 .
then , the energies of the mbr photons at the peak of the spectrum are @xmath69 , i.e. the ics of the mbr occurs still in the t regime .
therefore , the cut - offs in the ic spectrum calculated for the electron lorentz factors corresponding to the quiescent and flaring stages clearly show stronger variability .
we propose that the variable emission from the crab nebula on a time scale of a few days , recently observed by the agile and the fermi telescopes above 100 mev , can be understood assuming that the emission region is moving mildly relativistically to the observer .
we suggest that this @xmath0-ray emission comes from the region just behind the shock in the pulsar wind .
it could be produced by electrons accelerated to different maximum energies in the electric fields induced during the reconnection process of the magnetic field under the pressure exerted by the shocked wind .
these highest energy electrons produce synchrotron radiation in the mev - gev energy range and also the ic @xmath0-rays above @xmath70 tev by scattering of the mbr and the low energy synchrotron radiation .
the relative role of these two soft radiation fields in the ics process is difficult to estimate due to the lack of precise knowledge on the density of the ghz radio photons at the acceleration site in the pulsar wind .
we consider different parameters describing the spectrum of injected electrons and compare them with the observations in the mev - gev and above 1 tev energy ranges .
the flaring stage might result either due to the change of the maximum energies of accelerated electrons ( described by their lorentz factors at the break of the power law spectrum ) or the the change of the spectral index of electrons at the highest energies , or by both effects .
the change of only the break energy in the electron spectrum results in the tev @xmath0-ray ic spectra which show rather low level of variability at energies above @xmath2 tev .
therefore , in such a case it will be difficult to observe synchronous variations at the gev and tev energies even with the future cherenkov telescope array ( see fig .
the situation is more promising in the case of the instantaneous change of the break energy in the electron spectrum with the flattening of the electron spectrum .
in such a case , the clear variability of the ic spectrum through the broad range of the tev energies could be detected .
the possible discovery of flaring emission in the tev energies will be easier in the case of significantly stronger magnetic field at the acceleration region .
then , the maximum energies of accelerated electrons can be lower and the ic scattering of the mbr could still occur in the thomson regime . as a result ,
the end of ic spectrum should change quadratically with the change of energy of electrons .
the tev @xmath0-ray spectra calculated for the range of discussed parameters are compared with the best measurements of the crab nebula spectrum provided by the hegra ( aharonian et al .
2004 ) and the hess ( aharonian et al . 2006 ) .
the hegra spectrum extends up to @xmath71 tev showing no evidences of the cut - off . on the other hand ,
the hess spectrum is steeper and shows a cut - off at clearly lower energies .
calculated by us tev spectra from ic scattering of the mbr and synchrotron radio photons in the quiescent and flaring stages shows better consistency with the spectrum measured by the hegra collaboration ( aharonian et al .
2004 ) . the relatively steep tev @xmath0-ray spectrum reported by the hess collaboration ( aharonian et al .
2006 ) might be consistent with our calculations provided that it has been obtained during the stage of a relatively non - efficient acceleration of electrons and relatively strong magnetic field strength in the emission region .
the sub - gev synchrotron @xmath0-ray emission during such stage should be on the level below the @xmath0-ray flux reported by the recent measurements by the fermi - lat telescope ( abdo et al .
note however , that the @xmath0-ray emission reported by the egret telescope a decade ago ( e.g. kuiper et al .
2001 ) seems to be on the lower level than the present fermi - lat measurements .
this suggest that the sub - gev emission from the crab nebula can vary not only up but also down in respect to the so called quiescent level observed by the fermi .
the possible variability of the end of the tev @xmath0-ray emission from the crab nebula produced in the ics process by electrons may be partially hidden in the case of efficient acceleration of hadrons within the nebula .
these hadrons , coming from the surface of the crab pulsar , could be also accelerated in the scenario discussed here .
they interact with the matter within the crab nebula producing steady emission which is expected above @xmath2 tev , see e.g. amato et al .
( 2003 ) , bednarek & bartosik ( 2003 ) , horns et al .
we would like to thank maria giller for discussion .
this work is supported by the polish mniszw grant n n203 390834 and ncbir grant era - net - aspera/01/10 .
abdo , a.a .
2010 apj 708 , 1254 abdo , a.a .
et al . 2011
science , doi:10.1126/ science.1199705 ( arxiv:1011.3855 ) achterberg , a. , gallant , y.a . , kirk , j.g . ,
guthmann , a.w .
2001 mnras 328 , 393 aharonian , f.a . ,
atoyan , a.m. 1995 aph 3 , 275 aharonian , f.a .
2004 apj 614 , 897 aharonian , f.a .
2006 a&a 457 , 899 aielli , g. et al .
2010 atel 2921 albert , j. et al . 2008 apj 674 , 1037 amato , e. , guetta , d. , blasi , p. 2003
a&a 402 , 827 atoyan , a.m. , aharonian , f.a .
1996 mnras 278 , 525 baars , j.w .. m . , hartsuijker , a.p .
1972 a&a 17 , 172 balbo , m. , walter , r. , ferrigno , c. , bordas , p. 2010 , a&a in press ( arxiv:1012.3397 ) bednarek , w. , bartosik , m. 2003 a&a 405 , 689 blumenthal , g.r . , gould , r.j .
1970 rev.mod.phys .
42 , 237 buehler , r. et al .
2010 atel 2861 camus , n.f . ,
komissarov , s.s . ,
bucciantini , n. , hughes , p.a . 2009 mnras 400 1241 caraveo , p. et al .
2010 atel 2903 de jager , o.c . ,
harding , a.k . 1992 apj 396 , 161 de jager , o.c .
et al . 1996 apj 457 , 253 espinoza , c.m .
2010 atel 2889 hays , e. et al .
2010 atel 2879 hester , j.j .
2008 araa 46 , 127 horns , d. , aharonian , a.f . ,
santangelo , a. , hoffman , a.i.d . ,
masterson , c. 2006 a&a 451 , 51 kennel , c.f .
, coroniti f.v .
1984 apj 283 , 694 kirk , j.g . , skjaeraasen , o. 2003 apj 591 , 366 kirk , j.g .
2004 prl 92 , 181101 kirk , j.g . ,
reville , b. 2010 apj 710 , l16 komissarov , s.s . , lyubarsky , y.e .
2003 mnras 344 , l93 komissarov , s.s . ,
lyutikov , m. 2010 mnras , submitted ( arxiv:1011.1800 ) kuiper , l. et al . 2001 a&a 378 , 918 lyubarsky , y. 1996 a&a 311 , 172 mariotti , m. for the magic collab .
2010 atel 2967 meyer , m. , horns , d. , zechlin , h .- s . 2010 a&a 523 , 2 niemiec , j. , ostrowski , m. 2004 apj 610 , 851 niemiec , j. , ostrowski , m. , pohl , m. 2006 apj 650 , 1020 ong , r.a . for the veritas collab .
2010 atel 2968 ramanamurthy , p.v .
1995 apj 450 , 791 sakamoto , t. et al .
2010 atel 2893 shaposhnikov , n. et al .
2010 atel 2872 tanimori , t. et al .
1998 apj 492 , l33 tavani , m. et al .
2010 atel 2855 tavani , m. et al .
2011 science , doi:10.1126/ science.1200083 ( arxiv:1101.2311 ) the cta consortium , 2010 arxiv:1008.3703 tennant , a. et al .
2010 atel 2882 | recently the agile @xmath0-ray telescope has reported the enhanced @xmath0-ray emission above 100 mev from the direction of the crab nebula during a period of a few days .
this intriguing observation has been confirmed by the fermi - lat telescope .
this emission does not show evidences of pulsations with the crab pulsar .
it seems that it originates at the shock region created as a result of the interaction of the pulsar wind with the nebula .
we propose that such variable @xmath0-ray emission originate in the region behind the shock when the electrons can be accelerated as a result of the reconnection of the magnetic field compressed by the decelerating pulsar wind .
the natural consequence of such interpretation is the prediction that the crab nebula @xmath0-ray spectrum produced by electrons as a result of the inverse compton scattering of soft radiation to multi - tev energies should also show synchronous variability on the time scales as observed at gev energies by the agile and fermi - lat telescopes .
we calculate how the end of the ic component of the crab nebula @xmath0-ray spectrum should look like during the quiescent and the flare gev @xmath0-ray emission .
we conclude that the variability of the multi - tev @xmath0-ray spectrum from the crab nebula might in principle be responsible for the differences between the spectral features reported by the hegra and hess collaborations at the multi - tev energies .
[ firstpage ] neutron stars : nebulae individual : crab nebula radiation mechanisms : non - thermal gamma - rays : theory |
since the discovery of high-@xmath12 cuprate with its perovskite structure , a large number of measurements of thermal conductivity @xmath16 have been reported up to date , giving crucial information about transport properties in the superconducting and normal states@xcite .
the thermal conductivity of solids is usually separated into the phonon and electronic components @xmath17 and @xmath18 @xcite . for high-@xmath12 copper oxide superconductors yba@xmath3cu@xmath4o@xmath5(ybco ) ,
bi@xmath3sr@xmath3cacu@xmath3o@xmath19 and tl@xmath3sr@xmath3cacu@xmath3o@xmath19 , a rapid enhancement in @xmath20 is observed below @xmath12 , with decreasing @xmath21 , then @xmath20 reaches a maximum around a half of @xmath12 and finally shows a monotonous decrease at low temperatures .
tewordt et al .
ascribed the origin of the anomaly to the phonon contribution @xmath17 on the basis of the classical theory of thermal transport in superconductors @xcite . on the contrary , yu et al .
have proposed a new interpretation of the anomalous enhancement in @xmath20 as being responsible for the electron contribution @xmath18 @xcite .
following their proposal , a strong suppression in the quasiparticle scattering rate below @xmath12 causes the @xmath20 anomaly , which is consistent with an observed peak in measurements of microwave conductivity in yba@xmath3cu@xmath4o@xmath5 @xcite .
in addition , the magnetic field effect on the thermal conductivity in high-@xmath12 superconductors enables us to examine the scattering processes of thermal carriers due to magnetic vortex lines @xcite . in ref .
@xcite , the authors developed a two component method to analyze the field dependence of the thermal conductivity and concluded that only the electronic contribution is field sensitive . in a similar way , a procedure to check the linear field dependence of inverse electronic thermal conductivity was proposed by pogorelov et al .
. however , the origin of the anomalous peak in @xmath20 and its field dependence are still open questions . in this paper , we present measurements of the in - plane thermal conductivity @xmath11 of sm - substituted single crystals ( y@xmath1,sm@xmath2)ba@xmath3cu@xmath4o@xmath5 ( @xmath6=0 , 0.1 , 0.2 and 1.0 ) as functions of temperature and magnetic field . to our knowledge
, no measurement of thermal conductivity has been performed on a - site substituted single crystals ybco except for polycrystalline samples @xcite . here , we give some comments on the substitution effect of sm ion on the crystal growth of yba@xmath3cu@xmath4o@xmath5 . in a previous work , the substitution of sm for y led to a high growth rate for single crystal growth , which is about two times greater with a sm content @xmath6=0.12 than that of ybco @xcite . moreover , by controlling the sm content , one obtained high superconductivity for sm substituted ybco . the @xmath13 profile is fitted using the electron contribution model , to estimate the mean free path of quasi particles @xmath14 near a half of @xmath12 in the absence of field @xcite .
compared with the linear field dependence of the inverse @xmath18 , the square - root fit to the @xmath13data is applied .
in addition , we discuss the impurity substitution effect on the thermal conduction of large single crystals yba@xmath3cu@xmath4o@xmath5 .
a series of single crystalline ( y@xmath1,sm@xmath2)ba@xmath3cu@xmath4o@xmath5 ( @xmath6=0 , 0.1 , 0.2 and 1.0 ) and yba@xmath3(cu@xmath7zn@xmath8)@xmath4o@xmath5 ( @xmath9=0.0071 ) were grown by the crystal pulling method .
sm contents range from 0 , 0.1(0.119),0.2(0.215 ) and 1.0 .
as - grown crystals with @xmath6=0 , 0.1 , 0.2 and @xmath9(zn)=0.0071 were annealed at @xmath22c in the oxygen atmosphere ( for 1 to 2 weeks)to achieve the optimal @xmath12 with @xmath23 , as shown in fig.[mt ] .
smba@xmath3cu@xmath4o@xmath5 ( smbco ) crystal was oxidized at @xmath24c for 200h , giving a higher @xmath12 property .
the slight degradation in @xmath12 for smbco probably arises from the partial substitution of sm for ba site rather than oxygen defects because we annealed it in the oxygen atmosphere for a very long time .
the oxygen content for the as - grown crystal with @xmath6=0 was estimated to be @xmath25 using the iodometric analysis on one piece cut from the same batch .
sample preparations are described by yao et al.in @xcite .
sample dimensions are typically [email protected] mm@xmath27 in the @xmath28-plane and 1.7 mm along the @xmath10-axis .
the epma analysis showed that sm ion is uniformly distributed in single crystals .
the inductively coupled plasma ( icp ) technique revealed that single crystalline samples used in our measurements are close to the nominal compositions .
measurements of the thermal conductivity , in the @xmath28-plane and along the @xmath10-axis , were done with a conventional steady - state heat - flow method@xcite . the magnetic field effect on the @xmath28-plane thermal conductivity
was examined in the application of fields up to 14 t along the @xmath10-axis at the tsukuba magnet laboratory , the national institute for materials science and at the high field laboratory for superconducting materials , institute for materials research , tohoku university .
a temperature gradient across the sample was determined using a differential - type chromel - constantan thermocouple after a calibration procedure .
the magnetization measurement was carried out using the squid magnetometer .
the temperature variation of the resistivity was measured on the identical samples using a four - probe technique .
the in - plane thermal conductivity @xmath11 for single crystals ( y@xmath1,sm@xmath2)ba@xmath3cu@xmath4o@xmath5 ( @xmath6=0 , 0.1 , 0.2 and 1.0 ) and yba@xmath3(cu@xmath7zn@xmath8)@xmath4o@xmath5 ( y=0.0071 ) is shown as a function of temperature in fig.[kt ] . for comparison , the out - of - plane thermal conductivity @xmath29with @xmath6=0 and 1.0 is also depicted in the fig.[kt ] . for the identical crystals used for our experiment , measurements of the low - field magnetization and the in - plane resistivity are presented in fig.[mt ] , to check the superconducting properties .
the resistivity data for the lightly zn doped sample will be shown in the inset of fig.3(b ) .
first of all , the @xmath11 of ( y@xmath1,sm@xmath2)ba@xmath3cu@xmath4o@xmath5 showed a rapid enhancement upon crossing @xmath12 with decreasing temperature , while for @xmath29 no remarkable variation was detected near @xmath12 .
the substitution of y by sm strongly suppresses a maximum of @xmath30 from 160 mw / cmk at @xmath6=0 down to 50 mw / cmk at @xmath6=0.2 through 85 mw / cmk at @xmath6=0.1 .
however , the pronounced peak of @xmath11 below @xmath12 is still observed even for the @xmath6=0.2 crystal with its low thermal value . for the smbco crystal , @xmath11 exhibits a low thermal conductivity accompanied by the broad maximum near 40k .
this finding is closely related to a local lattice distortion due to the slight substitution of sm for ba as mentioned before . for free ions ,
the effective moment of sm@xmath31 is 0.85 @xmath32 and it is much smaller than the value of other rare earth ions ( 3.62 @xmath32 for nd@xmath31 and 7.96 @xmath32 for gd@xmath31 ) .
if the phonon scattering due to magnetic ions with the smaller moment is assumed to be in the presence of the crystal field for smbco , it will contribute a suppression observed in @xmath11 to some extent .
now , we comment on @xmath16 in the normal state for the superconducting crystals .
the thermal conductivity is separated into the electronic and phonon components .
first , the electronic thermal conductivity in the normal state @xmath33 is estimated from the resistivity data using the wiedemann - franz ( wf ) law @xmath34 with a lorentz number @xmath352.45@xmath36w@xmath37k@xmath38 .
for all superconducting crystals , the estimated value @xmath33 is almost independent of @xmath21 because of the @xmath21 linear dependence of @xmath39 , as shown in fig.[ke](a ) .
accordingly , when @xmath40 gives rise to the phonon component @xmath41 , the @xmath21 dependence of @xmath42 observed represents that of the phonon component in the normal state @xcite . for comparison ,
let us show in fig.[ke](b ) the @xmath11 data for the corresponding as - grown crystals of ( y@xmath1,sm@xmath2)ba@xmath3cu@xmath4o@xmath5 ( @xmath6=0 and 0.1 ) and yba@xmath3(cu@xmath7zn@xmath8)@xmath4o@xmath5 ( @xmath9=0.0071 ) . in the inset of fig.3(b ) , the as - grown @xmath39 data for the zn - doped sample are presented and the electronic component @xmath33 is neglected from the wf law using its resistivity data . for other as - grown crystals , we also expect an insulating property or a weakly metallic one with high resistivity .
we note that the as - grown crystals for our experiment were prepared at least under the similar crystal growth environment .
accordingly , we conclude that for all as - grown crystals used , @xmath43 .
the @xmath11 at @xmath6=0 follows a weakly @xmath44-like behavior , while for the sm - substituted crystal it is almost independent on @xmath21 .
this finding is probably caused by the enhanced phonon scattering due to the sm substitution . here , at high temperatures , the @xmath21 dependence of @xmath42 of the as - grown crystal
is expressed as @xmath45 , where @xmath46 and @xmath47 are thermal resistances due to defect scattering and phonon - phonon umklapp scattering , respectively .
moreover , for the light zn doping for the as - grown sample we observed a similar @xmath21 dependence of @xmath16 as that in the pure sample , indicating that a small amount of zn - impurity does not affect the phonon thermal conduction .
in contrast , the light zn doping for the superconducting crystal strongly suppresses @xmath12 down to 80k through the superconducting pair breaking , resulting in a reduced peak in @xmath42 in fig.[kt](a)@xcite .
we notice a common trend in the temperature dependence of phonon component between as - grown and superconducting samples .
for example , @xmath42 of both as - grown and oxygen - annealed samples with @xmath6=0 shows a weakly @xmath44-like behavior at high-@xmath21 .
in addition , the oxygen content does not affect the @xmath21 dependence of @xmath17 strongly except for highly insulating crystal yba@xmath3cu@xmath4o@xmath48(@xmath49 ) @xcite .
accordingly , let us assume that a minor modification of @xmath42 of as - grown crystals expresses the temperature variation of the phonon component below @xmath12 for superconducting crystals if we ascribe the @xmath16 peak to the electronic contribution . thus , the original @xmath16 data for as - grown crystals are scaled to coincide with @xmath40 for the corresponding superconducting crystals at high temperatures ( 150k ) .
the solid and dashed curves in fig.[kt ] represent the phonon contribution and , at the selected temperature of 40k , we obtain 43 @xmath50 at @xmath51 0 , 55 @xmath50 at @xmath52 0.1 and 71 @xmath50 at @xmath9(zn)=0.0071 .
( the corresponding electronic component is listed in table 1 ) this rough estimation in @xmath17 is comparable with the value ( @xmath53 near @xmath12/2 for yba@xmath3cu@xmath4o@xmath5 with @xmath54 ) evaluated by takenaka et al . , @xcite .
separately , a primitive estimation on the phonon component of an untwinned crystal ybco with @xmath12=90.5k gives a similar value @xcite .
next , we show in the fig.[kh ] the magnetic field dependence of the in - plane thermal conductivity @xmath55 of single crystals ( y@xmath1,sm@xmath2)ba@xmath3cu@xmath4o@xmath5 ( @xmath6=0 , 0.1 and 1.0 ) and yba@xmath3(cu@xmath7zn@xmath8)@xmath4o@xmath5 ( @xmath9=0.0071 ) .
the @xmath55 data are normalized by the zero field value . at low fields ,
the value of @xmath11 exhibits a rapid drop around nearly @xmath12/2 and it then tends to saturate at high fields up to 14 t. the applied field causes a substantial decrease in @xmath56 by @xmath57 at @xmath6=0 , while @xmath56 of sm - substituted samples shows a smaller degradation from @xmath58 at @xmath6=0.1 to @xmath59 at @xmath6=1.0 .
in the mixed state of conventional superconductors , we believe that thermal carriers ( quasiparticles and phonons ) are scattered by the magnetic vortices , resulting in a substantial decrease in @xmath16 .
now , we try to fit the @xmath55 data to the electronic contribution model @xcite .
the electronic thermal conductivity in magnetic fields , @xmath60 , is rewritten using both @xmath61and @xmath14 as follows ; @xmath62 here , the total mean free path of quasiparticles is expressed such as @xmath63 , where @xmath14 and @xmath64 are the mean free path of quasiparticles in zero field and in sufficiently strong fields , respectively .
accordingly , the field dependence of the thermal conductivity , @xmath65,is fitted using the following formula @xmath66 , where we set @xmath67 as @xmath68 and @xmath69 , for @xmath70=1 and 0.5 , respectively ( @xmath71 , the quantum flux ) . in the case of @xmath72
, @xmath73 represents a scattering cross section per vortex cores in transport @xcite . when the quasiparticle scattering from vortex cores is proportional to the number of the magnetic quantum flux , we get @xmath74=@xmath75@xcite . thus , for the exponent @xmath70=1 in eq .
( [ fit ] ) , we take into account for the mean free path of quasiparticles @xmath76 from the scattering due to vortex cores , in addition to @xmath14 for scattering processes in the meissner state . using the @xmath73 = 90 @xmath77 in ybco @xcite
, @xmath76 is @xmath78 t@xmath79 . on the other hand , in the case of @xmath70=0.5 , assuming the quasiparticle scattering due to screening currents around vortex cores in a disordered vortex array , the vortex contribution @xmath76 is proportional to the average lattice constant of a vortex lattice @xmath80 i.e. , @xmath81 , where @xmath82 is a constant independent of fields @xcite . using @xmath82=0.5 in @xcite , @xmath76 becomes @xmath83 t@xmath84 .
our reliable fits of the @xmath13 data to eq.([fit ] ) give the two parameters , @xmath85(=1-@xmath86 ) and @xmath87 , at selected temperatures as summarized in table 1 .
it is truly that our fitting procedures yield a better fit for the @xmath72 than that for the @xmath88 , but it is difficult to distinguish a @xmath89-linear or @xmath90-dependence in the denominator of @xmath13 in fig.[kh ] . in @xmath89-linear fits , we notice the underestimation in electron component for all crystals if we accept the phonon component estimated from the as - grown data . for example , in the pure crystal of @xmath6=0 , the @xmath89-linear fit yields @xmath91 , while in the @xmath90 fit we get @xmath92 .
this tendency in @xmath18 below @xmath12 appears in previous works on the field dependences of @xmath13 for underdoped yba@xmath3cu@xmath4o@xmath93 and tl@xmath3ba@xmath3cuo@xmath94@xcite .
such the underestimation in @xmath18 will enable us to consider a significant modification of the free - electron lorenz number in the wf law for the strongly electron correlated system such as high- @xmath12 cuprates , giving the reduced electron thermal conductivity @xcite .
finally , let us examine the fitting parameter @xmath95 listed in table[table1 ] . from @xmath96
, we obtain the mean free path of quasiparticles @xmath97 m for @xmath70=1 at the pure crystal . here
, we used the cross section of @xmath73 = 90 @xmath77 in ybco @xcite . in a similar way , for the @xmath70=0.5 , @xmath98 m with @xmath82=0.5 in @xcite .
both the former and latter fits give @xmath991000@xmath100 at 40k for pure @xmath6=0 crystal .
from table[table1 ] , we guess that @xmath14 at the sm substituted crystal is about one - third as long as the value of @xmath14 of the pure crystal , whether we use the @xmath89-linear fit or the @xmath90 . if the normal state @xmath101 is proportional to @xmath102 at @xmath103 , then the @xmath33 plots in fig.[ke](a)give the ratio of @xmath14 at @xmath6=0.1 to that at @xmath6=0 @xmath104 at higher @xmath21 , this value is not far from the above results in the @xmath14 ratio ( @xmath105 ) around a half of @xmath12 .
moreover , for the lightly zn - doped crystal , our fits to the data indicate the reduced mean free path of quasiparticles .
a pair - breaking effect due to zn - doping makes quasiparticles increase , giving the strongly electron - electron scattering enhanced below @xmath12 .
this finding has a close relationship with the reduced peak observed in @xmath30 of yba@xmath3(cu@xmath7zn@xmath8)@xmath4o@xmath5 with y=0.0071 .
we have studied the in - plane thermal conductivity of large single crystals ( y@xmath1,sm@xmath2)ba@xmath3cu@xmath4o@xmath5 ( @xmath6=0 , 0.1 , 0.2 and 1.0 ) and yba@xmath3(cu@xmath7zn@xmath8)@xmath4o@xmath5 ( y=0.0071 ) as functions of temperature and magnetic field .
the nonlinear field profile of @xmath13 observed around a maximum peak is discussed on the basis of the electron contribution model .
we tried to make the fits of the @xmath13 data to the @xmath89-linear or @xmath90 dependence of the inverse electronic thermal conductivity in eq.(2 ) .
the local distortion due to the sm substitution for y site suppresses both the phonon and electronic contributions , accompanied by the substantial decrease in @xmath14 at 40k in comparison with that of the @xmath6=0 pure sample .
the light zn doping for cuo@xmath3 planes does not contribute the phonon conduction but affects solely the electron component below @xmath12 , resulting in the reduced mean free path of quasiparticles in the superconducting state . for smbco ,
a suppressed maximum in @xmath20 arises from the partial substitution of sm for ba site , which is closely related to the slight degradation in @xmath12 .
one of authors ( x yao ) would like to thank nbrpce(no.2006cb601003 ) and sstc ( no.055207077 ) for financial support .
we are grateful to f.sato and prof.s.kambe for the iodometric analysis .
.[table1 ] the fitting parameters @xmath106 and @xmath95 from the fit of @xmath55 data to eq.([fit])for single crystals ( y@xmath1,sm@xmath2)ba@xmath3cu@xmath4o@xmath5 ( @xmath6=0 , 0.1 and 1.0 ) and yba@xmath3(cu@xmath7zn@xmath8)@xmath107 o@xmath5 ( @xmath9=0.0071 ) . for comparison
, we show the electronic component near @xmath12/2 , estimated from the @xmath16 data of as - grown crystals described in the text . on the last column ,
the electronic components in the normal state are given using the wf law . [
cols="^,^,^,^,^,^,^,^,^,^ " , ]
30 c.uher , j.supercond.3,337(1990 ) . j.m.ziman,electrons and phonons .
clarendon press , oxford,1960 .
l.tewordt and t.h.wolkhausen , solid state commun.70,839(1989 ) .
r.c.yu , m.b.salamon , j.p.lu , and w.c.lee , phys.rev.lett.69 , 1431(1992 ) .
d.a.bonn , ruixing liang , t.m.riseman , et al.,phys.rev.b47,11314(1993 ) .
s.d.peacor,j.l.cohn,and c.uher , phys.rev.b43,8721(1991 ) .
r.a.richardson , s.d.peacor,j.l.cohn,f.nori , and c.uher , phys.rev.lett .
67,3856(1991 ) .
f.yu , m.b.salamon , v.n.kopylov , n.n.kolesnikov , h.m.duan and a.m.hermann , physica c235 - 240 , 1489(1994 ) .
y.pogorelov , m.a.arranz,r.villar , and s. vieira , phys.rev.b51,15474(1995 ) .
s.t.ting,p.pernambuco-wise , and j.e.crow , phys.rev.b50,6375(1994 ) .
r.k.williams,p.m.martin,and j.o.scarbrough,phys.rev.b59,13639(1999 ) .
x.yao,m.nakamura,m.tagami,and y.shiohara , physica c272,227(1996 ) .
k.krishana,j.m.harris,and n.p.ong , phys.rev.lett.75,3529(1995 ) .
x.yao,t . izumi , and y.shiohara , supercond.sci .
technol.16,l13(2003 ) .
x.yao,k.ohtsu,s.tajima,h.zama,f.wang , and y.shiohara , j.mater.res.11,1120(1996 ) .
m.matsukawa,t.mizukoshi , k.noto,and y.shiohara , phys.rev.b53,r6034(1996 ) .
k.takenaka,y.fukuzumi,k.mizuhashi , and s. uchida , phys.rev.b56,5654 ( 1997 ) .
p.j.hirschfeld and w.o.putikka , phys.rev.lett.77,3909(1996 ) .
j.l.cohn,c.k.lowe-ma , and t.a.vanderah,phys.rev.b52,r13134(1995 ) .
r.m.cleary,phys.rev.175,587(1968 ) .
n.p.ong , k. krishana , y. zhang and z.a.xu , cond - mat/9904160 .
m.franz , phys.rev.lett.82 , 1760(1999 ) .
y.zhang,n.p.ong,z.a.xu,k.krishana,r.gagnon , and l.taillefer , phys.rev.lett.84 , 2219(2000 ) . for single crystals ( y@xmath1,sm@xmath2)ba@xmath3cu@xmath4o@xmath5 ( @xmath6=0 , 0.1 , 0.2 and 1.0 ) and yba@xmath3(cu@xmath7zn@xmath8)@xmath4o@xmath5 ( @xmath108 ) as a function of temperature . for comparison , the out - of - plane thermal conductivity @xmath29with @xmath6=0 and 1.0 is also depicted .
solid and dashed curves in ( a ) represent the estimated phonon component @xmath109 for @xmath6=0 and @xmath108,respectively ( see to the text ) . in a similar way
, a solid curve in ( b ) shows the phonon contribution @xmath109 of the @xmath6=0.1 crystal .
, width=453 ] under a zero - field cooling ( zfc ) scan and ( b ) in - plane resistivity @xmath110 , for single crystals ( y@xmath1,sm@xmath2)ba@xmath3cu@xmath4o@xmath5 ( @xmath6=0 , 0.1 , 0.2 and 1.0 ) . the @xmath111 data of yba@xmath3(cu@xmath7zn@xmath8)@xmath4o@xmath5 ( @xmath108 ) are given in the inset of ( a ) .
the inset of ( b ) represents magnified plots for the @xmath110 data of the sm - substituted samples .
, width=453 ] estimated from the resistivity data using the wiedemann - franz ( wf ) law @xmath34 with a lorentz number @xmath352.45@xmath36w@xmath37k@xmath38 .
( b ) the @xmath11 data for the corresponding as - grown crystals of ( y@xmath1,sm@xmath2)ba@xmath3cu@xmath4o@xmath5 ( @xmath6=0 and 0.1 ) and yba@xmath3(cu@xmath7zn@xmath8)@xmath4o@xmath5 ( @xmath9=0.0071 ) . in the inset of ( b ) , the @xmath110 data for the as - grown and oxidized samples with light zn doping are presented.,width=453 ] of single crystals ( y@xmath1,sm@xmath2)ba@xmath3cu@xmath4o@xmath5 ( @xmath6=0 , 0.1 and 1.0 ) and yba@xmath3(cu@xmath7zn@xmath8)@xmath4o@xmath5 ( @xmath9=0.0071 ) , at selected temperatures .
the applied field @xmath89 is parallel to the @xmath10-axis .
solid curves represent better fits of the @xmath13 data to the @xmath89-linear dependence of the inverse electronic thermal conductivity in eq.(2 ) . for comparison ,
the inset shows the curve fits to the @xmath90-dependence of the inverse @xmath15 in eq.(2 ) .
, width=453 ] | we have investigated the in - plane thermal conductivity @xmath0 of large single crystals of optimally oxygen - doped ( y@xmath1,sm@xmath2)ba@xmath3cu@xmath4o@xmath5 ( @xmath6=0 , 0.1 , 0.2 and 1.0 ) and yba@xmath3(cu@xmath7zn@xmath8)@xmath4o@xmath5 ( @xmath9=0.0071 ) as functions of temperature and magnetic field ( along the @xmath10 axis ) . for comparison ,
the temperature dependence of @xmath11 for as - grown crystals with the corresponding compositions are presented .
the nonlinear field dependence of @xmath11 for all crystals was observed at relatively low fields near a half of @xmath12 .
we make fits of the @xmath13 data to an electron contribution model , providing both the mean free path of quasiparticles @xmath14 and the electronic thermal conductivity @xmath15 , in the absence of field . the local lattice distortion due to the sm substitution for y suppresses both the phonon and electron contributions . on the other hand
, the light zn doping into the cuo @xmath3 planes affects solely the electron component below @xmath12 , resulting in a substantial decrease in @xmath14 . |
we consider a quantum network which constitutes a periodic two - dimensional lattice , see fig .
the nodes are located at lattice sites , and each bond connecting nearest neighboring sites correspond a pure entangled state shared by two nodes , @xmath0 , where @xmath1 and @xmath2 is assumed .
suppose we need to transmit some qubits from alice at node @xmath3 in the network to bob at node @xmath4 .
we follow the standard protocol : first , we convert each entangled state represented by a bond to a maximally entangled state .
the optimal conversion probability is @xmath5 , @xcite .
the quantum network then is mapped to a classical network with each bond appearing probability @xmath6 , see ref . @xcite and fig .
[ eqc ] ( a ) . secondly ,
if the bonds of maximally entangled states can form a path connecting @xmath3 and @xmath4 , by entanglement swapping ( entanglement repeating ) , a maximally entangled state can be generated , which constitutes a quantum channel , between @xmath3 and @xmath4 .
then alice can teleport a qubit to bob .
the entanglement percolation describes well whether this protocol can succeed or not @xcite .
however , different from classical case , the quantum channel can only be used once . after a qubit is teleported through a quantum channel ,
the resource of entanglement is consumed .
the bonds no longer carry the entanglement .
so the number of eqc of a quantum network is a key quantity in describing the efficiency of quantum communication , see fig .
[ eqc ] .
we define the eqc as , @xmath7 where @xmath8 is the expectation number of quantum channels connecting @xmath3 and @xmath4 , and @xmath9 which is actually the number of bonds connecting each node in one to one communication and thus is fixed for a lattice . by definition , eqc increases with @xmath6 , is upper bounded by 1 , and depends on the network and the protocol used .
greater eqc means higher efficiency of qubit transmission .
it is generally very hard to find eqc analytically .
however , as we can see below , the regular structures of lattices allow us to build a rather simple model which can approximate eqc well . after the singlet conversion . *
( a ) * blue bonds are states that has been successfully converted to maximally entangled states while gray bonds represent those failed . *
( b ) * the three independent quantum connections between a ( alice ) and b ( bob ) marked as green points are highlighted in red , yellow and purple ( also 1 , p2 and 3 ) . * ( c ) * a part of an infinite square network with @xmath10 and @xmath11 .
the three quantum connections between alice and bob ( green points ) are highlighted in red ( also 1 , 2 and 3 ) . * ( d ) * the attractive circles in the network .
each circle has 4 ` internal bonds ' .
outside the effective circles the network can be regarded simply as an isotropic ` medium ' .
so , whether a quantum channel can be formed or not mostly depends on how many ` internal bond ' is formed.,width=317 ]
now , we consider a quantum network with square lattice configuration .
we remark that our method is general and applicable for various quantum networks .
we also assume that @xmath12 , which is clear above the percolation threshold of square lattices , @xmath13 .
when @xmath6 is near the critical point , the quantum state transferring efficiency can be reasonably assumed to be low .
the effective approximation method may not be applicable , so we will not discuss this situation .
s. since the attractive radii pertain to relative change of eqc , we depict the relative eqc in which all eqc of @xmath14 are normalized as 1 . for the original image ,
see @xcite for details .
the corresponding @xmath15s for @xmath16 are 1.2 , 1.7 , 10.6 , 10.9 , with @xmath17 .
a rather fast convergence can be observed when @xmath18 . *
( b ) * monte carlo estimations of eqc from two nodes to infinity for different @xmath6s .
for convenience in demonstration of repulsive radii we still present the relative eqc .
all eqc of @xmath19 are normalized as 1 . for the original image ,
see @xcite for details .
the corresponding @xmath15s for @xmath16 are 0.64 , 1.04 , 1.59 , 2.37 , with @xmath20.,width=328 ] let alice and bob be on two lattice nodes with distance @xmath21 , which is defined as the linear distance and the distance of two nearest neighboring sites is assumed to be 1 .
now alice would like to send some qubits to bob via the network .
monte carlo estimations ( fig .
[ mc ] ) shows that when @xmath12 , eqc between alice and bob decreases and converges quickly to its asymptotic value when @xmath21 increases .
this fast convergence suggests us to define the attractive radius of an effective circle , the eqc can be fitted by an exponential function , @xmath22 where @xmath23 , @xmath24 are parameters depending on @xmath6 .
we then can relate the attractive radius with an exponent parameter , @xmath25 .
when @xmath26 , i.e. the attractive circles are not overlap , the second term in eq .
( [ fitfunc ] ) can be neglected since it is less than @xmath27 as we generally used physically .
thus , we know that @xmath23 represents @xmath28 asymptotically when distance @xmath21 is large , this is what we are interested
. we will present an estimation of @xmath23 below .
this formula agrees well with numerical results , see fig.[mc ] . for @xmath12
, we have a rough estimation for the attractive radius , @xmath29 .
this means that the effective circle of alice or bob includes only their nearest neighboring sites on a square lattice .
the physical interpretation of @xmath30 can be understood as the radius of an effective circle . when the distance of @xmath3 and @xmath4 is larger than @xmath31 , the eqc will be approximated being independent of their distance , @xmath32 which is a fixed value for a lattice .
it seems like that the two nodes are connected by first connecting from their positions to the borders of their effective circle , and second the two circles are connected through the isotropic medium .
this result is significant since it means that the number of qubits capable transmitted through a quantum network does not decrease with the increase of the distance .
it avoids the usual exponential decaying probability for long - distance quantum communication .
our result is also consistent with the conclusion that quantum percolation in quantum networks can succeed when @xmath6 is larger than a threshold .
the eqc further quantifies the number of quantum channels available in a quantum network .
when @xmath3 and @xmath4 are located near , it is possible that more quantum channels can be constructed
. so the attraction will be beneficial which stimulates the name ` attractive radius ' .
next , we consider a different scheme for quantum communication in quantum networks .
the scheme is that alice and bob at different nodes would like to send qubits to someone at positions of infinity in the network .
here we only concern about how many qubits can be sent out by alice and bob .
a simple consideration is that if alice and bob are located near in the quantum network , their ability of sending out qubits will be influenced by each other .
the problem is in what distance they can send out qubits independently .
we name the radius of this circle as the repulsive radius .
monte carlo estimations ( fig .
[ mc ] ) show that when @xmath12 , eqc for alice and bob together to transfer states to someone at infinity increases and converges quickly to its asymptotic value when @xmath21 increases , similar as the case of the attractive circles .
also , we fit the monte carlo data to an exponential function like eq .
( [ fitfunc ] ) .
we find that eqc of this type is also described well by an exponential function ( fig .
[ mc ] ) , @xmath33 where @xmath34 are parameters similar to those in previous consideration of attractive radii .
also we define the repulsive radius as , @xmath35 .
we emphasize that the eqc here means that alice and bob both send qubits to infinity . for @xmath12 , we have a rough estimation , @xmath36 .
it means that when distance of alice and bob is larger than @xmath37 , they can be considered independent in sending out qubits .
the repulsive radius @xmath38 also has a clear physical interpretation .
it describes the range of the bonds that have a significant influence to the quantum channels connecting a and b to infinity . for small @xmath21 , the paths which connect @xmath3 and @xmath4 with positions of infinity used to build the quantum channels will conflict with each other , which will decrease the number of quantum channels resulting in a smaller eqc . the key feature of the eqc as a function of distance is its rapid convergence to a stationary value .
it is worth mentioning that this is a universal feature for various lattices .
now we can easily approximate eqc analytically by using the physical meanings of @xmath30 and @xmath38 .
suppose alice needs to transmit qubits to bob .
two situations are considered : ( i ) point - to - point quantum communication , alice and bob both can access only one node in the network ; ( ii ) multipartite - to - multipartite quantum communication , they each can access @xmath39 nodes . here
we assume that each node is independent from other nodes .
that is , any pair distance of those @xmath39 nodes possessed by alice ( bob ) is larger than 2@xmath38 . and alice s group nodes are far from bob s group nodes . within this situation ,
our approach gives @xmath23 in eq .
( [ fitfunc ] ) .
henceforth , all eqc we mentioned here are @xmath23 since distance is supposed large . of triangle lattice(the proportion of nodes in the largest connected nodes ) @xcite
. we may observe that @xmath40 ( red linep ) corresponding to entanglement percolation is like a ` jump function ' demonstrating only success or failure , while the behavior of the proposed eqc is different , it in general increases monotonically with @xmath6 and approaches 1 finally when @xmath6 takes the limit of 1 . because the exact expression of @xmath40 is an infinite series and we picks first 30 terms here , the zero point of the curve is slightly different from the exact percolation threshold @xmath41 @xcite . but the ` jump function ' property is maintained . *
( b ) * lines with different symbols represent monte carlo estimation of eqc of square , triangle and hexagon lattices .
red lines are fitted curves according to eq .
p([fitform ] ) , with corresponding lattice indices @xmath42 , @xmath43 , @xmath44.,width=332 ] here we still consider the case of square lattice firstly . for the first situation ( i ) ,
when @xmath12 which means @xmath45 , the effective circle contains only 5 nodes , 4 internal bonds and 12 peripheral bonds connecting to the outside of the circle . when alice and bob are far away from each other , the quantum channels between @xmath3 and @xmath4 depend mostly on number of maximally entangled states induced by those 4 internal bonds for alice or bob
. only pairs of internal bonds belonging respectively to two sides can possibly form independent quantum channels .
the number of established pairs is denoted by @xmath46 which is a random number .
thus the expectation of the number of pairs is , @xmath47 we remark that when @xmath48 is omitted , the above equation equals to @xmath49 . when @xmath50 , the medium can be seen as fully connected , and we have an asymptotic limit , @xmath51 . for general @xmath12 ,
notice that for a connected neighboring point of @xmath3 , the probability that at least one peripheral bond is connected to the outside is @xmath52 . under this consideration ,
the probability that the quantum connection does not meet a dead end in the ` medium ' can be written as @xmath53 , where @xmath54 is like an index of the medium depending only on lattice which can be fitted by numerical data . in this way
, an estimated eqc can be written as follows , @xmath55 we can find that this formula agrees well with the data , see fig .
[ fit ] . the fig .
[ fit](a ) also shows difference between entanglement percolation and eqc .
it is worth mentioning that the eqc also experienced a phase transition at the percolation threshold .
next , we consider situation ( ii ) .
alice can access @xmath39 independent points @xmath56 , @xmath57 , ... , @xmath58 , bob can similarly access @xmath59 , @xmath60 , ... , @xmath61 .
the same reasoning can be applied , the eq.([exppairs ] ) can be generalized as , @xmath62 then we immediately have , @xmath63 the validity of this formula can be confirmed similarly as eq .
( [ fitform ] ) , and it agrees well with numerical results .
we remark that here we assume @xmath6 is far from the critical point .
higher order of correction terms become significant and thus not negligible when the considered @xmath6 gets closer to the critical point .
actually the number of correction terms tend to infinity in eq .
( 5 ) and eq .
( 7 ) when the estimation is carried out at the critical point of the network . for comparison between eq .
( [ kfit ] ) and monte carlo approach , see @xcite for figures .
data shows that multipartite - to - multipartite quantum communication offers a higher eqc .
this can be explained as that more bonds tend to be easier to be paired up , and the influence of the minimum function is weakened .
another important kind of situation is that alice and bob can access different number of nodes .
for example , alice can access one node while bob can access @xmath642 nodes .
this resembles a ` distribution ' scenario .
results show that eqc is very close to @xmath6 and is almost constant when @xmath65 .
this can be explained similar to the previous case by noting that the four bonds within the attractive effective circle of alice can almost always be paired in this case , see @xcite for figures .
lattice transformation by entanglement swapping at some nodes can enhance entanglement percolation properties @xcite .
this is done by altering the lattice structure by carrying out first at some nodes the optimal entanglement swapping protocol and correspondingly changing the percolation threshold of the network . in this section
we focus on the eqc of the networks undertaking various transformations and study whether the lattice transformations can increase or decrease the communication efficiency . for lattice transformation ,
we need to modify the definition of eqc as @xmath66 where @xmath67 refers to the expectation number of quantum channels connecting @xmath3 and @xmath4 in the new network , and @xmath68 is that of the old network with @xmath69 , regarded as fixed initial entanglement resources .
( if two or more bonds are connecting two neighboring nodes , consider the bond - separated situations ) .
this modification takes lattice transformation effects into account , and the eqc can be compared to the previously defined one to evaluate the effect of the transformations . by using monte carlo methods ,
we observe two typical situations when lattice transformations may be beneficial .
when the entanglement percolation have different thresholds for the corresponding lattices , however , the eqc curves may demonstrate two behaviors , * ( i ) * intersecting or * ( ii ) * non - intersecting . for the * ( i ) * non - intersecting case , lattice transformation can enhance the efficiency for all @xmath6 above the percolation threshold .
in contrast for the * ( ii ) * intersecting case , the lattice transformation induced by entanglement swapping is beneficial in some regions while in other regions is not , i.e. the eqc may decrease after lattice transformation . in the following ,
we consider two lattice transformations as examples .
_ double - bond hexagonal lattice._ we know that a double - bond hexagonal lattice with @xmath6 slightly less than its percolation threshold @xmath70 can be transformed into a single - bond triangular lattice with @xmath71 greater than its percolation threshold @xmath72 via entanglement swapping @xcite .
our approach by estimating eqc reveals that there is an intersection between two eqc curves .
the corresponding @xmath6 of the intersection can be estimated by rewriting eq.([fitform ] ) for triangle and hexagonal lattices @xmath73 , @xmath74 .
we can substitute @xmath75 and @xmath76 into those equations .
it is apparently that @xmath77 , while applying eq .
( [ fitform ] ) to triangle lattice , we find @xmath78 , which demonstrates an intersection in the curves of eqc ( see fig . [ latticetransform ] ) .
we also notice that for a network with more bonds connecting with one node , its eqc is closer to eq.(5 ) .
monte carlo estimations confirm the intersection .
our approach provides an explicit illustration for this lattice transformation : for @xmath79 , lattice transformation is beneficial not only for a lower critical point in entanglement percolation but a higher efficiency shown by eqc as well . but for @xmath80 , transformation will decrease the quantum communication efficiency represented by eqc .
another lattice transformation involves separating double bonds and turns the network into two identical single - bond networks .
this strategy can be applied for around @xmath81 by considering the eqc ( fig .
[ latticetransform ] ) , in which two separately - used bonds can give a higher @xmath6 .
_ hexagonal lattice with @xmath82 double bonds to square lattice._ for this example , a hexagonal lattice with @xmath83 double bond can be transformed into a single - bond square lattice ( fig .
[ latticetransform ] ) .
first we can easily conclude that the percolation threshold of square lattice is lower than that of both kinds of strategies of hexagonal lattice with @xmath82 double bonds .
monte carlo estimations show that the eqc of square lattice is higher than those two different strategies about the double bonds .
this means we can benefit from lattice transformation for the whole range of @xmath84 .
quantum network is currently one of the central topics of quantum information processing
. quantum communication by teleportation protocol should be one fundamental function of quantum networks where nodes share entanglements such as einstein - podolsky - rosen pairs as quantum channels .
however , despite its many advantages , teleportation is entanglement resource consuming which results in switching off the quantum channel when one state is teleported .
the number of independent or exclusive quantum channels shared by nodes of the quantum networks , which corresponds the amount of quantum information capable of being transmitted , becomes an important problem . in this work
, we systematically investigate this quantity and find some significant properties by both analytical and numerical methods . quantum communication capacity quantified by eqc between nodes of quantum networks
is , surprisingly , independent of the communicating distance .
this not only avoids the exponential decaying of standard quantum repeating communication , but also it actually does not suffer any decaying with distance .
we remark that this non - decaying property is different from the long - distance entanglement by that this quantum channel is established by a chain of entanglement swapping .
it naturally decays exponentially in one - dimensional case depending on the conversion probability @xmath6 .
exact formulae of eqc are obtained , and the involved parameters are studied .
we also show that eqc can be enhanced by lattices transformations performed by quantum measurements . at the same time , however , this may also induce the reduction of eqc in some region of @xmath6 .
eqc is a new concept proposed specifically for quantum networks .
many aspects of this quantity and some closely related topics can be explored further .
this opens a new avenue in studying quantum communication of quantum networks .
_ acknowledgements : _ this research is supported by ` 973 ' program ( 2010cb922904 ) , nsfc , china innovation funding and grants from cas .
see fig . [ abcd ] . and @xmath30 ( @xmath38)are two important parameters of given network and approach of information transmission , width=661 ] . * ( a ) * @xmath85 , the attractive radius @xmath86 , the long distance eqc limit @xmath87 .
* ( b ) * @xmath88 , the attractive radius @xmath89 , the long distance eqc limit @xmath90 . * ( c ) * @xmath85 , the repulsive radius @xmath91 , the long distance eqc limit @xmath92 . *
( d ) * @xmath88 , the repulsive radius @xmath93 , the long distance eqc limit @xmath94 [ abcd ]
in our consideration in the main text , the probability of a successfully formed internal bond of sender is @xmath6 .
if all these bonds can finally connected to receiver , this gives the upper bond of eqc , that is , @xmath95 .
we can make comparison of eqcs between different situations and investigate how much are they closed to @xmath96 .
[ e ] of scenario where alice and bob both can access to @xmath39 independent nodes .
black squares represent eqc ( i ) with @xmath97 . *
( ii ) * alice and bob both can access one nodes .
blue triangles represent eqc ( ii ) .
green line is the estimation function eq .
red line denotes @xmath96 , which is the maximum number of long - distance eqc according to eq.(4)-(7 ) in the main text , width=340 ] .
according to eq . ( 6 ) in the main text , the influence of the minimum function will be reduced , thus gives a higher eqc , when alice and bob can access different number of nodes .
it is because if the number of nodes accessed by receiver is larger than that of sender , the internal bonds of the sender can almost always be paired by internal bonds of receiver s .
[ f ] chou , c. w. , laurat , j. , deng , h. , choi , k. s. , de riedmatten , h. , felinto , d. , & kimble , h. j. functional quantum nodes for entanglement distribution over scalable quantum networks .
_ science _ * 316 * , 1316 ( 2007 ) .
papp , s. b. , choi , k. s. , deng , h. , lougovski , p. , van enk , s. j. , & kimble , h. j. characterization of multipartite entanglement for one photon shared among four optical modes . _ science _ * 324 * , 764 ( 2009 ) .
bennett , c. h. , brassard , g. , crpeau , c. , jozsa , r. , peres , a. & wootters , w. k. teleporting an unknown quantum state via dual classical and einstein - podolsky - rosen channels .
lett . _ * 70 * , 1895 ( 1993 ) .
lapeyre , g. john jr .
, wehr , j. & lewenstein , m. enhancement of entanglement percolation in quantum networks via lattice transformations .
a. _ * 79 * , 042324 ( 2009 ) see the supplementary materials for detailed information . | quantum state can be teleported to a remote site by only local measurement and classical communication if the prior einstein - podolsky - rosen quantum channel is available between the sender and the receiver .
those quantum channels shared by multiple nodes can constitute a quantum network .
yet , studies on the efficiency of quantum communication between nodes of quantum networks remain limited , which differs from classical case in that the quantum channel will be consumed if teleportation is performed . here
, we introduce the exclusive quantum channels ( eqc ) as the measure of efficiency of quantum information transmission .
it quantifies the amount of quantum information which can be teleported between nodes in a quantum network .
we show that different types of eqc are local quantities with effective circles .
significantly , capacity of quantum communication of quantum networks quantified by eqc is independent of distance for the communicating nodes .
thus , the quantum network can be dealt as the isotropic medium where quantum communication is no - decaying .
eqc are studied by both analytical and numerical methods .
the eqc can be enhanced by transformations of lattices of quantum network via entanglement swapping .
our result opens the avenue in studying the quantum communication of the quantum networks .
quantum information processing offers quantum algorithms which may surpass their classical counterparts .
recently , quantum network and its extension , quantum internet which might potentially be the next generation internet , have been attracting a great deal of interests @xcite .
the quantum networks are constituted by quantum nodes where quantum information can be produced , stored and processed locally .
those nodes are supposed to be linked by both quantum channels and classical channels so that quantum states can be teleported between different nodes .
this feature is also essential if each individual node in quantum network is restricted to finite size of state space , and probably the central quantum servers are available .
the quantum network may have the capabilities for quantum computation with a distributed style , simulation of quantum many - body systems @xcite , quantum metrology @xcite , quantum cloud computation with unconditional security by quantum key distribution protocols @xcite .
these exciting features provide the motivation to examine research related to quantum networks
. quantum state can be sent directly between any nodes by flying qubits like photons in quantum networks . on the other hand
, the prior shared entangled state between different nodes , which can be considered as the available quantum channel , can facilitate quantum communication by only local operations on each individual node with assistance of classical communication .
this is the well - known teleportation @xcite .
additionally , the teleportation protocol may increase the fidelity for state transmission in case entanglement purification is applied @xcite .
this is a fundamental function of the quantum network .
the efficiency of such a quantum network for quantum communication should be well described and quantified .
one pioneering work is about the entanglement percolation in quantum networks @xcite .
it is shown that the problem of establishing maximally entangled states between nodes can be related directly to classical percolation in statistical mechanics .
in particular , by applying appropriate local measurements on some nodes which changes the lattice configurations of quantum network and correspondingly changes the percolation threshold , the entanglement percolation can change from failure to success .
however , the entanglement percolation only describes whether a quantum channel between two nodes can be constructed or not , i.e. it only shows ` on ' or ` off ' . if it succeeds , it is still not clear what amount of quantum information can be teleported . in comparison with classical networks ,
this is crucial for quantum networks since once the pre - shared entanglement is consumed for telepotation , new quantum channel should be constructed .
thus the number of exclusive quantum channels , i.e. , independent maximally entangled pairs , become important for quantum networks . in this work
, we define the eqc as the expectation number of maximally entangled pairs , which equals to the number of individual qubits capable teleported between two nodes in an established quantum network , normalized by the number of bonds connecting each node .
we find that two types of eqc depend on the local structure and the bond singlet conversion probability of the quantum network .
particularly , they are independent of the distance of the studied nodes when they are located farther than their effective radii .
they can then be easily computed without considering the detailed connections between these nodes .
significantly , it also means that the eqc are non - exponentially decaying , thus avoiding the main hurdle for long - distance quantum communication .
the eqc can be increased by lattices transformation realized by quantum measurements on some nodes .
we show that this quantity defines effectively the efficiency of the quantum communication for the quantum network .
the eqc are universal quantities applicable in quantum networks with various geometries . |
for the reader s convenience , we present here analytical form of the expressions used in the main part of the paper . for @xmath5 given by ( [ eq : gauss ] ) and @xmath17 given by ( [ eq : nontrivialtransmission ] )
, we obtain the integral @xmath41 | in the context of the sensational results concerning superluminal velocities , announced recently by the opera collaboration , we have proposed a classical model yielding a statistically calculated measured velocity of a beam , higher than the velocity of the particles constituting the beam .
the two key elements of our model , necessary and sufficient to obtain this curious result , are a time - dependent `` transmission '' function and statistical method of the maximum - likelihood estimation . opera neutrino anomaly , superluminal neutrinos 06.30.gv velocity , acceleration , and rotation , 06.20.dk measurement and error theory , 07.05.kf data analysis : algorithms and implementation ; data management , 13.15.+g neutrino interactions , 14.60.lm ordinary neutrinos , 03.30.+p special relativity inspired by the amazing results of the opera collaboration @xcite , claiming that the velocity of light @xmath0 has been beaten by a beam of neutrinos , we propose a simple classical model , which yields the curious effect of a seeming increase of `` effective '' velocity ( see @xcite , for an earlier independent approach , and also @xcite , for a wave version ) .
there are dozens of papers , which have recently appeared , adopting various attitudes towards the results presented by the opera collaboration , for example @xcite ( see also an example of an older work on superluminal velocities @xcite ) .
the attitude of our paper is definitely skeptical . in view of the fact that we have found a ( `` mathematical '' ) model providing an `` artificial '' increase of real velocity , we are forced to put the results announced by the opera collaboration in doubt .
moreover , we have also proposed a simple working example , such that appropriately fitting its parameters one can simulate the controversial results of the opera collaboration .
strictly speaking , our paper does not indisputably invalidates the conclusions drawn by the opera collaboration , but it seriously weakens their argumentation , indicating a logical gap in their reasoning .
our model is purely classical and dynamics free .
no new physics , nor quantum mechanics , nor even ( classical ) wave mechanics is involved .
only standard classical kinematics notions as well as the statistical method of the maximum - likelihood estimation ( mle ) are used in our approach .
the assumptions of our model are the following .
a spatially homogeneous , lasting the period @xmath1 , beam of classical particles ( `` extraction '' ) moving with a constant speed @xmath2 ( @xmath3 ) travels from a source to a detector .
the distance between the source and the detector is @xmath4 .
the probability density function ( pdf ) of the time of emission of the particles within the duration @xmath1 of production of the beam is given by the function @xmath5 .
in an ideal situation ( none of the particles is lost ) we would obtain , according to classical kinematics , the measured data waveform @xmath6 , where @xmath7 .
now , let as suppose that the fraction of the particles emitted , measured by the detector , due to some physical mechanism , is given by the ( non - negative ) `` transmission '' function @xmath8 where @xmath9 is the number of the particles emitted at the time instant @xmath10 and @xmath11 is the number of the particles detected , which have been sent at the same time instant @xmath12 .
in other words , in general , not all particles emitted are detected ( @xmath13 ) obvious , and moreover @xmath14 conceivable . for simplicity
, we will assume that the numbers of the particles , @xmath15 and @xmath16 , are so large that we are allowed to use a continuous approximation .
then , the transmission function @xmath17 is a ( continuous ) function satisfying the condition @xmath18 . to be able to draw conclusions from experimental data
, we should implement some statistical methodology .
to be so precise as possible , in our approach , we adopt the method of the mle , as the opera collaboration has done @xcite . in the framework of the mle
, we introduce the likelihood function @xmath19 .
the logarithm @xmath20 of @xmath19 is given by the formula @xmath21\equiv\sum_{j}\log\left[w(t_{j}+\delta t)\right],\label{eq : genlikelihood}\ ] ] where @xmath22 are the time instants corresponding to the measurement events at the detector , and the time deviation @xmath23 we are interested in ( see @xcite , for details ) provides the maximum of @xmath20 ( and also of @xmath19 ) .
as the numbers of the measured events @xmath22 are large , in our continuous approach , instead of summation in ( [ eq : genlikelihood ] ) , we should use integration with an appropriate integration measure @xmath24 , where @xmath25 represents the time distribution of the experimental events detected by the opera .
in fact , @xmath25 is determined by product of two factors .
the first factor , @xmath26 , is proportional to the number of the particles sent , i.e. @xmath27 , and the second one is proportional to the transmission function @xmath17 .
then , @xmath28 finally , @xmath29f(t)w(t)dt.\label{eq : conlikelihood}\ ] ] to demonstrate that our idea actually works , we propose a specific example .
the parameters of the example are so fitted that it yields the time deviation @xmath30 ( for comparison , the opera result is @xmath31 ) . for calculational simplicity ,
we assume the gaussian form of the pdf ( see the solid line in fig.[fig:2curves ] ) , @xmath32 as well as the gaussian form of of the transmission function , @xmath33,\label{eq : nontrivialtransmission}\ ] ] where the one tenth in front of the exponent is reminiscent of `` @xmath34 variation '' in @xcite .
the time distribution of the `` detected experimental data '' @xmath25 , corresponding to @xmath5 of the form ( [ eq : gauss ] ) and to the transmission function @xmath17 of the form ( [ eq : nontrivialtransmission ] ) is presented in fig.[fig:2curves ] by the dashed curve . ] in this ( doubly ) gaussian case it is even possible to solve the problem analytically ( see appendix ) , but for our purposes a numerical value will do .
it appears , that the maximum of ( [ eq : conlikelihood ] ) is attained for the time deviation @xmath35 where s.d .
means the standard deviation .
one can easily translate the dimensionless ( [ eq : numdelt ] ) into a dimensionfull entity . in the opera experiment @xmath36 , whereas in our example we can reasonably assume , for definiteness , that @xmath1 equals ( in dimensionless units , or in the units of the standard deviation ) twice the two standard deviations , i.e. @xmath37 ( see two vertical intervals in fig.[fig:2curves ] ) .
then , the dimensionfull time deviation corresponding to ( [ eq : numdelt ] ) is @xmath38 in particular , our analysis confirms the findings of @xcite that the time deviation @xmath23 is independent of the distance @xmath4 , but depends on the shape of the beam .
obviously , the numerical coincidence ( [ eq : unitdelt ] ) has only an illustrative purpose .
the non - zero time deviation @xmath23 directly implies a modified velocity @xmath39 according to the elementary formula @xmath40 in our model the sign of @xmath23 ( positive in our example ) depends on details of the form of the transmission function @xmath17 .
e.g. , reversing the sign in front of the exponent in ( [ eq : nontrivialtransmission ] ) reverses the sign of @xmath23 .
we would like to stress that our specific example is not supposed to mimic a real situation in the opera experiment , but only to demonstrate that it is , in principle , possible to easily come to its conclusions without any reference to superluminal particles and/or some exotic phenomena .
any considerations concerning a possible physical mechanism governing the time - dependence of the `` transmission '' function @xmath17 are outside the scope of our paper .
we can only speculate that @xmath14 could be a property of the source , or of the detector or it could follow from interactions in the earth s crust .
moreover , measurement errors , always encountered in real experiments , do not enter our considerations as they have nothing to do with the discussed effect .
it is also possible to intuitively explain the non - zero value of @xmath23 .
namely , deforming the shape of the pdf @xmath5 with an appropriate ( time asymmetric ) transmission function @xmath17 shifts a portion ( its `` upper '' part ) of the `` waveform '' @xmath5 forward or backward , yielding positive or negative @xmath23 , respectively , which is next erroneously interpreted by the method of the mle as a `` real '' time deviation . in a sense ,
our approach is a `` corpuscular '' analog of the well - known situation of `` superluminal '' velocity of light in material media @xcite , where to some extent , the role of @xmath17 plays dispersion . in conclusion , we would like to emphasize that there are two elements of our model , responsible for the curious effect of the superluminal velocity : the time - dependent transmission function @xmath17 , which should `` favor '' leading edges and `` disfavor '' trailing edges of the beam , and statistical methodology , erroneously assuming the method of the mle .
this is why it is conceivable that the paradoxical results of the opera collaboration could be possibly avoided upon another approach , e.g. an approach giving preference to the front velocity or to another statistical method .
thus , no superluminal particles are necessary to explain the `` superluminal '' velocities derived by the opera collaboration . moreover , independently of the further fate of the conclusions drawn by the opera collaboration ( acceptance or refutation ) , we have demonstrated that standard statistical methods , e.g. mle , should be used with great care , as they can provide erroneous output . therefore , the results of our paper do not rely on the final solution of the opera paradox . |
understanding charge transfer dynamics in dna is of fundamental importance for biochemistry and molecular electronics . @xcite it can generate chemical reactions over long distances and has biological relevance in the formation of oxidative damage .
charge transfer in dna also plays an important role in anti - cancer drugs,@xcite electrochemical readout of micro arrays and molecular - electronic devices .
whereas dna conductivity is still disputed , there is no doubt that electron transfer through dna double strands takes place .
@xcite it is generally believed that the electron or hole migrates by super - exchange tunneling or multistep hopping in a donor - bridge - acceptor system depending on the energy difference between the bridge and donor / acceptor @xcite and the fluctuation of the environment .
@xcite most of the current understanding of electron transfer is derived from a single - excess charge where only two states are needed to model a donor - acceptor system . at high temperatures
the transfer is governed by thermal activation .
the rate can be described in the non - adiabatic limit by the semi - classical marcus theory @xcite and is determined by the hopping integral @xmath1 , the energy difference between product and reactant states @xmath2 , the reorganization energy @xmath3 and the temperature @xmath4
. at low temperatures , however , the transfer rate does not vanish but rather remains finite and temperature independent : the transfer is governed by quantum fluctuations and involves nuclear tunneling.@xcite in this regime the transfer is modeled by an electronic two - level system@xcite coupled to a dissipative environment as in the spin - boson model@xcite which has to be treated quantum mechanically .
one can identify three regions of dynamics depending on the coupling to the bosonic bath : coherent ( damped oscillations ) , incoherent ( exponential relaxation ) transfer and self - trapping .
however , in many cases charge transfer involves multiple excess charges which are correlated .
@xcite their transfer characteristics is changed drastically in comparison to the case of a single excess charge and requires a description by more than two states.@xcite two charges can migrate as pairs , sequentially or can be self - trapped .
their dynamics are governed not only by the coupling to a dissipative bath but also by the local and non - local coulomb interactions ( @xmath5 and @xmath6 , respectively.)@xcite in this paper we study the charge transfer dynamics of two electrons or two holes in different dna base - pair dimers based on guanine / cytosine ( gc ) or adenine / thymine ( at ) pairs .
we have used the base - pair specific coulomb matrix elements as calculated with ab - initio methods.@xcite we assume that two charges are initially prepared on one base - pair monomer ( donor ) this could be achieved , e.g. , by voltage gating .
then , the transfer of those two charges occurs either in a concerted manner , where an electron pair or hole pair is transferred between the donor ( d ) and the acceptor ( a ) , or both charges migrate sequentially but correlated .
this transfer of the first electron facilitates or hinders the transfer of the second one due to the electronic correlations .
we find that depending on the coulomb matrix elements a second charge can lead even to self - trapping or , in the other extreme , to partially activationless transfer .
recently , conductance measurements of single dna molecules were performed in aqueous solution .
@xcite the solution preserves the native confirmation and the transport properties can be associated with charge transfer reactions at least for weak coupling to the leads.@xcite at voltages equal to the on - site coulomb interaction we expect that more than one excess charge is relevant for the charge transport through a molecular device . consequently , the transport characteristics is altered considerably .
@xcite furthermore , two - electron transfer in guanine due to strong oxidants has been discussed in connections to in vivo lesions .
@xcite our detailed study of the base - pair specific transfer characteristics of multiple excess charges for realistic systems reveals the connection between the microscopic parameters and the transfer efficiency .
previously , only an electronic model with density - density interactions coupled to an ohmic bath has been investigated.@xcite in this paper we extend the calculations to a more general electronic model which includes an additional bond - bond interaction ( pair - transfer ) @xmath0 previously neglected as well as a bond - charge interaction @xmath7 ( correlated hopping . )
these matrix elements have been calculated for different dna base - pairs@xcite using ab - initio methods and serve as input for our calculations .
we have found that the additional bond - bond interaction @xmath0 favors concerted electron transfer from the donor to the acceptor but can also lead to a change of the ground state of the dimer .
the parameter @xmath0 has a profound impact on the ground state of the electronic subsystem with two excess charges : it lifts the degeneracy of singlet and the triplet state .
the latter becomes the ground state for @xmath8 and a positive effective coulomb matrix element .
since the hamiltonian considered here is spin - conserving , the system can only decay into an metastable steady state in this regime starting with two excess charges on the donor in a spin singlet state .
another important application of charge transfer are dna biosensors . during the sensing process
, electron transfer is used to produce electronic signals .
@xcite in general the signal magnitude depends on the efficiency of the electron transfer through the dna as well as the number of electrons .
it has been suggested that the transfer of multiple electrons would enhance those signals.@xcite however , our work shows that the transfer rate for a single electron in systems with two excess charges depends on the environment as well as the coulomb matrix elements : it may be much larger or much smaller than the transfer rate of a single excess charge depending on the microscopic parameters . for temperatures lower than the activation energy the bath must be treated quantum - mechanically to calculate the time - dependent population probabilities and the transfer rates . in this regime
, we employ the time - dependent numerical renormalization group ( td - nrg ) approach@xcite whereas in the thermal activation regime we compute the transfer rates using the marcus theory.@xcite the paper is organized as follows . in section [ model ]
we introduce the microscopic model as well as both methods for high and low temperature regimes .
we present a detailed description of the single and multiple charge dynamics in sec .
[ singlecharge ] and sec .
[ multcharge ] , respectively .
a summary and conclusions of our results is given in sec .
we consider a base - pair dimer represented by a two - site system as donor ( d ) and acceptor ( a ) with a lumo ( lowest unoccupied molecular orbital ) and a homo ( highest occupied molecular orbital ) coupled to an aqueous environment ( bosonic bath ) as sketched in fig .
[ fig_schema_dna ] .
the hamiltonian of the coupled system consists of three parts @xmath9 where @xmath10 denotes the electronic part of the hamiltonian , @xmath11 the bath and @xmath12 the coupling between both subsystems .
each base pair monomer is modelled as a single site with one orbital ( either lumo or homo depending on excess holes or electrons ) .
the energy gap between homo and lumo is about @xmath13 .
@xcite @xmath10 has been derived in ref .
[ ] and corresponds to a general two - site extended hubbard model : @xmath14 where @xmath15 and @xmath16 denote annihilation and creation operators for fermions with spin @xmath17 in an orbital localized at site @xmath18 .
the first and second term are the on - site energy and the transfer integral or single - charge hopping for nearest neighbors , respectively .
the latter defines the kinetic energy .
the matrix elements @xmath19 account for the coulomb repulsion between two electrons .
the dominant interactions in the dna dimers@xcite are @xmath20 the first two matrix elements denote the on - site and nearest neighbor electron - electron interaction , respectively .
the parameter @xmath7 labels the bond - charge interaction and is often called correlated hopping since it depends on the density of the nearest neighbor .
the last term is the bond - bond interaction which induces the pair hopping and the exchange interaction . for the donor - acceptor pair ,
we arrive at the following hamiltonian : @xmath21 with the number operator @xmath22 .
the last two terms correspond to hopping of electron pairs . in our previous study,@xcite only the matrix elements @xmath5 and @xmath6 have been taken into account . here
, we will investigate ( i ) the influence of the additional matrix elements @xmath7 and @xmath0 onto the transfer dynamics using ( ii ) realistic values for @xmath23 as well as for the coupling to the environment as will be discussed in the following .
the bosonic bath comprises the internal vibrations and solvent dynamics and leads to dephasing and dissipation of the electronic system .
the hamiltonian @xmath11 @xmath24 models the free bosonic bath .
the displacements of the polaronic bath couple to the change of the electron density , @xcite @xmath25 and give rise to an attractive contribution to the on - site coulomb interactions .
the renormalized interaction is defined by @xmath26 which determines the energy difference between the ionic and covalent states for @xmath27 and includes the influence of the environment .
the hamiltonian of the isolated dimer conserves the total charge @xmath28 , where @xmath29 is the total number of electrons .
we consider the case @xmath30 ( two holes ( 2h ) in the homo ) , @xmath31 ( two electrons ( 2e ) in the lumo ) as well as @xmath32 and @xmath33 ( single hole in the homo and single electron in the lumo , respectively ) .
furthermore , the square of the total spin as well as its @xmath34-component @xmath35 are conserved by the hamiltonian eq .
( [ eq : model ] ) .
the subspace of two holes / electrons ( @xmath36 ) and @xmath37 ( for @xmath38 no transfer is possible ) is spanned by 4 states , two ionic states with two electrons / holes on the donor @xmath39 or on the acceptor @xmath40 and the two covalent states @xmath41 the latter states correspond to the singlet and triplet states , respectively .
@xmath42 is an eigenstate of @xmath43 with energy @xmath44 and is decoupled from the other three states displayed in fig . [ fig_schema_states ] .
it is the ground state for @xmath45 ( @xmath46 ) . in the subspace of @xmath47 ,
the total number of electrons per spin , @xmath48 , is constant , and the correlated hopping amplitude @xmath7 renormalizes the single particle hopping to @xmath49 . for one electron or hole ( @xmath50 ) and @xmath51
the model is equivalent to the spin - boson model .
the charge is able to migrate between the two states @xmath52 and @xmath53 with one electron / hole on the donor and acceptor , respectively .
the coupling between electronic subsystem and bosonic bath is specified by the bath spectral function@xcite @xmath54 @xmath55 characterizes the bath and the system - bath coupling and reflects intramolecular vibrations and the solvent ( water ) .
the spectral function can be calculated by molecular dynamics simulations@xcite for a two - sphere da model embedded in an aqueous medium and shows a power law behavior at low frequencies .
the asymptotic low - temperature behavior is determined by the low energy part of the spectrum . discarding high - energy details
, we use the standard parametrization@xcite @xmath56 where @xmath57 is a dimensionless constant which characterizes the dissipation strength and @xmath58 is the cutoff frequency of the spectral function . we investigate an ohmic bath ( @xmath59 ) and @xmath60 as an example for the super - ohmic dissipation.@xcite as an realistic estimate for the cut - off frequency @xmath58 we use @xmath61 which correspond to the frequency of the oh stretch vibrations at about @xmath62.@xcite the classical reorganization energy @xmath3 for the single - electron transfer @xmath63 , which measures the energy relaxation following a sudden electronic transition , @xmath64 has been determined@xcite for dna in water as approximately @xmath65 . therefore , we have used @xmath66 and @xmath67 for the ohmic bath and super - ohmic ( @xmath60 ) bath , respectively , for modeling the coupling between the dna dimers and the environment .
the reorganization energy for the transfer @xmath68 in the sector of two excess electrons is given by @xmath69 and for the transfer @xmath70 given by @xmath71 .
the single - charge hopping @xmath1 as well as pair - hopping @xmath0 are renormalized by the bosonic bath . because only the ionic states are coupled to the bath ,
the pair - hopping is much stronger reduced than @xmath1 which connects ionic and covalent states .
the exchange interaction , however , which is also proportional to @xmath0 remains unaltered by the bath since it does not change the occupancy of the orbitals . in the following sections , the four time - dependent population probabilities , defined as the expectation values of the operators
@xmath72 are calculated .
they measure the single and double occupancy on the donor and the acceptor site ( ionic states ) as well as the equally populated states ( covalent states ) .
the marcus theory of charge transfer is connected to the quantum mechanical treatment of the bosonic bath in hamiltonian ( [ eqn_gen ] ) by a mean - field decoupling of the bosonic bath modes from the electronic degrees of freedom .
the local displacement @xmath73 is replaced by its expectation value @xmath74 , which is also called the reaction coordinate .
the resulting potential surfaces @xmath75 for the two states @xmath76 and @xmath77 in the @xmath50 sector are plotted schematically in the upper panel of fig .
[ fig_schema ] .
the potential surfaces @xmath75 of the four states in the @xmath36 sector , @xmath78 , @xmath79 and @xmath80 or @xmath81 are depicted in the lower panel of fig . [ fig_schema ] .
the single - electron transfer from @xmath82 to @xmath83 is determined by the hopping @xmath1 and the coupling constant @xmath57 ( or the reorganization energy @xmath3 ) . for @xmath36 ,
the transfer additionally depends on @xmath84 and @xmath0 . moreover , the effective bath coupling constant is four times larger due to a factor 2 in the coupling term . as a consequence
one would expect a slower transfer compared to the @xmath50 case .
however , the transfer of two charges is strongly affected by the energy difference between the ionic and covalent singlet states @xmath85
. the transfer of one charge can be even activationless if the difference @xmath85 is equal to the reorganization energy @xmath86 as indicated in the lower panel of fig .
[ fig_schema ] at finite values of @xmath74 . to address the thermal activation regime where the temperatures are larger than the activation energy ( @xmath87 ) ,
the dissipative environment is treated classically , and the rates are obtained from fermi s golden rule . generally the marcus rate is given by @xmath88 where in the franck - condon integral @xmath89 the energy difference between state @xmath18 and @xmath90 , @xmath91 , and the corresponding reorganization energy @xmath92 as described after eq .
( [ reo ] ) enters .
the population probabilities are determined by kinetic equations @xmath93 where @xmath94 is the probability that the system is found in state @xmath95 and @xmath96 is the marcus rate of the transition from state @xmath97 to state @xmath95 . at low temperatures ( @xmath98 )
the bosonic bath must be treated quantum mechanically . in this regime ,
coherent quantum oscillations may be present , and we have calculated the time - dependent population probabilities using the td - nrg .
@xcite for the models studied in this work , the impurity couples to a bosonic bath , for which the standard nrg approach has to be modified as described in refs . .
the basic features of the bosonic nrg are as follows : ( i ) the logarithmic discretization of the bath spectral function @xmath55 in intervals @xmath99 $ ] , with @xmath100 and @xmath101 the nrg discretization parameter ( all the results shown in this paper are calculated using @xmath102 ) ; within each of these intervals only one bosonic degree of freedom is retained as a representative of the continuous set of degrees of freedom .
( ii ) the mapping of the resulting hamiltonian onto a semi - infinite chain . (
iii ) the iterative diagonalization of the chain - hamiltonian via successively adding one site to the chain . for the results shown here
, we use @xmath103 bosonic basis states for each chain link and retain the @xmath104 lowest eigenstates after each iteration .
the set of all discarded states eliminated during the nrg procedure form a complete basis of the semi - infinite chain , which is also an approximate eigenbasis of the hamiltonian . in one nrg
run we compute the initial density matrix,@xcite in the second run the approximate eigenbasis of the hamiltonian eq .
( [ eq : model ] ) .
since the hamiltonian usually does not commute with the initial density matrix , the time evolution of the density matrix governs the time dependence of all expectation values .
the real - time dynamics of the population probabilities has been calculated by summing over all contributions from the sequence of reduced density matrices which contain the information of decoherence and dissipation at a given energy scale @xmath105 . to simulate a bath continuum we have averaged over different bath discretizations ( we used eight different realizations ) . for more details on the approach
see refs .
before analyzing the transfer of multiple excess charges we briefly review the single - charge transfer .
the two - state system is prepared such that the donor is occupied by one electron / hole at @xmath106 .
the transfer of a single charge is not effected by the coulomb matrix elements and eq .
( [ eq : hel ] ) reduces to the first two terms which is equivalent to the spin - boson model .
we use a typical hopping matrix element@xmath107 as obtained by ab - inito methods@xcite and @xmath66 as estimated above . in the subspace of a single charge .
the system is prepared such that at time t = 0 the charge occupies the donor .
the results are shown for @xmath108 ( blue ) , a decoupled charge ( dotted line ) for @xmath46 .
the inset shows a comparison with simple second - order perturbation theory in @xmath57 which is only valid for very short time scales .
, scaledwidth=48.0% ] the population of the donor @xmath109 calculated with the td - nrg for @xmath110 is shown in fig .
[ fig_single ] .
after about 200 fs the system relaxes to its ground state @xmath111 .
the frequency of the oscillations increased due to the strong coupling to the bosonic bath , which is counterintuitive at first .
usually , a reduction of the frequency in a damped harmonic oscillator is expected due to friction .
this has also been found in the spin - boson model in the limit@xcite of @xmath112 , when the bare oscillation frequency is much smaller than the bath cutoff energy . in that limit , all coherent oscillations are suppressed once @xmath113 . by comparing the result to the dynamics of a decoupled spin ( @xmath46 , dotted line )
we find indeed a slower spin - decay at ultra - short time scales of @xmath114fs see inset of fig .
[ fig_single ]
. however , the realistic dna parameters yield @xmath115 : the bare frequency of oscillation is of the order of the cutoff of the bosonic bath .
now , band edge effects become important , and the strong coupling to the bath ( @xmath66 ) lead to a level splitting in the first rg steps ( level repulsion ) . some local excitation energies
are shifted above the cutoff @xmath58 ; other contributions are renormalized further down to low frequencies in each rg step .
this is reflected in the fourier transformation of @xmath116 not shown here which peaks at a frequency of approximatly @xmath117 ; however , it also comprises a second broad contribution which has its maximum at a reduced frequency of @xmath118 . the analytic solution obtained in second - order perturbation theory in the bath - coupling @xmath119 agrees very well with the ultra - short dynamics of td - nrg .
this is depicted in the inset of fig .
[ fig_single ] : the perturbative and the td - nrg solution are identical for @xmath120fs for ohmic and super - ohmic baths not shown here .
in fact the perturbative solution shows even a stronger renormalization of the frequencies than the td - nrg .
however , the solution becomes less reliable for @xmath121fs due to the strong coupling effects ignored in the perturbation theory , which is only valid at very short time scales . the origin of this enhancement can be traced to a factor @xmath122 in the analytic integral kernel which causes a sign change of the contribution for @xmath123 to @xmath124 .
the position of @xmath1 relative to the cutoff @xmath58 and the power @xmath95 of @xmath55 determines the sign of the frequency renormalization .
the spin - boson model is usually only investigated in the limit @xmath112 .
the dna dimer system , however , appears to be in the complementary regime where @xmath125 . in the nrg as well as in the analytical solution
, we find coherent oscillations which prevail well above @xmath67 ( not shown here ) .
.coulomb interaction parameter taken from ref . [ ] .
the other parameters are @xmath126 , @xmath127 , @xmath128 , @xmath129 , @xmath130 and @xmath131 .
we vary @xmath0 between @xmath132 and @xmath133 . [ cols="^,^",options="header " , ]
to investigate the influence of the diagonal and the additional off - diagonal coulomb interactions on the transfer of two charges , we initially prepare the system such that two electrons / holes are localized on the donor at time @xmath106 .
this can be achieved experimentally , e.g. , through radiolytic or photolytic excitation , strong oxidation or voltage gating .
the hopping matrix element @xmath1 is renormalized to @xmath84 in the presence of a finite correlated hopping matrix element @xmath7 in the subspace of two charges . the base - pair specific coulomb interaction used in the following are stated in table i and are taken from ref .
since we are mainly interested in the influence of the coulomb interactions onto the transfer rates , we set the hopping matrix element to a typical value@xcite @xmath126 and a correlated hopping parameter @xmath127 for all base - pairs:@xcite the dynamics of a single charge is governed by the energy scale of @xmath134 .
it has been argued@xcite that the bond - bond interaction @xmath0 is either small or negligible .
therefore , we have investigated the influence of @xmath0 onto the electron transfer dynamics by varying @xmath0 between @xmath135 and @xmath136 . in the next section ,
we begin with @xmath27 . as defined in eq .
( [ steady ] ) at high temperatures ( black line ) and low temperatures obtained from nrg ( red points ) , as a function of @xmath137 .
the region of sequential and concerted pair transfer is indicated . for @xmath138 the pair transfer
is followed by a very slow transfer to the covalent state . at low temperatures three regions
are present : self trapping for an ohmic bath ( left of the dotted line ) , sequential transfer ( white region ) and pair transfer combined with slow sequential transfer ( shaded region for @xmath139 ) .
the low - temperature rates were obtained by fitting the population probabilities to an exponential function .
the points represent the rates for ( 1 ) at - at(2h ) , ( 2 ) gc - gc ( 2e ) and ( 3 ) gc - gc ( 2h ) . , scaledwidth=52.0% ] the marcus theory is applicable in the temperature regime where the bosonic bath can be replaced by a single reaction coordinate , and the physics is governed by thermal activation .
the theory predicts an exponential dependency of the transfer rates on the energy difference @xmath140 .
the concerted transfer of both electrons becomes relevant once the transfer rate between the ionic states exceeds the rate between ionic and covalent states .
the latter regime , however , is excluded by the parameters stated in tab .
[ table : i ] in the dna dimers considered in this paper .
the rate @xmath141 that at least one electron is transferred from the donor to the acceptor in the thermal activation regime is defined as @xmath142 and is plotted versus @xmath137 as a solid line in fig .
[ u4double ] .
the maximum at @xmath143 marks the optimal or activation - less regime and separates the normal regime , @xmath144 , from the inverted regime , @xmath145 . for @xmath146 and @xmath147 the pair transfer is the main process . for @xmath139
it is accompanied by a slow sequential transfer to the covalent state .
( starts at one ) and the covalent states @xmath148 ( starts at zero ) for at - at ( 2h ) ( red dotted line ) , gc - gc ( 2e ) ( black solid line ) and for gc - gc ( 2h ) ( blue dashed line ) as a function of time for an ohmic bath ( @xmath59 ) and @xmath27 using the td - nrg . at t=0
two electrons are localized at the donor .
two excess electrons on at - at are self - trapped and not displayed .
, scaledwidth=50.0% ] using the parameters of tab .
[ table : i ] , we note that the transfer occurs activationless for two holes in a gc - gc dimer because @xmath143 . the estimate of the transfer rate via eq .
( [ fgr ] ) at room temperature translates into an ultra - fast characteristic time scale of about 10fs . for two electrons on the gc - gc dimer ,
however , the bare @xmath5 is smaller , and the condition @xmath144 results in a smaller rate .
the transfer rates for two electrons or holes on an at - at dimer are exponentially suppressed and several orders of magnitude smaller than those on a gc - gc dimer . at low temperatures the mean - field decoupling of the bath modes
is not applicable any longer .
the quantum mechanical nature of bath modes must be taken into account , and nuclear tunneling is important : now coherent collective oscillations of the electrons become possible . as a consequence , the transfer characteristics
is changed , and the rates can become larger than in the thermal activation regime .
we employ the td - nrg to calculate the population probabilities @xmath149 and @xmath150 at a temperature @xmath151 for the parameters of the different base pairs coupled to an ohmic bath .
the real - time dynamics for short - time scales is plotted in fig .
[ fig_quantumu ] .
the transfer shows an incoherent behavior with a few remains of oscillations .
the decay rates are evaluated by fitting the result to the kinetic equations eq .
( [ kinetic ] ) . these fitted rates
are added to fig .
[ u4double ] as ( red ) points and are also depicted in fig .
[ fig_ratew ] .
the largest rate is found for gc - gc dimer with two electrons , @xmath152 , and the equilibrium is reached after about 40 fs , while the rate for gc - gc ( 2h ) is 25 times smaller corresponding to a characteristic time scale of @xmath153fs . for @xmath110 , the optimal regime
is already reached for @xmath144 and , therefore , earlier than predicted by the marcus theory .
such behavior has also been reported in the biased spin - boson model at low temperatures .
@xcite nevertheless , the marcus rates and the td - nrg rates are of the same order of magnitude in the optimal regime . using the classification of the marcus theory , the transfer in gc - gc ( 2e ) then would be already in the inverted region at low temperatures .
we would like to emphasize that these transfer rates are significantly larger than those reported for the single charge subspace .
the incoherent relaxation occurs on time scales between a few fs and 50fs while the envelope function of the single charge dynamics decays on times scales larger than 200fs as can be seen in fig .
[ fig_single ] . for at - at ( 2e ) the two electrons on the donor
are self - trapped in an ohmic bath . for at - at ( 2h ) one hole is transferred with a characteristic time scale of 146fs .
the transfer is much faster than in the thermal activation regime where the rate is exponentially suppressed . a summary of the td - nrg results is included in fig .
[ u4double ] .
we can identify three regions .
below @xmath154 ( left of the dotted line ) the system is localized@xcite for an ohmic bath , and the charges are self - trapped .
right of the dotted line
white region we found sequential transfer . with increasing @xmath137 more coherent transfer is observed , and the fraction of pair transfer is increasing . at low temperatures
pair transfer would only be present when @xmath137 exceeds @xmath13 for the given reorganization energy @xmath155 in an ohmic bath .
the charges remain trapped at the donor in the at - at ( 2e ) dimer coupled to an ohmic bath since @xmath156 lies in the localized regime .
this self - trapping is lifted in a super - ohmic bath : pair transfer is observed in super - ohmic baths as shown in fig .
[ fintepairw ] .
the bond - bond interaction has two effects : first it induces an effective pair hopping between @xmath157 and @xmath158 and second it lifts the degeneracy between the triplet and singlet states @xmath42 and @xmath159 , respectively .
the triplet state , which is decoupled from the other states , becomes the ground state for a finite @xmath8 and @xmath139 . since spin conservation prohibits its population when starting initially with two fermions on the donor site , the transfer involves only the covalent singlet state @xmath159 .
its dynamics is governed by the energy difference @xmath160 between @xmath161 and @xmath159 .
for a small @xmath162 as in at - at(2h ) the rate predicted by the marcus theory decreases with increasing @xmath0 because the energy difference changes its sign from @xmath163 to @xmath164 .
these two regimes are shown schematically in fig .
[ fig_marcusw ] for @xmath165 .
the dependency of the rate @xmath141 on @xmath0 obtained at low temperatures using the td - nrg follows the prediction of the marcus theory as shown in fig .
[ fig_ratew ] as ( red ) circles .
as discussed above , the gc - gc ( 2h ) dimer is located in the inverted region as a consequence of the large @xmath166 .
an increasing @xmath0 reduces the energy difference between the initial state and @xmath159 leading to an increase of @xmath141 as depicted in fig .
[ fig_ratew ] ( squares ) .
the transfer in gc - gc ( 2e ) shows almost no dependence on @xmath0 ( fig .
[ fig_ratew ] , triangles ) since it is located in the activationless regime .
an electron - pair in an at - at dimer remains self - trapped for @xmath167 in an ohmic bath . in a super - ohmic bath ( @xmath168 ) , however , coherent transfer is found at low temperatures .
the electrons oscillate coherently as pairs between the donor and the acceptor while the high lying singlet covalent state @xmath159 is only virtually occupied .
the oscillation frequency is approximately given by @xmath169 where @xmath170 is the renormalized pair - hopping in the presence of the bosonic bath : this frequency increases with increasing @xmath0 .
the occupation probabilities show slow oscillations with a period of the order of @xmath171 ns as depicted in fig .
[ fintepairw ] .
note that the time scale is multiple orders of magnitude larger compared to the fs time - scales of the gc - gc dimer .
the relaxation rate of the envelope function corresponds to a significantly longer time scale @xmath172 .
we have calculated the real - time dynamics of a redox system with multiple charges . at temperatures lower than the activation energy we have used the time - dependent numerical renormalization group approach ; at temperatures larger than the activation energy when the transfer is governed by thermal activation we have employed the marcus theory .
four different dna dimers have been considered : gc - gc and at - at with initially two excess electrons or holes on the donor site .
all results are obtained for realistic matrix elements obtained by ab - initio methods.@xcite the charge transfer characteristics of a system with two excess charges deviates substantially from those of a single excess charge .
the transfer occurs correlated and depends strongly on diagonal and off - diagonal coulomb matrix elements .
we predict that an additional charge in dna sensors or dna devices will show base - pair specific signal amplification or attenuation .
we have calculated the rate @xmath141 that at least one charge is transferred from the donor to the acceptor .
the time - dependent population probabilities of the singlet covalent state ( @xmath159 ) and two ionic states ( @xmath173 , @xmath174 ) depend on the difference between the effective diagonal coulomb interaction @xmath137 and the off - diagonal bond - bond interaction @xmath0 .
we have found a qualitative agreement between the marcus theory a high - temperatures and the td - nrg at low temperatures .
the transfer rate @xmath141 shows a dome - shape as function of @xmath137 in the marcus theory as depicted in fig .
[ u4double ] .
it is slightly shifted to lower values of @xmath137 and broadened at low temperatures as calculated via td - nrg .
however , self - trapping is only found in the td - nrg .
the correct application of the marcus theory requires the restriction to transitions compatible with spin - conservation : the triplet state @xmath42 remains inaccessible since we always start with two charges in a spin - singlet state on the donor .
the maximal transfer rate @xmath141 , defining the activation - less regime , is significantly larger in the subspace of two excess charges compared to the subspace of a single excess charge . at low temperatures ,
the largest rate is found for two excess electrons in gc - gc and for two excess - holes at room temperature .
electron - pairs initially on the donor in an at - at dimer are either self - trapped ( ohmic bath ) or are slowly oscillating as pairs between d and a ( super - ohmic bath ) at low temperatures .
the dependency of the transfer rate on a finite but small bond - bond interaction @xmath0 can be understood in terms of a simple shift of @xmath175 : the rate @xmath141 now peaks at @xmath176 .
consequently , the transfer rate increases for two holes on gc - gc which is located in the inverted regime while @xmath141 decreases for two holes on at - at with increasing @xmath0 .
it remains constant for two electrons on gc - gc since the rate appears to be rather insensitive to @xmath0 in the activation - less regime .
furthermore , the presence of the additional bond - bond interaction @xmath0 splits the degeneracy of the singlet and the triplet state , @xmath159 and @xmath42 , respectively . without any further spin relaxation process
, the system can not reach the electronic thermodynamic ground state @xmath42 for @xmath8 and @xmath177 since the triplet state remains unpopulated at all times .
the system remains trapped in an intermediate metastable state whose life - time will depend on the spin relaxation mechanism not considered here .
the transfer characteristics of two charges in a dna dimer depends strongly on the base pair specific coulomb interactions .
these effects will become important when such redox systems are weakly contacted .
they strongly influence the current - voltage characteristics once multiple excess charges are relevant at higher source - drain voltages.@xcite
this research was supported by the state of lower- saxony and the volkswagen foundation ( st , gz ) and by the dfg through sfb 608 ( rb ) and an275/6 - 2 ( fba ) . we acknowledge helpful discussions with e. starikow , a. nitzan and a. schiller . | we investigate the charge transfer characteristics of one and two excess charges in a dna base - pair dimer using a model hamiltonian approach .
the electron part comprises diagonal and off - diagonal coulomb matrix elements such a correlated hopping and the bond - bond interaction , which were recently calculated by starikov [ e. b. starikov , phil . mag .
lett . * 83 * , 699 ( 2003 ) ] for different dna dimers .
the electronic degrees of freedom are coupled to an ohmic or a super - ohmic bath serving as dissipative environment .
we employ the numerical renormalization group method in the nuclear tunneling regime and compare the results to marcus theory for the thermal activation regime . for realistic parameters ,
the rate that at least one charge is transferred from the donor to the acceptor in the subspace of two excess electrons significantly exceeds the rate in the single charge sector .
moreover , the dynamics is strongly influenced by the coulomb matrix elements .
we find sequential and pair transfer as well as a regime where both charges remain self - trapped .
the transfer rate reaches its maximum when the difference of the on - site and inter - site coulomb matrix element is equal to the reorganization energy which is the case in a gc - gc dimer .
charge transfer is completely suppressed for two excess electrons in at - at in an ohmic bath and replaced by damped coherent electron - pair oscillations in a super - ohmic bath . a finite bond - bond interaction @xmath0
alters the transfer rate : it increases as function of @xmath0 when the effective coulomb repulsion exceeds the reorganization energy ( inverted regime ) and decreases for smaller coulomb repulsion . |
with the static color charges one observes on the lattice a color - electric flux tube @xcite , which , if the distance between the static quarks is large , can be approximated as a string .
this string is nondynamical ( in the sense that its ends are fixed ) . in the light quark systems
a crucial aspect is a relativistic motion of quarks at the ends of a possible string .
there is no consistent theory of the dynamical qcd string with quarks at the ends .
it is even apriori unclear whether such a picture has something to do with reality . in this case
the chiral symmetry as well as its dynamical braking should be relevant .
both confinement and chiral symmetry breaking dynamics are important for the hadronic mass generation .
their interplay is responsible for a rather complicated structure of the hadron spectra in the light quark sector .
a large degeneracy is seen in the highly excited mesons @xcite , which is , however , absent in the observed spectrum below 1.8 gev .
the low - lying hadron spectra should be strongly affected by the chiral symmetry breaking dynamics . in order to disentangle the confinement physics from the chiral symmetry breaking dynamics we remove on the lattice from the valence quark
propagators the lowest - lying quasi - zero modes of the dirac operator keeping at the same time the gluon gauge configurations intact @xcite .
indeed , the quark condensate of the vacuum is related to a density of the lowest quasi - zero eigenmodes of the dirac operator @xcite : @xmath2 we subtract from the valence quark propagators their lowest - lying chiral modes , which are a tiny part of the full amount of modes , @xmath3 where @xmath4 and @xmath5 are the eigenvalues and the corresponding eigenvectors of the dirac operator .
given these reduced quark propagators we study the existence of hadrons and their masses upon reduction of the lowest - lying modes responsible for the chiral symmetry breaking .
in contrast to refs .
@xcite , where a chirally not invariant wilson lattice dirac operator was employed , we adopt now @xcite a manifestly chiral - invariant overlap dirac operator @xcite .
the quark propagators were generously provided by the jlqcd collaboration @xcite .
for some technical details of the work we refer the reader to ref .
at this first stage of the study we have investigated the ground state and the excited states of all possible @xmath6 isovector @xmath7 mesons , i.e. , @xmath8 , @xmath9 , @xmath10 .
an exponential decay of the correlation function is interpreted as a physical state and its mass is extracted .
the quality of the exponential decay improves with the reduction of the lowest modes . the evolution of masses of the ground and excited states upon reduction of the low - lying modes is shown in fig . 1 .
.the complete set of @xmath11 @xmath7 states classified according to @xmath0 .
the symbol @xmath12 indicates the states belonging to the same representation @xmath13 that must be degenerate in the chirally symmetric world . [ cols="^,^",options="header " , ] all possible multiplets of the @xmath0 group for the @xmath7 mesons are shown in table 1 @xcite . in the chirally symmetric world
there must be two independent @xmath15-mesons that belong to different chiral representations .
the first one is a member of the @xmath16 representation .
it can be created from the vacuum only by the operators that have the same chiral structure , e.g. by the vector - isovector current @xmath17 .
its chiral partner is the axial vector meson , @xmath18 , that is created by the axial - vector isovector current .
when chiral symmetry is restored this @xmath15-meson must be degenerate with the @xmath18 state .
another @xmath15-meson , along with its chiral partner @xmath19 , is a member of the @xmath20 representation .
it can be created only by the operators that have the same chiral structure , e.g. by @xmath21 . in the real world
( i.e. with broken chiral symmetry ) each @xmath15-state ( i.e. @xmath15 and @xmath22 ) is a mixture of these two representations and they are well split . upon subtraction of 10 lowest modes
we observe two independent degenerate @xmath15-mesons .
one of them couples only to the vector current and does not couple to the @xmath21 operator .
the other @xmath15 meson - the other way around .
this means that one of these degenerate @xmath15-states belongs to the @xmath16 multiplet and the other one is a member of the @xmath20 representation .
a degeneracy of the @xmath23 @xmath15-meson with the @xmath18 meson is a clear signal of the chiral @xmath0 restoration .
consequently , a similar degeneracy should be observed in all other chiral pairs from table 1 .
the @xmath1 symmetry transforms the @xmath24 state into the @xmath20 @xmath15-meson @xcite .
their degeneracy indicates a restoration of the @xmath1 symmetry .
we conclude that simultaneously both @xmath0 and @xmath1 symmetries get restored .
the restored @xmath25 symmetry requires a degeneracy of four mesons that belong to @xmath26 and @xmath20 chiral multiplets @xcite , see table [ tab : t1 ] .
this symmetry does not require , however , a degeneracy of these four states with other mesons , in particular with @xmath18 and its chiral partner @xmath15 .
we clearly see the latter degeneracy .
this implies that there is some higher symmetry , that includes @xmath27 as a subgroup .
this higher symmetry requires a degeneracy of all eight mesons from the table [ tab : t1 ] .
all these eight mesons can be combined to a reducible @xmath6 representation , which is a product of two fundamental chiral quark representations @xcite : @xmath28 \times [ ( 0,1/2)+(1/2,0 ) ] = \hspace{15 mm } \nonumber\\ ( 0,0 ) + ( 1/2,1/2)_a + ( 1/2,1/2)_b + [ ( 0,1)+ ( 1,0)]\;.\hspace{4mm}\end{aligned}\ ] ] they exhaust all possible chiralities of quarks and antiquarks , i.e. , their spin orientations , as well as possible spatial and charge parities for non - exotic mesons .
the observed degeneracy of all these eight mesons suggests that the higher symmetry , mentioned above , should combine the @xmath1 and the @xmath29 rotations in the isospin space with the @xmath30 spin - symmetry .
the latter one is due to independence of the energy on the orientations of the quark spins .
the higher symmetry group that combines all eight mesons from the table 1 into one multiplet of dimension 16 should be @xmath31 .
the quark - spin independence of the energy levels implies that there are no magnetic interactions in the system , i.e. , the spin - orbit force , the color - magnetic ( hyperfine ) and tensor interactions are absent @xcite .
the energy of the system is entirely due to interactions of the color charges via the color - electric field and due to a relativistic motion of the system .
we interpret ( or define ) such a system as a dynamical qcd string .
note a significant qualitative difference with the case of a motion of an electrically charged fermion in a static electric field or with a relative motion of two fermions with the electric charge .
in the latter cases there exist spin - orbit and spin - spin forces that are a manifestation of the magnetic interaction in the system . in our case
such a magnetic interaction is absent , which implies that both quarks are at rest with respect to the electric field ( that moves together with quarks ) .
it is this circumstance that suggests to interpret ( define ) our system as a dynamical qcd string .
the observed radial levels at the truncation energy 65 mev , at which we see the onset of the symmetry , are approximately equidistant and can be described through the simple relation @xmath32 the extracted value of the radial string excitation quantum amounts to @xmath33 mev . at the moment
we can not exclude , however , the quadratic relation @xmath34 , because the excited level can be shifted up due to rather small finite lattice volume .
there is an interesting aspect of the dynamical qcd string that crucially distinguishes it from the nambu - goto open string .
the energy of the the nambu - goto open bosonic string is determined by its orbital angular momentum @xmath35 , @xmath36 .
for the dynamical qcd string that contains chiral quarks at the ends the orbital angular momentum @xmath35 of the relative motion is not conserved @xcite .
for instance , two orthogonal @xmath15-mesons at the same energy level are represented by the mutually orthogonal fixed superpositions of the @xmath37- and @xmath38-waves .
@xmath39 i am very grateful to m. denissenya and c. b. lang for our common efforts in lattice simulations .
the jlqcd collaboration is acknowledged for their suggestion to use their overlap gauge configurations and quark propagators .
this work is partially supported by the austrian science fund ( fwf ) through the grant p26627-n16 .
g. s. bali , phys .
* 343 * , 1 ( 2001 ) .
l. y. glozman , phys .
b * 587 * , 69 ( 2004 ) .
l. y. glozman , phys .
rept . * 444 * , 1 ( 2007 ) . c. b. lang and m. schrck , phys .
d * 84 * , 087704 ( 2011 ) .
l. y. glozman , c. b. lang and m. schrck , phys .
d * 86 * , 014507 ( 2012 ) .
m. denissenya , l. y. glozman and c. b. lang , phys .
d * 89 * , 077502 ( 2014 ) . | mesons constructed from the quark propagators without the lowest - lying eigenmodes of the dirac operator reveal not only restored @xmath0 chiral and @xmath1 symmetries , but actually a higher symmetry .
all possible chiral and @xmath1 multiplets for the states of the same spin are degenerate , i.e. , the energy of the observed quantum levels does not depend on the spin orientation of quarks in the system and their parities .
the quark - spin independence of the energy levels implies absence of the magnetic interactions in the system .
the ultrarelativistic quark - antiquark system with only the color - electric interactions can be interpreted ( or defined ) as a dynamical qcd string . |
the study of computational devices based upon quantum mechanics , _ i.e. _ quantum computation , has drawn the attention of researchers in the last few decades @xcite .
the recent advances in technology that allow to construct and preserve almost perfectly quantum states , have opened the possibility of building useful quantum computing devices . however , relatively few quantum algorithms that outperform classical ones have been found @xcite .
the classical random walk is an example of stochastic motion that has found classical applications in many fields .
the quantum version of this problem has several features which are markedly different from the classical walk @xcite . as some classical algorithms
are based on random walks , it seems natural to ask whether quantum random walks might be a useful tool for quantum computation @xcite .
the discrete time quantum walk on the line was introduced as a generalization of the classical random walk to the quantum world . here
, we will focus on the discrete time quantum random walk on the line from a new perspective which emphasizes the role of coherence as the physical reason behind the striking differences found in the quantum version of the walk .
we first briefly introduce the basic notions and notation relative to the discrete time quantum walk on the line . consider a particle that can move freely over a series of interconnected sites . the discrete quantum walk on the line is implemented by introducing an additional degree of freedom ,
the chirality , which can take two values : `` left '' or `` right '' , @xmath0 or @xmath1 , respectively .
this is the quantum analog of the classical decision of the random walker . at every time step ,
a rotation ( or , more generally , a unitary transformation ) of the chirality takes place and the particle moves according to its final chirality state .
the global hilbert space of the system is the tensor product @xmath2 where the hilbert space associated to the motion on the line is @xmath3 and the chirality hilbert space is @xmath4 .
if one is only interested in the properties of the probability distribution , it has been claimed @xcite that it suffices to consider unitary transformations which can be expressed in terms of a single real angular parameter @xmath5 .
let us call the operators that translate the walker one site to the left ( right ) on the line in @xmath3 as @xmath6 ( @xmath7 ) , and @xmath8 and @xmath9 the chirality projector operators in @xmath4 .
we consider transformations of the form , @xmath10 where @xmath11 , @xmath12 is the identity operator in @xmath3 , and @xmath13 and @xmath14 are pauli matrices acting in @xmath4 .
the unitary operator @xmath15 evolves the state by one time step , @xmath16 one of the most remarkable characteristics of the quantum walk on the line is that it spreads over the line faster than its classical counterpart . in this work
, we apply a general approach which leads to an physical insight of why this is so . to do this ,
we rewrite the evolution equation as the sum of two separate terms , one responsible for the classical like diffusion and the other for the quantum coherence @xcite .
as we shall see , the terms responsible for the diffusion obey a master equation , as is typical of markovian processes , while the other includes the interference terms needed to preserve the unitary character of the quantum evolution . in section [ sec : derivation ] , we review the decomposition of the evolution in these two terms . in section [ sec : moments ] , we obtain analytical expressions for the first and second moments of the probability distribution for these terms , in the case of the quantum random walker . finally , in section [ sec : conclusions ] , we discuss our conclusions .
in a recent work @xcite , we have shown in detail how a unitary quantum mechanical evolution can be separated into markovian and interference terms . this approach provides a new intuitive framework which proves useful for analyzing the behavior of quantum systems in which decoherence plays a central role .
it is particularly suited to describe the evolution of quantum systems which have classically diffusive counterparts .
the unitary evolution associated to @xmath15 can be decomposed into a markovian term and an interference term .
we begin by expressing the wave vector , as the spinor @xmath17 where we have associated the upper ( lower ) component to the left ( right ) chirality and the states @xmath18 are eigenstates of the position operator corresponding to the site @xmath19 on the line . the unitary evolution for @xmath20 , corresponding to eq.([evol1 ] ) ,
can then be written as the map @xmath21 note that for the particular case @xmath22 , the usual hadamard walk on the line is obtained .
the cases @xmath23 and @xmath24 are trivial motions not considered here .
we define the left - distribution of position as @xmath25 and the right - distribution as @xmath26 . then , the probability distribution for the position is @xmath27 and these distributions satisfy the map @xmath28 where the interference term in eq.([mapa1 ] ) has been renamed as @xmath29 $ ] with @xmath30 indicating the real part of @xmath31 .
it will prove useful to write equations ( [ mapa1 ] ) in the form@xmath32 where @xmath33 take the two values of chirality @xmath34 , @xmath35 and @xmath36 .
the transition probabilities @xmath37 are defined as @xmath38 note that these transition probabilities satisfy the necessary requirements @xmath39 and @xmath40 .
now it is clear that if the interference term @xmath41 in eq.([master ] ) can be neglected , the time evolution of the occupation probability is described by a markovian process in which the transition probability @xmath42 in a time @xmath43 , is given by @xmath37 .
as is characteristic of markovian processes , the new position and chirality depend only on the previous values for position and chirality .
since the chirality is an auxiliary dimension introduced to implement the quantum walk , we focus on the evolution of the position distribution for the particle , @xmath44 , which can be obtained from eqs.([mapa1 ] ) as , @xmath45 \cos^{2}\theta -p_{i}(t-1)\cos2\theta\nonumber\\ & \qquad\qquad\qquad\qquad\qquad+\left [ \beta_{i+1}(t)-\beta_{i-1}(t)\right ] \sin2\theta . \label{probabilidad}\ ] ] we have used the fact that @xmath46 , a relation that is a consequence of the map ( [ mapa ] ) .
note that if the interference terms are neglected in eq.([probabilidad ] ) , the resulting evolution is markovian in two time steps . in the general case , eq.([probabilidad ] )
does not possess a continuum equivalent , even in the absence of interference terms .
however , there are two special cases where it is possible to obtain continuum limits@xcite .
one is if we take the increments in space ( @xmath47 ) and the increments in time ( @xmath48 ) in such a way that the velocity @xmath49 and the ratio @xmath50 are kept constant when @xmath51 ; in this case one obtains the telegraphist s equation @xcite .
@xmath52 \label{probabilidad1}\ ] ] in the second case , @xmath53 is kept constant as @xmath51 and @xmath5 can take any value
. then the more familiar classical diffusion equation is obtained .
@xmath54 where , in the previous two equations , @xmath55 is the diffusion coefficient , @xmath56 note that this expression reduces to the classical value , @xmath57 , for the hadamard walk .
the inclusion of the interference terms in the continuum equations ( [ probabilidad1],[probabilidad2 ] ) is done by adding @xmath58 to their right hand sides .
the evolution of the variance , @xmath59 , of the distribution @xmath60 is a distinctive feature of the quantum walk .
it is known @xcite that it increases quadratically in time in the quantum case , but only linearly in the classical case .
we obtain the evolution of the variance analytically from the evolution of the first and second moments , defined as @xmath61 and @xmath62 respectively . the evolution equation for these moments
is , from eq.([probabilidad ] ) , written as@xmath63 \ , -\cos2\theta\ , m_{2}(t-1)-4\sin2\theta\sum_{i } i\beta_{i}(t ) .
\label{energia}\ ] ] in differential terms , these equations become , @xmath64 if the sums in eqs.([energia1 ] ) corresponding to the interference terms can be neglected , the evolution of the moments of the distribution is given by @xmath65 the general solution for these equations is of the form @xmath66 where @xmath67 , @xmath68 , @xmath69 and @xmath70 are constants . for times larger than @xmath71 , the transient
exponential is negligible and the variance increases linearly in time with a slope given by eq.([eq : d_coef ] ) , @xmath72 this is consistent with the expected result for a classical random walk on a line .
we emphasize that we have obtained this result neglecting the interference terms in the exact quantum evolution , but this result is valid for any @xmath5 . if the interference terms in eqs.([energia1 ] ) are not neglected , the process is not markovian but unitary .
the solution for arbitrary @xmath5 is cumbersome and not particularly illuminating .
therefore , in this subsection , we particularize our solution for the case of the hadamard walk setting @xmath22 in the map ( [ mapa ] ) . using fourier analysis , the solutions for the amplitudes @xmath73 and @xmath74
can be obtained @xcite . for the particular initial conditions @xmath75 and @xmath76 these solutions are @xmath77 e^{-i\left ( w_{k}t+kj\right ) } , \nonumber\\ b_{j}(t ) & = \frac{1+\left ( -1\right ) ^{j+t}}{2}\int_{-\pi}^{\pi}\frac { dk}{2\pi}\left [ \frac{e^{ik}}{\sqrt{1+\cos^{2}k}}\right ] e^{-i\left ( w_{k}t+kj\right ) } .
\label{bes}\ ] ] where @xmath78 . from these amplitudes , an expression for the variance can be obtained in the long time limit .
the sums in eqs.([energia1 ] ) are evaluated , using the fourier expansion of the delta function , with the result @xmath79 the numerical constants are given by @xmath80 and @xmath81 . even though we have obtained these results in the long time limit , for finite times
they correctly describe the average trend .
we have checked this fact numerically , by computing the relative differences @xmath82 /a$ ] and @xmath83 /(-at+b)$ ] as a function of time .
the result is shown in fig .
[ fig : f1 ] . using eq.([energia ] )
, the evolution for the first and the second moments can be expressed as the maps @xmath84 the solutions for these maps are @xmath85 with @xmath86 and @xmath87 arbitrary constants .
the variance @xmath59 is then @xmath88 this result is consistent with previous those of previous works . in ref.@xcite , it is obtained through numerical simulations , in @xcite it is found through fourier analysis , while in @xcite , the same expressions are obtained through a summation over different paths .
however , it is important to emphasize that , from eqs.([lim3 ] ) and ( [ variance ] ) , a direct relation between a coherent evolution and the long time quadratic increase of the variance has been established . in a decoherent evolution @xmath89 and @xmath90 are negligible and the increase of the variance becomes linear in time , as seen in eq.([sol2 ] ) .
this is a particular instance of the markovian process discussed in the previous subsection . in this subsection
we present an alternative analytical approach which can be applied to the generalized quantum walk , i.e. for arbitrary values of @xmath5 .
we start by rearranging the original map , eqs.([mapa ] ) , to uncouple both chirality components .
the resulting independent evolution equations are @xmath91 \nonumber\\ b_{i}(t+2)-b_{i}(t ) & = \cos\theta\,\left [ b_{i+1}(t+1)-b_{i-1}(t+1)\right ] .
\label{discreta2}\ ] ] for long times the left hand sides of the above equations can be approximated by time derivatives .
the result can be expressed as @xmath92 , \label{bessel1}\ ] ] where @xmath93 stands for either chirality component , @xmath94 or @xmath95 .
we define the effective time @xmath96 , then eq.([bessel1 ] ) becomes @xmath97 which is the recursion relation satisfied by the bessel functions .
thus , we can write the general solution as @xmath98 where @xmath99 are the initial conditions to be used in the differential equation ( [ bessel2 ] ) .
these initial conditions are not necessarily the same as those to be used in the discrete map ( [ discreta2 ] ) , because the approximation of a difference by a continuous derivative does not hold for small times .
the solution ( [ solucion1 ] ) with appropriate initial conditions provides a good long time description of the dynamics of the discrete map ( [ discreta2 ] ) . in this context ,
long time implies many applications of the discrete map .
note that eq.([solucion1 ] ) provides the additional information @xcite that the long time propagation speed of the probability distribution is given by @xmath100 .
it is worth mentioning that the long time solutions for the spinor amplitudes , eq.([solucion1 ] ) , have the same form as the time - evolution of the amplitudes of the kicked rotor in a principal resonance @xcite .
in fact , if @xmath101 is the dimensionless strength parameter of the kicked rotor as defined in @xcite , the time evolution for these amplitudes is given by eq.([solucion1 ] ) after the substitution @xmath102 we can obtain analytically the increase in the variance implied by eq.([solucion1 ] ) .
the position probability distribution can be expressed as @xmath103 \,j_{i - l}(t\cos\theta)\,j_{i - l}^{\prime}(t\cos\theta ) .
\label{distribucion}\ ] ] the first and second moments of this distribution are @xmath104 + m_{1}(0 ) \label{moment1}\ ] ] and @xmath105 \right\ } \nonumber\\ & \qquad - t\,\cos\theta\ , \sum_{l } ( 2l-1)\re\left [ \tilde
a_{l}(0)\tilde a_{l-1}^{\ast}(0 ) + \tilde b_{l}(0)\tilde b_{l-1}^{\ast}(0)\right ] + m_{2}(0 ) . \label{moment2}\ ] ] respectively , where @xmath106 and @xmath107 are the moments of the initial distribution .
thus the variance @xmath59 increases quadratically in time for arbitrary initial conditions .
this result holds for arbitrary ( non trivial ) values of the parameter @xmath5 . in fig .
[ fig : f2 ] we show the time evolution of the relative difference @xmath108 where @xmath109 is the difference between the approximate variance @xmath110 obtained from eqs.([moment1 ] ) and ( [ moment2 ] ) and the exact variance @xmath111 from the original hadamard map , eq.([mapa ] ) .
the independence of the quadratic increase of @xmath111 on the initial conditions has been established numerically for the particular case of the hadamard walk @xcite .
here we have also demonstrated the independence of this quadratic increase on the parameter of the unitary transformation . this is a new result .
the quadratic increase in time of the variance for the discrete time generalized quantum walk has been obtained analytically .
this increase remains quadratic for arbitrary initial conditions and for all non trivial values of the parameter @xmath5 controlling the unitary transformation in the chirality subspace .
this quadratic increase results from interference effects and thus it is strongly dependent on the coherence of the quantum evolution . if , for any reason , the quantum evolution is decoherent then the increase in the variance becomes linear with time .
the general approach of separating the quantum evolution equation into a master equation supplemented by a term which takes into account quantum coherence effects provides a new perspective which is helpful in clarifying why the quantum evolution spreads faster than in the classical one . in the quantum case
, there is a superposition of a left - propagating wave and a right - propagating wave , with both wavefronts traveling with constant speed @xmath100 .
thus , their separation increases linearly in time and the variance does so quadratically . in the markovian approximation ,
as developed in this work , the particle moves a step either to the right or to the left , making its choice in a random way . due to the randomness of this motion , the variance increases only linearly with time
this process can be visualized as a frequent position measurement process , which amounts to reinitialize the system , at each measurement , in a different state .
the wave function collapse causes memory loss of the previous distribution .
thus , we have a markovian process which describes a series of random steps .
we have shown how the markovian approximation method can be applied to a generalized quantum walk , for arbitrary values of the parameter @xmath5 and arbitrary initial conditions . when the evolution becomes decoherent ,
the interference terms may be dropped and the evolution is described by a master equation implying a linear increase of the variance with time .
the diffusion coefficient is not , in general , the same as that for the classical random walk . only in the particular case of the hadamard
walk the diffusion coefficient is @xmath112 as in the classical case .
this formalism shows in a transparent form that the primary effect of decoherence is to make the interference terms negligible in the evolution equation and then the markovian behavior is immediately implied .
this is true for any evolution operator of the form given in eq.([ugen ] ) .
we have established the analogy between the generalized quantum walk and the resonant kicked rotor .
this is done by obtaining an expression for the time evolution of the chirality amplitudes of the quantum walk and showing that they have the same form as the angular momentum components of the wavefunction of the kicked rotor in a resonant regime .
we related the kicked rotor strength parameter to the parameter @xmath5 defining the unitary transformation of the chirality .
this analogy opens several interesting possibilities .
a quantum algorithm implements the evolution of the quantum kicked rotor in a quantum computer is known @xcite .
since the quantum random walk on the line has the same long time dynamics as the resonant kicked rotor , the same quantum algorithm may be used to describe the evolution of a generalized discrete time quantum walk on the line .
the quantum kicked rotor has been experimentally realized using ultra - cold atom traps and some experiments have focused on the resonant case @xcite .
this opens interesting possibilities for the experimental realization of the quantum walk using quantum optics .
finally , note that the markovian approach presented here provides a systematic way to find the classical analog associated to a given quantum random walk . by reformulating the quantum problem as described here and neglecting the interference terms , the markovian equation describing the equivalent classical problem can be readily obtained .
we acknowledge the comments made by v. micenmacher and the support of _ pedeciba _ and _ conicyt ( clemente estable proy .
r.d . acknowledges partial financial support from the brazilian research council ( cnpq ) .
a.r , g.a . and r.d .
acknowledge financial support from the _
brazilian millennium institute for quantum information_cnpq .
williams , m.p .
sadgrove , a.j.daley , r.n.c .
gray , s.m .
tan , a.s .
parkins and r. leonhardt , preprint arxiv : quant - ph/0209090 ( 2002 ) ; f.l .
moore , j.c .
robinson , c.f .
bharucha b. sundaram and m.g .
raizen , _ phys .
lett . _ * 75 * 4598 ( 1995 ) . | we analyze in detail the discrete time quantum walk on the line by separating the quantum evolution equation into markovian and interference terms . as a result of this separation ,
it is possible to show analytically that the quadratic increase in the variance of the quantum walker s position with time is a direct consequence of the coherence of the quantum evolution .
if the evolution is decoherent , as in the classical case , the variance is shown to increase linearly with time , as expected .
furthermore we show that this system has an evolution operator analogous to that of a resonant quantum kicked rotor . as this rotator
may be described through a quantum computational algorithm , one may employ this algorithm to describe the time evolution of the quantum walker .
hadamard walk ; markovian process ; quantum information + pacs : 03.65.yz ; 03.67.lx ; 05.40.fb |
the pristine setting of atomic bose - einstein condensates ( becs ) has offered a significant playground for the examination of numerous physical concepts @xcite .
one of the flourishing directions has been at the interface of the theory of nonlinear waves and such atomic ( as well as optical ) systems , concerning , in particular , the study of matter - wave solitons @xcite .
such coherent nonlinear structures have been observed in experiments and studied extensively in theory .
prototypical examples of pertinent studies include but are not limited to bright @xcite , dark @xcite and gap @xcite matter - wave solitons , as well as vortices @xcite , solitonic vortices and vortex rings @xcite . among these diverse excitations , dark solitons in repulsive becs
have enjoyed a considerable amount of attention in effectively one - dimensional settings due to numerous experiments leading ( especially , more recently ) to their well - controlled creation @xcite . moreover , numerous works considered dark solitons in higher - dimensional settings in order to experimentally explore their instability leading to the formation of vortex rings and vortex lines , as illustrated , e.g. , in refs . @xcite . more recently , such excitations have also been observed in fermionic superfluids @xcite .
on the other hand , multi - dimensional ( i.e. , non - planar ) variants of dark solitons , were studied as well cf .
e.g. , chapters 7 and 8 of ref .
examples of such structures are radially symmetric solitons , such as the ring dark solitons ( rds ) , initially proposed in atomic becs in ref .
@xcite ( see also refs .
@xcite for subsequent studies ) .
nevertheless , such structures were studied to a far lesser extend , arguably due to their generic identified instability , which was found to give rise to states such as alternating charge vortex polygons .
notice that , as was recently shown in ref .
@xcite , stabilization of rds is possible upon employing a proper potential barrier .
the three - dimensional non - planar generalization of dark solitons , namely spherical shell solitons , has received very limited attention .
the analysis of ref .
@xcite in the case of box potentials ( rather than the typical parabolic trap setting of becs ) revealed their potential existence , but did not pursue their stability analysis .
moreover , the adiabatic dynamics of these structures was studied analytically via a variational approach in ref .
@xcite without providing , however , relevant numerical results . nevertheless , it is important to note that in an experimental realization of vortex ring states , the transient observation of spherical shell solitons was reported @xcite . in addition
, they were also suggested as a potential outcome of collisions of vortex rings in the work of ref .
@xcite .
it is the purpose of the present manuscript to revisit these states and explore not only their existence in isotropic traps , but also perform their systematic bogolyubov - de gennes ( bdg ) stability analysis .
this is done by projecting perturbations onto eigenfunctions of the linear problem in the angular directions . using this basis ,
the relevant computations are greatly simplified , and the full 3d spectrum computation is reduced to an effectively small set of one - dimensional problems ( for the different modes ) . it should be mentioned here that the use of spherical harmonics as a basis in the angular variables for states bearing radial symmetry is rather common in terms of numerical schemes for linear and nonlinear schrdinger ( nls ) equations see , e.g. , ref .
@xcite for a a recent example .
earlier efforts along these lines in the context of linear schrdinger equations can be found , e.g. , in the works of refs .
@xcite . in 2d becs ,
analogous decompositions of azimuthal modes for radial states including rdss and vortices @xcite have appeared , and in 3d similar possibilities are starting to emerge @xcite . in this work , we find that indeed spherical shell solitons can be identified as stable 3d solutions of the gross - pitaevskii equation , sufficiently close to the linear ( small - amplitude ) limit of small chemical potential from which they bifurcate , as is shown below .
moreover , we provide a theoretical analysis of their dynamics as `` effective particles '' .
this approach allows to characterize the opposite limit of large chemical potential .
our numerical computations of existence and stability , complemented by direct numerical simulations of the spherical shell soliton dynamics , not only corroborate the above two limits , but also provide a systematic way to interpolate between the two analytically tractable regimes .
our presentation is organized as follows .
first , in sec .
[ setup ] , we introduce the model and describe the implementation of the linear stability analysis . our analytical approaches for the spherical shell dark solitons in the above mentioned analytically tractable limits are then presented in sec .
[ sec : sub : part ] .
next , in sec .
[ results ] , we present our systematic numerical results . finally , our conclusions and a number of open problems for future consideration are given in sec .
[ conclusion ] . in the appendix , a similar numerical decomposition method but using cylindrical ( rather than spherical ) symmetry
is also discussed whose results are also given for comparison in sec . [ results ] .
in the framework of lowest - order mean - field theory , and for sufficiently low - temperatures , the dynamics of a 3d repulsive bec , confined in a time - independent trap @xmath0 , is described by the following dimensionless gross - pitaevskii equation ( gpe ) @xcite : @xmath1 where @xmath2 is the macroscopic wavefunction of the bec and @xmath3 is the chemical potential ( subscripts denote partial derivatives ) . here
, we consider a harmonic trap of the form : @xmath4 where @xmath5 , @xmath6 and @xmath7 are the trapping frequencies along the @xmath8 plane and the vertical direction @xmath9 , respectively .
note that the potential has rotational symmetry with respect to the @xmath9-axis . in our numerical simulations , we focus on the fully symmetric ( isotropic , spherically symmetric ) case with @xmath10 , in which we can benchmark our numerical methods in effective 1d and 2d against those of the fully 3d setting .
we examine the dark spherical shell solitons , hereafter referred to as dss .
this single radial node state exists in this isotropic case from the linear limit onwards as a stationary state of the form @xmath11 where @xmath12 is the spherical radial variable . at the linear limit
, the waveform is an eigenmode of the quantum harmonic oscillator with chemical potential @xmath13 , and spatial profile : @xmath14 where the basis @xmath15 denotes the 3d harmonic oscillator quantum states in cartesian coordinates . the dss is generally expected to be unstable for high chemical potentials due to transverse modulational ( snaking - type ) instability , in a way similar to its planar and rds counterparts , as summarized , e.g. , in refs .
@xcite . however , for smaller chemical potentials , and especially near the linear limit , it is relevant to analyze stability of the dss and identify corresponding instabilities and where they may arise . in
what follows ( cf .
[ results ] below ) , we will turn to numerical computations , based on a fixed point iteration and our spherical harmonic decomposition of the spectral stability bdg problem , in order to efficiently identify the intervals of existence and stability of the relevant modes . in this section ,
we discuss our analysis of the linear stability of the dss states utilizing their effective 1d nature , namely their spherical symmetry . in spherical coordinates @xmath16
, the laplacian can be decomposed into radial and angular parts , denoted by @xmath17 and @xmath18 respectively , namely @xmath19 where @xmath20 the @xmath18 operator has eigenstates given by the spherical harmonics @xmath21 with eigenvalues @xmath22 , i.e. , @xmath23 for states with spherical symmetry ( @xmath24 ) , the @xmath18 part of the laplacian is not relevant for identifying the stationary state . thus , the stationary dss state @xmath25 of eq .
( [ gpe ] ) , defined in the domain @xmath26 , satisfies the following radial equation : @xmath27 now , let us consider the linear stability ( bdg ) spectrum of such a stationary state .
we consider small - amplitude perturbations expanded using the complete basis of @xmath21 ( the spherical harmonic eigen - basis ) as follows : @xmath28 , \label{perturbation}\ ] ] where the space- and time - dependent coefficients @xmath29 and @xmath30 determine the radial and temporal evolution of the perturbation .
then , substituting eq .
( [ perturbation ] ) into eq .
( [ gpe ] ) and retaining up to linear terms in the expansion , one notes that all the @xmath31 modes are mutually independent due to the spherical symmetry of the steady state . for mode @xmath31 , the coefficients @xmath32 and @xmath33 of the expansion
obey the following evolution equations : @xmath34 -i b_t&=&-\frac{1}{2 } \delta_r b+ \frac{\ell(\ell+1)}{2r^2}b + vb + 2 |\psi_0|^2b + \psi_0^{*2}a-\mu b , \end{array}\ ] ] where we have dropped for simplicity the subscripts @xmath35 . as we are interested in the stability of the steady - state
, we now expand the coefficients using @xmath36 where @xmath37 is an eigenvalue of the following matrix @xmath38 with @xmath39 m_{12}&=&-i\psi_0 ^ 2,\\[1.0ex ] m_{21}&=&i\psi_0^{*2},\\[1.0ex ] m_{22}&=&i\left(-\frac{1}{2 } \delta_r + \frac{\ell(\ell+1)}{2r^2 } + v + 2|\psi_0|^2-\mu\right ) .
\end{array}\ ] ] one can , therefore , compute the full spectrum by computing that of each mode @xmath31 independently , and then putting them all together to `` reconstruct '' the full spectrum . in what follows
we use @xmath40 for values of the chemical potential up to @xmath41 .
the reduction of dimensionality with the decomposition on a complete basis set is also available in 2d for states with rotational symmetry up to a topological charge along the @xmath9 axis . for clarity
, this is shown in the appendix .
it is worth mentioning that a large class of bec states belongs to this class .
finally , we discuss the application of the degenerate perturbation method ( dpm ) @xcite to the dss state .
this method is employed for the determination of the dss spectrum near the linear limit , and also for the identification of the nature of instabilities .
the method can be most readily understood from the 3d version of the matrix @xmath42 , where one can separate the free ( linear ) part with known eigenstates and eigenvalues and the rest of the terms can be treated as perturbations near the linear limit .
note that there are 2@xmath432 block matrices within @xmath42 and the complete basis consists of both `` positive '' and `` negative '' eigenstates .
this terminology is associated with quantum harmonic oscillator eigenstates leading to positive or negative eigenvalues for the linear analogs of the operators @xmath44 in eq .
( [ m ] ) i.e. , with @xmath45 set to @xmath46 . in dpm ,
the functional dependence of the eigenvalues on the chemical potential @xmath3 is dominated by the interplay of equi - energetic , degenerate states .
the result of the nonlinear perturbations ( @xmath47 ) in @xmath48 within eq .
( [ m ] ) is to cause the degenerate eigenvalues to depart from their respective linear limit , as the norm of the solution ( mass ) increases .
the free part provides the basis that is formed from `` up '' states and `` down '' states , of the form @xmath49 and @xmath50 with eigenvalues @xmath51 and @xmath52 respectively , where @xmath53 is the eigenenergy of the 3d quantum harmonic oscillator for eigenstate @xmath54 . the relevant states for eigenvalues near the spectrum of im(@xmath37)=0 , 1 and 2 are listed in table .
then , the effect of the nonlinearity on these eigenmodes and the corresponding eigenvalue corrections are calculated , as discussed in ref .
@xcite ( and originally spearheaded in this context in ref .
below , this is compared to the numerical spectrum in sec .
[ results ] . .
relevant states for eigenvalues near the spectrum of im(@xmath37)=0,1 and 2 for the degenerate perturbation theory .
here @xmath55 stands for eigenstates with all possible distinct permutations with the quantum numbers @xmath56 and @xmath57 .
[ para ] [ cols="^,^,^",options="header " , ]
we now turn to the limit of large chemical potential @xmath3 . in this so - called thomas - fermi ( tf )
limit of large @xmath3 , the ground state of the gpe is approximated as @xmath58 @xcite .
a natural way to obtain a reduced dynamical description of the dss analogously to what is done in lower - dimensional settings is to develop an effective particle picture for the dss in the tf limit . here , following ref .
@xcite , we focus on the equilibrium radius @xmath59 of the dss , as a prototypical diagnostic that we can compare to our numerical results . this equilibrium radius is determined as the critical value of the dss radius for which the restoring force due to the harmonic trap is counterbalanced by the effective force exerted due to the curvature of the dss .
( left panels ) and argument @xmath60 ( right panels ) of a dark soliton shell state at @xmath61 ; the top row illustrates the @xmath62 plane , while the middle row the @xmath63 plane .
the bottom panel shows an isocontour plot of the same state .
, title="fig:",width=302 ] ( left panels ) and argument @xmath60 ( right panels ) of a dark soliton shell state at @xmath61 ; the top row illustrates the @xmath62 plane , while the middle row the @xmath63 plane .
the bottom panel shows an isocontour plot of the same state .
, title="fig:",width=226 ] .
bottom panel : number of particles as a function of chemical potential @xmath3 , showing the continuation of states from the linear limit to the nonlinear regime .
, width=302 ] the first approach ( motivated also by the work of ref .
@xcite ) focuses on the stationary state using the ansatz @xmath64 , where one obtains an equation for the dss profile @xmath65 ( on top of the tf ground - state @xmath66 ) given by : @xmath67 where @xmath68 and primes denote derivatives with respect to @xmath69 . inspired by the profile of the 1d and 2d equivalents of the dss , namely the dark soliton and the rds ,
we seek a stationary dss solution in the form of @xmath70 $ ] . multiplying both sides of eq .
( [ neweq ] ) by @xmath71 and integrating over @xmath69 from @xmath72 to @xmath73 ( bearing in mind that the contribution of the integral from @xmath74 to @xmath75 is exponentially small ) , we find that the equilibrium radius is given by @xmath76 where @xmath77 .
this result , as in the case of the dark soliton ring in 2d @xcite , slightly overestimates the actual value for @xmath78 that , as we will see below , is accurately predicted by the second particle picture approach , which retrieves the precise value for @xmath78 to be @xmath79 ; this is confirmed by the numerical results shown below , in sec .
[ results ] .
the second approach relies on energy conservation and it is based on the analysis of ref .
@xcite . in this approach
, it is argued that the equation of motion can be derived by a local conservation law ( i.e. , an adiabatic invariant ) in the form of the energy of a dark soliton under the effect of curvature and of the density variation associated with it .
more specifically , knowing that the energy of the 1d dark soliton centered at @xmath80 is given by @xmath81 @xcite , the generalization of the relevant adiabatic invariant quantity in a 3d domain bearing density modulations reads : @xmath82 \\[1.0ex ] & = & 4 \pi r_0 ^ 2 \left [ \frac{4}{3 } ( \mu - v(r_0)-\dot{r}_0 ^ 2)^{3/2 } \right ] , \end{array } \label{ppnew1}\ ] ] where @xmath83 and @xmath84 are the ( radial ) location and velocity of the dss with initial conditions @xmath85 and @xmath86 . taking a time derivative on both sides of eq .
( [ ppnew1 ] ) and assuming that the dss has no initial speed ( @xmath87 ) , we obtain a newtonian particle equation of motion for the dss of the following form : @xmath88 .
\label{ppextra}\ ] ] from this equation , the equilibrium position for the dss yields @xmath89 , a result consistent with the numerical findings of the next section .
we now proceed to test these predictions , the stability analysis spectrum , and the equilibrium positions for the dss . .
( a)-(b ) spectra computed using our effectively 1d method ( in the radial variable , using a spherical harmonic decomposition in the transverse directions ) .
identical results are obtained when using the 2d variant of the method in cylindrical coordinates ( and decomposing only the azimuthal direction in suitable fourier modes ) .
both 1d and 2d numerics use a spacing of @xmath90 .
( c ) spectra computed using the full 3d numerics with spacing @xmath91 .
orange ( grey ) lines depict the real part of @xmath37 ( i.e. , unstable parts of the eigenvalues ) , while blue ( dark ) lines the imaginary part of @xmath37 . , title="fig:",width=264 ] .
( a)-(b ) spectra computed using our effectively 1d method ( in the radial variable , using a spherical harmonic decomposition in the transverse directions ) .
identical results are obtained when using the 2d variant of the method in cylindrical coordinates ( and decomposing only the azimuthal direction in suitable fourier modes ) .
both 1d and 2d numerics use a spacing of @xmath90 .
( c ) spectra computed using the full 3d numerics with spacing @xmath91 .
orange ( grey ) lines depict the real part of @xmath37 ( i.e. , unstable parts of the eigenvalues ) , while blue ( dark ) lines the imaginary part of @xmath37 . , title="fig:",width=264 ] .
( a)-(b ) spectra computed using our effectively 1d method ( in the radial variable , using a spherical harmonic decomposition in the transverse directions ) .
identical results are obtained when using the 2d variant of the method in cylindrical coordinates ( and decomposing only the azimuthal direction in suitable fourier modes ) .
both 1d and 2d numerics use a spacing of @xmath90 .
( c ) spectra computed using the full 3d numerics with spacing @xmath91 .
orange ( grey ) lines depict the real part of @xmath37 ( i.e. , unstable parts of the eigenvalues ) , while blue ( dark ) lines the imaginary part of @xmath37 .
, title="fig:",width=264 ]
let us start by providing some of the basic features of the dss . a typical dss state at @xmath61 is shown in fig .
[ state ] , where its spherical symmetry is apparent .
a radial plot of the wavefunction in the tf limit is depicted in the top panel of fig .
[ lc ] . as the chemical potential
is increased the number of atoms ( `` mass '' ) @xmath92 increases as depicted in the bottom panel of fig .
[ lc ] .
we now examine the stability of the dss .
the corresponding spectra illustrating the imaginary part of the eigenvalues [ dark ( blue ) lines ) and the real part [ light ( orange ) lines ] are shown in fig .
[ f : spectrum ] .
panel ( a ) depicts the spectrum obtained via our numerical method over a wide range of chemical potentials .
identical results were obtained when using the method in 2d , based on a cylindrical coordinate decomposition and a representation of the azimuthal variable dependence in the corresponding fourier modes . for reasons of completeness , a summary of this variant of the method is presented in the appendix . panels
( b ) and ( c ) depict a zoomed in region to contrast the results between these lower dimensional methods and the full 3d numerics .
panel ( b ) corresponds to the spectrum as computed by means of the 1d spherical coordinate decomposition method using spherical harmonics with spatial spacing of @xmath90 , while panel ( c ) depicts the results from direct full 3d numerics using an spatial spacing of @xmath91 .
a close comparison between the panels suggests small discrepancies of the imaginary parts at im@xmath93 and im@xmath94 , and a spurious unstable mode with re@xmath95 . these discrepancies ( and their amendment as the mesh resolution @xmath96 decreases see below ) are likely due to the large spatial spacing @xmath96 that needs to be chosen such that the 3d calculations are still `` manageable'' in which case
the dss radial profile is under - resolved .
it is crucial to note that the 3d calculations are far more expensive than their lower dimensional ( 1d or 2d ) counterparts considered above . .
the larger frequency corresponds to the oscillatory mode of the dss while the smaller one corresponds to the breathing mode of the background density .
, width=264 ] to confirm that the discrepancies are indeed caused by the mesh resolution , when we further decrease the lattice spacing to @xmath97 ( results not shown here ) and recalculate the spectrum in 3d , the branches at im@xmath93 indeed split into two non - degenerate curves and the spurious modes at im@xmath94 and re@xmath98 begin to get suppressed .
this suggests that the discrepancies are indeed attributable to resolution effects and that the spectrum computed with the effectively 1d numerical method is indeed accurate . ) from the full 3d numerical solution for @xmath99 .
the initial condition corresponds to the stationary dss at its equilibrium radius .
second set of panels : evolution of a dss shifted from its equilibrium position undergoing oscillatory motion over one oscillation period @xmath100 .
the white dashed circle denotes the starting location of the dss at the beginning of the period .
third panel : radial profiles for the oscillating dss over half a period .
bottom panel : oscillations of the dss radius vs. time .
blue dots represent the radius extracted from full 3d numerics while the solid red line is the best ( least - square ) fit to a linear combination of oscillations with frequencies @xmath101 and @xmath102 ( see red dots in fig .
[ dpm ] ) corresponding , respectively , to the frequencies of the dss oscillations and the breathing mode of the background cloud . for a movie depicting the stable oscillations of the dss
please see the supplemental material .
, title="fig:",width=257 ] ) from the full 3d numerical solution for @xmath99 .
the initial condition corresponds to the stationary dss at its equilibrium radius .
second set of panels : evolution of a dss shifted from its equilibrium position undergoing oscillatory motion over one oscillation period @xmath100 .
the white dashed circle denotes the starting location of the dss at the beginning of the period .
third panel : radial profiles for the oscillating dss over half a period .
bottom panel : oscillations of the dss radius vs. time .
blue dots represent the radius extracted from full 3d numerics while the solid red line is the best ( least - square ) fit to a linear combination of oscillations with frequencies @xmath101 and @xmath102 ( see red dots in fig .
[ dpm ] ) corresponding , respectively , to the frequencies of the dss oscillations and the breathing mode of the background cloud . for a movie depicting the stable oscillations of the dss
please see the supplemental material .
, title="fig:",width=257 ] ) from the full 3d numerical solution for @xmath99 .
the initial condition corresponds to the stationary dss at its equilibrium radius .
second set of panels : evolution of a dss shifted from its equilibrium position undergoing oscillatory motion over one oscillation period @xmath100 .
the white dashed circle denotes the starting location of the dss at the beginning of the period .
third panel : radial profiles for the oscillating dss over half a period .
bottom panel : oscillations of the dss radius vs. time .
blue dots represent the radius extracted from full 3d numerics while the solid red line is the best ( least - square ) fit to a linear combination of oscillations with frequencies @xmath101 and @xmath102 ( see red dots in fig .
[ dpm ] ) corresponding , respectively , to the frequencies of the dss oscillations and the breathing mode of the background cloud . for a movie depicting the stable oscillations of the dss
please see the supplemental material .
, title="fig:",width=257 ] ) from the full 3d numerical solution for @xmath99 .
the initial condition corresponds to the stationary dss at its equilibrium radius .
second set of panels : evolution of a dss shifted from its equilibrium position undergoing oscillatory motion over one oscillation period @xmath100 .
the white dashed circle denotes the starting location of the dss at the beginning of the period .
third panel : radial profiles for the oscillating dss over half a period .
bottom panel : oscillations of the dss radius vs. time .
blue dots represent the radius extracted from full 3d numerics while the solid red line is the best ( least - square ) fit to a linear combination of oscillations with frequencies @xmath101 and @xmath102 ( see red dots in fig .
[ dpm ] ) corresponding , respectively , to the frequencies of the dss oscillations and the breathing mode of the background cloud . for a movie depicting the stable oscillations of the dss
please see the supplemental material .
, title="fig:",width=257 ] the comparison between the stability spectrum predicted by the degenerate perturbation method ( dpm ) and the one obtained numerically near the linear limit is shown in fig .
as the figure shows , the dpm captures the behavior of the spectrum close to the linear limit .
let us now confirm the stability of the dss in this limit .
we have performed a direct numerical integration of the full 3d dynamics of the dss for @xmath99 , where it is predicted to be stable cf .
[ f : spectrum ] and [ dpm ] .
this evolution is depicted in the top set of panels of fig .
also , in this figure , we depict the stable oscillations of a perturbed dss by displacing its initial location away from the equilibrium radius . the evolution for such a dss performing oscillations is displayed in the second set of panels in fig .
[ rk4s ] . a movie for this evolution
is provided in the supplemental material .
to more clearly visualize these oscillations , we depict on the third panel of fig .
[ rk4s ] the transverse density cuts at four different stages over half a period of the oscillation .
these density cuts also reveal the excitation of the background cloud which performs breathing oscillations . from the density cuts we extracted the ( radial ) location of the dss along the evolution which displays a beating - type quasiperiodic oscillation as depicted by the blue dots in the bottom panel of fig .
in fact , by ( least - square ) fitting a linear combination of two harmonic oscillations ( see red line ) we extract the frequencies @xmath101 and @xmath102 . upon closer inspection , these frequencies correspond , respectively , to the frequency of the dss oscillations and the frequency of the breathing mode of the background cloud .
these frequencies are also depicted in fig .
[ dpm ] clearly showing that they indeed belong to the stability spectrum of the stationary dss state .
it is important to stress that the observed dss oscillations are supported by the stability of the dss at its equilibrium position .
therefore , despite strong perturbations of the background that undergoes breathing oscillations , the dss robustly persists and performs stable oscillations about its equilibrium position .
as the spectrum indicates , see figs .
[ f : spectrum ] and [ dpm ] , for larger values of the chemical potential the dss becomes unstable . to identify this instability ,
we have looked at the eigenvectors of the dpm .
there are high degeneracies in the eigenvalues in this case , rendering harder the identification of the nature of the instability .
nevertheless , the dpm is still helpful in that regard .
for instance , we know from the eigenvectors that the mode with the largest slope decreasing at im@xmath93 , which causes instabilities , is a linear combination of `` up '' states , and the `` down '' states are not involved .
therefore , the instability is due to a bifurcation rather than a collision with `` negative '' energy modes ; the latter may lead to oscillatory instabilities , as discussed , e.g. , in ref .
the former may be associated with symmetry breaking features and exhibits the instability via real eigenvalue pairs .
direct numerical integration of eq .
( [ gpe ] ) for chemical potentials @xmath103 and 10.6 , shows that the first instability is a bifurcation of a six vortex lines ( vl6 ) cage , while the rest of the instabilities are dominated by a cubic six vortex rings ( vr6 ) cage .
the first instability suggests a bifurcation of vl6 from the interactions of the dss and a dark soliton state with three plane nodes perpendicular to the @xmath8-plane , separated by angles of @xmath104 ; the dark soliton state ( ds3 ) can be written as @xmath105 and a linear eigenenergy @xmath106 . a two - mode analysis @xcite using this state and the dss state yields a prediction of the bifurcation as occurring at @xmath107 , which is in good agreement with the numerical result @xmath108 of fig .
[ f : spectrum ] .
a similar treatment of the vr6 cage turns out to be more challenging due to the difficulty in designing the corresponding dark soliton state .
instead , we have looked at the ds4 state , which can form an eight vortex line ( vl8 ) cage . the two - mode analysis
then predicts a bifurcation at @xmath109 , which is in fair agreement with the second instability at @xmath110 from fig .
[ f : spectrum ] .
the numerical results , however , show that the dynamically resulting state is a vr6 cage , not a vl8 cage .
presumably the vl8 cage can also cause instabilities of the dss but is less robust than the vr6 cage , possibly due to the intersections of the vortex lines , which are absent in vr6 .
further analysis with higher values of @xmath111 ( not shown here ) also suggests the progressively lesser relevance of the dsm state in causing instabilities . on the other hand
, the vr6 state seems to be very robust and dominates the instability in a wide range of chemical potentials from the onset of this instability around @xmath112 .
the vl6 and vr6 cages are shown in fig .
[ vl6vr6 ] . )
dynamics at @xmath113 and @xmath114 , respectively .
in particular , the vr6 cage seems to be robust and appears in the dynamics for a wide range of chemical potentials from the onset of the instability around @xmath112 .
, title="fig:",width=158 ] ) dynamics at @xmath113 and @xmath114 , respectively .
in particular , the vr6 cage seems to be robust and appears in the dynamics for a wide range of chemical potentials from the onset of the instability around @xmath112 .
, title="fig:",width=158 ] finally , we present the numerical results for the equilibrium location @xmath59 of the dss as a function of the chemical potential @xmath3 and compare the results with those of the two particle pictures in the large density limit ( see sec . [ sec : sub : part ] ) .
a plot of @xmath115 as a function of @xmath3 is depicted in fig .
[ rc ] . note that the first particle picture predicts a slightly larger equilibrium @xmath59 , while the particle picture based on energy conservation of the dss as an adiabatic invariant agrees very well with the numerical results .
it is interesting to note that the asymptotic behavior only sets in at about @xmath116 , deep inside the tf regime .
although this would be a rather computationally demanding parametric range when using 3d ( or even 2d ) methods , our effective 1d approach based on a spherical harmonic decomposition of the angular dependencies , allows us to reach such large values of @xmath3 .
, as a function of @xmath3 ( solid red line ) .
note that the numerical values reach the asymptotic value of @xmath117 ( dashed black line ) , as per the second particle picture , when @xmath3 is large .
the first particle picture slightly overestimates @xmath59 ( dash - dotted blue line ) .
, width=302 ]
in this work , we have revisited the theme of spherical shell dark solitons .
we have employed analytical methods to study these structures both in the vicinity of the linear limit , using a bifurcation analysis , and in the thomas - fermi limit , where they can be treated as effective particles . for the numerical simulations , we have used continuation methods , as well as a quasi - one - dimensional radial computation method , decomposing the angular dependence of the perturbations in spherical harmonics , to determine their stability .
we have found that spherical shell dark solitons can , in fact , be dynamically robust , spectrally stable solutions of the 3d gross - pitaevskii equation within a certain interval of chemical potentials ( and atom numbers ) ; this occurs sufficiently close to the linear limit of relatively small chemical potentials . their instabilities , emerging when the chemical potential increases , have been elucidated and their dynamical outcome , namely the breakup into vortex ring and vortex line cages , has been showcased .
we believe that these results , may not only be used to partially explain the transient observation of these structures in earlier experiments @xcite and numerical simulations @xcite , but may also pave the way for their consideration in future experiments .
our work suggests a number of interesting future research directions , concerning , in particular , related coherent structures , their physical relevance and properties .
for instance , this study suggests large scale stability computations for a wide range of pertinent steady states with radial symmetry , including ones with different numbers of radial nodes as well as a number of different external potentials .
furthermore , it would be relevant to generalize and apply the numerical ( and analytical ) techniques used herein to multicomponent bec settings .
for example , one can study `` symbiotic states '' composed of spherical shell dark - bright solitons , in the spirit of the dark - bright 2d rings of ref .
@xcite . on the other hand ,
states emerging robustly from the instabilities of these spherically ( or cylindrically ) symmetric ones , such as the vortex ring ( and line ) cages are worthwhile to investigate further in their own right .
efforts along these directions are currently in progress and will be presented in future publications .
w.w . acknowledges support from nsf ( grant nos .
dmr-1151387 and dmr-1208046 ) .
p.g.k . gratefully acknowledges the support of nsf - dms-1312856 , as well as from the us - afosr under grant fa950 - 12 - 1 - 0332 , and the erc under fp7 , marie curie actions , people , international research staff exchange scheme ( irses-605096 ) .
r.c.g . gratefully acknowledges the support of nsf - dms-1309035 .
we thank the texas a&m university for access to their ada cluster .
the greatest advantage of the 1d method involving the decomposition into spherical harmonics is that one can compute the spectrum for large chemical potentials without excessive computational costs .
the reason that typical computations for large chemical potentials become computationally prohibitive , is that in this limit the density is large and , thus , the width of the relevant localized structures ( proportional to @xmath118 ) becomes much smaller than the domain size ( proportional to @xmath119 ) ; hence , a large number of mesh points is necessary to resolve properly these configurations .
nonetheless , it is worth mentioning that , for the projection method in 1d , the form of states that one can study is also restricted by the spherical symmetry ; therefore , it is also relevant to employ an analog of this method in 2d , so as to encompass a wider class of states .
an example concerns states with a topological charge @xmath120 along the @xmath9 axis , i.e. , stationary states of the form @xmath121 .
note that in this framework , one can study states including but not limited to the ground - state , planar , ring , or spherical shell dark soliton states , vortex lines , as well as coplanar or parallel vortex rings and hopfions @xcite , in both isotropic and anisotropic traps . in 2d , and for cylindrical coordinates @xmath122 , the laplacian is decomposed as follows : @xmath123 where its components read : @xmath124 the @xmath125 operator has eigenstates @xmath126 with eigenvalues @xmath127 , i.e. , @xmath128 a stationary state @xmath129 , defined in the domain @xmath130 , and bearing a topological charge @xmath120 , satisfies the following equation : @xmath131 as in the 1d case , we construct the linear stability problem as follows .
let @xmath132 a rotationally symmetric stationary state , up to a topological charge @xmath120 , along the @xmath9 axis perturbed using the complete basis of @xmath126 : @xmath133 \right ] .
% \label{perturbation2 } \nonumber\ ] ] substituting this expansion into the gpe of eq .
( [ gpe ] ) , and upon linearizing and matching the basis expansion on both sides , one can see that the different @xmath111 modes are mutually independent .
an equivalent derivation as in 1d shows that @xmath37 are the eigenvalues of the matrix @xmath134 where @xmath135 m_{12}&=&-i\psi_0 ^ 2,\\[1.0ex ] m_{21}&=&i\psi_0^{*2},\\[1.0ex ] m_{22}&=&i\left(-\frac{1}{2 } \delta_h + \frac{(m - s)^2}{2\rho^2 } + v + 2|\psi_0|^2-\mu\right ) .
\end{array}\ ] ] as before , one can therefore compute the full spectrum by computing each mode @xmath111 independently and then putting them together . in the results based on this method and presented in sec .
[ results ] , we use @xmath136 .
it is worth commenting here that the matrix for the 2d eigenvalue problem with a charge @xmath120 is less symmetric than matrices corresponding to the 1d projection and the 2d case with no charge .
therefore , the full 2d problem with charge appears to demand more computational work .
however , it should be noted that , in the 2d case , the eigenvalues of the full matrix @xmath42 corresponding to the projections over the @xmath137 and @xmath138 modes are related via an orthogonal transformation and , thus , the two set of eigenvalues are complex conjugates of each other .
therefore , it is only necessary to compute the non - negative @xmath111 set and , hence , the charged and the uncharged states actually have similar computational complexity .
p. g. kevrekidis , d. j. frantzeskakis , and r. carretero - gonzlez ( eds . ) , _ emergent nonlinear phenomena in bose - einstein condensates . theory and experiment
_ ( springer - verlag , berlin , 2008 ) ; r. carretero - gonzlez , d. j. frantzeskakis , and p. g. kevrekidis , nonlinearity * 21 * , r139 ( 2008 ) .
j. denschlag , j.e .
simsarian , d.l .
feder , c.w .
clark , l.a .
collins , j. cubizolles , l. deng , e.w .
hagley , k. helmerson , w.p .
reinhardt , s.l .
rolston , b.i .
schneider , and w.d .
phillips , science * 287 * ( 2000 ) 97101 .
b. p. anderson , p. c. haljan , c. a. regal , d. l. feder , l. a. collins , c. w. clark , and e. a. cornell , phys . rev .
lett . * 86 * , 2926 ( 2001 ) .
p. engels and c. atherton , phys .
lett . * 99 * , 160405 ( 2007 ) .
i. shomroni , e. lahoud , s. levy and j. steinhauer , nature phys .
* 5 * , 193 ( 2009 ) .
s. middelkamp , p. g. kevrekidis , d. j. frantzeskakis , r. carretero - gonzlez , and p. schmelcher , physica d * 240 * , 1449 ( 2011 ) .
w. wang , p. g. kevrekidis , r. carretero - gonzlez , d. j. frantzeskakis , t. j. kaper , and m. ma , phys .
a * 92 * , 033611 ( 2015 ) . | in this work we study spherical shell dark soliton states in three - dimensional atomic bose - einstein condensates .
their symmetry is exploited in order to analyze their existence , as well as that of topologically charged variants of the structures , and , importantly , to identify their linear stability bogolyubov - de gennes spectrum .
we compare our effective 1d spherical and 2d cylindrical computations with the full 3d numerics .
an important conclusion is that such spherical shell solitons can be stable sufficiently close to the linear limit of the isotropic condensates considered herein .
we have also identified their instabilities leading to the emergence of vortex line and vortex ring cages .
in addition , we generalize effective particle pictures of lower dimensional dark solitons and ring dark solitons to the spherical shell solitons concerning their equilibrium radius and effective dynamics around it . in this case too , we favorably compare the resulting predictions such as the shell equilibrium radius , qualitatively and quantitatively , with full numerical solutions in 3d . |
in our previous paper @xcite , we proved a conjecture of lisca @xcite , establishing a bijective correspondence between the milnor fibers of the smoothing components of a cyclic quotient singularity and the stein fillings of the corresponding contact lens space , the boundary of the singularity .
as a particular case of our results , the milnor fibers corresponding to distinct smoothing components of the reduced miniversal base space of the cyclic quotient singularity are pairwise non - diffeomorphic . here
one has to understand the diffeomorphisms _ in a strong sense _ : namely , there are natural identifications of the boundaries of the milnor fibers up to isotopy , and we showed that there are no diffeomorphisms extending those identifications . having in mind the construction of de jong and van straten in @xcite regarding the milnor fibers of sandwiched singularites , it is natural to try to extend the above result for such singularities .
one of the main obstructions of this program is that at the present moment the precise description of all the smoothing components of sandwiched singularities ( or / and the classification of all stein fillings of the corresponding contact boundaries ) is out of hope .
nevertheless , we will prove a slightly weaker version of the above result . in order to explain it we need some preparation .
the class of sandwiched singularities received its name in @xcite from the fact that the corresponding singularities are analytically isomorphic to the germs of surfaces which may be `` sandwiched '' birationally between two smooth surfaces ( see also subsection [ sand ] ) .
they form a subclass of the rational surface singularities and their study may be reduced in many respects to the study of plane curve singularities . indeed , de jong and van straten showed that this was the case for their deformation theory as well .
namely , one may encode a sandwiched singularity @xmath0 by a _ decorated curve _
@xmath1 , which is a germ of reduced plane curve whose components @xmath2 are decorated by sufficiently high positive integers @xmath3 .
one of the main theorems of @xcite states that the deformation properties of a sandwiched singularity reduces to the deformation properties of any such decorated curve which encodes it .
in particular , the 1-parameter smoothings of @xmath0 correspond bijectively to the so - called _ picture deformations _ of @xmath1 , which are @xmath4-constant deformations of @xmath5 with generic fibers having only ordinary singularities , accompanied with flat deformations of @xmath6 ( seen as a subscheme of the normalization of @xmath5 ) which generically are reduced and contain the preimage of the singular locus of the generic fiber .
the milnor fiber corresponding to a smoothing component of @xmath0 may be reconstructed up to strong diffeomorphism from the embedding in a 4-ball of the generic fiber of a corresponding picture deformation .
part of the topological structure of the generic fiber of the deformation of @xmath1 may be encoded in an associated _ incidence matrix _ between the irreducible components of the deformation of @xmath5 and the images of the support of @xmath6 seen as a subscheme of the normalization .
this matrix is well - defined up to permutation of columns .
as it is proved in @xcite , the validity of lisca s conjecture implies that , _ for cyclic quotient singularities _ , the milnor fibers which correspond to distinct incidence matrices ( up to permutation of columns ) are not strongly diffeomorphic .
the aim of this paper is to show that _ the same is true for all sandwiched surface singularities _ ( see theorem [ distinc ] ) . in fact , for any fixed realization of @xmath0 by a decorated curve @xmath1 , we provide a canonical method to reproduce all the incidence matrices ( associated with @xmath1 ) from the milnor fibers .
hence , each realization @xmath1 provides a ` test ' to separate the milnor fibers ( in fact , precise criteria to recognize them ) .
recall also that de jong and van straten predict ( see ( * ? ? ?
* ; * ? ? ?
* ( 4.3 ) ) ) that , for any fixed @xmath1 , the set of smoothing components of the deformation space of @xmath0 injects into the set of incidence matrices associated with @xmath1 . in particular , for any class of singularities , when this is indeed the case , we obtain via our result that there is no pair of smoothing components with ( strongly ) diffeomorphic milnor fibers . in order to connect our main result with the set of stein fillings of the contact boundary
, we have to face a new aspect of the problem : sandwiched singularities are not necessarily taut ( that is , their analytical structure is not determined by their topology ) .
moreover , fixing a topological type , for different analytic types one might have different structure of the miniversal deformation space ( therefore different sets of milnor fibers ) .
nevertheless , for a fixed topological type , the induced natural contact structure on the boundary is independent of the analytic realization and is invariant up to isotopy by all orientation - preserving diffeomorphisms ( cf .
@xcite ) . therefore ,
in theorem [ lowbound ] we are able to extend the above recognition criterion of the milnor fibers via incidence matrices by considering all the smoothings associated with the deformations of all the decorated germs of plane curves with a fixed topology . in this way
we get a lower bound for the number of stein fillings of the contact boundary of a sandwiched surface singularity , considered up to orientation - preserving diffeomorphisms fixed on the boundary .
all the differentiable manifolds we consider are oriented : any letter , say @xmath7 , denoting a manifold denotes in fact an oriented manifold .
we denote by @xmath8 the manifold obtained by changing the orientation of @xmath7 , and by @xmath9 its boundary , canonically oriented by the rule that the outward normal _ followed _ by the orientation of @xmath9 gives the orientation of @xmath7 .
we work exclusively with homology groups _ with integral coefficients_.
if @xmath7 is 4-dimensional , the intersection number in @xmath10 is denoted by @xmath11 .
an element of @xmath10 is called _ a @xmath12-class _ if its self - intersection is equal to @xmath13 .
if @xmath14 and @xmath15 are two oriented manifolds with boundary , endowed with a fixed isotopy class of orientation - preserving diffeomorphisms @xmath16 , we say that @xmath14 and @xmath15 are _ strongly diffeomorphic _
( with respect to that class ) if there exists an orientation - preserving diffeomorphism @xmath17 whose restriction to the boundary belongs to the given class .
the aim of this section is to recall briefly the needed facts about deformation theory , @xmath4-constant deformations and the topology of surface singularities . in what follows
, a _ singularity _ will design a germ of reduced complex analytic space .
let @xmath0 be an isolated singularity .
choose one of its representatives embedded in some affine space : @xmath18 .
denote by @xmath19 the compact ball of radius @xmath20 centered at the origin and by @xmath21 its boundary . by a general theorem ( see @xcite ) , there exists @xmath22 such that @xmath23 is transversal to the euclidean spheres @xmath21 for any @xmath24 $ ] .
@xmath25 is then called a _
milnor ball _ for the chosen embedding and such a representative @xmath26 is called a _ milnor representative _ of @xmath0 .
the manifold @xmath27 , oriented as boundary of the complex manifold @xmath28 is well - defined up to orientation - preserving diffeomorphisms , and is called the _ ( abstract ) boundary _ of @xmath0 .
we denote it by @xmath29 .
[ miniv ] let @xmath0 be a singularity .
a _ deformation _ of @xmath0 is a germ of flat morphism @xmath30 together with an isomorphism between @xmath0 and the special fiber @xmath31 .
1-parameter smoothing _ is a deformation over a germ of smooth curve such that the generic fibers are smooth .
a deformation @xmath30 of @xmath0 is _ versal _ if any other deformation is obtainable from it by a base - change .
a versal deformation is _ miniversal _ if the zariski tangent space of its base @xmath32 has the smallest possible dimension .
a _ smoothing component _ is an irreducible component of the miniversal base space over which the generic fibers are smooth .
if @xmath0 is a germ of reduced complex analytic space _ with an isolated singularity _ , then the following well - known facts hold : \(i ) ( schlessinger @xcite , grauert @xcite ) the miniversal deformation @xmath33 exists and is unique up to ( non - unique ) isomorphism .
\(ii ) there exist ( milnor ) representatives @xmath34 and @xmath35 of the reduced total and base spaces of @xmath33 such that the restriction @xmath36 is a trivial @xmath37-fibration .
hence , for each smoothing component of @xmath0 , the oriented diffeomorphism type of the oriented manifold with boundary @xmath38 does not depend on the choice of the generic element @xmath39 : it is called _ the milnor fiber _ of that component . moreover ,
its boundary is canonically identified with the boundary @xmath29 by an orientation - preserving diffeomorphism , _ up to isotopy_. therefore , if one takes the milnor fibers corresponding to two distinct components of @xmath35 , one may ask whether they are _ strongly diffeomorphic _ ( a notion introduced in subsection [ conv ] ) with respect to the previous natural identification of their boundaries .
let @xmath40 be a ( not necessarily plane ) curve singularity .
its @xmath4-_invariant _ is by definition : @xmath41 where @xmath42 is a normalization morphism ( in particular , @xmath43 denotes a multi - germ ) . if @xmath5 is a reduced curve , by definition , its @xmath4-_invariant _ is the sum of the @xmath4-invariants of all its singular points . as an example , consider an _ ordinary @xmath44-tuple point _ of a reduced curve embedded in a smooth surface , that is a point where the curve has exactly @xmath44 local irreducible components , all smooth and meeting pairwise transversally .
at such a point , the @xmath4-invariant of the curve is equal to @xmath45 .
more generally , if @xmath40 is a germ of plane curve , one has the following classical formula : @xmath46 where @xmath47 varies among all the points infinitely near @xmath48 on the ambient surface and @xmath49 denotes the multiplicity of the strict transform of @xmath5 at such a point @xmath47 ( see ( * ? ? ?
* ; * ? ? ?
* and 393 ) ) .
denote by @xmath50 a deformation of @xmath40 over a smooth germ of curve .
it is called @xmath4-_constant _ if there exists a milnor representative of @xmath33 such that the @xmath4-invariant of its fibers is constant . by a characterization of teissier @xcite
( see also @xcite ) , a deformation @xmath33 is @xmath4-constant if and only if the normalization morphism of @xmath51 has the form @xmath52 , and it is the _ simultaneous normalization _ of the fibers of @xmath33 . in particular , if one has a @xmath4-constant embedded deformation of a germ of _ plane _ curve @xmath40 such that the general fiber @xmath53 has only ordinary multiple points , then after a choice of a milnor representative of the morphism , the general fibers are immersed _
discs _ in a @xmath54-ball .
consider a normal surface singularity @xmath0 .
it has a unique minimal normal crossings resolution , that is , a resolution whose exceptional set @xmath55 is a divisor with normal crossings and which looses this property if one contracts any @xmath12-curve of it .
consider also the _ weighted dual graph _ of this resolution .
each vertex is weighted by the genus and by the self - intersection number of the corresponding irreducible component of the divisor @xmath55 . from the resolution
, the abstract boundary @xmath29 inherits a _ plumbing structure _ , that is , a family of pairwise disjoint embedded tori whose complement is fibered by circles and such that on each torus , the intersection number of the fibers from each side is @xmath56 . the tori correspond to the edges of the dual graph and the connected components of their complement correspond to the vertices . for more details , see @xcite or @xcite .
neumann proved in @xcite that the knowledge of the boundary of @xmath0 , up to orientation - preserving homeomorphisms , is equivalent to the knowledge of the weighted dual graph of the minimal normal crossings resolution , up to isomorphisms . in (
* ; * ? ? ?
* theorems 9.1 and 9.7 ) was proved the following enhancement of neumann s result : [ invar ] the plumbing structure corresponding to the minimal normal crossings resolution is determined by the oriented boundary ( as abstract manifold ) ; in paticular , it is invariant , up to isotopy , by all orientation - preserving diffeomorphisms of the boundary .
it is important to note that in the definition of a plumbing structure , the circles of the fibration of the complement of the tori are considered _ unoriented _ ( otherwise the previous theorem would not be true ) .
but in the case of the boundary of a surface singularity , once a fiber is oriented , there is a canonical way to orient all of them ( see ( * ? ? ?
* ; * ? ? ?
* corollary 8.6 ) ) .
note also that any fiber has a natural framing , given by the nearby fibers .
we say that it is _ the framing coming from the plumbing structure_.
de jong and van straten related in @xcite the deformation theory of sandwiched surface singularities to the deformation theory of so - called _ decorated plane curve singularities_. they showed that 1-parameter deformations of particular decorated curves provide 1-parameter deformations for sandwiched singularities , and that all of these later ones can be obtained in this way .
moreover , the milnor fibers of those which are smoothings can be combinatorially described by the so - called _ picture deformations_. the aim of this section is to recall briefly this theory .
a normal surface singularity @xmath0 is called _ sandwiched _ ( see @xcite )
if it is analytically isomorphic to a germ of algebraic surface which admits a birational map @xmath57 .
see also @xcite for other view - points .
sandwiched singularities are rational , therefore their minimal normal crossings resolution coincides with their minimal resolution and the associated weighted graph is a negative - definite tree of rational curves .
they are characterized ( like the rational singularities ) by their dual resolution graphs , which allows to speak about _ sandwiched graphs _ : [ charsand ] a graph @xmath58 is sandwiched if by gluing ( via new edges ) to some of the vertices of @xmath58 new rational vertices with self - intersections @xmath13 , one may obtain a ` smooth graph ' , i.e. the dual tree of a configuration of @xmath59 s which blows down to a smooth point .
the way to add such @xmath12-vertices is not unique .
but once such a choice was done , for each sandwiched singularity with that topology , one may embed a tubular neighborhood of the exceptional set @xmath55 of the minimal resolution in a larger surface which contains also some @xmath12-curves , whose union with @xmath55 can be contracted and has as dual graph the chosen ` smooth graph ' .
this procedure will be used in the next subsection .
any sandwiched singularity may be obtained from a weighted curve @xmath1 . here
@xmath60 denotes a reduced germ of plane curve _ with numbered branches _ ( irreducible components ) @xmath61 and @xmath62 .
let us explain this .
consider the minimal abstract resolution of @xmath5 obtained by a sequence of blowing - ups of ( closed ) points .
if one realizes those blowing - ups inside the smooth ambient surface , one does not necessarily obtain an embedded resolution of @xmath5 , that is , the total transform of @xmath5 is not necessarily a divisor with normal crossings .
the _ multiplicity sequence _ associated with @xmath2 is the sequence of multiplicities on the successive strict transforms of @xmath2 , starting from @xmath2 itself and not counting the last strict transform .
the _ total multiplicity _
@xmath63 of @xmath2 with respect to @xmath5 is the sum of the sequence of multiplicities of @xmath2 defined before .
* ; * ? ? ?
* ( 1.3 ) ) [ decgerm ] a _ decorated germ _ of plane curve is a weighted germ @xmath1 such that @xmath64 for all @xmath65 .
the point is that , starting from a decorated germ , one can blow up iteratively points infinitely near @xmath48 on the strict transforms of @xmath5 , such that the sum of multiplicities of the strict transform of @xmath2 at such points is exactly @xmath3 .
such a composition of blow - ups is determined canonically by @xmath1 .
if @xmath3 is sufficiently large ( in general this bound is larger than @xmath63 , see below ) , then the union of the exceptional components which do not meet the strict transform of @xmath5 form a _ connected _ configuration of ( compact ) curves @xmath66 . after the contraction of @xmath66 , one gets a sandwiched singularity @xmath67 , determined uniquely by @xmath1 ( for details see @xcite ) .
using proposition [ charsand ] , for any sandwiched singularity @xmath0 one can find @xmath1 such that @xmath0 can be represented as @xmath67 . indeed , choose an extension of the sandwiched graph to a smooth graph by adding @xmath12-vertices ( different choices provide different realizations @xmath67 ) .
fix a geometrical realization of this extension starting from the minimal resolution of the given sandwiched singularity ( by gluing tubular neighbourhoods of the new @xmath12-curves ) .
consider then a _ curvetta _ ( that is , a smooth transversal slice through a general point ) for each @xmath12-component and blow down the union of these curvettas to the smooth surface obtained by contracting the configuration with smooth graph .
number then the components of the plane germ @xmath5 obtained in this way and consider the function @xmath6 describing the reverse process of reconstruction of the initial configuration of compact curves and curvettas .
one obtains a decorated germ @xmath1 such that @xmath67 is the original singularity .
we introduce the following definition : @xmath1 is called _ a standard decorated germ _ if it was obtained in the previous way .
an example of the previous process applied to a cyclic quotient singularity is indicated in figure [ fig : extgraf ] .
a representative of the standard decorated germ obtained like this is @xmath68 , where @xmath69 is defined by @xmath70 , @xmath71 is defined by @xmath72 and @xmath73 .
one has @xmath74 . at 3 112 at 58 112 at 111 112 at 238 112 at 291 112 at 347 112 at 401 112 at 323 37 at 637 141 at 510 72 at 662 28 at 662 82 we wish to emphasize again , that for standard decorated germs the integers @xmath3 usually are larger than @xmath63 .
the ` right ' bound is the following .
consider the _ minimal embedded resolution _
of @xmath75 by blowing - ups of closed points .
that is , one does not only ask that the strict transform of @xmath5 be smooth , but also that its total transform be a normal crossings divisor . for each @xmath76 ,
denote by @xmath77 ( see @xcite ) the sum of multiplicities of the strict transforms of @xmath2 , before arriving at this embedded resolution ( for instance , for the germ represented in figure [ fig : extgraf ] , one gets @xmath78 and @xmath79 ) .
one may see easily that @xmath1 is a standard decorated germ if and only if @xmath80 , for all @xmath76 .
consider again an _ arbitrary _ decorated germ @xmath1 .
the total multiplicity of @xmath2 with respect to @xmath5 may be encoded also as the unique subscheme of length @xmath63 supported on the preimage of @xmath48 on the normalization of @xmath2 .
this allows to define the _ total multiplicity scheme _
@xmath81 of any reduced curve contained in a smooth complex surface , as the union of the total multiplicity schemes of all its germs .
the deformations of @xmath1 considered by de jong and van straten are : \(i ) ( * ? ? ?
* ; * ? ? ?
* ( 4.1 ) ) [ defcurve ] given a smooth complex analytic surface @xmath51 , a pair @xmath1 consisting of a reduced curve @xmath82 and a subscheme @xmath6 of the normalization @xmath83 of @xmath5 is called _ a decorated curve _ if @xmath81 is a subscheme of @xmath6 .
\(ii ) @xcite a @xmath84-_parameter deformation _ of a decorated curve @xmath1 over a germ of smooth curve @xmath32 consists of : \(1 ) a @xmath4-constant deformation @xmath85 of @xmath5 ; \(2 ) a flat deformation @xmath86 of the scheme @xmath6 , such that : \(3 ) @xmath87 , where the _ relative total multiplicity scheme _ @xmath88 of @xmath89 is defined as the closure @xmath90 .
\(iii ) a @xmath84-parameter deformation @xmath91 is called a _ picture deformation _ if for generic @xmath92 the divisor @xmath93 is reduced .
it is an immediate consequence of the definitions that for a picture deformation @xmath94 , the singularities of @xmath95 are only ordinary multiple points .
therefore , it is easy to draw a real picture of such a deformed curve , which motivates the name .
as an example , in figure [ fig : pictdef ] we represented a generic fiber @xmath96 and the image of the support of the subscheme @xmath93 of its normalization for one of the picture deformations of the standard decorated curve obtained in the figure [ fig : extgraf ] . at 332 135 at 12 52
de jong and van straten proved that part of the deformation theory of sandwiched surface singularities may be reduced to that of @xmath84-parameter deformations of decorated germs : ( * ? ? ?
* ; * ? ? ?
* ( 4.4 ) ) all the 1-parameter deformations of @xmath67 are obtained by @xmath84-parameter deformations of the decorated germ @xmath1 .
moreover , _ picture deformations _
provide all the smoothings of @xmath67 .
consider a decorated germ @xmath1 and one of its picture deformations @xmath91 .
fix a closed milnor ball @xmath97 for the germ @xmath40 . for @xmath98
sufficiently small , @xmath53 will have a representative in @xmath97 , denoted by @xmath99 , which meets @xmath100 transversally . as explained in subsection [ delta ] , it is a union of _ immersed discs _
@xmath101 canonically oriented by their complex structures ( and whose set of indices correspond canonically to those of @xmath102 ) .
the singularities of @xmath99 consist of ordinary @xmath44-tuple points , for various @xmath103 .
denote by @xmath104 the set of images in @xmath97 of the points in the support of @xmath93 .
it is a finite set of points which contains the singular set of @xmath99 ( because @xmath105 for @xmath106 ) , but it might contain some other ` free ' points as well . there is a priori no preferred choice of their ordering .
hence , the matrix introduced next is well - defined only _ up to permutations of columns_. by contrast , the ordering of lines is determined by the fixed order of the components of @xmath5 .
@xcite [ incid ] the _ incidence matrix _ of a picture deformation @xmath91 is the matrix @xmath107 ( with @xmath20 rows and @xmath108 columns ) whose entry @xmath109 at the intersection of the @xmath110-th row and the @xmath111-th column is equal to the multiplicity of @xmath112 as a point of @xmath113 .
such a matrix satisfies the following necessary conditions : @xmath114 where @xmath115 denotes the @xmath4-invariant of the branch @xmath2 ; @xmath116 where @xmath117 stands for the intersection multiplicity at @xmath48 of the branches @xmath2 and @xmath118 ; @xmath119 the previous equations are consequences of definition [ defcurve ] ( ii ) and of the fact that @xmath96 has only ordinary singularities .
it is an open problem to determine , in general , the set of incidence matrices associated with a given decorated germ , among the matrices which satisfy simultaneously ( [ delt ] ) , ( [ inters ] ) and ( [ nopoints ] ) .
it is also an open problem to determine the union of the sets of incidence matrices associated with all decorated germs with a given topology .
the milnor fiber of a smoothing associated to a picture deformation of a standard decorated germ @xmath1 is recovered as follows .
let : @xmath120 be the simultaneous blow - up of the points @xmath112 of @xmath99 .
here @xmath121 , where @xmath122 is the strict transform of the disc @xmath113 by the modification @xmath123 .
consider pairwise disjoint compact tubular neighborhoods of the discs @xmath124 in @xmath125 and denote their interiors by @xmath126 .
* ; * ? ? ?
* ( 5.1 ) ) [ recofiber ] let @xmath1 be a standard decorated germ .
then the milnor fiber of the smoothing of @xmath67 corresponding to the picture deformation @xmath91 is orientation - preserving diffeomorphic to the compact oriented manifold with boundary @xmath127 ( whose corners are smoothed ) .
moreover , one can show that in this realization , the canonical identification of the boundaries of the milnor fibers corresponding to different smoothing components is the unique identification ( up to isotopy ) which extends the identity morphism on the complement in @xmath128 of some tubular neighborhood of @xmath129 .
even more , by this presentation ( for a fixed @xmath1 ) one can also see the canonical identification of all the boundaries of the milnor fibers with the boundary @xmath29 , cf .
also lemma [ handles ] and the paragraphs following it .
we mention that although many results regarding decorated curves are valid for any @xmath64 , in proposition [ recofiber ] one needs @xmath130 , i.e. standard decorated curves
( a fact omitted in ( * ? ? ?
* ; * ? ? ?
* ( 5.1 ) ) ) . without this hypothesis , proposition [ recofiber ]
is false , as one may see by considering more than one curvetta attached to a @xmath12-curve .
in this section we prove our main results , theorem [ distinc ] and theorem [ lowbound ] .
the main idea of the proofs , inspired by the article @xcite of lisca , is to close all the milnor fibers with the same ` cap ' ( a 4-manifold with boundary ) which is glued each time in the ` same way ' ( this make sense thanks to the canonical identifications up to isotopy of the boundaries of the milnor fibers ) .
the cap we use emerges naturally from the view - point of de jong and van straten .
[ sanfold ] an oriented 3-manifold is called a _ sandwiched manifold _ if it is orientation - preserving diffeomorphic to the oriented boundary @xmath29 of a sandwiched surface singularity @xmath0 .
any such singularity is called a _ defining singularity _ of @xmath131 . by neumann s theorem recalled in subsection [ toponorm ] ,
sandwiched 3-manifolds determine the sandwiched graph associated with any defining singularity . by theorem [ invar ] ,
they may also be endowed canonically , up to an isotopy , with a plumbing structure @xmath132 corresponding to the minimal resolution of a defining singularity .
we say that the connected components of the complement in @xmath131 of the tori of @xmath132 are the _ pieces _ of @xmath131 .
suppose now that @xmath133 , where @xmath1 is a standard decorated germ with @xmath60 .
let @xmath134 be the composition of blow - ups of points infinitely near @xmath48 determined by @xmath1 .
one may write : @xmath135 where @xmath66 is the exceptional divisor of the minimal resolution of @xmath0 and @xmath136 are all the @xmath12-curves contained in the exceptional divisor @xmath137 , numbered such that @xmath138 is the unique irreducible component of @xmath137 which intersects the strict transform of @xmath2 .
denote by @xmath139 the unique irreducible component of @xmath66 which intersects @xmath138 . to @xmath139
corresponds a well - defined piece of @xmath131 . in this way
, one gets a map from the set @xmath140 to the set of pieces of @xmath131 .
as the disc - configuration @xmath99 is obtained by deforming @xmath5 , its boundary @xmath141 is isotopic as an oriented link to @xmath142 .
therefore , we can isotope @xmath99 outside a compact ball containing all the points @xmath112 till its boundary coincides with the boundary of @xmath5 . from now on ,
@xmath99 will denote the result of this isotopy .
let @xmath143 be a second copy of @xmath144 , and define ( see figure [ fig : glue ] ) : @xmath145 here @xmath146 is the oriented 4-sphere obtained by gluing the boundaries of @xmath97 and @xmath147 by the tautological identification , and @xmath148 , where @xmath149 is obtained by gluing @xmath113 ( perturbed by the above isotopy ) and @xmath150 along their common boundaries . recall that @xmath150 is a topological disc , while @xmath113 an immersed disc .
moreover , one can also glue @xmath151 with @xmath152 in such a way that the blow - up morphism @xmath123 of ( [ blowpoints ] ) may be extended by the identity on @xmath147 , yielding : @xmath153 here @xmath154 , where @xmath155 denotes the strict transform of @xmath149 , i.e. @xmath156 . in particular ,
@xmath155 is a topologically _ embedded _
2-sphere in @xmath157 , which is smoothly embedded in @xmath158 ( it is smoothly embedded everywhere if and only if the branch @xmath2 is smooth ) . by blowing - up a point of multiplicity @xmath163 of a compact complex curve on a smooth complex surface , its self - intersection drops by @xmath164 .
this is also true if one looks simply at a @xmath165-cycle on an oriented 4-manifold which becomes holomorphic in an adequate chart containing the point to be blown - up . applying this last fact to @xmath166 , which is holomorphic as seen inside a neighborhood of the points @xmath167
, we get : @xmath168 since @xmath169 by the fact that @xmath170 .
the first family of formulae is proved . since @xmath175 ( cf .
[ recofiber ] ) , the closed oriented 4-manifold @xmath157 is obtained by closing the boundary of @xmath7 by the cap @xmath176 .
this cap is independent of the chosen picture deformation : [ handles ] @xmath176 depends only on @xmath1 and is independent of the chosen picture deformation ( therefore one may close all the different milnor fibers using the same @xmath176 ) .
in fact , each @xmath126 in @xmath176 is a @xmath54-dimensional handle of index @xmath165 glued to @xmath147 along the knot @xmath177 endowed with the @xmath178-framing .
the chosen framing index is equal to the self - intersection number of a surface obtained by gluing the core of the handle to a 2-chain contained in the 4-ball @xmath147 and whose boundary is @xmath179 . by taking @xmath124 as core disc and @xmath150 as 2-chain , we get the sphere @xmath155 .
we apply then the first family of formulae stated in lemma [ intsp ]
. moreover , the cap @xmath176 is always glued in the same way , up to isotopy , to the boundaries of the milnor fibers corresponding to the various picture deformations , identified by the natural diffeomorphisms . to see this ,
note first that @xmath126 may be seen as a 2-handle attached to @xmath147 , with core disc @xmath124 .
take a co - core @xmath180 .
it is a core of @xmath126 seen as a 2-handle attached to @xmath7 . in order to describe the attaching of @xmath176 to @xmath7 ,
it is enough to describe the link @xmath181 in @xmath7 and its associated framing up to isotopy .
[ attach ] let @xmath1 be a standard decorated germ . consider on @xmath9 the plumbing structure corresponding to the minimal resolution of @xmath67 .
consider then an ordered collection of @xmath20 regular fibers @xmath182 , where @xmath183 is the generic fiber of the piece of @xmath7 marked by @xmath110 .
each knot @xmath183 is endowed with the framing coming from the plumbing structure , increased by @xmath84 .
then one obtains a framed link isotopic to the attaching framing of @xmath184 .
note that the orientation of @xmath9 ( as induced from @xmath7 ) coincides with the orientation induced from @xmath185 . in what follows
, we will reason inside @xmath186 .
consider on @xmath187 the natural complex structure , as a copy of @xmath97 .
then @xmath188 is holomorphic with respect to this complex structure and we may consider the sequence of blow - ups above @xmath189 dictated by the copy @xmath190 of @xmath1 .
let @xmath191 be this total blow - up morphism and @xmath192 be the strict transform of @xmath193 by the morphism @xmath194 . denote by @xmath195 the multiplicity sequence associated to @xmath196 .
the total multiplicity of @xmath2 with respect to @xmath5 is given by @xmath197 .
one has @xmath198 , where @xmath199 , which implies that : @xmath200 but @xmath201 by formula ( [ deltgen ] ) , which implies that @xmath202 . therefore : @xmath203 from formula ( [ comp ] ) and lemma [ intsp ] , we deduce that @xmath204 .
the total space @xmath205 appears then as a tubular neighborhood of a configuration of spheres whose dual graph is obtained from the dual graph of the exceptional divisor of @xmath194 by attaching a @xmath206-vertex to each @xmath12-vertex .
the co - cores of the handles @xmath207 appear as curvettas for @xmath192 .
an easy application of plumbing calculus ( see @xcite ) finishes the proof . up to permutations
, there exists a unique basis @xmath208 formed by @xmath12-classes of @xmath209 such that the matrix : @xmath210\cdot e_j)_{1\leq i \leq r , 1\leq j \leq n}\ ] ] has only non - negative entries .
denote by @xmath211 the 2-spheres embedded in @xmath157 obtained by blowing - up the points @xmath112 .
we suppose them oriented by the complex structure existing on @xmath125 .
their homology classes are @xmath12-classes in @xmath209 . as @xmath146 is diffeomorphic to a 4-sphere
, we deduce that @xmath212)_{1 \leq j \leq n}$ ] is a basis of the free group @xmath209 .
moreover , the spheres being disjoint , the intersection form on @xmath213 is in canonical form ( the opposite of a sum of squares ) in this basis .
this shows that any @xmath12-class is equal to some @xmath214 $ ] . by construction , if we take the basis @xmath212)_{1
j \leq n}$ ] , the matrix @xmath215\cdot [ \sigma_j])_{1\leq i \leq r , 1\leq j \leq n}$ ] has only non - negative entries ( in fact , elements of @xmath216 ) .
suppose that we change the sign of one of the classes @xmath217 $ ] .
as the corresponding column contains at least one non - zero entry ( because the point @xmath112 lies on at least one of the curves @xmath113 ) , we deduce that the new matrix has some negative entries in that column .
this proves the lemma .
the previous proof shows that , up to permutation of columns : @xmath221\cdot [ \sigma_j])_{1\leq i \leq r , 1\leq j \leq n}.\ ] ] but @xmath222\cdot [ \sigma_j]$ ] is equal to the incidence number between the curve @xmath113 and the point @xmath112 , that is , to the @xmath223-entry @xmath161 of the matrix @xmath218 .
the proposition is proved .
[ markinc ] the incidence matrix @xmath218 associated to a picture deformation of a standard decorated germ @xmath1 is determined ( up to a permutations of its columns ) by the associated milnor fiber and the marking of @xmath224 .
consider the compact 4-manifold with boundary @xmath176 , described in lemma [ handles ] .
attach it to the milnor fiber @xmath225 as described in lemma [ attach ] .
this attaching is determined by the marking of @xmath224 . then use proposition [ eqmat ] .
consider now two decorated germs of plane curves .
one says that they are _ topologically equivalent _ if one may choose milnor representatives of them inside euclidean balls centered at the origin of @xmath226 and homeomorphisms between the balls which restrict to homeomorphisms between the representatives and which additionally respect the numberings of their branches and the decoration functions @xmath6 .
one may show easily that topologically equivalent decorated germs give rise to topologically equivalent sandwiched singularities with induced diffeomorphisms of their boundaries preserving the associated markings .
look at the diffeomorphisms between the boundaries of the two singularities which preserve the orientations and the markings .
one may consider those diffeomorphisms between fillings of their boundaries which restrict to such special diffeomorphisms on the boundaries . then : [ distinc ] consider two topologically equivalent standard decorated germs of plane curves , and for each one of them a picture deformation .
if their incidence matrices are different up to permutation of columns , then their associated milnor fibers are not diffeomorphic by an orientation - preserving diffeomorphism which preserves the markings of the boundaries . in section [ general ]
, we recalled the notion of abstract boundary @xmath29 of an isolated singularity @xmath0 of arbitrary dimension .
in fact , by considering the field of maximal complex subspaces of the tangent space to a representative of the boundary , one gets a contact structure .
we denote it @xmath227 and we call it the _ contact boundary _ of @xmath0 .
it is well - defined up to contactomorphisms which are unique up to isotopy ( see @xcite and @xcite ) . by the main theorem of @xcite , for normal _ surface _ singularities
the contact structure @xmath228 is a topological invariant of the germ , that is , the contact boundaries of two singularities with homeomorphic oriented boundaries are contactomorphic by an orientation - preserving diffeomorphism .
moreover , as proved in @xcite , this ( unoriented ) contact structure is left invariant by all orientation - preserving diffeomorphisms of the boundary @xmath29 , when this boundary is a rational homology sphere .
this happens in particular for sandwiched singularities .
therefore , we can speak about the _
standard contact structure _ of a sandwiched 3-manifold , well - defined _ up to isotopy_. by taking milnor fibers of possibly non analytically equivalent but topologically equivalent sandwiched singularities , one gets stein fillings of the same contact manifold ( the boundaries being identified by any orientation - preserving diffeomorphism ) .
we get : [ lowbound ] let @xmath229 be a sandwiched 3-manifold endowed with its standard contact structure .
fix the topological type of a defining standard decorated germ .
then there are at least as many stein fillings ( up to diffeomorphisms fixed on the boundary ) of @xmath229 as there are incidence matrices ( up to permutations of columns ) realised by the picture deformations of all the decorated germs with the given topology .
we consider the sandwiched singularities @xmath67 where @xmath5 is an ordinary 6-uple singularity and @xmath6 takes the constant value @xmath230 .
the total multiplicity of each component of @xmath5 is equal to @xmath84 with respect to @xmath5 , therefore one has to blow up 4 times on the strict transforms of the branches of @xmath5 , starting from their separation produced by the blow - up of @xmath48 . as a consequence ,
@xmath1 is a standard decorated germ and the sandwiched singularity @xmath67 has as dual graph of its minimal resolution a star as in figure [ fig : sixuple ] , where all the vertices distinct from the central one have attached self - intersection @xmath231 . in particular , one may choose for @xmath5 the union of six pairwise distinct lines passing through the origin of @xmath232 . as explained in @xcite ,
one gets _ in general _
@xmath233 possibilities for the incidence matrix of a picture deformation obtained by simply moving the lines by parallelism .
but for special positions of the lines , one gets _ one more _ incidence matrix , the configuration generating it being drawn in figure [ fig : speconf ] . by theorem [ lowbound ] , we see that the contact boundary of such sandwiched singularities has at least @xmath234 stein fillings , up to orientation - preserving diffeomorphisms fixed on the boundary .
we expect that these are _ all _ the stein fillings of that contact manifold , up to strong diffeomorphisms .
popescu - pampu , p. _ the geometry of continued fractions and the topology of surface singularities_. in _ singularities in geometry and topology 2004_. advanced studies in pure mathematics * 46 * , 2007 , 119 - 195 . | the _ sandwiched _ surface singularities are those rational surface singularities which dominate birationally smooth surface singularities . de jong and van straten showed that one can reduce the study of the deformations of a sandwiched surface singularity to the study of deformations of a 1-dimensional object , a so - called _ decorated _ plane curve singularity . in particular , the milnor fibers corresponding to their various smoothing components may be reconstructed up to diffeomorphisms from those deformations of associated decorated curves which have only ordinary singularities .
part of the topology of such a deformation is encoded in the _ incidence matrix _ between the irreducible components of the deformed curve and the points which decorate it , well - defined up to permutations of columns . extending a previous theorem ofours , which treated the case of cyclic quotient singularities ,
we show that the milnor fibers which correspond to deformations whose incidence matrices are different up to permutations of columns are not diffeomorphic in a strong sense .
this gives a lower bound on the number of stein fillings of the contact boundary of a sandwiched singularity . |
quantum correlations have played a central role from the very beginning of quantum mechanics , and currently the concept of entanglement has become a key resource in the research on quantum information and quantum computation @xcite .
the availability of entangled quantum states and their characterization are among the most important questions in quantum information . in this sense
, there have been a considerable number of theoretical works that have allowed for characterizing quantum entanglement . from a mathematical point of view ,
a quantum state of a pair of particles is separable if this can be written as a convex sum of product states @xcite , @xmath0 , where @xmath1 and @xmath2 correspond to different particles in the pair and @xmath3 .
actually , separability criteria @xcite provides us with necessary and sufficient conditions to state if a bipartite quantum state is separable or not .
however , these criteria do not allow us to quantify exactly the amount of entanglement of a given system .
bennet _ et.al . _
@xcite have characterized the necessary and sufficient channel resources to transmit quantum states .
this leads to a measure of entanglement that is called _ entanglement of formation_. in other important work , wootters @xcite has shown that the entanglement of formation of an arbitrary state of two qubits is defined by an exactly calculable quantity called _
concurrence_. on the other hand , recent advances in the manipulation of collections of atoms have led to the possibility of considering inhomogeneous coupling , for example , in systems of atomic clouds .
some contributions have emerged inspired by this fact .
for instance , decoherence of collective atomic states due to the inhomogeneous coupling between the atoms and external fields has been studied by sun _ et.al . _
also , kuzmich _ et.al .
_ @xcite has pointed out the nonsymmetric character of the entanglement of multi - atom quantum states due the inhomogeneity .
recently , decoherence in process of quantum information storage in atomic clouds has been studied @xcite , where the phenomenon of electromagnetically induced transparency ( eit ) is present .
atomic cloud physics could be conveniently described by the dicke model @xcite , when considering the interaction of atoms with light in free space .
the tavis - cummings ( tc ) model @xcite , is suitable when the coupling takes place inside a cavity . in most applications of the tc model
, a constant coupling between the atoms and the radiation field is assumed .
this simplification is , of course , essential when an analytical description of the coupled system is required @xcite , at least when a small number of atoms is involved .
for the case of many atoms , the dynamics can still profit from the associated @xmath4 group structure and numerical solutions can be found .
however , the situation is drastically different when we consider the more realistic inhomogeneous coupling . in this case
, there is no possibility to access the hilbert space in a simple manner , because all angular momenta representations are mixed along the dynamics and no analytical approach is known .
the entanglement properties of the ground state for the dicke model has recently been studied by authors in ref .
@xcite . in such case , when the coupling between atoms and field is homogeneous , the concurrence between two atoms is independent of the pair considered due to the symmetry of the dicke states .
a similar situation should arise in the case of the tavis cummings model . from the point of view of the model properties
, it would be interesting to study the quantum correlations between atoms under more general conditions , as is for the case of inhomogeneous coupling .
an interesting question to answer would be , how does the spatial profile of the coupling reflect on entanglement between atoms ? .
in this work we aim to describe the entanglement properties in the spectrum eigenstates of the tc model @xcite by considering inhomogeneous coupling of the atoms to the quantum electromagnetic field .
specifically we study the bipartite atomic concurrence by tracing out @xmath5 particles .
this work is organized as follows : in sec .
ii we present the inhomogeneous tavis - cummings ( itc ) model and the properties of the hilbert space are shown . in sec .
iii , we study the properties of the eigenstates of the model , the bipartite concurrence between atoms is analyzed for different number of excitations in the system . in sec .
iv , we present our concluding remarks .
the hamiltonian describing the inhomogeneous interaction of @xmath6 atoms with a single mode of a quantum electromagnetic field in the on resonance regime is given by @xmath7 where @xmath8 besides we have taken @xmath9 , @xmath10 and @xmath11 are the atomic collective operators with @xmath12 .
the number of excitation operator @xmath13 is an integral of motion for the system @xcite .
in addition @xmath14 is the inhomogeneous coupling of atoms to the field . in this work
, we will consider @xmath15 .
we notice that due to the inhomogeneous coupling , the collective atomic operators do not satisfy a closed lie algebra @xmath16 , leading to a larger total hilbert space , so that , the description in terms of symmetric dicke states is not allowed anymore .
an effective approach to study the dynamics of systems with inhomogeneous coupling can be developed @xcite .
the basic idea of this approach is to follow the hilbert space that the coupled system will visit along the evolution and to implement a truncation criteria based on a probabilistic argument . in the homogeneous case
the eigenstates of @xmath17 are built as superpositions of eigenstates of @xmath18 , that is , states with a fixed number of excitations .
thus , the ground state corresponds to the state with all atoms in the lower state .
the first excited state corresponds to a superposition of one photon state with all atoms in the ground and the state with zero photons and the symmetric state with one excitation . in the inhomogeneous case
we have to find which are the states with a fixed number of excitations .
for example , let us consider the state @xmath19 photons and @xmath20 denotes the collection of @xmath6 atoms in the ground state @xmath21 .
let us see now how the state @xmath22 is coupled to other states under the interaction hamiltonian ( [ ham ] ) which conserves the number of excitations .
we see that the state @xmath23 is not coupled by the interaction hamiltonian to other states of the global system , that is , @xmath24 this state is coupled through the term @xmath25 so that @xmath26 where @xmath27 represents the state where the @xmath28-th atom is excited .
we define the normalized state @xmath29 such that the matrix element of the hamiltonian between @xmath30 states is given by @xmath31 as we can see , the number of excitations is conserved .
the resulting state is coupled to the state @xmath32 in the system , for example @xmath33 , one would expect that the hilbert space will be spanned by the vectors @xmath34 and @xmath35 however , it can be shown that the state @xmath36 is coupled through @xmath37 to the state @xmath38 where @xmath39 , which is different from @xmath40 , so we can say that @xmath41 has a component along @xmath42 and a component along a state @xmath43 perpendicular to the state @xmath40
thus , if the number of excitations keep growing , then it is possible to generate a sequence of collective atomic states and its respectively perpendicular states as shown by fig .
[ esquema ] , where the coupling among atomic collective states are shown .
it is possible to show that the states @xmath44 ( first row in fig .
[ esquema ] ) can be obtained from the repeated application of the operator @xmath45 over the state @xmath46 @xmath47 this implies that the collective atomic states @xmath44 are given by @xmath48 with @xmath49 . on the other hand , the sequence of states @xmath50 in the second row of fig .
[ esquema ] are given by @xmath51 , \ ] ] where the states @xmath52 are written as @xmath53 and @xmath54 an accurate description of the model should consider all the states we have found associated with the inhomogeneity .
however , based on numerical calculations we have obtained that the spectrum of the hamiltonian is not sensitive to perpendicular states further than the first row . to the energy as a function of the number of atoms @xmath6 .
@xmath55 , where @xmath56 are the energies for @xmath57 excitations using @xmath58 or @xmath59 rows of fig . [ esquema].,width=264 ] this is shown in fig .
[ deltae ] , where the effects of the second row on the energies obtained from the first row are quantified . in this figure
we observe that the contribution of the second row in fig .
[ esquema ] is less than @xmath60 .
although the contribution of the second row in fig .
[ esquema ] is higher when the number of excitations increases , decreases when the number of atoms is increased , as can be seen in fig .
[ deltae ] . in other words
, the contribution of the second row to the energy , will increase only when @xmath61 increases , where @xmath57 is the number of excitations .
thus , we can conclude from fig .
[ deltae ] that only the first row of fig .
[ esquema ] is needed in order to characterize the spectrum of the itc model for the parameters used in this work . on the other hand , it is important to note that this technique of taking into account only a few rows in fig .
[ esquema ] is also valid when the quantum dynamics of the system is studied . in that case
the approximation is valid within a given window of time .
the size of this window of time will depend on both @xmath6 and @xmath57 @xcite .
in the previous section we have discussed the necessary ingredients to deal with the itc model .
particulary , we have shown that only a part of the hilbert space associated to the system is necessary for a suitable description . in this section
we study the properties of the eigenstates of the system , by presenting some analytical calculations that allow us to characterize the behavior of the entanglement between a pair of particles .
the spectrum and bipartite atomic concurrence @xcite , can be analytically calculated in the inhomogeneous case for @xmath62 and @xmath59 excitations respectively .
the energies for these cases are given by : @xmath63 where @xmath64 and @xmath65 , correspond to the normalization of the states with one and two excitations in eq .
( [ nbar ] ) respectively .
these energies are associated with the eigenstates @xmath66 from the expressions for eigenstates in eq .
( [ autoestados ] ) we are able to calculate an analytical expression for the bipartite concurrence of a pair of atoms @xmath28 , @xmath67 by tracing out respect to the other @xmath5 atoms . for this cases the concurrence is given by : @xmath68 , \end{array } \label{conc1}\ ] ] with @xmath69 , and @xmath70 .
it is important to note from these expressions for the concurrence that while in the homogeneous tc model the bipartite concurrence of a pair of atoms is independent of the chosen pair , the apparition of inhomogeneity in the coupling between the atoms to the quantum field causes the bipartite concurrence to be dependent on the pair of atoms that we choose .
as is apparent from these equations , the entanglement between a pair of atoms is proportional to the coupling constants of atoms to the field .
thus a pair of atoms located in a region of strong coupling to the field will be more entangled . as we will see these dependence of entanglement on the coupling constant appears considering more excitations . between the first and the @xmath67-th atom for @xmath71 excitation for different number of atoms @xmath6.,width=264 ] fig .
[ c1j ] shows the bipartite concurrence @xmath72 , i.e. , the concurrence between the first and the @xmath67-th atom for different total numbers of atoms @xmath6 .
all atoms are equally spaced for each case .
we observe in this picture the effect of the inhomogeneous coupling of atoms with the field .
the concurrence exhibit a reminiscence of the spatial profile of the single mode electromagnetic field .
the maximum of entanglement happens for the strongest coupling of the @xmath67-th atom , which occur for this atom located at the center the cavity .
comparing the entanglement of the first atom with the atom located around the center of the cavity , we realize that as the number of atoms @xmath6 inside the cavity is increased , entanglement decreases .
this is in agreement with that found in the homogeneous case . in the most general case
, the state with @xmath57 excitations can be written as @xmath73 where @xmath74 is the collective atomic state with @xmath75 excitations defined in eq .
( [ nbar ] ) and @xmath76 is a coefficient arising from the eigenvalue problem .
tracing out the quantum field , the reduced density matrix for the atomic subsystem is given by @xmath77 now , tracing out @xmath5 atoms , the bipartite concurrence between atoms @xmath78 will be given by @xmath79^{1/2 } ] , \label{conc}\end{aligned}\ ] ] where we have defined @xmath80 between the first and the last atom for @xmath81 excitations as a function of the number of atoms @xmath6.,width=264 ] between the first and the @xmath67-th atom for @xmath81 excitations with @xmath82 atoms.,width=264 ] as we can see in the general expression for the concurrence ( [ conc ] ) for @xmath57 excitations and @xmath6 atoms , the entanglement between two of these atoms still depends directly on the coupling constant .
as shown by this result this is one important feature of entanglement that the tc model exhibit for the inhomogeneous case .
this makes the difference with the the situation for the homogeneous case where only depends on the total number of atoms @xmath6 for a given number of excitations @xmath57
@xcite . fig .
[ c1n2 ] shows the concurrence between the first atom and the last atom as a function of the number of atoms inside the cavity . in this picture
we can observe how the concurrence decreased when the number of total excitations @xmath83 in the system increases .
however , when @xmath84 the concurrence becomes less dependent on the number of excitations tending to a fixed value . fig .
[ c1jk ] , as fig .
[ c1j ] shows the concurrence @xmath72 between the first and the @xmath67-th atom in the cavity .
the effects of the inhomogeneous coupling and the loss of entanglement when the number of excitation is increased are also clear in this figure .
in the present work we have found that although the symmetry @xmath4 no longer exits in the itc model and consequently a treatment in terms of symmetric dicke states is not feasible , there are still some features that can be analyzed . in particular , the bipartite concurrence of the eigenstates of the model can be obtained exhibiting an explicit dependence on the inhomogeneity in addition to the dependence on the number of atoms and the number of excitations . c. h. bennett , g. brassard , s. popescu , b. schumacher , j. a. smolin , w. k. wootters phys.rev.lett . * 76*,722 - 725 ( 1996 ) ; c. h. bennett , d. p. divincenzo , j. a. smolin , w. k. wootters , phys.rev .
a , * 54 * 3824 - 3851(1996 ) | in this work we study the properties of the atomic entanglement in the eigenstates spectrum of the inhomogeneous tavis - cummings model .
the inhomogeneity is present in the coupling among the atoms with quantum electromagnetic field .
we calculate analytical expressions for the concurrence and we found that this exhibits a strong dependence on the inhomogeneity . |
high temperature superconductivity was discovered in cuprates in 1986.@xcite the rapid raising of the transition temperature to well above the melting point of nitrogen @xcite shattered the old record of 23 k. furthermore , the fact that high @xmath0 superconductivity was discovered in a rather unexpected material , a transition metal oxide , made it clear that some novel mechanism must be at work .
the intervening years have seen great strides in high @xmath0 research .
the growth and characterization of cuprate single crystals and thin films have advanced to the point where sample quality and reproducibility problems which plagued the field in the early days are no longer issues . at the same time , basically all conceivable experimental tools have been applied to the cuprates .
indeed , the need for more and more refined data has spurred the development of experimental techniques such as angle resolved photoemission spectroscopy ( arpes ) and low temperature scanning tunneling microscopy ( stm ) .
today the cuprate is arguably the best studied material outside of the semiconductor family and a great deal of facts are known .
it is also clear that many of the physical properties are unusual , particularly in the metallic state above the superconductor .
superconductivity is only one aspect of a rich phase diagram which must be understood in its totality .
it is often remarked that there is no consensus for the mechanism of high @xmath0 superconductivity .
this may be true but i must emphasize that a lack of consensus is not synonymous with a lack of understanding or lack of progress .
in fact , i will argue that the basic physics of the cuprate family is well understood . while there are hundreds of high @xmath0 compounds , they all share a layered structure which contains one or more copper - oxygen planes .
the low energy physics of these planes can further be simplified to a model of electrons hopping on a square lattice called the one band hubbard model.@xcite ( many of the details left out in this paper can be found in a more comprehensive review.@xcite ) @xmath4 where @xmath5 is the creation operator of an electron with spin @xmath6 on a square lattice , @xmath7 , @xmath8 is the hopping matrix element between sites @xmath9 and @xmath10 , ( we shall denote nearest neighbor hopping by @xmath11 and further neighbor hopping by @xmath12 , @xmath13 etc . ) , and @xmath14 is the repulsive energy cost due to screened coulomb interaction to put two electrons with opposite spin on the same site . at half - filling ( one electron per site ) , there is a metal - to - insulator transition as the ratio @xmath15 is in increased . the insulator is called a mott insulator,@xcite as opposed to a band insulator , because its existence is driven by strong repulsion and is not described by band theory . the latter would require the state to be metallic at half - filling . the term `` strong correlation physic '' is now used to describe the state of affairs when interaction energy dominates the kinetic energy in controlling the electron motion . as seen in fig . 1 , for large enough @xmath15 the electrons prefer
to be localized on the lattice site because any hopping to reduce the kinetic energy @xmath11 requires double occupation of some site , which costs @xmath14 .
this insulator is also predicted to be antiferromagnetic ( af ) , because af alignment permits virtual hopping to gain an energy @xmath16 by second order perturbation theory , whereas hopping is strictly forbidden by pauli exclusion for parallel spins .
it is now generally agreed that the parent compound of the high @xmath0 cuprate is a mott insulator .
copper atoms sit on a square lattice with oxygen atoms in between .
the electronic structure is simplified to a one band model shown on the right , with electrons hopping with matrix element @xmath11 .
there is an antiferromagnetic exchange @xmath17 between spins on neighboring sites.,scaledwidth=50.0% ] things get interesting when electron vacancies ( called holes ) are introduced into the copper - oxygen layers in a process called hole doping , i.e. a charge reservoir away from the copper - oxygen plane is introduced which removes electrons for the plane .
we denote the concentration of holes by @xmath18 .
the resulting phase diagram is schematically shown in fig . 2 .
the af order is rapidly destroyed by a few percent of holes , beyond which superconductivity appears .
the transition temperature reaches a maximum around 15% doping , which is called optimal doping .
the dome - shaped @xmath0 is characteristic of all hole doped cuprates , even though the maximum @xmath0 seems to be clustered around two groups .
it is about 40 k in the la@xmath19sr@xmath20cuo@xmath21 ( lsco ) family and 93 k and higher in a second family which includes yba@xmath22cu@xmath23o@xmath24 ( ybco ) and ba@xmath22sr@xmath22cacu@xmath22o@xmath25 ( bi-2212 ) .
the highest @xmath0 at ambient pressure of 135 k was reached in hgba@xmath22ca@xmath22cu@xmath23o@xmath26 in 1993.@xcite before high @xmath0 superconductors were discovered in 1986 , almost all superconductors were believed to be @xmath27-wave bcs superconductors . the requisite attractive interaction to form pairs
is provided by electrons exchanging a phonon .
there were a few potential exceptions in a class of strongly correlated metals called the heavy fermions , but @xmath0 was very low , typically less than 1 k.@xcite . this is why the discovery of superconductivity in a system where the repulsion is strong enough to create a mott insulator was such a surprise .
however , in the intervening years , we know of many examples of non @xmath27-wave superconductors .
these all occur in strongly correlated materials and are clearly not driven by electron - phonon coupling . in the heavy fermion systems ,
experimental progress has removed any doubt about the non-@xmath27-wave nature of a large number of compounds.@xcite other new examples include sr@xmath22ruo@xmath21 exhibiting triplet pairing which breaks time - reversal symmetry;@xcite hydrated cobaltates,@xcite , and several systems near the quantum critical point of a zero temperature ferromagnetic ( uge@xmath22)or antiferromagnetic ( cepd@xmath22si@xmath22 , cein@xmath23)phase transition.@xcite the transition temperature of these systems remain low , less than 5 k. of particular interest to the present discussion is the superconductivity discovered in a series of layered organic molecular solids.@xcite this is because these compounds also live in the vicinity of the mott transition . unlike the cuprates , they are not doped and remain at half filling . in these materials
the effective hopping parameter @xmath11 is sensitive to pressure and to the choice of anion molecules .
the ratio @xmath15 can be tuned right through the mott transition.@xcite amazingly it was discovered that when the mott insulator is destroyed , the system immediately becomes a superconductor , before becoming a metal at even higher pressure .
furthermore , the transition temperature reaches 11.6 k , the highest known among the organics .
there is also strong evidence that these superconductors have @xmath28-wave pairing symmetry.@xcite materials .
the antiferromagnet ( af ) is rapidly destroyed by doped holes .
the @xmath28-wave superconductor is subject to strong phase fluctuations below the dotted line , where the proliferation of vortices has been detected by the nernst effect .
a pseudogap region extends up to high temperatures in the underdoped region.,scaledwidth=40.0% ] i would argue that 11.6 k for an organic metal qualifies it as an example of a high @xmath0 superconductor !
the reason is that the electronic energy scale for organic solids is much smaller that that for ordinary solids . for example , the hopping matrix element @xmath11 is about 0.05 ev compared with 0.4 ev for the cuprates .
thus the ratio @xmath29 is about the same for both systems . to emphasize this point , in fig 3
i have put both materials on the same phase diagram in the parameter space @xmath15 and @xmath18 of the hubbard model .
is the @xmath28-wave superconductor that appears with doping connected with the one that appears under pressure ?
we do not have the answer at present .
my point is that with so many `` unconventional '' examples , our mindset today should be different from that of 20 years ago , and we should be more receptive to the idea that superconductivity may be a highly competitive ground state in a pure repulsive model such as the hubbard model .
cuprates and organic superconductors in the hubbard model phase diagram . at half
filling , the antiferromagnetic insulator onsets when @xmath15 exceeds a critical value @xmath30 , where the mott transition occurs .
high @xmath0 superconductivity occurs when holes are doped into the mott insulator over a concentration range between 6% and 25% . in certain organic compounds ,
12 k superconductivity lives on the boundary between the mott insulator and the metal .
the ratio @xmath31 is about @xmath32 for both systems .
whether the two superconducting regions are connected is not known and indicated by the question mark.,scaledwidth=45.0% ] with that remark let us return to the cuprates and examine the phase diagram in more detail .
the region between the disappearance of af and the onset of superconductivity is complicated by disorder effects , even though heroic efforts to make pure samples of ybco have yielded interesting new information.@xcite we shall not discuss this region further .
the regions of the phase diagram with doping to the left and right of optimal is called underdoped and overdoped , respectively .
the metallic state above @xmath0 in the underdoped region has been under intense study and exhibits many unusual properties not encountered before in any other metal .
this is shown below the dashed line in fig .
2 and has been called the pseudogap phase .
it is not a well - defined phase in that a definite finite temperature phase boundary has never been found , so the dashed line should be regarded as a cross - over .
there is now broad agreement that the high @xmath0 problem is synonymous with that of doping of a mott insulator .
it then makes sense to focus on the underdoped region , where the battle line between mott insulator and superconductivity is drawn . since we are interested in the case where @xmath14 is sufficiently large compared with @xmath11 for the hubbard model to be in the mott insulator phase , it is useful to expand in @xmath33 .
the leading order result is the @xmath11-@xmath17 model @xmath34 p .\ ] ] the second term is the af heisenberg exchange between local spins @xmath35 discussed earlier .
the nontrivial part of the @xmath11-@xmath17 model resides in the projection operator @xmath36 which restricts the hilbert space to exclude the doubly occupied states .
the strong coulomb repulsion now becomes a constraint of no double occupation .
compared with the hubbard model , the hilbert space is reduced from four states per site to three , namely spin up , spin down or empty .
the parameters of the @xmath11-@xmath17 model appropriate for the cuprates is also well established .
ev @xmath38 k , @xmath39 and @xmath40 is negative , of order @xmath41 , and is believed to vary somewhat from compound to compound.@xcite equations ( 1 ) and ( 2 ) are deceptively simple looking hamiltonians which have defied accurate numerical or analytic solution .
nevertheless , the belief among many workers in the field is that they contain the rich physics of the high @xmath0 phase diagram .
the situation is not unlike qcd , where the lagrangian is known , but precise understanding of confinement and the mass spectrum has just begun to emerge from quantum monte carlo after decades of hard work . to make matters worse ,
the high @xmath0 problem at finite doping is analogous to the qcd problem with finite quark density,@xcite where accurate numerical solution is so far not possible due to the fermion sign problem . on the other hand , unlike the quark - gluon problem the high @xmath0 problem has a lot more experimental constraint and input . as a result we know a lot about the high @xmath0 phenomenology which severely limits the theoretical options .
let us start with some simple common sense arguments to gain some insight into the nature of the problem of a doped mott insulator .
consider a single hole hoping in an af background as shown in fig . 1 .
after one hop we find a spin surrounded by ferromagnetic neighbors , costing an energy of @xmath42 from the 3 ferromagnetic bonds if the spins are treated as classical @xmath43 .
there is a competition between the exchange energy @xmath17 and the desire of the hole to hop in order to gain the kinetic energy @xmath11 per hole . for large enough doping the kinetic energy wins and we expect a metallic state with some short range af correlation . by comparing @xmath44 and @xmath17
, we expect this to onset at @xmath45 , in good agreement with the experimental finding .
this state should be a fermi liquid state .
there is a powerful theorem in landau fermi liquid theory commonly called the luttinger theorem@xcite which states that the area of the fermi surface is the same as that of free fermions , i.e. , it is determined by the total density of electrons in the unit cell . in our case
the area is @xmath46 where @xmath47 is the area of the brillouin zone ( bz ) .
this is exactly what is found experimentally . in fig .
4(d ) we show an example of the measured fermi surface .
the precise shape can be fitted with a hopping model with further neighbor hopping .
the opposite limit of a few holes @xmath48 hopping in an af background is less trivial , but by now reasonably well understood .
the competition with the af exchange causes the effective hopping matrix element to be renormalized downward from @xmath11 to @xmath17.@xcite the quasiparticle nevertheless manages to form coherent bands .
the bands have minima at @xmath49.@xcite with finite doping the fermi surfaces are ellipses centered at @xmath49 as shown in fig .
note that the unit cell is doubled because of af ordering and the bz is reduced to the diamond in fig .
applying luttinger theorem to the doubled unit cell , the total area of the fermi surface in the reduced bz is now @xmath50 where @xmath51 .
therefore we conclude that the area of each ellipse ( hole pocket ) is @xmath52 .
physically it makes sense that transport properties are determined only by @xmath18 carriers occupying small fermi pockets .
the theory of a few holes in af background is quite well developed , and recently papers applying effective field theory approach borrowed from the particle physics literature are particularly notable.@xcite we have good understanding of @xmath53 and @xmath54 . what happens in between ?
here we run into a dilemma . we know that af order is destroyed for @xmath55 , beyond which points we have no indication of unit cell doubling .
if fermi liquid theory were to hold , what would happen to luttinger theorem ?
recall that the nice physical picture of small hole pockets rely on the unit cell doubling .
once that is absent , luttinger theorem forces us to have a `` large '' f.s .
, i.e. one with area proportional to @xmath56 . in that case
it will be difficult to see how transport properties will continue to look as if it is given by @xmath18 holes .
we note that while the original derivation of the luttinger theorem was perturbative in the interaction strength , the modern derivation by oshikawa@xcite is a topological one and relies on very few assumptions , not much beyond the statement that well defined quasiparticles exist . in principle , the fermi liquid can develop a heavy mass @xmath57 so that the conductivity spectral weight @xmath58 , but experimentally there is no evidence of such heavy mass formation
. parenthetically we point out that the three dimensional example of doped mott insulator la@xmath19sr@xmath20tio@xmath23 appears to take the heavy mass route.@xcite it turns out that nature solves this problem in an extremely clever and unexpected way .
as far as the ground state is concerned , the question is moot because it appears that once af is destroyed the system becomes superconducting and the luttinger theorem can not be applied .
what about the normal state above the superconducting @xmath0 ?
the extensive work using angle resolved photoemission spectroscopy ( arpes ) has shown that the gapless excitations lie on an arc.@xcite anywhere apart from the arc , the excitations are gapped . ) .
colors indicate the intensity of low energy excitations .
data from plat _ et al._@xcite ( e ) experimental data showing the fermi arc in one quadrant of figure 4(c ) in underdoped na@xmath19ca@xmath20cu@xmath22o@xmath22cl@xmath22 @xmath59 .
data from k. shen _
et al._@xcite , scaledwidth=50.0% ] this situation is sufficiently strange that requires a bit more explanation in terms of the experimental observation .
arpes measures the spectrum of occupied electron states which can be removed by excitations with a photon , i.e. it measures the hole spectral function .
a spectrum is measured at every @xmath60 point . in a fermi liquid
the spectrum consists of a quasiparticle peak at energy @xmath61 .
as one moves along line a in fig .
4(c ) , the peak approaches the fermi energy and disappears as @xmath61 crosses the fermi surface , thus locating its position .
this is how the fermi surface in fig.4(b ) is mapped out . in the underdoped case ,
what happens is that along line b , the quasiparticle peak ( now quite broad ) approaches the fermi energy but stops before crossing it .
instead it loses weight and disappears .
you might say that this also happens in fig.4 ( a ) , if path b misses the hole pocket .
the important difference is that in the case of the hole pocket where there is unit cell doubling , in the extended zone scheme we expect the occupied band to live also outside the reduced bz .
thus if we follow path a we should see a quasiparticle peak appearing at the second crossing of the ellipse .
further along path a this peak will then move down in energy away from the fermi energy .
in other words , the back side of the hole pocket should be visible in arpes as an occupied quasiparticle state rises up to meet the fs .
the surprise is that in fig .
4(c ) , there is no back side to the ellipse , and one is left with what is called a fermi arc .
the spectrum is gapped everywhere except on the arc , and the gap reaches a maximum near @xmath62 . in the superconducting state
the arc also becomes gapped , leaving a single gapless point along @xmath63 called the nodal point . in this way the quasiparticle spectrum of a @xmath28 wave superconductor is smoothly formed out of the pseudogap normal state .
the size of the maximum gap has been measured as a function of doping , and is found to increase with decreasing @xmath18 , in a way which tracks the onset of the pseudogap phase shown in fig . 2 .
( a word of caution : the electron spectral function near @xmath62 , the so called anti - nodal direction , is extremely broad and , unlike the nodal directions , does not show a quasiparticle peak even in the superconducting state in strongly underdoped samples .
so the gap is measured as the pull - back of the leading edge of the spectrum for the fermi energy .
this picture is corroborated by extensive scanning tunneling microscopy ( stm ) work , which has excellent energy and spatial resolution , but no momentum space information .
the fermi arc and anti - nodal gap is one part of a rich phenomenology associated with the pseudogap phase .
the first evidence of an energy gap comes from nmr data , which found that in underdoped samples the knight shift , which measures the spin susceptibility , does not follow pauli s prediction of being temperature independent , but starts dropping around room temperature and has lost about 80% of its value by the time @xmath0 is reached.@xcite the gap also shows up in @xmath64-axis frequency dependent conductivity,@xcite where electrons move from one layer to the next , but the transport within the plane remains metallic , being dominated by the quasiparticles near the fermi arc .
what is the origin of the large gap at @xmath62 ?
one suggestion is that it is simply a @xmath28-wave superconducting gap and the pseudogap phase should be understood as a superconductor destroyed by strong phase fluctuations .
phase fluctuation is controlled by the superfluid stiffness which is given by the superfluid density .
thus @xmath0 should be proportional to the superfluid density , i.e. decreases with decreasing @xmath18.@xcite this picture is made more quantitatively accurate if the reduction of the superfluid density due to thermal excitations of nodal quasiparticles is taken into account.@xcite the opposite trend of the energy gap and @xmath0 is explained , but the origin of such a large superconducting gap close to an insulator becomes even a greater puzzle .
experimentally this is now ample evidence that the picture of fluctuating superconductivity indeed survives up to much higher temperature compared with a conventional metal : perhaps 2 or 3 times @xmath0 .
however , that scale decreases with decreasing doping and does not reach up to the pseudogap temperature .
the key experiment that mapped out this region is the nernst effect , which is sensitive to mobile superconducting vortices.@xcite for this reason we refer to this region as the nernst region in fig . 2 .
if the pseudogap is not a pairing gap , it presents a great challenge for theory , because while the fermi arc scenario interpolates between the small hole pocket and large fermi surface beautifully , it is not allowed by conventional band theory or fermi liquid theory .
fermi surfaces do not simply terminate .
this is a situation not encountered before in solid state physics .
this is why the psuedogap phenomenon is considered one of the central mysteries of the high @xmath0 story .
we shall return to this issue in section v.
while the pseudgap phenomena is really strange , to a large extent it was anticipated by theory . here
i am referring to the concept of the resonating valence bond ( rvb ) introduced by p.w . anderson@xcite and the slave boson mean field theory and elaborations which followed . here
we provide a brief review .
we explained in the last section that the nel spin order is incompatible with hole hopping .
the question is whether there is another arrangement of the spin which achieves a better compromise between exchange energy and the kinetic energy of the hole . for @xmath65
it appears possible to take advantage of the special stability of the singlet state .
the ground state of two spins @xmath66 coupled with antiferromagnetic heisenberg exchange is a spin singlet with energy @xmath67 .
compared with the classical large spin limit , we see that quantum mechanics provides an additional stability in the term unity in @xmath68 and this contribution is strongest for @xmath65 .
let us consider a one - dimensional spin chain .
a nel ground state with @xmath69 gives an energy of @xmath70 per site . on the other hand , a simple trial wavefunction of singlet dimers already gives a lower energy of @xmath71 per site .
this trial wavefunction breaks translational symmetry and the exact ground state can be considered to be a linear superposition of singlet pairs which are not limited to nearest neighbors , resulting in a ground state energy of 0.443 j. in a square and cubic lattice the nel energy is @xmath72 and @xmath73 per site , respectively , while the dimer variational energy stays at @xmath71 .
it is clear that in a 3d cubic lattice , the nel state is a far superior starting point , and in two dimensions the singlet state may present a serious competition .
historically , the notion of a linear superposition of spin singlet pairs spanning different ranges , called the resonating valence bond ( rvb ) , was introduced by @xcite and @xcite as a possible ground state for the @xmath65 antiferromagnetic heisenberg model on a triangular lattice .
the triangular lattice is of special interest because an ising - like ordering of the spins is frustrated .
subsequently , it was decided that the ground state forms a @xmath74 superlattice where the moments lie on the same plane and form @xmath75 angles between neighboring sites.@xcite soon after the discovery of high @xmath0 superconductors , anderson @xcite revived the rvb idea and proposed that with the introduction of holes the nel state is destroyed and the spins form a superposition of singlets .
the vacancy can hop in the background of what he envisioned as a liquid of singlets and a better compromise between the hole kinetic energy and the spin exchange energy may be achieved .
many elaborations of this idea followed , but here we argue that the basic physical picture described above gives a simple account of the pseudogap phenomenon .
the singlet formation explains the decrease of the uniform spin susceptibility .
the vacancies are responsible for transport in the plane .
the conductivity spectral weight in the @xmath76 plane is given by the hole concentration @xmath18 and is unaffected by the singlet formation . on the other hand , for @xmath64-axis conductivity , an electron is transported between planes .
since an electron carries spin @xmath3 , it is necessary to break a singlet .
this explains the gap formation in @xmath77 and the energy scale of this gap should be correlated with that of the uniform susceptibility . in photoemission ,
an electron leaves the solid and reaches the detector , the pull back of the leading edge simply reflects the energy cost to break a singlet .
a second concept associated with the rvb idea is the notion of spinons and holons , and spin charge separations .
anderson postulated that the spin excitations in an rvb state are @xmath65 fermions which he called spinons .
this is in contrast with excitations in a nel state which are @xmath78 magnons or @xmath79 gapped singlet excitations .
-axis conductivity experiment .
this necessitates the breaking of a singlet .
, scaledwidth=50.0% ] initially the spinons are suggested to form a fermi surface , with fermi volume equal to that of @xmath56 fermions.@xcite later it was proposed that the fermi surface is gapped to form @xmath28-wave type structure , with maximum gap near @xmath80.@xcite this @xmath60 dependence of the energy gap is needed to explain the momentum dependence observed in photoemission .
the concept of spinons is a familiar one in one - dimensional spin chains where they are well understood to be domain walls . in two dimensions
the concept is a novel one which does not involve domain walls .
instead , a rough physical picture is as follows .
if we assume a background of short range singlet bonds , forming the so - called short - range rvb state , a cartoon of the spinon is shown in fig .
if the singlet bonds are `` liquid , '' two @xmath81 formed by breaking a single bond can drift apart , with the liquid of singlet bonds filling in the space between them .
they behave as free particles and are called spinons .
the concept of holons follows naturally @xcite as the vacancy left over by removing a spinon .
a holon carries charge @xmath82 but no spin .
is there any calculated tool or mathematical formalism to put some meat into the physical picture of rvb described in the last section ? as far as computation is concerned , the use of the projected wavefunction has enjoyed considerable success .
the idea is to write down a trial wavefunction of the type @xmath83 where @xmath84 is called the gutzwiller projection and @xmath85 is any hartree fock or bcs wavefunction , usually suggested by mean field theory described below .
the role of the gutzwiller projection is to remove all doubly occupied states in @xmath85 .
equation ( 3 ) is a suitable variational wavefunction for the @xmath11-@xmath17 model because the constraint is satisfied by definition , and its expectation values and correlation functions can be computed by efficient monte carlo algorithms.@xcite the mean field parameters can be treated as variational parameters .
the projection wavefunction gives excellent ground state energy and sublattice magnetization at half - filling , capturing the important quantum fluctuations of the nel ordered state . with doping it predicted correctly the @xmath28-wave pairing ground state,@xcite even though the prediction of the co - existence of superconductivity with af up to @xmath86 is not in agreement with experiment . putting aside the question of whether it is the ground state , a comparison of the physical properties of the projected @xmath28-wave pairing states with a variety of experiments was successfully made.@xcite the trial wavefunction
can be further improved by the lanczos method of repeatedly hitting it with the hamiltonian .
there is some controversy as to whether the ground state of the @xmath11-@xmath17 model with nearest - neighbor hopping only is a @xmath28-wave superconductor,@xcite but it is clear that superconductivity is a highly competitive state , as found by other numerical methods such as density matrix renormalization group,@xcite cluster dynamical mean field theory @xcite and variational cluster approximation.@xcite recently it was found that introduction of @xmath12 considerably stabilizes the @xmath28-wave superconducting state.@xcite at present , i would stay that there is strong numerical evidence that the @xmath28-wave superconductor is a strong contender for the ground state of the @xmath11-@xmath17 model .
what about analytic theory and where does the mean field @xmath85 come from ?
a useful method is called the gutzwiller approxmation , which imposes the constraint approximately by treating the available configuration for hopping and exchange in a statistical basis.@xcite this is clearly related to the slave - boson method which we discus below .
the slave - boson method was developed for the kondo problem@xcite it has enjoyed great success as the best way to undersand the properties of a remarkable class of materials called the heavy fermion compounds , where fermi liquid theory has been stretched to the extreme , with effective mass as large as several thousand times the free electron mass.@xcite the idea is to write the electron operator as a product of the boson and fermion which carries the spin index @xmath87 with the condition @xmath88 this constraint can be enforced with a lagrangian multiplier @xmath89 .
note that eq .
( 4 ) is not an operator identity and the right - hand - side does not satisfy the fermion commutation relation .
rather , the requirement is that both sides have the correct matrix elements in the reduced hilbert space with no doubly occupied states .
for example , the heisenberg exchange term is written in terms of @xmath90 , @xmath91 only @xcite @xmath92 we then decouple the exchange term in both the particle - hole and particle - particle channels via the hubbard - stratonovich ( hs ) transformation .
then the partition function is written in the form @xmath93 where @xmath94 \\ & + & \tilde{j } \left [ \sum_{\langle\bm i\bm j\rangle } \delta_{\bm i\bm j } \left ( f^\dagger_{\bm i\uparrow}f^\dagger_{\bm j\downarrow } - f^\dagger_{\bm i\downarrow } f^\dagger_{\bm j\uparrow } \right ) + c.c .
\right ] \nonumber \\ & + & \sum_{\bm i } b_{\bm i}^\ast(\partial_\tau - i\lambda_{\bm i } + \mu_b ) b_{\bm i } - \sum_{\bm i\bm j}{t}_{\bm i\bm j}b_{\bm i}b_{\bm j}^\ast f_{\bm i\sigma}^\dagger f_{\bm j\sigma } , \nonumber \end{aligned}\ ] ] with @xmath95 representing fermion hopping and @xmath96 representing fermion pairing corresponding to the two ways of representing the exchange interaction in terms of the fermion operators .
@xmath97 is chosen to reproduce the mean field self - consistent equation which is obtained by the feynman variational principle .
mean field theory corresponds to the saddle point solution to the functional integral .
the mean field conditions are @xmath98 let us write @xmath99 and ignore the fluctuation of the amplitude . furthermore , in the last term in eq .
( 8) we replace @xmath100 by @xmath101 .
then the rather complicated lagrangian of eq .
( 8) has a rather simple interpretation .
it describes fermions and bosons hopping on a lattice with hopping matrix element @xmath101 .
the phase @xmath102 lives on the links , and plays the role of the spatial components of a lattice gauge field , while the @xmath103 fields introduced to enforce the constraint become the time components .
note that both fermions and bosons are coupled to the same gauge field .
in addition , the fermions may have singlet pairing amplitude given by @xmath104 .
thus we come to the conclusion that the @xmath11-@xmath17 model is equivalent to a lattice gauge theory with non - relativistic fermions and bosons coupled with a _
compact _ @xmath1 gauge field ( compactness simply refers to the fact that the gauge field @xmath102 is a phase defined module @xmath105 ) . at this point the gauge field has no dynamics .
it fluctuates freely and is in the infinite coupling limit .
interesting dynamics emerge upon integrating out some of the matter field but one is left with a lattice gauge theory with strong coupling .
the mapping is basically exact , but the question remains as to how to deal with such a model . before discussing the importance of gauge fluctuations , let us examine some examples of mean field solutions . 1
. @xmath28-wave pairing states . + here
@xmath106 is constant and @xmath107 for @xmath108 bonds along @xmath18 and @xmath109 for @xmath108 along @xmath110 , i.e. it has @xmath28-wave pairing geometry . without pairing ,
the fermions hop on a tight binding band with dispersion + @xmath111 + with pairing we have the classic @xmath28-wave dispersion + @xmath112 + where @xmath113 is the fermion chemical potential and @xmath114 .
the bosons see the same band dispersion and condense at the bottom of the band minimum at low temperatures . in mean field theory
the bose condensation temperature is proportional to the boson density @xmath18 .
below this temperature we have electron pairing , because the bcs order parameter @xmath115 where @xmath116 .
the mean field phase diagram is shown in fig . 6 and captures some key features of the high @xmath0 phase diagram shown in fig . 2 .
in particular , the @xmath28-wave superconducting state appears at intermediate doping and a spin gap state ( region ii ) where a @xmath28-wave - like gap exists for spin excitations but not charge excitations , anticipates many of the properties of the pseudogap phase .
+ mean field theory .
the solid line denotes the onset of the uniform rvb state @xmath117 .
the dashed line denotes the onset of fermion pairing @xmath118 and the dotted line denotes mean field bose condensation @xmath119 .
the four regions are ( i ) fermi liquid @xmath120 , @xmath121 ; ( ii ) spin gap @xmath120 , @xmath122 ; ( iii ) @xmath28-wave superconductor @xmath120 , @xmath122 , @xmath121 ; and ( iv ) strange metal @xmath120 . from lee and nagaosa.@xcite , scaledwidth=40.0% ] 2 . the staggered flux state .
+ early in the development of the mean field theory , a variety of mean field states were discovered which give identical dispersion .
notable among these is the staggered flux state @xcite in this state the hopping @xmath95 is complex , @xmath123 , and the phase is arranged in such a way that it describes free fermion hopping on a lattice with a fictitious flux @xmath124 threading alternative plaquettes . remarkably , the eigenvalues of this problem are identical to that of the @xmath28-wave superconductor given by eq .
( 12 ) , with @xmath125 and @xmath126 the case @xmath127 , called the @xmath128 flux phase , is special in that it does not break the lattice translation symmetry . as we can see from eq .
( 13 ) , the corresponding @xmath28-wave problem has a very large energy gap and its dispersion is shown in fig .
the key feature is that the energy gap vanishes at the nodal points located at @xmath129 . around
the nodal points the dispersion rises linearly , forming a cone which resembles the massless dirac spectrum . for the @xmath128 flux state the dispersion around the node is isotropic . for @xmath130 less than @xmath131
the gap is smaller and the dirac cone becomes progressively anisotropic .
the anisotropy can be characterized by two velocities , @xmath132 in the direction towards @xmath63 and @xmath133 in the direction towards the maximum gap at @xmath80 .
figure shown is for the special case of @xmath128-flux . in general
, the nodal spectra becomes anisotropic . with doping fermi pockets
are formed when the fermi energy crosses the energy spectrum.,scaledwidth=40.0% ] + the reason various mean - field theories have the same energy at half - filling was explained by affleck _ et al_.@xcite and dagotto _
et al_.@xcite as being due to a certain @xmath134 symmetry .
it corresponds to the following particle - hole transformation @xmath135 note that the spin quantum number is conserved .
it describes the physical idea that adding a spin - up fermion or removing a spin - down fermion are the same state after projection to the subspace of singly occupied fermions .
it is then not a surprise to learn that the gutzwiller projection of the @xmath28-wave superconductor and that of the staggered flux state gives the same trial wavefunction , up to a trivial overall phase factor , provided @xmath125 and eq .
( 13 ) is satisfied . a simple proof of this is given by zhang _
et al_.@xcite .
the energy of this state is quite good .
the best estimate for the ground state energy of the square lattice heisenberg antiferromagnet which is a nel ordered state is @xmath136 j.@xcite the projected @xmath128 flux state @xcite gives @xmath137j , which is excellent considering that there is no variational parameter .
+ in the presence of a hole , the dispersion of the staggered - flux phase is still given by eq .
( 12 ) with @xmath125 , but the fermi level is now located at @xmath138 which lies below the node .
this is shown in fig .
if the bosons are condensed , this becomes a fermi liquid state with small hole pockets just like what is shown in fig .
this state has higher energy than the @xmath28-wave superconductor because in a superconductor the node is shifted in @xmath139 space away from @xmath140 towards the origin , but its energy is always tied to the fermi level .
the staggered flux state was proposed by hsu , marston and affleck@xcite in 1991 to be the origin of the pseudogap state .
they pointed out that with doping , the staggered flux state has the remarkable property that orbital currents flow around each square plaquette in a staggered way , so that the physical order parameter of this state is the staggered orbital current .
this proposal did not receive much attention because the appearance of an ordered state with ising order parameter requires a finite temperature phase transition , which has never been seen experimentally .
furthermore , the model requires hole pockets instead of the arcs which were later observed . while the matrix element effect can reduce the spectral weight of the backside,@xcite the model requires sharp quasiparticle peaks which should be observable , especially near the end of the arc , where the weight reduction is only a factor of 2 . in 2002 , chakravarty _
et al._@xcite revived this proposal , arguing that disorder effect may round the transition .
they named this state the @xmath28-density wave ( ddw ) state .
since this state is in fact identical to the staggered flux state of hsu _
et al._@xcite , the two names are sometimes used interchangeably in the literature . however , the philosophy of their approach is very different .
take a phenomenological landau theory approach .
for them the superconducting state is an entirely different order parameter and the two orders compete with each other .
i believe this view misses the key fact that all this is happening close to the mott insulator .
while the @xmath28-wave superconductor and the staggered flux ( ddw ) states are very different states at the mean field level , the mean field states do not respect the no double occupation constraint .
the story changes if we enforce the constraint by applying the gutzwiller projection .
as mentioned earlier , the projected states become identical at half filling and by continuity must share a lot of similarity slightly away from half filling .
for example , we found that at @xmath141 the projected @xmath28-wave superconductor has short - range orbital current order@xcite and presumably the projected staggered flux state has short range @xmath28-wave superconductor order .
both have significant short range af order@xcite .
this kind of consideration motivated wen and i to introduce the @xmath134 gauge theory in 1996 as an improvement over the hsu _
_ proposal.@xcite instead of an ordered state , the pseudogap phase is considered to be a fluctuating state which included the staggered flux state and @xmath28-wave superconductivity on equal footing . this will be described in more detail in the next section .
sf stands for staggered flux . from ivanov.@xcite , scaledwidth=40.0% ] + it is interesting to examine the energetics of various projected states as shown in fig .
it was found that the best state is a projected @xmath28-wave superconductor and the sublattice magnetization is nonzero for @xmath142 , where @xmath143 for @xmath144.@xcite the projected staggered - flux state always lies above the projected @xmath28-wave superconductor , but the energy difference is small and vanishes as @xmath18 goes to zero , as expected .
the staggered - flux state also prefers antiferromagnetic order for small @xmath18 , and the critical @xmath145 is now 0.08 , less than that for the projected @xmath28-wave superconductor .
the projected staggered flux state is the lowest energy nonsuperconducting state that has been constructed so far .
the lesson from examining projected wavefunctions is that the close relationship between different mean field states must be taken seriously .
in fact , the @xmath28-wave state and the staggered flux state are not the only states which have similar energies after projection .
there is a continuous family of states related by the @xmath134 symmetry given by eq .
wen and i developed a formulation to take this into account by introducing an @xmath134 doublet of bosons @xmath146 instead of the single boson in the @xmath1 gauge theory.@xcite for our purpose this is just a technical way of generating different mean field states which are parametrized by rotating the quantization axis @xmath147 in the @xmath134 space.@xcite this way different states can be visualized and smoothly connected to each other in space and time . in fig .
9 we show such a representation .
the north and south poles correspond to the two degenerate staggered flux ( ddw ) states which break translational symmetry .
the equator is the @xmath28-wave superconducting state . in between
are many states which share both kinds of order .
one advantage of this approach is that we can construct a model of the vortex core of an @xmath148 vortex.@xcite this takes the form of a meron , or half a skyrmion . as shown in fig .
10 the center of the vortex is occupied by the staggered flux state .
this solves a serious problem with the original @xmath1 formulation which favors the @xmath149 vortex@xcite because it is energetically favorable to retain the fermion pairing and make the boson wind by @xmath105 around the vortex .
that approach ignored the fact that another state , the staggered flux state , is available to take the place of the normal state inside the core , allowing us to construct an @xmath148 vortex that costs very little energy .
our model of the vortex core also explains a puzzling experimental observation , i.e. stm tunneling found that an energy gap remains when tunneling into the core region.@xcite this is opposite to what was found in a conventional superconductor , where bound states are formed inside the core which fills in the gap.@xcite in the @xmath134 gauge theory .
the north and south poles correspond to the staggered flux phases with shifted orbital current patterns .
all points on the equators are equivalent and correspond to the @xmath28-wave superconductor . , scaledwidth=40.0% ] the @xmath134 construction allows us to construct a simple picture of the finite temperature phase diagram which accounts for the nernst region and the pseudogap phase.@xcite this is shown in fig . ( 11 ) . at low temperature
the @xmath134 quantization vector @xmath147 is in the @xmath18-@xmath110 plane and we have the @xmath28-wave superconductor ground state .
as the temperature is raised , vortices are created with staggered flux cores and the proliferation of these vortices drives the berezinskii - kosterlitz - thouless transition in the standard way . above this transition ( which is the true superconducting @xmath0 ) ,
vortex and anti - vortex proliferates giving rise to the nernst phase . at even higher temperature
the @xmath147 vector is completely disordered and this is our picture of the pseudogap phase .
we emphasized that in this theory the pseudogap phenomenon and superconductivity are intimately connected .
there is no separate pairing mechanism for superconductivity .
what drives superconductivity is the coherence of the boson to select the true ground state out of a myriad of fluctuating possibilities .
this is very different from the competing order scenario which does require a separate pairing mechanism and a completely separate energy gap scale .
this dichotomy has spurred a debate concerning one gap vs. two gaps , i.e. whether a smaller energy gap appears which scales with @xmath0.@xcite the debate is often couched in a black and white language , with one gap favoring a superconducting gap destroyed by phase fluctuations , and two gaps implying the need for some kind of competing order . in my view , the truth is likely to be more complicated . in the mean field
rvb picture , a sharp quasiparticle peak appears below @xmath0 with weight @xmath18 which follows the @xmath28-wave dispersion with only one gap .
the mean field picture is probably too simplistic .
for example , it is possible that the large gap at @xmath80 as a spin gap which can remain broad while the low energy quasiparticles near the nodes become coherent below @xmath0 . to a first approximation
, the coherent nodal quasiparticle has a dispersion which extrapolates to the large pseudogap at @xmath80 .
there could be a coherence energy scale which scales with @xmath0 , but exactly how this coherence scale affects the density of states and how it develops as a function of temperature is an open question .
the issue of one gap / two gaps is not settled experimentally . for moderately underdoped bi2212
( @xmath150 k ) the evidence from arpes is that the quasiparticle in the superconducting state is reasonably peaked even near the antinodes and obey the @xmath28-wave dispersion with a single gap which increases with decreasing doping .
this is supported by low temperature thermal conductivity data which measures the ratio @xmath151 where @xmath133 is the quasiparticle velocity in the direction of @xmath80.@xcite it is found that @xmath133 increases with decreasing @xmath18 and extrapolates to the antinodal gap measured by arpes . on the other hand , for severely underdoped samples and one layer cuprates with low @xmath0 s , there are claims based on arpes that the energy gap in the fermi arc region near the nodal point does not scale with the pseudogap at @xmath80 , which increases with decreasing doping as mentioned before.@xcite instead , it seems to stay constant or increase with decreasing doping .
it is argued that this reveals a new gap scale associated with superconductivity .
i should caution that deeply underdoped samples are known to be strongly disordered , and the disorder increases with reduced doping .
furthermore , the lineshape remains very broad in the antinodal direction even in the superconducting state .
thus it is risky to draw strong conclusion from lineshapes without an understanding of disorder effects and of the lineshape .
the @xmath134 quantization axis points for the north pole at the center ( forming a staggered flux vortex core ) and rotates smoothly towards the equatorial plane as one moves out radially.,scaledwidth=44.0% ] other support for two gaps comes from andreev reflection studies@xcite and raman scattering.@xcite in a superconductor - normal metal junction in conventional superconductors , normal electrons incident on the junction has an extra channel for transport , by tunnelling as a cooper pair into the superconductor and andreev reflected as a hole .
this leads additional conductance below an energy scale of the energy gap .
such extra conductance was observed in underdoped cuprates , but the energy scale observed is much lower than the pseudogap and is more related to @xmath0 .
i note that in contrast to conventional tunnelling , andreev reflection does not simply measure the density of states , but requires coherence of the quasiparticle in its interaction with the condensate .
what is seen in the andreev data may be this coherence scale . in different parts of the phase diagram shown in fig . 2
( a ) in the superconducting phase @xmath152 is ordered in the @xmath18-@xmath110 plane .
( b ) in the nernst phase , @xmath152 points to the north or south pole inside the vortex core .
( c ) the pseudogap corresponds to a completely disordered arrangement of @xmath152 .
( @xmath152 is a three dimensional vector and only a two dimensional projection is shown.),scaledwidth=50.0%,scaledwidth=33.0% ] i must emphasize that the simple cartoon shown in fig .
11 is only an approximate picture .
we have assumed that the bosons are locally condensed and can be treated as a @xmath64-number which varies in space and time .
however , even at @xmath153 , the vortex configurations shown in fig .
10 can tunnel between each other and destroy the staggered flux order at some time scale .
i think the correct answer requires a quantum mechanical treatment of the boson strongly coupled to gauge fields , which is not available at present .
in particular , we have not yet been able to compute the arpes spectrum and make a satisfactory comparison with experiment .
we make crude approximations such as assuming a binding of the bosons with fermions via gauge fluctuations.@xcite as an example of an alternative aproach , ribeiro and wen@xcite introduced a new formulation which hybridizes the physical hole with the spin carrying fermions and have had success in understanding the higher energy spectra .
their theory seems to favor the two gap scenario .
the truth is that the theory of the spectral function is not under control at present : the problem of a fermion and boson strongly interacting with a gauge field at finite temperatures is too daunting with currently known techniques . rather than trying to explain data in detail
, it may be more fruitful to step back and attempt a classification of the pseudogap state .
we have developed a view that the pseudogap phase belongs to the deconfined side of the lattice gauge theory , so that discussion in terms of fermions , bosons and _ noncompact _ gauge fields makes sense .
more explicitly , we propose that the pseudogap is best understood as the doping of a particular spin liquid called the algebraic spin liquid .
this is explained in the next section .
let us return to the gauge theory formulation described in section iii . as it stood eq .
( 8) is pretty much an exact reformulation of the @xmath11-@xmath17 model .
the constraint is enforced exactly upon integrating out the gauge fields .
however , the gauge field is fluctuating strongly because there is no restoring force , and critics of the gauge theory approach have pointed to this as evidence that this approach can not be trusted to give meaningful results . at a deeper level ,
i think what lies behind this objection is the implicit assumption that a phenomenon called confinement , familiar in qcd , necessarily takes place when the gauge coupling is large . in this case
, fermions are tightly bound to bosons by the gauge field , and one just recovers the original intractable electron problem with constraint .
in contrast , our way of thinking assumes that confinement does not happen , in which case the fermions , bosons and gauge fields retain their identity , albeit with strong interaction .
the ultimate strongly coupled fixed point is nontrivial and different from that of free particles , but as is usual in the strong coupling problem , it may be possible to access their behavior using artificial expansion parameters such as @xmath154 where @xmath155 is the number of copies of the matter field . even though the physical problem may correspond to @xmath156 2 or 4 , as long as confinement does not take place the behavior of the physical system is smoothly connected to the large @xmath155 system and the physical behavior of the system
can be understood , even though the critical exponents can not be computed quantitatively . in the case of a compact @xmath1 gauge
, it is well known that in the 2 + 1 dimension the theory is confining even for arbitrarily small coupling , due to the proliferation of instantons .
the point is that in a compact gauge field , a @xmath105 flux can pop up through an elementary plaquette .
the space - time point where this happens is called an instanton event , which is also called a magnetic monopole in euclidean space - time .
if these monopoles proliferate , the total flux through the sample is not conserved .
the strongly fluctuating gauge flux leads to confinement between external charges . at first sight
, this bodes poorly for the gauge theory approach .
however , the existence of matter field can change the story completely .
there is considerable experience in the study of a gauge field coupled to relativistic fermions and bosons , but little is known in the non - relativistic case . in the past few years there has been significant progress in the case of half - filling where there are no bosons .
consider the saddle point with @xmath128 flux through a plaquette .
the fermionic spectrum are massless dirac fermions with nodes at @xmath157 . due to the additional @xmath134 symmetry , these fermions are minimaly coupled to a set of @xmath134 gauge fields .
if the coupling to the gauge field is confining , this leads to what is known in qcd as mass generation and chiral symmetry breaking , which translates to gap formation and af order in our language.@xcite thus the @xmath128 flux phase is our route to af order . on the other hand , consider the staggered flux state .
this mean field ansatz breaks the @xmath134 symmetry and the gauge field is broken down from @xmath134 to @xmath1 .
this motivates the study of th problem of @xmath155 2-component massless dirac fermions coupled to a @xmath1 gauge field in 2 + 1 dimensions .
this model is often called qed@xmath23 in the literature . since the staggered flux state has two spins and two nodes , the physical problem corresponds to @xmath158 .
( it was shown that the velocity anisotropy of the dirac cone is an irrelevant variable and the low energy physics scales to the isotropic fixed or the qed@xmath23 model.@xcite ) assuming deconfinement , this class of states has been studied by @xmath154 expansion and is called the algebraic spin liquid.@xcite hermele _ et al._@xcite showed , using results borrowed from the field theory literature , that for sufficiently large @xmath155 , the deconfined state can be stable .
in other words , the large @xmath155 fixed point has no relevant perturbation , including the appearance of monopoles .
this is at least a proof of the principle that deconfinement is possible in the presence of matter field .
it is not known whether the critical @xmath155 is greater than or smaller than 4 , and here is where the qcd monte carlo community can help.@xcite recently we received some encouraging news from experiments .
it has long been thought that the kagome lattice with @xmath43 has sufficient frustration to support a spin liquid ground state .
this seems to be realized by recent experiments on zucu@xmath23(oh)@xmath159cl@xmath22 where the cu ions occupy kagome sites and the system shows no af order down to 30 mk , despite an af exchange of @xmath160 k.@xcite on the theoretical side , it is found that the projected wavefunction of a certain flux state gives the ground state energy obtained by exact diagonalization to within error.@xcite this is remarkable for a trial wavefunction with no variation parameter , and is much better than what projected flux states did for the square lattice .
the low energy physics of this state is that of two 2-component massless dirac fermions with spins , coupled to a @xmath1 gauge field .
the confirmation of this picture for the kagome lattice will give us reason to believe that the critical @xmath155 is less than 4 in this case and perhaps in the case of the algebraic spin liquid as well .
prior to the kagome example , there was strong evidence that a spin liquid state exists in the organic compound @xmath161-(et)@xmath22cu@xmath22(cn)@xmath23.@xcite here the active sites are organic molecular dimer molecules which form a triangular lattice to a good approximation .
as mentioned in section iii , the heisenberg hamiltonian on a triangular lattice is expected to order .
nevertheless , experimentally no af is found down to 32 mk .
the explanation lies in the fact that this system sits just on the insulating side of the mott transition , so that charge fluctuations and ring exchange terms lead to a more complicated effective hamiltonian which may favor the spin liquid state.@xcite numerical work has given an adequate account of the phase diagram , including the appearance of superconductivity under pressure as shown in fig .
3.@xcite on the triangular lattice the low energy lagrangian is expected to be an almost circular fermi sea of spinons ( fermions which carry @xmath65 but no charge ) coupled to @xmath1 gauge fields .
this model has even more low lying fermionic excitations than the dirac sea , and there is reason to believe that deconfinement occurs . experimentally the observations of a constant spin susceptibility and linear specific heat at low temperatures are highly unusual for an insulator , and support the notion of a spinon fermi surface .
the appearance of particles , such as spinons which carry different quantum numbers compared with the original electron , is a phenomenon called `` fractionalization . ''
it is remarkable that the fermions originally introduced as a formal device in eq .
( 4 ) take on a life of their own at the end of the day .
many people find this a hard concept to swallow and here is where explicit examples of a spin liquid will be a great help in convincing skeptics .
i must emphasize that the new structure in the theory ( spinons , gauge fields , etc . )
emerges in the low energy physics and what emerges is independent of the way the problem was formulated .
the constraint of no double occupation could have been enforced using @xmath162 , @xmath1 or @xmath134 gauge fields , but the low energy structure will be the same .
one particular formulation may simply be the convenient way to expose this emergent structure .
the discovery of experimental examples of spin liquids is a very important development because for the first time , the deconfined ( fractionalized ) states are low temperature states which in principle can be studied in great detail
. the success of the gauge theory method in these materials will give us confidence in its application to the more complex problem with holes .
it is fortunate that the two promising examples of spin liquids are closely related to those discussed in the high @xmath0 context for underdoped and optimal doping , respectively . finally , i would like to mention an approach to the pseudogap problem from a more general perspective .
senthil and i @xcite proposed that the underdoped cuprate should be considered as proximate to a spin liquid state which is then doped . in this view
the pseudogap is the finite temperature region controlled by the quantum critical point of doping a spin liquid by varying the chemical potential .
the most promising spin liquid is in fact the algebraic spin liquid with its low energy massless dirac fermion and @xmath1 gauge field .
this point of view does not help solve the doped spin liquid problem , but does raise and answer the question : what is the signature of a deconfined state .
the answer is that in the presence of a matter field , the only signature left is the irrelevance of instantons , i.e. the total gauge flux is conserved , much like the conservation of magnetic flux in our world .
the experimental implication is subtle , but can be probed at least in principle in a suggested experiment.@xcite
is there a region of the phase diagram where the picture is less complicated and precise predictions can be made and tested ?
one natural idea is to apply a large magnetic field perpendicular to the layer to kill the superconductivity and ask what is the next most stable ground state which emerges .
when @xmath163 reaches @xmath164 the vortex core overlaps and the core state becomes the homogeneous ground state .
recent experiments indicate that we may be probing this regime and have caused a lot of excitement.@xcite it is believed that @xmath164 for underdoped cuprates is approximately 100 t , well beyond what is available in the laboratory . the resistive transition , on the other hand
, is controlled by flux flow and is much more accessible to experiment .
recently , shubnikov - de haas oscillations are reported in an underdoped ybco compound ( @xmath165 k ) in the flux flow region , in a magnetic field range of 40 t to 62 t. from the period of oscillation of the resistivity and hall resistivity vs. @xmath166 the area of the fermi surface pocket is extracted and found to contain 0.038 holes per pocket .
if we assume the doubled unit cell scenario , there are two pockets in the reduced bz and this corresponds to @xmath167 .
the effective mass can also be measured from the temperature dependence , and @xmath168 @xmath169 .
the doping concentration of ybco is difficult to estimate , because the doping is from the oxygen chains and its charge state is not independently measurable .
the best estimate by the experimentalists places @xmath18 at 10% .
more recently , similar quantum oscillations are reported in yba@xmath22cu@xmath21o@xmath170 , the so - called y124 compound ( @xmath171 k ) with magnetic field from 55 t to 85 t @xcite and shubnikov - de - haas oscillations were reported in the range 45 t to 61 t.@xcite the hole concentration per pocket is now 0.05 and the inferred @xmath18 of 0.1 is again lower than the value of 0.125 that is believed to characterize this material .
the effective mass is now @xmath172 3 @xmath169 .
quantum oscillation is considered the best way to measure the fermi surface area , provided the sample is sufficiently free of defects that the electron can complete a cyclotron orbit without being scattered . even at 50 t , the cyclotron orbit circumference is more than a thousand @xmath173 in size ( see fig .
12 ) , thus making it clear that disorder is really not an issue in the ybco family . .
figure is drawn almost to scale at @xmath174 , when the discs occupy half of the total area ( @xmath175 ) .
a cyclotron orbit is shown , also roughly to scale .
its area encloses about 10 vortices for the experiment on ybco .
the long dimension of the orbit is about @xmath176.,scaledwidth=45.0% ] i think there is an excellent chance that the experiment is accessing the normal state that lies beyond @xmath164 .
it is well known experimentally that quantum oscillations persist to @xmath177.@xcite the frequency of the oscillation is not changed upon crossing @xmath164 , only its magnitude is diminished .
this phenomenon is particularly striking in layered materials where the flux flow regime can be very wide .
for example , a recent experiment on organics observed quantum oscillations down to @xmath178 and is even more striking in that the expected reduction of the amplitude was not even observed.@xcite a rough physical explanation may be as follows .
let us take a cartoon picture of the vortex core as discs of radius @xmath179 . at @xmath164
the discs overlap and we define @xmath180 where @xmath181 is the superconducting flux quantum .
note @xmath182 where @xmath183 is the coherence length . for @xmath184 t , @xmath185 = 25 @xmath173 .
for @xmath186 the disc density is rather high , as drawn in fig .
12 , almost to scale .
let us estimate the size of the semiclassical orbit .
very generally , the real space area @xmath187 is related to the momentum space pocket size @xmath188 by @xmath189 it is useful to introduce the landau length @xmath190 note that the flux through an area @xmath191 is @xmath192 , so that @xmath193 .
the landau level number @xmath194 at the fermi level is given by the number of full flux quantum @xmath195 which penetrates @xmath187 , i.e. @xmath196 we find @xmath197 for @xmath198 t , using the experimentally measured @xmath199 @xmath200 for ybco .
since 50 t corresponds to @xmath201 , this also means that on average , 10 vortex cores are inside the spatial orbit .
since the cyclotron motion is fast compared with that of the vortices , we can assume a static picture of the vortex core , which either forms a hexagonal lattice or somewhat distorted from it .
assuming the orbit to be an ellipse with aspect ratio @xmath202 ( for reasons to be explained later ) , we sketch a snapshot of the orbit in relation to the vortex core in fig .
12 . in order to see quantum oscillations ,
what is required is that the states in the cores are in phase , whether it is af order or staggered - flux order . in the latter case , the @xmath147 vector in all the vortices inside the cyclotron orbit are either pointing up or pointing down , i.e. the coherence length of the order must exceed the orbit size .
for the aspect ratio chosen , the long dimension of the orbit is about 800 @xmath173 .
what is required for quantum oscillations is strong tunnelling between the vortex cores .
then the quasiparticles are extended states which carry information of the uniform normal state for @xmath203 because the random phase coming from tunnelling through the superconducting region will cancel .
given the dense packing of vortices , this seems quite reasonable . what is the origin of the pockets that give rise to the quantum oscillations ?
here we are on rather uncertain grounds . in general , we can classify the state as conventional or exotic . by conventional
i mean there is some new order giving rise to unit cell doubling and a fermi liquid ground state , and exotic means everything else .
while exotic scenarios have been suggested @xcite here i will concentrate on discussing conventional possibilities , knowing that we immediately encounter a problem concerning the relationship of the measured area to doping estimates .
either the doping estimate is incorrect , or there exists undetected electron or hole pockets .
let me forge ahead and assume that the doping concentrations are actually @xmath204 and 0.10 and correspond to the pocket areas measured in the two experiments and discuss several candidates for the unit cell doubling order . 1 .
we saw in section i that at very low doping , hole pockets appear around @xmath205 in the background of af order .
the question is whether this narrow region @xmath206 will open up in a strong magnetic field and extend up to 10 or 12% doping .
this is probably what most people have in mind when they refer to af in the core and we shall call states of this kind `` doped af . ''
i think this scenario is unlikely . from neutron scattering
we know that there is a sharp triplet resonance at @xmath207 mev in this doping range at @xmath205 and a sharp drop in spectral weight below it.@xcite af order requires the condensation of this triplet mode .
a 50 t field corresponds to an energy scale of 6 mev , hardly enough to perturb the resonance mode .
thus i think the af is energetically highly unfavorable . nevertheless , there are experimental ways to distinguish doped af from the staggered flux state , which i will return to later .
2 . staggered flux state . as we argued before , this state
is energetically favored to appear once superconductivity is destroyed .
i want to point out that in principle a small amount of af order can co - exist with this state . as seen in fig . 8 ,
this is favorable for @xmath208 .
however the driving force and excitation spectrum of this state is completely different from the doped af . in that case
the holes live in the lower hubbard band , i.e. the band is a downward parabola , separated by a gap of order of the mott gap with energy scale 1 ev .
3 . valence bond solid ( vbs ) , or more general states which break translation symmetry without breaking time reversal as in af .
there has been discussion of this state living in the vortex core.@xcite roughly speaking , the electronic spectrum may be quite similar to the staggered flux state except that it is expected to be gapped .
4 . spin density wave ( sdw).@xcite this seems very attractive phenomenologically . from nmr
there are claims for enhanced antiferromagnetism in the core.@xcite in the doped la@xmath22cuo@xmath21 system it is known from neutron scattering that incommensurate sdw is induced in the vicinity of the vortex core.@xcite this is quite understandable because inelastic neutron scattering indicates soft excitations at precisely these incommensurate wave vectors , typically around @xmath209 from @xmath210 .
this is often connected to the notion of stripes near the concentration @xmath211 . however ,
static sdw has never been seen in the ybco family .
neutron scattering has reported dynamical fluctuations at 20 mev at the incommensurate wave vectors @xmath212 .
while recent data on more underdoped samples ( @xmath213 k ) sees dyamic scattering down to 12 mev , it has much less spectral weight than in the doped la@xmath22cuo@xmath21
system.@xcite thus it is questionable whether a magnetic field can induce sdw order .
furthermore , the prevailing view is that the sdw is not tied to fermi surface nesting , and it is unclear whether hole pockets should be expected .
+ + how can we distinguish the staggered flux state from the doped af state ?
a useful experimental test is to do a tunnelling experiment to measure the density of states in high fields .
since spatial resolution is not needed , we do not need stm , which is probably impossible under these pulsed field conditions , but we propose either break - junction tunnelling or other tunnelling geometry where electrons can be injected into the a - b plane from the side . schematically
the density of states of the staggered flux state vs. the doped af is sketched in fig .
of course , we expect much of the higher energy features to be smeared out , but strong particle - hole asymmetry and a dip in the density of states above @xmath214 will support the staggered flux picture . on the other hand , if this state is the continuation of the doped af state observed at @xmath215 for zero field , we expect a large mott gap above the fermi energy , which has been estimated to be 30 mev in the case of y124.@xcite thus tunnelling data is a sensitive test of the notion of hole pockets in the lower hubbard band .
+ next we extract some physical parameters from the data .
let us parametrize the dirac spectrum of the staggered flux state near each node by + @xmath216 + where @xmath217 in the direction perpendicular ( parallel ) to the large fermi surface . by definition + @xmath218 + by writing the density of states @xmath219 in terms of the velocities along the fermi surface
, we find the anisotropic generalization of the relation @xmath220 : + @xmath221 + equation ( 20 ) applies equally well to the anisotropic dirac spectrum and the anisotropic parabolic band .
the right hand side of this equation are directly measured .
putting in the numbers for ybco with @xmath222 = 1.9 @xmath169 , we find @xmath223 to be 0.48 ev@xmath173 .
the fermi velocity can be measured by arpes , even though the error bar is substantial .
taking the value @xmath132
= 1.4 ev@xmath173 from campuzano _
et al._,@xcite we find the anisotropy to be @xmath224 .
this compares with the value 7.9 directly extracted from thermal conductivity data @xcite in their @xmath225 sample . in that paper
they extract a gap value @xmath226 of 71 mev , using @xmath132 = 1.65 ev@xmath173 . with our numbers
we expect @xmath227 mev , in very good agreement with arpes estimate of the pseudogap for a similarly doped bi2212 sample . a similar exercise using y124 data @xcite yields @xmath228 and @xmath229 mev .
the gap value is on the low side . given that these estimates are sensitive to the square of @xmath222 and @xmath132 and their associated errors , these are reasonable numbers .
more importantly , it shows the correct trend of increasing anisotropy and reduced gap with increasing doping .
it is worth noting that the ellipse drawn in fig .
1(b ) by doiron - leyraud _
et al._@xcite has an aspect ratio of @xmath230 . noting that the length of the ellipse scale as @xmath231 , our numbers indicate a more elongated ellipse and a small departure of the inside edge of the pocket away from the fermi arc . for y124 ,
the ellipse almost reaches the saddle point at @xmath80 . beyond this point
the pockets connect to form large fermi surfaces which are presumably unobservable due to disorder scattering , and we expect the quantum oscillation to disappear .
there will be a transition to the uniform large fermi surface ( area @xmath56 ) state , but the exact nature of this transition is an open issue .
we reiterate that eq .
( 20 ) applies to the parabolic band of the doped af as well .
the only difference is that in that case there is no reason to identify @xmath132 with the normal state fermi velocity . + quite generally , if this state is a fermi liquid in a nonmagnetic background , we expect zeeman splitting of the up- and down - spin pockets . for free fermions the splitting is entirely determined by @xmath222 which gives the density of states .
let us define @xmath232 to the difference in the up- and down - spin pockets .
we find + @xmath233 + using @xmath234 with @xmath235 and @xmath236 t , we find for y124 a splitting of @xmath237 .
experimentally a splitting of @xmath238 was observed .
if we attribute this to spin splitting , we will need a reduction of the predicted magnetization due to residual interactions . in fermi liquid theory , this is given by the landau parameter @xmath239 , so that @xmath240 .
thus we need @xmath241 which is quite reasonable for a strongly interacting system .
+ in contrast , for the doped af , we do not expect a similar spin splitting .
this was pointed out to me by t. senthil .
this is because the af will ca nt with sublattice magnetization perpendicular to the applied field and spin along the field is no longer a good quantum number .
if splitting due to interlayer tunnelling can be ruled out , this is another argument against the doped af .
+ finally , we discuss the cyclotron resonance experiment . in principle , this can help distinguish between the dirac spectrum of the staggered flux state and the parabolic spectrum expected for the doped af.@xcite as made famous by recent experiments on graphene,@xcite the landau level in a dirac spectrum is given by @xmath242 which scales as @xmath243 , instead of the standard landau level spacing @xmath244 which is linear in @xmath245 in conventional parabolic bands . using measured parameters for ybco
we find the splitting between the @xmath246 and @xmath197 landau level to be 3.1 mev , compared with 3.3 mev for a parabolic band , an unfortunate coincidence . in principle , one can check the @xmath243 vs. @xmath245 dependence , but this may be a challenging experiment to carry out in pulsed fields
. + corresponds to the nernst region in fig .
2.,scaledwidth=50.0% ] + to summarize , with insight gained from nernst effect measurements , we propose in fig .
14 a phase diagram for the underdoped cuprate in a field .
the staggered flux state appears as a new phase in high field and low temperature .
since this state requires the phase coherence of the bosons just as the superconducting state , we expect its transition temperature to be comparable to the superconducting @xmath0 .
we expect a signature in transport properties at this transition temperature .
the @xmath164 line is a cross - over line which characterizes the onset of a vortex liquid state .
this is the extension of the nernst region shown in fig . 2 to finite @xmath163 .
the true superconducting transition is to a vortex solid or glass state .
whether the staggered flux state terminates in the vortex liquid or solid phase is not known .
below a large energy scale @xmath226 lies the spin gap phase which is the extension of the pseudogap phase in fig . 2 to finite @xmath163 .
we emphasize that while the model of hsu _
et al._@xcite and the more phenomenological approach of chakravarty _
et al._@xcite also predict a staggered flux ( ddw ) state at high field , they key point they made is that this state is smoothly connected to the pseudogap phase , i.e. the solid phase boundary in fig .
14 is pushed up to an energy scale @xmath226 .
great effort has been made to look for the moment generated by the orbital currents by neutron scattering and nothing convincing has been found .
there is no sign of a phase transition even in the highest purity materials .
it is difficult to see how the fermi arc can be consistent with a fermi liquid picture of sharp quasiparticles . in my opinion
, existing data at low field is sufficient to rule out the ddw scenario .
in contrast , in our picture the pseudogap phase is separated from the high field staggered flux fermi liquid state by a phase boundary .
while a detailed description of the fermi arc is not available , it is at least possible to imagine that different physics is at play and the fermi arc picture can be consistent with the experimental observation of quantum oscillations in high fields .
+ finally , the proposed phase diagram is heavily based on mean field theory and the notion of boson condensation and should be viewed simply as a starting point for discussion of the experiment .
it is possible that the orbital current can survive fluctuations , but it is also possible that tunnelling events wipe out the ordering pattern , leaving us with an exotic ground state .
other order , such as weak af , could also emerge to take its place . the overall picture , i think , is more general than the survival of the orbital currents .
the large spin gap at @xmath80 and features in the tunnelling density of states sketched in fig .
13(a ) may survive .
clearly more experiments in high fields will be needed to shed light on these important issues .
instead of a conclusion , i would like to pose a number of questions and attempt to answer them . 1 .
what is unique about the cuprates ? what is the pairing mechanism ?
what controls @xmath0 ? can @xmath0 be raised ? + the cuprate is the prime example of a two - dimensional mott insulator which can be doped .
the copper @xmath247 ion is in a @xmath248 configuration , making it a spin @xmath3 system .
the layer symmetry breaks the degeneracy of the @xmath28 orbitals dow to a single @xmath249 orbital , so that orbital issues which cause complications in most other transition metal oxides do not play a role here . in most other examples , strong coupling of the charge state to distortion of the oxygen cage causes doped carriers to localize . for example , if we replace cu by ni , la@xmath22nio@xmath21 requires 50% hole doping before it becomes metallic . in many ways ,
the cuprate is the ideal model system for a doped hubbard model .
the parent compoud is one of the few layered @xmath43 antiferromagnet with no orbital degeneracy that is known .
it is hard to find a material that is so simple ! + searching for a pairing mechanism may be the wrong question , because implicitly the assumption is made that pairing occurs via the conventional route of well defined quasiparticles exchanging some bosonic glue.@xcite we know that underdoped cuprates do not fit this description .
instead superconductivity is the ground state which emerges as the best compromise between the kinetic energy of hole motion and the antiferromagnetic exchange .
the energy gap is set by the exchange energy @xmath17 while @xmath0 is controlled by phase coherence and is of order @xmath44 .
thus the optimal @xmath0 is set by @xmath17 .
the cuprate family already has the highest @xmath17 among transition metals .
( the ladder compound made up of the same copper - oxygen bond is the only example i know which has a higher @xmath17 than the planar cuprate . ) the hopping matrix element @xmath11 is also large because of the strong covalent bonding between the @xmath249 orbital with the oxygen @xmath250 orbital .
thus the cuprates are blessed with a large energy scale .
the variation of @xmath0 on a few tens of degrees between ybco and the hg compound seems to be attributable to changes in the further neighbor hopping matrix elements.@xcite + if the superconductivity is due to the electronic energy scale such as @xmath17 and @xmath11 , why is nt @xmath0 higher ? as we saw in section i , it is only @xmath251 or @xmath252 for both cuprates and organics .
what accounts for the small numerical factor ?
one consequence of doping a mott insulator is that the carrier density and therefore the superfluid density is small .
thus phase fluctuation is important and limits @xmath0 . in a @xmath28-wave superconductor ,
the thermal excitation of nodal quasiparticles further suppresses the superfluid stiffness on a short distance scale , as seen by experiment.@xcite this reduces the @xmath0 further .
this is why @xmath0 is such a small fraction of @xmath17 . besides increasing the overall energy scale ,
one possible avenue for enhancing @xmath0 is to try to stabilize superconductors with a full gap such as @xmath253 states .
for example , on a triangular lattice , the @xmath253 state is believed to be favored . in this case
, the quasiparticle suppression of the superfluid density is negligible and @xmath0 will be higher , everything else being equal .
2 . what about other approaches ? why is there no consensus after all these years ? + in a review of this kind which adopts a particular viewpoint ( going under the names strong correlation , rvb , gauge theory , etc . )
i feel obligated to give my honest opinion on other approaches
. this may not be the best way to win friends , but let me proceed .
first let me point out that within the rvb approach there are many variants that have not been covered here .
i emphasized the formulation based on a fermionic representation of the spins and a bosonic representation of the holes . the opposite formulation with bosonic spinons @xcite and a more complicated formulation with flux attachment @xcite has also been pursued .
below let me focus on approaches outside the rvb `` big tent '' by separating the most prominent work into four major categories and discussing them in turn
. 1 . spin fluctuation models .
+ historically the possibility of @xmath28-wave superconductivity based on the exchange of af spin fluctuations was discussed prior to the discovery of high @xmath0 and a number of workers apply these ideas to the cuprates.@xcite these discussions are either based on random phase approximation treatment for the hubbard model , or phenomenological coupling between spins and electrons treated as separate entities . the latter may be correct in heavy fermion systems , where there indeed exists separate itinerant and local moments , but is more questionable in the hubbard model where there is only one kind of electronic state .
the best chance for this model to work is in the overdoped region , where well defined quasiparticles exist in the normal state and the onset of superconductivity appears quite conventional .
however , from neutron scattering it is known that spin fluctuations become very weak , with spin correlation length of 10 @xmath173 or less beyond optimal doping .
a variant of this idea ascribes the boson being exchanged to the sharp triplet resonance discovered by neutron scattering .
however , this resonance is very narrow in momentum space and carries very little spectral weight , making it unlikely to be the powerful glue needed.@xcite it is clear that in order for the spin fluctuation picture to work , one has to deal with the strong coupling problem and include spin fluctuations at all energy scales up to @xmath17 .
numerical methods may be the only option .
recent numerical studies based on cluster dynamical mean field theory shows promise in this direction in that a pairing kernel was identified in the triplet channel which peaks at @xmath254 and resembles the spin fluctuation channel.@xcite i think the most serious limitation of this line of ideas is that it completely fails to address the pseudogap issue . as we have seen
, there is simply no well defined quasiparticle above @xmath0 to pair them by whatever mechanism . by avoiding the issue of proximity to a mott insulator
, this approach misses the key physics and the most interesting aspect of high @xmath0 .
microscopic inhomogeneity , stripes , etc .
+ the basic idea is that one way the doped mott insulator can resolve the competition between kinetic energy and af exchange is for the holes to phase segregate . in particular
, the holes may concentrate into one dimensional regions where af order is suppressed , separated by an af region where the hole density is suppressed.@xcite experimentally , it was discovered that a particularly stable configuration where the holes form a quarter filled chain ( one hole per two sites along the chain ) separating af regions as an anti - phase boundary indeed exists in la@xmath19ba@xmath20cuo@xmath21 where @xmath211.@xcite this configuration and variants thereof has been called stripes . the stripe explains the suppression of @xmath0 in the doped la@xmath22cuo@xmath21 system near @xmath211 .
it also receives theoretical support from numerical solution of the @xmath11-@xmath17 model using density matrix renormalization group method.@xcite thus there is good reason to believe that stripes form a strongly competitive ground state , especially near @xmath211 .
however , the ordering temperature for the charge order which preceded spin order is low , of order 40 k , and the spin order ( incommensurate sdw ) is even lower .
furthermore , static stripes are seen only in the doped la@xmath22cuo@xmath21 family and not in the ybco family , and in the latter case , even dynamical stripe fluctuations have weak intensities , as discussed earlier .
thus it is my belief that stripes represent low energy physics and emerge as the ground state only as a result of delicate competition .
recently there is a remarkable report that the arpes in la@xmath255ba@xmath256cuo@xmath21 reveals quasiparticles which @xmath28-wave dispersion with a maximum gap of 20 mev , as if the existence of stripes has no effect on pairing of the pseudogaps except at very low energy.@xcite this reinforces my view that stripe order is a secondary effect .
+ static stripes are clearly detrimental to superconductivity .
nevertheless , there has been a strong push , notably by s. kivelson and co - workers , for the idea that fluctuating stripes may be responsible for superconductivity and the pseudogap phenomenon .
i have always found the motivation of this idea puzzling .
fortunately a recent article by kivelson and fradkin@xcite lays out the physical ideas and motivation in a very clear way .
they share our basic assumption that doping a mott insulator is the key question and they share our notion that superconductivity comes from strong repulsion .
however , they reject the possibility that this can happen in a homogeneous phase .
as far as i can tell , the reason offered is that `` the enormous effort has been devoted to numerical searches for superconductivity in various uniform hubbard and @xmath11-@xmath17 related models , with results that are , at least , ambiguous '' and the feeling that if that were the right route , `` unambiguous evidence of it would have been found by now . ''
instead they focused on models where superconductivity from a repulsive hubbard model can be demonstrated , namely the ladder system .
the ladder system can be considered an example of a valence band solid , where spins on the rung form singlets and doped holes tend to form pairs on the rung .
then pair hopping between ladders provides bulk superconductivity .
this is all well and good but kivelson and fradkin are the first to admit ( in fact insist ) that this toy model has little to do with cuprates .
for example , it is hard to imagine how quasiparticles can traverse this system or ladder ( or any kind of stripes ) at a @xmath257 angle and become the gapless nodal particle . as long as this problem is not addressed adequately , i fail to see how this approach can be a viable route to superconductivity in the cuprates .
3 . phonons .
+ electron phonon coupling is typically strong in transition metal oxides involving the @xmath258 orbitals , because of strong coupling to the distortion of the oxygen cage .
as mentioned earlier , in most cases this leads to localization of the doped hole . while holes in the cuprate escape this fate
, the electron phonon coupling must be strong .
nevertheless , we have argued that the dominant energy in this problem is coulomb repulsion .
hence the discussion of electron - phonon coupling must be made in this context .
a combination of numerical and analytic work has shown that short range repulsion has the effect of reducing electron - phonon scattering at large momentum transfer because charge fluctuations at short distance is suppressed.@xcite it is also claimed that under certain circumstances , @xmath28-wave pairing is possible with electron - phonon coupling , contrary to conventional wisdom .
experimentally , phonon sidebands appear visible in stm tunnelling spectra .
all this says that electron - phonon coupling is there ( no great surprise ) but does not provide any evidence that it is the essential driving force behind pairing .
+ the isotope effect is often quoted as evidence for the phonon mechanism for high @xmath0 .
it turns out that isotope effects affect @xmath0 via the superfluid density , and not the order parameter itself.@xcite thus the isotope effect in fact supports the view that @xmath0 is controlled by phase fluctuations in the underdoped region .
there is no isotope effect at optimal doping and beyond .
4 . three band model .
+ since the copper oxygen layer involves one copper and two oxygen per unit cell ( excluding the apical oxygen ) , the minimal microscopic electronic model requires a @xmath249 cu orbital and two oxygen @xmath250 orbitals .
this is called the three band model.@xcite various cluster calculations indicate that the low energy physics ( below a scale of @xmath259 mev ) can be adequately described by a one band hubbard model.@xcite over the years this has become the majority view , but there are still workers who believe that the three band model is required for superconductivity .
i think the original motivation for this view came from a period of nearly ten years when much of the community believed that the pairing symmetry is @xmath27-wave .
it certainly is true that a repulsive hubbard model can not have an @xmath27-wave superconducting ground state .
three band models with further neighbor repulsion were introduced to generate the requisite effective attraction .
the @xmath27-wave story was overturned once and for all by phase sensitive experiments in 1994 but some of the three band model proponents persevered .
in particular , varma introduced the idea of intra - cell orbital currents , i.e. a current pattern flowing between the cu - o and o - o bonds as a model for the pseudogap.@xcite this has the virtue of leaving the unit cell intact and this kind of order is very difficult to detect .
with such a complicated model it is difficult to make a convincing case based on theory alone and a lot of attention has been focused on experimental detection of time reversal symmetry breaking or spontaneous moments due to orbital currents .
unfortunately , the expected signals are very small and can easily be contaminated by a small amount of minority phase with af order . up to now there is no reliable results in support of this kind of orbital currents .
why is the high @xmath0 problem hard ?
why is the high @xmath0 problem important ?
+ one of the ironies about the high @xmath0 problem is that the ground states are rather conventional .
it ranges from af to @xmath28-wave superconductors to fermi liquid as the doping is increased .
the exciting new physics happens at finite temperatures where fermi arcs , pseudogaps and other novel phenomena appear .
this region is not characterized by order in the traditional landau sense , and it is not possible to make precise statements as one could if one has an exotic ground state . from the theoretical side , what we can do so far to classify the pseudogap as the deconfined phase of a gauge theory , where new particles , spinons and holons and @xmath1 gauge fields emerge in the low energy physics . these particles and gauge fields are still strongly coupled and we are not able to compute and compare with experiments in a quantitative way .
we can make various caricatures ( such as local boson condensation and fluctuations between different phases ) which mimic the observations in a qualitative manner .
nevertheless , i believe the problem is so strongly constrained by data that this path is the only viable option at the moment .
i must emphasize that the gauge field is an integral part of this story , and not a technical artifact of our particular formulation . in principle
it is possible to formulate the problem with spin carrying bosons and fermions and holes , and reach the same low energy state described here . in this sense
this proposed solution of the high t@xmath260 problem represents the emergence of new physics .
+ two recent developments are encouraging .
first , it appears possible to probe the `` normal state '' at zero temperature by applying a large magnetic field . while the state again appears to be a fermi liquid
, it may expose a new kind of order and reveal what is fluctuating in the pseudogap phase .
secondly , the elusive spin liquid ground state may finally have been realized .
even though this is an early stage in its development , the study of the low energy excitations of these ground states may finally allow us to access spinons and gauge fields .
the effective low energy theory of the spin liquid problem is also more amenable to treatment because bosonic holes are not involved .
it is our view that the study of spin liquids and high @xmath0 opens a new chapter in condensed matter , one that defies the dual landau paradigm ( local order parameter and fermi liquid theory ) that has been the backbone of our science for three quarters of a century . to me , this is why the high @xmath0 problem is important and exciting .
i thank naoto nagaosa and xiao - gang wen for their collaboration on many of the topics discussed here and i am particularly grateful to t. senthil for sharing his insights and for helping shape many of the thoughts that have gone into this review .
i acknowledge support by nsf grant number dmr0517222 . | this review gives a rather general discussion of high temperature superconductors as an example of a strongly correlated material .
the argument is made that in view of the many examples of unconventional superconductors discovered in the past twenty years , we should no longer be surprised that superconductivity emerges as a highly competitive ground state in systems where coulomb repulsion plays a dominant role .
the physics of the cuprates is discussed , emphasizing the unusual pseudogap phase in the underdoped region .
it is argued that the resonating valence bond ( rvb ) picture , as formulated using gauge theory with fermionic and bosonic matter fields , gives an adequate physical understanding , even though many details are beyond the powers of current calculational tools .
the recent discovery of quantum oscillations in a high magnetic field is discussed in this context . meanwhile ,
the problem of the quantum spin liquid ( a spin system with antiferromagnetic coupling which refuses to order even at zero temperature ) is a somewhat simpler version of the high @xmath0 problem where significant progress has been made recently .
it is understood that the existence of matter fields can lead to de - confinement of the @xmath1 gauge theory in @xmath2 dimensions , and novel new particles ( called fractionalized particles ) , such as fermionic spinons which carry spin @xmath3 and no charge , and gapless gauge bosons can emerge to create a new critical state at low energies .
we even have a couple of real materials where such a scenario may be realized experimentally .
the article ends with answers to questions such as : what limits @xmath0 if pairing is driven by an electronic energy scale ?
why is the high @xmath0 problem hard ?
why is there no consensus ? and
why is the high @xmath0 problem important ? |
accurate measurements of flow velocities are essential for exploring and understanding fluid dynamics .
such measurements raise important technical problems whenever one needs them to be nonintrusive .
indeed , many applications where direct access to the fluid is impossible perclude the use of local probes such as miniature piezoelectric pressure probes , hot films or hot wires .
in other cases , the introduction of a probe inside the fluid may perturb the flow or even the fluid structure itself , and may lead to incorrect interpretations of the measurements .
thus , in general , much insight can be gained from some remote , noninvasive sensing of the flow field . in the following ,
we focus on the case of complex fluids under simple shear .
a `` complex '' fluid may be characterized by the existence of a `` mesoscopic '' length located somewhere between the size of individual molecules and the size of the sample @xcite .
for instance , this intermediate length scale is the particle diameter in a colloidal suspension , the diameter of an oil droplet in an emulsion , or the radius of gyration of polymer coils .
the existence of such a supramolecular organization ( or `` microstructure '' ) can lead to very complicated behaviors under flow and to inhomogeneous velocity profiles under simple shear @xcite .
since the microstructure of a complex fluid is very sensitive to local deformations , it is quite obvious that a nonintrusive technique is required in order to measure velocity profiles .
the most popular nonintrusive techniques , namely particle imaging velocimetry ( piv ) or laser doppler velocimetry ( ldv ) , are based on the interaction between light and seeding particles following the flow @xcite .
however , many complex fluids , such as emulsions , slurries , or pastes , may not be transparent enough to allow the use of piv or ldv .
nuclear magnetic resonance ( nmr ) offers the possibility to image opaque media @xcite but requires the use of powerful magnets and remains expensive and tricky to set up . on the other hand , ultrasound appears as an efficient , cost - effective tool to measure velocity profiles in a large range of fluids . in this paper
, we adapt a speckle - based ultrasonic velocimetry technique @xcite to complex fluid flows .
although based on the classical principle of backscattering by particles suspended in the flow , our technique brings an original contribution to both the fields of acoustical flow measurements and of complex fluid studies . indeed , by using high - frequency pulses ( frequencies larger than 20 mhz ) , we show that we are able to measure velocity profiles in complex fluids sheared between two plates separated by 1 mm with a spatial resolution of about 40 @xmath0 m . depending on the required accuracy , a full velocity profile can be obtained typically in 0.02 s to 2 s , which makes it possible to resolve transient regimes or temporal fluctuations of the flow .
the paper is organized as follows .
we first explain in more details why local velocity measurements are crucial to the understanding of complex fluids .
examples are given that show that inhomogeneous flows may occur even in simple shear geometries and at small imposed stresses .
the third section is devoted to a brief review of existing ultrasonic techniques for measuring flow velocities .
we then present the electronic setup and the data analysis used for high - frequency ultrasonic speckle velocimetry ( usv ) .
section [ s.calib ] deals with the calibration step necessary to obtain quantitative estimates of the velocity . in section [ s.lamellar ] ,
high - frequency usv is applied to a particular complex fluid : an aqueous solution of surfactant .
the ultrasonic data reveal the existence of inhomogeneous flows , with both wall slip and shear bands , as well as complex spatio - temporal behaviors during transient regimes . finally ,
in light of the present results , the technique is compared to other nonintrusive tools used in the field of complex fluid flows .
we emphasize on the promising spatio - temporal resolution of usv and discuss its possible future applications .
it is well - known that shear flows of simple fluids are laminar at low velocities and may become unstable above a critical reynolds number due to hydrodynamic instabilities @xcite .
a classical shear geometry is the _ couette geometry _ in which the fluid is sheared between two concentric cylinders as sketched on fig .
[ f.couette ] .
usually the inner cylinder of radius @xmath1 is rotating while the outer cylinder of radius @xmath2 remains fixed .
the two cylinders are thus called `` rotor '' and `` stator '' respectively .
the distance @xmath3 between the two walls is called the `` gap '' of the couette cell .
this couette geometry will be used throughout the paper as well as the cylindrical coordinates @xmath4 .
we will denote @xmath5 the distance from the inner cylinder and @xmath6 the velocity vector .
a couette device can work in two different modes : either a torque @xmath7 is imposed on the rotor and the rotor angular velocity @xmath8 is measured , or @xmath8 is imposed and the torque @xmath7 is measured . for a newtonian fluid at small velocities ,
the flow is stationary and purely orthoradial @xmath9 . the tangential velocity @xmath10 decreases from the rotor velocity @xmath11 down to zero on the outer wall . in the small gap approximation _
i.e. _ when @xmath12 , the velocity profile is linear as shown in fig .
[ f.exs](a )
. the shear rate @xmath13 ( @xmath14-component of the rate - of - strain tensor ) and the shear stress @xmath15 ( tangential force per unit surface _ i.e. _ @xmath14-component of the stress tensor ) are then almost uniform across the gap and are simply given by : @xmath16 where @xmath17 is the height of the couette cell .
the viscosity @xmath18 of the sample can thus be computed from the rheological data @xmath13 and @xmath15 by @xmath19 , as long as the flow remains laminar and stationary . in complex fluids ,
the picture of the couette flow may be radically different .
for instance , when a complex fluid is confined between two plates and depending on the roughness properties of the plates , the fluid velocity close to the walls may strongly differ from that of the walls ( see fig .
[ f.exs](b ) ) : the fluid _
slips_. slippage is usually explained by the existence of two thin lubricating layers in which the fluid structure is very different from that in the bulk .
for instance , in a colloidal suspension , electrostatic or steric effects next to the walls may induce a depletion of the particles on a small distance , leaving a much less viscous fluid film at the walls and leading to this apparent wall slip @xcite .
in concentrated oil - in - water emulsions , wall slip is due to the presence of very thin water layers close to the walls .
even if the bulk fluid remains newtonian , the velocity profile between the two plates may be quite different from that expected in a uniformly sheared fluid that does not slip ( see fig .
[ f.exs](a ) ) .
wall slip in complex fluids raises important industrial problems and is often difficult to detect and to assess quantitatively . in particular , computing the viscosity from eqs .
( [ e.gammarheo ] ) and ( [ e.sigmarheo ] ) can be very misleading since the effective shear rate in the bulk is sometimes orders of magnitude smaller than @xmath20 @xcite .
moreover , even in the absence of wall slip and even for newtonian fluids , above a critical rotation speed of the rotor , the laminar flow becomes unstable and vortices develop due to the centrifugal force ( _ taylor - couette instability _ ) . upon further increasing the rotation speed
, this three - dimensional , inhomogeneous but still stationary flow turns into a time - dependent turbulent flow @xcite .
the procedure based on eqs .
( [ e.gammarheo ] ) and ( [ e.sigmarheo ] ) for measuring @xmath13 and @xmath15 and computing the viscosity then completely fails because the flow is no longer laminar and stationary .
about fifteen years ago , complex fluids such as entangled polymer solutions were shown to present inhomogeneous flows even when sheared at rotor velocities far below the onset of the taylor - couette instability .
these unstable flows result from purely _ elastic instabilities _ @xcite .
although due to the elasticity of the fluid rather than visco - inertial effects , the phenomenology of such instabilities is very similar to classical hydrodynamic instabilities but occur at very low reynolds numbers . beside those inertial and elastic instabilities , more subtle effects may come into play that are due to the very nature of complex fluids .
indeed , application of shear to a complex fluid may completely change its microstructure leading to a new _ shear - induced structure_. such a phenomenon may be called a _ shear - induced transition _ or a _ material instability _ in the sense that it results from a strong coupling between the shear flow and the microstructure of the fluid @xcite .
a material instability usually shows up on rheological data as a sudden jump in the viscosity when shear is increased .
for instance , if the shear - induced structure is much less viscous than the complex fluid at low shear , then a large drop of the visocity may be observed and the transition is called a shear - thinning transition . during such a transition , the fluid is assumed to phase - separate into two states : a shear - induced state that flows at a given shear rate and coexists with the old structure , thus giving rise to two shear bands @xcite .
the corresponding _ shear - banded _
velocity profile is sketched in fig .
[ f.exs](c ) . in order to investigate those transitions ,
most previous studies have focused on global measurements such as the viscosity of the sample as a function of time @xcite .
indeed , the time series @xmath21 sometimes display complex temporal fluctuations on time scales characteristic of the microstructure rearrangements .
however , such global measurements do not provide any information on the spatial structure of the flow @xcite .
thus , other studies have focused on the local characterization of the fluid structure using birefringence neutron , x - ray or light scattering and showed the coexistence of bands of different microstructures in the vicinity of shear - induced transitions @xcite .
as far as the flow field is concerned , rather few local velocity measurements are reported in the complex fluid literature @xcite .
the presence of inhomogeneous flows in some surfactant mixtures ( `` wormlike micelles '' ) was first unveiled using nmr @xcite .
however , the existence of shear bands in the velocity profiles was firmly ascertained only recently using dynamic light scattering ( dls ) in heterodyne mode , a technique close to ldv @xcite . from the various examples cited above ,
it is quite clear that local measurements are essential , for instance to precisely follow the shear bands in both space and time . in general , a velocity profile with about 20 points accross a 1 mm gap , _
i.e. _ with a resolution of 50 @xmath0 m , will carry enough spatial information on the flow to allow quantitative conclusions on possible inhomogeneous flow profiles .
contrary to electromagnetic waves , the _ phase _ of acoustic waves is easily accessible using piezoelectric transducers . moreover
, ultrasound may propagate deeply into optically opaque media .
these two properties have led to the spectacular development of ultrasonic imaging techniques since the early 1950s particularly in the _ biomedical _ domain .
when an ultrasonic wave travels through biological tissues , it gets scattered by density and/or compressibility inhomogeneities @xcite .
if the scatterer moves in a flow , like for instance red blood cells inside blood vessels , its motion induces a doppler shift in the wave frequency .
early systems for measuring blood velocities were based on the estimation of this doppler frequency shift using a monochromatic ultrasonic wave @xcite .
unfortunately , even if these `` true doppler '' systems helped to detect the occlusion of blood vessels , they did not offer any spatial resolution . indeed ,
if one uses a continuous wave , one gets a very good resolution on the frequency shift ( and thus on the velocity ) but one also loses any temporal information on the echo arrival time ( and thus on the scatterer position , as explained below ) .
continuous doppler systems may thus be very efficient in the case of a single scatterer suspended in a flow whose position is measured independently @xcite . in order to discriminate between different scatterers in space and to measure velocity profiles , a solution is to use short acoustic _
pulses_. indeed , if an ultrasonic pulse is sent by a transducer through a scattering medium , a series of backscattered echoes can be recorded on the same transducer .
the precise nature of the backscattered ( bs ) signal will be discussed at length in sec .
[ s.usv ] . at this point , it is only important to note that the arrival times @xmath22 of the echoes are directly linked to the position @xmath23 of the scatterer along the acoustic beam by @xmath24 , where @xmath25 is the effective sound speed and the factor 2 accounts for the round trip from the transducer to the @xmath26 inhomogeneity and back to the transducer .
the idea is then to follow the scatterer motion through the evolution of the bs signals : between successive pulses , echoes will move along with the flow and their positions may be tracked in time .
although the doppler effect is not used in pulsed wave systems , these are still mistermed `` pulsed doppler systems '' in the literature and the technique is referred to as `` ultrasonic doppler velocimetry '' ( udv ) or `` ultrasonic velocity profiler '' ( uvp ) @xcite .
this confusion may be due to the fact that , owing to the constraint of real - time display in commercial devices , bs signals are sampled only at a few given depths and that these sampled signals are analyzed in the frequency domain @xcite .
udv has been successfully applied to various classical problems in hydrodynamics such as the taylor - couette instability @xcite , cylinder wakes @xcite , or magnetic fluid flows @xcite .
the fluid has to be seeded with small particles that scatter ultrasound .
such acoustic `` contrast agents '' are assumed to follow the flow as lagrangian tracers .
the uvp has also been used fairly recently for in - line rheological studies of concentrated suspensions in pipe flows @xcite .
if one drops the constraint of real - time display , the whole bs signals may be stored for post - processing .
direct tracking of the bs signals in the time domain is the subject of recent technical developments such as ultrasonic speckle velocimetry which will be addressed in the next section .
when the moving medium contains a lot of scatterers per unit volume , the bs signal results from the interferences of all the backscattered waves and appears as a complex , high - frequency signal called _ ultrasonic speckle _ ( see fig . [ f.signal ] for a typical speckle signal and sec .
[ s.analysis ] for more details , in particular about the possibility of multiple scattering ) .
the development of fast digitizers with large memories has allowed the full recording of backscattered bs signals for post - processing .
various time - domain algorithms for tracking the motion of ultrasonic echoes have been proposed @xcite .
so far , such _ speckle tracking _
techniques have mainly been applied to the measurement of tissue motion or blood flow @xcite .
they have also been adapted to the uvp system and tested on pipe flows to provide high time resolution ( of about 0.5 ms per profile ) @xcite .
recently , one of us introduced a combination of a 2d echographic system and speckle tracking that we called 2d ultrasonic speckle velocimetry ( usv ) @xcite .
this system allowed us to image strong vortical flows in newtonian fluids in two dimensions @xcite .
here we propose to adapt such a technique to complex fluids confined in a couette cell of gap @xmath27 mm .
as explained in section [ ss.inhom ] , one needs to measure velocity profiles with a spatial resolution of typically 50 @xmath0 m .
however , frequencies commonly used in commercial ultrasonic velocimeters are in the range @xmath2810 mhz , and with a typical sound speed @xmath29 m.s@xmath30 , this yields a resolution of about @xmath311.5 mm .
thus , conventional ultrasound does not permit very fine measurements and one has to turn to utrasound with frequencies larger than 20 mhz .
again , high frequencies have first attracted the biomedical community .
high - resolution ultrasonic images have been recently obtained in dermatology , ophtalmology and stomatology @xcite or for measuring blood flow in the microcirculation @xcite . to our knowledge , no such high - frequency measurements ( above 20 mhz ) have been reported so far in classical hydrodynamics nor in the field of complex fluids .
note that the resolution can not be increased indefinitely by increasing the frequency because _ attenuation _ of sound waves sharply increases with frequency ( as @xmath32 in water ) .
attenuation is due to the combination of thermo - viscous absorption and scattering .
the signal - to - noise ratio rapidly deteriorates as @xmath33 increases .
we found that for our application to complex fluids , @xmath34 mhz realizes a good compromise between resolution and attenuation over propagation distances of a couple of centimeters .
finally , since transducer arrays are not available at such a high frequency , the technique will have to be restricted to one - dimensional measurements .
figure [ f.system ] presents our usv electronic system .
focused ultrasonic pulses are generated by a pvdf piezo - polymer immersion transducer of central frequency @xmath34 mhz ( panametrics pi 50 - 2 ) .
the focal distance is 11.6 mm and the active element diameter is 6.3 mm .
the transducer bandwidth is 11 mhz at -3 db .
the axial and lateral resolutions given by the manufacturer at -3 db are 30 @xmath0 m and 65 @xmath0 m respectively and the depth of field is about 1 mm .
the transducer is controlled by a pulser - receiver unit ( panametrics 5900pr ) .
the pulser generates 220 v pulses with a rise time of about 1 ns .
the pulse repetition frequency ( prf ) is tunable from 0 to 20 khz .
the receiver is equipped with a 200 mhz broadband amplifier of maximum voltage gain 54 db as well as a set of selectable high - pass and low - pass filters .
bs signals are sampled at @xmath35 mhz , stored on a high - speed pci digitizer with 8 mb on - board memory ( acqiris dp235 ) , and later transferred to the host computer for post - processing .
the shear flow is generated in a plexiglas couette cell with @xmath36 mm , @xmath37 mm , and @xmath38 mm .
the rotation of the inner cylinder is controlled by a standard rheometer ( ta instruments ar1000 ) .
the whole cell is surrounded by water whose temperature is kept constant to within @xmath39c . the thickness of the stator is 2 mm everywhere except for a small rectangular window where the minimal thickness is 0.5 mm in order to avoid additional attenuation due to the plexiglas .
ultrasonic pulses are incident on the stator with a given angle @xmath40 relative to the normal to the window in the stator as sketched in fig .
[ f.setup ] .
they travel through plexiglas and enter the gap with an angle @xmath41 that is given by the law of refraction . since the sound speed of the working fluid may differ from that of water , and since the precise value of @xmath41 depends on the exact arrangement of the acoustic beam relative the stator , @xmath41 will be taken as an unknown until a careful calibration procedure is completed ( see section [ s.calib ] ) .
once inside the fluid , ultrasonic pulses get scattered by inhomogeneities that can be either naturally present ( oil droplets in an emulsion for instance ) or artificially introduced to enhance the acoustic contrast . the total round trip for a pulse travelling from the transducer to the rotor and back to the transducer lasts about 15 @xmath0s .
the position of the transducer is tuned so that the 1 mm gap lies into the focal spot in order to optimize the signal - to - noise ratio .
typical recorded bs signals are 1000 point long , which corresponds to a transit time of 2 @xmath0s or equivalently to a depth of roughly 1.5 mm inside the medium . allowing for a fraction of a millimeter before the stator and after the rotor helps to locate the two walls .
note that spurious reflections on water plexiglas or plexiglas fluid interfaces are minimized by carefully choosing the angle @xmath42 . in any case
, even in the absence of interfaces , such an angle is necessary to get a non - zero projection of the velocity vector along the acoustic axis as explained below .
figure [ f.signal ] shows a typical signal backscattered by a dilute suspension of polystyrene spheres and normalized by the maximum amplitude .
the noise level is about 10 mv in all our experiments and typical signal amplitudes are in the range 50200 mv corresponding to a signal - to - noise ratio of 1030 db . under the assumption of _ single scattering _ , the speckle signal received between times @xmath43 and @xmath44 can be interpreted as interferences coming from scatterers located between @xmath45 and @xmath46 , where @xmath47 , @xmath48 , and @xmath45 is the distance from the transducer along the acoustic beam .
the approximation that no multiple scattering takes place is thus essential in order to use the echographic conversion rule @xmath47 .
multiple scattering is avoided by carefully controlling the amount and the properties of the scatterers ( see sec .
[ s.calib ] ) . when the fluid is submitted to a shear flow , the interference pattern changes as the scatterers move along .
figure [ f.signal_zoom ] shows two signals corresponding to two pulses separated by @xmath49 ms and plotted over 4 acoustic periods . a global shift to the right of @xmath50
ns is clearly visible . with @xmath29 m.s@xmath30
, this corresponds to @xmath51 m and to a velocity @xmath52 mm.s@xmath30 projected along the acoustic axis .
in order to get an accurate estimate of the time - shift @xmath53 , we use the simple cross - correlation algorithm described below .
two consecutive signals @xmath54 and @xmath55 recorded after pulses sent at times @xmath56 and @xmath57 are cross - correlated over small time windows of width @xmath58s , where @xmath59 is the acoustic period . more precisely , the following cross - correlation coefficient is computed according to : @xmath60 where the @xmath61 time window is centered around @xmath62 and @xmath63s is a reference time that corresponds to the beginning of the gap .
@xmath64 depends on experimental parameters such as the working temperature and on the exact arrangement of the transducer relative to the couette cell ( see sec . [ s.calib ] ) .
the times @xmath22 correspond to the centers of the various time windows over which the signals are split . they will be later converted into positions @xmath23 at which the velocity is measured using @xmath65 .
such windows correspond to slices of width @xmath66 m along the acoustic axis separated by @xmath67 m . as shown in fig .
[ f.correl ] , the correlation function @xmath68 is then interpolated around its main maximum by a parabola and the value @xmath53 of @xmath69 that maximizes this parabola is extracted : @xmath70 where @xmath71 denotes the parabolic interpolation of @xmath72 .
note that @xmath53 depends both on the position @xmath22 of the time window ( fast `` ultrasonic time '' ) and on the time @xmath56 for which the cross - correlation is performed ( slow `` pulse time '' ) .
thus @xmath53 will be noted as a function of the two variables @xmath22 and @xmath56 below . finally , it is straightforward to convert time - shifts @xmath53 into velocities using : @xmath73 where @xmath74 is the position of the center of the window along the acoustic axis . the origin @xmath75 is taken at the stator fluid interface .
equation ( [ e.vy ] ) thus yields the projection @xmath76 of the velocity vector along the acoustic axis at the discrete positions @xmath23 and at time @xmath56 . the procedure based on the interpolation of @xmath68 allows us to measure time - shifts as small as @xmath77 ns that correspond to displacements @xmath78 m .
since the prf @xmath79 can be made as small as desired , infinitesimal velocities could in principle be measured .
however , at very small shear rates , decorrelation of the bs signals may occur due to brownian motion of the scatterers or to low - frequency mechanical vibrations in the experimental setup .
consequently , we will avoid using prfs smaller than 10 hz and we estimate the minimum measurable velocity at about 1 @xmath0m.s@xmath30 .
on the other hand , the maximum measurable velocity is reached when @xmath80 m .
indeed , displacements greater than the acoustic wavelength can not be measured using the cross - correlation algorithm described above unless phase unwrapping and continuity conditions are implemented , which will not be considered here @xcite . with the highest achievable prf @xmath81 khz
, a 40 @xmath0 m maximum displacement means that velocities above 0.8 m.s@xmath30 may not be accessed using usv .
note however that this limiting range of [ 1 @xmath0m.s@xmath30 0.8 m.s@xmath30 ] applies to the velocity @xmath76 projected along the acoustic axis .
this range may be extended by varying the incidence angle @xmath41 ( see sec .
[ s.conversion ] and eqs .
( [ e.v])([e.x ] ) below ) .
finally , eq . ( [ e.vy ] ) shows that a velocity profile could in principle be obtained by cross - correlating only two successive bs signals _
i.e. _ within a time interval @xmath83 that can be made very small depending of the prf . in practice , as seen on fig .
[ f.signal ] , the speckle amplitude is never uniform .
locally , destructive interferences or the absence of scatterers may lead to signal levels too small to be analyzed ( see the signal of fig .
[ f.signal ] around @xmath84 or @xmath85s for instance ) .
time windows where the signal amplitude does not reach a given threshold ( typically 20 % of the maximum amplitude ) are thus left out of the analysis . to recover a full velocity profile , some averaging
is then performed over several sucessive cross - correlations : @xmath86 figure [ f.data ] shows twenty successive bs signals recorded in a sheared newtonian suspension with @xmath87 khz .
the ultrasonic pulses leave the stator and enter the gap at @xmath63s and a small fixed echo corresponding to the rotor position may be seen at @xmath88s ( as well as a stronger echo at about 15.2 @xmath0s that corresponds to a multiple reflection from the plexiglas window in the stator ) .
the slopes of the traces left by the echoes in this two - dimensional @xmath89 diagram are inversely proportional to the local velocities according to eq .
( [ e.vy ] ) .
thus , the signature of shear is rather clear : velocities increase from the stator , where the scatterers remain almost fixed , to the rotor .
moreover , the presence of moving echoes for times @xmath90 is not surprising : these echoes simply correspond to scattering of the wave reflected on the rotor .
one can see that the whole range of times from @xmath64 to @xmath91 gets covered by some significant speckle signal in about 20 ms in that case .
if better accuracy is required , more averaging can be performed to improve the statistical convergence of the method ( see fig .
[ f.converg ] below ) . depending on the trade - off between accuracy and temporal resolution
, we found that usv allows us to obtain a full velocity profile every 0.022 s.
in order to test and calibrate the experiment , much effort was devoted to developing `` contrast agents '' that feature good scattering properties for high - frequency usv .
indeed , too weak a scatterer yields a signal level too low to work with , whereas too strong a scatterer leads to multiple scattering .
when the scatterer is small compared to the acoustic wavelength ( rayleigh approximation ) , three parameters control the amount of scattering : the scatterer diameter , its compressibility and its density @xcite . moreover ,
if one wants the contrast agents to remain lagrangian tracers of the flow , their density has to be matched to that of the fluid .
we found that homemade polystyrene spheres of diameter @xmath9210
@xmath0 m were large and hard enough to scatter 36 mhz pulses efficiently but soft enough to prevent multiple scattering when diluted at a weight concentration of 1 % in water .
such polystyrene spheres were obtained by polymerization following refs .
a 1 % wt .
solution of benzoyl peroxide in divynylbenzene is first prepared .
then , a polydisperse emulsion is obtained by stirring 40 g of the previous mixture in 59 g of water and 1 g of sodium dodecyl sulfate at room temperature .
finally , the emulsion is polymerized at 90@xmath93c for two hours .
the prepared polystyrene spheres were washed with water three times before drying .
the 1 % wt .
suspension of such polystyrene spheres is newtonian and has the same viscosity as water . in order to use eqs .
( [ e.vy ] ) and ( [ e.yk ] ) , the sound speed @xmath25 of the working fluid has to be measured .
we used a classical _ transmission _ setup consisting of two transducers facing each other .
one of them , the emitter , remains fixed while the other works as a receiver and can be moved by a computer - controlled actuator .
pulses are recorded for various displacements @xmath94 of the receiver and averaged over 100 sweeps .
the time - shifts @xmath53 between the arrival times of the pulses are measured and the sound speed @xmath25 is given by the slope of @xmath94 vs. @xmath53 ( see fig . [ f.soundspeed ] ) .
due to technical constraints , the measurements of @xmath25 were performed with 25 mhz transducers .
temperature is controlled to within 0.1@xmath93c by a water circulation around the sample .
figure [ f.soundspeed ] shows that @xmath95 m.s@xmath30 leads to a very good approximation of the data obtained in the suspension of polystyrene spheres at our working temperature of 32@xmath93c .
this value of @xmath25 is in good agreement with the measurements reported in the literature for the sound speed of water vs. temperature @xcite . in the following
, we will assume that this value does not vary significantly with frequency between 25 and 36 mhz and all the data will be taken at the same temperature of 32@xmath93c .
finally , note that the measurement of @xmath25 using the transmission of ultrasonic pulses between two transducers also allows us to check that no significant multiple scattering takes place in our fluids by looking at the transmitted waveform : the received signal remains always as short as the emitted pulse and no long - lasting echoes typical of multiply scattered waves were ever recorded during the measurements of @xmath25 .
another important parameter is @xmath64 that gives the position of the stator
fluid interface and strongly depends on the experimental conditions .
figure [ f.brut ] presents the results of our speckle tracking algorithm averaged over 50 series of 20 pulses like that shown in fig .
[ f.data ] .
the rotor velocity @xmath96 was varied between 2.9 and 47.0 mm.s@xmath30 .
the linear behavior of the time - shift @xmath53 with the position @xmath43 of the correlation window is a direct signature of the uniform shear flow inside the newtonian suspension .
indeed , all the velocities considered here are far below the onset of the taylor - couette instability which occurs when @xmath97 ( where @xmath98 [email protected]@xmath30 is the kinematic viscosity of our calibration fluid ) @xcite _ i.e. _ when @xmath100 mm.s@xmath30 . for our dilute suspension
, no wall slip is expected and @xmath53 should go to zero at the stator : @xmath101 . by linearly fitting @xmath102 vs. @xmath43 for various values of @xmath96 , different estimates of @xmath64
are obtained that yield an average value @xmath103s .
the uncertainty on @xmath64 is due to the presence of the stator which leads to a small fixed echo on the signals and to small edge effects around @xmath64 ( see inset of fig .
[ f.brut ] ) .
it corresponds to an uncertainty of about @xmath104 m on the position of the stator
fluid interface .
inserting the values of @xmath25 and @xmath64 found above in eqs .
( [ e.vy ] ) and ( [ e.yk ] ) yields the velocity profile @xmath106 .
thus , so far , we obtained measurements of the velocity vector projected along the acoustic axis . in order to get some physically more relevant information
, we will assume from now on that the radial velocity is zero ( or at least negligible when compared to the orthoradial velocity ) .
this is of course the case for our newtonian suspension below the onset of the taylor - couette instability . in complex fluids , inhomogeneous flows with
wall slip or shear banding are purely orthoradial ( see sec .
[ ss.inhom ] ) so that the above approximation remains valid as long as hydrodynamic or elastic instabilities do not occur .
note that in the geometry discussed here , usv measurements are not affected by a non - zero vertical component of the velocity field . assuming that @xmath107 and using standard trigonometric relations ( see fig . [ f.convert ] )
, one gets : @xmath108 where @xmath109 in the small gap approximation , eqs .
( [ e.v ] ) and ( [ e.x ] ) reduce to @xmath110 and @xmath111 .
thus the last parameter to calibrate is the angle @xmath41 .
this is done by looking for a value of @xmath41 for which @xmath105 coincides with the velocity profile expected for a newtonian fluid : @xmath112\simeq{v_{\scriptscriptstyle 0}}\left(1-\frac{x}{e}\right)\ , , \label{e.vnewt}\ ] ] where the last term results from the small gap approximation .
figure [ f.calib ] shows that @xmath113 leads to very good results for all the rotor velocities @xmath96 considered here .
normalized data @xmath114 vs. @xmath115 show that the newtonian velocity profiles @xmath116 are recovered for all our data with the set of parameters @xmath25 , @xmath64 , and @xmath41 inferred from the above calibration ( see inset of fig . [ f.calib ] ) .
this calibration method is very sensitive to small variations of @xmath41 and allows its estimation to within @xmath117 .
note that @xmath41 depends on both the temperature and the nature of the fluid under study through @xmath25 .
as long as the temperature remains the same , the new value @xmath118 of the incidence angle in a fluid whose sound speed @xmath119 differs from that of the calibration fluid is simply given by snell s law : @xmath120 where @xmath25 and @xmath41 are the parameters measured for the calibration fluid .
finally , fig .
[ f.converg ] shows the value of the @xmath76 measured at two different positions along the acoustic beam as more and more averaging is performed .
it is clear that convergence is reached when @xmath121 pulses are taken into account .
the standard deviation of the measurements decreases from about 6 % for @xmath122 to less than 3 % for @xmath123 .
let us now turn to an application of high - frequency usv to a complex fluid .
we chose to test usv in a lyotropic lamellar phase composed of sds ( 6.5 % wt . ) , octanol ( 7.8 % wt . ) and brine ( 20 g.l@xmath30 nacl ) .
this surfactant mixture is known to display a series of structural shear - induced transitions @xcite . at equilibrium ,
the system is composed of surfactant bilayers stacked with a smectic distance of about 15 nm . under shear flow ,
this lamellar phase was shown to form a closely packed assembly of multilamellar vesicles @xcite .
the diameter of these vesicles , commonly called `` onions , '' is typically a few microns . at low shear velocities ,
the diffraction pattern observed with light scattering is an isotropic ring indicating the absence of any positional order .
however , when the velocity @xmath96 is increased , bragg peaks appear in the diffraction pattern showing the apparition of long - range order in the vesicle positions . as sketched in fig .
[ f.transit ] , this disorder - to - order transition corresponds to a sharp drop in the sample viscosity by a factor of about 3 @xcite .
such a shear - thinning transition is also observed in colloidal suspensions and is called a `` layering '' transition @xcite . as explained in sec .
[ ss.inhom ] , inhomogeneous velocity profiles are expected in the vicinity of this shear - induced transition where the two states ( ordered and disordered states ) may coexist in the gap of a couette cell . as for its ultrasonic behavior ,
the `` onion '' assembly does not scatter 36 mhz pulses .
thus , the lamellar phase was seeded with the polystyrene spheres used in sec .
[ s.calib ] in order to get some speckle signal from the fluid .
we checked that the `` layering '' transition is not significantly modified by the addition of such contrast agents .
figure [ f.photos ] presents two pictures of the seeded lamellar phase as seen under crossed polarizers and after shearing : the granular texture characteristic of disordered multilamellar vesicles does not seem perturbed by the addition of polystyrene spheres at 1 % wt . as shown on fig .
[ f.soundspeed ] , the sound speed of the seeded lamellar phase was measured to be @xmath124 m.s@xmath30 .
since the experiments and the calibration are performed at the same temperature and since the sound speeds of both fluids are the same ( within less than 1 % ) , we used @xmath95 m.s@xmath30 , @xmath125s and @xmath113 to obtain the velocity profiles shown in figs .
[ f.lam1 ] and [ f.lam2 ] .
the two main features of the velocity profiles are as follows : [ [ wall - slip .
] ] wall slip .
+ + + + + + + + + + as clearly shown by the insets of figs .
[ f.lam1 ] and [ f.lam2 ] , the velocity close to the walls is never equal to that of the walls . on the contrary ,
@xmath126 may differ from @xmath96 by more than 30 % .
one can also notice that slip velocities are comparable at the rotor and at the stator for the smallest rotor velocity but become highly dissymmetric when @xmath127 mm.s@xmath30 .
a detailed study of wall slip in this system is left for future work . [ [ shear - banding . ] ] shear banding .
+ + + + + + + + + + + + + + figure [ f.lam1 ] also unveils the presence of shear bands when @xmath128 mm.s@xmath30 .
a highly sheared band nucleates close to the rotor and progressively fills the gap as @xmath96 is increased ( see fig .
[ f.lam2 ] for the highest velocities ) . the shear rate in the high shear band is about three times larger than that in the low shear band .
this ratio remains roughly constant as long as two bands coexist . for @xmath129 mm.s@xmath30
, the highly sheared region has invaded the whole gap .
when @xmath130
mm.s@xmath30 , the flow is laminar again and looks similar to that observed before the `` layering '' transition at @xmath131 mm.s@xmath30 .
simultaneously to the present ultrasonic measurements , dls data obtained in our group have confirmed the existence of shear - banded flows and strong wall slip effects in the same lamellar phase without seeding polystyrene particles @xcite .
this last observation shows _ a posteriori _ that our `` contrast agents '' do not perturb the flow and can be treated as simple lagrangian tracers . both usv and dls data are in good qualitative agreement with classical theoretical pictures of shear - banded flows @xcite .
our purpose here is only to show the relevance of high - frequency usv in the field of sheared complex fluids and a more quantitative analysis of shear banding in light of dls results can be found in ref .
@xcite .
finally let us show how usv may be used to follow the dynamics of a complex fluid under shear .
until now , we have considered time - averaged velocity profiles . however ,
during the `` layering '' transition , noticeable fluctuations of the flow field occur and result in large error bars for @xmath132 or 29.4 mm.s@xmath30 for instance .
actually , some very rich dynamics were reported in this sheared lamellar phase when the _ torque _ applied on the rotor is imposed instead of the _ velocity _
@xcite . in that case , the viscosity @xmath21 displays large , complex fluctuations on long time scales of the order of 100 s. similar ( although faster ) temporal behaviors were also observed in other surfactant systems @xcite . the origin of these fluctuations and their precise statistical properties remain unclear .
one possibility is that the flow presents shear bands that may couple to each other and lead to the observed behavior of the viscosity .
thus the dynamics could be not only temporal as observed on @xmath21 but also spatial as suggested in ref .
@xcite .
figure [ f.lam3 ] shows a transient regime recorded after the torque applied to the rotor was suddenly increased from 83 to 87 mn at time @xmath133 .
the response of the rotor velocity @xmath134 is shown in fig .
[ f.lam3](a ) @xcite .
a maximum of about 38 mm.s@xmath30 is reached at @xmath135 s before @xmath134 slowly relaxes with small fluctuations .
velocity profiles recorded during that transient show that : ( i ) during the initial increase of @xmath134 , a shear - banded flow is generated with _
three _ bands : two highly sheared bands at both the rotor and the stator and a low shear region in between ( see fig .
[ f.lam3](b)(d ) ) ; ( ii ) the band located at the stator disappears when @xmath134 decreases , leaving only _ two _ bands similar to figs .
[ f.lam1 ] and [ f.lam2 ] ( see fig . [ f.lam3](e)(f ) ) ; ( iii ) states where the highly sheared band almost disappears can be observed ( see fig . [ f.lam3](g ) ) even though the rotor velocity @xmath136 mm.s@xmath30 indicates that a band should extend over almost half the gap if the average picture at imposed @xmath96 remained valid when the torque is imposed ( see the @xmath137 symbols corresponding to @xmath138 mm.s@xmath30 in fig .
[ f.lam1 ] ) .
the above results show that high - frequency usv is a powerful tool for measuring velocity profiles in sheared complex fluids .
we have successfully tested the technique on a lyotropic lamellar phase that undergoes a shear - induced `` layering '' transition . when the rotor velocity , _
i.e. _ the global shear rate , is imposed , shear banding is observed : time - averaged velocity profiles display a highly sheared region that nucleates from the rotor and progressively invades the gap as @xmath96 is increased .
wall slip can also be very large especially during the transition , which makes global measurements such as the sample viscosity tricky to analyze . moreover , usv allowed us to record velocity profiles in 1 s typically .
this acquisition time is short enough ( compared to the intrinsic time scales of our complex fluid ) to follow the velocity profiles in time .
we showed the existence of more complex , three - band flows during a transient at imposed torque _
i.e. _ imposed shear stress .
this raises the question of the structure of the flow during the spontaneous oscillations of the viscosity observed in the same lamellar phase @xcite .
usv may thus provide a `` time - resolved '' tool for investigating spatially the dynamical behaviors of @xmath21 . in this paper
, we have shown that we were able to measure velocity profiles with a spatial resolution of about @xmath139 m and every 0.022 s. high - frequency usv compares very well with other local techniques for measuring flow fields .
ldv , or equivalently dls , provides in principle a slightly better spatial resolution of 1050 @xmath0 m but only yields a pointlike measurement . in order to obtain a full velocity profile with ldv or dls
, one has to mechanically scan the whole gap of the cell @xcite .
this usually takes about 1 min and may raise major interpretation problems when the flow evolves on time scales shorter than 1 min .
nmr appears as a very promising technique for complex fluids @xcite since no seeding is required and two - dimensional images of the flow are obtained .
however , acquisition times are rather long too ( from about 20 s to more than an hour depending on the number of points in the image ) , making it rather difficult to access dynamical local information .
the main drawback of usv is that it needs a good control of the scattering properties of the fluid and that it may require that the flow be seeded with contrast agents .
unless their mechanical properties is strictly controlled , most concentrated suspensions or emulsions will lead to multiple scattering of high - frequency pulses .
thus , further application of usv to other complex fluids will involve a careful study of their acoustic properties .
however , we believe that usv is likely to extend the range of complex fluids in which local measurements may be performed , in particular to optically opaque fluids . the biggest advantage of usv lies undoubtedly in its ability to perform nonintrusive velocity measurements fast enough to follow the dynamics of complex fluids under low shear . in principle , in an ideal scattering environment , the temporal resolution of usv could even get as short as 1 ms per measurement .
future research directions on usv are as follows : ( i ) more measurements on seeded lamellar phases will be performed and compared to dls data ; more precisely , we will focus on the spatio - temporal flow dynamics in regimes where sustained oscillations of the viscosity are observed ; ( ii ) we will investigate shear flows of other fluids such as emulsions , gels , or granular pastes ; ( iii ) we also intend to apply high - frequency usv to the case of very small deformations in order to perform some `` local linear rheology '' and understand how a complex fluid starts to flow in the transition from linear to nonlinear rheology . the authors wish to thank the `` cellule instrumentation '' at crpp for designing and building the mechanical parts of the experimental setup .
we are very grateful to d. roux , j .- b .
salmon , and r. wunenburger for fruitful discussions .
e. cappelaere , j .- f berret , j .- p .
decruppe , r. cressely , and p. lindner .
rheology , birefringence , and small - angle neutron scattering in a charged micellar system : evidence of a shear - induced phase transition . , 56:18691878 , 1997 .
d. h. turnbull , b. g. starkoski , k. a. harasiewicz , j. l. semple , l. from , a. k. gupta , d. n. sauder , and f. s. foster .
a 40100 mhz b - scan ultrasound backscatter microscope for skin imaging .
, 21:7988 , 1995 .
d. a. christopher , p. n. burns , b. g. starkoski , and f. s. foster . a high - frequency pulsed - wave doppler ultrasound system for the detection and imaging of blood flow in the microcirculation .
, 23:9971015 , 1997 . | high - frequency ultrasonic pulses at 36 mhz are used to measure velocity profiles in a complex fluid sheared in the couette geometry .
our technique is based on time - domain cross - correlation of ultrasonic speckle signals backscattered by the moving medium .
post - processing of acoustic data allows us to record a velocity profile in 0.022 s with a spatial resolution of 40 @xmath0 m over 1 mm . after a careful calibration using a newtonian suspension ,
the technique is applied to a sheared lyotropic lamellar phase seeded with polystyrene spheres of diameter 310 @xmath0 m .
time - averaged velocity profiles reveal the existence of inhomogeneous flows , with both wall slip and shear bands , in the vicinity of a shear - induced `` layering '' transition .
slow transient regimes and/or temporal fluctuations can also be resolved and exhibit complex spatio - temporal flow behaviors with sometimes more than two shear bands . |
the violation of the cp symmetry is now established in the b meson sector and the parameter @xmath3 is measured with precision by babar @xcite and belle @xcite . however , the measurement of the other angles of the unitarity triangle are necessary for a more comprehensive study of cp violation .
these measurements suffer from several difficulties related to the uncertainties of the theory , the limited statistics and/or ambibuities in their extraction .
we have recently proposed a new method @xcite involving three - body b meson decays , which circumvent these problems and allows to determine the angle @xmath4 . in this paper
we extend this method to the weak phase @xmath5 , which can be identified to the combination @xmath6 of the unitarity triangle angles @xcite .
important constraints on the theory will be obtained from the measurement of this phase . in the following ,
we describe a new _ time dependent _ method involving the entire area of the dalitz plot of the 3-body decays @xmath7 . +
the method involves no reconstruction into @xmath8 eigenstates and avoids problems with interfering dcsd .
[ sec : method - infinite - stat ] we investigate a way to circumvent the color suppression penalty by using @xmath1 decay modes which potentially offers significantly large branching fractions and cp asymmetries .
the particular decays considered here are of the type .
these three body decays may be obtained by popping a @xmath10 pair in color allowed decays .
although modes where one or more of the three final state particles is a vector can also be used , for clarity and simplicity only the mode @xmath7 is discussed here . to the diagrams leading to the final states of interest are shown in fig .
[ fig : diagrams ] . as can be seen , the diagrams ( fig .
[ fig : diagrams]a and [ fig : diagrams]c ) are both color - allowed and of order @xmath11 in the wolfenstein parameterization @xcite , where @xmath12 is the cabibbo mixing angle .
the general formalism describing how the weak angle is extracted from observation of the time - dependent cp violation has been first presented in @xcite . however , in the present analysis additional information is used to resolve the inherent ambiguities related to the method . selecting a particular point @xmath13 in the dalitz plot ,
let us write the amplitudes corresponding to the transitions in fig .
[ fig : diagrams]a and b on the one hand and [ fig : diagrams]c and d on the other hand as @xmath14 and their cp conjugates @xmath15 where @xmath16 is the relative phase of the ckm matrix elements involved in this decay , and @xmath17 ( @xmath18 ) and @xmath19 ( @xmath20 ) are the real amplitudes and cp - conserving strong interaction phases
. + the time dependent probability density function ( pdf ) of the transition @xmath21 is @xmath22 where @xmath23 , @xmath24 with @xmath25 = @xmath26 and @xmath27 .
we have used the approximation @xmath28 .
+ the pdf of the transition @xmath29 is @xmath30 from the total number of events in each of the four final states ( [ eq : btodm ] ) and ( [ eq : btodp ] ) and a global fit of their time dependence , it is possible to extract the quantities , @xmath31 where @xmath32 and @xmath33 .
one obtains : @xmath34 hence , in the limit of very high statistics , one would extract @xmath35 for each point @xmath13 of the dalitz plot .
however , for every point of the dalitz plot , @xmath36 is obtained with an eight - fold ambiguity in the range @xmath37 $ ] , because of the invariance of the @xmath38 terms in eq .
( [ eq : btodm ] ) and ( [ eq : btodp ] ) under the three symmetry operations @xmath39 when the multiple measurements made in different points of the dalitz plot are combined , some of the ambiguity will be resolved , in the likely case that the strong phase @xmath40 varies from one region of the dalitz plot to the other .
this variation can either be due to the presence of resonances or because of a varying phase in the non - resonant contribution . in this case
, the exchange symmetry @xmath41 is numerically different from one point to the other , which in effect breaks this symmetry and resolves the ambiguity .
similarly , the @xmath42 symmetry is broken if there exists some _ a - priori _ knowledge of the dependence of @xmath40 on the dalitz plot parameters .
this knowledge is provided by the existence of broad resonances , whose breit - wigner phase variation is known and may be assumed to dominate the phase variation over the width of the resonance . to illustrate this , let @xmath13 and @xmath43 be two points in the dalitz plot , corresponding to different values of invariant mass of a particular resonance .
one then measures @xmath44 at point @xmath13 and @xmath45 at point @xmath43 , where @xmath46 is known from the parameters of the resonance .
it is important to note that the sign of @xmath46 is also known , hence it does not change under @xmath42 .
therefore , should one choose the @xmath42-related solution @xmath47 at point @xmath13 , one would get @xmath48 at point @xmath43 .
since this is different from @xmath45 the @xmath42 ambiguity is resolved .
this is illustrated graphically in eq .
[ eq : ssign ] , where @xmath49 is the weak phase : @xmath50 bw \downarrow & & \downarrow bw \\[-0.5ex ] \sin(\phi \pm ( \delta\delta_i \!+\ ! \alpha_{ij } ) ) & \stackrel{\ssign}{\not\!\!\!\longleftrightarrow}\ ! & \sin(\pi - \phi \mp ( \delta\delta_i \!-\ ! \alpha_{ij } ) ) \end{array } \hspace*{-1ex } \label{eq : ssign } \vspace*{-2ex}\ ] ] thus , broad resonances reduce the initial eight - fold ambiguity to the two - fold ambiguity of the @xmath51 symmetry , which is never broken . fortunately , @xmath51 leads to the well - separated solutions @xmath36 and @xmath52 , the correct one of which is easily identified when this measurement is combined with other measurements of the unitarity triangle .
since experimental data sets will be finite , extracting @xmath16 will require making use of a limited set of parameters to describe the variation of amplitudes and strong phases over the dalitz plot
. the consistency of this approach can be verified by comparing the results obtained from fits in a few different regions of the dalitz plot , and the systematic error due to the choice of the parameterization of the data may be obtained by using different parameterizations .
a fairly general parameterization assumes the existence of @xmath53 breit - wigner resonances , as well as a non - resonant contribution : @xmath54 where @xmath55 represents the dalitz plot variables , @xmath56 is the breit - wigner amplitude for a particle of spin @xmath57 , normalized such that @xmath58 , @xmath59 and @xmath60 ( @xmath61 and @xmath62 ) are the magnitude and cp - conserving phase of the non - resonant @xmath63 ( @xmath64 ) amplitude , and @xmath65 and @xmath66 ( @xmath67 and @xmath68 ) are the magnitudes and cp - conserving phase of the @xmath63 ( @xmath64 ) amplitude associated with resonance @xmath43 . the functions @xmath69 and @xmath70 may be assumed to vary slowly over the dalitz plot , allowing their description in terms of a small number of parameters .
the decay amplitudes of @xmath71 mesons are identical to those of eqs .
( [ eq : amp - param ] ) , with @xmath16 replaced by @xmath72 .
the decay amplitudes of eqs .
( [ eq : amp - param ] ) can be used to conduct the full data analysis . given a sample of @xmath73 signal events , @xmath16 and the other unknown parameters of eq .
( [ eq : amp - param ] ) are determined by minimizing the negative log likelihood function @xmath74 where the amplitude @xmath75 is given by one of the expressions of eqs .
( [ eq : amp - param ] ) , or their cp - conjugates , depending on the final state @xmath76 , and @xmath77 are the dalitz plot variables of event @xmath13 . in
what follows , we discuss important properties of the method by considering the illustrative case , in which the @xmath63 decay proceeds only via a non - resonant amplitude , and the @xmath64 decay has a non - resonant contribution and a single resonant amplitude .
for concreteness , the resonance is taken to be the @xmath78 .
we take the a priori @xmath55-dependent non - resonant phases to be @xmath79 .
[ sec : simulation ] to study the feasibility of the analysis using eq .
( [ eq : chi2 ] ) , we conducted a simulation of the decays @xmath80 and @xmath81 .
events were generated according to the pdf in eqs .
( [ eq : btodm ] ) and ( [ eq : btodp ] ) , with the parameter values given in table [ tab : reference ] . in this table and throughout the rest of the paper
, we use a tilde to denote the `` true '' parameter values used to generate events , while the corresponding plain symbols represent the `` trial '' parameters used to calculate the experimental @xmath82 . for simplicity ,
additional resonances were not included in this demonstration .
however , broad resonances that are observed in the data should be included in the actual data analysis .
.parameters used in the simulation .
the value of @xmath83 is chosen so as to roughly agree with the measurement of the corresponding branching fraction @xcite . [ cols="^,^,^,^",options="header " , ] in fig .
[ fig : sig - gamma - vs - gamma ] we present @xmath84 , the statistical error on @xmath36 as a function of @xmath85 and @xmath86 , obtained by fitting simulated event samples .
each point is obtained by repeating the simulation 250 times , to minimize sample - to - sample statistical fluctuations .
all the parameters of table [ tab : reference ] were determined by the fit , and the arrow in these figures indicates the point corresponding to the parameters of table [ tab : reference ] .
one observes that @xmath84 does not depend strongly on the value @xmath36 . as expected , a strong dependence on @xmath87 is seen .
however , it should be noted that @xmath88 changes very little for all values of @xmath89 above 0.4 .
this suggests that the potential for a significantly sensitive measurement is high over a broad range of parameters .
we have shown how @xmath36 may be measured in the color - allowed decays , focusing on the simplest mode @xmath7 .
the absence of color suppression in the @xmath63 amplitudes is expected to result in relatively large rates and significant cp violation effects , and hence favorable experimental sensitivities .
while the dalitz plot analysis implies some experimental complication , it should reduce the eight - fold ambiguities , which are a serious problem with other methods for measuring @xmath36 . as a result of these advantages ,
this method is likely to lead to a measurement of @xmath36 even with the current generation of @xmath90 factory experiments .
the authors thank francois le diberder for his fruitful ideas and help with simulation .
this work was supported by cea and the danish research academy .
9 babar collaboration ( b. aubert _ et al .
_ ) , phys . rev .
* 86 * ( 2001 ) , 2515 .
belle collaboration ( b. abe _ et al .
_ ) , phys . rev .
* 87 * ( 2001 ) , 091802 .
r. aleksan , t. petersen and a. soffer , phys .
d. * 67 * ( 2003 ) 096002 .
r. aleksan , b. kayser and d. london , phys .
lett . * 73 * , ( 1994 ) 18 .
l. wolfenstein , phys .
* 51 * ( 1983 ) 1945 .
r. aleksan , i. dunietz , b. kayser , f. le diberder , nucl .
b361 ( 1991 ) 141 .
the cleo collaboration , phys .
88 , 101803 ( 2002 ) . | we present a method to measure the weak phase @xmath0 in the three - body decay of neutral @xmath1 mesons to the final states @xmath2 .
these decays are mediated by interfering amplitudes which are color - allowed and hence relatively large . as a result , large cp violation effects can be observed with high statistical significance .
in addition , the three - body decay helps to resolve discrete ambiguities present in measurements of the weak phase .
the experimental implications of conducting these measurements are discussed , and the sensitivity of the method is evaluated using a simulation . |
in both high energy and condensed matter physics , there is a long tradition of studying systems which exhibit topological excitations .
recently , this field has received a new impetus , since it has been realized that such topological excitations may permit fault tolerant storage and manipulation of quantum information @xcite . in connection with this , there are current experimental efforts to prove the existence of nontrivial topological phases in the fractional quantum hall effect@xcite and to construct such phases in josephson junction networks @xcite .
topological excitations are usually introduced at the classical level as solutions to the equations of motion and the observables that distinguish them are directly linked to topologically invariant properties of these solutions . this places topological particles in marked contrast to the more usual perturbative ( quasi)particles . the latter are described as low energy quantum fluctuations over a given vacuum state .
the group of symmetries of the system that fixes the vacuum state will act on the fluctuations and cause them to organize into multiplets . as a result these perturbative particle states form irreducible representations of the symmetry group . for topological particles
no such labeling is obviously present .
a similar dichotomy exists when considering the ground states of different phases , or the order parameters that distinguish between phases .
traditionally , when a phase exhibits ground state degeneracy , the different ground states would be related by the action of symmetry operators , but in topological phases , ground state degeneracies appear for models on topologically nontrivial spatial manifolds without the obvious intervention of any symmetry , and in fact the different ground states can often not be mixed by any local operator .
analogously , traditional phases can be distinguished by the expectation values of local order parameters , while different topological phases may exist which are not distinguished by any local order parameter . as a result
a number of indicators for _ topological order _ which are not based on symmetry have emerged , notably the dimensions of the spaces of ground states on spatial surfaces of non - trivial topology @xcite and the topological entanglement entropy @xcite . despite the fact that topological phases may not be fully characterized by their symmetries ( or at least not by symmetries represented by local operators ) , one may often still organize the excitation spectra of such phases by using ` symmetries ' that are not obvious from the hamiltonian or lagrangian of the system and which may in fact not be realized locally .
thus one may hope for a generalization or analogue of the theory of symmetry breaking phase transitions which applies topological phases , by allowing for such ` topological symmetries ' .
one of the main goals of this paper is to set up such a formalism for the particular case of phase transitions which occur due to the formation of a condensate of bosonic quasiparticles . before we go into a further description of this formalism ,
let us make some remarks which we hope may prevent confusion in reading the rest of the paper .
first of all , we do not want to limit ourselves to ` strictly topological phases ' , which have no nontrivial symmetries represented by local operators .
in fact we will include theories which have a discrete symmetry represented by local operators but no nontrivial topology as a special case .
we would like to point out that in gauge theories , electric charges , which are supposedly ` non - topological particles ' , coming from the locally represented gauge symmetry of the system , can have nontrivial topological interactions with magnetic fluxes through the aharonov - bohm effect , so in order to describe the full topological order of gauge theories it is necessary to take the usual gauge symmetry into account .
secondly , we will often speak of ` symmetry breaking ' when some of the symmetries involved may be gauge symmetries .
gauge theories can be interpreted to a certain extent as constrained systems ; some of the gauge degrees of freedom are auxiliary and could in principle be eliminated at the price of introducing very complicated interactions among the true physical degrees of freedom . in a theory with a gauge symmetry the physical states are the gauge invariant states , so the spectrum does not manifestly exhibit the degeneracies of nontrivial representations in the spectrum and one may wonder at the idea of a symmetry breaking phase transition . however , despite the absence of gauge - variant states , the physics of gauge theories certainly depends on the invariants characterizing the representations that are present in the model and one speaks of gauge symmetries as ` hidden symmetries ' .
a similar situation occurs in topological field theory , where the particles often do not have internal degrees of freedom on which a symmetry could act , but nevertheless , their fusion rules can be described by the representation theory of a quantum group .
hidden symmetry breaking is to a large extent analogous to the usual breaking of global , non - gauge symmetries . in the global case , there is typically a local order parameter that breaks the symmetry and as a result , in the broken phase the degeneracies due to the original symmetry are ( partially ) lifted , and the spectrum is now organized in representations of the smaller residual symmetry group .
a gauge symmetry can not be broken by a local order parameter ( by elitzur s theorem@xcite ) . yet
, condensates with invariant order parameters are allowed and the hidden symmetry can effectively be reduced due to such a condensate .
this phenomenon is usually referred to as the higgs effect or the ` breaking ' of a hidden ( or local ) symmetry . bearing this warning in mind
our philosophy is to use the term ` breaking ' in this cavalier way .
a well known example of topological symmetry occurs in gauge theories with gauge group @xmath0 defined on a lattice in dimensions@xcite in such gauge theories , the spectrum consists of charges , whose internal state transforms under the @xmath0 gauge group , magnetic fluxes ( where flux is the topological quantum number ) , which are gauge invariant and composites of charge and flux , called dyons .
one sees immediately that the @xmath0 representation label on the particles is not enough to completely fix the sector , since it is blind to the flux .
however , one may introduce a second dual @xmath0 symmetry which is not a gauge symmetry , and which acts on fluxes in the same way that the original gauge group acts on charges and which leaves states without flux invariant .
the topological sectors of the theory are then completely distinguished by their behavior under the full @xmath1 symmetry and the addition of flux and charge quantum numbers is also captured by the tensor product of @xmath1 representations .
one may even include the aharonov - bohm braid interactions between charges and fluxes by introducing a new structure on @xmath2 called the universal @xmath3-matrix .
in general , one can not expect to capture the full particle spectrum and topological interactions of a physical system using only group theory .
still , it is believed that every type of topological order in 2 + 1 dimensional systems can be described using the representation theory of a modular tensor category or , dually , a quantum group .
the different types of quasiparticle correspond to the irreducible representations of the quantum group and the fusion and braiding interactions are described by the tensor product of representations and the @xmath3-matrix respectively .
this asks for generalization of the theory of symmetry breaking phase transitions to include topological symmetries described by quantum groups . in earlier work@xcite ,
we have developed such a theory of ` spontaneous ' quantum group symmetry breaking and applied it to discrete gauge theories .
this theory was later refined and applied to phase transitions in quantum nematics and other systems@xcite .
recent work of bombin and martin - delgado @xcite also provides interesting realizations of such transitions in models based on kitaev s toric code model@xcite , which exhibits the same topological order as the discrete gauge theories . that may have irreps which are in fact confined .
the low energy effective theory of the condensed phase is based on the unconfined algebra @xmath4.,width=307 ] the general features of our symmetry breaking scheme are as follows ( see figure [ fig : symscheme ] ) . before condensation ,
the system is described by a quantum group @xmath5 and each particle carries an irreducible representation ( irrep ) of @xmath5 .
when particles carrying the representation @xmath6 condense , the condensate will have an order parameter which is a state in the module of the representation @xmath6 .
the symmetry of the condensed phase should leave this order parameter invariant and hence the quantum group @xmath5 is broken down to a hopf subalgebra @xmath7 whose representations characterize the excitations of the condensed phase . depending on their braiding interaction with the condensed particles , these excitations may or may not be confined . in particular ,
if an excitation over the condensate has nontrivial braiding with the condensed particle then the order parameter of the condensate will not be single valued near this excitation and the excitation will pull a string or wall in the condensate and be confined ( the energy required for the creation of the string will be linear in the string s length , since the condensate is destroyed near the string ) .
the non - confined particles are particles in the true sense of the word , that is , point - like excitations , and their interactions are described by the representation theory of a ` hopf quotient ' @xmath4 of @xmath8 ( @xmath4 is the image of @xmath8 under a surjective map that preserves the hopf algebra structure ) .
the strings pulled by the confined particles can also be studied and they are classified by the representations of a subalgebra of @xmath8 which is determined by the hopf map from @xmath8 onto @xmath4 and which is analogous to the kernel of a homomorphism between groups .
and @xmath9 representations of @xmath10 ) .
the states on the outer edge belong to representations that are not confined in the outer region ( the @xmath9 and the @xmath11 of @xmath12).,width=240 ] in the sequel , we will devise a treatment of hopf symmetry breaking which generalizes the treatment given in our earlier papers , while keeping the amount of quantum group or hopf algebra theory that is needed to a minimum .
therefore , it will not be necessary to flesh out the details of the algebras contained in diagram [ fig : symscheme ] ( detailed definitions can be found in the original papers ) .
however , the structure of the diagram will be preserved , in that there will still be two levels to our treatment of condensation - induced phase transitions .
first , the symmetry is broken , leading to a spectrum of excitations that occur in the broken phase , or on an interface between domains that support the broken and unbroken phases .
then these excitations are separated into confined and non - confined excitations , where the word ` confined ' means that the excitations are either bound to a boundary between a domain in the broken phase and a domain in the unbroken phase , or bound together like quarks in hadrons .
this should lead us finally to a description of the fusion and braiding properties of the non - confined particles and ultimately also to a description of the strings pulled by the confined particles and a classification of hadronic composites of confined particles .
the breaking schemes we discuss may also have interesting applications in describing the physics of spatial geometries where interfaces between different topological phases occur , that may be enforced by external means ( by applying different magnetic fields for example ) .
we give an example of such a two phase geometry in figure [ fig : edgeinterface ] , where the interface carries only those edge states of the interior disc ( with phase i ) , which are confined in phase ii of the outer region , the outer edge carries ( non - confined ) states that are also allowed in the outer region .
we will return to the specific phases indicated in the figure later on . in line with this application one may also draw conclusions on the boundary theory of certain two layer systems as will be explained in sections [ sec : embeddings2 ] and [ sec : doublec - s ] .
let us quickly review the minimal knowledge of @xmath13-dimensional topological field theory that we will need for the rest of the paper . for much more detail , the reader may consult for instance refs . .
first of all , we assume that the theory has a finite number of topological sectors , labeled by some finite set of labels . we will call these _ anyonic charges _ , or in some cases _ topological charges _ , especially when it is not obvious if we are dealing with charges characterizing point - like excitations .
we can think of these charges as topological quantum numbers , but also as charges related to some ( group ) symmetry and sometimes , like in the case of @xmath0 gauge theory described before , we can think of them either way .
in many physical situations , it is necessary to introduce superselection sectors which correspond to the same topological charges but which have different non - topological quantum numbers ( these may for instance characterize short - range interactions ) . for the purposes of this paper , we will ignore these and consider such sectors to be the same . as a result
we can describe theories which include electric charge and other gauge charges as if they have only finitely many sectors .
the basic interactions between topological sectors in 2 + 1 dimensions are _ fusion _ and _ braiding_. fusion may be summarized using _ fusion rules _ of the form @xmath14 here @xmath15 and @xmath16 are the two topological charges which are to be fused , the labels @xmath17 are the possible overall topological charges of the result of the fusion and the integer coefficients @xmath18 indicate the number of independent couplings between @xmath15 and @xmath16 that give @xmath17 .
concretely , a zero fusion coefficient means that the charges @xmath15 and @xmath16 can not fuse to @xmath17 and a nonzero coefficient means they can .
one may think of the fusion coefficient @xmath18 as the dimension of the space of low energy states of a piece of topological medium which has overall topological charge @xmath17 and which contains two topological excitations with charges @xmath15 and @xmath16 .
a physical requirement on fusion rules is that they must be associative , that is @xmath19 we also require that there is a unique vacuum sector , labeled @xmath20 or @xmath21 , depending on the context , which has the property that @xmath22 for any other sector @xmath15 .
finally , we require that any sector @xmath15 has a charge conjugate sector , denoted @xmath23 , with which it can fuse to the vacuum in a unique way , i.e. @xmath24 for systems with well defined braiding interactions ( that is , all two - dimensional systems with pointlike excitations ) , we will also have ` symmetry of the fusion interaction ' , that is @xmath25 .. ] any topological charge @xmath15 has a _ spin factor _ @xmath26 associated with it .
this is a phase factor that the wave function of the anyonic system picks up when the anyon is rotated ( twisted ) over a @xmath27 angle .
we can think of the particle as being in an eigenstate of two dimensional spin and the spin factor is the effect of a @xmath27 rotation on this eigenstate .
we will also use the spin @xmath28 of the particle , which is related to the spin factor by @xmath29 . for systems with finitely many topological sectors ,
the spins are always rational@xcite .
adiabatic exchanges of the particles ( without twisting ) also have an effect on the wave function of the system .
this is the analogue of the statistical interaction of fermions or bosons ( for bosons the fact that there is no effect of the exchanges actually tells us a lot about collective behaviour ) . in two dimensions
these exchanges are governed by the braid group , rather than the permutation group .
in particular this means that left over right exchanges are not the same as their inverses , the right over left exchanges .
the product of two right over left exchanges is often called the _
monodromy_. it returns the excitations to their original positions , but may nevertheless have a nontrivial effect on the state of the system .
this effect may be described in terms of fusion and twisting , using the so called ` ribbon equation ' whose pictorial representation is shown in figure [ fig : ribbon ] .
the braiding process is topologically the same as a full twist of the region containing both charges ( i.e. a full twist of their fusion product ) , combined with full twists of the charges themselves in the opposite direction .
hence , given two anyonic charges @xmath15 and @xmath16 , the effect of the monodromy on the fusion channels that yield overall charge @xmath17 is to introduce a phase factor @xmath30 , or @xmath31 .
like the monodromy , many of the other properties of the topological phase may be obtained directly from the fusion rules and the spin factors .
examples are the quantum dimension @xmath32 of a particle labeled @xmath15 , the modular matrices @xmath33 and @xmath34 and the topological central charge @xmath17 .
we give formulas for all these quantities here .
the quantum dimension @xmath32 of the sector @xmath15 gives the asymptotic number of fusion channels available when many particles of type @xmath15 are fused together .
if there are @xmath35 such particles , then the total number of fusion channels scales asymptotically as @xmath36 for large @xmath35 .
we may find @xmath32 as the perron - frobenius eigenvalue of the fusion matrix @xmath37 whose elements are defined by @xmath38 .
the quantum dimensions are real and positive and they have the important property that they are conserved under fusion , that is @xmath39 the total quantum dimension @xmath40 is defined by @xmath41 the quantity @xmath42 can be considered the quantum dimension of the quantum group that underlies the system .
the topological entanglement entropy of the ground state of a system in a topological phase is proportional to the logarithm of @xmath40 . the topological central charge @xmath17 is defined by the following formula .
it is only determined up to a multiple of @xmath43 . @xmath44
any conformal field theory whose topological sectors have these quantum dimensions and spin factors must have conformal central charge @xmath45 equal to @xmath17 modulo @xmath43 .
it is in fact impossible to find out more about the conformal central charge of a cft from the tqft corresponding to the cft than its value up to multiples of @xmath43 . to see this
one may consider the @xmath46 wzw model at level @xmath21 .
this is a cft at @xmath47 which has only one primary field , or topological sector .
tensoring any cft with copies of this @xmath48 model allows one to change the central charge of the cft by multiples of @xmath43 without changing the corresponding tqft . on a surface of nontrivial topology
, a tqft will have a set of degenerate ground states . on the torus
there is one state for every topological sector ( for anyon models described by a _
modular _ tensor category ) .
the mapping class group of the surface acts projectively on this ground state space . in the case of the torus
this means we can work with ordinary representations of the double cover of the mapping class group , the famous modular group generated by two elements @xmath33 and @xmath34 subject to the relations @xmath49 here @xmath50 is an element of order @xmath51 .
on the basis of states labeled by the particle sectors @xmath50 is represented by the charge conjugation matrix , that is @xmath52 . in a cft that realizes this tqft ,
the modular @xmath33-matrix and @xmath34-matrix are given by@xcite @xmath53 the numbers @xmath54 can also be defined within tqft , as the trace of the monodromy acting on particles with labels @xmath55 .
this , together with the ribbon equation , leads to the formula above .
the formula for @xmath34 is interesting , because it depends on the value of the conformal central charge @xmath45 modulo @xmath56 .
we just noted that the value of the central charge @xmath17 in a tqft is only determined modulo @xmath43 , so it seems that there might be a problem defining the above modular group representation for tqfts .
however , it turns out that any choice of @xmath45 modulo @xmath56 , given that @xmath57 modulo @xmath43 , gives a good representation of the double cover of the modular group .
the different choices just change the action of @xmath34 by a third root of unity and this factors through the relations ( [ eq : modulargroup ] ) that @xmath33 and @xmath34 must satisfy .
still , it is remarkable that cfts are able to see topological information that tqfts are blind to , namely a factor of a third root in the action of the modular matrix @xmath34 .
of course one might include this third root into the definition of a tqft .
if this is done , then there will be three tqfts with only one sector , realized in cft for example by the trivial , @xmath48 level @xmath21 and @xmath58 theories . taking this third root into account in any definition of _
topological order _ would also necessarily mean that topologically ordered phases arise whose ground states on the torus are not degenerate , but do transform nontrivially under the action of @xmath34 .
such phases can be realized by @xmath48 chern
simons theories at levels @xmath21 and @xmath59 respectively .
fusion rules , spin factors and the quantities that can be expressed in terms of these are the only elements of topological field theory that will be essential for an understanding of the main thread of the rest of this paper . an important question one may ask is thus whether fusion rules and spin factors are information that can be accessed through experiment . experimental probes of topological systems ( notably quantum hall systems ) that have been proposed in recent years include measurements of the tunneling current at point contacts and especially interferometric measurements of tunneling currents through double point contacts @xcite . in the case of fractional quantum hall states with abelian topological order , experiments with double point contacts have in fact recently been performed@xcite
. the tunneling current at a single point contact in a hall system is dominated at weak tunneling by the quasiparticle with the lowest scaling dimension . by observing the behavior of the tunneling current as a function of temperature , voltage or the size of the system@xcite
, it should be possible , in principle , to extract this scaling dimension and from that the topological spin of the particle .
extracting the spins of the other particles will probably be considerably more difficult .
interference experiments with double point contacts are governed by the _ _ monodromy matrix__@xcite .
this is just a normalized version @xmath60 of the @xmath33-matrix , given by @xmath61 it seems likely that at least some elements of the monodromy matrix @xmath60 will become available through interferometric measurements in any topological system that allows these . on the other hand
, it should be noted that for example in the hall systems , it will be difficult to get elements that do not involve the quasiparticle of lowest scaling dimension , which naturally dominates the tunneling .
if enough of the matrix elements of @xmath60 are known , it might be possible to reconstruct the whole @xmath60 matrix and its close relative , the s - matrix .
the @xmath60 and @xmath33-matrices are strongly constrained by various tqft - identities , so a relatively small number of measured elements may be enough to determine them completely . from the @xmath33-matrix
one may reconstruct the fusion rules using verlinde s formula , @xmath62 the @xmath33-matrix also gives information on the spins of the particles .
for example , given the @xmath33-matrix and the fusion rules , the second line of equation ( [ modularmatrices ] ) becomes a system of equations for the spin factors @xmath63 .
another set of equations for the spin factors comes from the fact that @xmath33 and @xmath34 satisfy the modular group relation @xmath64 .
these equations do not always allow for a full determination of the @xmath63 : two theories which have the same @xmath33-matrix , but different spin factors are the ising model and the @xmath65 chern simons theory at level @xmath51 .
these theories are distinguishable by their central charges , which are @xmath66 and @xmath67 , respectively .
however , there also exist examples of distinct theories with the same @xmath33-matrix as well as equal central charge , but with different spin factors , for example the two @xmath68 theories based on the quantum doubles of the finite groups @xmath69 ( the symmetry group of a square ) and @xmath70 ( the group of unit quaternions ) .
nevertheless , it is clear from ocneanu rigidity ( see section [ sec : reconstructionphantasies ] ) that for given fusion rules ( and hence for any given @xmath33-matrix ) , there can only be finitely many possible solutions for the spinfactors .
the mathematics of topological phases obviously involves more than just fusion rules and spin factors .
the requirements on fusion we have given may be fleshed out with more mathematical structure to give the definition of a tensor category .
similarly , including braiding and spin , we may get to the definition of a ribbon tensor category .
the hilbert spaces and transition amplitudes of the topological systems we are interested in may then be viewed as coming from representations of such categories .
the categories themselves in turn may be viewed as the representation categories of ( appropriately generalized ) quantum groups . while such structures will certainly be of importance for a more mathematically rigorous treatment of transitions between topological phases , that is not the aim of this paper and we intend instead to go into these details in a separate publication@xcite .
still , one may ask at this point wether knowledge of the fusion rules and spin factors would allow one to reconstruct the full tqft describing the system . an important step towards answering
this question is a mathematical theorem which is usually referred to as ocneanu rigidity @xcite .
this theorem states that , given a set of fusion rules , there can only be a finite set of inequivalent tqfts , ribbon tensor categories , or just plain tensor categories corresponding to it .
since we are given not only the fusion rules but also the spin factors , it seems likely that a tqft will be uniquely determined by this information in most cases .
in fact , we are not aware of any example of a pair of inequivalent tqfts with the same fusion rules and spin factors and even if such pairs do exist in nature , it will be difficult to separate them by experimental observation , for instance because they have the same @xmath60-matrix .
in 3 + 1 or more dimensions , bosons can be characterized either as particles with integer spin or as particles which have trivial exchange interactions , that is , wave functions for many identical bosons are invariant under exchanges of the bosons .
these two properties are equivalent by the spin statistics theorem @xcite . in 2 + 1 dimensions , the requirements of integer spin and trivial statistics are no longer equivalent . there may be particles which have the property that when two of them are fused together , multiple fusion products may arise and the braiding of the original particles is trivial or nontrivial depending on the fusion channel they are in .
therefore it is not completely obvious what constitutes a boson in dimension 2 + 1 .
two necessary conditions for a particle @xmath15 to be a boson are the following .
* @xmath15 should have trivial spin , that is @xmath71 , or @xmath72 .
* @xmath15 should have partially ( or completely ) trivial self - monodromy . by partially trivial self - monodromy , we mean that there should be _ at least _
one fusion channel in @xmath73 for which the monodromy factor equals @xmath21 .
in other words , if @xmath74 , then there should be at least one charge @xmath17 in the sum such that @xmath75 . since we already required that @xmath71 , this comes down to the the requirement that there is a fusion channel @xmath17 with trivial twist , @xmath76 .
both of these conditions are special cases of a more general condition , * for every number @xmath35 of identical particles of type @xmath15 , there should be at least one state in the hilbert space for @xmath77 which is completely invariant under monodromy .
this condition is a reasonable requirement for particles which should be able to condense , because for any particle number , it provides at least one state which will not notice any ` stirring ' .
this is analogous to the requirement that any ` order parameter ' for the condensate should be single valued .
the general condition is much more difficult to check than the two special cases mentioned earlier .
a thorough treatment would also require that we introduce much more of the formalism of topological field theory .
however , we can make two useful remarks .
first of all , for any particle @xmath15 with quantum dimension @xmath78 , one may show that the requirement of trivial spin actually implies the general condition above ( and hence it also implies trivial self - monodromy ) .
secondly , in a number of cases with @xmath79 , we have been able to show explicitly that there are in fact states with trivial monodromy for any number of particles of type @xmath15 . for the rest of this paper
, we will ignore the general condition and work with condensates of particles with trivial spin and partially or fully trivial self - monodromy .
these conditions seem to be sufficient for the condensate transitions we have studied .
finally let us note that we have asked only for trivial monodromy and not for completely trivial braiding .
this means that in principle , the exchanges in multi - particle states may be represented non - trivially ( with eigenvalues @xmath80 ) , so that the particles would behave in some ways like fermions , though spinless ones .
we could of course exclude this behavior , but this restriction is unnecessary for our purposes and so we will allow for the more general situation .
it turns out that bosons , that is , particles with trivial spin and partially trivial self - monodromy , are a very common occurrence in topological field theories and particularly in the theories that arise from proposed models for topological quantum computation .
let us give a few examples .
the quantum hall effect is currently the only physical system where theory predicts the existence of anyonic excitations in a parameter regime that is accessible in the laboratory .
recently , direct observation of anyonic statistics has been claimed @xcite and much indirect experimental evidence is also available .
anyons with non - abelian braiding are also expected to exist at filling fraction @xmath81 and predicted at a number of other filling fractions , including @xmath82 .
these non - abelian hall states are currently the most experimentally advanced avenue toward topological quantum computation .
hall states are described by chern - simons theory in the bulk of the sample and by a corresponding chiral conformal field theory on the edge . in the case of non - abelian theories ,
the cft on the edge is often better understood than the bulk theory .
the moore - read ( mr ) state@xcite for @xmath81 and its generalizations , the read - rezayi ( rr ) series of states @xcite , which includes a candidate wave function for @xmath82 , are described by an @xmath83 parafermionic cft coupled to a @xmath84 theory describing an electrically charged chiral boson .
electrically neutral excitations of these models may be described more simply using the even spin subsectors of an @xmath83 wess - zumino - witten model .
the @xmath65 wzw - model at level @xmath85 has @xmath86 primary fields which we will label @xmath87 by their @xmath65 weights .
the field labeled @xmath88 corresponds to a topological sector with spin factor given by @xmath89 , with @xmath90 .
in particular , the field labeled by @xmath91 has spin factor @xmath92 , which equals @xmath21 whenever @xmath93 divides @xmath85 .
it also has quantum dimension equal to @xmath21 , so trivial spin is enough to guarantee that this field is a boson .
we may also see directly that this field has trivial self - monodromy , since it fuses to the vacuum sector with itself .
thus the rr - states based on the @xmath65 wzw - theories at levels which are a multiple of @xmath93 all have bosons . if we want to consider excitations that have nonzero electric charge we have to introduce labels corresponding to the @xmath84 part of the cft describing the rr - states , in addition to the @xmath65 labels . taking the full spectrum into account
, we then find even more bosons .
however , the condensation of such charged bosons should lead to superconductivity as well as a change in the filling fraction and so it may be physically more difficult to imagine .
it is perhaps interesting to note that the @xmath94 rr - state , which could describe the hall plateau at @xmath82 , is one of the few low - lying states in the read - rezayi series that does not admit any bosons .
we will present an inventory of bosons in the proposed non - abelian hall states and the expected condensed phases that result from them in a separate publication@xcite .
most of the known local models with anyons are in fact non - chiral ; they have vanishing central charge .
important examples of this class of model are kitaev s toric code models for discrete groups@xcite and levin and wen s string net condensates@xcite ( the loop and dimer models of refs . can also be viewed as a special case of these@xcite ) .
the toric code models exhibit the same topological order as the discrete gauge theories@xcite , described by a quantum group called the quantum double @xmath95 of the finite gauge group @xmath96 .
we have treated quantum group symmetry breaking in these models in our earlier works@xcite , so we will discuss them only briefly here . in the model with gauge group @xmath96 , topological sectors are labeled by a conjugacy class @xmath97 and by an irreducible representation @xmath98 of the centralizer group @xmath99 of an element @xmath100 of @xmath101 ( the choice of @xmath100 does not matter ) .
the spin factor of the sector labeled @xmath102 is @xmath103 , where @xmath104 is the dimension of the irrep @xmath98 .
depending on the group @xmath96 there may be many bosonic sectors , but in general there are two classes of bosons that are always present : the electric sectors which have @xmath105 ( where @xmath106 is the unit element ) and the magnetic sectors which have @xmath107 , the trivial representation of @xmath108 .
it is clear that both electric and magnetic sectors have trivial spin . under fusion
, the electric sectors only produce new electric sectors , so they also have trivial monodromy ( in fact not just for two particle states , but for arbitrary numbers of particles ) .
the purely magnetic sectors may fuse to give sectors which have nontrivial electric charges ( that is , nontrivial centraliser labels ) , called cheshire charges. the sectors with cheshire charge will usually have nontrivial spin and so one may wonder if the magnetic particles are always true bosons .
however , it is not difficult to show that the fusion of identical magnetic particles always contains at least one channel without cheshire charge , so that the requirement of partially trivial self - monodromy is satisfied .
in fact , one may go further and show that the topological hilbert space for arbitrarily many identical magnetic particles always contains at least one state which has completely trivial monodromy .
this state is basically the gauge invariant magnetic condensate state proposed earlier@xcite , but to make contact with the present formalism one must project this onto a subspace of the hilbert space with fixed total topological charge ( for example the space of topological singlets ) .
string net condensates are also described by quantum doubles , but more typically by quantum doubles of quantum groups like @xmath109 and its generalizations .
these models are directly related to doubled chern - simons or wzw - models , with gauge groups @xmath110 , where @xmath111 is now a lie group and the left and right hand copies of @xmath111 occur with the opposite levels , i.e. these are @xmath112 theories .
the topological sectors of such models are labeled by pairs of representations of @xmath111 that are admissible at level @xmath85 .
the spin of the sector labeled @xmath113 is given by @xmath114 and we see that all diagonal fields @xmath113 with @xmath115 have trivial spin .
the fusion of two diagonal fields always yields at least one diagonal field , so the requirement of partially trivial self - monodromy is also satisfied .
however , as with the case of the toric code , one may show that there are in fact states with totally trivial self - monodromy for any number of identical diagonal fields .
the reason is that the monodromy of any state with @xmath116 diagonal fields @xmath117 may be described using the tensor product of the braid group representation associated with @xmath116 copies of @xmath118 and its dual braid group representation .
this tensor product contains a canonical singlet representation of the braid group and the corresponding state has totally trivial monodromy .
suppose we can change the parameters of the microscopic hamiltonian underlying our anyonic system in such a way that the particle with charge @xmath15 condenses ( @xmath15 should be a boson in order for this to happen ) .
then we can ask what the overall effect of this condensation on the topological excitation spectrum of the system will be .
the basic idea of this paper is that the condensate breaks down a symmetry underlying the spectrum .
that is , before condensation the charge sectors correspond to irreducible representations of some quantum group @xmath5 , the fusion rules are described by decomposition of the tensor products of these irreps etcetera
. then condensation breaks @xmath5 down to a subalgebra @xmath8 and afterward the excitations carry irreps of @xmath8 . when a quantum group is broken down to a subalgebra
, two things generically happen to its irreducible representations .
first of all , some irreps of the original quantum group @xmath5 will not be irreducible as representations of the subalgebra @xmath8 .
these representations will split into multiple irreps of @xmath8 .
secondly , some representations which are inequivalent under the full @xmath5-action will have equivalent @xmath8-actions and hence become identified .
more concretely , all this can be described by _ branching rules _ of the form @xmath119 where @xmath15 is an irreducible representation of the original quantum group @xmath5 , the @xmath120 s are irreps of the algebra @xmath8 that is left after symmetry breaking and the @xmath121 are multiplicities .
we will now make an important conceptual step and put these representation labels and branching rules center stage , forgetting for the moment about quantum groups and their subalgebras .
so assuming that we have a set of labels @xmath122 which characterise the charge sectors of the unbroken theory , as well as fusion rules and spin factors for these labels , then we will say that symmetry breaking means that to each of these labels we associate a branching rule @xmath123 we will call the right hand side of this equation the _ restriction _ of @xmath15 ( we are still thinking of it as the restriction of a representation @xmath15 of @xmath5 to @xmath8 ) .
the @xmath124 are again multiplicities and we have introduced a new notation where instead of labeling the components of the restriction of @xmath15 directly by sectors of the broken theory ( which we think of as labels of irreps of @xmath8 ) , we simply label them @xmath125 , @xmath126 etc .
of course the new labels @xmath127 that occur will most likely not all correspond to distinct sectors of the new theory , for different choices of @xmath15 .
however , the notation introduced here is quite useful in the process of finding out exactly what the new set of sectors actually is .
we want the new labels to be the labels for the excitations of the broken phase , so we will make the physical assumption that they have their own set of fusion rules satisfying the requirements of section [ sec : tqftsec ] , namely , associativity , existence of a vacuum label and conjugate representations , and a unique way for each conjugate pair of labels to annihilate to the vacuum
. we will not require the new fusion rules to be symmetric , that is , we may allow that @xmath128 for some pairs @xmath129 .
we also do not require a well defined spin or monodromy of the @xmath127 at this point .
the reason that we do not impose these requirements is that we want the new set of labels to capture not only pointlike excitations of the condensate vacuum , but also topological excitations which pull strings or alternatively , excitations which are confined to a boundary between the broken and unbroken phase . the string pulling excitations are expected to be the same as the excitations which occur only on the boundary , since a confined boundary excitation may be visualized as a thread or string extending from the boundary into the broken phase , ending at a string - pulling excitation of that phase . apart from the requirement that the new theory has sectors with associative fusion and unique duals , there are two more important assumptions that go into the determination of the new set of labels and and their fusion rules .
first of all , the sector that contains the condensed excitation should be indistinguishable from the vacuum sector in the condensed phase .
hence we require that the restriction of the condensed sector @xmath17 of the original theory contains the vacuum label @xmath21 of the new theory .
in other words @xmath130 secondly , we require that the fusion of the old and new labels is compatible with the branching , that is , restriction and fusion commute and we have @xmath131 and @xmath132 the equations above , together with the uniqueness of the unit of the new theory also imply that @xmath133 here , we introduce a slight abuse of notation that we will utilize throughout , namely , if a sector branches to a unique new sector , we will denote the old and new sectors by the same label , as long as the meaning is clear from the context .
the compatibility of fusion and restriction has another important consequence : it implies that the quantum dimensions are preserved under branching , that is , for every label @xmath15 of the unbroken phase , we have @xmath134 we will now give a simple example of how one can determine the set of labels for the broken phase and their fusion rules directly , given the assumptions above . consider the representation theory of @xmath136 at @xmath137 .
this is the quantum group for the @xmath10 wzw model of conformal field theory and for the @xmath65 chern - simons theory at level @xmath93 .
there are five different topological sectors in this theory which are simply denoted by dynkin labels @xmath138 , with @xmath20 denoting the vacuum . the quantum dimensions , spins and fusion rules for these sectors are given in table [ tab : su24 ] and from this table , we may read off that the sector labeled @xmath93 is bosonic ( we have @xmath139 , @xmath140 and @xmath141 ) .
@xmath142 if an excitation in the @xmath93-sector condenses , then @xmath93 will have to branch to the new vacuum and possibly other new labels .
however , since @xmath143 and the quantum dimension of the new vacuum is also necessarily equal to @xmath21 and quantum dimensions are preserved under branching , we find that @xmath144 from here , we may conclude immediately that the restrictions of @xmath9 and @xmath21 must equal each other , since @xmath145 and @xmath146
. also , the restriction of @xmath21 ( or @xmath9 ) can only have one part , because each part would contribute at least a numerical value of @xmath21 to the quantum dimension of the label @xmath21 and the value of this quantum dimension is less than @xmath51 .
now let us look at the fusion of the restriction of @xmath51 with itself .
we have @xmath147 since the vacuum appears twice on the right hand side , @xmath51 must branch into at least two parts ( if there was only one part , it would be able to annihilate with itself in two different ways ) .
since the quantum dimension of @xmath51 equals @xmath51 , this is possible , and in fact there must be exactly two parts @xmath148 and @xmath149 , each with quantum dimension @xmath21 .
note that neither @xmath148 nor @xmath149 can be identified with the vacuum sector @xmath21 , since this would imply the splitting of @xmath21 through the fusion rule @xmath150 and this is impossible , since @xmath151 .
looking back at the fusion @xmath152 we then conclude also that @xmath153 .
we have now completely identified the branching rules for this transition and we turn to the fusion rules .
these are straightforward for the new labels @xmath20 and @xmath21(using ( [ tensreq ] ) ) , but for @xmath148 and @xmath149 , we have two options , in principle . either these sectors are both self dual , giving @xmath154 , or they are dual to each other , giving @xmath155 . now rewriting equation ( [ eq:2x2 ] ) with our current knowledge
, we see that @xmath156 hence if we assume that @xmath148 and @xmath149 are self - dual , it follows that either @xmath157 and @xmath158 or @xmath159 and @xmath160 . in either case ,
one quickly checks that associativity of the fusion rules is violated , by evaluating @xmath161 and @xmath162 .
hence @xmath148 and @xmath149 must be dual to each other .
now we just have to decide whether @xmath163 equals @xmath148 or @xmath149 ( and similarly for @xmath164 ) . a similar associativity argument as before quickly
yields that we must have @xmath165 and @xmath166 .
hence we can straightforwardly obtain the full new set of sectors , as well their fusion rules .
we summarize these results in table [ tab : su24broken ] @xmath167 note that while the fusion rules of the broken theory turn out to be symmetric , we did not put this in by hand and it is in fact just a particular feature of this theory that is not reproduced in general .
not all of the excitations of the broken phase will be pointlike ; some will pull strings in the condensate .
these excitations will be confined , since a string is just a part of the medium where the original symmetry is restored and will cost an amount of energy proportional to its length . the excitations which do not pull a string will be the particlelike excitations of the new phase and they should have well - defined fusion and braiding interactions , in particular well defined monodromies and spin factors .
intuitively , an excitation should pull a string when it has nontrivial monodromy with the condensed excitation , since such nontrivial braiding would lead to a branch cut singularity in the condensate order parameter .
though not very rigorous , it is probably best to say that it is not possible to have a smooth single valued order parameter field enclosing a particle which has a nontrivial monodromy , and that therefore that particle has to pull a string upon entering such a phase .
one expects at least that the presence of the condensate does not interfere with the monodromy of the non - confined particles .
in particular , we expect to be able to assign spin factors to the non - confined sectors by lifting them into the unbroken phase .
the lifts of a sector @xmath16 of the broken theory are just all labels @xmath168 of the original theory that have @xmath16 in their restriction .
a necessary condition for a sector to not be confined is the following : * if a sector @xmath16 is not confined then all its lifts @xmath169 must have equal spin factors . in the other direction ,
sectors which do not satisfy this condition will be confined .
note that it is only natural that we should not be able assign spin factors to string - like excitations , since twisting such an excitation leads to a physically observable twist in the string connected to it and we should not expect that such a change can be absorbed by a change of the phase of the wave function .
there are a number of other physical criteria on the set of non - confined particles which we could impose separately , but which in practice turn out to be implied by the simple requirement above in all cases we have investigated .
first of all , the non - confined sectors must form a closed set under fusion , since pointlike excitations must fuse to pointlike excitations .
also , this set must contain the vacuum .
in particular , this means that all lifts of the vacuum must have trivial spin .
this is of course a criterion that is intimately related to the nature of the condensate ; we can impose it already at the symmetry breaking stage , or even view it as part of the definition of a bosonic condensate .
finally , we can go so far as to require that there is a unitary braided tensor category describing the fusion and spins of the set of unconfined excitations .
proving such a thing is beyond the scope of this paper , but again it does turn out to be true in all our examples . also , we would like to stress once more that it is likely that if such a braided tensor category exists , it will actually be fixed uniquely by the fusion and spins of the unconfined sectors . from our assignment of spin factors , we may derive the monodromy of the non - confined particles using the ribbon equation ( assuming that the set of non - confined particles closes under fusion ) .
the resulting monodromy is just the same as the monodromy of the lifts of the particles .
more specifically , let @xmath15 , @xmath16 and @xmath17 be sectors of the broken theory which are not confined and let @xmath170 .
also , let @xmath171 , @xmath172 and @xmath173 be arbitrary lifts of @xmath15 , @xmath16 and @xmath17 with the property that @xmath174 .
then the monodromy of @xmath171 and @xmath172 in the fusion channel @xmath173 is given by the combination of spin factors @xmath175 , but since the spin factors of the lifts of @xmath15 , @xmath16 and @xmath17 are all equal , this factor does not actually depend on the choice of lifts @xmath171 , @xmath172 and @xmath173 and we may as well write @xmath176 , which is the monodromy of @xmath15 and @xmath16 in the fusion channel @xmath17 .
an important special case of this argument is the case @xmath177 . in this case
we are looking at the monodromy of the lifts of @xmath15 with the lifts of the vacuum , which are of course the condensed sectors .
the argument we just gave now says precisely that the lifts of the non - confined particle @xmath15 have trivial braiding with the condensed particles , so we have managed to give a more precise meaning to the intuition about confinement that we mentioned at the start of this section .
we could in fact turn the whole argument above around and start by requiring that all lifts of a non - confined sector should have trivial monodromy with lifts of the vacuum sector ( i.e. with condensed sectors ) . from that assumption
we can get back to the confinement criterion given above if two conditions are satisfied .
first of all , all lifts of the vacuum of the broken theory should have trivial spin ( we also required this before ) and secondly , it must be possible to obtain all lifts of a sector of the broken theory by fusion with lifts of the vacuum . as we remarked , the first requirement is intimately connected with the nature of the condensate and with the question what exactly constitutes a boson in @xmath178 dimensions . if the second requirement does not hold , then it would seem that we have identified sectors which should be distinguishable .
it is not clear to us at this point whether these two requirements follow from the conditions on the condensate and on the symmetry breaking scheme that we had imposed already , though they do in all our examples . in any case , assuming that these two requirements do hold , we can regain our previous confinement criterion .
any lift of the fusion channel @xmath179 is of the form @xmath180 and the monodromy factor in this lift is @xmath181 . now using the two conditions above
, we see that @xmath182 for all @xmath183 and also , for any @xmath183 , the possible @xmath184 run through all lifts of @xmath16 .
thus for all these monodromy factors to be equal to @xmath21 is equivalent to @xmath185 for all @xmath186 and @xmath85 .
in other words , excitations in the @xmath16 factor are not confined precisely when all lifts of @xmath16 have the same spin factor and we are back at our original criterion . applying the confinement criterion to our @xmath10 example
, we see that the sectors labeled @xmath20 , @xmath148 and @xmath149 are not confined . for @xmath148 and @xmath149
this is immediate , since they have a unique lift , and @xmath20 lifts to either @xmath20 or @xmath93 , both of which have spin factor @xmath21 . the sector with label @xmath21
is confined , since @xmath21 lifts to @xmath21 and @xmath9 , which have different spin factors ( @xmath187 and @xmath188 respectively ) .
this gives the results in table [ tab : su24nonconf ] .
@xmath189 this result actually fixes the topological order of the non - confined sector of the broken theory uniquely , since there is only one solution to the consistency conditions for topological field theories ( notably the pentagon and hexagon equations@xcite ) which has these fusion rules and spin factors@xcite .
we have given a description of the spectrum of topological excitations in a theory which has undergone a condensation transition .
we have seen that the broken theory has excitations which are pointlike as well as confined excitations which pull strings .
these confined excitations can exist as boundary excitations , when their strings are attached to a phase boundary , or as ` hadronic ' composites , when two or more confined excitations form a cluster whose overall topological charge is not confined . in such clusters , the confined particles are joined together by their strings .
it is interesting to try and characterize the different types of string themselves in some non - redundant way .
a redundant labeling is given by the set of labels of confined particles .
many confined particles will likely pull the same type of string in the condensate vacuum , since each confined particle can be fused with any non - confined particle to give some other confined particle , and this should not change the type of string that occurs .
therefore , we propose to label the different types of string by equivalence classes of confined sectors modulo fusion with excitations from non - confined sectors . to be more precise , let us introduce an equivalence relation on the sectors of the broken theory as follows ,
@xmath190 we clearly have @xmath191 , just take @xmath17 and @xmath192 trivial .
also @xmath193 , since the definition of the relation is symmetric . finally if @xmath194 and @xmath195 then @xmath196 . to see this , note that from @xmath194 , we have unconfined sectors @xmath197 with @xmath198 and @xmath199 .
similarly , from @xmath195 , we have unconfined sectors @xmath200 with @xmath201 and @xmath202 .
hence , we have @xmath203 and @xmath204 and since @xmath205 , @xmath206 , @xmath207 and @xmath208 are not confined , neither are @xmath209 or @xmath210 , so @xmath196 and we have a good equivalence relation .
the different types of string should be uniquely labeled by the equivalence classes , which are some sort of ` orbits ' under fusion with non - confined excitations . as a check
, we note that all non - confined representations are equivalent to each other , which is what we want , since they all correspond to the situation with no string . to see this note that if @xmath15 and @xmath16 are not confined , then neither are @xmath211 and @xmath212 . but @xmath213 and @xmath214 , so indeed @xmath194 . for our @xmath10 example
the classification of strings is rather trivial , since there is only one type of confined particle .
hence there are just two classes under the equivalence above , the class consisting of the confined particle @xmath21 , which pulls a string and the class consisting of the unconfined particles @xmath20 , @xmath148 and @xmath149 , which pull no string . in the previous sections we have given general principles for the treatment of the phenomenon of breaking a quantum symmetry @xmath215 through the formation of a boson condensate , as well as a detailed example .
the analysis proceeds in three stages : + 1 . _
criteria for a condensate_. we formulated some criteria that have to be satisfied for a field @xmath17 to be a boson , and to serve as a possible candidate to form a condensate .
two necessary conditions are that it should have trivial spin factor ( @xmath76 ) and partially trivial self monodromy , i.e. there is at least one fusion channel @xmath216 with @xmath217 .
_ consistent branching_. our analysis is based on the construction of a set of branching rules giving the decomposition of the topological sectors of the unbroken theory into sectors of the broken theory .
we can think of this as branching representations of the quantum group @xmath215 describing the unbroken phase into representations of an intermediate algebra @xmath218 ( whose structure is not discussed a priori ) .
there are a number of consistency conditions on these branching rules that have to be met and these in fact appear to determine the possible branchings uniquely .
a crucial condition is that branching commutes with fusion .
this implies in particular that the total quantum dimension is preserved under the branching rule and that the old vacuum branches into the new .
we furthermore require that the condensate has the new vacuum in its branching .
confinement_. we observe that we can determine which representations in the broken phase have a nontrivial braiding with the condensate , and it is clear that these will pull a string in the new vacuum and hence are confined .
the effective topological low energy theory is then described by the fusion and braiding rules of the non - confined representations ( these must form a closed fusion ring ) and these can then presumably be identified as the irreducible representations of some quantum group @xmath219
. to our knowledge applying these conditions and performing these steps determines the breaking pattern uniquely . in previous papers on this subject
we restricted our attention mostly to theories described by finite dimensional quasitriangular hopf algebras , especially the so - called discrete gauge theories , whose hidden symmetry corresponds to the quantum double of the discrete gauge group . in those cases
the analysis of the breaking phenomena was done by explicitly considering the algebraic structure of @xmath218 and @xmath219 .
however following that route directly in case one is dealing with representations that carry non - integer quantum dimensions is problematic , and that is why in this paper our analysis is based on the dual route , directly studying the breaking pattern on the level of topological sectors ( representations ) and their branching rules .
indeed , in the explicit example we treated in the previous section of this paper we showed that the present approach , using the branching rules directly , can also be applied to topological systems which have sectors carrying non integer quantum dimensions , for example the systems described by the quantum groups based on quantum deformations of semisimple lie algebras which show up in relation to conformal field theories of the wess - zumino - witten ( wzw ) type .
this is an important extension of the possible applications of the breaking mechanism which will allow applications in physical contexts like the fractional quantum hall effect .
let us uncover some features of the condensate transition that are mostly independent of the details of the topological phase we start with .
one can get surprisingly far with this if the condensed sector is a _
simple current_. a simple current in cft is a primary field @xmath220 whose fusion rules are such that the fusion of @xmath220 with any other field contains only one channel , i.e. for any primary field @xmath221 , the fusion rule @xmath222 has only one primary field on the right hand side .
we will use the analogous definition in the context of tqft .
it is easy to see that a topological sector labeled @xmath220 is a simple current precisely if @xmath223 .
first of all , since there are only finitely many sectors , we must have @xmath224 for some @xmath225 .
we call this integer @xmath225 the order of @xmath220 and denote it @xmath226 . now using formula ( [ dimtens ] ) repeatedly and noting that @xmath227 ( which also follows from formula ( [ dimtens ] ) ) , we see that @xmath228 .
but since @xmath229 is real and positive , it follows that @xmath223 . for the converse ,
let us assume that @xmath223 .
then it is immediate that all fusion powers of @xmath220 are dimension @xmath21 sectors and there will be some @xmath220 for which @xmath224 .
now if there would be some sector @xmath16 for which the fusion @xmath230 has multiple channels , then this would imply that @xmath231 also has multiple channels , or multiplicities greater than @xmath21 .
however since @xmath232 , this is a contradiction and hence @xmath220 is a simple current .
if a bosonic simple current @xmath220 condenses , we can immediately see that the restrictions of a number of fields of the original theory will be identified .
first of all , the fusion powers of @xmath220 all branch to the vacuum .
more generally , for any sector @xmath15 , there is an orbit of @xmath15 under the action of fusion with powers of @xmath220 and the restrictions of the fields @xmath233 in this orbit are all identified .
if the orbits are all of the maximal size , @xmath226 , then these identifications lead directly to a new fusion theory , without any further identifications or splittings .
the @xmath220-orbits of the old theory correspond to the excitations of the condensed theory .
the lifts of the new vacuum sector are precisely the sectors @xmath234 of the old theory and using that @xmath235 , we find that @xmath236 for all @xmath237 , so all lifts of the vacuum have trivial spin . more generally , the non - confined excitations of the new medium are precisely those @xmath220-orbits for which the spin factor is the same for all particles in the orbit
. if there are @xmath220-orbits of less than maximal size , the sectors in these orbits will split . to see this ,
let @xmath15 be sector in a non - maximal @xmath220-orbit and let @xmath225 be the smallest integer for which @xmath238 ( note that @xmath225 must divide @xmath226 ) .
then we have @xmath239 and so @xmath240 .
an analogous argument shows that we must have @xmath241 and so the restriction of @xmath242 contains at least @xmath243 copies of the new vacuum sector , which implies that @xmath15 ( and @xmath23 ) must split .
if there are no multiplicities @xmath244 greater than @xmath21 in the restriction of @xmath15 , then it must in fact split into at least @xmath243 parts , but there may be extreme cases where @xmath15 restricts to @xmath245 copies of the same sector of the new theory . to obtain the fusion rules for the parts of the split sectors , new input about the theory is necessary ; we have examples where two parts obtained in this way are dual to each other ( like @xmath148 and @xmath149 in the broken @xmath10 theory ) as well as examples where they are self dual ( see for instance the discussion of @xmath246 in section [ su28sec ] ) .
when the condensed sector is not a simple current it becomes much more difficult to say anything general about the symmetry broken and confined theories . in this case
the sector @xmath17 that condenses branches to a number of copies of the vacuum and possibly to other sectors , i.e. @xmath247 , where we have chosen @xmath248 . here @xmath249 and some of the @xmath250 with @xmath251 may be greater than zero .
if @xmath252 is not an integer , then there have to be such non - vacuum components of the restriction of @xmath17 , in order to preserve the quantum dimension .
in fact , some of the restrictions of @xmath17 may be confined .
one may heuristically interpret this partial condensation by thinking of particles in the topological sector labeled by @xmath17 as having a hidden internal hilbert space of dimension @xmath253 and condensing in a particular state in this internal space .
the condensed state ( and possibly some other states in the internal space ) will then branch to the vacuum , but other internal states will not and may even be confined . in our earlier work , where we restricted ourselves to a class of theories with integer quantum dimensions ,
this interpretation could be made completely rigorous .
however , in the present context , this seems more difficult .
it is likely better to think of the number @xmath254 as a measure for the number of states in a system of @xmath35 identical particles of type @xmath17 which would be indistinguishable from the vacuum in the condensed phase ( one would expect this number to grow as @xmath255 ) . with the condensate @xmath17
not a simple current , there will still be identifications , but they are more difficult to obtain . in general , the fusion rules @xmath256 just tell us that the components of the restriction of @xmath15 are identified with some of the components of the restrictions of the sectors @xmath16 appearing on the right . similarly ,
if a fusion @xmath257 contains the condensed sector @xmath17 , then this tells us that some component of the restriction of @xmath15 must be identified with a component of the restriction of @xmath258 . to get more information
, we need to use the requirement that the broken theory is once again a good fusion theory .
it is possible to make some general predictions on splitting , though not as strong as the ones for simple currents .
if , for some sector @xmath15 , the fusion @xmath259 contains @xmath260 copies of @xmath17 , then the restriction of this sector must split in order to produce the at least @xmath261 copies of the new vacuum in the restriction of @xmath259 .
again , if the components of the restriction are all distinct then there must be at least @xmath261 of them , but in extreme cases , we may have just one component with multiplicity @xmath262 . note that if @xmath17 is a simple current , then @xmath263 , since if @xmath264 , then we would have @xmath265 , contradicting the axiom of fusion that says that sectors can only fuse the the vacuum in a unique way .
also , in the simple current case , one may see easily that the splittings deduced here are a special case of the splittings of sectors in non - minimal orbits discussed before .
similar arguments to the above show that for any pair of sectors @xmath15 , @xmath16 for which @xmath266 , at least one of @xmath15 and @xmath16 must have a restriction which splits .
another case in which splitting of a sectors @xmath15 occurs for arbitrary condensates @xmath17 is if the fusion @xmath267 contains @xmath15 and no other sectors whose quantum dimension is greater than or equal to that of @xmath15 .
of course if @xmath17 is a simple current this just says that @xmath15 is a fixed point . in the general case
the restriction of the fusion @xmath268 must contain at least @xmath269 copies of the new vacuum sector , one from the @xmath17 in @xmath270 and @xmath271 from the copies of @xmath259 in @xmath272 .
now if @xmath15 ( and @xmath23 ) do not split , we see that there must be at least @xmath269 copies of the restriction of @xmath15 in the restriction of the fusion of @xmath267 . however , none of the components of the restrictions of other fields in @xmath267 can be identified with the restriction of @xmath15 , since the quantum dimensions of these other fields are smaller than @xmath32 by assumption .
hence , it follows that @xmath15 must split . from the examples we have calculated , we observe that the central charges and total quantum dimensions of anyon models seem to follow certain general rules under condensation , if the anyon model that one starts with is _
modular_. modularity is equivalent to the requirement that the monodromy is nondegenerate , that is , for every topological sector @xmath15 there is at least one topological sector @xmath16 such that @xmath15 and @xmath16 have nontrivial monodromy ( see ref . ,
section e.5 ) . another useful characterization of modularity is that the s - matrix of the theory must be unitary .
this requirement is satisfied for many models that have been studied in physics , for example for all models coming from conformal field theories with bosonic chiral algebras .
however , there are examples where modularity does not apply , notably in systems with excitations which behave like the vacuum under monodromy but which have nontrivial , necessarily fermionic , exchange behavior
. typical examples of such excitations are the actual electrons in quantum hall systems . given modularity , we observe that * the topological central charges of the unbroken theory and the unconfined theory are equal * denoting the total quantum dimensions of the original , broken and unconfined theories by @xmath273 , @xmath274 and @xmath275 , we have @xmath276
. we also note that generally ( independently of modularity ) we have @xmath277 . in the remainder of this paper
we will study connections between the quantum group symmetry breaking scheme we have described so far and constructions in conformal field theory .
we will argue that quantum group symmetry breaking in cft is dual to conformal extension of the chiral algebra .
this should also clarify the observation that central charge is conserved .
it appears more difficult to get an intuition for the identity between total quantum dimensions from the cft side .
in an idealized system with topological order , topological quantum numbers can not be changed by the application of any local operator . in other words ,
topological observables are conserved quantities which commute with the full algebra of local observables .
of course in realistic systems , the situation is often more complicated than this .
first of all , the topological features are often emergent only at low energies and different topological sectors of the hilbert space may be mixed by high energy ( virtual ) excitations .
secondly , any real system has a finite size , which implies that a product of finitely many local operators can actually become a topologically nontrivial operator and relate states with different topological quantum numbers . in other words , there is no clean separation between local and topological observables .
however , in the setting of conformal field theory , such a separation does exist . here
, the role of the local algebra is played by the chiral algebra , which can be the virasoro algebra , or some more complicated algebra like a kac - moody algebra or w - algebra .
the hilbert space of the theory splits into sectors on which this chiral algebra acts , but which are not mixed with each other by this action .
these chiral sectors correspond to the topological sectors of the cft .
hence it is natural to expect that there should be a tqft , or modular tensor category , which describes the fusion and exchange interactions of states from the different sectors .
it is also natural to introduce operators for topological charges which commute with the full chiral algebra and one may in fact hope to find a quantum group whose representation category is precisely the modular tensor category that describes the system s topological interactions and which has an action on the hilbert space of the cft that commutes with the action of the chiral algebra .
operators for topological charges can then be the analogs of casimir operators for this quantum group .
while this picture of a chiral algebra and a quantum group with commuting actions seems to be part of the lore of cft , there does not appear to be a detailed mathematical understanding of this picture for general cfts .
we will nevertheless attempt to flesh it out for wess - zumino - witten models in the remainder of this section , and note in advance that the picture just sketched does provide useful intuition about the connection between quantum group symmetry breaking and some well known constructions in cft .
since the quantum group and the chiral algebra are morally each other s commutants , we expect that breaking down quantum group symmetry from a large quantum group to a smaller one should be accompanied by an extension of the chiral algebra on the local side of things .
in fact , we find that there is a beautiful connection between quantum group symmetry breaking and conformal extension of the chiral algebra . in conformal extensions , we start with a chiral algebra that has a representation which is bosonic , but topologically nontrivial
. then we enlarge the algebra by adding an intertwining operator between the vacuum representation and this nontrivial bosonic representation ( i.e. a creation operator for a topologically nontrivial particle ) . before the conformal extension
, the theory would have a topological ( or more precisely , chiral ) sector corresponding to this bosonic field , but afterwards , this sector has become part of the vacuum sector of the new chiral algebra . on the quantum group side of the story we can interpret this merger of a bosonic sector with the vacuum sector as condensation of the bosonic particle and
describe its effects using the formalism proposed here .
in retrospect , this intuitive argument explains the similarity of some of our constructions and criteria to those mentioned for conformal extensions in moore and seiberg s famous work on the classification of rational cfts@xcite .
since many of the common constructions of cfts , most notably the coset construction@xcite , can be described in terms of conformal extension of the chiral algebra , these constructions now obtain a physical interpretation as being due to condensation of bosonic quasiparticles .
in fact , this suggests that some cft constructions will have a direct physical realization in phase transitions which occur in systems described by cfts .
after this somewhat abstract discussion let us turn to wzw models .
in these models@xcite the physical states are organized into integrable representations of an extended conformal symmetry , a kac - moody algebra based on a finite dimensional lie algebra @xmath111 , at a certain level @xmath85 , which we will denoted as @xmath278 .
these representations correspond to chiral primary fields and there is a finite number of them .
this theory has a central charge equal to @xmath279 with @xmath280 is the dual coxeter number of g. the chiral primary fields are operators which create the lowest energy states of the different topological sectors of the theory from the vacuum and one may obtain the fusion rules and braiding of the topological sectors directly from the cft by calculating the correlators , or more precisely the conformal blocks , of these chiral primary fields .
there is a one to one correspondence between chiral primary fields in the wzw model and irreducible representations of the quantum group @xmath281 , where @xmath282 and in fact , it is known@xcite that the fusion and braid relation obtained in this way are exactly the fusion and braiding of these quantum group representations .
explicit representations of the quantum group in terms of operators acting on the hilbert space of the theory can also be obtained , within the coulomb gas formalism@xcite .
all of this goes a long way toward establishing the picture that we sketched earlier in this section , of a chiral algebra and a quantum group normalizing each other .
the relation between wzw theory and quantum groups is quite useful even as a calculational tool , because many properties of multi(quasi)particle states in conformal field theory , such as their braiding properties , can be determined by just using the properties of the quantum group ( see e.g. ref . ) . in the remainder of the paper
we will pursue the relation between the breaking mechanism and cft in some detail , exhibiting the connections between the two formalisms in explicit examples , most of them based on wzw theories .
we will find that the breaking of quantum symmetries is indeed closely related to conformal extensions and we will show how well known constructions like the coset construction@xcite , conformal embeddings@xcite and orbifolding@xcite acquire a direct physical relevance and interpretation in the present context .
conformal embeddings are embeddings of affine lie algebras @xmath283 with the property that the central charges are equal , @xmath284 as a result , the corresponding cosets @xmath285 have central charge equal to zero and are therefore trivial .
we will have more to say about this in section [ sec : cosets ] . here
we will focus on the embeddings themselves .
general ( non - conformal ) embeddings of affine lie algebras do not conserve the central charge , but for all embeddings , the levels @xmath85 and @xmath286 are related by the dynkin index @xmath237 of the corresponding embedding of @xmath96 into @xmath111 , one has @xmath287 .
the conformal embeddings of affine lie algebras have been classified in refs .
; they form a number of infinite series and a finite list of special cases . in these papers
it is proved that for a conformal embedding the level of @xmath288 is always unity : @xmath289 .
conformal embeddings also have the remarkable property that the ( infinite dimensional ) highest weight representations of @xmath288 branch to finitely many highest weight representations of @xmath290 @xcite .
let us return to the breaking of @xmath10 that we studied in section [ sec : qbreak ] and let us show how it is connected to the well known conformal embedding of @xmath291 in @xmath292 ; indeed conformal because both have central charge @xmath293 . in table [ tab : su3_1_and_so5_1 ]
we give the representations and fusion algebra of @xmath12 .
@xmath294{|l| } \hline su(3)_1 \\
\hline \hline \begin{array}{l|ll } 1&d_1=1&h_0=0\\ 3&d_3=1&h_3=\frac{1}{3}\\ \bar{3 } & d_{\bar{3 } } = 1&h_3=\frac{1}{3}\\ \end{array}\\ \hline \hline \begin{array}{llll } 3\times 3=\bar{3}~ & & & \\
3\times \bar{3}=1&\bar{3}\times \bar{3}=3~ & & \\
\end{array}\\ \hline \end{array } & \begin{array}[t]{|l| } \hline so(5)_1 \\
\hline \hline \begin{array}{l|ll } 1&d_1=1&h_0=0\\ 4&d_4=\sqrt{2}&h_4=\frac{5}{16}\\ 5 & d_{5 } = 1&h_5=\frac{1}{2}\\ \end{array}\\ \hline \hline \begin{array}{llll } 4\times 4= 1 + 5~ & & & \\ 4\times 5=4 & 5\times 5=1~ & & \\
\end{array}\\ \hline \end{array } \end{array}$ ] from the embedding we obtain the branching of the corresponding kac - moody representations : @xmath295 indeed , this finite branching is possible because the @xmath20 and @xmath93 representations of @xmath291 are degenerate , in the sense that their conformal weights differ by an integer .
so one way to understand the conformal embeddings is to say that because the @xmath290 representations are degenerate there is a larger symmetry realized in the spectrum , i.e. @xmath288 .
interestingly , the ( bosonic ) singlet module of @xmath288 decomposes into bosonic representations under @xmath290 , as we see in the first line of ( [ eq : branchingsu31 ] ) and if we now return to our analysis of section iii we see that it is exactly the nontrivial bosonic component in the branching ( i.e. the 4 of @xmath291 ) that is assumed to form the condensate .
on the other hand we make the remarkable observation that after breaking and subsequent confinement the residual symmetry @xmath219 , i.e. the representations and their fusion rules as given in table [ tab : su24nonconf ] are precisely those of @xmath292 !
the conclusion is that there is a unique correspondence between the conformal embedding @xmath296 and the breaking of the quantum group for @xmath297 , where it should be noted that on the side of the chiral algebras , the embedded algebra is smaller , while on the quantum group side the residual quantum group corresponding to @xmath298 is smaller than the one for the embedded alegebra @xmath290 .
all this is in good agreement with our intuition that the fusion algebra is somehow the normalizer of the chiral algebra in the operator product algebra of the cft .
in fact , we can think of the chiral algebra of @xmath12 as an extension of the chiral algebra of @xmath10 by the intertwining operator between the vacuum sector of the @xmath10 theory and the sector labeled by @xmath299 .
the breaking of quantum symmetries is thus related to enlarging the conformal symmetry , and the construction of new conformal models with larger chiral symmetries , starting with models related to kac
moody algebras , has in the present context acquired a very direct physical meaning and relevance , namely the formation of a bosonic condensate in the phase with the smaller chiral symmetry .
it is instructive to discuss another example of a conformal embedding , where applying the breaking formalism is less straightforward .
let us consider the conformal embedding @xmath300 both with @xmath301 . we have listed the quantum dimensions , spins and fusion of the @xmath302 theory in table [ tab : su3_1_and_so5_1 ] . the spins and quantum dimensions for @xmath303
are given in table [ tab : su2_10 ] .
@xmath304 the fusion rules for @xmath83 , and in particular for @xmath303 , are given by @xmath305 where the sum runs over those @xmath88 in the indicated range for which @xmath306 is even ( i.e. @xmath88 is incremented by @xmath51 ) .
let us now consider the breaking mechanism .
the @xmath307 representation is the only nontrivial bosonic representation , and it has a trivial self - braiding channel because the fusion product with itself contains the identity representation .
we see that it has a quantum dimension @xmath308 which tells us that we have to split the representation @xmath309 where we assume @xmath310 to have unit quantum dimension and to be the component that condenses ( indeed : @xmath311 ) while we consider the @xmath312 component with quantum dimension @xmath313 for the moment as an independent field in the broken phase .
starting with this splitting of the @xmath307 and using the fusion rules in a similar fashion as we did in section iii we see that also other representations have to split and furthermore other identifications have to be made .
the net result of this straightforward analysis is given in table [ tab : su210identifications ] .
@xmath314 it is easy to see that the new representations have the following quantum dimensions : @xmath315 and @xmath316 . at this intermediate ( broken ) level
we are left with five representations which have the fusion rules given in table [ tab : su210brokenfusions ] : @xmath317 these fusion rules together with the conformal weights of the parent representations in the unbroken phase now allow us to determine which representation will be confined in the broken phase . as we said before , representations will not be confined if all their lifts have equal spin factors , or equal spins up to integers .
for example if we want to know whether @xmath318 will be confined , we have to check whether @xmath319 for all combinations @xmath55 with @xmath15 and @xmath16 taken from the list of fields that restrict to @xmath318 according to table [ tab : su210identifications ] , i.e. @xmath320 . since these have conformal weights @xmath321 and @xmath322 respectively , we conclude that the @xmath318 representation will be confined .
for the @xmath323 which is identified with the @xmath324 we have parent conformal weights @xmath325 and
@xmath326 so that that representation will _ not _ be confined .
the upshot of this analysis is that only the @xmath20 , the @xmath323 and the @xmath327 survive after confinement , of course with the fusion rules given in table [ tab : su210brokenfusions ] .
we see that indeed our residual set of fields and their fusion and spin factors are isomorphic to the @xmath302 algebra under the map @xmath328 and @xmath329 .
clearly the fusion algebra is also isomorphic to the ising model or the @xmath330 model , but these have to be rejected because the conformal weights do not match .
if we furthermore look at the branching rules for the conformal embedding : @xmath331 we confirm that they are fully consistent with this correspondence .
note that in these rules we clearly have matching ( modulo @xmath332 ) of the conformal weights .
representations can only branch into representations with the same conformal weights up to integers and hence the spin factors of the representations are preserved under the branching .
the conclusion is that also in this more complicated situation we find that the quantum group @xmath219 , which appears after breaking @xmath333 by a bose condensate and subsequent confinement , is the expected quantum group @xmath334 appearing in the conformal embedding . using the quantum group symmetry breaking formalism
, we can now conjecture new conformal embeddings which are not contained in the classification of conformal embeddings of refs .
, for example because one or both of the theories involved in the embedding is not a wzw theory . we can start with an arbitrary tqft or cft which has a boson , condense the boson , find the theory describing the nonconfined excitations of the broken phase and then conjecture that the original theory can be conformally embedded in a cft with the same topological order as the unconfined broken theory .
as an example let us consider @xmath246 .
the quantum dimensions and spin factors for this theory are given in table [ su2_8_spintab ] .
the fusion rules for @xmath246 follow from formula ( [ su2_k_fusion ] ) .
@xmath335{|l| } \hline su(2)_{8}\\ \hline \hline \begin{array}{l|ll } 0&d_0=1&h_0=0\\ 1&d_1= \sqrt{\frac{5+\sqrt{5}}{2}}&h_1=\frac{3}{40}\\ 2&d_2= \frac{3+\sqrt{5}}{2}&h_2= \frac{1}{5}\\ 3&d_3= \sqrt{5 + 2\sqrt{5}}&h_3=\frac{3}{8}\\ 4&d_4= 1+\sqrt{5}&h_4= \frac{3}{5}\\ 5&d_5= \sqrt{5 + 2\sqrt{5 } } & h_5= \frac{7}{8}\\ 6&d_6= \frac{3+\sqrt{5}}{2}&h_6=\frac{6}{5 } \\
7&d_7= \sqrt{\frac{5+\sqrt{5}}{2}}&h_7= \frac{63}{40}\\ 8&d_8= 1&h_8= 2 \\
\end{array}\\ \hline \end{array } $ ] the @xmath336 representation is the only bosonic field that meets the requirements for a condensate .
analysis of the fusion rules after condensation of this field , using the methods of section [ generalmethodsec ] leads to identifications of the @xmath337 with the @xmath338 sectors for @xmath339 , while the @xmath93 has to split : @xmath340 , in two parts which have equal quantum dimension . without going through the details we summarize the branching and the fusion rules of the symmetry broken theory , in table [ tab : su(2)_8_broken ] .
@xmath341 the @xmath342 and @xmath9 representations will become confined , so that we are left with four fields : @xmath343 and @xmath344 .
we see that the fusion rules of these non - confined fields are just those of the direct product of two fibonacci theories .
the proper identification of the algebra @xmath219 is in fact the quantum group @xmath345 with the identifications @xmath346 , @xmath347 , @xmath348 , @xmath349 .
this quantum group has identical fusion rules , quantum dimensions and conformal weights as the ones given in the tables .
alternatively , one might use @xmath350 .
all of this strongly suggests that there should be a conformal field theory with the same topological order as @xmath350 which has the property that the @xmath246 theory can be conformally embedded into it . in the previous section
, we have shown that the quantum group breaking allows us to conjecture many new conformal embeddings .
now we will show how these conformal embeddings can be used to generate non - diagonal modular invariants for certain conformal field theories .
in fact for what follows , it will not be necessary to know the exact cft describing the theory into which the embedding takes place ( i.e. the symmetry broken theory ) : it is enough to know the corresponding modular group representation , which is precisely what we get from the construction in the previous section . in standard cases of conformal embeddings
the branchings of representations can be used to construct the non - diagonal invariants for @xmath290 using the standard diagonal modular invariant for @xmath288@xcite .
for example for the simple conformal embedding @xmath291 in @xmath292 discussed in section [ sec : confemb ] we had the branching rules ( [ eq : branchingsu31 ] ) , and substituting these branchings into the modular invariant partition function for @xmath292 : @xmath351 yields the exceptional @xmath291 invariant : @xmath352 this is the lowest member of the socalled @xmath101 series series of non diagonal @xmath353 invariants with @xmath354 in the classification of invariants by cappelli , itzykson and zuber@xcite : @xmath355 these invariants follow from the breaking scheme of the quantum group @xmath356 with a condensate in the highest i.e. @xmath357 representation .
this representation corresponds to a simple current under which the representations @xmath358 and @xmath359 get identified , while the @xmath360 representation has to split as @xmath361 , furthermore the odd representations with @xmath362 get confined .
this leaves us with a fusion algebra @xmath219 of some cft with the modular invariant partition function given above .
@xmath363{|c|}\hline { su}(3)_2 \otimes { su}(3)_2 \supseteq { su(2)}_{8 } \\ \hline \hline \begin{array}{ll } ( 1,1 ) \rightarrow 0 + 8 & h_{1,1}=0\\ ( 8,1 ) \rightarrow 4_1 & h_{8,1}=\frac{3}{5}\\ ( 1,8 ) \rightarrow 4_2 & h_{1,8}=\frac{3}{5}\\ ( 8,8 ) \rightarrow 2 + 6 & h_{8,8}=\frac{6}{5}\\ \end{array } \\ \hline \end{array } &
\begin{array}[t]{|c| } \hline { su(3)}_4/{\mathbb{z}}_3 \supseteq { su(2)}_{12 } \\
\hline \hline \begin{array}{ll } 1 \rightarrow 0 + 12 & h_0=0\\ 8 \rightarrow 2 + 10 & h_8=\frac{3}{7}\\ 10 \rightarrow 6_1 & h_{10}=\frac{6}{7}\\ \bar{10 } \rightarrow 6_2 & h_{\bar{10}}=\frac{6}{7}\\ 27 \rightarrow 4 + 8 & h_{27 } = \frac{8}{7}\\ \end{array}\\ \hline \end{array}\\ \end{array } $ ] let us give some details for the cases @xmath364 and @xmath9 .
the first case is the formation of a condensate in the @xmath336 representation in @xmath246 .
here we have the identifications @xmath365 and @xmath366 while the @xmath299 splits @xmath367 .
we have discussed this case already in detail in section [ sec : confemb ] on conformal embeddings , in particular tables [ su2_8_spintab ] and [ tab : su(2)_8_broken ] .
this leaves us after confinement of the odd representations with four fields described by a @xmath368 theory , or equivalently a theory with the same topological order as @xmath369 .
the branchings are given in the left part of table [ tab : confbranchings ] . finally the case @xmath370 .
one is left with a theory with five primary fields , which is easily identified as the @xmath371 ( for identification one may consult for example ref . ) .
this theory has the fields corresponding to the @xmath372 and @xmath373 dimensional representations with the branchings and conformal weights given in the right hand part of table [ tab : confbranchings ] .
we see that the breaking mechanism allows us to systematically construct many new conformal embeddings and thereby it will also generate a large number of non - diagonal modular invariants for non - chiral cfts .
the coset construction@xcite is a way to construct a new conformal field theory , starting from @xmath298 and @xmath290 wzw models based on lie groups @xmath111 and @xmath96 with @xmath374 .
given an embedding of @xmath96 into @xmath111 with dynkin index @xmath237 , this embedding will fix the relation between the levels @xmath286 and @xmath85 as @xmath287 .
this also implies that @xmath375 .
the canonical generators of the conformal algebra for the coset are just the differences of the conformal generators of the wzw theories ( which are sugawara bilinears in the currents of the chiral algebras ) .
equivalently , the energy momentum tensor of the coset is defined as the difference of the energy momentum tensors for the @xmath111 and @xmath96 theories , @xmath376 this gives the coset central charge as @xmath377 .
one physical interpretation of the coset models is that they correspond to gauging the @xmath96 subgroup of @xmath111 in the wzw model based on @xmath111@xcite .
coset cfts play an important role in for example the description of fractional quantum hall states with non - abelian anyonic excitations .
for example the moore - read state and the series of read
rezayi states involve the cosets : @xmath378 with central charges @xmath379 it is in general nontrivial to the determine the full chiral algebra and the set of primary fields of a coset theory and to determine their fusion and braiding properties .
one way to approach this problem is through the character theory of affine lie algebra representations ( see for instance ref . ) . the @xmath278 highest weight representations @xmath380 branch into @xmath290 representations @xmath381 .
both the @xmath382 and the @xmath381 are infinite dimensional and in most cases the branching of @xmath382 yields either infinitely copies of @xmath381 or no copies at all . however , the subspaces of the @xmath382 and the @xmath381 at any fixed eigenvalue of @xmath383 , the chiral hamiltonian , are finite dimensional .
the character of an affine lie algebra representation is just a generating function for the dimensions of the eigenspaces of @xmath383 in that representation .
hence there is an identity between the characters of the integrable @xmath278 representations and the integrable @xmath290 representations into which they decompose .
we have @xmath384 where @xmath385 and @xmath386 are the characters of the representations @xmath382 and @xmath381 and the @xmath387
are so called _ branching functions_. one approach to coset models is to consider the branching functions directly as characters of the representations of the coset theory . in other words , one does not explicitly construct the coset chiral algebra , but instead one says that there is a non - zero chiral primary field of the coset theory for any nonzero branching function @xmath387 .
the requirement that the branching function should be nonzero means that there will not be a coset primary field for any combination @xmath388 but only for those combinations allowed by the branching rules . on top of the branching rules
, there are so called _ field identifications _ which say that some of the coset primary fields may be labelled by various different combinations of weights @xmath388 , or in other words , some of the pairs @xmath388 are identified if they are used as labels for coset primaries .
basically the pairs @xmath389 and @xmath390 are identified when the corresponding branching functions are equal , but often it is much easier to find the identfications by arguments involving the modular transformations of the characters and the automorphisms of @xmath298 and @xmath290@xcite , rather than by explicit calculation of the branching functions .
an alternative way to find the branching rules and field identifications of coset theories is through the action of the _ identification group _ @xmath391@xcite . for a @xmath285 coset , this identification group is defined as the group of bosonic simple current primary fields in the tensor product theory @xmath392 . here
the bar indicates that we should use the conjugate representation of the usual mapping class group representation for the @xmath290 theory .
in particular , the conformal weight of a @xmath392 primary field labeled by @xmath393 is the difference @xmath394 of the @xmath298 and @xmath290 conformal weights and bosonic simple currents are those simple currents for which this difference is an integer .
the group product on @xmath391 is given by the fusion of the simple currents .
@xmath391 also acts on the labels of the branching functions by fusion .
if the orbits of branching functions under the @xmath391 action all have the same number of elements , then one may describe the field identifications and branching rules of the coset in a very simple way : all fields in a single @xmath391 orbit are identified and the branching rules allow precisely those combinations @xmath395 such that the corresponding primary field of the @xmath392 theory has trivial monodromy with the elements of @xmath391 .
there is obviously a strong similarity between this procedure for finding branching rules and field identifications in coset theories and the procedures we have described for quantum group symmetry breaking and confinement , particularly with the special case of our symmetry breaking scheme described at the beginning of section [ sec : simplecurrents ] , where the condensed fields are simple currents and the orbits under the action of these simple currents are all of maximal size . in such cases , the procedure for finding the spectrum
, fusion and modular properties of coset fields reduces precisely to the procedure we have described for the condensation of the bosonic fields in the group @xmath391 , in the tqft corresponding to the @xmath392 wzw theory .
field identifications appear at the symmetry breaking stage , as the @xmath392 related by fusion with the condensed fields from @xmath391 turn out to have the same restriction , whereas the coset branching rules are due to confinement ; only fields that have trivial monodromy with the fields in @xmath391 are not confined . as an illustration of this relation between breaking a quantum symmetry and the coset construction
, we discuss the simplest example of the series ( [ eq : suncosets ] ) , the case @xmath396 . in this case
, the coset is the chiral ising cft , which plays a fundamental role in the construction of the moore - read fractional quantum hall state , as well as in the hierarchy of non - abelian hall states based on it@xcite .
we have to consider a boson condensate in @xmath397 .
the properties of representations of the factors of this product are given below .
@xmath398{|c| } \hline su(2)_1 \\
\hline \hline \begin{array}{l|ll } 0&d_0=1&h_0=0\\ 1&d_1=1&h_1=\frac{1}{4}\\ \end{array}\\ \hline \hline \begin{array}{l } 1\times 1=0 \\
\end{array}\\ \hline \end{array}\;\;\;\;\ ; & \;\;\;\;\ ; \begin{array}[t]{|l| } \hline su(2)_2 \\
\hline \hline \begin{array}{l|ll } 0&d_0=1&h_0=0\\ 1&d_1=\sqrt{2}&h_1=\frac{3}{16}\\ 2&d_2=1&h_2=\frac{1}{2}\\ \end{array}\\ \hline
\hline \begin{array}{ll } 1\times 1=0 + 2~ & \\
1\times 2=1&2\times 2=0~ \\ \end{array } \\
\hline \end{array } \end{array}$ ] all together there are @xmath399 fields which we denote by @xmath400 .
there is one nontrivial boson @xmath401 which we assume to condense .
it is a simple current because @xmath402 .
we now have to identify the fields which form orbits under fusion with the condensed field : @xmath403 where we have used an obvious notation to save space . at this point we are left with @xmath307 fields which we will label as the ones on the left .
now we have to determine which of the remaining fields will be confined .
using the conformal weights given in the tables we see that for example that the identified fields ( 00;1 ) and ( 11;1 ) have conformal dimensions @xmath404 and @xmath405 respectively which differ by 1/2 .
this as we explained before , means that this field has to be confined .
similarly one finds that @xmath406 and @xmath407 are confined .
we are then left with three non - confined fields and as expected these correspond exactly to the coset model , which is the ising model , as indicated in the following table .
@xmath408 at this point it is natural to ask how the correspondence between the coset construction and quantum group symmetry breaking fits into the general picture of quantum group symmetry breaking as dual to conformal extension that we sketched before .
it would appear that there is something of a mismatch .
the quantum group symmetry breaking picture for the coset @xmath285 starts from the @xmath409 topological data and condenses the available bosonic simple currents ( i.e. the fields in @xmath391 ) .
naively , the dual chiral algebra extension should start from the chiral algebra for a @xmath409 wzw model and extend this by the currents in @xmath391 . however , this chiral algebra is not the chiral algebra of the coset . in the construction of the coset theory ,
the @xmath290 chiral algebra is embedded in the @xmath298 chiral algebra so there is a priori no tensor product of the two .
in fact , the interpretation of the coset theory as a gauged wzw model and the identification of the coset primary field with branching functions strongly suggest that the chiral algebra of the coset theory should be the commutant of the @xmath290 chiral algebra in ( some extension of ) the @xmath298 chiral algebra .
nevertheless the description of the coset based on the identification group strongly suggests that , while the coset theory and the @xmath409 conformally extended by @xmath391 may be different as conformal field theories , they nevertheless have identical topological data . as a result , one should be able to describe the topological phase in @xmath178 dimensions whose @xmath410-dimensional boundary is described by the @xmath285 coset model using the topological data obtained from breaking @xmath409 by condensation of the bosons constituting @xmath391 .
this claim is also supported by the work of moore and seiberg@xcite .
they study the chern - simons theory based on the gauge group @xmath411 , where @xmath412 is the common center of @xmath111 and @xmath96 , with chern simons terms at level @xmath85 for @xmath111 and at level @xmath413 for @xmath96 , and they show that this theory has precisely the gauged wzw description of the @xmath285 coset as its boundary theory .
so far we have only discussed the very simplest cosets , which have the property that the identification group orbits are all the same size . however in general , there will be orbits of different sizes . in this case one
speaks of field identification fixed points , since some of the elements of @xmath391 will now fix some of the pairs @xmath395 labeling the coset primaries .
it turns out that in such cases it becomes necessary to introduce extra coset primary fields , and to view the branching functions corresponding to the identification fixed points as linear combinations of the characters for these fields .
this is analogous to the situation we describe in the latter part of section [ sec : simplecurrents ] , where we show that fields that are fixed under fusion with a simple current condensate must split under restriction . in the context of coset cfts
, special techniques have been developed to deal with fixed points@xcite , but it appears quantum group symmetry breaking takes care of fixed points without any changes to the procedure we have described already ( although of course the actual calculations involved in carrying out the procedure do become more complicated when fixed points appear ) .
fixed points are not the only complicating factors that may appear in the description of coset cfts .
there are in fact cosets for which there are more field identifications and more restrictive branching rules than one would expect from the action of the identification group ( one may show this for example by explicit calculation of the branching functions ) .
the first examples of such _ maverick cosets _ were found in refs . in 1992 , and several more have been found since@xcite . in the quantum group symmetry breaking formalism ,
such maverick cosets can be explained by the condensation of a bosonic field which is not a simple current , again , without any change to the framework we have described .
the simplest maverick coset is @xmath414 .
the central charge of this coset is @xmath415 , which means the coset theory must be related to the unitary minimal model @xmath416 .
in fact it turns out that the coset primary fields are in one to one correspondence with the subset of the chiral primary fields of @xmath416 which appear in the description of the critical point of the three - state potts model .
we will work out the quantum group symmetry breaking point of view on this coset in some detail .
we listed the conformal weights and quantum dimensions of the @xmath246 already in [ su2_8_spintab ] and we give those for the @xmath417 fields in table [ su3_2_spintab ] .
we label the @xmath246 fields in the usual way ( by their dynkin labels ) and we label the @xmath417 fields using a notation based on the dimensions of the corresponding @xmath418 representations .
the correspondence between the dynkin labels of the highest weights of these @xmath418 representations and the labels that we use is as follows : @xmath419 , @xmath420 , @xmath421 , @xmath422 , @xmath423 and @xmath424 .
@xmath335{|l| } \hline su(3)_{2}\\ \hline \hline \begin{array}{l|ll } 1&d_1=1&h_1=0\\ 6&d_6=1&h_6=1\\ \bar{6}&d_{\bar{6}}=1 & h_{\bar{6}}=1\\ 8&d_8= \frac{1+\sqrt{5}}{2}&h_8=\frac{3}{5}\\ 3&d_3= \frac{1+\sqrt{5}}{2}&h_3=\frac{2}{3}\\ \bar{3}&d_{\bar{3}}=\frac{1+\sqrt{5}}{2 } & h_{\bar{3}}=\frac{2}{3}\\ \end{array}\\ \hline \end{array } $ ] the fusion rules for @xmath246 can be read off from formula ( [ su2_k_fusion ] ) .
the fusion rules for @xmath417 are given in table [ tab : su(3)2_fusion ] .
@xmath425 from table [ su2_8_spintab ] , we read off that there is a single nontrivial identification current in the @xmath426 theory , namely the field with labels @xmath427 . in other words ,
there is also a bosonic field which is not a simple current , the field labeled @xmath429 .
let us first investigate what happens when we condense only the simple current field @xmath427 and not the field @xmath429 . since @xmath427 acts trivially on the @xmath418 part of the theory , we can just search for the restrictions of the pure @xmath246 fields , that is , the fields labeled @xmath430 for some @xmath88 .
the restrictions for more general fields will be similar .
the problem of breaking @xmath246 by condensation of the @xmath43-sector was already considered in section [ su28sec ] , and the results of the breaking were presented in table [ tab : su(2)_8_broken ] .
after symmetry breaking and confinement , @xmath246 reduces to a @xmath431 theory , with @xmath93 sectors , labeled in table [ tab : su(2)_8_broken ] as @xmath20 , @xmath51 , @xmath327 and @xmath344 .
this means that condensation of the @xmath427 field in the @xmath426 theory will lead , after condensation and confinement , to a new theory with @xmath56 distinct sectors , each labeled by an @xmath417 representation and a label from the broken and confined remnant of the @xmath246 theory .
this @xmath56 sector theory is clearly not the right description of the coset @xmath414 .
the full virasoro minimal model @xmath432 at @xmath433 only has @xmath434 sectors , so @xmath56 sectors is clearly too many and also some of the conformal weights we find are not compatible with the conformal weights of @xmath432 sector model , it just means that this cft would have to occur at some central charge which is different from @xmath415 , necessarily by some nonzero multiple of @xmath43 . ] . to describe the coset at @xmath433
, we must condense the non - simple current field @xmath429 in addition to the field @xmath427 .
this suggests that for the general description of coset models , we should condense all available bosons .
note however that in the case treated here , we will show that condensation of @xmath429 actually implies that @xmath427 condenses as well .
we sketch the calculation of the branching rules .
first of all , we note that there will be many additional splittings which do not occur when only @xmath427 is condensed .
for example the fusions @xmath435 for @xmath436 all contain the field @xmath429 , which means the restrictions of these fusions contain the vacuum at least twice and hence all the fields @xmath437 with @xmath438 split . as a result the fields @xmath439 and @xmath440 with @xmath88 in the same range also split under restriction , since these fields can be obtained from the @xmath437 fields by fusion with the simple currents @xmath441 and @xmath442 .
there are also @xmath443 sectors which will certainly not split under restriction , because they have quantum dimensions less than @xmath51 .
these are all sectors with labels of the form @xmath444 or @xmath445 as well as the @xmath307 fields with labels @xmath446 . the sector labeled @xmath447 could in principle split into two sectors @xmath448 and @xmath449 of quantum dimensions @xmath21 and @xmath450 respectively
. given such a splitting , we know that the restriction of @xmath451 must contain the vacuum twice and we find @xmath452 and comparing the first and last lines , we notice that either @xmath453 or on of the components of @xmath447 must branch to the vacuum .
however , this can not be allowed , since @xmath453 and @xmath447 have nontrivial spin and hence can not condense ( alternatively we might say this would confine the vacuum ) . in conclusion
, we find that the sector @xmath447 does not split under restriction .
this is a crucial piece of information in what follows .
now let us consider the fusion of the condensed field @xmath429 with the @xmath246 type fields @xmath430 .
whenever @xmath88 is even , we have @xmath454 .
since @xmath429 has the trivial field in its restriction , we see that the restriction of @xmath429 must contain the restriction of the dual of some component of each of the sectors @xmath430 with @xmath88 even .
since the sectors @xmath430 are self - dual , we see in fact that the restriction of @xmath429 contains the restrictions of @xmath455 , @xmath456 , @xmath457 and @xmath458 and at least one component of the restriction of @xmath459 .
similarly , we may consider the fusion of @xmath429 with fields labeled @xmath437 and derive that @xmath429 contains at least one component of the restrictions of @xmath460 , @xmath461 , @xmath429 , @xmath462 and @xmath463 . from this point
simple arguments using quantum dimensions give us much information on splittings and field identifications .
the quantum dimension of @xmath429 is @xmath464 .
we know that the restriction of @xmath429 contains at least the vacuum @xmath465 , which has quantum dimension @xmath21 and the full restriction of @xmath447 , which has quantum dimension @xmath464 ( it must contain the full restriction , since @xmath447 does not split ) .
hence it follows that @xmath429 splits into @xmath9 components , the vacuum @xmath466 the restriction of @xmath456 which we just denote @xmath467 as well and a third component @xmath468 of quantum dimension @xmath450 .
but in the previous paragraph we noted that the restriction of @xmath429 contains the restriction of @xmath427 and since this has quantum dimension equal to @xmath21 , it must be the same as the restriction of @xmath465 , in other words , we have shown that @xmath427 is condensed , as we promised earlier .
this immediately fixes the restrictions of the fields in the pure @xmath246 sector , as in our treatment in section [ su28sec ] and consequently also the restrictions of the fields with labels @xmath469 and @xmath470 . to find the restrictions of the remaining fields , consider the fusion @xmath471 .
we have @xmath472 where @xmath473 and @xmath474 are the two components of @xmath453 which result from the condensation of @xmath427 ( these correspond to @xmath327 and @xmath344 in table [ tab : su(2)_8_broken ] ) . comparing the two lines
, we see that the restriction of @xmath460 must equal either @xmath473 or @xmath474 .
we can in fact make an arbitrary choice between these two options , because the fusion rules of the broken @xmath246 theory are invariant under the exchange of the sectors @xmath473 and @xmath474 .
choosing @xmath475 , we can write @xmath476 using the fusion rules for the broken @xmath246 theory given in table [ tab : su(2)_8_broken ] . comparing this with the previous equation , we finally get the full branching of @xmath429 , namely @xmath477 . from here
, we can easily produce the branchings for @xmath478 , @xmath461 and @xmath479 .
we have @xmath480 we summarize the branchings for the fields of the forms @xmath430 and @xmath437 in table [ tab : su(3)2xsu(2)8broken ] .
they branch to a set of fields which is in one to one correspondence with the fields of the broken @xmath246 theory and which also has the same fusion rules ( see table [ tab : su(2)_8_broken ] ) .
the full set of fields for the broken @xmath481 theory has @xmath443 fields corresponding to the products of the @xmath482 group of simple currents @xmath483 with the broken @xmath246 fields @xmath484 .
the branching for the full theory can easily be obtained from the branchings for the fields given in table [ tab : su(3)2xsu(2)8broken ] .
for example , for fields of the form @xmath439 , we write @xmath485 and conclude that if the branching for @xmath437 was given by @xmath486 then the branching for @xmath439 is given by @xmath487 .
a similar statement holds for all other branchings .
the fusion rules of the full broken theory can also be described easily ; they are just the product of the @xmath482 fusion rules for @xmath483 and the broken @xmath246 fusion rules .
@xmath488 using the branching rules , we may now check for confinement in the usual way .
note that some of the fields that were not confined in the broken @xmath246 theory are now confined , because they appear in more branching rules than before and no longer have well defined spin factors as a result .
it turns out that there are @xmath307 non - confined fields , which have precisely the conformal weights and fusion rules of the chiral three state potts model , or equivalently of the @xmath307 fields involved in the non - diagonal modular invariant for the @xmath416 minimal model .
hence , we have reproduced the topological data of this maverick coset , using precisely the same quantum group symmetry breaking formalism as for standard cosets .
@xmath335{|l| } \hline su(3)_{2}/su(2)_8\\ \hline \hline \begin{array}{l|ll } ~&d&h\\ \hline ( 1,0)&1&0\\ ( 6,0)&1&\frac{2}{3}\\ ( \bar{6},0)&1 & \frac{2}{3}\\ ( 1,4_2)&\frac{1+\sqrt{5}}{2}&\frac{2}{5}\\ ( 6,4_2)&\frac{1+\sqrt{5}}{2}&\frac{1}{15}\\ ( \bar{6},4_2)&\frac{1+\sqrt{5}}{2}&\frac{1}{15}\\ \end{array}\\ \hline \end{array } $ ] since conformal embeddings @xmath283 conserve the central charge , the cosets @xmath285 coming from these embeddings have conformal central charge @xmath68 and must be trivial . still , confirming the triviality of these cosets using the identification group may not be so trivial , because the procedure can involve resolution of fixed points and even dealing with more complicated issues like those which occur for the ( nontrivial ) maverick cosets . from the quantum group symmetry breaking perspective , conformal embeddings are in fact just mavericks for which the coset happens to come out trivial .
the quantum group symmetry breaking perspective on these cosets adds information to the usual treatment , because , while the effective theory for the non - confined excitations of the coset is of course trivial , the symmetry breaking approach also gives a description of the confined excitations , which we can view as boundary excitations between a phase with @xmath409 topological order and a topologically trivial phase .
this boundary theory will be nontrivial .
in fact , one may expect that the boundary theory is the same as the boundary theory for a boundary between @xmath298 and @xmath290 phases .
the reason for this is that the @xmath409 theory can be thought of as a two layer theory , where a piece of material with @xmath290 topological order has been folded under a piece with @xmath298 topological order .
this folding converts the boundary between the region with @xmath298 order and the region with @xmath290 order into a boundary between a region with @xmath409 topological order and a region with trivial topological order .
however , since folding is just a geometric deformation of the medium it should not change the topological order on the boundary and so we expect the two boundary theories to be the same .
we will now demonstrate triviality of the coset , as well as this correspondence of boundary theories for our favourite example , @xmath489 .
forming the product @xmath490 , we see that there are three nontrivial bosons , labeled @xmath453 , @xmath491 and @xmath492 in our usual labeling conventions for @xmath12 and @xmath10 .
only the field @xmath453 is a simple current and , as with the maverick @xmath414 , we find that condensing only this simple current does not give the desired result , that is , the effective theory after breaking and confinement is still nontrivial .
however , if we condense all bosonic fields , we find that we do obtain the correct ( trivial ) coset theory , as in the case of the maverick .
we give the results of the symmetry breaking calculation in table [ tab : su3_1xsu2_4broken ] .
after symmetry breaking , there are @xmath93 sectors left , the vacuum sector and the restrictions of @xmath493 , @xmath494 and @xmath495 .
the fields @xmath453 , @xmath491 and @xmath492 branch to the vacuum , so they are indeed condensed . also , it is easy to check , using the weights of @xmath10 and of @xmath12 ( see tables [ tab : su24 ] and [ tab : su3_1_and_so5_1 ] ) , that all nontrivial broken sectors are confined , leaving just the vacuum sector and hence confirming that the coset is trivial .
finally , the @xmath93 sectors of the boundary theory have quantum dimensions @xmath21 , @xmath21 , @xmath21 and @xmath496 , which fixes the fusion rules and indeed , we see that there is a one to one correspondence with the boundary theory between @xmath10 and @xmath12 given in table [ tab : su3_1xsu2_4broken ] , as expected .
in section [ sec : bosons ] we saw that many examples of anyon models with bosons can be obtained either from kitaev s toric code construction@xcite or from gauge theories by breaking the gauge group to a discrete subgroup@xcite .
these theories can also be realized as conformal field theories , namely as orbifolds of topologically trivial cfts@xcite .
all the topological information in these models can be described using the representation theory of the quantum doubles of finite groups@xcite .
probably the simplest example which allows for non - abelian braiding is the model based on the quantum double @xmath498 of the smallest non - abelian group @xmath499 , the symmetry group of the regular triangle , or equivalently , the permutation group of three objects .
this model has been shown to allow for universal quantum computation , if some measurements are allowed as operations in addition to braiding@xcite .
an implementation of this model using josephson junctions has been proposed in refs . .
there are @xmath43 topological sectors in the model , each labeled as described in section [ sec : bosons ] , by a conjugacy class of @xmath499 and a representation of the centralizer group of one of the elements in that conjugacy class ( the elements all have isomorphic centralizer groups ) .
@xmath499 has three conjugacy classes , the class of the trivial element @xmath206 , a class we denote @xmath500 , containing the nontrivial rotations of the triangle ( three - cycles as permutations ) and a class called @xmath501 which contains the reflections ( exchanges ) .
the trivial class has all of @xmath499 as its centraliser , leading to three particle sectors @xmath502,@xmath503 and @xmath504 corresponding to the three irreducible representations @xmath21 , @xmath220 and @xmath98 of @xmath499 . here
@xmath21 denotes the trivial representation , making @xmath502 the vacuum sector , and @xmath220 and @xmath98 denote the nontrivial one dimensional and two dimensional irreducible representations respectively .
the centraliser of @xmath500 is the @xmath505 generated by the rotations , giving sectors @xmath506 and @xmath507 and the centraliser of @xmath501 is a @xmath508 giving two sectors @xmath509 and @xmath510 .
the spin factors and quantum dimensions of these sectors are given in table [ d3_spintab ] .
[ d3sec ] @xmath511 the fusion rules of the irreps of @xmath512 have been worked out in ref . .
for @xmath498 , we have of course @xmath513 ( for all @xmath102 ) and furthermore @xmath514 for the fusion rules of the @xmath515 , we have @xmath516 finally , we have @xmath517 a look at the spin factors and fusion rules confirms immediately that there is a wealth of bosons in the theory . for the odd dihedral groups ( including @xmath499 ) , we have analyzed all possible choices of condensate using our earlier quantum group based approach@xcite and we will not repeat that exercise here .
however , it will be good to check in an example that we can actually reproduce the results obtained there .
this will also serve to illustrate some of the more interesting things that may happen on condensation .
in particular , we will see an example with non - abelian fusion rules after symmetry breaking ( but before confinement ) .
we will investigate what happens when an excitation in the @xmath504 sector condenses . in our earlier treatment of quantum group symmetry
breaking , there were two non - equivalent ways of condensing this sector , because we could choose different ( non gauge equivalent ) internal states of the @xmath504 particles to form the condensate ; note that since all quantum dimensions are integers here , it makes sense to talk about internal hilbert spaces for single particles .
we will find that our current methods produce the same two unconfined theories .
however at the level of the broken theory including confined excitations , we find one extra solution to the requirements set out in this paper , in addition to the two solutions produced by our previous methods .
the extra solution is almost certainly spurious and due to the fact that the requirements we give here are not completely sufficient to determine the broken theory ( before confinement ) in this case .
it is not surprising that this can sometimes happen , since we have restricted our attention to a relatively crude level of description of topological order in this paper , looking only at fusion rules and spin factors . for theories with only integer quantum dimensions
our old methods allow an approach to the problem which makes full use of the underlying hopf algebra theory . also , theories with many integer quantum dimensions can be relatively complicated to handle with the methods of this paper because integers allow for so many different splittings into smaller integers .
now let us sketch the calculations which lead to the above results .
given that @xmath504 condenses , we know that @xmath504 branches to the vacuum sector and some other one dimensional sector . since @xmath518 , it follows that in fact @xmath504 branches to @xmath21 plus the restriction of @xmath503 .
now we have two possibilities : either @xmath503 restricts to the vacuum or it does not .
both possibilities cases lead to a consistent theory for the confined and non - confined excitations .
let us first assume that @xmath503 does not branch to the vacuum , but rather to some nontrivial sector which we will still call @xmath503
. then @xmath519 .
comparing @xmath520 with @xmath521 , we see immediately that all @xmath522 must branch to the same new sector , which we will simply call @xmath523 .
we also see that @xmath524 appears on the right hand side of the fusion rules for the fields @xmath509 and @xmath510 , so these must both split . after completing the calculation , we find the branching given in table [ tab : dd3broken1 ] .
the unconfined fields are the vacuum , @xmath503 , @xmath525 and @xmath526 .
these have spins @xmath21 , @xmath21 , @xmath21 and @xmath59 respectively and @xmath527 fusion rules , which fixes the topological order to be that of the @xmath508 discrete gauge theory or toric code model . in the discrete gauge theory
, we can interpret the transition as a higgs effect which has broken the @xmath499 gauge group down to a @xmath508 subgroup . on the cft side it looks like we have partially unorbifolded the original theory , leaving a cft where only the @xmath508 subgroup of the @xmath499 is orbifolded .
this could be interpreted as due to extension of the orbifold conformal algebra by some of the chiral primaries for the twisted sectors .
@xmath528 now let us consider the possibility that not only @xmath504 , but also @xmath503 condenses .
in this case the entire electric part of the spectrum becomes trivial , or in other words , the gauge symmetry is fully broken .
as before , we note that the restrictions of the @xmath522 must all be equal and now because @xmath529 , we see that the restrictions of @xmath510 and @xmath509 also equal each other . in the fusion rules
@xmath530 , we see that @xmath531 appears on the right hand side , implying that the @xmath522 split into two parts of quantum dimension @xmath21 .
similarly , from @xmath532 , we see that @xmath509 and @xmath510 split into three parts , also of quantum dimension @xmath21 .
we note that none of the @xmath522 can branch to the vacuum because not all the @xmath522 have trivial spin and they all have the same restriction .
similarly @xmath510 ( and hence @xmath509 ) can not branch to the vacuum . since all sectors after breaking are simple currents , they form a group under fusion .
moreover , one sees easily that the compomnents of the @xmath522 , together with the vacuum @xmath21 , form a subgroup .
if the components of the restriction of @xmath522 were equal , this would have to be a @xmath508 subgroup , but this is inconsistent with the fusion rules of the @xmath522 , so there must be two different components in the restriction of @xmath522 and with the vacuum , these form the group @xmath482 under fusion . from the fusion rules for @xmath509 and @xmath510 we now read off that the three components of the restriction of these fields must also be distinct .
this means in particular that at least one of them is a component that does not occur in the @xmath482 that we already uncovered .
but since we are looking for a group with a @xmath482 subgroup that has at least @xmath93 and most @xmath307 elements , we now have only two possibilities left for the entire group of fields of the broken theory : it is isomorphic either to @xmath533 or to @xmath499 itself . in the first case , the @xmath482 would be the subgroup of even elements of @xmath533 and in the second case , it would be the @xmath482 subgroup of rotations ( or three - cycles in the permutation representation of @xmath499 ) .
it turns out that both options are fully consistent with the fusion rules of the original theory .
also , both lead to the same unconfined theory , namely the trivial theory with one sector ( in other words , all sectors of the broken theory save the vacuum sector are confined ) . however , from basic intuition about discrete gauge theory or toric code models , as well as from our formalism for theories with integer quantum dimensions based on hopf algebra theory@xcite , we know that the correct broken ( but confined ) theory should be the one whose sectors fuse according to the group multiplication of @xmath499 . in other words
, the @xmath533 is the spurious solution to the requirements posed in this paper that we already announced .
the fact that we find a theory with fusion rules described by the group multiplication of @xmath499 is also interesting in itself , since it shows that our formalism can produce boundary theories which have non - abelian fusion rules .
looking at the situation from a cft perspective , it appears that we have fully unorbifolded the theory this time and this might lead one to conjecture that quantum group symmetry breaking accomplishes a sort of inverse of orbifolding .
however , this seems premature and we expect that a more complicated picture emerges when one studies different condensates and orbifolds of topologically nontrivial cfts .
one may envision many constructions of new topological field theories and corresponding conformal field theories based on the principle of quantum group symmetry breaking .
perhaps the simplest thing one may do is tensor a number of known tqfts or cfts together in such a way that the tensor product theory has some bosonic sectors and then condense some or all of these bosonic sectors .
the coset construction is of course a special case of such a construction , but more generally we do nt have to require that the tensor product is of the form @xmath392 with @xmath283 .
it is not difficult to come up with simple examples of such theories which are not cosets .
an example of potential interest in the quantum - hall context corresponds to the product @xmath534 , where @xmath535 is the unitary minimal model at @xmath536 .
after condensation of the single nontrivial simple current in this theory we obtain precisely the spins and fusion rules of the @xmath417 parafermions .
this may be connected with the interface or transition between the spin - polarized moore - read state , which is based on the ising model , and a non - abelian spin singlet state ( nass ) proposed by schoutens and ardonne@xcite , based on the @xmath417 parafermionic cft@xcite .
we will return to this in detail elsewhere@xcite .
there are many other examples one can think of , e.g. one may take @xmath537 and condense the bosonic simple current @xmath538 , or alternatively , one may take @xmath539 and condense the bosonic simple current @xmath540 .
it is often not at all obvious what the conformal field theories corresponding to these constructions should look like .
for example the case @xmath541 yields topological central charge @xmath542 , but if there is a corrresponding unitary cft , then its conformal central charge can not be equal to @xmath543 , since there is no unitary minimal model with this central charge , and it must differ from @xmath543 by some multiple of @xmath43 .
on the other hand , we do expect , in analogy to the chern - simons description of coset theories@xcite , that the topological data for such theories should be described by a chern - simons theory whose gauge group is a quotient of the product of the groups appearing in the construction , with chern - simons terms for these groups at the appropriate levels
. the quotient would be by a finite group characterizing the simple currents which are condensed ( for situations where the condensed sectors are not all simple currents , the situation may be more complicted ) .
if this conjectured chern simons description is correct , then this suggests that a cft with the same topological order can be obtained as a boundary theory of this chern - simons theory .
many constructions of this type could start from products of the form @xmath544 , which are just doubled chern - simons theories , or more generally from products @xmath545 where @xmath5 represents a tqft which is not of the @xmath298 type .
such theories are important in the description of string net condensates@xcite and picture tqfts@xcite . as we discussed in section
[ sec : bosons ] , these theories all have bosons , namely the diagonal sectors with labels of the form @xmath546 .
we should be able to produce many new theories by condensing some of these bosons .
if we condense all diagonal fields , then we should expect that the broken phase is topologically trivial , while the boundary between the broken and unbroken phases will be described by @xmath298 ( or more generally @xmath5 ) itself .
we give an intuitive argument for this first and then sketch a proof .
intuitively , condensing all bosons in the theory should implement the coset construction which in this case gives the completely trivial theory .
the boundary between the completely trivial theory and the @xmath547 theory should be described by an @xmath5 or an @xmath548 theory according to the folding principle for boundaries that we introduced in section [ sec : embeddings2 ] .
that is , we can think of the @xmath545-theory as just a @xmath5-theory on a plane which has been folded over to give two layers .
the boundary between this folded plane and empty space is not really a boundary on the plane before folding and so it should have the same excitations as the plane itself , in other words , it should be described by an @xmath5 theory or an @xmath548 theory .
the @xmath5 and @xmath548 theories of course have the same fusion rules but opposite spins .
the ambiguity in the spins explains the complete confinement of these boundary excitations .
one might expect less severe confinement if the @xmath5 theory itself has bosons .
now let us give an argument that comes closer to a proof .
first of all consider the fusions @xmath549 .
since the right hand side has the vacuum in its restriction , we see that the restriction of @xmath550 must be identified with the restriction of @xmath551 .
now more generally , we have that @xmath552 and hence the restrictions of all sectors can be written in terms of the restrictions of the sectors @xmath553 alone , confirming that the broken theory before confinement is simply a single copy of the @xmath5 theory itself .
now let us consider confinement .
in order for the restriction of @xmath553 to be unconfined , we must have @xmath554 . hence unconfined particles have trivial spin .
in fact , unconfined particles should satisfy much more stringent conditions . if @xmath553 is not confined , then for any @xmath555 such that @xmath556 , we must have @xmath557
. however , from this , it follows also that for any @xmath558 , @xmath559 such that @xmath560 , we have @xmath561 and using the ribbon equation , we see that @xmath88 must have totally trivial monodromy with all other fields in the @xmath5 theory . however , this can only happen if the tensor category describing @xmath5 is not modular ( see ref . ,
section e.5 ) , so we have argued that there is indeed complete confinement if @xmath5 is modular .
perhaps the simplest example of condensation in a doubled theory which does not lead to complete confinement occurs in the doubled ising model .
we label the sectors of the ising model in the usal way by @xmath21 , @xmath562 and @xmath563 .
their spins and quantum dimensions are given in table [ tab : su22 ] .
the @xmath564 theory has three bosonic fields , the vacuum @xmath495 and the diagonal fields @xmath565 and @xmath566 .
it is possible to have condensation of @xmath566 without condensing @xmath565 .
this leads to the branching in table [ tab : doubled_ising_broken ] .
the broken theory has @xmath307 sectors and its fusion is the same as that of an @xmath567 tensor product theory .
there are four simple currents , @xmath495 , @xmath568 and the two components of the restriction of @xmath565 .
these are also the unconfined fields and they form a @xmath569 group under fusion .
this is actually not completely straightforward to derive since the requirements for symmetry breaking we have stated in this paper would also be consistent with @xmath570 fusion rules .
however , one may check that the values of the spins of these fields are consistent only with @xmath569 fusion rules and in fact , from the spins and the fusion we see that the unconfined theory has precisely the topological order of a @xmath508 discrete gauge theory or toric code model .
@xmath571 this suggests that in any local model which realizes the doubled ising model one might expect a transition to an abelian topological phase of @xmath569 type .
loop gases with ground states reflecting topological order of doubled ising type have been constructed@xcite , but it has been shown recently@xcite that these loop states can not be ground states of a gapped local hamiltonian .
they are in fact associated with gapless critical points and it is possible to drive the models away from these critical points and into in a gapped abelian topological phase with the same topological order as the @xmath508 toric code or @xmath508 discrete gauge theory
. it would be interesting to study if this can be viewed as due to the condensation of bosonic excitations of @xmath566 type .
we have given some simple principles and requirements relating the spectra and topological interactions of topological phases that may be obtained from one another by condensation of a bosonic quasiparticle , based on the the idea of quantum group symmetry breaking .
these turn out to be surprisingly powerful and practical tools in determining the topological field theory describing the condensed phase from the tqft that describes the phase without condensate and also in describing the boundaries between condensed and uncondensed topological phases .
we have worked out a number of examples in detail and shown connections between our quantum group symmetry breaking scheme and various constructions in conformal field theory .
* more detailed study of systems which are or may soon be accessible by experiment , notably the various proposed non - abelian fractional quantum hall states . *
explicit realization of the predictions on phase transitions in topological models that we make here in local models that exhibit topological phases .
one aspect of this would be the introduction of ( necessarily non - local ) order parameters which signal a non - zero condensate density@xcite .
string - net condensed phases would be a good laboratory for this . * a more mathematically rigorous treatment of the material presented here .
this would hopefully allow for the systematic construction of the full unitary braided tensor category describing the condensed theory .
also one would like to prove some of the observations on @xmath17 and @xmath40 in section [ sec : observations ] . *
more in - depth treatment of the relation between quantum group symmetry breaking and chiral algebra extension , i.e. explicit investigation of the action of the chiral algebra on the hilbert space of the cft before and after condensation .
conformal embeddings would be the obvious place to start such a program , but it would be especially interesting if one could find a natural cft counterpart for some of the tqft constructions mentioned in section [ sec : new_constructions ] . | we investigate transitions between topologically ordered phases in two spatial dimensions induced by the condensation of a bosonic quasiparticle . to this end
, we formulate an extension of the theory of symmetry breaking phase transitions which applies to phases with topological excitations described by quantum groups or modular tensor categories .
this enables us to deal with phases whose quasiparticles have non - integer quantum dimensions and obey braid statistics .
many examples of such phases can be constructed from two - dimensional rational conformal field theories and we find that there is a beautiful connection between quantum group symmetry breaking and certain well - known constructions in conformal field theory , notably the coset construction , the construction of orbifold models and more general conformal extensions .
besides the general framework , many representative examples are worked out in detail . |
this work is supported by rfbr , grant 05 - 02 - 17477 , grant ur 02.01.299 and infn , grant is / pi13 .
work of s. a. and d. e. is partially supported by cyt fpa , grant 2004 - 04582-c02 - 01 and by cirit gc , grant 2001sgr-00065 .
9 a. bramon , e. etim and m. greco , _ phys .
lett . _ * b41 * , 609 ( 1972 ) ; m. greco , _ nucl .
phys . _ * b63 * , 398 ( 1973 ) ; j. j. sakurai , _ phys . lett . _ * b46 * , 207 ( 1973 ) .
a. v. anisovich , v. v. anisovich and a. v. sarantsev , _ phys .
rev . _ * d62 * , 051502 ( 2000 ) .
s. s. afonin , a. a. andrianov , v. a. andrianov and d. espriu , _ jhep _ * 04 * , 039 ( 2004 ) .
g. t hooft , _ nucl .
_ * b72 * , 461 ( 1974 ) ; e. witten , _ nucl .
* b160 * , 57 ( 1979 ) .
m. shifman , hep - ph/0009131 ; hep - ph/0507246 . m. a. shifman , a. i. vainstein and v. i. zakharov , _ nucl
. phys . _ * b147 * , 385 , 448 ( 1979 ) .
a. a. andrianov and v. a. andrianov , hep - ph/9705364 ; hep - ph/9911383;a .
a. andrianov , d. espriu and r. tarrach , _ nucl .
_ * b533 * , 429 ( 1998 ) .
m. golterman and s. peris , _ jhep _ * 01 * , 028 ( 2001 ) . s. r. beane , _ phys . rev . _ * d64 * , 116010 ( 2001 ) .
a. simonov , _ phys .
_ * 65 * , 135 ( 2002 ) .
particle date group ( s. eidelman _ et al .
_ ) , _ phys . lett . _ * b592 * , 1 ( 2004 ) . | we investigate the possible corrections to the linear regge trajectories for the light - quark meson sector by matching two - point correlators of quark currents to the operator product expansion .
we find that the allowed modifications to the linear behavior must decrease rapidly with the principal quantum number . after fitting the lightest states in each channel and certain low - energy constants the whole spectrum for meson masses and residues
is obtained in a satisfactory agreement with phenomenology . the perturbative corrections to our results
are discussed .
the observed masses squared of mesons with given quantum numbers form linear trajectories @xcite depending on the number of radial excitation @xmath0 .
this is a strong indication that qcd admits an effective string description , as this type of spectrum is characteristic e.g. of the bosonic string . in the bosonic string model
the slope of all trajectories must be equal since this quantity is proportional to the string tension depending on gluedynamics only .
however , there exist sizeable deviations from the string picture . in the present analysis we examine possible corrections to the linear trajectories in the vector ( v ) , axial - vector ( a ) , scalar ( s ) , and pseudoscalar ( p ) channels @xcite .
our method is based on the consideration of the two - point correlators of v , a , s , p quark currents in the large-@xmath1 limit of qcd @xcite . on the one hand , by virtue of confinement
they are saturated by an infinite set of narrow meson resonances , that is , they can be represented by the sum of related meson poles in euclidean space : @xmath2 expressing the quark - hadron duality@xcite . here @xmath3 ; @xmath4 .
further we denote @xmath5 . on the other hand ,
their high - energy asymptotics is provided by the perturbation theory and the operator product expansion ( ope ) with condensates @xcite .
matching these two approaches results in the chiral symmetry restoration sum rules representing a set of constraints on meson mass parameters @xcite .
we performed the further analysis in the chiral limit and firstly in the leading order of perturbation theory .
let us consider the linear ansatz for the meson mass spectra with a non - linear correction @xmath6 : @xmath7 the last term in eq .
( [ acor1 ] ) signifies a possible deviation from the string picture in qcd .
the condition of convergence for the generalized weinberg sum rules @xcite imposes the equality of slopes and a falloff of non - linear correction .
the asymptotic freedom leads to the relation between residues and masses : @xmath8 .
our analysis showed that the analytical structure of the ope admits , however , exponentially small deviations from this relation ( in contrast to to @xcite ) , @xmath9 with certain constants @xmath10 and @xmath11 to be fitted .
it is plausible to suppose that , for masses , the dynamics under these exponentially small corrections is governed mostly by gluons and thereby does not depend on flavor .
thus , we keep the exponent @xmath12 the same for all channels . for the same reason we regard @xmath13 as independent of parity . in tables 1 and 2
we show an example of numerical fits resulting from our approach where in the s , p - channels two possibilities are considered : @xmath14-meson belongs to the radial regge trajectory ( @xmath14-in ) and does not ( @xmath14-out ) .
the inputs general for all tables ( if any ) are : @xmath15 , @xmath16 , @xmath17 , @xmath18 , @xmath19 , @xmath20 .
the units are : @xmath21 , @xmath22 , @xmath23 mev ; @xmath24 mev@xmath25 ; @xmath26 , @xmath27 , @xmath28 mev@xmath29 .
let us discuss the impact of running coupling constant at next - to - leading order of perturbation theory .
we have in this case the contribution to the imaginary part of correlator which is related to the full correlator through the dispersion relation . in the large-@xmath1 limit for the vector and axial - vector case
this leads to : @xmath30 we perform summation in eq . by applying the euler - maclaurin summation formula
this provides smoothness of this expression .
finally we obtain the relation : @xmath31 eq .
can be approximated by finite differences .
for the first two states this reads : @xmath32 substituting experimental values @xcite into eq .
one arrives at the estimate : @xmath33 in a good agreement with the phenomenology . in order to reproduce the running coupling behaviour in eq .
one should accept the following ansatz for the residues : @xmath34 however this ansatz introduces into the ope the additional terms @xmath35 ( and powers of these terms ) which are absent in the standart ope .
thus ansatz must be improved .
this problem is under our studies now .
let us summarize the results of our analysis : + * 1 ) * the convergence of the generalized weinberg sum rules requires the universality of slopes and intercepts for parity conjugated trajectories .
+ * 2 ) * the matching to the ope can not be achieved by a simple linear parameterization of the mass spectrum , the linear trajectory ansatz .
there must exist deviations from the linear trajectory ansatz triggered by chiral symmetry breaking .
these deviations must decrease at least exponentially with @xmath0 .
+ * 3 ) * for heavy states , the d - wave vector mesons have to decouple from asymptotic sum rules .
this fact implies the exponential ( or faster ) decreasing the corresponding decay constants @xmath36 .
+ * 4 ) * our results seem to exclude a light @xmath37 particle as a quarkonium state and rather favor the non - linear realization of chiral symmetry with the lightest scalar of mass @xmath38 gev , its chiral partner being the @xmath39 .
+ * 5 ) * in our approach the quantities @xmath40 and @xmath41 are obtained , in satisfactory agreement with the phenomenology .
+ * 6 ) * perturbative corrections can be systematically treated making , nevertheless , small effects on the fits presented . |
the optoelectronic properties of 2d transition metal dichalcogenide ( tmd ) monolayers are dominated by excitons @xcite . distinguishing features of the excitons in 2d metal dichalcogenides are the large exciton binding energies and the strong exciton - photon interactions @xcite .
recently , exciton - polaritons have been also studied experimentally and theoretically in these materials @xcite .
the strong exciton - photon coupling results in spontaneous emission radiative lifetimes in the hundreds of femtoseconds range @xcite .
the strong exciton - photon coupling suggests that the rates for interlayer energy transfer between excitons in parallel tmd monolayers would also be fast . in electronically coupled 2d tmd monolayers , ultrafast energy transfer via
interlayer charge transfer has been observed @xcite . in this work
, we study the rate of transfer of energy between excitons in parallel tmd monolayers as a result of electromagnetic coupling . the mechanism for this energy exchange could be both radiative ( mediated by propagating photons for exciton states within the light cone
@xcite ) or non - radiative ( mediated by evanescent photons that are bound to the exciton states outside the light cone in an isolated tmd layer as exciton - polaritons but can mediate energy exchange between two tmd layers if the two tmd layers are close enough ) .
the latter mechanism is the same as the well known forster resonance energy transfer ( fret ) mechanism due to dipole - dipole coupling @xcite .
however , the use of the standard fret dipole - dipole energy exchange formulas in the present case gives erroneous results since it ignores the retarded nature of the exciton interlayer interaction @xcite .
the quantum electrodynamic green s function approach used here treats both these mechanisms on equal footing while taking into account the corrections to the longitudinal and the transverse exciton energy dispersion relations due to coupling with the radiation .
the energy transfer times depend on the exciton momentum , exciton intralayer scattering rates , and the interlayer separation .
our results show that the large exciton optical oscillator strengths in tmd monolayers result in energy transfer times shorter than 100 fs for longitudinal excitons for interlayer spacings smaller than 10 nm .
average energy transfer times for a thermal ensemble of longitudinal and transverse excitons can be smaller than a picosecond .
we also consider localized excitons and find that localized longitudinal excitons can also have energy transfer times shorter than a picosecond for interlayer spacings smaller than 10 nm .
if the exciton scattering rates are fast , energy transfer involves a simple decay of energy from one layer to the other .
if the exciton scattering rates are slow , energy transfer dynamics can exhibit coherent oscillations .
conditions for observing such oscillations are discussed in this paper .
our results show that electromagnetic interlayer energy transfer can be an efficient mechanism for energy exchange between electronically uncoupled tmd monolayers and can even compete with the interlayer charge transfer mechanism in the case of electronically coupled tmd layers .
section [ sec : prelim ] discusses exciton states in tmds , longitudinal and transverse excitons in tmds , and the hamiltonian terms that describe exciton - photon interaction in tmds .
sections [ sec : rate ] and [ sec : dis ] present the main results of this paper on the energy transfer rates between two tmd layers . section [ sec : num ] presents numerical results for energy transfer rates between two mos@xmath0 layers .
section [ sec : coh ] considers the case where the exciton intralayer scattering and dephasing rates are small and energy transfer dynamics exhibit coherent oscillations .
sections [ sec : app1 ] and [ sec : app2 ] discuss some special cases , including that of localized excitons and energy transfer between excitons and free electron - hole pairs in tmd heterolayers .
the appendix discusses exciton self - energies and energy dispersions in tmd monolayers in the presence of exciton - photon interaction .
although most of the numerical results presented in this paper are for mos@xmath0 monolayers , the analysis and the results presented are expected to be relevant to all tmds , and are expected to be useful in realizing new kinds of metal dichalcogenide optoelectronic devices based on energy transfer .
the conduction and valence bands in monolayer tmds near the @xmath1 and @xmath2 points in the brillouin zone are described by the hamiltonian @xcite , @xmath3 \label{eq : h1}\ ] ] here , @xmath4 is related to the material bandgap , @xmath5 stands for the electron spin , @xmath6 stands for the @xmath1 and @xmath2 valleys , @xmath7 is the splitting of the valence band due to spin - orbit coupling , @xmath8 , and the velocity parameter @xmath9 is related to the coupling between the orbitals on neighboring @xmath10 atoms . from density
functional theories @xcite , @xmath11 m / s . the wavevectors are measured from the @xmath1(@xmath2 ) points .
the d - orbital basis used in writing the above hamiltonian are @xmath12 and @xmath13 @xcite .
we will use the symbol @xmath14 for the combined valley ( @xmath15 ) and spin ( @xmath16 ) degrees of freedom .
the intravalley momentum matrix element between the conduction and valence band states near @xmath1(@xmath2 ) points follows from the above hamiltonian , @xmath17 \nonumber \\
\label{eq : matrix}\end{aligned}\ ] ] here , @xmath18 is the free electron mass , @xmath19 is the phase of the wavevector @xmath20 , and , @xmath21 near the band extrema , @xmath22 and @xmath23
. exciton states in tmds have been discussed in detail in several published works @xcite .
we assume an undoped intrinsic tmd layer .
the creation operator for an exciton state with center - of - mass momentum @xmath24 is defined as @xcite , @xmath25 @xmath26 and @xmath27 are the destruction operators for the conduction band and valence band electron states , respectively , @xmath28 is the exciton relative wavefunction , and @xmath29 is the area of the tmd monolayer .
@xmath30 , @xmath31 , @xmath32 , where @xmath33 , @xmath34 , and @xmath35 are the effective masses of electrons , holes , and excitons , respectively @xcite .
the subscript @xmath36 corresponds to the different bound exciton levels
. only optically active exciton levels ( whose relative wavefunction have a non - zero amplitude at zero relative distance ) will be considered here .
the exciton hamiltonian can be written approximately as , @xmath37 here , @xmath38 is the exciton energy measured with respect to the ground state consisting of a filled valence band and an empty conduction band .
the quantized radiation field is @xcite , @xmath39 \frac{e^{i\vec{q}.\vec{r}}}{\sqrt{v } } \nonumber \\ & = & \sum_{\vec{q},j } \sqrt{\frac{\hbar}{2\epsilon_{o } \omega_{q } } } \vec{r}_{\vec{q},j } \frac{e^{i\vec{q}.\vec{r}}}{\sqrt{v } } \label{eq : field}\end{aligned}\ ] ] here , @xmath40 for @xmath41 are the field polarization vectors and @xmath42 is the field destruction operator for a mode with wavevector @xmath43 and frequency @xmath44 .
we assume a tmd monolayer occupying the @xmath45 plane located at @xmath46 along the @xmath46-axis .
the interaction between the electrons and photons is given by the hamiltonian , @xmath47 where , @xmath48 and @xmath49^{\dagger}$ ] .
we have expressed the field wavevector @xmath43 in terms of the in - plane component , @xmath50 , and the out - of - plane component , @xmath51 .
excitons belong to a particular valley ( @xmath1 or @xmath2 ) are not the eigenstates in the presence of long - range dipole - dipole interactions @xcite .
it is therefore convenient to switch excitons basis .
transverse and longitudinal exciton states , which are superspositions of exciton states belonging to the two valleys , are a much better choice .
transverse exciton exciton states couple with only te polarized ( or s - polarized ) radiation and longitudinal exciton states couple with only tm polarized ( or p - polarized ) radiation @xcite .
we define creation operators for transverse ( @xmath52 ) and longitudinal ( @xmath53 ) exciton states as follows , @xmath54 \nonumber \\
b^{\dagger}_{\vec{q}_{\parallel},t,\alpha } & = & \frac{i}{\sqrt{2 } } \left [ e^{-i\phi_{\vec{q}_{\parallel } } } b^{\dagger}_{\vec{q}_{\parallel},\tau=1,\alpha } - e^{i\phi_{\vec{q}_{\parallel } } } b^{\dagger}_{\vec{q}_{\parallel},\tau=-1,\alpha } \right ] \nonumber \\\end{aligned}\ ] ] we also define exciton operator @xmath55 and photon operator @xmath56 as , @xmath57 the optical matrix element can be expressed in terms of @xmath58 , @xmath59 using these definitions , the exciton - photon hamiltonian becomes , @xmath60 where , @xmath61 the basis transformation decouples the longitudinal and transverse exciton - polaritons . in what follows we will assume that @xmath62 is real . if it had a phase , it could be absorbed in @xmath63 used to define the operators @xmath64 above .
the interaction hamiltonian used above ignores the quadratic vector potential terms @xcite .
following girlanda et al .
@xcite , the form of these extra term is found to be , @xmath65 \nonumber \\ & & \times \left [ \frac{e}{m_{o } } \sum_{\vec{k } } \delta_{\vec{k}_{\parallel},\vec{q } } \sqrt{\frac{a\hbar}{v \epsilon_{o } \omega_k } } e^{-ik_{z}z } \frac{|\chi_{ex}(0,\vec{k}_{\parallel})|}{e_{ex,\alpha}(\vec{k}_{\parallel } ) } \frac{|k_{z}|}{k } r_{-\vec{k},l } \right ] \nonumber \\ & & \hat{h}'_{t } = \sum_{\vec{q},\alpha } e_{ex,\alpha } ( \vec{q } ) \nonumber \\ & & \times \left [ \frac{e}{m_{o } } \sum_{\vec{q } } \delta_{\vec{q}_{\parallel},\vec{q } } \sqrt{\frac{a\hbar}{v \epsilon_{o } \omega_q } } e^{iq_{z}z } \frac{|\chi_{ex}(0,\vec{q}_{\parallel})|}{e_{ex,\alpha}(\vec{q}_{\parallel } ) } r_{\vec{q},t } \right ] \nonumber \\ & & \times \left [ \frac{e}{m_{o } } \sum_{\vec{k } } \delta_{\vec{k}_{\parallel},\vec{q } } \sqrt{\frac{a\hbar}{v \epsilon_{o } \omega_k } } e^{-ik_{z}z } \frac{|\chi_{ex}(0,\vec{k}_{\parallel})|}{e_{ex,\alpha}(\vec{k}_{\parallel } ) } r_{-\vec{k},t } \right ] \nonumber \\
\end{aligned}\ ] ] these terms must be added to the interaction hamiltonian when expressions for the exciton self - energies are calculated @xcite .
we define the exciton green s function @xmath66 as follows @xcite , @xmath67 the angular brackets indicate averaging with respect to an ensemble of excitons . in the frequency domain
, @xmath68 here , @xmath69 is the exciton spectral density function and @xmath70 is the exciton occupation factor and equals the bose - einstein factor in thermal equilibrium .
most other exciton green s functions can be obtained from the spectral density function @xcite , which incorporates effects due to exciton - photon interaction as well as intra - layer exciton scattering .
the average exciton number is , @xmath71 the spectral density functions satisfy the sum rule , @xmath72 the energy dispersions @xmath73 and the spectral density functions of the transverse and longitudinal excitons in the presence of exciton - photon interaction can be found from the corresponding retarded green s functions @xcite , as shown in the appendix .
a convenient phenomenological choice for @xmath74 is a lorentzian , @xmath75 the fwhm exciton linewidth is @xmath76 .
the one instance where the above simple lorentzian form does not work well is in the case of the transverse excitons right when the spectral weight shifts between the two branches of the polariton dispersion when moving from inside the light cone to outside the light cone ( see the appendix ) .
we consider two parallel ( not necessarily identical ) electronically decoupled ( but electromagnetically coupled ) tmd monolayers , labeled @xmath77 and @xmath78 , located at @xmath79 and @xmath80 , respectively , as shown in the figure [ fig : fig1 ] .
we assume that the exciton intralayer scattering and dephasing rates are fast so that energy transfer dynamics can be described as a simple decay .
the case where exciton scattering and dephasing rates are slow and coherent dynamics are important is discussed later in this paper in section [ sec : coh ] .
we calculate the average of the rate of change of the number of excitons with in - plane momentum @xmath24 in layer @xmath77 as result of electromagnetic coupling to layer @xmath78 .
the desired heisenberg operator is , @xmath81\end{aligned}\ ] ] note that the layer index ( @xmath77 or @xmath78 ) is added to the subscripts .
we calculate the average of the operator @xmath82 using the non - equilibrium green s function technique @xcite , @xmath83 \rangle\end{aligned}\ ] ] here , @xmath84 stands for operator contour ordering along the keldysh contour @xmath85 that runs from time @xmath86 to @xmath87 and back @xcite .
the angled brackets stand for averaging with respect to the initial density matrix at time @xmath86 @xcite .
the above expression can be evaluated in terms of green s functions using standard perturbation techniques @xcite .
the lowest order non - zero terms in the perturbative expansion above give the rate of decrease of the exciton number due to spontaneous emission into free - space , as discussed by wang et al .
the terms relevant to the present discussion correspond to the feynman diagrams depicted in figure [ fig : fig2 ] which represent energy transfer between the excitons in the two layers .
the bare exciton green s functions of each layer in figure [ fig : fig2 ] are dressed from , ( i ) intra - layer photon interactions ( or intra - layer long - range dipole - dipole interactions ) , as discussed by wang et al .
@xcite , and from ( ii ) intra - layer interactions responsible for exciton scattering assuming these interactions are included in the hamiltonian .
the final result for the energy transfer rate @xmath88 can be written in terms of the spectral density functions of the excitons in the two layers .
for the transverse excitons we get , @xmath89 \label{eq : maint}\end{aligned}\ ] ] and for the longitudinal excitons we obtain , @xmath90 \label{eq : mainl}\end{aligned}\ ] ] note that the expressions above are valid for @xmath91 ( radiative transfer ) as well as for @xmath92 ( non - radiative transfer ) provided in the latter case the replacement @xmath93 is made .
the above expressions represent the main results of this work .
the following points regarding the expressions above need to be noted , 1 .
the rate of energy transfer depends on the overlap of the spectral density functions of the excitons in the two layers . 2 .
outside the light cone , when @xmath92 , the energy transfer is mediated via evanescent fields and the rate of transfer decreases exponentially with interlayer separation @xmath94 as @xmath95 .
the energy transfer rates depend inversely on the exciton linewidth ( and , therefore , the exciton intralayer scattering rates ) via the exciton spectral density functions .
the energy transfer rates depend on the dielectric constants ( or the refractive indices ) of the media surrounding the two monolayers . in the simple case when the media on either side of the layers and also in between the layers have the same dispersionless refractive index @xmath96 , the impedance @xmath97 and the speed of light @xmath85 in the above expressions get replaced by @xmath98 and @xmath99 , respectively .
5 . since the exciton state belonging to one valley can be considered a superposition of transverse and longitudinal exciton states , its energy transfer rate will be the average of the energy transfer rates for the transverse and longitudinal excitons . 6 . in the static limit , @xmath100 , the expressions for the energy transfer rates obtained for the longitudinal excitons have the same form as those obtained previously for quantum well excitons using the static dipole - dipole interaction hamiltonian @xcite , which is to be expected .
for numerical evaluations of the results , we assume two identical and parallel mos@xmath0 layers at a distance @xmath94 .
the electronic and optical parameter values used for mos@xmath0 monolayers are the same as those given previously @xcite .
we first calculate the energy dispersions for the lowest energy @xmath101 longitudinal and transverse excitons ( as described in the appendix ) , and then use these to compute the energy transfer rates . in numerical calculations , a momentum - independent scattering - limited fwhm exciton linewidth of @xmath10230 mev is assumed .
figure [ fig : fig3 ] shows the calculated energy transfer times for both longitudinal and transverse excitons as a function of the exciton in - plane momentum @xmath103 for different values of the interlayer separation @xmath94 ( @xmath94=5 , 10 , 25 and 50 nm ) .
the value of the momentum @xmath104 , defined by @xmath105 , is @xmath1029.6 1/@xmath106 m ( see the appendix ) . for @xmath107 ( inside the light cone ) ,
the energy transfer is via the radiative mechanism and the energy transfer time is almost independent of @xmath103 . when @xmath108 , the cusps in the energy transfer times follow the trends in the radiative lifetimes and the optical conductivities of longitudinal and transverse excitons ( see the appendix ) .
when @xmath109 ( outside the light cone ) , the energy transfer is via the non - radiative mechanism ( via evanescent waves ) . in the case of the transverse excitons , the energy transfer rate decreases ( and the energy transfer time increases ) exponentially with the product @xmath110 ( for large @xmath103 ) . in the case of the longitudinal excitons ,
the energy transfer time first decreases with @xmath103 because the in - plane component of the evanescent radiation also increases with @xmath103 , and then the energy transfer time increases exponentially with the product @xmath110 ( for large @xmath103 ) .
note that for interlayer separations @xmath94 less than 10 nm , the energy transfer times for the longitudinal excitons can be in the hundreds of femtoseconds range or even smaller .
these results show the efficacy of the exciton energy transfer mechanism in tmd monolayers .
simple expressions for the energy transfer times @xmath111 can be found for two identical tmd layers when @xmath112 and @xmath109 ( i.e. away from @xmath113 ) assuming lorentzian spectral density functions , @xmath114 where @xmath115 , @xmath116 , and @xmath117 when @xmath118 outside the light cone . for @xmath112 ( radiative energy transfer )
the above expressions can be written as , @xmath119 here , @xmath120 is the spontaneous emission radiative lifetime of excitons in a single tmd layer @xcite ( see the appendix ) . in the case of two mos@xmath0 monolayers , @xmath121 is around 200 fs ( for @xmath101 excitons ) @xcite .
assuming fwhm exciton linewidths of @xmath10230 mev @xcite and @xmath10210 mev , the radiative energy transfer times are @xmath1023.6 ps and @xmath1021.2 ps , respectively .
although these times seem short , the radiative energy transfer process will not be very efficient ( efficiency less than @xmath10215% ) because the spontaneous emission time is also very short and only a small fraction of the photons emitted from one layer get absorbed by the other layer . despite the fast energy transfer rates for the longitudinal excitons when
@xmath122 , their contribution to the energy transfer process is expected to be limited by the relatively small density of the longitudinal excitons in a thermal ensemble of excitons since @xmath123 for @xmath122 ( see the appendix ) .
figure [ fig : fig4a ] shows the calculated average energy transfer times for a thermal ensemble of ( @xmath101 ) longitudinal and transverse excitons as a function of the exciton temperature .
the exciton density is assumed to be dilute enough such that the exciton chemical potential is less than the lowest exciton energy level by at least several @xmath124 .
the exciton fwhm linewidth is assumed to be momentum - independent and equal to 30 mev .
figure [ fig : fig4b ] shows the same results assuming that the exciton fwhm linewidth is 10 mev .
the results show that when the interlayer separation is small then as the temperature increases , and the density of the longitudinal excitons also increases relative to the transverse excitons , the average energy transfer time decreases . however , when the interlayer separation is large , and the energy transfer is by excitons with only small momenta ( @xmath125 ) , an increase of the temperature results in an increase of the energy transfer time because the exciton thermal distribution spills to larger momenta .
these results also show that the average energy transfer times scale with the exciton fwhm linewidth and can be shorter than a picosecond for interlayer separations smaller than 10 nm and exciton linewidths narrower than 10 mev .
in the previous section we assumed that the intralayer exciton scattering and dephasing rates are slow and energy transfer dynamics can be described as a simple decay . here
we quantify this notion and also discuss coherent interlayer energy transfer dynamics .
first , we evaluate corrections to the exciton dispersions as a result of interlayer radiative and non - radiative interactions .
we consider two parallel and _ identical _ electronically decoupled ( but electromagnetically coupled ) tmd monolayers , labeled @xmath77 and @xmath78 , located at @xmath79 and @xmath80 , respectively , as shown earlier in figure [ fig : fig1 ] . the analysis is greatly simplified if we define operators for the in - phase ( + exciton ) and out - of - phase ( - exciton ) excitons in the two layers as follows @xcite , @xmath126 the + and - excitons have their dipole moments in - phase and out - of - phase , respectively .
the retarded green s functions and self - energies for the + and - excitons can be found using the methods described in the appendix , @xmath127 } { ( \hbar\omega)^{2}-e_{ex,\alpha}(\vec{q}_{\parallel})^{2 } - 2\frac{(\hbar \omega)^{2}}{e_{ex,\alpha}(\vec{q}_{\parallel } ) } \sigma^{or}_{\vec{q}_{\parallel},l / t,\pm,\alpha}(\omega ) } \nonumber \\ & \approx & \frac{2e_{ex,\alpha}(\vec{q}_{\parallel})}{(\hbar\omega)^{2}-e_{ex,\alpha}(\vec{q}_{\parallel})^{2 } - 2\frac{(\hbar \omega)^{2}}{e_{ex,\alpha}(\vec{q}_{\parallel } ) } \sigma^{or}_{\vec{q}_{\parallel},l / t,\pm,\alpha}(\omega ) } \nonumber \\\end{aligned}\ ] ] here , @xmath128\ ] ] @xmath129 is the retarded self - energy for excitons in a single tmd layer and its expression was given previously @xcite ( also see the appendix ) .
the expression above is valid for @xmath130 as well as for @xmath131 provided in the latter case the replacement @xmath132 is made .
it is clear from the above expression for the self - energy that in the limit @xmath133 the out - of - phase - excitons do not radiate whereas the radiative rates of the in - phase + excitons are twice as fast as those of excitons in a single tmd layer .
the energy splitting between the + and - excitons due to interlayer interactions can be estimated as , @xmath134_{\hbar \omega = e_{ex , l / t,\alpha}(\vec{q}_{\parallel } ) } \nonumber \\\end{aligned}\ ] ] since an exciton state in any one of the two tmd layers can be considered a superposition of the in - phase and the out - of - phase exciton states , if the energy splitting @xmath135 due to interlayer interactions is much larger than the exciton linewidth due to intralayer scattering and dephasing then coherent energy oscillations between exciton states in the two layers are expected at the frequency @xmath136 , and energy transfer between the layers can not be described as a simple decay of energy from one layer to the other .
the calculated energy splittings between the + and - excitons are plotted in figure [ fig : figa2 ] for the lowest energy ( @xmath101 ) exciton state in two suspended mos@xmath0 monolayers as a function of the in - plane momentum @xmath103 for different values of the interlayer spacing @xmath94 ( @xmath94=5 , 10 , 25 and 50 nm ) . inside the light cone , the energy splittings are small ( less than 1 - 2 mev ) for all values of @xmath94 considered .
the energy splittings are smaller than the natural linewidth of excitons in a single mos@xmath0 layer due to radiative decay .
consequently , coherent energy oscillations are not expected for excitons inside the light cone . outside the light cone , the energy splittings for
the longitudinal excitons become large reaching values larger than @xmath10225 mev for @xmath94 less than 5 nm
. however , these large energy splittings occur at large values of the exciton momenta where exciton intralayer scattering is also expected to be fast and the condition for coherent oscillations might be difficult to meet .
if , outside the light cone , exciton scattering rates are small then the coherent dynamics of the average exciton layer number @xmath137 can be modeled by the following equation similar to that of a damped simple harmonic oscillator , @xmath138 \langle n_{\vec{q},l / t , a / b,\alpha}(t ) \rangle = \nonumber \\ & & \frac{1}{2 } \frac{\delta^{2}_{\vec{q},l / t,\alpha}}{\hbar^{2}}\end{aligned}\ ] ] here , @xmath115 is related to the exciton scattering and dephasing rate and the appropriate boundary conditions are , @xmath139 if @xmath140 , then the above equation predicts damped oscillations at the frequency @xmath141 .
on the other hand , if @xmath142 , then the above equation gives a simple exponential decay of energy from layer @xmath77 to layer @xmath78 at the rate @xmath143 .
this latter result is almost exactly what was obtained earlier in ( [ eq : simple1 ] ) and ( [ eq : simple2 ] ) . in the absence of quantitative models or experimental data for exciton scattering in tmds
it is difficult to say if coherent energy oscillations are possible in tmds . in the case of localized excitons ,
discussed next , momentum spread due to localization also contributes to the decoherence of the oscillations .
the analysis in the previous sections shows that the longitudinal excitons with momenta @xmath103 in the @xmath144 range have the shortest energy transfer times but the density of such excitons is relatively small in a thermal ensemble at low temperatures thus limiting the average energy transfer rates .
localized excitons , whose wavefunction is a superposition of exciton states of different momenta , could overcome these limitations and exhibit fast energy transfer rates even at low temperatures .
we consider an initial exciton state in layer @xmath77 that is localized in space in a region of size @xmath145 .
we assume that the exciton wavefunction for the center of mass coordinate in real and fourier spaces is , @xmath146 a localized exciton state can be constructed from the ground state @xmath147 , corresponding to a filled valence band and an empty conduction band , as follows @xcite , @xmath148 with hwhm linewidth of @xmath149 and centered at the energy @xmath150 , the rate of energy transfer to free excitons in layer @xmath78 for the transverse case is found to be , @xmath151 \end{aligned}\ ] ] and for the longitudinal case we obtain , @xmath152 \end{aligned}\ ] ] again note that the expressions above are valid for @xmath92 ( non - radiative transfer ) provided the replacement @xmath93 is made .
simpler expressions can be obtained in some special cases .
suppose there exists an exciton state @xmath153 with momentum @xmath154 and a corresponding energy @xmath155 in layer @xmath78 which satisfies the energy conservation relation @xmath156 .
if the exciton in layer @xmath77 is strongly localized such that the energy spread of the free excitons in layer @xmath78 corresponding to the momentum spread of the localized exciton in layer @xmath77 is much greater than @xmath157 , coherent oscillations will not be possible .
if @xmath158 and @xmath159 ( i.e. away from @xmath160 ) then , assuming lorentzian spectral density functions , the energy transfer times can be written as , @xmath161 where @xmath162 , and @xmath163 when @xmath164 outside the light cone . here , @xmath165 is the density of states of free excitons in layer @xmath78 .
note that in this case the energy transfer times do not depend on the scattering / dephasing rate given by @xmath166 .
assuming @xmath167 , the expressions above differ from the corresponding expressions for free excitons , given earlier in ( [ eq : simple1 ] ) and ( [ eq : simple2 ] ) , by a multiplicative factor of @xmath168 .
for strongly localized excitons , this factor is of the order of unity and therefore energy transfer times for strongly localized @xmath101 excitons in mos@xmath0 , when plotted as a function of @xmath154 , are expected to be similar to those appearing in figure ( [ fig : fig3 ] ) .
we now consider the case in which the final exciton state in layer @xmath78 is also localized .
the center of mass wavefunctions of the excitons in layer @xmath77 and @xmath78 in real space are centered at the in - plane vectors @xmath169 and @xmath170 , respectively , and in momentum space these wavefunctions are @xmath171 and @xmath172 , respectively , where @xmath173 are as given earlier in ( [ eq : psicom ] ) .
the vector @xmath174 connects the center of the exciton states , @xmath175 .
the rate of energy transfer for the transverse case is found to be , @xmath176 \end{aligned}\ ] ] and for the longitudinal case we get , @xmath177 \end{aligned}\ ] ] an interesting case is that of extremely localized excitons for which @xmath178 and @xmath179 .
assuming wavevector independent values of @xmath180 and using the results , @xmath181 the above expressions for @xmath88 , in the limit @xmath182 , give a @xmath183 dependence of the energy transfer rate for the transverse case and a @xmath184 dependence for the longitudinal case .
it is satisfying to note that the former result corresponds to the classical inverse square law for radiative energy transfer and the latter corresponds to the standard forster s result for non - radiative energy transfer via dipole - dipole interaction @xcite . in the longitudinal case ,
if one integrates the energy transfer rate over the in - plane position @xmath170 of the final exciton state , then the total energy transfer rate will scale as @xmath185 with the interlayer separation .
in many cases of practical interest where the optical bandgaps and/or the exciton binding energies in two different tmd monolayers are very different , the energy transfer can be from the excitons in the wider bandgap tmd layer to the free electron - hole pairs in the narrower bandgap tmd layer .
for example , this could be the case in two parallel monolayers of mos@xmath0 and mote@xmath0 @xcite .
we assume that an exciton in layer @xmath77 with momentum @xmath24 decays into a free electron - hole pair in layer @xmath78 .
we assume that the energy emission and absorption remains close to the conduction and valence band edges at @xmath1 and @xmath2 valleys so that the standard optical selection rules are not violated .
since a free electron - hole pair can be considered an unbound exciton , the expressions for the rate of energy transfer between excitons given in the main text are also valid for energy transfer between excitons in layer @xmath77 and free electron - hole pairs in layer @xmath78 provided the relative exciton wavefunction in layer @xmath78 is assumed to be a plane wave , the exciton energy dispersion in layer @xmath78 is replaced by that of a free electron - hole pair , and the summation over the final exciton states ( i.e. over @xmath153 ) in layer @xmath78 is replaced by a phase space integral over the relative wavevector .
we assume that the free electron - hole pair in layer @xmath78 is described by the spectral density function @xmath186 with hwhm linewidth @xmath187 . here
, @xmath20 is the relative momentum of the electron - hole pair and @xmath24 is the center of mass momentum and the spectral density function is centered at the energy @xmath188 .
@xmath189 is the reduced electron - hole mass in layer @xmath78 , @xmath190 is the exciton mass in layer @xmath78 , and @xmath191 is the bandgap of layer @xmath78 .
for the case of the transverse excitons in layer @xmath77 we get , @xmath192 \right .
\nonumber \\ & & \left .
- f_{c}(\vec{k } + \lambda_{e , b}\vec{q } ) \left [ 1 - f_{v}(\vec{k } - \lambda_{h , b}\vec{q } ) \right ] \right\ } \end{aligned}\ ] ] and for the longitudinal excitons we obtain , @xmath193 - f_{c}(\vec{k } + \lambda_{e , b}\vec{q } ) \left [ 1 - f_{v}(\vec{k } - \lambda_{h , b}\vec{q } ) \right ] \right\ } \nonumber \\\end{aligned}\ ] ] @xmath194 are the layer @xmath78 conduction and valence band electron occupation factors , and @xmath195 is related to the interband momentum matrix element in layer @xmath78 by the expression , @xmath196 } \label{eq : chi2}\end{aligned}\ ] ] we define the momentum @xmath197 and the energy @xmath155 by the energy conservation relations , @xmath198 and @xmath199 .
given the number of possible final states ( corresponding to different values of @xmath20 ) and the fast electron and hole scattering rates in tmds , we do nt expect coherent oscillations .
assuming that the narrower bandgap material is in the ground state with a full valence band and an empty conduction band , the energy transfer times can be expressed as , @xmath200 where @xmath201 , and @xmath202 when @xmath203 outside the light cone . here , @xmath204 is the joint density of states for the creation of free electron - hole pairs ( per valley / spin ) in layer @xmath78 . again
note that the energy transfer times do not depend on the scattering / dephasing rate .
the expressions above differ from the corresponding expressions for free excitons , given earlier in ( [ eq : simple1 ] ) and ( [ eq : simple2 ] ) , by a multiplicative factor of @xmath205 .
this factor is expected to be much smaller than unity for most tmd pairs .
still , the energy transfer times for longitudinal excitons in layer @xmath77 can range from a picosecond to tens of picoseconds ( depending on the initial exciton momentum @xmath24 and the magnitude of the joint density of states for free electron - hole pair creation in layer @xmath78 ) for interlayer spacings smaller than 10 nm . as an example
, we consider the case of mos@xmath0 and mote@xmath0 layers .
the exciton energy in mos@xmath0 is @xmath1021.9 ev and the quasiparticle bandgap @xmath191 of mote@xmath0 is @xmath1021.7 ev . assuming an exciton fwhm of 30 mev in mos@xmath0 and parameters of mote@xmath0 as given in the literature @xcite , the value of @xmath206 is found to be in the 0.07 - 0.08 range ( for different momenta of the initial exciton state in mos@xmath0 ) .
therefore , the energy transfer times between ( @xmath101 ) excitons in mos@xmath0 and free electron - hole pairs in mote@xmath0 will be approximately 12 - 14 times those given in figure [ fig : fig3 ] for the case of @xmath101 excitons in two identical mos@xmath0 layers .
in this paper we presented results on the energy transfer rates between excitons in 2d tmd monolayers .
the results show that the energy transfer rates can be very fast .
exciton energy transfer can potentially be used to design novel optoelectronic devices with tmd monolayers . to date , the authors are not aware of an experimental studies in this area .
however , the results presented in this paper can be easily tested experimentally . the theory presented in this paper has certain limitations and care needs to exercised when interpreting the results and comparing these results with experiments : 1 .
the technique used in this paper is valid provided the exciton optical conductivity @xcite @xmath207 satisfies @xmath208 , which is typically the case in tmds @xcite .
if @xmath209 , the vacuum field in the vicinity of the tmd layers will get modified and the expression for the field given in ( [ eq : field ] ) will no longer be valid .
the field would then need to be quantized in the presence of the tmd layers @xcite , a task beyond the scope of this paper .
2 . in plotting all the results ,
a momentum - independent exciton fwhm linewidth was used .
exciton intralayer scattering and dephasing rates are expected to depend on the exciton momentum . since the exciton energy transfer rates depend , in most cases
, on the exciton scattering rates , a quantitative theory or experimental data for momentum - dependent exciton scattering rate in tmds is needed for a better understanding of the dependence of the energy transfer rates on exciton momenta .
it is well known that radiation emission and absorption rates are affected by the presence of dielectric interfaces @xcite .
most experiments on tmd layers are performed with the layers placed on dielectric substrates .
the influence of nearby dielectrics would need to be taken into account in comparing theory with experiments .
the authors would like to acknowledge helpful discussions with jared strait , paul l. mceuen and michael g. spencer , and support from ccmr under nsf grant number dmr-1120296 , afosr - muri under grant number fa9550 - 09 - 1 - 0705 , and onr under grant number n00014 - 12 - 1 - 0072 .
in this section , we calculate the exciton self - energy in tmd monolayers . for the sake of simplicity
we will assume that there is only one significant exciton level labeled by @xmath36 .
the non - interacting ( bare ) retarded green s function for the exciton field is @xcite , @xmath210\rangle\ ] ] the exciton operator @xmath211 was defined earlier in ( [ eq : fields ] ) . in the fourier domain
the green s function is , @xmath212 the non - interacting ( bare ) retarded radiation green s function is defined as , @xmath213\rangle \nonumber \\ d^{or}_{\vec{q},l / t}(\omega ) & = & \frac{2\hbar \omega_{q}}{(\hbar\omega)^{2 } - ( \hbar \omega_{q})^{2 } + i\eta } \nonumber \\\end{aligned}\ ] ] first , we dress the radiation green s function with term @xmath214 in the exciton - photon interaction hamiltonian that is quadratic in the vector potential ( see section ( [ sec : quad ] ) .
the resulting dressed green s functions @xmath215 can be expressed in a form that will be useful later , @xmath216 the denominator on the right hand side in the above equations is generally small and may be neglected .
but it plays an important role when off - shell exciton self - energies are desired , as shown below .
the green s functions for the radiation field are gauge - dependent .
it is convenient to choose the temporal gauge in which the scalar potential is set equal to zero ( and need not be taken into account separately ) @xcite .
henceforth , all results will be given for the temporal gauge .
the dyson equation for the exciton green s function is , @xmath217 \end{aligned}\ ] ] where the retarded self - energies are found to be , @xmath218 the dressed exciton green s functions become , @xmath219 } { ( \hbar\omega)^{2}-e_{ex,\alpha}(\vec{q}_{\parallel})^{2 } - 2\frac{(\hbar \omega)^{2}}{e_{ex,\alpha}(\vec{q}_{\parallel } ) } \sigma^{or}_{\vec{q}_{\parallel},l / t,\alpha}(\omega ) } \nonumber \\ & \approx & \frac{2e_{ex,\alpha}(\vec{q}_{\parallel})}{(\hbar\omega)^{2}-e_{ex,\alpha}(\vec{q}_{\parallel})^{2 } - 2\frac{(\hbar \omega)^{2}}{e_{ex,\alpha}(\vec{q}_{\parallel } ) } \sigma^{or}_{\vec{q}_{\parallel},l / t,\alpha}(\omega ) } \nonumber \\\end{aligned}\ ] ] here , @xmath129 is the same as @xmath220 given in ( [ eq : self ] ) above except that the bare radiation green s function @xmath221 is used in place of @xmath215 .
finally the exciton spectral density function @xmath222 can be related to the retarded green s function as follows , @xmath223 when @xmath130 ( inside the light cone ) , both the self - energies @xmath220 have a vanishingly small real part and a large magnitude of the imaginary part .
the latter corresponds to the radiative lifetime of the exciton .
the situation is reversed when @xmath224 and then the magnitude of the real part of the self - energy becomes large and the imaginary part vanishes .
therefore , when @xmath225 ( inside the light cone ) one can ignore corrections to the exciton dispersion , and the radiative lifetime of the exciton can be related to the imaginary part of the self - energy evaluated on the shell , @xmath226 and , we get @xcite , @xmath227 when @xmath224 and the self - energies are real , we get @xcite , @xmath228 note that the corrections obtained from @xmath214 , which was quadratic in the vector potential , are important for obtaining the correct expressions for the exciton self - energies far off - shell .
the exciton self - energy expressions given above can be used to obtain energy dispersions , @xmath229 , for the longitudinal and transverse excitons .
the results are shown in figure [ fig : figa1 ] for the lowest energy ( @xmath101 ) exciton state in a suspended mos@xmath0 monolayer .
we define @xmath104 by the equation @xmath230 .
the dispersion for the transverse exciton consists of two curves ( corresponding to the distinct poles of the green s function ) .
the upper curve has all the spectral weight for @xmath231 and the lower curve gets all the weight when @xmath232 , and the spectral weight shifts from the upper curve to the lower curve around @xmath233 .
the dispersion for the longitudinal exciton consists of only a single curve and the energy dispersion is linear in @xmath103 for @xmath232 . for @xmath122 , @xmath123 .
therefore , in a thermal ensemble of excitons , transverse exciton density is expected to exceed the longitudinal exciton density .
note that there are negligibly small corrections to the exciton dispersion within the light cone for both the transverse and longitudinal excitons .
k. f. mak , c. lee , j. hone , j. shan , and t. f. heinz , phys .
105 , 136805 ( 2010 ) .
k. f. mak , k. he , j. shan , and t. f. heinz , nat .
7 , 494 ( 2012 ) .
j. s. ross , s. wu , h. yu , n. j. ghimire , a. m. jones , g. aivazian , j. yan , d. g. mandrus , d. xiao , w. yao , and x. xu , nat . comm . 4 , 1474 ( 2013 ) .
t. c. berkelbach , m. s. hybertsen , and d. r. reichman , phys .
b 88 , 045318 ( 2013 ) .
a. chernikov , t. c. berkelbach , h. m. hill , a. rigosi , y. li , o. b. aslan , d. r. reichman , m. s. hybertsen , t. f. heinz , phys .
lett . , 113 , 076802 ( 2014 ) . c. zhang , h. wang , w. chan , c. manolatou , f. rana , phys .
b , 89 , 205436 ( 2014 ) . h. wang , c. zhang , w. chan , c. manolatou , s. tiwari , f. rana , phys . rev .
b , 93 , 045407 ( 2016 ) .
x. liu , t. galfsky , z. sun , f. xia , e. lin , y. lee , s. kna - cohen , v. m. menon , nature photonics , 9 , 30 ( 2015 ) .
m. i. vasilevskiy , d. g. santiago - prez , c. trallero - giner , n. m. r. peres , a. kavokin , phys .
b 92 , 245435 ( 2015 ) .
y. n. gartstein , x. li , c. zhang , phys .
b , 92 , 075445 ( 2015 ) . g. moody , c. k. dass , k. hao , c .- h .
chen , l .- j .
li , a. singh , k. tran , g. clark , x. xu , g. berghuser , e. malic , a. knorr , x. li , nature communications , 6 , 8315 ( 2015 ) . c. poellmann , p. steinleitner , u. leierseder , p. nagler , g. plechinger , m. porer , r. bratschitsch , c. schller , t. korn , r. huber , nature materials , 14 , 889 ( 2015 ) .
x. marie , b. urbaszek , nature materials , 14 , 860 ( 2015 ) .
h. fang , c. battaglia , c. carraro , s. nemsak , b. ozdol , j. seuk kang , h. a. bechtel , s. b. desai , f. kronast , a. a. unal , g. conti , c. conlon , g. k. palsson , m. c. martin , a. m. minor , c. s. fadley , e. yablonovitch , r. maboudian , a. javey , pnas , 111 , 6198 ( 2014 ) .
x. hong , j. kim , s. shi , y. zhang , c. jin , y. sun , s. tongay , j. wu , y. zhang , f. wang , nature nanotechnology , 9 , 682686 ( 2014 ) .
a. f. rigosi , h. m. hill , y. li , a. chernikov , and t. f. heinz , nano letters , 15 , 5033 ( 2015 ) .
h. wang , j. h. strait , c. zhang , w. chen , c. manolatou , s. tiwari , f. rana , phys .
b 91 , 165411 ( 2015 ) .
t. cheiwchanchamnangij and w. r. l. lambrecht , phys . rev .
b 85 , 205302 ( 2012 ) .
a. kormanyos , v. zolyomi , n. d. drummond , p. rakyta , g. burkard and v. i. falko , phys . rev .
b 88 , 045416 ( 2013 ) .
d. xiao , gui - bin liu , w. feng , x. xu , and w. yao , phys . rev
108 , 196802 ( 2012 ) .
s. savasta , r. girlanda , solid state communications , 96 , 517 ( 1995 ) .
h. haug , s. w. koch , _ quantum theory of the optical and electronic properties of semiconductors _ , world scientific publishing , singapore ( 1990 ) .
g. d. mahan , _ many particle physics _ , springer , ny ( 2000 ) . c. cohen - tannoudji , j. dupont - roc , g. grynberg , _ photons and atoms : introduction to quantum electrodynamics _ , wiley ,
ny ( 1989 ) .
d. l. andrews , d. s. bradshaw , european journal of physics , 25 , 845 ( 2004 ) .
t. forster , annals of physics , 437 , 55 ( 1948 ) .
a. tomita , j. shah , phys .
b , 53 , 10793 ( 1996 ) .
s. k. lyo , phys .
b , 62 , 13641 ( 2000 ) .
h. wang , c. zhang , f. rana , nano letters , 15 , 339 ( 2015 ) .
e. guadagnini , il nuovo cimento , 57a , 294 ( 1980 ) .
j. yang , t. lu , y. w. myint , j. pei , d. macdonald , j. zheng , y. lu , acs nano , 9 , 6603 ( 2015 ) .
r. matloob , r. loudon , s. m. barnett , j. jeffers , phys .
a 52 , 4823 ( 1995 ) .
h. khosravi , r. loudon , proc .
london a , 433 , 337 ( 1991 ) .
d. s. citrin , phys .
b , 49 , 1943 ( 1994 ) . | we present results on the rates of interlayer energy transfer between excitons in two - dimensional transition metal dichalcogenides ( tmds ) .
we consider both radiative ( mediated by real photons ) and non - radiative ( mediated by virtual photons ) mechanisms of energy transfer using a unified green s function approach that takes into account modification of the exciton energy dispersions as a result of interactions .
the large optical oscillator strengths associated with excitons in tmds result in very fast energy transfer rates .
the energy transfer times depend on the exciton momentum , exciton linewidth , and the interlayer separation and can range from values less than 100 femtoseconds to more than tens of picoseconds . whereas inside the light cone the energy transfer rates of longitudinal and transverse excitons are comparable , outside the light cone
the energy transfer rates of longitudinal excitons far exceed those of transverse excitons .
average energy transfer times for a thermal ensemble of longitudinal and transverse excitons is temperature dependent and can be smaller than a picosecond at room temperature for interlayer separations smaller than 10 nm .
energy transfer times of localized excitons range from values less than a picosecond to several tens of picoseconds .
when the exciton scattering and dephasing rates are small , energy transfer dynamics exhibit coherent oscillations .
our results show that electromagnetic interlayer energy transfer can be an efficient mechanism for energy exchange between tmd monolayers . |
following eq . 6 in the main text ,
the time - ordered bare susceptibility is a @xmath49 matrix @xmath110 defined as , ^0_i , j(,)=_i(,)^_j(,0 ) , where the density operators are defined in eq .
3 in the main text .
first , we perform the wicks decomposition and then convert the @xmath111 operators to the band basis , c_r(,)=_u_r_()a _ ( , ) .
after we perform a fourier transformation to matsubara space , sum over the internal matsubara frequency , and do the analytic continuation , we obtain the final expression of retarded bare susceptibility as | we investigate the charge excitations of a weyl semimetal in the axionic charge density wave ( axionic cdw ) state .
while it has been shown that the topological response ( anomalous hall conductivity ) is protected against the cdw state , we find that the long wavelength plasmon excitation is radically influenced by the dynamics of the cdw order parameter . in the normal state , we show that an undamped collective mode should exist at @xmath0 if there is an attractive interaction favoring the formation of the cdw state .
the undamped nature of this collective mode is attributed to a gap - like feature in the particle - hole continuum at @xmath0 due to the chirality of the weyl nodes , which is not seen in other materials with cdw instability . in the cdw state ,
the long wavelength plasmon excitations become more dispersive due to the additional interband scattering not allowed in the normal state . moreover , because the translational symmetry is spontaneously broken , umklapp scattering , the process conserving the total momentum only up to @xmath1 with @xmath2 an integer and @xmath3 the ordering wave vector , emerges in the cdw state .
we find that the plasmon excitation couples to the phonon mode of the cdw order via the umklapp scattering , leading to two branches of resonant collective modes observable in the density - density correlation function at @xmath4 and @xmath0 .
based on our analysis , we propose that measuring these resonant plasmon - axion excitations around @xmath4 and @xmath0 by the momentum - resolved electron energy loss spectroscopy ( m - eels ) could serve as a reliable way to detect the axionic cdw state in weyl semimetals .
_ introduction_ the weyl semimetal@xcite is a new gapless state of matter attracting a lot of attention in the past few years due to its rich topological properties.@xcite in the wake of the study on topological insulators@xcite , the search for semimetals exhibiting linear band dispersion near the fermi energy has been one of the most active fields in condensed matter physics .
one typical example is the dirac semimetal@xcite in which the conduction and valence bands become degenerate at certain points in the brillouin zone , namely the dirac nodes , and the low energy single particle states near the dirac nodes can therefore be described by dirac equations with a four - component spinor .
if the dirac semimetal has a zero mass , the dirac equation can be further reduced to two independent sets of weyl equations with two - component spinors . as a result ,
the weyl semimetal is a special case of the dirac semimetal with zero mass , which could be obtained by breaking either the inversion or the time - reversal symmetries in a dirac semimetal .
in addition , it can be proved@xcite that weyl nodes must appear in pairs with opposite chirality if the weyl equations are imposed on a lattice .
therefore , if a crystalline material is a weyl semimetal , the low energy single particle excitations can be described by a minimal model of the weyl fermions around two weyl nodes obeying the weyl equations : h_0= v_fd _ = 1 - _ , [ h0 ] where @xmath5 is the fermi velocity , and @xmath6 is the chirality of the weyl nodes .
very recently , the experimental realization of the weyl semimetal has been found in taas.@xcite the spontaneous symmetry breaking phases in the weyl semimetal induced by the electron - electron correlation have also been investigated.@xcite one intriguing aspect unique to the weyl semimetal is the chiral anomaly accompanying the charge - density - wave ( cdw ) state.@xcite if the pair of the weyl nodes described in eq . [ h0 ] is located at @xmath7 in the brillouin zone , the electron - electron interaction could result in a cdw state with an ordering wavevector @xmath8 .
since in the cdw state the electron occupation number at @xmath9 is different , the chiral symmetry is also broken .
it can be further shown that the phase of the cdw order parameter can be identified as the axion field which couples to the electromagnetic fields and exists only in the presence of the chiral anomaly.@xcite in this paper , we study the charge excitations of the weyl semimetal in the axionic cdw state . in the normal state
, we find an undamped collective mode at @xmath0 if there is an attractive interaction favoring the formation of the cdw state . in the cdw state , we find that the nature of the long wavelength plasmon excitation is dramatically changed .
the phonon mode of the cdw order , or equivalently the collective mode of the axion field , couples to the plasmon excitation via the umklapp process , resulting in two resonant collective modes which can be seen in the density - density correlation function at both @xmath4 and @xmath0 .
this resonant feature does not emerge in the case where the chiral symmetry is broken by the application of a pairing of the electric @xmath10 and the magnetic @xmath11 fields.@xcite we propose that measuring these resonant plasmon - axion excitations around @xmath4 and @xmath0 by the momentum - resolved electron energy loss spectroscopy ( m - eels)@xcite could serve as a reliable way to detect the cdw state as well as the nature of the axion field in a weyl semimetal . _ hamiltonian and formalism _ we start from the effective hamiltonian @xmath12 describing the weyl fermions around two weyl nodes at @xmath9 given in eq.[h0 ] together with electron - electron interactions : h&=&h_0 + h_cdw + h_lc , + h_cdw & = & v_cdwd ( + _ cdw)(--_cdw ) , + h_lc & = & d v_()(- ) . where @xmath13 is the 3d coulomb interaction in momentum space .
it has been shown in previous study that the cdw instability is one of the leading instabilities in the weyl semimetal induced by repulsive interactions@xcite , which justifies using @xmath14 to stduy the cdw fixed point .
the charge density operator can be expanded around the proximity of the weyl nodes , leading to & & ( ) _ ^d c^_r ( + ) c_r()+c^_l ( ) c_l ( - ) , + & & ( --_cdw ) _ ^d c^_l ( - ) c_r ( ) [ den ] where @xmath15 are small momenta compared to @xmath16 , @xmath17 is the cut - off in the integration over the momentum , and @xmath18
. since we are only interested in the long - wavelength plasmon excitations and the charge fluctuations near the cdw ordering wave vector @xmath3 , the integration over @xmath19 is limited to the region of small @xmath20 we perform the standard mean - field theory to decouple @xmath14 with the axionic cdw order parameter defined as @xmath21 , and consequently the mean - field hamiltonian in the cdw state can be written as@xcite @xmath22 , where ( ) & = & ( cc - _ + & ^ + & - - _ - ) , + & = & ( cc m_1 - im_2&0 + 0&m_1 -im_2 ) , , and @xmath23 $ ] .
@xmath24 and @xmath25 are the creation operators near @xmath26 and @xmath27 respectively , and they are related via @xmath28 , where @xmath8 is the cdw ordering wave vector .
@xmath29 can be diagonalized as @xmath30 , where @xmath31 , and @xmath32 $ ] are the corresponding eigenvectors which are related to @xmath33 via a unitary matrix @xmath34 such that @xmath35 .
the cdw order parameter is self - consistently obtained by solving the mean - field equation : = d _
u^*_r,()u_l , n_f(e _ ) , which supports a non - zero order parameter if @xmath36 is negative and smaller than a critical value ( @xmath37 ) . clearly , in the normal state
, the density fluctuations at @xmath38 are independent of those at @xmath19 , thus they do not affect the long wavelength plasmon excitations at all .
if the system is in the cdw state , a non - zero order parameter induces a coupling between the density fluctuations at @xmath19 and @xmath38 .
such a coupling can be described by the umklapp scattering in which the total momentum is conserved only up to @xmath1 with @xmath2 an integer , and significant changes in the properties of the long wavelength plasmon excitations occurs .
note that in principle all the density fluctuations in different umklapp channels ( i.e. , different @xmath2 ) are coupled , but for the demonstration of principle , it is sufficient to consider the two channels given in eq . [ den ] . to take the umklapp process into account
, we introduce a matrix form to represent the density - density correlation function written as @xmath39 , where @xmath40 and @xmath41 refer to @xmath42 and @xmath43 respectively .
in general , @xmath44 refers to the charge excitations inside each weyl nodes , and @xmath45 refers to those between different weyl nodes . performing a generalized rpa which has been widely used in studying the coupling between different collective excitations@xcite , we obtain ^grpa ( , ) = ^-1 where @xmath46 , @xmath47 , and @xmath48 are @xmath49 matrices , ^0_i , j ( , ) & = & - _ , = 1 ^ 4 ^d m_i , j^ , ( , ) + & & ( ) , + [ chi0 ] @xmath50 is the matrix element which is given in the supplementary materials , and @xmath51 is the interaction kernel . for the calculations , we adopt the lorentz cut - off , i.e. , @xmath52 and @xmath53 .
since we only focus on the excitations near the weyl nodes , @xmath54 .
furthermore , we choose @xmath55 and @xmath56 to be the units of momentum and energy in all of our calculations .
( a ) real ( blue solid ) and imaginary ( red dashed ) parts of @xmath57 .
the parameters used are @xmath58 , @xmath59 and @xmath60 .
the particle - hole excitations from the intraband scatterings are greatly suppressed , resulting a gap - like feature in @xmath61 .
the intraband and interband scatterings in @xmath62 and @xmath63 are schematically illustrated in ( b ) and ( c ) respectively.,title="fig:",width=278 ] ( a ) real ( blue solid ) and imaginary ( red dashed ) parts of @xmath57 . the parameters used are @xmath58 , @xmath59 and @xmath60 .
the particle - hole excitations from the intraband scatterings are greatly suppressed , resulting a gap - like feature in @xmath61 .
the intraband and interband scatterings in @xmath62 and @xmath63 are schematically illustrated in ( b ) and ( c ) respectively.,title="fig:",width=278 ] _ collective charge excitation in normal state _
in the normal state without cdw , @xmath64 . as a result , @xmath65 and @xmath66
are decoupled .
@xmath65 has an undamped delta - function - like peak corresponding to the plasmon excitations , which has been studied previously@xcite . as for @xmath66 , we find that an undamped delta - function - like peak could emerge if @xmath36 is negative .
the occurrence of this undamped mode is due to a gap - like feature appearing in the bare susceptibility @xmath62 shown in fig .
[ fig : chi022](a ) .
because the weyl nodes have opposite chirality , the intraband scattering is largely suppressed while the interband scattering is robust in @xmath62 as demonstrated in fig .
[ fig : chi022](b ) .
this unusual chirality effect leads to @xmath67 for @xmath68 , where @xmath69 is the particle - hole continuum edge for the interband scattering .
in contrast , @xmath63 is dominated by the intraband scattering as sketched in fig .
[ fig : chi022](c ) , and consequently @xmath70 is finite for @xmath71 , where @xmath72 is the particle - hole continuum edge for the intraband scattering , and almost zero for @xmath73 . by kramers
kronig relations , it can be proved that @xmath74 increases monotonically from @xmath75 until @xmath76 . if @xmath36 is negative , a collective mode with the frequency @xmath77 satisfying @xmath78 exists , which is corresponding to the undamped delta - function - like peak obtained in @xmath66 .
@xmath77 decreases with the increase of @xmath79 , and the normal state remains stable until @xmath80 where cdw state starts to emerge .
@xmath81 as a function of @xmath79 is plotted in fig .
[ fig : chirpa22 ] .
we notice that the critical interaction strength for cdw can be obtained by @xmath82 which is roughly proportional to @xmath83 .
this dependence of @xmath17 comes from the diverging density of states for the interband scattering in @xmath62 , which has also been observed in previous study.@xcite we emphasize that the gap - like feature in @xmath84 is a direct consequence of the chirality of the weyl nodes . as a result ,
the undamped delta - function - like peak in the @xmath81 in the normal state is a unique feature for the weyl semimetal , and it is rarely seen in other materials with cdw instability in which the peak is significantly damped by the particle - hole continuum . the undamped collective mode in @xmath85 associated with the cdw instability in the normal state .
the peak moves toward zero frequency as @xmath36 approaches @xmath86 .
the parameters used are @xmath58 , @xmath87 and @xmath60.,width=278 ] _ resonant plasmon - axion excitation in cdw state _ in the cdw state , we find two new effects in the bare susceptibilities .
first , since the cdw order parameter mixes up the states with different chiralities , all the scatterings have finite probability . as a result , @xmath70 develops new spectral weights within the interband particle - hole continuum ( @xmath73 ) .
these extra contributions in @xmath88 push the plasmon frequency toward the lower edge of the particle - hole continuum , resulting in the plasmon frequency @xmath89 being more dispersive in the cdw state than in the normal state@xcite .
second , because @xmath90 is non - zero in the cdw state , the frequencies of the collective excitations are now determined by ( + ^0 ( , ) _ ) = 0 , [ plas ] which has _ two _ solutions as explained below .
eq .
[ plas ] can be rewritten as : k_11(,)k_22 ( , ) - v_v_cdw^0_12(,)^2 = 0 , where [ k1122 ] k_11(,)&= & 1+v _
^0_11(,)a(1- ) , + k_22(,)&= & 1+v_cdw ^0_22(,)b(1-),and @xmath89 ( @xmath91 ) is the plasmon ( cdw phonon ) frequency without @xmath90 .
note that we have expanded @xmath92 and @xmath93 in the region without uncorrelated particle - hole excitations , i.e. , @xmath94 .
this is also the region where the undamped collective excitations could exist . substituting eq .
[ k1122 ] into eq .
[ plas ] , it can be proved straightforwardly that two collective excitations exist with frequencies near @xmath95 and @xmath96 respectively .
since these two collective excitations are mixtures of the plasmon excitation and the cdw phonon mode , they should result in prominent peaks in both @xmath97 and @xmath98 , emerging as resonant modes between these two channels .
fig .
[ fig : plascdw ] plots the frequencies of these two resonant modes solved from eq .
[ plas ] for different @xmath19 together with @xmath99 and @xmath100 .
note that because magnitude of the spectral weight due to the interband scattering generally scales with @xmath101 ( @xmath102 ) , we could not draw a quantitative conclusion on the amount of the spectral weight for each resonant mode .
nevertheless , the qualitative features of the collective excitations are robust regardless of the @xmath17 chosen . the frequencies of the two branches of collective excitations due to the mixture of the plasmon excitation and cdw phonon ( axion ) mode .
the parameters used are @xmath58 , @xmath103 , @xmath104 , @xmath105 , and @xmath60 .
the solid lines represent the particle - hole continuum edges for the intraband ( @xmath99)and interband ( @xmath100 ) scatterings respectively .
the branch marked by the circle has more plasmon characters while the one marked by the triangle has more cdw phonon ( axion ) characters .
both branches should show up in the density - density correlation functions around @xmath4 and @xmath0 , which in principle can be detected by m - eels.,width=278 ] it is remarkable to see that different mechanism to break the chiral symmetry in a weyl semimetal could result in different behaviors of the plasmon excitation .
zhou _
et al .
_ studied the plasmon excitation of a weyl semimetal with the chiral symmetry broken by the application of a pair of the electric @xmath10 and the magnetic @xmath11 fields.@xcite in that case , the chirality - dependent chemical potential @xmath106 is introduced , which leads to a @xmath19-dependent damping effect of the plasmon excitation .
however , since @xmath106 does not provide any coupling between different weyl nodes , the plasmon excitation does not have the resonant nature as discussed in the cdw state .
it can be seen that the cdw phonon mode has a dispersion of @xmath107 as predicted by the goldstone theory . however , the reason why we obtain the gapless goldstone mode is that the effective hamiltonian in eq .
[ h0 ] has the continuous translational symmetry .
if the lattice effect is included , we will obtain a gapped cdw phonon mode .
nevertheless , the weyl semimetals discovered recently@xcite have quite small @xmath108 , thus the wavelength of the cdw state @xmath109 is much longer than the lattice constant .
the lattice effect can therefore be argued to be quite small , and the cdw phonon mode should have a very small gap . moreover , because the phase of the cdw order parameter is shown to be the axion field@xcite , the cdw phonon mode is in fact a collective excitation of the axion field . as a result , measuring these resonant plasmon - axion excitations around @xmath4 and @xmath0 could serve as a reliable way to detect the axion field .
_ conclusion _ in this paper , we have investigated the charge excitations of a weyl semimetal in the charge - density - wave ( cdw)state . we have found that in the normal state , an undamped collective mode could emerge in the density - density correlation function at the cdw ordering wavevector @xmath3 if there is an attractive interaction favoring the formation of the cdw state . in the cdw state , we have found that the plasmon excitation becomes more dispersive compared to the behavior in the normal state .
moreover , we have shown that due to the umklapp scattering enabled by the cdw state , the plasmon excitation couples to the cdw phonon ( axion ) mode , leading to two branches of collective excitations .
both collective excitations can be seen in the density fluctuations at @xmath4 and @xmath0 simultaneously , exhibiting an interesting resonant nature between two distinct channels .
we have proposed that measuring the resonant plasmon - axion excitations around @xmath4 and @xmath0 by m - eels could serve as a reliable way to detect the cdw state as well as the nature of the axion field in a weyl semimetal .
_ acknowledgement _ this work is supported by a start up fund from binghamton university . |
in the last years there has been a renewed interest in the analysis of interacting field theories in de sitter spacetimes .
on the one hand , there is observational evidence that supports the existence of periods of exponential or quasi - exponential expansion of the universe , both during the inflationary era before the radiation dominated period in the early universe , and in the present accelerated expansion .
quantum fields may play a relevant role during both periods . on the other hand , the high degree of symmetry of de sitter spacetime ,
make it a very interesting arena to analyze conceptual problems of semiclassical and quantum gravity .
there are several unsettled questions regarding interacting quantum fields in de sitter spacetime .
in particular , perturbative calculations show the existence of secular growth and/or infrared ( ir ) divergences @xcite .
the appearance of terms that secularly grow with time , which can lead to a breakdown of perturbation theory at late times , might be due to a deficiency of the perturbative approach .
therefore , it is clear that in order to make reliable predictions , such infrared effects would have to be understood , and for this the use of nonperturbative techniques seems to be unavoidable .
there are several works in the literature attempting to understand the physical effects of the secular growth @xcite .
recently , there has been much progress in the context of single - field inflationary models where there are only adiabatic perturbations , and the variable @xmath1 that describes the curvature perturbations remains constant on super - horizon scales @xcite .
working directly with the evolution operator for @xmath1 , it has been shown @xcite that despite the existence of individual contributions to @xmath1 correlation functions that secularly grow with time , when all of them are added up cancellations occur so that the final result does not exhibit a secular growth .
for this , the use of symmetries and nonperturbative techniques have shown to be crucial and the computations to be very subtle , even for the case in which only scalar perturbations are considered . a complete explicit calculation that addresses
the renormalization procedure is extremely complex , and still lacking .
an alternative two - step procedure to perturbatively compute loop corrections to @xmath1 correlation functions has been also applied ( see for instance @xcite ) .
the first step consists in calculating correlation functions of the inflaton fluctuations , and the second in obtaining the ones of @xmath1 by performing a gauge transformation @xcite . in this case , even though the correlation functions of the inflaton fluctuations can show secular growth , the use of nonperturbative schemes is avoided by performing the gauge transformation soon after the field modes exited the horizon ( which would be well before the secular growing effect becomes relevant ) .
nevertheless , to our knowledge , when loop corrections are included there is still no formal general proof of the equivalence between the two procedures .
the effect of secular growth is less clear for massless interacting test fields in de sitter space ( which in the context of inflationary models could be interpreted as an approximation to entropy modes ) .
this kind of fields have been considered by several authors in different frameworks ; for instance , in the stochastic approach @xcite , in the context of euclidean de sitter space @xcite , using dynamical renormalization group resummations @xcite , or other nonperturbative approximation schemes @xcite . in particular , it has been pointed out that in the presence of interactions a scalar field dynamically acquires a mass @xcite , which screens perturbative ir divergences . the dynamical generated mass has been obtained in the different approaches and , although the numerical value does not coincide for some of the approximations , they agree on that it is non - analytic in the coupling constant . also , it has been argued that for scalar fields presenting a symmetry breaking tree - level potential , the presence of large infrared effects could give rise to a symmetry restoration @xcite ( see also @xcite ) .
nevertheless , the conditions under which the symmetry is restored and the order of the phase transition are still unclear @xcite .
we will come back to these points later .
nonperturbative quantum field theory is a difficult topic even in flat spacetime .
the most widely used approximations involve the selective summation of feynman diagrams of infinite perturbative order , like the hartree approximation , or expansions in @xmath2 , where @xmath3 is the number of quantum fields ( or the number of colors in qcd ) .
nonperturbative calculations can be addressed using the effective action for composite operators introduced a long time ago by cornwall , jackiw and tomboulis @xcite , also called in the literature two - particle irreducible effective action ( 2pi ea ) .
indeed , this is a general and systematically improvable approach for resumming classes of feynman diagrams . in its lowest approximation level
, it corresponds to the so called hartree ( or gaussian ) approximation .
the gaussian approximation can also be introduced without referring to the 2pi ea , by means of a variational principle .
it has been the subject of many works after the papers by stevenson @xcite , in which the author showed that it is possible to choose carefully the bare constants of the @xmath0 theory allowing for the emergence of a nontrivial quantum theory in the limit of infinite cutoff ( it was shown afterwards that a different choice of the parameters that involve a wave function renormalization is also possible , resulting in a completely different quantum theory @xcite ) .
both possibilities have their corresponding counterparts in curved spacetimes @xcite .
the quantum theories so derived have different drawbacks , which originate from the fact that in the perturbative regime ( where the renormalized coupling constants are very small ) those theories do not approach the ones defined by the usual perturbation theory in terms of feynman diagrams , which are based on a loop - by - loop renormalization procedure .
moreover , for finite values of the cutoff the theory is unstable , since the bare coupling constant is negative and goes to zero as the cutoff tends to infinite @xcite .
in addition , there is no clear way to systematically improve the approximation based on such variational principle .
the use of the 2pi ea solves some of these drawbacks by changing the renormalization procedure .
indeed , in this formalism , there are multiple definitions of a given @xmath4-point function , which are equivalent in the exact theory but not necessarily if some approximation is considered .
when the theory is truncated , it is mandatory to impose some consistency relations between the different counterterms , and this is possible only if one allows for more bare constants than usual , which are to be fixed in terms of the usual number of renormalized parameters ( the ones that are in principle measurable ) .
in other words , for a given renormalized parameter one includes different counterterms .
there has been much progress in understanding the renormalization properties of general truncation of the 2pi effective action , and a systematic method for a consistent renormalization has been developed @xcite . in this paper
we will analyze interacting quantum fields in general spacetimes using the 2pi ea with the consistent renormalization procedure of ref .
we will show that the renormalization procedure can be generalized to curved spacetimes , and we will find the explicit renormalized form for the mean value equation and for the propagator of the fluctuations .
we will then analyze the particular case of de sitter spacetime .
the study of the mean value equation along with the corresponding effective potential will allow us to show that , under this nonperturbative approximation and renormalization , depending on the values of the parameters , spontaneous symmetry breaking may take place in de sitter spacetime .
this is to be contrasted with the results based on the approach of ref .
@xcite , in which one can prove that spontaneous symmetry breaking is not compatible with de sitter symmetries @xcite ( a similar incompatibility holds to leading order in the @xmath2 approximation to @xmath5 models , as mentioned in @xcite ; see also @xcite ) .
our results show discrepancies with those of ref.@xcite , where the effective potential in de sitter spacetime has been also recently considered using the consistent renormalization procedure .
as we will see , the discrepancies come from the milder version of the consistency conditions used in that work . before describing the details of the calculations ,
it is worth to stress here that the aim of this paper is not to arrive at a conclusive answer to these points . indeed , previous works based in the hartree approximation in flat spacetime suggest that this is not possible .
for example , it is well known that the hartree approximation predicts a first order phase transition when applied to @xmath5 models , while other nonperturbative methods give a second - order phase transition @xcite , in agreement with expectations based on general arguments @xcite .
these drawbacks of the hartree approximation can be cured by including two loops corrections to the 2pi ea @xcite .
a similar situation could arise in curved spacetimes , in particular when discussing symmetry breaking in de sitter spacetime .
therefore , one of our purposes here is to push forward a rigorous and critical analysis of the situation , by starting with an understanding of the simplest nonperturbative approximation , along with its limitations , but having in mind the necessity of an improvement .
this subject is clearly worth studying in detail , to the end of analyzing its eventual relevance not only for correlation functions , but also for the evolution of the universe itself through its contribution to the energy momentum tensor in the semiclassical einstein equations ( see ) , which is the subject of a forthcoming paper ii @xcite .
the paper is organized as follows . in sec .
[ 2pisect ] we introduce the 2pi ea in curved spacetimes , and make contact with previous formulations based on the use of the equations for the mean value of the field and its two - point function . in sec .
[ rengapsect ] we discuss in detail the renormalization of the field and gap equations , using the consistent renormalization procedure of @xcite .
we show that the renormalized equations are independent of the scale @xmath6 introduced by the dimensional regularization , once we express the equations in terms of the renormalized mass and coupling constants defined in minkowski spacetime . in sec .
[ dssect ] we analyze the field and gap equations in a de sitter spacetime , exploring the possibility of having spontaneous symmetry breaking . in sec .
[ rp ] we generalize all the previous results to the case in which the renormalization point is taken for a given fixed de sitter metric .
this change of the renormalization point allows us to explore further the parameter space where spontaneous symmetry breaking is possible . in sec .
[ conclusect ] we present a summary and a discussion of the results obtained in the paper .
the appendices contain some details of the calculations .
everywhere we set @xmath7 and adopt the mostly plus sign convention .
the definition of the 2pi ea along with the corresponding functional integral can be found in both papers and textbooks ( see for instance @xcite ) . in this section
, we briefly summarize the main relevant aspects of the formalism for the case of a self - interacting scalar field @xmath8 in a general curved spacetime .
the 2pi generating functional can be written as @xcite @xmath9 = s_0[\phi_0,g^{\mu \nu } ] + \frac{i}{2 } \textup{tr } \ln(g^{-1 } ) + \frac{i}{2 } \textup{tr } ( g_0^{-1 } g ) + \gamma_{int}[\phi_0,g , g^{\mu \nu } ] , \label{gamma-2pi - alt}\ ] ] where @xmath10 is quadratic part of the classical action @xmath11 without any counterterms and @xmath12 = \frac{i}{2 } \phi_a g_{0,ab}^{-1 } \phi_b i g_0^{ab}(x , x ' ) = \frac{1}{\sqrt{-g } } \frac{\delta^2 s_0[\phi_0,g^{\mu\nu}]}{\delta \phi_a(x ) \delta \phi_b(x ' ) } \frac{1}{\sqrt{-g'}},\ ] ] with @xmath13 the time branch indices ( with index set @xmath14 in the usual notation ) corresponding to the ordering on the contour in the `` closed - time - path''(ctp ) or schwinger - keldysh @xcite formalism .
this formalism is in principle needed in order to obtain real and causal evolution equations , although in the approximation we will use in this paper the details of the ordering along the contour will not be needed .
the interaction part of the 2pi ea is given by @xmath15 = s_{int}[\phi_0,g^{\mu \nu } ] + \frac{1}{2 } \tr \left [ \frac{\delta^2 s_{int}}{\delta \phi_0 \delta \phi_0 } g \right ] + \gamma_2[\phi_0,g , g^{\mu \nu}],\ ] ] where @xmath16 , the functional @xmath17 is @xmath18 times the sum of all two - particle - irreducible vacuum - to - vacuum diagrams with lines given by @xmath19 and vertices obtained from the shifted action @xmath20 , which comes from expanding @xmath21 $ ] and collecting all terms higher than quadratic in the fluctuating field @xmath22 .
the equations of motion are obtained by extremizing the effective action @xmath23 the usual 1pi ea is then obtained by replacing the solution of the gap equation @xmath24 $ ] back in the 2pi ea @xmath25 = \gamma_{2pi}[\phi_0,\bar{g}[\phi_0],g^{\mu \nu } ] .
\label{1pi-2pi}\ ] ] as mentioned in the introduction , in the 2pi formalism there are various possible @xmath4-point functions that can be obtained from the several possible ways of functionally differentiating @xmath26 $ ] with respect to @xmath27 and @xmath28 .
however in the exact theory it can be shown that the different @xmath4-point functions are not independent , and must satisfy certain consistency conditions .
these are thoroughly derived in @xcite , of which we provide a brief sketch in appendix a. to the purposes of this paper , it is enough to consider a simplified version of these conditions which are only valid for a theory with @xmath29 symmetry , and taking advantage of further simplifications when evaluating at @xmath30 .
the two important conditions in our case are @xmath31 and @xmath32}{\delta \phi_1 \delta \phi_2 \delta \phi_3 \delta \phi_4}\bigg|_{\phi_0=0 } = 2 \left [ \frac{\delta^2 \gamma_{int}}{\delta g_{12 } \delta g_{34 } } \bigg|_{\bar{g},\phi_0=0 } + perms(2,3,4 ) \right ] - \frac{1}{2 } \frac{\delta^4 \gamma_{int}}{\delta \phi_1 \delta \phi_2 \delta \phi_3 \delta \phi_4 } \bigg|_{\bar{g},\phi_0=0 } , \,\,\,\,\,\ , % 2 \left [ \frac{\delta^2 \gamma_{int}}{\delta g_{12 } \delta g_{34 } } \bigg|_{\bar{g } } + \frac{\delta^2 \gamma_{int}}{\delta g_{13 } \delta g_{24 } } \bigg|_{\bar{g } } + \frac{\delta^2 \gamma_{int}}{\delta g_{14 } \delta g_{23 } }
\bigg|_{\bar{g } } \right ] , \label{4-pt - relation}\end{aligned}\ ] ] where expression is used to establish a connection between the 1pi effective action on the left hand side and @xmath33 on the right hand side .
it is worth to recall that , due to an artifact of the approximation , these conditions can be violated when applied to a particular truncation of @xmath34 .
however , if the approximation is good any departure from these conditions should be small . nevertheless , since we are dealing with formally divergent quantities , the latter statement is non - trivial , and therefore a consistent renormalization procedure that takes these relations into account is required .
in fact , an essential step in the consistent renormalization procedure of ref .
@xcite is to impose these conditions to the particular truncation of @xmath34 at a given renormalization point ( for which we choose @xmath30 ) , which allows to fix the different counterterms . for a non -
minimally coupled scalar field with quartic self - coupling in a curved background with metric @xmath36 the classical action reads @xmath37 = -\int d^4 x \
, \sqrt{-g } \left [ \frac{1}{2 } \phi \left ( -\square + m^2_b + \xi_b r \right ) \phi + \frac{1}{4 ! } \lambda_b \phi^4 \right ] , \label{classical - s}\ ] ] where @xmath38 , @xmath39 .
then the shifted action that defines the interaction vertices needed to construct the 2-particle irreducible vacuum diagrams is @xmath40 = - \frac{\lambda_b}{6 } \int d^4 \,x \
, \sqrt{-g } \left [ \frac{1}{4 } \varphi^4 + \phi_0 \varphi^3 \right ] .
\label{shifted - action}\ ] ] although we are dealing with a scheme in which there is an infinite resummation of diagrams , we can not compute the effective action completely because we still have to perform an infinite sum of 2pi vacuum diagrams in the @xmath17 term .
therefore we must resort to some kind of approximation . at the lowest order one can drop the @xmath17 term altogether , which corresponds to the 1-loop approximation in which case
there is no difference with the 1pi effective action .
this is because the 2pi vacuum diagrams start at 2-loops , and it is from this order onwards that the 2pi effective action gives a nontrivial result compared to the 1pi effective action . with the shifted action @xmath41 given by eq . , there are two diagrams that contribute at the 2-loop order .
these are the double - bubble and the sunset shown in fig .
[ diagrams ] .
( 3,0 ) ( 0,0 ) ( -0.48,0)(1,0)0.95 ( 2,0 ) ( 2.94,0 ) the hartree approximation corresponds to taking into account only the local contribution ( double - bubble diagram ) . in this case
the 2pi effective action is @xmath42 & = & -\int d^4 x \
, \sqrt{-g } \left [ \frac{1}{2 } \phi_0 \left ( -\square + m_{b2}^2 + \xi_{b2 } r \right ) \phi_0 + \frac{1}{4 ! } \lambda_{b4 } \phi_0 ^ 4 \right ] + \frac{i}{2 } \tr \ln ( g^{-1 } ) \nonumber \\ & & - \frac{1}{2 } \int d^4 x \ , \sqrt{-g } \left [ -\square + m_{b0}^2 + \xi_{b0 } r + \frac{1}{2 } \lambda_{b2 } \phi_0 ^ 2 \right ] g(x , x ) \label{2pi - lambdaphi4 } \\
\nonumber & & - \frac{\lambda_{b0}}{8 } \int d^4 x \ , \sqrt{-g } \ ,
g^2(x , x ) . % \notag\end{aligned}\ ] ] note that in each contribution to the effective action we allow the counterterms to be different , denoted as different subscripts in the bare parameters that refer to the power of @xmath43 in the corresponding term of the action .
this allows for the possibility of adjusting how each diagram contributes to the cancellation of divergences , which turns out to be a crucial point for the renormalization procedure to respect the 2pi consistency relations of the exact theory .
the relationship between the different counterterms will be fixed by imposing the consistency relations on the 2 and 4-point functions in the next section .
it is important to note that , within the hartree approximation , the ctp formalism gives the same equations of motion than the usual _ in - out _
formalism @xcite .
therefore , in eq .
( [ 2pi - lambdaphi4 ] ) and in what follows we drop the time branch indices .
the equations of motion for the mean field and the exact propagator ( gap equation ) are obtained by taking the variation of the @xmath44 $ ] with respect to @xmath43 and @xmath19 respectively . in the hartree approximation that we are considering these read @xmath45
these equations are similar to those obtained when considering a gaussian approximation at the level of the mean field equations @xcite . in that case
the starting point is the classical equation of motion for @xmath8 , which is then separated into a mean field @xmath46 and a fluctuation @xmath47 .
then taking the expectation value of the classical field equation one gets a pair of coupled equations for @xmath43 and @xmath22 , or equivalently for @xmath43 and the propagator of the fluctuations @xmath48 . finally upon the assumption of gaussian states , which implies @xmath49 the resulting equations are similar to those shown above .
the difference is , however , that there is neither a diagrammatic interpretation for each of the contributions that allows for different counterterms , nor a set of consistency relations to fix them .
therefore in this approach it would seem unnatural to use different counterterms .
in the analysis of the renormalization of the hartree approximation , we will use the following parametrization of the bare couplings : @xmath50 [ type - of - counterterms ] which corresponds to the ms scheme ( i.e. , the counterterms @xmath51 , @xmath52 and @xmath53 ( @xmath54,@xmath55 ) contain only divergences and no finite part ) . as mentioned in the previous section , in order to fix the different counterterms within this level of approximation we impose the consistency relations and on the different 2pi 2 and 4-point kernels .
these kernels are computed from functional derivatives of @xmath56 $ ] evaluated at @xmath30 , which by comparison of and reads @xmath57 & = & -\int d^4 x \
, \sqrt{-g } \left [ \frac{1}{2 } \left ( \delta m_2 + \delta \xi_2 r \right ) \phi_0 ^ 2 + \frac{1}{4 ! } \left ( \lambda + \delta \lambda_4 \right ) \phi_0 ^ 4 \right ] \notag \\ & & - \frac{1}{2 } \int d^4 x \ ,
\sqrt{-g } \left [ \delta m_0 + \delta \xi_0 r + \frac{1}{2 } \left ( \lambda + \delta \lambda_2 \right ) \phi_0 ^ 2 \right ] g(x , x ) \label{gammaint - lambdaphi4 } \\ & & - \frac{1}{8 } \int d^4 x \
, \sqrt{-g } \left ( \lambda + \delta \lambda_0 \right ) g^2(x , x ) , \notag\end{aligned}\ ] ] where we have used that the inverse free propagator @xmath58 in is defined as @xmath59 after some straightforward functional differentiations , the 2-point kernels at @xmath60 are given by @xmath61 \delta(x - x ' ) , \\ 2 \frac{\delta \gamma_{int}}{\delta g(x , x ' ) } \bigg|_{\phi_0=0 } & = & - \sqrt{-g}\left [ \delta m_0 + \delta \xi_0 r + \frac{1}{2}(\lambda + \delta \lambda_0 ) g(x , x ) \right ] \delta(x - x').\end{aligned}\ ] ] hence the condition implies that @xmath62 since both mass counterterms are equal , as also both curvature coupling counterterms are , we drop the subindexes in those cases from now on .
we turn to the second consistency relation eq . .
the equation relates the different 4-point kernels at @xmath60 , two of which can be calculated by carrying on differentiating @xmath56 $ ] : @xmath63 where we used @xmath64 as a notational shorthand .
the other quantity that must be specified is the fourth functional derivative of the 1pi effective action .
its value at @xmath30 is then easily interpreted as the renormalized self - interaction coupling @xmath65 , since the effective potential is proportional to the effective action at a constant value of @xmath43 .
therefore @xmath66}{\delta \phi_1 \delta \phi_2 \delta \phi_3 \delta \phi_4 } \bigg|_{\phi_0=0 } = - \sqrt{-g } \ , \lambda_r \delta_{12 } \delta_{13 } \delta_{14},\ ] ] is equivalent to @xmath67 it is important to note that , since the 1pi effective action depends on the background geometry , in general it is not possible to impose these consistency conditions exactly for any arbitrary background metric .
we expect an analogous situation to happen in minkowski space if for instance the background field is time dependent .
however , in that case , in order to define the renormalized parameters , one can still impose the consistency conditions for a particular constant value of the field .
this is enough to fix the counterterms .
then , once made finite , any deviation from the consistency conditions is expected to be small , or more precisely to be of the same order as the contributions neglected in the approximation considered for @xmath68 .
the corresponding choice in our case , is to consider a spacetime with constant curvature .
we will come back to this point in sec .
[ rp ] . feeding the above ingredients to the second consistency relation
fixes the remaining counterterm .
here it is important to take into account the symmetry properties of the kernels to evaluate the different permutations .
the resulting relation is @xmath69 at this point one can see more clearly the consequences of this `` arbitrary '' truncation of the 2pi effective action , i.e. the hartree approximation , if we insist on enforcing the properties that are valid for the exact theory . turning back to the renormalization of the mean field and gap equations , instead of considering the equation for the feynman propagator @xmath48
, we will consider the equation for the hadamard propagator @xmath70 , which contains the same information but is more convenient for our purposes @xmath71 \right ) g_1(x , x ' ) = 0,\ ] ] where @xmath72 = g_1(x , x ) = 2 g(x , x ) = 2 \langle \varphi^2 \rangle$ ] .
renormalizability of the mean field and gap equations means that they can be rendered finite by a suitable choice of counterterms .
if this is the case , the resulting equations can be expressed in terms of a finite physical mass plus a coupling to the curvature term , namely @xmath73 the physical mass @xmath74 is a spacetime dependent scalar function determined by @xmath75 , \label{phys - mass}\ ] ] which is a self - consistent equation , as @xmath74 also enters the right hand side through @xmath72 $ ] . in this expression
we have already used the relationships among the counterterms as dictated by the consistency conditions .
the divergences come from @xmath72 $ ] , and must be cancelled by a suitable choice of the counterterms @xmath76 , @xmath77 and @xmath78 . to the end of exposing and isolating the divergences
, we use a schwinger - dewitt type expansion for @xmath72 $ ] for a free field with variable mass @xmath79 and coupling to the curvature @xmath80 on a general @xmath4-dimensional background with metric @xmath81 : @xmath82 & = & \frac{1}{8 \pi^2 } \left ( \frac{m_{ph}^2}{\mu^2 } \right)^{\epsilon/2 } \sum_{j \geq 0 } \ , [ \omega_j ] ( m_{ph}^2)^{1-j } \
, \gamma\left ( j-1 - \frac{\epsilon}{2 } \right ) \notag\\ & \equiv & \frac{1}{4\pi^2 \epsilon } \left [ m_{ph}^2 + \left(\xi_r - \frac{1}{6 } \right ) r \right ] + 2\ , t_f ( m_{ph}^2,\xi_r , r,\tilde{\mu } ) .
\label{adiab - g1 } \end{aligned}\ ] ] here @xmath83 , @xmath84 is the gamma function and the coefficients @xmath85 $ ] are scalars of adiabatic order @xmath86 built from the metric and its derivatives and satisfy certain recurrence relations .
we use the results for these coefficients presented in @xcite . in the second line
we have split the sum by isolating the zeroth and second adiabatic orders , which are the only divergent terms .
then we expanded for @xmath87 and redefined @xmath88 to absorb some constant terms .
the finite part is @xmath89 \ln \left ( \frac{m_{ph}^2}{\tilde{\mu}^2 } \right ) + \left(\xi_r - \frac{1}{6 } \right )
r \\ & - & 2 f(m_{ph}^2,\{r\ } ) \bigg{\}}\end{aligned}\ ] ] where the function @xmath90 contains the adiabatic orders higher than two and it is independent of @xmath91 and @xmath92 .
the dependence on @xmath74 involves also its derivatives , while that denoted by @xmath93 is to be understood as a dependence on curvature invariants constructed from contractions of the riemann tensor and its derivatives .
this function satisfies the following properties @xmath94 [ f - props ] then inserting eq . back in eq .
, we have @xmath95 \\ & + & \frac{1}{2 } ( \lambda + \delta
\lambda_2 ) t_f.\end{aligned}\ ] ] to separate unambiguously the divergent part from the finite part it is necessary to choose a subtraction scheme . in the ms scheme ,
the divergent part must vanish , and then the following equations are satisfied independently @xmath96 + \frac{1}{2 } \delta \lambda_2 t_f \bigg\}.\end{aligned}\ ] ] these conditions will determine the ms counterterms @xmath76 , @xmath77 and @xmath78 . using the first equation to express the @xmath97 and replace it in the second , we end up with an expression that depends on the counterterms and the finite constants @xmath98 + \left [ \delta \xi + \frac{1}{16\pi^2 \epsilon } \left ( \xi - \frac{1}{6 } \right ) ( \lambda + \delta
\lambda_2 ) \right ] r\nonumber \\ & + & \frac{1}{2 } \left [ \delta \lambda_2 + \frac{\lambda}{16\pi^2 \epsilon } ( \lambda + \delta \lambda_2 ) \right ] \left ( \phi_0 ^ 2 + t_f \right),\end{aligned}\ ] ] where each pair of square brackets must vanish independently .
the resulting counterterms are @xmath99 and the bare parameters become @xmath100 [ bare - couplings ] once made finite , the equation for the physical mass is @xmath101 \ln \left ( \frac{m_{ph}^2}{\tilde{\mu}^2 } \right ) \right.\nonumber\\ & + & \left .
\left(\xi_r - \frac{1}{6 } \right ) r - 2 f(m_{ph}^2,\{r\ } ) \right\}. \label{phys - mass - fin}\end{aligned}\ ] ] this is a self - consistent equation for @xmath102 , whose result shall then be inserted into the field equation and solved for @xmath43 . expressed in this way
, the result depends on a mixture of the finite ms - parameters @xmath103 , @xmath104 and @xmath105 and the renormalized parameters @xmath80 and @xmath65 ( this last one coming from the consistency condition through the field equation ) , as well as in the regularization scale @xmath106 .
the renormalized parameters are those that characterize the effective potential @xmath107 and will not be equal to the ms - parameters in general .
both sets of parameters will be related by @xmath106 .
it is convenient then to express the @xmath74-equation in terms of only one set of parameters , for which we use the effective potential as identified through its derivative in the field equation @xmath108 in order to find relations between both sets .
as it was discussed in the previous section , the consistency conditions take a particularly simple form only at @xmath60 , so we will use that choice of renormalization point .
it is also necessary to fix the background geometry at a constant curvature . in this section
we will choose minkowski spacetime as the renormalization point , for which @xmath109 .
with these to conditions we define the renormalized parameters as @xmath110 [ ren - quantities ] where the @xmath111 subindex means both zero curvature and @xmath60 .
the definition is the same as . from these definitions and eqs . and , it is straightforward to arrive at expressions that relate the renormalized parameters @xmath112 , @xmath80 and @xmath65 to the ms finite parameters @xmath103 , @xmath104 and @xmath105 and @xmath106 ( see appendix b ) .
the resulting relations are @xmath113 } , \label{mr - mu } \\ \left ( \xi_r - \frac{1}{6 } \right ) & = \frac{\left ( \xi - \frac{1}{6 } \right ) } { \left[1 - \frac{\lambda}{32\pi^2 } - \frac{\lambda}{32\pi^2 } \ln \left ( \frac{m_r^2}{\tilde{\mu}^2 } \right ) \right ] } , \label{xir - mu } \\
\lambda_r & = \frac{\lambda}{\left [ 1 - \frac{\lambda}{32\pi^2 } - \frac{\lambda}{32\pi^2 } \ln \left ( \frac{m_{r}^2}{\tilde{\mu}^2 } \right ) \right]}. \label{lambdar - mu}\end{aligned}\ ] ] [ ren - mu ] putting these together we can find some useful @xmath106-independent combinations of the ms parameters : @xmath114 and @xmath115 where we have introduced @xmath116 as a notational shorthand , with the property @xmath117 for @xmath118 . note that , although in principle @xmath103 could take negative values , the parameter @xmath112 is positive by construction .
this is so because @xmath112 is a solution of the gap equation in minkowski space at @xmath30 , which is based on the existence of an stable propagator for the fluctuations ( see eq .
( [ adiab - g1 ] ) ) .
therefore , these equations tell us that in this case @xmath103 must also be positive , provided that @xmath116 and @xmath105 are positive .
this is a consequence of having defined the renormalized parameters in minkowski spacetime , and as we will see in sec .
v , this restriction can be relaxed by taking the renormalization point in de sitter spacetime . in any case
, it is worth to remark that the consistency conditions impose nontrivial constraints on the finite parameters of the theory .
these restrictions are not considered ( or at least not apparently ) in the approach of ref .
@xcite where , working with minimal subtraction , it is assumed from the beginning that @xmath119 .
on one hand , as we are working with an approximation to the ea , one could argue that it is not necessary to impose the consistency conditions exactly at the renormalization point , but only to use the conditions to fix the proportionality constant between the counterterms , because anyway they are not expected to be exact beyond that point .
this is in principle correct . however , unless a set of renormalization conditions is specified , the interpretation of the finite parameters is unclear and the equations are @xmath6-dependent , as happen in ref .
. moreover , if we assume that @xmath119 , and then we define the renormalization conditions as derived from the effective potential ( as we are doing ) , it turns out that the gap equation can not be entirely written in terms of only the renormalized parameters , and hence it is @xmath6-dependent .
this can be easily seen by noticing that it is due to the combination on the rhs of eq .
( [ lambdas2 - 4 ] ) that it is @xmath65 and not @xmath105 which appears in eq .
( [ field - eq ] ) . on the other hand ,
if the consistency conditions are imposed , the relation @xmath119 implies that @xmath120 , which yields a particular choice of the parameter @xmath6 .
our analysis shows that , taking minkowski spacetime as the renormalization point , the choice @xmath121 is not compatible with the consistency conditions .
similar restrictions will appear when considering a more general definition for the renormalized parameters ( see section v ) .
the above relations between the parameters can be used to rewrite the equation for @xmath74 in terms of the renormalized parameters only .
after a bit of algebra we arrive at @xmath122 \ln \left ( \frac{m_{ph}^2}{m_r^2 } \right ) - 2 f(m_{ph}^2,\{r\ } ) \right\}. \label{mph - eq}\end{aligned}\ ] ] this is the main result of this section .
it shows that @xmath74 can be completely expressed in terms of the renormalized parameters , showing a manifest invariance under changes of the regularization scale @xmath106 . as a consequence both the equation for @xmath43 and the gap equation
will also show these properties .
furthermore , it can be easily seen that in the free field limit , @xmath123 , we have @xmath124 and consequently the physical mass becomes the renormalized mass , @xmath125 .
we close this section by defining the non - ms counterterms associated to the renormalized parameters in the following way @xmath126}{1+\frac{\lambda}{16\pi^2\epsilon } } - m_r^2 = - \frac{\lambda_r^*}{32 \pi^2 } \frac{m^2}{\left(1+\frac{\lambda}{16\pi^2\epsilon}\right ) } \left [ \frac{2}{\epsilon } + \ln \left ( \frac{m_{r}^2}{\tilde{\mu}^2 } \right ) \right ] \notag \\ & = & - m_b^2 \frac{\lambda_r^*}{32\pi^2 } \left [ \frac{2}{\epsilon } + \ln \left ( \frac{m_{r}^2}{\tilde{\mu}^2 } \right ) \right],\end{aligned}\ ] ] where we have used both eqs . and for @xmath127 and @xmath103 , respectively , and then eq . .
similarly we obtain @xmath128 , \label{deltatildexi } \\
\delta \tilde{\lambda } & \equiv & \lambda_{b2 } - \lambda_r = - \lambda_b \frac{\lambda_r}{32\pi^2 } \left [ \frac{2}{\epsilon } + 1 + \ln \left ( \frac{m_{r}^2}{\tilde{\mu}^2 } \right ) \right].\end{aligned}\ ] ] these counterterms contain not only the poles in @xmath91 but also a @xmath106-dependent finite term . in this expressions
the 1-loop limit can be easily taken by replacing @xmath129 , @xmath130 and @xmath131 in their right - hand - sides .
in this section we apply the above results to de sitter spacetimes .
we consider the cosmological patch of de sitter space in flat coordinates @xmath132 the de sitter spacetime is a solution to the vacuum einstein equations with positive constant curvature , and as a consequence of its large degree of symmetry a possible solution to the mean field and gap equations in this background is that with both @xmath43 and @xmath133/2 $ ] constant . in this geometry
infrared divergences appear on @xmath134 for massless minimally coupled free fields , while in the case of interacting fields , the perturbative expansion is ill defined for light fields .
we can study interacting fields in de sitter spacetime in the hartree approximation without much hustle by exploiting the fact that the gap eq .
is actually the equation for the propagator of a free field with mass @xmath74 ( which is constant in de sitter ) and coupling to the curvature @xmath80 .
the solution of such an equation is known exactly in de sitter for an arbitrary number of dimensions @xmath4 .
the most common form found in the literature for the coincidence limit @xmath72 $ ] is @xmath135 = \frac{2 h^{n-2}}{(4\pi \mu^2)^{n/2 } } \ , \gamma \left ( 1 - \frac{n}{2 } \right ) \frac{\gamma \left ( \frac{n-1}{2 } + \nu_n \right ) \ , \gamma \left ( \frac{n-1}{2 } - \nu_n \right)}{\gamma \left ( \frac{1}{2 } + \nu_n \right ) \ , \gamma \left ( \frac{1}{2 } - \nu_n \right)},\ ] ] where @xmath136 and @xmath137 . to make use of the results of sec .
ii we can extract the function @xmath138 from this expression , defined through the adiabatic expansion in eq . .
this is achieved by setting @xmath139 and expanding for @xmath87 , holding @xmath140 fixed . from this comparison ,
detailed in appendix c , we can extract the exact form of the function @xmath90 for de sitter spacetime , @xmath141 \notag \\ & & - \left ( \xi_r -\frac{1}{6 } \right ) - \frac{1}{18 } \biggr\ } , \label{f - ds } \end{aligned}\ ] ] with @xmath142 and where @xmath143 , @xmath144 is the digamma function and @xmath145 .
the @xmath146 function has all the expected properties , that is , it is written only in terms of renormalized parameters , it is independent of @xmath91 and @xmath106 , and it satisfies the correct limits eqs . .
then turning back to the physical mass equation , we can plug the expression for @xmath146 in the general expression for the renormalized @xmath74-equation , to obtain @xmath147 \left [ \ln \left ( \frac{r}{12 m_{r}^2 } \right ) + g\left ( m_{ph}^2/r + \xi_r \right ) \right ] \nonumber\\ & - & \frac{\lambda_r^*}{32 \pi^2 } \left ( \xi_r - \frac{1}{9 }
\right ) r .
\label{mph - eq - ds}\end{aligned}\ ] ] note that in de sitter this equation depends only on the combination @xmath148 since we are interested in the infrared effects we consider @xmath149 and expand the function @xmath150 for @xmath151 , @xmath152 this gives a pole in @xmath153 that can immediately be identified as the origin of the infrared divergence in @xmath134 .
hence , the solutions to the field and gap equations must be such that @xmath154 . after expanding , multiplying by @xmath155 and keeping terms
up to quadratic order in @xmath155 the gap equation reads @xmath156\ , y + c_{mk } = 0,\ ] ] where the coefficients are given by the following expressions , @xmath157 , \\
b_{mk } & = - \left ( \frac{m_r^2}{r } + \xi_r \right ) + \frac{\lambda_r^*}{32\pi^2 } \left [ \frac{1}{6 } a\left(r/12m_r^2
\right ) + \xi_r + \frac{5}{36 } \right ] , \nonumber \\
c_{mk } & = - \frac{\lambda_r^*}{768 \pi^2 } , \end{aligned}\ ] ] [ cuadeq - dsren ] with @xmath158 the @xmath159 subindex indicates we are using the minkowski renormalization point .
note that @xmath160 is always negative , while in principle @xmath161 and @xmath162 can have either sign .
it can be seen that in order to have a real and positive solution @xmath163 for all @xmath43 , it is necessary that @xmath164 , while @xmath162 is unconstrained .
the solutions are @xmath165 ^ 2 - 4 r^2\ , a_{mk } c_{mk } } } { 2 a_{mk}}. \label{mph - phi}\ ] ] which branch is the appropriate one depends on the sign of @xmath162 , but we can see that the solution is unique , as the other branch will be always negative . for
now we keep both branches .
the effective potential is defined for fixed @xmath140 through its @xmath43-derivative that can be read from the field equation for constant @xmath43 .
this allows to find it by integration with respect to @xmath43 , @xmath166 \phi_0 \ , d\phi_0 = \frac{1}{2 } \int \mathcal{m}_{ph}^2(\phi_0 ^ 2,r ) \ , d\phi_0 ^ 2 - \frac{1}{12 } \lambda_r \phi_0 ^ 4.\ ] ] using eq .
the result is @xmath167}{\lambda_r^ * } -\frac{\lambda_r \phi_0 ^ 4}{12}. \label{veff - integrated}\end{aligned}\ ] ] this is a function of both @xmath43 and @xmath140 which is well defined for all @xmath43 as long as @xmath164 , as discussed above .
the effective potential has an extreme at @xmath168 , which is the trivial solution to the field equation for constant @xmath43 .
it can be seen immediately that this must be a minimum , as the second derivative of @xmath169 at @xmath60 is @xmath170 , which must be positive in the hartree approximation .
we are interested in the question of whether there are other minima that break the @xmath29 symmetry , that is , @xmath171 . according to equation ,
this happens when @xmath172 using this condition can be recast as a quadratic equation for @xmath173 , whose solutions are @xmath174 .
\label{bar - phi}\ ] ] as mentioned above , in the hartree approximation the effective potential has always a minimum at @xmath175 .
this implies that in order for another minimum to exist at @xmath171 , there must be also a maximum at smaller value of @xmath43 , i.e. in between the two minima .
for this reason we must consider both branches of , and look for the conditions under which both solutions are real and positive .
first of all , it can be seen that @xmath176 when @xmath177 , so the conditions are , in turn , @xmath178 and @xmath179 .
these conditions can be put together in the following one @xmath180 note that imposing @xmath179 selects the upper branch in and .
so now we can look for values of the parameters that satisfy simultaneously the conditions for having a well defined potential , @xmath164 , and for symmetry breaking .
the coefficients @xmath161 , @xmath162 and @xmath160 depend on @xmath181 , @xmath80 and @xmath65 .
we plot in fig .
[ fig : regionplotnmc - mk ] the regions in the @xmath65-@xmath181 plane for which each of these conditions hold , considering both the minimally coupled case @xmath182 , as well as @xmath183 .
the @xmath184 case is qualitatively similar to the minimally coupled case , but the region where symmetry breaking is possible shrinks .
0.5 ( low density stripes ) and where the conditions for the existence of symmetry breaking solutions are met ( high density stripes ) , as functions of @xmath65 ( horizontal axis ) and @xmath185 ( vertical axis ) , for @xmath182 and @xmath186 .
the first condition is met everywhere in both cases , while the spontaneous symmetry breaking exists for small @xmath181 . the symmetry breaking region of the first plot moves further down and to the right when increasing @xmath80 to positive values.,title="fig:",scaledwidth=90.0% ] 0.5 ( low density stripes ) and where the conditions for the existence of symmetry breaking solutions are met ( high density stripes ) , as functions of @xmath65 ( horizontal axis ) and @xmath185 ( vertical axis ) , for @xmath182 and @xmath186 .
the first condition is met everywhere in both cases , while the spontaneous symmetry breaking exists for small @xmath181 . the symmetry breaking region of the first plot moves further down and to the right when increasing @xmath80 to positive values.,title="fig:",scaledwidth=90.0% ] in fig .
[ fig : veff ] we show several curves of the effective potential for fixed parameters but varying values of @xmath140 .
it can be seen that the effective potential always has a minimum in @xmath187 , while sometimes it can also have a second minimum for @xmath188 . .
in all cases @xmath189 and @xmath190 . there is always a minimum for @xmath60 , while depending on the value of @xmath181 there can also be a minimum for @xmath191 with a maximum in between.,scaledwidth=60.0% ] we note that in order to see the symmetry breaking minimum it was necessary to consider a value of @xmath112 several orders of magnitude , more than 20 , below @xmath140 .
hence , this symmetry breaking minimum might be understood as a peculiarity of approaching the massless limit .
it is also worth mentioning that this is not the usual kind of symmetry breaking / restoration scenario seen elsewhere in the literature where the symmetry is restored as @xmath140 increases , since in this case the @xmath192dependence is just the opposite .
it is important to note that the reason why the @xmath171 solutions are allowed is the presence of the @xmath65 term in , which comes as a consequence of imposing the 2pi consistency relations .
otherwise it would happen as in the @xmath2 approximation or the usual gaussian approach , where the absence of such term requires that for @xmath171 we had @xmath193 , and for that case there is no de sitter invariant vacuum @xcite . in the following section we will consider a more general choice of renormalization point , namely , de sitter spacetime .
in the previous sections we fixed the consistency conditions in minkowski space by defining the renormalized parameters at @xmath109 , and this was enough to show for example that the renormalized equations can be expressed in a @xmath106-independent way . however , it can be objected that we are fixing the consistency conditions in a background different from the one we then want to study . as it was mentioned in sec .
iii , it is not possible to impose that the consistency relations hold exactly for an arbitrary background , unless the spacetime has a constant curvature .
nevertheless , a de sitter renormalization point with scalar curvature @xmath194 may feel more natural when the spacetime of interest is of frw type , where one could fix the consistency conditions at given time by matching @xmath194 to the corresponding value of @xmath140 at that time .
the consistency conditions could then be incrementally violated as time progresses . of course ,
even if we are studying the equations in de sitter itself , although time - independent , the matching can be done exactly only if the variables @xmath140 and @xmath43 are fixed beforehand .
however , this is not the case we are considering here , as our purpose is to analyze the dependence of the physical mass and effective potential on these two variables .
therefore , the natural thing to do is to impose the consistency conditions at a given renormalization point .
another point in favor of generalizing to de sitter is that , as seen in the previous section , the minkowski renormalization point did not allow for certain values of the ms - parameters that may be of interest , such as @xmath195 .
the aim of this section is to generalize the previous results to the case where the renormalization point is taken for a fixed de sitter metric , applying it first to a general curved spacetime and then particularly to de sitter itself .
we start by considering the consistency relations for the 4-point functions , which gives a new definition of @xmath65 @xmath66}{\delta \phi_1 \delta \phi_2 \delta \phi_3 \delta \phi_4 } \bigg|_{\phi_0=0,r = r_0 } = - \lambda_r \delta_{12 } \delta_{13 } \delta_{14},\ ] ] where the notation @xmath196 implicitly implies we are evaluating in de sitter spacetime , and also we avoided the use of a different notation for @xmath65 .
the consistency relation for the 2-point functions eq .
remains unchanged . keeping this change in the definition in mind , the minimal subtraction renormalization proceeds in the same way as before and the bare couplings are found again to be given by eqs . ,
and the finite gap equation in terms of the finite ms parameters is again eq . .
the derivative of effective potential is read from the field equation as before , eq . , but now we change the definitions of the renormalized parameters by evaluating at @xmath196 , @xmath197 [ ren - quantities - dsren ] this redefinitions naturally lead to a generalization of eqs . relating the ms - parameters to the renormalized ones with an explicit dependence on @xmath194 , @xmath198}{\left [ 1 - \frac{\lambda}{32 \pi^2 } \ln \left ( \frac{m_r^2}{\tilde{\mu}^2 } \right ) \right ] } , \label{mr - mu - dsren } \\ \left ( \xi_r - \frac{1}{6 } \right ) & = \frac{\left ( \xi - \frac{1}{6 } \right ) - \frac{\lambda}{16\pi^2 } \frac{df_{ds}}{dr}\big|_{m_r^2,r_0 } } { \left[1 - \frac{\lambda}{32\pi^2 } - \frac{\lambda}{32\pi^2 } \ln \left ( \frac{m_r^2}{\tilde{\mu}^2 } \right ) \right ] } , \label{xir - mu - dsren } \\
\lambda_r & = \frac{\lambda}{\left [ 1 - \frac{\lambda}{32\pi^2 } - \frac{\lambda}{32\pi^2 } \ln \left ( \frac{m_{r}^2}{\tilde{\mu}^2 } \right ) - \frac{\lambda}{32\pi^2 } \left ( \frac{(\xi_r - \frac{1}{6 } ) r_0}{m_r^2 } - 2 \frac{df_{ds}}{d m_{ph}^2}\big|_{m_r^2,r_0 } \right ) \right]}. \label{lambdar - mu - dsren}\end{aligned}\ ] ] [ ren - mu - dsren ] the original minkowski spacetime equations are easily obtained by setting @xmath199 , which makes all the terms involving the @xmath146 function or its derivatives to vanish according to its expected properties . from now on we can use the explicit expression for @xmath146 given in eq .
, and so the previous expressions can be recast as @xmath200 } , \label{mr - mu - dsren2 } \\ \left ( \xi_r - \frac{1}{6 } \right ) & = \frac { \left ( \xi - \frac{1}{6 } \right ) + \frac{\lambda}{32\pi^2 } \left [ y_0 - \frac{1}{6 } - \left(y_0 - \frac{1}{6 } \right)^2 g'(y_0 ) - \frac{1}{18 } \right ] } { 1 - \frac{\lambda}{32\pi^2 } \left [ \ln \left ( \frac{r_0}{12\tilde{\mu}^2 } \right ) + g(y_0 ) + \left(y_0 - \frac{1}{6 } \right ) g'(y_0 ) \right ] } , \label{xir - mu - dsren2 } \\
\lambda_r & = \frac{\lambda } { 1 - \frac{\lambda}{32\pi^2 } \left [ \ln \left ( \frac{r_0}{12\tilde{\mu}^2 } \right ) + g(y_0 ) + \left ( y_0 - \frac{1}{6 } \right ) g'(y_0 ) \right]}. \label{lambdar - mu - dsren2}\end{aligned}\ ] ] [ ren - mu - dsren2 ] here we are using the notation @xmath201 , and @xmath202 . also , eqs . and are not direct equivalents of eqs . and , but rather a mixing between them that will come in handy in what follows . following the procedure outlined in minkowski space
, we can combine these equations in order to find @xmath106-independent relations between the ms - parameters and the renormalized parameters , namely @xmath203 and @xmath204 .
\label{xi - lambda - ds}\ ] ] these are as generalizations of eqs . and respectively .
in minkowski spacetime the first relation showed that the parameter @xmath103 could not be negative , while in de sitter spacetime this relation is modified in such a way that makes that case possible , although not under general conditions . more specifically , putting both of the above relations together we find that @xmath205 + \frac{1}{6}\left(\frac{1}{\lambda}-\frac{1}{\lambda_r}\right).\ ] ] then , considering that the hartree approximation demands @xmath206 , in which case it can be seen that @xmath207 , we conclude that the term in square brackets in the right hand side should be positive definite , and hence the combination of ms - parameters given by @xmath208 must be positive when @xmath209 .
we want to emphasize the importance of this result .
the validity of the consistency conditions ( which involve both finite and divergent parts ) forbids to set @xmath210 and @xmath211 simultaneously , as was done in the literature when analysing spontaneous symmetry breaking in de sitter space within the hartree approximation @xcite . as a consequence of this result
, one must allow for @xmath212 , making it unavoidable to appeal to the effective potential in order to fix the finite part of the consistency condition eq . .
carrying on with the renormalization in de sitter space , we use eqs . to rewrite the gap eq . in terms of the new renormalized parameters and @xmath146 , @xmath213 \ln \left ( \frac{m_{ph}^2}{m_r^2 } \right ) \notag \\ & & + \left ( m_{ph}^2 - m_r^2 \right ) \left [ 2 \frac{df_{ds}}{d m_{ph}^2}\big|_{m_r^2,r_0 } - \frac{(\xi_r - \frac{1}{6 } ) r_0}{m_r^2 } \right ] \label{mph - eq - dsren } \\ & & + 2\left [ f_{ds}(m_r^2,r_0 ) + \frac{df_{ds}}{dr}\big|_{m_r^2,r_0 } \left ( r - r_0 \right ) - f(m_{ph}^2,\{r\ } ) \right ] \biggr\}. \notag\end{aligned}\ ] ] setting the background metric @xmath36 to de sitter , that is , @xmath214 , and using again eq . and the variable @xmath215 , the gap equation reads @xmath216 \notag\\ & - & \left ( 1 - \frac{r_0}{r } \right ) \left(y_0 - \frac{1}{6 } \right ) + \left(y_0 - \frac{1}{6 } \right)^2 g'(y_0 ) \biggr\}.\label{gap - eq - ds } \end{aligned}\ ] ] in order to solve it for @xmath217 we consider the small mass case @xmath149 ( @xmath218 ) .
the resulting quadratic equation now has the following coefficients [ cuadeq - dsren ] with @xmath220 defined in eq . .
the analysis of this equation and its solutions now proceeds exactly the same as in minkowski spacetime , but now there is one new parameter that must be taking into account , @xmath194 .
let us first consider @xmath196 .
the first interesting fact is that the gap equation does not depend explictly on @xmath80 when @xmath196 , but rather only through the combination @xmath221 .
this simplifies the analysis by leaving only two parameters , @xmath222 and @xmath65 . in fig .
[ fig : regionplot - ds - r0 ] we show in the @xmath65-@xmath223 plane that while the effective potential can be well defined in some region ( low density stripes ) , the symmetry breaking conditions are not met .
we emphasize that there is no other free parameter when @xmath196 , as all the combinations of @xmath224 and @xmath80 such that @xmath206 are accounted for .
, the region in which the effective potential is well defined for all @xmath43 ( low density stripes ) in the @xmath65-@xmath223 plane . there is no region in which the symmetry breaking conditions are met .
there is no other free parameter.,scaledwidth=45.0% ] finally , we consider the case @xmath225 .
we show in fig .
[ fig : regionplot - ds ] plots in the @xmath226-@xmath227 plane of the regions of interest for fixed @xmath80 and @xmath228 .
note that these plots are in a different plane of the parameter space in contrast to the previous plots we analysed . in the minimally coupled case ,
both regions overlap only at @xmath229 and with @xmath224 in a certain range . in particular
this implies that in the low mass limit @xmath230 there is no possible symmetry breaking .
moreover , the effective potential is not even well defined for all @xmath43 .
note that this is different to what we have seen in fig .
[ fig : regionplotmc - mk ] in minkowski space , where the situation was @xmath231 , that is , the flat limit was taken first .
it is well known that the flat and massless limits in de sitter do not commute .
0.5 ( low density stripes ) and where the conditions for the existence of symmetry breaking solutions are met ( high density stripes ) , as functions of @xmath226 ( horizontal axis ) and @xmath227 ( vertical axis ) , for @xmath228 . for @xmath182 both regions only overlap at @xmath229 and with @xmath224 in a certain range , while for @xmath232 the effective potential is well defined everywhere and spontaneous symmetry breaking exists for small @xmath233 and @xmath224 .
the plots do not change qualitatively by varying @xmath65.,title="fig:",scaledwidth=90.0% ] 0.5 ( low density stripes ) and where the conditions for the existence of symmetry breaking solutions are met ( high density stripes ) , as functions of @xmath226 ( horizontal axis ) and @xmath227 ( vertical axis ) , for @xmath228 . for @xmath182 both regions only overlap at @xmath229 and with @xmath224 in a certain range , while for @xmath232 the effective potential is well defined everywhere and spontaneous symmetry breaking exists for small @xmath233 and @xmath224 .
the plots do not change qualitatively by varying @xmath65.,title="fig:",scaledwidth=90.0% ] if we now allow for @xmath234 , we find a different situation .
both regions can overlap for certain values of the parameters , allowing for a well defined effective potential with symmetry breaking .
some examples of such a potential are shown in fig .
[ fig : potds ] for different values of @xmath233 .
symmetry breaking occurs for small @xmath233 and then the symmetry is restored for larger @xmath233 . ,
@xmath235 and @xmath228 , for different values of @xmath233 .
symmetry breaking is exhibited for small @xmath233 , while for larger @xmath233 the symmetry is restored .
the critical value in this case is @xmath236.,scaledwidth=60.0% ]
in this section we will summarize the results obtained in this paper and will discuss the relation with previous works on the subject .
we have considered a single , selfinteracting scalar field with @xmath29 symmetry in curved spacetimes , using a nonperturbative approach based on the 2pi ea . in the lowest approximation ( a local approximation of the 2pi ea ) the formalism reproduces the usual hartree ( or gaussian ) approximation , that can also be derived using a variational approach or a resummation of a particular class of feynman diagrams . however ,
when considered in the context of the 2pi ea , there are some consistency conditions which , although automatically satisfied in the full theory , are not fulfilled for particular approximations ( this is the case when the approximation is not a systematic expansion in powers of a small parameter ) .
this fact is well known in flat spacetime , and the consistency conditions can be forced by allowing for more than one counterterm for each mass or coupling constant of the theory @xcite . our first goal in this paper has been to show that this consistent renormalization procedure " can be applied to the mean field and gap equations in general curved spacetimes .
several explicit calculations previously performed in flat spacetime used as regulator a momentum cutoff ( see for instance @xcite ) .
this is problematic in curved backgrounds , and in order to maintain the covariance of the regularized theory we used dimensional regularization .
the consistent renormalization procedure has been partially extended to curved spacetimes in some recent works @xcite , where the renormalization of the mean field and gap equations has been analyzed .
our results in sec .
iii are more complete than those in ref.@xcite .
we have included the full adiabatic expansion of the propagator , we have written the renormalized equations in terms of the physical parameters defined from the effective potential , and we have shown explicitly that , when expressed in terms of these physical parameters , the equations are independent of the regularization scale @xmath6 introduced by dimensional regularization . in the last part of the paper ( secs .
iv and v ) , we have focused on de sitter spacetime .
we have written the explicit form of the mean field and gap equations for this particular metric , and we have computed the effective potential . for this , we have considered the renormalized parameters at @xmath30 as defined from the effective potential corresponding , in turn , to minkowski spacetime and de sitter spacetime with a given curvature @xmath194 . then , we performed an analysis of the effective potential for different values of the curvature of the original de sitter spacetime @xmath140 .
this analysis is somehow analogous to the one done in @xcite for a self - interacting scalar field at finite temperature , where the renormalization point is chosen for a fixed finite value of the temperature .
a relevant conclusion of our work is that the hartree approximation and the application of the consistency conditions impose restrictions on the ms- parameters of the theory , as discussed in secs iv and v. on the one hand , the renormalized mass @xmath112 must be positive . on the other hand , the definition of the renormalized @xmath65 in terms of the fourth derivative of the effective potential and the consistency condition for the 4-point function imply @xmath6-independent relations between ms and renormalized parameters . from them , it is clear that some choices of the ms - parameters may not be compatible with the consistency conditions . in particular
, one can not take @xmath120 , @xmath237 , and @xmath121 simultaneously .
our results show that the @xmath29 symmetry can be spontaneously broken when one uses the consistent renormalization , although not under general conditions .
this is to be contrasted with previous results obtained using a standard renormalization of the theory , as can be easily seen from the gap and mean value equations . indeed , when the consistency relations are not taken into account and the presence of different counterterms is not allowed , eqs .
( [ field - eq ] ) and ( [ gap - eq ] ) become @xcite @xmath238 ( note that the extra term in eq .
( [ field - eq ] ) comes from the consistency conditions ) .
if there were spontaneous symmetry breaking in de sitter spacetime , one would have a solution to these equations with a constant and non - vanishing @xmath43 . if this were the case , one should have @xmath239 however , it is well known that there is no de sitter invariant propagator for a minimally coupled and massless scalar field , and therefore a constant @xmath240 can not be a solution of eqs .
( [ field - eq - usual ] , [ gap - eq - usual ] ) .
thus , there is no spontaneous symmetry breaking when one uses the standard renormalization .
a similar situation holds in the large @xmath3 limit of @xmath5 theories . in this case , the consistency relations are satisfied automatically order by order in powers of @xmath2 , and there is no need for additional counterterms . the mean field and gap equations
are once more given by eqs .
( [ field - eq - usual ] ) and ( [ gap - eq - usual ] ) @xcite , and the same argument applies @xcite . the results of this paper are not conclusive about the occurrence of spontaneous symmetry breaking of the @xmath29 symmetry .
it is not clear whether the existence of solutions with @xmath240 is an artifact of the hartree approximation or not .
it is plausible that the inclusion of the setting sun diagram in the computation of the 2pi ea restores the @xmath29 symmetry .
we hope to address this interesting issue in a forthcoming work . however , there are several technical complications to be sorted . on the one hand , the use of the ctp formalism will be unavoidable when considering nonlocal terms in the 2pi ea .
on the other hand , the inclusion of higher loops in the 2pi ea induces some subtle points in the renormalization even in flat spacetime @xcite , that will have their counterpart in curved spaces .
finally , it is important to show that this consistent renormalization procedure can be extended for general curved spacetimes , to make finite not only the mean field and gap equations of the matter sector of the theory , but also the gravitational sector .
this is the main goal of the forthcoming paper ii @xcite .
this research was supported in part by anpcyt , conicet and uba . f.d.m and l.g.t would like to thank the hospitality of ictp , where part of this work has been done .
we would like to thank guillem prez nadal for useful discussions at the initial stages of this work .
in this section we briefly review the derivation of the consistency conditions we used to establish a relation between the different counterterms , which are eqns .. ( [ 2-pt - relation ] ) and ( [ 4-pt - relation ] ) .
let us start by recalling that the 1pi resummed effective action is obtained by evaluating the 2pi ea in the solution @xmath241 of the gap equation , @xmath242 = \gamma_{2pi}[\phi_0,\bar{g}(\phi_0 ) ] .
\label{gamma-1pi}\ ] ] from the gap equation is , formally , @xmath243 or , equivalently , @xmath244 using the definition of the self - energy @xmath245 the first derivative of @xmath246 $ ] with respect to @xmath43 , which equated to zero gives the field equation , can be written as @xmath247}{\delta \phi_1 } \bigg|_{\bar{\phi_0 } } = \frac{\delta \gamma_{2pi}}{\delta \phi_1 } + \frac{\delta \gamma_{2pi}}{\delta g_{ab } } \bigg|_{\bar{g } } \ , \frac{\delta \bar{g}_{ab}}{\delta \phi_1 } = \frac{\delta \gamma_{2pi}}{\delta \phi_1 } \nonumber\\ & = & i g_{0,1a}^{-1 } \phi_a + \frac{\delta \gamma_{int}}{\delta \phi_1 } \bigg|_{\bar{g } } , \label{1-pt } \end{aligned}\ ] ] where in the third equality we used eq .
( [ gaux ] ) . as mentioned in the main text ,
there are multiple definitions of a given @xmath4-point function .
we focus here on the two- and four - point functions . on one side
, we have the two- and four - point functions that can be obtained by taking functional derivatives of @xmath248 $ ] with respect to @xmath43 : @xmath249}{\delta \phi_1 \delta \phi_2 } \bigg|_{\bar{\phi_0}},\ ] ] @xmath250}{\delta \phi_1 \delta \phi_2 \delta \phi_3 \delta \phi_4 } \bigg|_{\bar{\phi_0}}.\ ] ] functionally deriving eq .
, we obtain the following expression for the 2-point function @xmath251}{\delta \phi_1 \delta \phi_2 } \bigg|_{\bar{\phi_0 } } = i g_{0,12}^{-1 } + \frac{\delta^2 \gamma_{int}}{\delta \phi_1 \delta \phi_2 } \bigg|_{\bar{g } } + \frac{\delta^2 \gamma_{int}}{\delta \phi_1 \delta g_{ab } } \bigg|_{\bar{g } } \frac{\delta \bar{g}_{ab}}{\delta \phi_2}.\ ] ] using that @xmath252 and that from eq .
we have @xmath253 we can write @xmath251}{\delta \phi_1 \delta \phi_2 } \bigg|_{\bar{\phi_0 } } = i g_{0,12}^{-1 } + \frac{\delta^2 \gamma_{int}}{\delta \phi_1 \delta \phi_2 } \bigg|_{\bar{g } } + \frac{\delta^2 \gamma_{int}}{\delta \phi_1 \delta g_{ab } } \bigg|_{\bar{g } } \bar{g}_{ac } \bar{g}_{bd } \frac{\delta \bar{\sigma}_{cd}}{\delta \phi_2}. \label{2-pt}\ ] ] moreover , one can show that @xmath254 where @xmath255 is defined by @xmath256 hence , @xmath257 satisfies a self - consistent equation . in a similar way , one can define a 4-point vertex function as the self - consistent solution of the following equation @xmath258 which in matrix form reads @xmath259 where the last equality follows from the symmetry properties of @xmath260 ( which are the same as @xmath261 ) and @xmath262 . in order to obtain the 4-point function one should take two more derivatives , however , this would become very cumbersome .
therefore , let us restrict ourselves to the case of @xmath29 symmetric theories where a simplification occurs . in this case ,
@xmath4-point functions with odd n vanish at @xmath30 , for instance @xmath263 taking this into account , the 4-point function results @xmath264 from eq . , using the @xmath29 symmetry , we obtain a self - consistent equation @xmath265 whose solution can be expressed in terms of the function @xmath266 , solution of , @xmath267 where we used the definition @xmath268 then , the 4-point function can be written as @xmath269,\nonumber\end{aligned}\ ] ] or , in matrix form @xmath270 .
\label{4-pt - matr}\ ] ] in the exact theory , the following relation is satisfied : @xmath271 which is the first consistency relation we used . differentiating with respect to @xmath19
, we obtain @xmath272 that is @xmath273 given this , the 2-point function results @xmath274 where the gap equation was used to arrive at the last equality . in a similar way , but now for the 4-point function , using and we obtain @xmath275 \notag \\ & = & \frac{\delta^4 \gamma_{int}}{\delta \phi^4 } \bigg|_{\bar{g } } + \frac{i}{2 } \bar{v } \bar{g}^2 \bar{\lambda}^{\dagger } + perm \\ & = & \frac{\delta^4 \gamma_{int}}{\delta \phi^4 } \bigg|_{\bar{g } } + \left [ \bar{v } - \bar{\lambda } + perm \right]\end{aligned}\ ] ] or , more explicitly @xmath276 .
\label{4-pt - relation2}\ ] ] this last equation is the second consistency condition we were looking for . in order to implement this ,
it is necessary to consider another relation which is valid for the exact theory : @xmath277 then the previous expression can be rewritten as @xmath278 - \frac{1}{2 } \frac{\delta^4 \gamma_{int}}{\delta \phi_1 \delta \phi_2 \delta \phi_3 \delta \phi_4 } \bigg|_{\bar{g},\phi_0=0 } , \,\,\,\,\,\ , % 2 \left [ \frac{\delta^2 \gamma_{int}}{\delta g_{12 } \delta g_{34 } } \bigg|_{\bar{g } } + \frac{\delta^2 \gamma_{int}}{\delta g_{13 } \delta g_{24 } } \bigg|_{\bar{g } } + \frac{\delta^2 \gamma_{int}}{\delta g_{14 } \delta g_{23 } } \bigg|_{\bar{g } } \right ] , \label{4-pt - relation3}\end{aligned}\ ] ] where the symmetry properties of @xmath279 where used .
this is the second consistency relation we used .
in this section we summarize the calculation of the renormalized parameters defined from the effective potential , and we compute their relation with the ms finite parameters .
for @xmath112 we use together with and , giving @xmath280 or , equivalently @xmath281}. \label{mr - mu2}\ ] ] next , for @xmath80 , we see from that we must impose the condition @xmath282 . for that we take the @xmath140-derivative of @xmath283 \ln \left ( \frac{m_{ph}^2}{\tilde{\mu}^2 } \right)\right .
\nonumber\\ & + & \left.\left [ m_{ph}^2 + \left ( \xi_r -\frac{1}{6 } \right ) r \right ] \frac{1}{m_{ph}^2 } \frac{dm_{ph}^2}{dr } - 2 \frac{df}{dr}\right\}.\end{aligned}\ ] ] we can now evaluate for @xmath60 and @xmath109 , use and , and impose the condition previously mentioned .
this leads to @xmath284 or , with some algebra @xmath285 } .
\label{xir - mu2}\ ] ] finally for @xmath65 we must take two @xmath43-derivatives of , @xmath286 \frac{1}{m_{ph}^2}\nonumber\right.\\ & - & \left .
2 \frac{df(m_{ph}^2,\{r\})}{d m_{ph}^2 } \right\ } \frac{dm_{ph}^2}{d\phi_0}.\end{aligned}\ ] ] before taking the second derivative , we evaluate this expression for @xmath60 and @xmath109 as it will be needed later @xmath287 = 0\ ] ] which implies @xmath288 .
now , deriving once more @xmath289 \frac{1}{m_{ph}^2 } \right.\nonumber\\ & -&\left .
2 \frac{df(m_{ph}^2,\{r\})}{d m_{ph}^2 } \right\ } \frac{d^2 m_{ph}^2}{d\phi_0 ^ 2 } + \left [ \dots \right ] \frac{d m_{ph}^2}{d\phi_0},\end{aligned}\ ] ] then evaluating for @xmath60 and @xmath109 , using the previous result and we arrive at @xmath290}.\ ] ]
here we provide some details of the calculation of the function @xmath90 for de sitter spacetime .
we must expand the de sitter coincident propagator for @xmath87 and fixed @xmath140 , and for this reason we rewrite @xmath136 as @xmath291 .
then expanding for @xmath87 : @xmath82 & = & \frac{r}{96 \pi^2 } \frac{\gamma \left ( \frac{3}{2 } + \nu_4 \right ) \
, \gamma \left ( \frac{3}{2 } - \nu_4 \right)}{\gamma \left ( \frac{1}{2 } + \nu_4 \right ) \ , \gamma \left ( \frac{1}{2 } - \nu_4 \right ) } \biggl\ { \frac{2}{\epsilon } - \frac{13}{6 } + \gamma_e + \ln \left ( \frac{r}{48\pi \mu^2 } \right ) + \psi\left ( \frac{3}{2 } + \nu_4 \right ) + \psi\left ( \frac{3}{2 } - \nu_4 \right ) \nonumber\\ & + & 2 \left [ \psi\left ( \frac{1}{2 } - \nu_4 \right ) - \psi\left ( \frac{1}{2 } + \nu_4 \right ) + \psi\left ( \frac{3}{2 } + \nu_4 \right ) - \psi\left ( \frac{3}{2 } - \nu_4 \right ) \right ] \frac{d\tilde{\nu}_n}{dn}\bigg|_{n=4 } \biggr\ } + \mathcal{o}(\epsilon ) \end{aligned}\ ] ] here @xmath144 is the digamma function , @xmath92 is an arbitrary mass scale introduced to maintain the usual units as @xmath292 .
furthermore , although @xmath293 , the derivatives @xmath294 . from now on @xmath91 can be set to @xmath111 in the non - divergent terms . using the properties of the gamma and digamma functions we can simplify the pre - factor and the expression between square brackets , leading to @xmath135 = \frac{r}{96 \pi^2 } \biggl\ { \left ( \frac{1}{4 } - \nu_4 ^ 2 \right ) \left [ \frac{2}{\epsilon } - \frac{13}{6 } + \gamma_e + \ln \left ( \frac{r}{48\pi \mu^2 } \right ) + \psi_{+ } + \psi_{- } \right ] - 4 \nu_4 \frac{d\tilde{\nu}_n}{dn}\bigg|_{n=4 } \biggr\}\ ] ] where we use the condensed notation @xmath295 . now replacing for @xmath296 and @xmath297 we obtain @xmath82 & = & \frac{1}{8 \pi^2 }
\biggl\ { \left [ m_{ph}^2 + \left ( \xi_r - \frac{1}{6 } \right ) r \right ] \left [ \frac{2}{\epsilon } - \frac{13}{6 } + \gamma_e + \ln \left ( \frac{r}{48\pi \mu^2 } \right ) + \psi_{+ } + \psi_{- } \right ] \nonumber\\ & - & \frac{r}{4 } + \frac{7}{6 } \left ( m_{ph}^2 + \xi_r r \right ) \biggr\ } \end{aligned}\ ] ] or , rearranging slightly and substituting @xmath298 , @xmath82 & = & \frac{1}{4 \pi^2 \epsilon } \left [ m_{ph}^2 + \left ( \xi_r - \frac{1}{6 } \right ) r \right ] + \frac{1}{8 \pi^2 } \biggl\ { \left [ m_{ph}^2 + \left ( \xi_r - \frac{1}{6 } \right ) r \right ] \nonumber\\ & \times&\left [ \ln \left ( \frac{r}{12 \tilde{\mu}^2 } \right ) + \psi_{+ } + \psi_{- } \right ] - \frac{r}{18 } \biggr\}\end{aligned}\ ] ] in this form this expression is directly comparable to the first line of eq . .
the function @xmath90 for de sitter spacetime , @xmath299 \left [ \ln \left ( \frac{r}{12 m_{ph}^2 } \right ) + \psi_{+ } + \psi_{- } \right ] \nonumber\\ & + & \frac{1}{2 } \left ( \xi_r - \frac{1}{6 } \right ) r + \frac{r}{36 } \end{aligned}\ ] ] where @xmath143 .
this function has all the expected properties , that is , it is written only in terms of renormalized parameters , it is independent of @xmath91 and @xmath106 , and it satisfies the correct limits , and . to verify these limits
it is useful to consider that @xmath300 & = 0 \\ \lim_{r \to 0 } \left [ m_{ph}^2 + \left ( \xi_r - \frac{1}{6 } \right ) r \right ] \left [ \frac{1}{r } + \left ( \psi_{+}^ { ' } - \psi_{-}^ { ' } \right ) \frac{d\tilde{\nu}_4}{dr } \right ] & = \left ( \xi_r - \frac{1}{6 } \right ) + \frac{1}{18}\end{aligned}\ ] ] s. weinberg , phys .
d * 72 * , 043514 ( 2005 ) [ arxiv : hep - th/0506236 ] ; ibidem phys .
d * 74 * , 023508 ( 2006 ) [ arxiv : hep - th/0605244 ] .
n. c. tsamis and r. p. woodard , nucl .
* b724 * , 295 ( 2005 ) [ arxiv : gr - qc/0505115 ] .
m. van der meulen and j. smit , jcap * 0711 * , 023 ( 2007 ) [ arxiv:0707.0842 [ hep - th ] ] .
d. seery , jcap * 0711 * , 025 ( 2007 ) [ arxiv:0707.3377v3 [ astro - ph ] ] ; ibidem jcap * 0802 * , 006 ( 2008 ) [ arxiv:0707.3378 [ astro - ph ] ] .
d. seery , class .
quantum grav .
* 27 * , 124005 ( 2010 ) [ arxiv:1005.1649 [ astro-ph.co ] ] .
t. tanaka and y. urakawa , arxiv:1306.4461 [ hep - th ] .
burgess , l. leblond , r. holman , and s. shandera , jcap * 1003 * , 033 ( 2010 ) [ arxiv:0912.1608 [ hep - th ] ] ; ibidem jcap * 1010 * , 017 ( 2010 ) [ arxiv:1005.3551 [ hep - th ] ] .
b. garbrecht and g. rigopoulos , phys .
d * 84 * , 063516 ( 2011 ) [ arxiv:1105.0418 [ hep - th ] ] .
a. rajaraman , phys .
d * 82 * , 123522 ( 2010 ) [ arxiv:1008.1271 [ hep - th ] ] .
s. hollands , ann .
h. poincar * 13 * , 1039 ( 2012 ) [ arxiv:1105.1996 [ gr - qc ] ] .
m. beneke and p. moch , phys .
d * 87 * , 064018 ( 2013 ) [ arxiv:1212.3058 [ hep - th ] ] .
a. a. starobinsky and j. yokoyama , phys .
d * 50 * , 6357 ( 1994 ) .
l. senatore and m. zaldarriaga , jhep * 1012 * , 008 ( 2010 ) [ arxiv:0912.2734 [ hep - th ] ] .
g. l. pimentel , l. senatore and m. zaldarriaga , jhep * 1207 * , 166 ( 2012 ) [ arxiv:1203.6651 [ hep - th ] ] .
l. senatore and m. zaldarriaga , arxiv:1210.6048 [ hep - th ] .
t. tanaka and y. urakawa , arxiv:1301.3088 [ hep - th ] .
v. assassi , d. baumann , and d. green , jhep * 02 * , 151 ( 2013 ) [ arxiv:1210.7792 [ hep - th ] ] . s. p. miao and r. p. woodard , jcap * 07 * , 008 ( 2012 ) [ arxiv:1204.1784 [ astro-ph.co ] ] .
e. dimastrogiovanni and n. bartolo , jcap * 0811 * , 016 ( 2008 ) [ arxiv:0807.2790 [ astro - ph ] ] ; n. bartolo , e. dimastrogiovanni , and a. vallinotto , jcap * 1011 * , 003 ( 2010 ) [ arxiv:1006.0196 [ astro-ph.co ] ] .
g. lazzari and t. prokopec , arxiv:1304.0404 [ hep - th ] ; t. prokopec , jcap * 1212 * , 023 ( 2012)[arxiv:1110.3187v2 [ gr - qc ] ] .
e. t. akhmedov , jhep * 1201 * , 066 ( 2012 ) [ arxiv:1110.2257 [ hep - th ] ] ; e. t. akhmedov amd ph .
burda , phys .
d * 86 * , 044031 ( 2012 ) [ arxiv:1202.1202 [ hep - th ] ] ; e. t. akhmedov , f. k. popov , and v. m. slepukhin , arxiv:1303.1068 [ hep - th ] .
a. riotto and m. s. sloth , jcap * 0804 * , 030 ( 2008 ) [ arxiv:0801.1845 [ hep - ph ] ] .
d. boyanovsky , phys .
d * 85 * , 123525 ( 2012 ) [ arxiv:1203.3903 [ hep - ph ] ] .
a. youssef and d. kreimer , arxiv:1301.3205 [ gr - qc ] .
t. arai , class .
* 29 * , 215014 ( 2012 ) ; ibidem , phys .
d * 86 * , 104064 ( 2012 ) ; ibidem , arxiv:1304.5631 [ hep - th ] .
r. parentani and j. serreau , phys .
d * 87 * , 045020 ( 2013 ) [ arxiv:1212.6077 [ hep - th ] ] ; ibidem , arxiv:1302.3262 [ hep - th ] .
j. serreau , phys .
. lett . * 107 * , 191103 ( 2011 ) [ arxiv:1105.4539 [ hep - th ] ] ; ibidem , arxiv:1306.3846 [ hep - th ] .
b. ratra , phys .
d * 31 * , 1931 ( 1985 ) .
j. m. cornwall , r. jackiw and e. tomboulis , phys .
d * 10 * , 2428 ( 1974 ) .
p. m. stevenson , phys .
d * 32 * , 1389 ( 1985 ) .
p. m. stevenson and r. tarrach , phys .
b176 * , 436 ( 1986 ) ; p. m. stevenson , b. alles , and r. tarrach , phys .
d * 35 * , 2407 ( 1987 ) ; p. m. stevenson , z. phys .
c * 35 * , 467 ( 1987 ) .
f. d. mazzitelli and j. p. paz , phys . rev .
d * 39 * , 2234 ( 1989 ) .
g. amelino - camelia , s. -y .
pi , phys .
d * 47 * , 2356 ( 1993 ) .
j. berges , sz .
borsanyi , u. reinosa , and j. serreau , annals phys . * 320 * , 344 ( 2005 ) .
f. cooper , j. f. dawson , and b. mihaila , phys .
d * 67 * , 056003 ( 2003 ) ; j. r. espinosa , m. quiros , and f. zwirner , phys
b291 * , 115 ( 1992 ) .
n. tetradis and c. wetterich , nucl . phys . *
b398 * , 659 ( 1993 ) .
g. marko , u. reinosa , and zs .
szep , phys .
d * 86 * , 085031 ( 2012 ) ; ibidem , arxiv 1303.0230 [ hep - ph ] .
d. l. lpez nacir , f. d. mazzitelli , and l. g. trombetta , in preparation .
e. calzetta and b. l. hu , nonequilibrium quantum field theory , " cambridge university press ( 2008 ) .
s. a. ramsey and b. l. hu , phys .
d * 56 * , 661 ( 1997 ) [ arxiv : gr - qc/9706001 ] .
paz and f. d. mazzitelli , phys .
d * 37 * , 2170 ( 1988 ) . u.reinosa and z. szep , phys .
d * 83 * , 125026 ( 2011 ) [ arxiv:1103.2689 [ hep - ph ] ] . | we consider a quantum scalar field with @xmath0 interaction in curved spacetimes .
the quantum effects are taken into account nonperturbatively using the hartree approximation to the 2pi effective action .
although this approximation has been considered in many previous works , we reconsider it using a consistent nonperturbative renormalization procedure , which we extend to general curved spacetimes .
we obtain the renormalized equations for the mean field and for the propagator of the fluctuations , showing explicitly their independence on the arbitrary scale introduced by the regularization scheme .
we apply our results to the particular case of de sitter spacetime and discuss spontaneous symmetry breaking .
the results depend strongly on the renormalization procedure . |
a relativistic treatment is one of the essential ingredients that should be incorporated in describing hadronic systems .
the hadrons have an intrinsically relativistic nature since the quantum chromodynamics ( qcd ) governing the quarks and gluons inside the hadrons has _ a priori _ a strong interaction coupling and the characteristic momenta of quarks and gluons are of the same order , or even very much larger , than the masses of the particles involved .
it has also been realized that a parametrization of nuclear reactions in terms of non - relativistic wave functions must fail . in principle , a manifestly covariant framework such as the bethe - salpeter approach and its covariant equivalents can be taken for the description of hadrons . however , in practice , such tools are intractable because of the relative time dependence and the difficulty of systematically including higher order kernels .
a different and more intuitive framework is the relativistic hamiltonian approach . with the recent advances in the hamiltonian renormalization program , a promising technique to impose
the relativistic treatment of hadrons appears to be light - front dynamics ( lfd ) , in which a fock - space expansion of bound states is made at equal light - front time @xmath2 . the reasons that make lfd so attractive to solve bound - state problems in field theory
make it also useful for a relativistic description of nuclear systems .
light - front quantization @xcite has already been applied successfully in the context of current algebra @xcite and the parton model @xcite in the past . for the analysis of exclusive processes involving hadrons , the framework of light - front ( lf ) quantization @xcite is also one of the most popular formulations . in particular , the light - front or drell - yan - west ( @xmath3 )
frame has been extensively used in the calculation of various electroweak form factors and decay processes @xcite . in this frame
@xcite , one can derive a first - principle formulation for the exclusive amplitudes by choosing judiciously the component of the light - front current .
as an example , only the parton - number - conserving ( valence ) fock state contribution is needed in @xmath4 frame when a good " component of the current , @xmath5 or @xmath6 , is used for the spacelike electromagnetic form factor calculation of pseudoscalar mesons .
one does nt need to suffer from complicated vacuum fluctuations in the equal-@xmath2 formulation due to the rational dispersion relation .
the zero - mode contribution may also be avoided in drell - yan - west ( dyw ) frame by using the plus component of current @xcite . the perturbative qcd ( pqcd ) factorization theorem for the exclusive amplitudes at asymptotically large momentum transfer
can also be proved in lfd formulated in the dyw frame .
however , caution is needed in applying the established drell - yan - west formalism to other frames because the current components do mix under the transformation of the reference - frame @xcite . especially , for the spin systems , the light - front helicity states are in general different from the ordinary ( or canonical ) helicity states which may be more appropriate degrees of freedom to discuss the angular momentum conservation .
as the spin of the system becomes larger , the number of current matrix elements gets larger than the number of physical form factors and the conditions that the current matrix elements must satisfy are essential to test the underlying theoretical model for the hadrons .
thus , it is crucial to find the relations between the ordinary helicity amplitudes and the light - front helicity amplitudes in the frame that they are computed . in this work
, we use the general collinear frames which cover both breit and target - rest frames to find the relations between the ordinary helicity amplitudes and the light - front helicity amplitudes . using the conservation of angular momentum
, we derive a general angular condition which can be applied for any spin system .
the relations among the light - front helicity amplitudes are further constrained by the light - front parity and time - reversal consideration .
for example , the spin-1 form factor analysis requires in general nine light - front helicity amplitudes although there are only three physical form factors .
thus , there must be six conditions for the helicity amplitudes . using the light - front parity relation
, one can reduce the number of helicity amplitudes down to five .
the general angular condition gives one relation among the five light - front helicity amplitudes , leaving four of them independent .
one more relation comes by applying the light - front time - reversal relation , also having the effect that the general anglar conditon can be reduced to the usual angular condition relating only four helicity ampituds .
consequently , only three helicity amplitudes are independent each other , as it should be because there are only three physical form factors in spin-1 systems .
we also apply the general angular condition to the electromagnetic transition between spin-1/2 and spin-3/2 systems and find the relation among the helicity amplitudes that can be used in the analysis of the n-@xmath1 transition .
in particular , the angular condition provides a strong constraint to the n-@xmath1 transition indicating that the suppression of the helicity flip amplitude with respect to the helicity non - flip amplitude for the momentum transfer @xmath7 in pqcd is in an order of @xmath8 or @xmath9 rather than @xmath10 , where both nucleon mass @xmath11 and delta mass @xmath12 are much larger than the qcd scale @xmath13 .
thus , one may expect that the applicability of leading pqcd could be postponed to a larger @xmath0 region than one may naively anticipate from leading pqcd .
the same consideration can apply for the deuteron form factor analysis from the spin-1 angular condition .
this work presents further discussions on the scaling law and the subleading power corrections in the breit and light - front frames .
the paper is organized as follows . in the next section ( section ii ) , we present the derivation of transformation laws between the ordinary helicity amplitudes and the light - front helicity amplitudes and obtain a general angular condition on the current matrix elements using the rotational covariance of the current operator . since we start from the definition of states in a general collinear frame , our derivation may be more physically transparent than any other formal derivation . in section iii
, we present the light - front discrete symmetries and derive the parity and time - reversal rules for the helicity amplitudes . in section iv , we discuss the consequences from these findings of the general angular condition and the light - front discrete symmetry relations . the reduction of number of independent helicity amplitudes is shown for a few example spin systems .
the current matrix elements of the spin-1 system and the spin-1/2 to spin-3/2 transition are shown as explicit examples .
the subleading power corrections are obtained from the general angular condition and the scaling laws are derived for the ordinary breit frame helicity amplitudes and the light - front helicity amplitudes .
summary and conclusion follow in section v.
our subject is relations among matrix elements or helicity amplitudes for the process @xmath14 , where @xmath15 is an off - shell photon of momentum @xmath16 , and @xmath17 and @xmath18 are hadrons with momenta @xmath19 and @xmath20 , respectively .
( results will be easily extendable for other incoming vector bosons . )
calculations may be done in the light - front frame , which is characterized by having @xmath21 , and may be done in the breit frame , which is characterized by having the photon and hadron 3-momenta along a single line .
each frame has its advantages . in the light front frame with @xmath22 , and for matrix elements of the current component @xmath5 , the photon only couples to forward moving constituents ( quarks ) of the hadrons and
never produces a quark - antiquark pair .
thus one only needs wave functions for hadrons turning into constituents going forward in ( light - front ) time , and can develop a simple parton picture of the interaction . on the other hand , the breit frame , being a collinear frame , makes it easy to add up the helicities of the incoming and outgoing particles and to count the number of independent non - zero amplitudes . by transforming
efficiently back and forth one can realize the advantages of both frames .
hence our first goal in this section will be to find the relations between the light - front and breit frame helicity amplitudes , and then to use those relations to derive in a transparent way the general relation among the light - front amplitudes that is usually referred to as the `` angular condition . '' connecting light - front and breit helicity amplitudes is facilitated by finding frames that are both simultaneously realized .
one excellent and easy example is the particular light - front frame where the target is at rest .
this is also a breit frame , since with the target 3-momentum zero , the remaining momenta must lie along the same line .
we are perhaps extending the idea of a breit frame , but are doing so in a way that leaves invariant the breit frame helicity amplitude .
that is , one normally thinks of a breit frame as one where the incoming and outgoing hadron have oppositely directed momenta , along the same line .
sometimes one specifies that the line is the @xmath23-axis .
however , since helicities are unaffected by rotations @xcite , one can choose any line at all .
further , helicities are unaffected by collinear boosts @xcite that do nt change the particle s momentum direction .
one can also boost along the direction of motion until one of the hadrons is at rest , provided one defines the positive helicity direction for the particle at rest to be parallel to the momentum the particle would have in conventional breit frame . with this natural helicity direction choice ,
the breit frame helicity amplitude in a target rest frame is ( if we use relativistic normalization conventions , as we shall always do ) precisely the same as the breit frame amplitude in a conventional breit frame with the same helicity labels .
thus , the light - front frame with the target at rest is both a breit frame and a light - front frame .
there is a continuum of such frames .
another useful example is a breit frame with the incoming and outgoing hadrons moving in the negative and positive @xmath24-directions , adjusted to have equal incoming and outgoing energies . in this case ,
@xmath25 and @xmath26 are individually zero , so that @xmath4 and we have also a light - front frame . even in a frame that is simultaneously light - front and breit , the connection between the two types of helicity amplitudes can be a bit involved .
this is because the definitions of the light - front and ordinary helicity states are not the same and the general conversion between them for a moving state involves a rotation by an angle that is not trivial to determine . our plan will be to use the rest of this subsection to define our notation , state the main result for the light - front to breit and vice - versa helicity amplitude conversion formulas , and show how one obtains the general angular condition from this result .
then in the next subsection , we will give the details of the derivation . for light - front amplitudes
one uses light - front helicity states , which for a momentum @xmath19 are defined by taking a state at rest with the spin projection along the @xmath23-direction equal to the desired helicity , then boosting in the @xmath23-direction to get the desired @xmath27 , and then doing a light - front transverse boost to get the desired transverse momentum @xmath28 .
we call this state @xmath29 and it is defined by formula in the next subsection
. the spin of the particle , @xmath30 , is understood but not usually written , @xmath31 is the light - front helicity of the particle , and the normalization is @xmath32 the light - front helicity amplitude @xmath33 is a matrix element of the electromagnetic current @xmath34 given by @xmath35 in the breit frame , we use ordinary helicity states , which are defined by starting with a state at rest having a spin projection along the @xmath23-direction equal to the desired helicity , then boosting in the @xmath23-direction to get the desired @xmath36 , and then rotating to get the momentum and spin projection in the desired direction .
( we shall generally keep our momenta in the @xmath24-@xmath23 plane , so we do not need to worry about the distinction between , for example , the jacob - wick @xcite helicity states and the somewhat later wick states @xcite . )
the state will be denoted , @xmath37 where @xmath38 is the helicity or spin projection in the direction of motion , and the subscript `` @xmath39 '' reminds us which frame we use these states in . except for momenta directly along the @xmath23-axis , the light - front helicity and regular helicity states are not the same , but if the 4-momenta are the same they can be related by a rotation .
the breit frame helicity amplitude @xmath40 is , @xmath41 where @xmath34 is the same electromagnetic current .
a main result is the relation between @xmath33 and @xmath40 , which is @xmath42 a sum on repeated helicity indices is implied .
the @xmath43-functions are the usual representations of rotations about the @xmath44-axis for particles whose spins are given by the superscript .
the angles are given by @xmath45 and @xmath46 where @xmath11 is the mass of the incoming hadron , @xmath12 is the mass of the outgoing hadron , @xmath47 , @xmath48 and @xmath49 and we assume that @xmath50 is spacelike ( negative , in our metric ) . for the elastic case , @xmath51 , the angles @xmath52 and @xmath53 are the same .
( it may seem peculiar to have a minus sign inserted twice , as it appears in eqs .
( [ main ] ) and ( [ thetaprime ] ) , but it will seem more sensible when one sees how the angles arise , in the next subsection . ) in the breit frame , since it is collinear , the sum of spin projections along the direction of motion must be conserved , so that if @xmath54 is the helicity of the photon , @xmath55 even if the photon is off - shell , it can not have more than one unit of helicity in magnitude . hence there is a constraint on the breit frame helicity amplitude , @xmath56 this induces a constraint on the light front amplitudes , and this constraint is what is called the angular condition , given generally by @xcite @xmath57 and most often applied when the lorentz index @xmath58 is @xmath59 .
one sees that the result follows from angular momentum conservation and the limited helicity of the photon .
we will check in section [ consequences ] that upon expressing the @xmath43-functions in terms of @xmath0 and mass , one obtains the known angular condition for electron - deuteron elastic scattering , and will we also obtain in terms of masses and @xmath0 the angular condition for the n-@xmath60 electromagnetic transition .
we gave two examples of frames that were simultaneously breit and light - front frames .
it turns out that half the work we need to do is very easy in one of these frames , and that visualizing one ensuing equality is quite easy in the other .
clearly , one can write @xmath61 so that the problem reduces to finding the overlaps of the light - front helicity and ordinary helicity amplitudes . we shall start using the light - front frame with the target at rest .
for the initial state , being at rest , the light - front helicity state is identical to the state with spin quantized in the positive @xmath23-direction , @xmath62 the helicity state , however , should be quantized along a direction antiparallel to the momentum of the entering photon ; see fig .
the photon four - momentum is @xmath63 and it makes an angle @xmath52 with the @xmath23-axis , where @xmath64 equivalent to the half - angle version given earlier , eq .
( [ thetahalf ] ) . with
@xmath52 taken positive , it is also the rotation angle from the breit frame helicity state to the light - front state , @xmath65 which leads to @xmath66 for the outgoing hadron , the helicity state is ( see fig .
[ one ] to get the angle ) , @xmath67 where @xmath68 is the boost operator for the @xmath23-direction and @xmath69 is a rapidity given in terms of the energy @xmath70 of the outgoing hadron , @xmath71 the light - front state is given by first boosting to the correct @xmath72 for outgoing hadron ( a boost in the negative @xmath23-direction , if @xmath73 ) , followed by a boost to get the correct transverse momentum , which is the same as for the photon .
one has @xmath74 where @xmath75 and @xmath76 is the light - front transverse boost @xmath77 thus the overlap is @xmath78 where we know the product of the four operators can only be a rotation because the rest four momentum is undisturbed .
consistent with our previous choice , we define @xmath79 as the angle rotating from the rest state connected to the breit frame helicity state to the corresonding state connected to the light - front state .
our method for finding @xmath79 is to choose a representation for the operators , namely @xmath80 where the @xmath81 are the usual @xmath82 pauli matrices , and then to multiply the operators out explicitly .
the result is @xmath83 equivalent to the useful half - angle version given earlier , eq .
( [ thetaprime ] ) . putting the pieces together gives the light - front to breit frame helicity amplitude conversion formula , quoted in eq .
( [ main ] ) .
the inverse of this relation follows using @xmath84 and is @xmath85
the discrete symmetries of parity inversion and time reversal are not compatible with the light - front requirement that @xmath22 .
however , putting all momenta in the @xmath24-@xmath23 plane , we can compound the usual parity and time reversal operators with 180@xmath86 rotations about the @xmath44-axis to produce useful and applicable light - front parity and time reversal operators @xcite .
let @xmath87 be the ordinary unitary parity operator that takes @xmath88 and @xmath89 .
define the light - front parity operator by @xcite @xmath90 since @xmath91 commutes with operators @xmath76 and @xmath68 , one has that @xmath91 acting on a light - front state gives @xmath92 further , @xmath93 where @xmath94 is the intrinsic parity of the state . then using @xmath95 , one gets for the states @xmath96 for current component @xmath5 , since @xmath91 is unitary , @xmath97 hence
, the parity relation for light - front helicity amplitudes is @xmath98 the parity relation for the usual ( breit frame ) helicity ampitudes is known @xcite , and is usually given in terms of amplitudes with definite photon helicity , which we define in section [ equivalence ] .
we shall only note that we can derive the relation from the light - front result just above , and quote for completeness , @xmath99 let @xmath100 be the ordinary time reversal operator which takes @xmath101 and @xmath102 and which is antiunitary . by known arguments , time reversal acting on a state at rest
reverses the spin projection , and one has @xmath103 ( by way of review , one starts with @xmath104 , and recalls that the states with different @xmath31 are related by the angular momentum raising and lowering operators @xmath105 .
one shows that @xmath106 changes sign as the spin projection changes by one unit by considering how @xmath100 commutes with the raising and lowering operators .
that only leaves @xmath107 to be fixed .
since @xmath100 is antiunitary , one can change @xmath107 by changing the phase of the state , and one chooses the phase of the state so that @xmath107 is one . )
define a light - front time reversal operator by @xmath108 giving @xmath109 this also works for moving light - front states . since @xmath110 is antiunitary , @xmath111 from which we see , @xmath112 we use time reversal first to show that the light - front amplitudes are real , for current component @xmath5 , still remembering that @xmath110 is antiunitary , @xmath113 or , @xmath114 ( for momenta in the @xmath24-@xmath23 plane ) . in general , it is not useful to reverse the initial and final states because the particles are different .
but for the elastic case we can use further time reversal to relate amplitudes with interchanged helicity . first note that for the light - front frame with the target at rest , the initial and final particles have the same @xmath27 , so to get a state with the final momentum requires just the transverse boost , @xmath115 beginning by applying the previous time reversal result to the elastic case , and recalling that @xmath76 commutes with `` @xmath59 '' components of four - vectors , leads to @xmath116 thus when the incoming and outgoing particles have the same identity , time reversal gives @xmath117 similarly to the close of the last subsection , we record the time reversal result for the helicity amplitudes in the breit frame , for identical incoming and outgoing particles , @xmath118 note that @xmath119 for the equal mass case . while some of the transformations are easy in the target rest frame , where we calculated
, visualizing this result is not . for this purpose , the @xmath24-breit frame , where the incoming and outgoing particles are both along the @xmath24-direction , works well .
the momenta are , in @xmath120 notation , @xmath121 and the incoming states are defined by , @xmath122 the outgoing states have the same longitudinal boosts , but have opposite transformations for getting the transverse momentum . ( the boost parameters are not the same .
the transformation with @xmath69 gives a momentum along the @xmath23-direction with the final energy ; the transformation with @xmath123 gives a momentum along the @xmath23-direction with the final @xmath27 , but with energy and @xmath124 different from the final ones . from the kinematics given by eq .
( [ x - breit - kinematics ] ) , we find @xmath125 and @xmath126 . )
formally , one defines angle @xmath52 from @xmath127 with a corresponding equation involving the final states and angle @xmath79 . using the representation given earlier in eq .
( [ representation ] ) , @xmath128 conjugating the above equation with @xmath129 ( i.e. , taking @xmath130 ) gives @xmath131 and @xmath119 .
pictorially , we draw the momenta in fig .
[ xbreit ] , and for the helicity states the particle spins point along the direction of the momenta .
the incoming and outgoing light front states both start with a boost in the @xmath23-direction , and then receive symmetrically opposite transverse boosts which rotate the spin vectors in opposite directions by the same amount .
the angles @xmath52 and @xmath79 are indicated in the figure .
one can see both that the size of the angles should be the same and that the senses should be opposite .
the general angular condition for current component @xmath5 reads @xmath132 say that @xmath133 . by changing the sign of both @xmath38 and
@xmath134 it looks like we could get another angular condition , @xmath135 however , using the first of the identities @xmath136 and the light - front parity relation , eq .
( [ parity ] ) , one can show by a series of reversible steps that each angular condition with @xmath137 is equivalent to one with @xmath138 .
hence , we only need to consider cases where @xmath139 .
we shall implement the general angular condition in a couple of special cases , rewriting the angular dependence in terms of @xmath0 and masses .
for the deuteron , the angular condition comes only from @xmath140 and we have @xmath141 for the equal mass case , the arguments of the @xmath43-functions are the same , @xmath142 ( @xmath143 is the deuteron mass ) . using light - front parity , eq .
( [ parity ] ) , and the @xmath43-function identities , eq .
( [ d_function ] ) , one gets @xmath144 substituting for the @xmath145 s and @xmath146 , and using the light - front time reversal result @xmath147 , leads to the angular condition in its known form @xcite , @xmath148 where @xmath149 . for the record , we have removed an overall factor , @xmath150 .
recently , bakker and ji @xcite obtained two constraints on the deuteron helicity amplitudes by noting that there were five amplitudes , and that all five could be derived from three independent form factors . both constraints they called angular conditions .
they appeared differently in different frames ; their drell - yan - west frame results can be most directly compared to our present results . the constraint they call `` ac1 '' is , for momenta in the @xmath24-@xmath23 plane , just @xmath151 . in the present paper this follows from light - front time reversal invariance .
their constraint `` ac2 '' is then precisely the same as the angular condition here .
perturbative qcd predicts , as we shall review below , that the hadron helicity conserving amplitude @xmath152 is the leading amplitude at high @xmath7 and that @xmath153 to leading order in @xmath154 .
no statement is initially made about the size of @xmath155 and @xmath156 .
one may go further , following chung _
et al . _
@xcite or brodsky and hiller @xcite ( who interestingly mention the work of carlson and gross @xcite in this regard ) , to argue that the scale of qcd is given by @xmath157 and that we can implement this in the light - front frame by saying that @xmath158 a consequence of this , written in terms of the deuteron charge , magnetic and quadrupole form factors @xcite , is that to good approximation one gets the `` universal ratios '' @xcite , @xmath159 this agrees with the leading power of @xmath0 result @xcite that @xmath160 , but goes beyond it and also gives a prediction for @xmath161 .
we have so far in this subsection used only three light - front helicity amplitudes .
there are more that are not zero , and we find a difficulty when we discuss a fourth .
amplitude @xmath162 is related to the others by the angular condition quoted above .
also , the perturbative qcd arguments that give the scaling behavior of the other helicity amplitudes give for @xmath162 at very high @xmath0 , @xmath163 ( helicity is conserved , but other spin dependent rules @xcite dictate a two power asymptotic suppression of @xmath162 .
this is also consistent with a naturalness condition discussed in ref .
@xcite . )
the angular condition to leading order now reads , @xmath164 the hypothesis that @xmath157 sets the scale of the subleading amplitudes would suggest that @xmath165 as well as @xmath155 is of @xmath166 .
given the angular condition result just above , this can not be right ; at least one of @xmath155 and @xmath165 must be @xmath167 .
hence the hypothesis is not generally workable , and one needs to consider thinking the same about the next - to - leading corrections in the `` universal ratios '' expression , eq .
( [ universal ] ) .
the @xmath169(1232 ) transition is an important reaction that involves final and initial states with different spins and masses .
this makes working out the angular condition more involved technically , but not unduly so , as we shall demonstrate .
there is one angular condition , @xmath170 using the light - front parity .
explicit substitution for the @xmath43-functions yields
@xmath171 finally , removing the overall factors and substituting for the trigonometric functions gives the angular condition for the @xmath168-@xmath1 transition , @xmath172 g^+_{l,-3/2,1/2 } + \sqrt{3 } m q ( m - m ) g^+_{l,-1/2,1/2 } \nonumber \\ & + & \sqrt{3 } m q^2 g^+_{l,1/2,1/2 } + q \left [ q^2 -m(m - m ) \right ] g^+_{l,3/2,1/2},\end{aligned}\ ] ] where @xmath11 is the nucleon mass and @xmath12 is the @xmath1 mass . for the record , we have removed another overall factor , @xmath173 .
the asymptotic scaling rules , cited in the next subsection , say that @xmath174 goes like @xmath175 at high @xmath7 , that @xmath176 and @xmath177 go like @xmath178 , and that @xmath179 goes like @xmath180 .
if we write @xmath181 modulo logarithms at high @xmath7 , then the leading @xmath7 part of the angular condition says @xmath182 the idea of `` good currents '' and `` bad currents '' is native to the light - front frame . in analyzing the power law scaling behavior at high @xmath0 , for a given helicity amplitude , it is often thought to be safest to stay in the light - front frame and use only `` good currents . ''
we shall here derive the breit frame helicity amplitude scaling behaviors > from their light - front counterparts .
note here @xmath22 both in the light - front frame and the breit frame that we discuss in this subsection .
all the @xmath22 frames are related to each other only by the kinematical operators that make the light - front time @xmath183 intact .
we will find , nicely enough , that the scaling behaviors are the same as one would have found using the breit frame only .
that is , one can get the correct leading power scaling behavior from a breit frame analysis alone . for a light - front helicity amplitude ,
the scaling behavior at high @xmath7 is @xmath184 where @xmath185 is the number of quarks in the state , @xmath11 is a mass scale , and @xmath186 is the minimum helicity of the incoming or outgoing state ( i.e. , 0 or 1/2 for bosons or fermions , respectively ) . regarding the breit frame , we have thus far given its helicity amplitudes in terms of @xmath187 where @xmath58 is a lorentz index .
it is usual to substitute a photon helicity index for the lorentz index , using ( for incoming photons ) @xmath188 with polarizations ( in @xmath189-type notation ) @xmath190 where @xmath52 is the angle between @xmath191 and the negative @xmath23-direction , as shown in fig .
one can work out that @xmath192 \ , \ ] ] so that @xmath193 \
, \ ] ] where the superscripts on @xmath40 will in the rest of this subsection refer to photon helicity unless explicitly stated otherwise . with the general relation , eq .
( [ main ] ) , this gives directly the expression we use to obtain the scaling behavior of the breit amplitudes , @xmath194 we can select terms on the left - hand - side by choice of @xmath134 and @xmath38 , since breit amplitudes are non - zero only for @xmath195 .
the @xmath43-functions can be written in terms of sines and cosines of half angles , so we record that at high @xmath7 , @xmath196 where @xmath11 inside the @xmath197 symbol is a generic mass scale .
the @xmath43-functions can be expanded as @xcite @xmath198 where @xmath199 are numerical coefficients with @xmath200 .
thus for large @xmath7 , @xmath201 on the right - hand - side of eq .
( [ expression ] ) , there is no term that falls slower than the term that has @xmath202 , and @xmath203 .
thus the breit amplitude leading falloff at high @xmath7 is @xmath204 these are the same results one can get by directly analyzing amplitudes in the breit frame for various photon helicities @xcite . by way of example , we will give the breit and light - front frame helicity amplitudes for elastic electron - nucleon scattering . in terms of the standard dirac , pauli , and sachs form factors
one may work out @xmath205 for the light - front states and @xmath206 for the breit frame states .
the scaling rules predict that @xmath161 , @xmath207 , and @xmath208 scale as @xmath175 and that @xmath209 scales as @xmath180 , consistent with the relations @xmath210 ( there is recent data @xcite and commentary @xcite on the helicity flip scaling results . )
the purpose of the present paper has been largely kinematical .
we have examined the relationship between the helicity amplitudes in the breit and light - front frames .
one particular result has been a clear view of where the angular condition comes from .
the angular condition is a constraint on light - front helicity amplitudes .
it follows from applying angular momentum conservation in the breit frame , where the application of angular momentum conservation to the helicity amplitudes is elementary .
one consequence is that an amplitude must be zero if it requires the photon to have more than one unit magnitude of helicity , and this statement cast in terms of light - front amplitudes is the angular condition @xcite .
another set of constraints follows from parity and time - reversal invariance .
neither of these symmetries can be used directly on the light - front because the light - front has a preferred spatial direction .
however , each of them can be modified to give a valid symmetry operation ( at least for strong and electromagnetic interactions ) for the light - front @xcite . to define a light - front parity ,
choose the @xmath24-@xmath23 plane to contain all the momenta and then consider mirror reflection in the @xmath24-@xmath23 plane .
this reflection leaves the momenta unchanged but reverses the helicities @xcite .
technically , it is the same as ordinary parity followed by a 180@xmath86 rotation about the @xmath44-axis , and helicity amplitude relations that follow from it were given in section [ discrete ] .
similarly one defines light - front time reversal as ordinary time reversal followed by the 180@xmath86 rotaton about the @xmath44-axis .
time reversal invariance implies that the amplitudes are always real for momenta in the @xmath24-@xmath23 plane , with additional relations possible for elastic scattering , as detailed also in section [ discrete ] .
the general angular condition appears compactly in terms of @xmath43-functions , the representations of the rotation operators .
it can be rewritten in terms of masses and momentum transfer .
we gave the translations for two cases .
for electron - deuteron elastic scattering , the result is well known @xcite .
nonetheless , it does have an unchronicled ( we believe ) consequence .
that is that the mass scale associated with the asymptotic power - law falloff of non - leading amplitudes must generally be of order of the nucleon or deuteron mass .
it had been hoped that the light - front was a favored frame where the non - leading amplitudes would have small numerical coefficients : asymtotically of order @xmath211 , to an appropriate power , times the leading amplitude .
as the angular condition contradicts this , it also takes away the motivation that underlay the suggestion of the universal behavior for spin-1 form factors .
additionally , we gave the angular condition explicitly for the @xmath168-@xmath1 electromagnetic transition .
we believe this result is also new .
the angular conditon here fixes precisely the leading power of one subleading amplitude .
the power law scaling of the helicity amplitudes can be analyzed , with a definite power of @xmath154 given in terms of the number of constituents in the wave function and in terms of the helicities of the incoming and outgoing states , and this can be done in either the light - front frame or in the breit frame .
a preference for one or the other is sometimes given .
our final `` application '' was to obtain the breit frame scaling from the light - front scaling using the transformation between them given earlier in section [ relatons ] , and to find that the result is the same as one obtains doing the analysis directly in the breit frame .
cec thanks the doe for support under contract de - ac05 - 84er40150 , under which the southeastern universities research association ( sura ) operates the thomas jefferson national accelerator facility , and also thanks the nsf for support under grant phy-9900657 .
crj thanks the doe for support under contract de - fg02 - 96er40947 and the nsf for support under grant int-9906384 , and also thanks the north carolina supercomputing center and the national energy research scientific computer center for the grant of supercomputing time .
s.fubini , g. furlan , physics * 1 * , 229 ( 1965 ) ; s. weinberg , phys . rev . * 150 * , 1313 ( 1966 ) ; j. jersak and j. stern , nucl .
* b7 * , 413 ( 1968 ) ; h. leutwyler , in springer tracts in modern physics vol .
g. hhler , ( berlin , 1969 ) .
d. e. soper , phys .
d * 5 * , 1956 ( 1972 ) .
h. osborn , nucl .
b * 38 * , 429 ( 1972 ) ; h. leutwyler and j. stern , annals phys . *
112 * , 94 ( 1978 ) ; i. l. grach and l. a. kondratyuk , sov .
j. nucl .
phys . * 39 * , 198 ( 1984 ) [ yad .
* 39 * , 316 ( 1984 ) ] ; b. d. keister , phys .
d * 49 * , 1500 ( 1994 ) [ arxiv : hep - ph/9303264 ] . c. e. carlson , j. r. hiller and r. j. holt , ann .
rev . nucl . part .
* 47 * , 395 ( 1997 ) .
b. l. bakker and c. r. ji , phys . rev .
d * 65 * , 073002 ( 2002 ) [ arxiv : hep - ph/0109005 ] .
p. l. chung , w. n. polyzou , f. coester and b. d. keister , phys .
c * 37 * , 2000 ( 1988 ) .
s. j. brodsky and j. r. hiller , phys .
d * 46 * , 2141 ( 1992 ) . c. e. carlson and f. gross , phys .
* 53 * , 127 ( 1984 ) .
see for example r. g. arnold , c. e. carlson and f. gross , phys .
c * 21 * , 1426 ( 1980 ) .
edmonds , _ angular momentum in quantum mechanics _ , princeton university press , 1957 .
o. gayou _ et al . _
[ jefferson lab hall a collaboration ] , = 5.6-gev**2 , phys .
lett . * 88 * , 092301 ( 2002 ) [ arxiv : nucl - ex/0111010 ] ; m.
k. jones _ et al . _
[ jefferson lab hall a collaboration ] , p(pol . ) , phys .
lett . * 84 * , 1398 ( 2000 ) [ arxiv : nucl - ex/9910005 ] .
ralston and p. jain , arxiv : hep - ph/0207129 ; j. p.
ralston , p. jain and r. v. buniy , aip conf .
proc . * 549 * , 302 ( 2000 ) [ arxiv : hep - ph/0206074 ] ; g. a. miller and m. r. frank , phys .
c * 65 * , 065205 ( 2002 ) [ arxiv : nucl - th/0201021 ] ; s. j. brodsky , talk at workshop on exclusive processes at high momentum transfer , newport news , virginia , 15 - 18 may 2002 , arxiv : hep - ph/0208158 ; a. v. belitsky , x. ji and f. yuan , arxiv : hep - ph/0212351 . | we analyze the current matrix elements in the general collinear ( breit ) frames and find the relation between the ordinary ( or canonical ) helicity amplitudes and the light - front helicity amplitudes . using the conservation of angular momentum , we derive a general angular condition which should be satisfied by the light - front helicity amplitudes for any spin system .
in addition , we obtain the light - front parity and time - reversal relations for the light - front helicity amplitudes . applying these relations to the spin-1 form factor analysis , we note that the general angular condition relating the five helicity amplitudes is reduced to the usual angular condition relating the four helicity amplitudes due to the light - front time - reversal condition .
we make some comments on the consequences of the angular condition for the analysis of the high-@xmath0 deuteron electromagnetic form factors , and we further apply the general angular condition to the electromagnetic transition between spin-1/2 and spin-3/2 systems and find a relation useful for the analysis of the n-@xmath1 transition form factors .
we also discuss the scaling law and the subleading power corrections in the breit and light - front frames . |
the study of multilinear pseudodifferential operators goes back to the pioneering works of r. coifman and y. meyer , @xcite , @xcite , @xcite and @xcite .
since then , there has been a large amount of work on various generalisations of their results , as well as studies of bilinear operators with symbols satisfying different conditions to those in the standard bilinear coifman - meyer classes .
the literature in this area of research is vast and any brief summary of it here would not do the authors justice .
therefore we confine ourselves to mention only those works with a direct connection to the present paper .
r. coifman and y. meyer , in @xcite and @xcite , proved the boundedness from @xmath2 to @xmath3 of multilinear pseudodifferential operators with symbols in the class @xmath4 ( see definition [ bilinear multiplier defn ] below ) for @xmath5 and @xmath6 with @xmath7 . in the seminal paper @xcite , l. grafakos and r. torres systematically developed the theory of multilinear caldern - zygmund operators .
they proved a multilinear @xmath8-theorem which they applied to generalise the result above to @xmath9 .
as a further application , they demonstrated the boundedness in lebesgue spaces of multilinear pseudodifferential operators which , together with each of the adjoint operators , belonged to @xmath10 ( see definition [ def4 ] ) .
however , in @xcite , a. bnyi and r. torres showed that there exist symbols in @xmath11 that do not give rise to bilinear operators which are bounded from @xmath12 to @xmath3 for @xmath13 such that @xmath14 .
in particular , there is no analogue of the caldern - vaillancourt theorem in the bilinear setting .
moreover , the class of operators @xmath15 is not closed under transposition . in contrast , @xcite demonstrates that @xmath16 is closed under transposition .
recently , in @xcite , a. bnyi , d. maldonado , v. naibo and r. torres proved that @xmath17 is closed under transposition for @xmath18 and @xmath19 . in particular , given an operator in @xmath20 , its adjoints are also in @xmath20 .
since @xmath21 , it follows that symbols in @xmath22 give rise to bounded operators , by applying the result of @xcite quoted above . in summary
, we see that @xmath23 are bounded on appropriate lebesgue spaces when @xmath24 ( that is , the caldern - zygmund case ) , but in general they fail to be bounded when @xmath25 .
the purpose of this paper is to address the following question , which is of interest for @xmath26 in - between these values , ` given @xmath27 , what @xmath28 is sufficient to ensure that symbols in @xmath29 give rise to bounded operators ? '
this question is in the spirit of questions asked in @xcite .
we will study this question for two different symbol classes .
first , in section [ main1 ] , we will consider a larger symbol class which does not require any differentiability in the spatial variable at all .
that is , we study the multilinear symbol class @xmath30 ( see definition [ def2 ] ) which , in particular , contains @xmath29 for any @xmath31 .
our main result in this context is theorem [ multi_one ] , which generalises a result obtained by the present authors in @xcite regarding the linear case .
the study of such symbol classes originates in @xcite , where c. kenig and the third author studied linear operators . in the context of multilinear operators ,
results regarding mildly regular bilinear operators have been proved previously . in particular ,
d. maldonado and v. naibo established in @xcite boundedness properties of bilinear pseudodifferential operators on products of weighted lebesgue , hardy , and amalgam spaces .
the regularity they require in the spatial variables is only that of dini - type .
section [ mixed norms ] deals with linear operators on mixed - norm lebesgue spaces , and is a corollary to the proof of theorem [ multi_one ] .
the second topic we will study is the bilinear symbol class @xmath32 . in section [ lpsmrho and applications ]
we adapt methods used to study symbols in @xmath30 to weaken the requirement on @xmath33 necessary to prove boundedness on lebesgue spaces of operators in @xmath17 for @xmath34 .
this is formulated as theorem [ smooth pseudo bilinear ] . in section [ outwith ] , although we can not show boundedness for general operators arising from symbols in @xmath35 , we can prove boundedness on a suitable subclass .
this is stated as theorem [ modulationtype ] , which is a result of the same flavour as that proved by f. bernicot and s. shrivastava in @xcite regarding a subclass of @xmath36 , albeit proved by more straight - forward methods .
a related result regarding @xmath37 was also proved in @xcite .
we begin the main body of the paper with section [ prelim ] where we set out some definitions , fix some notation and recall some well - known results that we will use later .
we study the following type of _ multilinear pseudodifferential operator_. given a function @xmath38 we define the @xmath39-linear operator @xmath40 to act on @xmath39 functions @xmath41 belonging to the schwartz class @xmath42 as @xmath43 here @xmath44 are all variables in @xmath45 , @xmath46 and @xmath47 denotes the fourier transform @xmath48 of @xmath49 .
we refer to the function @xmath50 , which has @xmath51 variables , as the _ symbol _ of the operator @xmath40 .
we set @xmath52 with @xmath53 , and define @xmath54 where @xmath55 denotes the standard euclidean norm of @xmath56 also , here and in the sequel we shall use @xmath57 to denote the set of nonnegative integers . + we will use a standard littlewood - paley partition of unity @xmath58 in @xmath59 by letting @xmath60 be a smooth radial function which is equal to one on the unit ball centred at the origin and supported on its concentric double .
setting @xmath61 and @xmath62 for @xmath63 we have @xmath64 and @xmath65 for @xmath66 .
one also has , for all multi - indices @xmath67 and @xmath68 , @xmath69 , and @xmath70 [ def2 ] given @xmath71 , @xmath72 and @xmath27 the symbol @xmath73 is said to belong to @xmath30 when for each multi - index @xmath67 there exists a constant @xmath74 such that @xmath75 in the case @xmath76 we also use the notation @xmath77 for the class of symbols of the linear pseudo - pseudodifferential operators , see @xcite . [ def4 ] given a class of symbols @xmath78 , operators which arise from elements in @xmath78 are denoted by @xmath79 .
that is , we say @xmath80 when there exists a symbol @xmath81 such that @xmath82 , as defined in .
consequently , for @xmath83 we say @xmath84 . for a non - negative function @xmath85 , which we refer to as a _ weight _
, we define @xmath86 to be the closure of @xmath87 in the norm @xmath88 when @xmath89 we write simply @xmath90 to mean @xmath91 and @xmath92 is the class of functions which belong to @xmath93 for each @xmath85 which is the characteristic function of a compact set .
we wish to study the boundedness from @xmath94 to @xmath95 of the operator @xmath40 , initially defined for schwartz functions @xmath96 via , for particular exponents @xmath97 and weights @xmath98 .
although the integral in may not be absolutely convergent for @xmath41 which do not decay sufficiently rapidly , if we can prove bounds on the operator norm which depend only on @xmath99 and @xmath50 , then it is a straight - forward exercise to show that @xmath40 has a unique extension to @xmath94 which agrees with for @xmath100 .
this is the sense in which we will refer to the boundedness of @xmath40 .
given @xmath101 , the @xmath102 maximal function @xmath103 is defined by @xmath104 where the supremum is taken over balls @xmath105 in @xmath106 containing @xmath107 .
clearly then , the hardy - littlewood maximal function is given by @xmath108 an immediate consequence of hlder s inequality is that @xmath109 for @xmath110 .
we shall use the notation @xmath111 for the average of the function @xmath112 over @xmath105 .
one can then define the class of muckenhoupt @xmath113 weights as follows .
[ weights ] let @xmath114 be a positive function .
one says that @xmath115 if there exists a constant @xmath116 such that @xmath117 one says that @xmath118 for @xmath119 if @xmath120 the @xmath113 constants of a weight @xmath118 are defined by @xmath121_{a_1}:= \sup_{b\ , \textrm{balls in}\,\ , \r^{n}}\,w_{b}\vert w^{-1}\vert_{l^{\infty}(b)},\ ] ] and @xmath121_{a_p}:= \sup_{b\ , \textrm{balls in}\,\ , \r^{n}}\,w_{b}(w^{-\frac{1}{p-1}})_{b}^{p-1}.\ ] ] the following results are well - known and can be found in , for example , @xcite .
[ maxweight ] for @xmath122 , the hardy - littlewood maximal operator is bounded on @xmath123 if and only if @xmath124 .
consequently , for @xmath125 , @xmath126 is bounded on @xmath123 if and only if @xmath127 [ convolve ] suppose that @xmath128 is integrable non - increasing and radial
. then , for @xmath129 , we have @xmath130 for all @xmath131 . we will need the following multilinear version of the hausdorff - young theorem due to a. benedek and r. panzone @xcite .
[ hausdorff - young ] suppose that @xmath132 and @xmath133 then @xmath134 as is common practice , we will denote constants which can be determined by known parameters in a given situation , but whose value is not crucial to the problem at hand , by @xmath135 .
such parameters in this paper would be , for example , @xmath33 , @xmath26 , @xmath136 , @xmath137 , @xmath138_{a_p}$ ] , and the constants @xmath74 in definition [ def2 ] .
the value of @xmath135 may differ from line to line , but in each instance could be estimated if necessary .
we also write @xmath139 as shorthand for @xmath140 .
the following lemma will be useful in obtaining pointwise estimates for the kernel of operators in @xmath142 for @xmath143 with @xmath144 and @xmath145 with @xmath146 we define @xmath147 [ zlemma ] let @xmath148 with @xmath149 and @xmath150 $ ] . given any @xmath151 such that the set @xmath152 \,|\ , |z_j|{\geqslant}1\ } \neq \emptyset$ ] , one has @xmath153 for all @xmath154 and @xmath155 setting @xmath156 , and using the definition of @xmath157 , inequality and the leibniz rule we see that @xmath158 where in we have also used the assumption @xmath27 .
we claim that @xmath159 for all @xmath160 . integrating by parts and using with @xmath161 yields @xmath162 now summing both sides of the above estimate over all @xmath163 with @xmath164 and using the straightforward inequality @xmath165
, we obtain . for the integrals containing @xmath166 , integration by parts and
yield @xmath167 therefore , if @xmath168 then @xmath169 from this and the definition of the set @xmath170 , by taking @xmath168 and @xmath171 , it follows that @xmath172 for @xmath173 and @xmath174 such that @xmath175 .
the estimate for @xmath176 follows by combining the estimates and .
this proves the lemma . as an immediate corollary we have the following kernel estimates [ zlemma2 ]
let @xmath177 and suppose @xmath107 and @xmath178 are such that @xmath179 \,|\ , |x - y_j|{\geqslant}1\ } \neq \emptyset$ ] .
then one has @xmath180 provided either @xmath181 and @xmath72 , or @xmath182 and @xmath183 .
when @xmath184 , this follows from lemma [ zlemma ] by setting @xmath185 .
an examination of the proof of lemma [ zlemma ] reveals that it can be easily modified for the case @xmath25 provided @xmath183 .
the following theorem is the main result of this section .
[ multi_one ] fix @xmath186 $ ] for @xmath187 and let @xmath188 with @xmath189 and @xmath190 .
then there exists a constant @xmath135 , depending only on @xmath137 , @xmath191 , @xmath33 , @xmath26 and a finite number of the constants @xmath74 in definition @xmath192 , such that latexmath:[\[\label{pointws } for all @xmath131 .
consequently , for @xmath194 and @xmath195 such that @xmath196 , @xmath40 is a bounded operator from @xmath94 to @xmath197 whenever @xmath198 and @xmath199 .
the boundedness follows immediately from the pointwise estimate by theorem [ maxweight ] . to prove we use the littlewood - paley partition of unity as in
, we decompose the symbol as @xmath200 with @xmath201 , @xmath202 .
first we consider the operator @xmath203 .
we have @xmath204 with @xmath205 now estimate yields @xmath206 for each @xmath207 and hence for @xmath208 .
therefore theorem [ convolve ] implies @xmath209 .
now let us analyse @xmath210 for @xmath66 .
we note , just as before , that @xmath211 can be written as @xmath212 with @xmath213 one observes that @xmath214 where the weight functions @xmath215 will be chosen momentarily . therefore , hlder s inequality yields @xmath216 where @xmath217 .
now for an @xmath218 , we define @xmath215 by @xmath219 we now wish to estimate @xmath220 by splitting each @xmath221-integral as integration over @xmath222 and @xmath223 . considering an arbitrary case of the @xmath224 possibilities
, we can estimate this portion of the integral by @xmath225 where the sum is taken over multi - indices @xmath163 ( each with @xmath137 components ) such that @xmath226 if @xmath222 and @xmath227 if @xmath223 . without loss of generality we may assume @xmath228 , so by theorem [ hausdorff - young ] and the estimate , this in turn is majorised by @xmath229 furthermore , once again using theorem [ convolve ] , we have @xmath230 with a constant that only depends on the dimension @xmath137 . combining these facts with yields @xmath231 summing in @xmath232 , and using and , we obtain @xmath233 we observe that the series above converges if @xmath234
. this proves and , with it , the theorem .
we remark in passing that the case @xmath235 theorem [ multi_one ] follows from its linear predecessor , that is theorem 3.3 in @xcite , with @xmath236 replacing @xmath45 .
in this section we show that a modification of the proof of theorem [ multi_one ] yields a mixed norm boundedness result for a class of linear pseudodifferential operators .
let @xmath237 with @xmath238 for @xmath239 to a symbol @xmath240 we associate a linear operator @xmath241 a - priori defined on functions @xmath112 in @xmath242 given by @xmath243 where @xmath244 is the fourier transform of @xmath112 in @xmath236 : @xmath245 we define the space @xmath246 to be the mixed norm space which is the closure of @xmath247 in the norm @xmath248 we will also need the following notation .
for a function @xmath249 we define @xmath250 to be the @xmath102 maximal function acting only in the @xmath251-variables .
that is , @xmath252 for @xmath187 , where the supremum is taken over balls @xmath105 in @xmath45 containing @xmath251 .
[ multi_two ] fix @xmath186 $ ] for @xmath187 and let @xmath253 with @xmath189 and @xmath190 .
then there exists a constant @xmath135 , depending only on @xmath137 , @xmath191 , @xmath33 , @xmath26 and a finite number of the constants @xmath74 in definition @xmath192 , such that @xmath254 .
consequently , @xmath40 is a bounded operator from @xmath255 to @xmath256 whenever @xmath257 @xmath258@xmath187@xmath259 and @xmath260 .
we repeat the prove of theorem [ multi_one ] , but with the linear operator above and observe that is replaced by @xmath261 we control the first factor on the right - hand side of as before . to control the second factor in we can use theorem [ convolve ] to show that it is majorised by @xmath262 by combining these estimates
we obtain . using the boundedness of the maximal function and
minkowski s inequality repeatedly , we obtain the boundedness of @xmath40 on the mixed norm space defined above .
we now consider linear pseudodifferential operators acting on functions on @xmath45 .
[ def5 ] let @xmath85 be a weight @xmath258that is , a non - negative function@xmath259 , and @xmath264 , @xmath72 and @xmath27 be parameters .
a symbol @xmath265 belongs to the class @xmath266 if for each multi - index @xmath267 there exists a constant @xmath74 such that @xmath268 when the weight @xmath269 then we use the notation @xmath270 for @xmath271 .
[ lrsym ] suppose @xmath272 , @xmath273 $ ] and @xmath274 with conjugate @xmath275 @xmath258for which @xmath276@xmath259 satisfy the relation @xmath277 . suppose further that @xmath27 and @xmath278 .
let @xmath279 and @xmath85 and @xmath280 be weights with @xmath281 .
then there exists a constant @xmath135 , depending only on @xmath137 , @xmath33 , @xmath26 , @xmath136 , @xmath282 , @xmath138_{a_{q / p'}}$ ] and a finite number of @xmath74 from @xmath283 @xmath284 , such that 1 . [ weightedlp ] if @xmath285 and @xmath286 , then @xmath287 where @xmath288 ; and 2 .
[ unweightedlp ] if @xmath289 or @xmath290 , then @xmath291 where @xmath292 .
first , we define @xmath293 it is then easy to check that @xmath294 provided @xmath295 is sufficiently large .
we consider the littlewood - paley pieces @xmath296 of the operator @xmath297 , where @xmath298 .
using hlder s inequality , the hausdorff - young inequality and theorem [ convolve ] , we compute @xmath299 consequently , under the hypotheses of @xmath300 , hlder s inequality with exponents @xmath301 and @xmath302 , the weighted boundedness of the hardy - littlewood maximal function and minkowski s inequality show us that @xmath303 we can then sum in @xmath232 to find that @xmath304 which completes the proof of @xmath300 .
the proof of @xmath305 is similar .
when @xmath289 , @xmath306 , so we can not use the boundedness of the hardy - littlewood maximal function in .
so before applying theorem [ convolve ] in , we instead first take the @xmath307 norm of the inequality , apply hlder s inequality as before , and then apply young s inequality to the factor involving @xmath308 .
when @xmath290 , @xmath309 , so once again we can not use the weighted boundedness of the hardy - littlewood maximal function , but we can use its boundedness on @xmath310 .
as an application of theorem [ lrsym ] we now establish our main result regarding the boundedness of smooth bilinear pseudodifferential operators that fall outwith the scope of the bilinear caldern - zygmund theory .
we shall prove a further result concerning a subclass of these operators in section [ outwith ] .
[ bilinear multiplier defn ] given @xmath311 , @xmath312 and @xmath313 $ ] , a symbol @xmath73 belongs to the class @xmath314 if , for each pair of multi - indices @xmath67 , and @xmath315 , there exists a constant @xmath316 such that @xmath317 in particular , a symbol @xmath318 belongs to the class @xmath319 if , for each triple of multi - indices @xmath320 @xmath321 and @xmath322 , there exists a constant @xmath323 such that @xmath324 define the adjoint operators @xmath325 and @xmath326 of a bilinear operator @xmath327 via the identities @xmath328 , where @xmath329 denotes the dual pairing on @xmath102 ( @xmath330 ) .
we will use theorem 1 in @xcite repeatedly , so we record it here .
[ 1in2 ] assume that @xmath18 , @xmath19 and @xmath331 .
then @xmath332 for some @xmath333 and @xmath334 .
[ smooth pseudo bilinear ] let @xmath331 , with @xmath18 , @xmath19 and @xmath335 for @xmath336 $ ] @xmath258see figure @xmath337@xmath259 .
then @xmath338 for @xmath136,@xmath282 and @xmath339 such that @xmath277 .
observe that the condition @xmath340 ensures that we can not have both @xmath136 and @xmath282 less than @xmath341 .
theorem [ 1in2 ] tells us that the adjoint operators @xmath342 and @xmath342 are operators in the same class as @xmath40 . considering these adjoint operators if necessary
, we can reduce the proof of the theorem to the special case @xmath343 .
we consider a littlewood - paley partition of unity @xmath58 defined as in with @xmath344 we then set @xmath345 , so @xmath346 .
we can write @xmath347 where @xmath348 is a linear operator for each fixed @xmath349 . in order to take advantage of the smoothness of the symbol while viewing the operator as written in , we must deal with each piece of the symbol @xmath350 depending on how the size of @xmath351 relates to the size of @xmath352 .
more precisely , for @xmath353 we have that @xmath354 on the @xmath355-support of @xmath350 .
fixing an @xmath356 sufficiently small , we can conclude that @xmath357 for @xmath358 such that @xmath359 .
thus , by @xcite , if @xmath360 then we have that @xmath361 this shows that @xmath362 therefore , using the fact that @xmath363 an application of theorem [ lrsym ] with the assumption @xmath364 yields @xmath365 for the case @xmath366 , we can repeat the same argument , but reverse the roles of @xmath351 and @xmath352 , and @xmath136 and @xmath282 , to obtain once again .
we then sum in @xmath367 and @xmath232 to obtain boundedness from @xmath368 to @xmath3 provided @xmath369 this result can be improved by using duality and once again applying theorem [ 1in2 ] .
we are concerned with triples of reciprocals of exponents @xmath370 such that @xmath371 and @xmath336 $ ] .
the set of such triples is a closed triangle with vertices @xmath372 , @xmath373 and @xmath374 in the plane @xmath375 , which itself lies in @xmath376 ( see figure [ figure ] ) . considering the edge of the triangle with end - points @xmath372 and @xmath373 ( that is , where @xmath377 ) we have proved that @xmath40 maps @xmath378 to @xmath379 ( of course , here we require @xmath380 ) for @xmath381 this agrees with the statement of the theorem for these exponents .
equally the theorem in the case @xmath382 ( corresponding to the point @xmath374 ) is also included in the condition . considering the adjoint @xmath326 , using duality and applying theorem [ 1in2 ] allows us to conclude that @xmath40 maps @xmath383 to @xmath384 for the same range on @xmath33 . once again
, this agrees with the statement of the theorem for these exponents , but this time corresponds to triples @xmath370 on the line with end - points @xmath372 and @xmath374 .
similarly , considering the adjoint @xmath325 , using duality and applying theorem [ 1in2 ] allows us to conclude that @xmath40 maps @xmath385 to @xmath386 , again , for the same range of @xmath33 . yet again , this agrees with the statement of the theorem for these exponents , but now corresponds to triples @xmath370 on the line with end - points @xmath373 and @xmath374 .
thus , we have proved the theorem on the edges of the triangle .
finally , the bilinear version of the riesz - thorin interpolation theorem ( see @xcite ) allows us to complete the proof on the interior of the triangle .
in this section we will establish the boundedness of a subclass of bilinear pseudodifferential operators with symbols in the class @xmath388 with @xmath389 $ ] . with this goal in mind
, the following lemma will prove to be useful .
[ notquite ] given a smooth function @xmath390 , define an operator @xmath170 by @xmath391 where @xmath392 denotes the fourier transform in @xmath393 @xmath258that is , with @xmath394@xmath259 .
the operator @xmath170 is bounded from @xmath395 to @xmath396 if and only if @xmath397 moreover , @xmath398 the boundedness of @xmath170 is equivalent to the boundedness of @xmath399 , where @xmath400 is the adjoint operator of @xmath170 : @xmath401 we can readily see that @xmath402 is given by @xmath403 so @xmath404 thus , @xmath405 is a multiplier and the condition of the lemma is exactly the condition required for its boundedness on @xmath396 .
[ modulationtype ] if @xmath406 and @xmath407 are such that , for each multi - index @xmath408 , @xmath409 where @xmath410 is the unit ball centred at @xmath411 , then @xmath40 is bounded operator from @xmath412 to @xmath413 for @xmath414 $ ] and @xmath415 $ ] such that @xmath371 .
this corresponds to @xmath370 contained in the closed triangle of @xmath416 @xmath337 with vertices @xmath417 , @xmath418 and @xmath419 .
first , let us observe that it suffices to prove the boundedness from @xmath420 to @xmath379 . indeed , assuming this and using theorem [ 1in2 ] , together with duality arguments and the multilinear riesz - thorin theorem ( see @xcite ) , yield the theorem .
let us now suppose that the symbol @xmath50 has compact @xmath107-support , say contained in the unit ball @xmath421 .
we follow @xcite and write @xmath422 and so @xmath423 where @xmath424 and @xmath425 is a multiplier operator for each @xmath426 . therefore , since @xmath427 we have that @xmath428 and so @xmath429 this means that , by our hypotheses , @xmath430 for any @xmath431 . consequently , by lemma [ notquite ] , @xmath432 once again using the support properties of @xmath50 , we conclude that @xmath433 which proves the theorem under the extra hypothesis that @xmath50 has compact @xmath107-support .
+ to remove the hypothesis that @xmath50 has compact @xmath107-support , we follow the argument of @xcite .
observe that it suffices to show that , for each @xmath434 , @xmath435 for all @xmath436 indeed , choosing @xmath437 , integrating in @xmath411 , using the cauchy - schwarz inequality and interchanging the order of integration produces the estimate @xmath438 to prove we introduce the cut - off function @xmath439 , which is identically one on @xmath440 and zero outside its concentric double @xmath441 .
define @xmath442 and @xmath443 and @xmath444 .
let @xmath445 be a second nonnegative cut - off function which is identically one on @xmath446 and supported in @xmath440 . using the cut - off
function @xmath447 , we can write @xmath448 where @xmath449 now using the cut - off function @xmath450 our previous boundedness result concerning bilinear operators with compact spatial support yields @xmath451 and this is , in turn , controlled by @xmath452 for any @xmath68 , because of the support properties of @xmath453 and @xmath454 . to estimate the contribution of the remaining term in , we use the kernel estimate of corollary [ zlemma2 ] . we need to estimate @xmath455 but since @xmath456 is supported outside @xmath457 , corollary [ zlemma2 ] yields that for all @xmath458 , is majorised by @xmath459 and using the cauchy - schwarz inequality and the fact that @xmath460 , we have @xmath461 provided @xmath462 this completes the proof of and with it , the theorem . | we consider two types of multilinear pseudodifferential operators .
first , we prove the boundedness of multilinear pseudodifferential operators with symbols which are only measurable in the spatial variables in weighted lebesgue spaces .
these results generalise earlier work of the present authors concerning linear pseudo - pseudodifferential operators .
secondly , we investigate the boundedness of bilinear pseudodifferential operators with symbols in the hrmander @xmath0 classes .
these results are new in the case @xmath1 , that is , outwith the scope of multilinear caldern - zygmund theory . |
continuous development of information communication technology and large scale proliferation is leading to an increase in victims of digital crime , as well as tools and evidence in criminal investigations [ 1 ] .
many crimes , even traditionally non - digital crimes such as murder , now normally have some sort of digital component [ 2,3 ] .
even so much so that some law enforcement agencies claim that there are rarely cases that do not involve some sort of digital device [ 4 ] .
because of this , the ability of law enforcement to handle electronic evidence is vital during entire investigative procedure .
a survey recently conducted by the united nations ( u.n . ) reported that more than 90% of respondent countries have at least some capability to conduct digital investigations [ 5 ] .
however , the u.n .
report also showed a global need for expansion of digital investigation capability .
key areas of concern were with the investigation capabilities in less developed countries , as well as the capability of organizations to conduct international investigations .
this mirrors huber [ 6 ] , who claimed that `` digital forensics is very much an interstate and international issue '' .
general cybercrime investigation , computer forensics and digital evidence handling were the most commonly cited areas that require more development and assistance . currently , many public and private organizations around the world are investing in the infrastructure of developing countries .
however , international assistance in regards to digital crime investigation has been slower that other sectors , such as information technology [ 7 ] .
many countries , developing and developed alike , are beginning to invest in the capability of their investigators to investigate digital crimes .
this has caused a rapid growth in the number of digital crime investigation programs offered at training institutions and universities around the world .
these programs are targeted at improving an investigator s _ capability _ in dealing with certain types of digital crime ; however , improvement of capability does not necessarily improve investigation _ capacity_. hekim , gul et al .
[ 8 ] previously examined investigation capacity in law enforcement to assess whether information technology improves investigation capacity . in their work , they define investigation capacity as the case clearance rate . james and jang [ 9 ] , however , claim that the case clearance rate alone does not adequately represent an investigator or unit s investigation capacity .
they instead define investigation capacity as case clearance rates as related to time , where : + @xmath0 where : * @xmath1 is the time span for which capacity should be measured capability and capacity are related terms that have to do with quality . in investigations a highly capable investigator may be able to conduct quality investigations unless there are more investigation requests than he or she has the capacity to deal with at a time , at which point quality may be reduced .
many investigators have to cope with a lack of resources [ 10 ] , meaning that they are normally working at if not over capacity . because of this , when attempting to educate investigators both expansion of capability and capacity should be considered and measured .
this work proposes a holistic training development model specifically focused on cybercrime investigation that is based on improving investigator capability while also considering the capacity of the investigator or unit . along with a training development model , a cybercrime investigation capacity assessment framework
is given for attempting to measure capacity throughout the education process .
first , a training development model is proposed that focuses on the expansion of investigation capability as well as capacity of investigators and units .
next , a capacity assessment model is given to evaluate the effectiveness of the training program .
a description of how the proposed model is being applied to the development of training programs for cybercrime investigators in developing countries will then be given , as well as already observed challenges . finally , concluding remarks as well as proposed future work will be discussed .
the growing number and complexity of digital devices related to criminal acts means that digital investigators must attempt to increase their knowledge of a broad range of technologies , as well as increase the pace at which such devices can be analyzed [ 11 ] .
there are a number of challenges that digital investigators face , and education and training is a primary challenge for many investigation units today [ 7,10 ] .
not only is receiving training sometimes a challenge , but the way training is organized and delivered can also cause issues .
further , in many training programs the content may be such that investigators do nt or ca nt implement trained material in practical investigations
. such a situation could be the product of a training program that did not meet the needs of the trainees ; however , without post - training assessment there is no way to know whether skills and knowledge are being transferred to practice .
the korean national police , for example , are active contributors for building investigation capability and capacity with developing countries ; however , few development programs have adequately evaluated the effectiveness of such initiatives . with many organizations
, evaluation may be at the end of a training session and perhaps award the trainee a simple qualification upon completion .
however , short term trainings do not necessarily lead to long - term knowledge gain , even if a trainee does well on an examination immediately after the course [ 7 ] . also ,
because no further evaluation is normally done , training organizations have no way of assessing if knowledge or skills have been practically transferred . for the issues of generally targeting the training to the needs of the trainees , and ensuring that knowledge and skills
are being practically transferred for the improvement of capability and capacity , this work proposes a holistic training development methodology to increase cybercrime investigation capability and capacity .
as shown in figure [ fig : pdmdiagram ] , the method begins with an extensive assessment of the organization , attempting to determine what their goals , processes and current knowledge are . during the observation phase
, a training group should gain understanding of the current state of knowledge and skills of the organization to be trained as well as how the organization can / will implement new knowledge . in our experience , it is common for the trainers to find that organizations to be trained are either already knowledgeable about many of the areas , or that the organization has no skills in the area and may require remediation .
it is relatively uncommon to have a group , even within the same organization , in which all participants have the same level of knowledge and skills in a particular area . once a better understanding of how the organization functions has been established , the needs of the organization should be determined .
this stage should be approached with some skepticism as many organizations tend to have difficulty differentiating between wants and needs , and in some cases are simply ignorant of their actual needs .
based on knowledge gained in the observation phase , paired with lists of the organization s needs , a training developer should attempt to assess identified needs in the context of the organization s goals and current practices .
many organizations , for example , often identify ` more equipment ' as a need to expand capacity .
capacity , however , can be expanded in many ways , such is more efficient processes .
the training developer , then , should attempt to determine major problem areas at the national , organizational , divisional and personal levels , and evaluate what is needed to make a lasting positive impact on the organization . again considering more equipment as a example
, additional equipment may make up for inefficient processes in the short - term , but will exacerbate inefficient processes as capacity again reaches its peak .
once needs of the organization have been established , program development can begin .
there are many course development strategies , but we tend to prefer the successive approximation model ( sam ) .
sam is essentially a breadth first approach to curricula design , where the design is iterative from general to specific for approximately three iterations [ 12 ] .
sam is composed of three specific phases : preparation , design and development . *
preparation * * information gathering * * * background information * * * needs analysis * interactive design * * iterate through the design , prototype and review phases * * `` savvy start '' * * project planning * * additional design * interactive development * * iterate through development , implementation and evaluation * * design proof * * alpha version * * beta version * * final ` gold ' version sam is general and intuitive enough to be easily applied to many different areas .
iteration in the development and design phases means that the developer can be more abstract and focus in once they have more information and feedback , rather than picking a design and sticking strictly with it .
further , iterations of sam include an evaluation phase ensuring quality standards are maintained . in this case
, sam is used for course design for the manager and train - the - trainer programs .
as discussed , many training and education initiatives focus on the expansion of capabilities of investigators .
however , to improve capability and capacity , management must also be involved . in the proposed model
, management receives in - house training designed to suit the organization s identified needs .
such training could focus on improving work flows , data management , evidence intake , etc . as can be seen by a work flow process analysis in
[ 13 ] .
once initial in - house training has been given , management is expected to form action plans that begin to apply what they have learned .
created action plans should be practically implementable in their organization
. implementation of new knowledge is assessed and supplemented via practical management on - site training .
on - site training allows the trainer to better understand how or if the trainees are attempting to apply what was learned to improve processes in the organization .
this also allows the trainer to help point out potential areas of improvement , and answer practical implementation questions with the organization as an example .
the outputs of management training are crime management , investigation and prevention strategies that are beginning to be implemented .
this includes management s , possibly newly created , goals for investigator training and investigation needs .
training for investigators is developed in parallel with manager training , again using sam .
the first content to be created for investigators is the e - learning component .
e - learning allows busy organizations to complete some training at a time that is convenient for them .
it also allows trainers to assess trainees current level of knowledge , and provide remedial education , if necessary .
this will allow all participants to have a similar starting baseline knowledge , which is missing in many training programs .
next , in - house training for investigators is developed around the previously identified needs , using the same themes as the management training .
once , skills and knowledge for investigators have been developed , and practical feedback has been given from the management training , then investigators undergo domestic training that feeds directly into their investigation process .
these processes continue , and are supplemented with on - site training .
oftentimes when organizations receive training and education , trainees may make informal social network connections that may help in the application of learned skills and knowledge .
however , this support may not be responsive , consistent , or the best suited for helping organizations .
for this reason , online consultation with experts associated with the training organization is provided in this model . in this case , if trainees have questions , they can consult with experts in the knowledge domain rather than other practitioners that may also be learning these processes .
with such support , the trainees may be more confident about attempting to develop and implement better investigation processes and techniques .
the cybercrime investigation capacity assessment framework includes observation and assessment throughout the training program .
observation of the organization is conducted before and after each training initiative for the whole organization . the observation phase is composed of a survey of the organization to determine the basic level of skills and knowledge for the organization .
this survey , along with knowledge about organization processes determined during the first observation phase give the training organization a baseline with witch to compare skills and knowledge development .
after a training session has been delivered either in - house or on - site an assessment of the trainees is given to attempt to evaluate their short term understanding of the objectives . from three to six months after training has been received , the survey will again be administered to assess capability , and an assessment of changed processes will be conducted to evaluate efficiency improvements . finally , the previously described capacity measurement model [ 9 ] will be used to assess improvements in capacity over time . by using a qualitative survey , an observation of process changes , and a quantitative approach to capacity assessment , greater understanding of how the organization s process has change , and what the effects of this change are in terms of capacity can be determined .
this process has been developed in response to shortcomings observed in current cybercrime investigation training initiatives , especially in the area of assessment .
the overall process is currently being applied to three ( 3 ) year cybercrime investigation capacity building programs for 5 asian countries . in many of these countries
, such a project will be the first attempt to measure the countries cybercrime investigation capacity , and evaluate an organization s investigation process in relation with training .
such an initiative , however , is only a first effort towards measuring the effects of training on cybercrime investigators .
therefore the model , although currently general , is expected to be modified with experience .
this is especially true from the cybercrime investigation capacity assessment framework as better metrics for metrics are identified .
this work proposed a holistic training development model specifically focused on cybercrime investigation that is based on improving investigator capability while also considering the capacity of the investigator or unit .
along with a training development model , a cybercrime investigation capacity assessment framework was given for attempting to assess investigation capability and capacity throughout the education process .
such a model is proposed to a attempt to improve pre and post training assessment , as well as allowing training developers more information about the organization that will receive the training so such training can be better targeted , or remediation can take place before training begins if necessary . as discussed previously , this development and assessment model
is currently being applied to training programs with five asian countries .
ideally this model will allow training developers to understand the individual needs of each country , but will result in a relatively generalized program that meets the needs of each group .
future work will attempt to assess the weaknesses of each step of the model , especially with metrics used in the capacity assessment framework .
casey , e. ( 2004 ) . digital evidence and computer crime : forensic science , computers and the internet ( 2nd ed . , pp .
xviii , 690p . ) .
amsterdam , london : elsevier academic .
2 . rte .
text message evidence in murder trial .
rte news . retrieved from http://www.rte.ie/news/2010/0423/drimnagh.html 3 .
nguyen , l. ( 2012 ) .
tori stafford trial : cellphone record shows gap during abduction , murder . postmedia news .
retrieved from http://www.canada.com + /life
/ tori+stafford+trial+cellphone+record+shows+during+abduction+ + murder/6486178/story.html 4 .
philipp , j. ( 2013 ) . nearly every nyc crime involves cyber , says manhattan da . the epoch times .
retrieved from http://www.theepochtimes.com + /n2/united - states / nearly - every - nyc - crime - involves - cyber - says - manhattan - da-355692.html 5 .
\(2013 ) `` comprehensive study on cybercrime '' .
united nations office on drugs and crime .
http://www.unodc.org/documents/commissions + /ccpcj_session22/13 - 80699_ebook_2013_study_crp5.pdf 6 .
huber , e. ( 2010 ) .
certification , licensing , and accreditation in digital forensics . a fistful of dongles . retrieved from http://ericjhuber.blogspot.com + /2010/11/certification - licensing - and.html 7 .
james , j. i. , & gladyshev , p. ( 2013 ) .
challenges with automation in digital forensic investigations , 17 .
computers and society . retrieved from http://arxiv.org/abs/1303.4498 8 .
hekim , h. , gul , s. , & akcam , b. ( 2013 ) .
police use of information technologies in criminal investigations .
european scientific journal , 9(4 ) , 221 - 240 .
retrieved from http://eujournal.org/index.php/esj/article/view/778 9 .
james , j. i. , & jang , y. j. ( 2013 ) . measuring digital crime investigation capacity to guide international crime prevention strategies .
unpublished paper in the 7th international symposium on digital forensics and information security . 10 .
gogolin , g. ( 2010 ) .
the digital crime tsunami . digital investigation , 7(1 - 2 ) , 3 - 8 .
doi:10.1016/j.diin.2010.07.001 11 .
casey , e. , ferraro , m. , & nguyen , l. ( 2009 ) .
investigation delayed is justice denied : proposals for expediting forensic examinations of digital evidence .
journal of forensic sciences , 54(6 ) , 1353?1364 .
doi:10.1111/j.1556 - 4029.2009.01150.x 12 .
allen , m. w. ( 2003 ) .
michael allen s guide to e - learning : building interactive , fun , and effective learning programs for any company .
wiley . 13 .
james , j. i. , & gladyshev , p. ( 2013 ) .
a survey of digital forensic investigator decision processes and measurement of decisions based on enhanced preview .
digital investigation , 1 - 10 . | digital technologies are constantly changing , and with it criminals are finding new ways to abuse these technologies .
cybercrime investigators , then , must also keep their skills and knowledge up to date .
this work proposes a holistic training development model specifically focused on cybercrime investigation that is based on improving investigator capability while also considering the capacity of the investigator or unit .
along with a training development model , a cybercrime investigation capacity assessment framework is given for attempting to measure capacity throughout the education process .
first , a training development model is proposed that focuses on the expansion of investigation capability as well as capacity of investigators and units .
next , a capacity assessment model is given to evaluate the effectiveness of the training program .
a description of how the proposed model is being applied to the development of training programs for cybercrime investigators in developing countries will then be given , as well as already observed challenges .
finally , concluding remarks as well as proposed future work is discussed .
* keywords : * _ cybercrime investigation capacity assessment , cybercrime investigation capability , training development , education , law enforcement _ |
cosmic baryogenesis stands as one of the unresolved problems of particle cosmology .
most models address baryogenesis after the inflationary epoch .
recently the authors of ref .
@xcite demonstrated that the baryon asymmetry can be generated during inflation from gravity waves . in this model
the lepton number was generated by a quantum expectation value of the chern - simons density from ultra - violet ( uv ) , birefringent gravitational waves during the inflationary epoch . in a subsequent paper
the authors showed that this model can be embedded in string theory , in a model independent manner , through the green - schwarz mechanism @xcite . in the stringy embedding
there was a huge enhancement of the lepton asymmetry due to a hierarchy in the fundamental string scale and the four dimensional planck scale .
other investigators have searched for parity violation in the cosmic microwave background ( cmb ) which ultimately leads to a birefringence in gravitational waves @xcite . in this note , we shall study the super - horizon power spectrum and tensor to scalar ratio of scalar and tensor birefringent perturbations produced during inflation . specifically , we study the spectrum of super - horizon gravity waves whose uv counterparts were responsible for leptogenesis .
is it possible to see a signature of the leptogenesis mechanism in the super - horizon power spectrum ? to address this question we shall derive the tensor to scalar ratio and show that it contains a direct signature of the leptogenesis mechanism which occurred in the uv . furthermore we show that the scalar to tensor ratio contains the string scale in a model independent way and is in an observable window to this physics .
the paper is organized as follows . in section
ii we derive the equations for the gravitational waves in the presence of the chern - simons term . in section iii we provide the exact solutions at various scales and
derive the power spectrum as well as the corrected tensor to scalar ratio . in section
iv we relate this modified tensor to scalar ratio to the stringy embedding of gravitational leptogenesis and we conclude with some open issues concerning a consistent quantization and further directions .
the starting point of inflationary leptogenesis is the einstein - hilbert action coupled to the gravitational chern - simons term , which is necessarily present in string theory .
this last term can be written as @xmath3 where @xmath4 , @xmath5 being the planck mass .
we proceed to linearize the einstein - hilbert action with the chern - simons term in a friedmann - lematre - robertson - walker ( flrw ) background in the presence of tensor perturbations ( _ i.e .
_ in presence of gravitational waves ) .
the corresponding metric tensor takes the form ( assuming that the space - like sections are flat ) @xmath6 \ , , \end{aligned}\ ] ] with @xmath7 , being a transverse and traceless tensor , i.e. @xmath8 , @xmath9 and @xmath10 , the flrw scale factor , being a function of the conformal time @xmath11 . due to the symmetries of the flrw metric , the inflaton field @xmath12 in eq .
( [ csaction ] ) is also a function of the conformal time only .
expanding the action up to second order in the gravitational waves tensor @xmath7 ( which is necessary in order to obtain first order equations of motion ) , after lengthy but straightforward calculations , one obtains the following expression @xmath13 -f'\epsilon ^{ijk } \biggl[\left(h^q{}_i\right)'\left(\partial _
jh_{kq}\right ) ' -\left(\partial ^rh^q{}_i\right)\partial _ j\partial _
rh_{kq}\biggr]\biggr\}\ , , \end{aligned}\ ] ] where a prime stands for a derivative with respect to conformal time and @xmath14 , @xmath15 being the totally antisymmetric tensor . in the above expression
, one recognizes the standard ( _ i.e . _ einstein - hilbert )
expression of the perturbed action ( first term between squared brackets ) while the term proportional to @xmath16 represents the correction coming from the chern - simons contribution . varying this action with respect to the gravitational waves tensor ,
one obtains the first order equation of motion which reads @xmath17 & = & 0 \ , , \end{aligned}\ ] ] next , following ref .
@xcite , we define the tensor @xmath18 by the following equation @xmath19 and , then , the equation of motion takes the form @xmath20=0 \ , , \ ] ] where we have defined @xmath21 .
this equation is similar to eqs . (
11 ) and ( 12 ) of ref .
@xcite , except that we have written the equation in terms of the conformal time rather than in terms of the cosmic time .
the next step consists in going to the fourier space . for this purpose ,
we write the metric tensor as @xmath22 in the above expression , @xmath23 is the linear polarization tensor ( @xmath24 corresponds to @xmath25 ) . concretely ,
if the wave - vector is written in polar coordinates as @xmath26 , then two vectors perpendicular to @xmath27 are given by @xmath28 for the first vector and @xmath29 for the second vector but only if @xmath30 . if @xmath31 , _ i.e. _
if the wave - vector points to the bottom , the expression of the second vector should in fact read @xmath32 .
it is also interesting to mention how these quantities transform under the change @xmath33 .
it is easy to see that this corresponds to the transformation @xmath34 .
then , we have @xmath35 and @xmath36 .
finally , the polarization tensor can be written as @xmath37 due to the properties of the vectors @xmath38 and @xmath39 established above , it is easy to check that @xmath40 and @xmath41 . using these properties and the fact that @xmath7 is real , @xmath42 , one can also establish that @xmath43 the next step consists in defining two other states of polarization , the so - called right and left polarization states .
the corresponding polarization tensors are given by @xmath44 from the above expressions , using the properties of the linear polarization tensors , one can show that @xmath45 these expressions are of course valid only if the polarization tensors are evaluated for the same wave - number .
we also have @xmath46 with @xmath47 . then , using the expression of the vectors @xmath48 and @xmath49 , it is easy to show that @xmath50 where @xmath51 and @xmath52 and where the upper sign refers to @xmath30 while the lower one refers to @xmath31 .
we are now in a position where one can re - write the gravitational waves tensor in terms of the left and right polarization states .
this gives @xmath53 where we have introduced the definitions @xmath54 let us notice that it is straightforward to demonstrate that @xmath55 and @xmath56 .
the next step consists in introducing the new expansion of the gravitational waves tensor given by eq .
( [ gwpolalr ] ) into the equation of motion and in using eq .
( [ eigenpola ] ) to arrive at a new form of the equation of motion .
one obtains @xmath57 finally , the last step consists in introducing the quantity @xmath58 defined by @xmath59 and the new amplitude @xmath60 defined by @xmath61 .
then , the equation of motion for @xmath62 has the traditional form of the equation of motion for a parametric oscillator , namely @xmath63 the effective potential @xmath64 depends on time , on polarization ( birefringence ) but also on the wave - number which is an important difference with respect to the standard case where the effective potential depends on conformal time only .
this equation has been derived for the first time in ref .
@xcite ; see eq .
( 15 ) of that paper . however , in ref .
@xcite , it is also assumed that the effective potential takes the form @xmath65 where @xmath66 is a constant . in particular , one notices that , with this ansatz , the scale dependence of the effective potential has disappeared .
this permits to find simple solutions in terms of bessel functions .
however , we will see that , in the present context , the effective potential is different and more complicated .
let us now calculate the effective potential explicitly . using the formulas established previously
, one obtains that the exact expression of the potential can be written as @xmath67 ^ 2}{\left[1-\lambda ^sk\left(f'/a^2\right)\right]^2}\ , .\end{aligned}\ ] ] to go further , we need to postulate the function @xmath68 . following ref .
@xcite , we choose @xmath69 where @xmath70 is the reduced planck mass and @xmath2 a number that we discuss in more details in the last section and that can be related to the string scale . with this definition @xmath71
is dimensionless , as it should , if the scale factor has the dimension of a length ( which is our convention ) . in terms of the slow - roll parameters
@xmath72 , @xmath73 and @xmath74 ( a dot means a derivative with respect to cosmic time ) , we have at leading order in the slow - roll parameters , see also ref .
@xcite @xmath75 from this expression , one deduces that ( at leading order in the slow - roll parameters ) @xmath76 because @xmath77 . from that expression , one arrives at the two following formulas which are useful in order to calculate the effective potential @xmath78 and @xmath79 inserting the above equations into the formula giving the potential , namely eq .
( [ poteff ] ) , one obtains @xmath80^{-1 } \nonumber \\ & & -\frac{k^2}{4}\frac{{\cal n}^2}{256\pi ^4}\left(\frac{h_{_{\rm inf}}}{m_{_{\rm pl}}}\right)^4(2\epsilon ) \left[1-\lambda ^s\frac{{\cal n}}{16\pi ^2 } \left(\frac{h_{_{\rm inf}}}{m_{_{\rm pl}}}\right)^2 \sqrt{2\epsilon } \times ( k\eta ) \right]^{-2}\ , , \end{aligned}\ ] ] where we have ignored sub - dominant term in the slow - roll parameters .
it is important to notice that , in order to obtain the above equation , we have never expanded a term like @xmath81 in the slow - roll parameters .
we notice the presence of @xmath82 in the numerator of the second term .
this is in full agreement with ref .
@xcite where it has been noticed that a term like @xmath83 in the effective potential necessarily implies a new characteristic scale . here , the characterized scale defined in ref .
@xcite could be written as ( at this level , the two situations are not yet totally equivalent because the above effective potential is not exactly similar to the potential studied in ref
. this will be the case below . )
now let @xmath84 where we have defined @xmath85 by the following relation .
( see eq .
( 13 ) in ref .
@xcite ) : @xmath86 in the present context , somehow , the characteristic scale @xmath87 `` depends on the scale '' ( _ i.e . _ on @xmath82 ) . however , we see that the large - scale limit , as defined in ref .
@xcite _ i.e. _ @xmath88 , corresponds in the present context to the condition @xmath89 .
the only way to satisfy this condition is to have a large @xmath2 which could compensate the smallness of @xmath90 and of the slow - roll parameter . for convenience ,
we now introduce the variable @xmath91 defined by @xmath92 .
then , the equation of motion takes the form @xmath93\mu = 0\ , , \ ] ] with @xmath94 the functions @xmath95 and @xmath96 are represented in fig .
[ pot ] . from this figure ,
the different behavior of the two states of polarization is apparent .
the l mode ( dashed line ) undergoes a `` kick '' at @xmath97 where the effective potential blows up . at the same point
the potential of the r mode is perfectly regular ( solid line ) .
therefore , we expect the r mode function to propagate smoothly through @xmath97 while the behavior of the l mode function can be more problematic .
we now turn to this question in more details .
let now us study the equation of motion for the left mode in the vicinity of @xmath98 .
it is easy to check that a very good approximation of the potential is @xmath99 in fact the approximation is good even far form @xmath98 provided @xmath100 since , on small scales , _
i.e. _ in the limit @xmath101 , we have @xmath102 . in the limit
, the solution can be written as @xmath103 where @xmath104 and @xmath105 are two constants that are fixed by the choice of the initial conditions .
usually , one requires that , on sub - hubble scales @xmath106 this prescription completely fixes the coefficients @xmath104 and @xmath105 which read @xmath107 in the above equation , @xmath108 is the planck length and @xmath109 is some initial time at the beginning of inflation .
the knowledge of this time is not important since it will drop out from the final result . with the potential given by eq .
( [ approxpotdiv ] ) , the equation of motion can be solved exactly .
indeed , if we define @xmath110 then the equation of motion takes the form @xmath111\mu _ { \mathbf k}^{_{\rm l}}=0\ , .\ ] ] this is the well - known whittaker equation , see eq .
( 9.220.1 ) of ref .
the corresponding solution , correctly normalized , see eqs . ( [ cidiv ] ) , reads @xmath112\ , , \ ] ] where @xmath113 is the whittaker function .
let us now study how the mode function behaves when @xmath114 .
the whittaker function can be expressed in terms of the confluent hypergeometric function , see eq .
( 13.1.33 ) of ref .
one obtains @xmath115\ , , \end{aligned}\ ] ] where @xmath116 is the above - mentioned confluent hypergeometric function . using eq .
( 13.5.9 ) of ref .
@xcite which says that , when @xmath117 , @xmath118/\gamma ( a)$ ] , where @xmath119 is the euler s integral of the second kind and where @xmath120 , see ref .
@xcite , one deduces that @xmath121 but what really matters is not the intermediate variable @xmath122 but in fact the amplitude of the gravitational waves itself given by @xmath123 , see eq .
( [ defz ] ) . since @xmath124 ,
one obtains @xmath125 the conclusion is that the amplitude of the mode @xmath126 blows up at the time corresponding to @xmath97 , that is to say at the time @xmath127 defined by @xmath128 at this point the linear theory of cosmological perturbations breaks down and becomes non linear .
an important feature of @xmath129 is that it is scale dependent .
this means that the physical wavelength of the fourier modes @xmath130 , at time @xmath131 , are all equal to the same physical length .
explicitly , one has @xmath132 somehow , this is reminiscent of one of the possible formulations of the trans - planckian problem of inflation @xcite where it is postulated that a mode of comoving wavenumber @xmath82 is `` created '' when its physical wavelength equals a given new fundamental scale in the theory ( the idea being to test the robustness of the inflationary predictions to short distance modifications of the theory ; therefore , it is typical in this context to consider that the new scale is the planck length , see refs .
@xcite for more details ) .
it is then easy to show that the `` time of creation '' is inversely proportional to @xmath82 as it is the case for @xmath129 , see in particular the fifth paper in ref .
@xcite . as a consequence ,
we see from eq .
( [ lengththeta ] ) that @xmath85 defines in fact a new scale the value of which depends on the inflation scale but also on the string scale since we will see that the string scale is hidden into the number @xmath2 which participates to the definition of @xmath85 , see eq .
( [ deftheta ] ) . a possible way out to the question of the divergence would be to push the problem in the trans - planckian regime . from the above equation
, this means that the parameter @xmath133 should satisfy @xmath134 therefore , this boils down to a quite stringent constraint on @xmath133 , typically @xmath135 .
unfortunately , we will see in the following that , for such small values of @xmath85 , the modifications on large scales are not observable .
if one wants to consider larger values of @xmath85 , it seems that much more refined ( _ i.e .
_ non - linear ) calculations are necessary .
this calculation are obviously beyond the scope of the present paper which is just exploratory .
let us conclude this subsection by stressing the fact that , a priori , trans - planckian effects do not play a deep role in the chern - simons theory under considerations in this work . here
, it is merely a technical trick which allows to avoid the non - linear regime and to adopt the common assumption that the fourier modes emerge from the trans - planckian region in the vacuum state . but clearly ,
if @xmath136 , then the non - linear calculation is in principle feasible without any trans - planckian considerations .
let us now study what happens on very large scales , _
i.e. _ in the limit where @xmath91 vanishes .
in the situation , the effective potentials can be very well approximated by the following equations @xmath137 where we remind that @xmath51 and @xmath52 .
birefringence enters this equation via the term proportional to @xmath138 which changes its sign according to the polarization state under considerations .
the term proportional to @xmath139 is the standard slow - roll term .
the corresponding equation of motion takes the form @xmath140\mu _ { \mathbf k}^s=0\ , .\ ] ] once again , we have to deal with a whittaker equation .
in fact this equation ( and the corresponding power spectrum ) has been studied in details in ref .
@xcite , see eq .
( 8) , and the corresponding power spectrum has been derived in that reference , see eq .
therefore , in the present paper we can use the results obtained in ref .
@xcite and follow the procedure utilized in that reference .
let us introduce the new definitions @xmath141 in the following , we will consider that @xmath142 ( see the discussion at the end of the previous subsection ) and , as a consequence , will simply approximate @xmath143 by @xmath144 . with the new definitions taken into account , the equation of motion takes the form @xmath145\mu _ { \mathbf k}^s=0\ , , \ ] ] which is again the whittaker equation .
the situation is exactly similar to the one studied around eq .
( 12 ) of ref .
the exact general solution to this equation is given in terms of whittaker functions @xmath146 where @xmath147 and @xmath148 are two constants fixed by the choice of the initial conditions .
as discussed in the preceding subsection , we will fix the initial conditions in the region @xmath149 which is free of divergences . in this regime ,
the only natural choice that we have is to postulate a plane wave .
this is equivalent to postulating that the non - linear phenomena occurring around the divergence of the effective potential , provided they happen in the trans - planckian region , will not affect the standard choice of the initial conditions in the region @xmath150 .
somehow , this is the same assumption that is made in the standard inflationary scenario . indeed , despite the fact that the modes of astrophysical interest today originate from the trans - planckian region @xcite , the vacuum is assumed to be the correct initial state .
let us stress , however , that a possible weakness of the above comparison is that , in the case of the trans - planckian problem of inflation @xcite , one can show that the final result can be robust to changes in the short distance physics @xcite ( under some conditions , _
i.e. _ adiabatic evolution of the fourier modes ) . in the present context
, however , it is more difficult to imagine that the non - linearities will not affect the initial conditions . on the other hand , in the absence of a second - order calculations and as a first approach to the problem , this seems to be quite reasonable . as shown in ref .
@xcite , see eqs .
( 14 ) , this choice amounts to @xmath151 where we have used the fact that , see eq .
( 9.227 ) of ref .
@xcite , @xmath152 .
the sign of the argument in the exponential depends on the polarization state considered , as expected .
we conclude that the solution to the mode equation on very large scales is now known explicitly .
usually , the power spectrum is given by the two - point correlation function calculated in the vacuum state .
another way to calculate the same quantity is to view it as a classical spatial average .
since a fully consistent quantum formulation of the present theory is not yet available , we adopt the second point of view .
therefore , the two - point correlation function can be written as @xmath153 with @xmath154 is the total volume .
using the properties of the polarization tensor , straightforward calculations show that @xmath155 from which we deduce the power spectrum @xmath156 let us notice that , usually , the power spectrum is proportional to the factor @xmath157 .
here we do nt have the factor @xmath158 because we consider the two states of polarization separately ( _ i.e .
_ usually , these two states are summed and produce the factor @xmath158 ) . a priori
, using the solution obtained in the previous subsection , we can calculate the spectrum exactly in terms of the whittaker function .
but only the spectrum on large scales is needed and in this regime one has ( for details , see ref .
@xcite ) @xmath159 let us notice that we have neglected the factor @xmath160 because @xmath161 is proportional to @xmath162 and hence negligible on large scales .
the above expression is similar to eq .
( 15 ) of ref .
@xcite . at this stage
, the only thing which remains to be done is to expand the above expression at first order in the slow - roll parameter .
after lengthy but straightforward calculations , one obtains the following result @xmath163 \ , , \end{aligned}\ ] ] with , @xmath164 at this point some remarks are in order .
as required one can check that , when @xmath165 , the standard inflationary result is recovered .
this is the case because @xmath166 and @xmath167 .
as already mentioned , a factor @xmath168 is left because the ( now identical ) contribution from the two states of polarization should be added .
the function @xmath169 describes the dominant modification in the amplitude of the power spectrum ( the contribution originating from the function @xmath170 is clearly sub - dominant since it is proportional to the slow - roll parameter @xmath171 ) . for small values of @xmath85
we have @xmath172 where @xmath173 and @xmath174 .
therefore , the amplitude of the right polarization state is reduced while the one of the left polarization state is enhanced . however , for small values of @xmath85 , the effect is clearly not very important .
another conclusion that can be obtained from the above spectrum is that , at leading order in the slow roll parameter , the spectral index remains unmodified .
indeed , one has @xmath175 for each polarization state . finally , let us now compute how the ratio @xmath0 is modified .
the scalar power spectrum is not modified ( see also ref .
@xcite ) and reads @xcite @xmath176\ , .\ ] ] therefore , we conclude that the consistency check of inflation , at leading order in the slow - roll parameters , can now be written as @xmath177 \\ & \simeq & 16\epsilon \times \left[1+\left(\frac{\pi ^2}{384}-\frac{1}{256}\right)\theta ^2\right]\ , .\end{aligned}\ ] ] unfortunately , the linear corrections in @xmath85 cancels out and we are left with a correction which is quadratic in @xmath85 .
another way to express the above result is to calculate the ratio of @xmath0 with the chern - simon modification taken into account to @xmath0 obtained in the standard case .
one gets @xmath178 it is clear from this expression that the modification is not observable at all since we have seen before that , typically , @xmath179 in order for the calculations presented here to be consistent ( _ i.e .
_ for the divergence of the effective potential to be in the trans - planckian region ) .
we have evaluated the super - hubble power spectrum and the tensor to scalar ratio for birefringent gravitational waves produced during inflation .
the power spectrum exhibits two interesting regimes , linear and non - linear .
the non - linear regime occurs when @xmath180 because the effective potential controlling the evolution of the linear perturbations blows up . at this point ,
the linear theory of cosmological perturbations that we used is no longer valid .
this divergence occurs for all modes ( _ i.e .
_ for all comoving wavenumber @xmath82 ) but at different times . in this present investigation we only considered the linear regime since at the present moment we were not able to perform a rigorous analysis of the non - linear phenonema
we found corrections which survive to second order in @xmath181 .
therefore , in this regime the tensor to scalar ratio gets corrected by @xmath181 but this effect is very small .
if @xmath135 , one can push the non - linear regime ( _ i.e . _
the divergence in the effective potential ) into the trans - planckian region where , anyway , other effects ( for instance , non perturbative stringy effects ) are likely to become important . somehow , this corresponds to the standard situation where the evolution from the planck scale to the super - horizon scales is under control and where the perturbations are assumed to emerge from the trans - planckian regime in the vacuum state , thus ignoring the modifications of the initial conditions that the trans - planckian physics could cause ( in the very same way that we have ignored the effect of the divergence in the potential , provided it is in the trans - planckian region ) .
however , it is important to keep in mind that this is mostly a technical trick which allows us to work with the linear theory . at a deeper level
, the trans - planckian effects are not expected to play a more important role than in the standard situation . in particular ,
if the divergence is not in the trans - planckian regime , only the non - linear theory of cosmological perturbations is necessary in order to calculate the modified @xmath0 irrespectively of any trans - planckian effects .
it is interesting to note that the linear regime ( where @xmath182 ) is compatible with the stringy embedding of inflationary baryogenesis ( seb ) @xcite . in this context
, the value of @xmath181 enhances and gives the resonant frequency associated with the observed baryon asymmetry .
as already mentioned before , this value is completely fixed by the string scale and coupling in a model independent fashion .
explicitly , the value of the number @xmath2 which appears in the definition of @xmath85 , see eq .
( [ deftheta ] ) , is given by @xmath183 where @xmath184 is the ten - dimensional fundamental scale and @xmath185 is the string coupling .
therefore , we established a direct link between stringy quantities and cmb anisotropies .
explicitly , eq . ( [ cccheck ] ) can be re - written as @xmath186 in a recent paper the authors of ref .
@xcite found that for reasonable values of string coupling ( _ i.e . _
weak ) and the string scale set to @xmath187 , both @xmath85 can be as small as @xmath188 and the observed baryon asymmetry can be generated .
of course the stringy embedding admits much larger values of @xmath181 putting our analysis into the non - linear regime . if @xmath189 , a non - linear calculation is mandatory and one can hope to obtain a significative modification of the ratio @xmath0 , maybe observable by future high accuracy cmb experiments .
clearly , the present situation is not very satisfactory since the regime for which sizable effects are expected turns out to be very complicated from the technical point of view .
furthermore , we suspect that the scale associated to the divergence of the effective potential corresponds to a resonant production of lepton number .
we wish to report on this issue in a future paper .
m. lemoine , m. lubo , j. martin and j. p. uzan , phys .
d * 65 * , 023510 ( 2002 ) , hep - th/0109128 ; j. c. niemeyer , phys .
d * 63 * , 123502 ( 2001 ) , astro - ph/0005533 ; a. kempf , phys .
d * 63 * , 083514 ( 2001 ) , astro - ph/0009209 ; r. easther , b. r. greene , w. h. kinney and g. shiu , phys .
d * 64 * , 103502 ( 2001 ) , hep - th/0104102 ; j. martin and r. h. brandenberger , phys .
d * 68 * , 063513 ( 2003 ) , hep - th/0305161 .
j. martin and c. ringeval , phys .
d * 69 * , 083515 ( 2004 ) , astro - ph/0310382 ; j. martin and c. ringeval , phys .
d * 69 * , astro - ph/0402609 ; j. martin and c. ringeval , astro - ph/0405249 . | in this work we show that the gravitational chern - simons term , aside from being a key ingredient in inflationary baryogenesis , modifies super - horizon gravitational waves produced during inflation .
we compute the super - hubble gravitational power spectrum in the slow - roll approximation and show that its overall amplitude is modified while its spectral index remains unchanged ( at leading order in the slow - roll parameters ) .
then , we calculate the correction to the tensor to scalar ratio , @xmath0 . we find a correction of @xmath0 which is dependent on @xmath1 ( more precisely quadratic in @xmath2 ) , the parameter characterizing the amplitude of the chern - simons terms . in a stringy embedding of the leptogenesis mechanism , @xmath1 is the ratio between the planck scale and the fundamental string scale .
thus , in principle , we provide a direct probe of leptogenesis due to stringy dynamics in the cosmic microwave background ( cmb ) .
however , we demonstrate that the corresponding correction of @xmath0 is in fact very small and not observable in the regime where our calculations are valid . to obtain a sizable effect
, we argue that a non - linear calculation is necessary . |
the detection of compact signals ( sources ) embedded in a background is a recurrent problem in many fields of science .
some common examples in astronomy are the separation of individual stars in a crowded optical image , the identification of local features ( lines ) in noisy one - dimensional spectra or the detection of faint extragalactic objects in microwave frequencies . regarding the detection of point sources on maps of the cosmic microwave background radiation ( cmb ) , several techniques based on different linear filters
have been proposed in the literature : the mexican hat wavelet ( mhw cayn et al.@xcite , vielva et al.@xcite ) , the classic _ matched _ filter ( mf , tegmark and de oliveira - costa@xcite ) , the adaptive top hat filter ( athf , chiang et al.@xcite ) and the scale - adaptive filter ( saf , sanz et al.@xcite , herranz et al.@xcite ) . a certain deal of controversy has appeared about which one , if any , of the previous filters is _ optimal _ for the detection of point sources in cmb data . in order to answer that question it is necessary to consider first a more fundamental issue , the concept of _ detection _ itself .
the detection process can be posed as follows : given an observation , the problem is to _ decide _ whether or not a certain signal was present at the input of the receiver .
the decision is not obvious since the observation is corrupted by a random process that we call ` noise ' or ` background ' .
formally , the _ decision _ is performed by choosing between two complementary hypotheses : that the observed data is originated by the background alone ( _ null hypothesis _ ) , and the hypothesis that the observation corresponds to a combination of the background and the signal .
to decide , the detector should use the totality of the available information in terms of the probabilities of both hypotheses given the data . the _ decision device _ separates the space @xmath0 of all possible observations in two disjoint subspaces , @xmath1 and @xmath2 , so that if an observation @xmath3 the null hypothesis is accepted , and if @xmath4 the null hypothesis is rejected , that is , a source is ` detected ' .
hence , we will call any generic decision device of this type a _
detector_. any detector can produce two kinds of errors : on the one hand , it can produce a _ false alarm _ or _ spurious detection _ when an observation in which no source was present is assigned to the subspace @xmath1 .
the probability of this kind of error depends on the statistical properties of the background and the choice of the detector . on the other hand ,
a signal that is present in the observation can be missed by the detector ( i. e. the observation is wrongly assigned to the subspace @xmath2 ) .
this error is often referred as _ false dismissal_. the probability of false dismissal depends on the statistical properties of the background , the choice of the detector and the properties of the signal ( for example , its intensity ) .
in general , it is not possible to decrease the incidence of both types of error at the same time : one of them can be reduced at expense of increasing the other .
the goodness of a given detector must be established by taking into account the balance between these two types of error .
the most simple example of detector , and one that has been exhaustively used in astronomy , is _
thresholding_. thresholding considers that the space @xmath0 of observations consists of all the possible values of the measured intensity @xmath5 ( in the case of an astronomical image ) and subdivides this space into two simple regions @xmath6 and @xmath7 .
the value @xmath8 is an arbitrarily chosen _ threshold _ that is often expressed as a number of times the standard deviation of the background , @xmath9 . thresholding works on the assumption that the probability of finding a value of @xmath5 due to the background decreases as the value of @xmath5 increases . in the case of a gaussian background ,
this assumption has a very precise meaning and it allows us to straightforwardly control the probability of occurrence of spurious detections simply by setting a large enough threshold .
however , this may lead to a very high probability of false dismissals .
unfortunately , in many cases the sources are very faint and this makes very difficult to detect them : a high threshold means that the number of detections will be very small .
here is where _ filtering
_ enters in scene .
the role of filtering is to transform the data in such a way that a detector can perform better than before filtering .
for example , a filter can be designed in order to reduce the fluctuations of the background so that we can safely use lower detection thresholds and , hopefully , increase the number of detections without increasing the number of spurious detections .
we remark that _ a filter is not a detector _ : the decision device we call ` detector ' can be applied after the application of any imaginable filter , or even no filter at all , while the use of any filter without a posterior detection criterion means nothing
. however , the two different steps in the process ( filtering and detection ) are not independent . in the thresholding example , the use of a filter that cancels most of the fluctuations in the background allows us to change the detection threshold from its original high value to a lower one .
given an adopted detector and a background , it is licit to ask which is the filter that creates the most favorables conditions in the filtered background for the detector to perform . in other words ,
the ` optimality ' of a filter for detection depends on the type of detector chosen which , in turn , depends on the specific _ goal _ the observer has in mind : in certain cases the observer will accept a relatively large number of spurious detections in order to have a large number of true targets , whereas in other cases it could be more important to be certain that the detections are all of them reliable , and so on .
for example , let us consider that we have chosen thresholding as our detection device . in that case , it has been shown that the optimal linear filter is the matched filter .
it produces the maximum amplification of the signal with respect to the background fluctuations , so that the threshold for a given probability of spurious detections is minimum , allowing the thresholding detector to find more sources than would be detected if we filtered with any other filter .
a sub - optimal approach , using only thresholding , is to select a priori a filter and adapt its scale in order to produce a maximum amplification in a given background .
such is the case of the mexican hat wavelet at the optimum scale ( mho , vielva et al.@xcite ) and the adaptive top hat filter ( athf , chiang et al.@xcite ) .
thresholding has a number of advantages , among them the facts that it is straightforward , it has an obvious meaning in the case of gaussian backgrounds , and it has been successfully used for many years in many fields of science .
it , however , does not use all the available information contained in the data to perform decisions .
the question is then : is it possible to devise a detector that uses additional information apart from mere intensities and that produces better results than thresholding ? and , if so , which is the filter that optimizes the performance of such a detector ?
let us focus on the case of one - dimensional data ( such as stellar spectra , or time - ordered scannings of the sky in cmb experiments ) and linear filters .
data in a one - dimensional array is entirely described by two quantities , namely the position in the array ( corresponding to the spatial or temporal coordinate , for simplicity we will refer it as _ spatial _ information ) and the value ( intensity ) at each position . thresholding uses only the intensity distribution to make the decision .
clearly , the inclusion of spatial information in a detector should be useful .
for example , it could help to distinguish the sources from fluctuations in the background with similar scale but a different shape . a full description of this `
spatial information ' should include the probability distribution of events ( both due to background and sources ) in space , with all its infinite moments .
we will somewhat relax this demand of information assuming that the background is homogeneous and isotropic , and asking at each point for some information about the shape of the sources and the autocorrelation of the background ( for example , the curvature of the peaks ) .
in fact , even a simple filtering - and - thresholding scheme uses implicitly some degree of spatial information .
both the mho and the athf adapt to the scale at which the contrast between sources and background produces the maximum amplification .
the mf includes as well the information on the profile of the sources in order to amplify the structures whose shape correlates with the shape of the sources .
saf goes a step further in constraining additionally the scale of the filter .
moreover , in most cases the detection is performed not in all the points of the data but only in the peaks , that is , _ in those points where the curvature is positive_. in a recent work , barreiro et al.@xcite propose a detection criterion based on the neyman - pearson decision rule that uses the information of both the intensities @xmath5 and the curvatures @xmath10 of the peaks in a data set . in that work the performances of several filters ( saf , mf and mhw ) is compared in terms of their _ reliability _ , defined as the ratio between the probability density of true detections over the probability density of spurious detections .
they find that , on the basis of this quantity , the choice of the optimal filter depends on the statistical properties of the background . for the case of backgrounds that can be described with a power spectrum of the form @xmath11 , the saf outperforms the other two filters for the case @xmath12 , whereas in the range @xmath13 the mf is the most reliable .
the mhw is the most reliable filter in this sense when @xmath14 .
the reliability , defined in the previous sense , could not be a valid measure of performance of the filters , since it favors a situation in which the number of the detections is very low in order to keep a ` safe ' number of spurious detections .
a different approach can be used in which the number density of spurious detections ( or , alternatively , of true detections ) is fixed for all the filters , and then the number density of true detections ( or spurious detections ) is compared for all the filters . in this work ,
we first clearly define the _ goal _ of our experiment : fixed a certain number density of spurious detections , to obtain the maximum possible true detections from the data .
note that this goal is not universal : in other applications , an observer may desire to work on the basis of the reliability described in barreiro et al.@xcite , or define its own requirements .
once fixed the goal , we will develop a detection criterion that corresponds to a bayesian generalized neyman - pearson test .
then , a particular filter from the family of matched - type filters ( mtf ) that optimizes the performance of the detector under the optimality conditions set by our goal will be obtained .
finally , the performance of the optimal mtf will be compared with the standard mf using simulations .
the overview of this paper is as follows : in section 2 , we introduce two useful quantities : number of maxima in a gaussian background in the absence and presence of a local source . in section 3 , we introduce the detection problem and define the region of acceptance . in section 4
, we obtain different analytical and numerical results regarding point sources and scale - free background spectra and compare the performance of a new family of matched - type filters . in section 5 , we describe the numerical simulations performed to test some theoretical aspects and give the main results .
finally , in section 6 , we summarize the main results and applications of this paper .
let us assume a 1d background ( e. g. one - dimensional scan on the celestial sphere or time ordered data set ) represented by a gaussian random field @xmath15 with average value @xmath16 and power spectrum @xmath17 : @xmath18 , where @xmath19 is the fourier transform of @xmath15 and @xmath20 is the 1d dirac distribution .
the distribution of maxima was studied by rice@xcite in a pioneer article . the expected number density of maxima per intervals @xmath21 , @xmath22 and @xmath23 is given by @xmath24 being @xmath25 the expected total number density of maxima ( i. e. number of maxima per unit interval @xmath26 ) @xmath27 @xmath28 where @xmath29 and @xmath30 represent the normalized field and curvature , respectively .
@xmath31 is the moment of order @xmath32 associated to the field .
@xmath33 are the coherence scale of the field and maxima , respectively .
if the original field is linear - filtered with a circularly - symmetric filter @xmath34 , dependent on @xmath35 parameters ( @xmath36 defines a scaling whereas @xmath37 defines a translation ) @xmath38 we define the filtered field as @xmath39 then , the moment of order @xmath40 of the linearly - filtered field is @xmath41 being @xmath42 the power spectrum of the unfiltered field and @xmath43 the fourier transform of the circularly - symmetric linear filter . now , let us consider a gaussian source ( i. e. profile given by @xmath44 ) embedded in the previous background .
then , the expected number density of maxima per intervals @xmath21 , @xmath22 and @xmath45 , given a source of amplitude @xmath46 in such spatial interval , is given by ( barreiro et al.@xcite ) @xmath47 where @xmath29 and @xmath30 , @xmath48 is the normalized amplitude of the source and @xmath49 is the normalized curvature of the filtered source .
the last expression can be obtained as @xmath50 we consider that the filter is normalized such that the amplitude of the source is the same after linear filtering : @xmath51 .
we want to make a decision between filters based on _ detection_. to make such a decision , we will focus on the following two fundamental quantities : a ) the number of spurious sources which emerge after the filtering and detection processes and b ) the number of real sources detected . as seen in the previous section , these quantities are properties of the gaussian field and source that can be calculated through equations ( [ nbackground ] ) and ( [ nsource ] ) .
as we will see , the previous properties are not only related to the signal - to - noise - ratio gained in the filtering process but depend on the filtered momenta to 4th - order ( in the 1d case ) , i. e. the amplification and the curvature of the source . let us consider a local peak in the 1d data set characterized by the normalized amplitude and curvature @xmath52 .
let @xmath53 represents the _ null _ hypothesis , i. e. the local number density of background maxima , and @xmath54 represents the _ alternative _ hypothesis , i. e. the local number density of maxima when there is a compact source with normalized amplitude and curvature @xmath55 .
given the data @xmath56 , we can associate to any region @xmath57 two number densities @xmath58 and @xmath59 @xmath60
@xmath61 then , @xmath58 is the number density of spurious sources , i. e. due to the background , expected inside the region @xmath57 , whereas @xmath59 is the number density of maxima expected in the same region of the @xmath62 space in the presence of a local source .
hereinafter , we will call it the _ number density of detections_. we will assume a bayesian approach : at a concrete pixel we get the number of source detections weighting with the _ a priori _ probability @xmath63 .
@xmath1 is called the _ acceptance _ region .
we remark that in order to get the true number of real source detections such a number must be multiplied by the probability to have a source in a pixel in the original data set .
we will assume a bayesian generalized neyman - pearson decision rule using number densities instead of probabilities : the acceptance region @xmath1 giving the highest number density of detections @xmath59 , for a given number density of spurious @xmath58 , is the region ( criterion for detection ) @xmath64 we decide that the signal is present , whereas if @xmath65 we decide that the signal is absent .
once we have assumed the previous decision rule for detection , the region of acceptance @xmath1 is given by @xmath66 or equivalently by the sufficient linear detector @xmath67 where @xmath68 is a constant and @xmath69 is given by @xmath70 we remark that the assumed criterion for detection leads to a _ linear _ detector @xmath69 ( i. e. linear dependence on the threshold @xmath71 and curvature @xmath10 ) .
once obtained the region of acceptance @xmath1 in the previous subsection , one can calculate the number density of spurious sources and the number density of detections as given by equations ( [ eq : nbstar ] ) and ( [ eq : nstar ] ) .
@xmath72 , \nonumber \\ h&=&\sqrt{\frac{1-\rho^2}{2(1 - 2\rho y_s+y_s^2 ) } } , \label{naceptb}\end{aligned}\ ] ] @xmath73e^{- \frac{(1 - \rho^2)\varphi^2 } { 2{(1 - \rho y_s)}^2}},\end{aligned}\ ] ] @xmath74 then , one can invert the equation for the number of spurious to get @xmath75 that allows to rewrite the equation for the number of detections as @xmath76 .
we will considerer as application the detection of compact sources characterized by a gaussian profile @xmath77 , though the extension to other profiles will be considered in the future .
such a profile is physically and astronomically interesting because represents the convolution of a point source ( dirac @xmath78 distribution ) with a gaussian beam . by introducing a circularly - symmetric filter , @xmath34
, we are going to express the conditions in order to obtain a matched filter for the detection of the source @xmath79 at the origin taking into account the fact that the source is characterized by a single scale @xmath80 .
the following conditions are assumed : @xmath81 , i. e. @xmath82 is an _ unbiased _ estimator of the amplitude of the source ; @xmath83 the variance of @xmath84 has a minimum at the scale @xmath80 , i. e. it is an _ efficient _
estimator @xmath85 this will be called _ matched _ filter as is usual in the literature . for the case of a gaussian profile for the source and a scale - free power spectrum given by @xmath86
, the previous formula leads to the following matched filter @xmath87 for the mf the parameters @xmath88 and the curvature of the source @xmath89 are given by @xmath90 we remark that the linear detector @xmath91 is reduced to @xmath92 i.e. the curvature does not affect the region of acceptance for the mf and the sufficient detector @xmath69 reduces to the plain thresholding detector .
let us modify the filtering scale as @xmath93 to introduce a family of matched - type filters .
in particular , if we consider the white noise case ( @xmath94 ) , this family is given by @xmath95 therefore , this family allows one to filter at scales different from the one of the source .
obviously , for @xmath96 the usual mf is recovered .
the parameters characterizing the background and source are in this case modified in the following way @xmath97 @xmath98 in this case , @xmath99.\ ] ] this allows one to obtain @xmath100,\ \ u\equiv \frac{\varphi } { \sqrt{2(y^2_s + \mu ) } } .\ ] ] hereinafter , we shall consider the case of white noise @xmath101 for the background and a uniform distribution for the point sources in an interval @xmath102 $ ] .
@xmath103 has been chosen to produce a threshold of 2 for the filtered field using the standard mf .
the corresponding thresholds for the mmf can be easily obtained using equation ( [ eq : nualpha ] ) .
we would like to point out that this reflects a situation where we want to detect very weak sources , which were below the @xmath104 level in the original unfiltered map .
the region of acceptance @xmath1 is defined by the sufficient linear detector @xmath69 .
let us consider equation ( [ naceptb ] ) , giving the number density of spurious , then one can invert to get @xmath105 . in fig .
[ fig : teoricas ] we show some results for the cases @xmath36=3 , @xmath106 and @xmath36=4 , @xmath107 .
the two panels on the left correspond to the first case and the ones on the right to the second one .
the top figures show the expected number density of detections versus the @xmath108 parameter of the filter .
it is interesting to remark that such a number decreases with @xmath108 .
in fact , the largest number of detections are obtained for @xmath109 which corresponds approximately to filter at the pixel scale .
the relative ratio to the standard mf defined as @xmath110 is shown in the bottom panels .
for the first case , the ratio takes values up to @xmath111 per cent whereas for the second it goes up to @xmath112 per cent .
these results clearly indicate that , under certain conditions , the standard mf can be improved by simply modifying the scale of the filter .
we remark the importance of the curvature defining the acceptance regions for the mtf . [ cols="^ " , ]
this methodology is optimal for weak sources embedded in white noise .
regarding bright sources , the amplification plays a major role as compared to curvature entering the acceptance region .
therefore , the use of the mf is well justified in this case .
regarding the behavior of the detector with respect to the scale @xmath36 , we found that if @xmath36 is very low the sources become more and more point - like and they are more easily mistaken with the fluctuations of the background .
therefore , in these cases the curvature does not help to distinguish among them and the detector tends again to the thresholding case which favors the use of the mf .
the detection of compact sources on a background is a relevant problem for many fields , in particular for astronomy .
several detection techniques based on the use of linear filters thresholding - based detectors are standard . here
, we have considered an approach to the problem of detector design based on a bayesian generalization of the neyman - pearson rule that includes _ a priori _ information of the source distribution and the number densities of maxima ( background and background plus source ) to define the acceptance region .
our approach based on maxima includes both the amplitude and the curvature .
therefore , the chances of detection do not depend only on the amplification of the sources produced by the filtering but also on the filtered momenta up to the fourth order .
this determines in a strong way the designing of the linear filters that are used to help the detection .
we have applied our technique to a family of matched - type filters ( mtf ) by modifying the scale of the standard matched filter .
we considered the very interesting case of white noise to represent the background .
as an example , we have considered a uniform distribution of sources in the interval @xmath113 in the filtered field , i.e. weak sources .
we have shown that the curvature plays an important role defining the acceptance region and we have proven that the number of detections in the case of a filter with a scale similar to the pixel size beats the number of detections in the case of the standard mf .
this result has been tested with numerical simulations for a uniform distribution and white noise .
the ideas presented in this paper can be generalized : application to other profiles ( e. g. multiquadrics , exponential ) and non - gaussian backgrounds is physically and astronomically interesting .
the extension to include several images ( multi - frequency ) is relevant .
the generalization to two - dimensional data sets ( plane and spherical maps ) and nd images is also very interesting .
finally the application of our method to other fields is without any doubt .
we are currently doing research in some of these topics .
the authors thank enrique martnez - gonzlez and patricio vielva for useful discussions .
rbb thanks the ministerio de ciencia y tecnologa ( mcyt ) and the universidad de cantabria for a ramn y cajal contract .
dh acknowledges support from the european community s human potential programme under contract hprn - ct-2000 - 00124 , cmbnet .
mlc thanks the mcyt for a predoctoral fpi fellowship .
we acknowledge partial support from the spanish mcyt projects esp2001 - 4542-pe and esp2002 - 04141-c03 - 01 and from the eu research training network ` cosmic microwave background in europe for theory and data analysis ' cayn , l. , sanz , j. l. , barreiro , r. b. , mart nez - gonzlez , e. , vielva , p. , toffolatti , l. , silk , j. , diego , j. m. & argeso , f. , `` isotropic wavelets : a powerful tool to extract point sources from cosmic microwave background maps '' , _ mnras _ * 315 * , pp .
757 - 761 , 2000 .
vielva , p. , barreiro , r. b. , hobson , m.p . , martnez - gonzlez , e. , lasenby , a. , sanz , j. l. & toffolatti , l. , `` combining maximum - entropy and the mexican hat wavelet to reconstruct the microwave sky '' , _ mnras _ * 328 * , pp . 1 - 16 , 2001 .
chiang , l.y . ,
jrgensen , h.e . ,
naselsky i.p . ,
naselsky p.d .
, novikov i.d .
& christensen , p.r . , `` an adaptive filter for the construction of the planck compact source catalogue '' , _ mnras _ * 335 * , pp . 1054 - 1060 , 2002 .
herranz , d. , sanz , j. l. , hobson , m. , barreiro , r. b. , diego , j. m. , martnez - gonzlez , e. lasenby , a.n .
, `` filtering techniques for the detection of sunyaev - zeldovich clusters in multifrequency maps '' _ mnras _ * 336 * , pp .
1057 - 1068 , 2002 .
herranz , d. , gallegos , j. , sanz , j. l. & martnez - gonzlez , e. , `` point source detection and extraction from simulated planck time - ordered data using optimal adaptive filters '' , _ mnras _ * 334 * , pp .
533 - 541 , 2002 . | we consider filters for the detection and extraction of compact sources on a background .
we make a one - dimensional treatment ( though a generalization to two or more dimensions is possible ) assuming that the sources have a gaussian profile whereas the background is modeled by an homogeneous and isotropic gaussian random field , characterized by a scale - free power spectrum .
local peak detection is used after filtering .
then , a bayesian generalized neyman - pearson test is used to define the region of acceptance that includes not only the amplification but also the curvature of the sources and the a priori probability distribution function of the sources .
we search for an optimal filter between a family of matched - type filters ( mtf ) modifying the filtering scale such that it gives the maximum number of real detections once fixed the number density of spurious sources .
we have performed numerical simulations to test theoretical ideas . |
the vacuum breakdown in external electric fields by the emission of the lightest charged particle - antiparticle pairs ( schwinger effect ) is an unobserved prediction of quantum electrodynamics ( qed ) @xcite . until recently
, this pair production mechanism seemed to be a mere academic problem owing to the required electric field strength of the order of @xmath0 .
recent efforts have raised the hopes that a direct observation might become feasible at ultra - high intensity laser facilities @xcite .
alternatively , there have been suggestions to study the schwinger effect in graphene @xcite or to quantum simulate it in analogue cold atom systems @xcite . the pair production problem in qed
can not be solved in full glory but requires an approximate treatment .
assuming ultra - high intensity lasers , it is well - justified to neglect quantum fluctuations of the electromagnetic field and to treat it as a classical , external one .
in addition , since the spatio - temporal laser scales are orders of magnitude different from the qed scale , which is set by the electron mass @xmath1 , the focal region is typically modeled by a spatially homogeneous , time - dependent electric field . in this case , quantum kinetic equations are especially well - suited to study the pair production problem numerically since they can be formulated as a coupled system of ordinary first - order differential equations @xcite , which can be solved very efficiently . for more complicated field configurations including spatial inhomogeneities or non - vanishing magnetic fields , more sophisticated approaches such as worldline methods @xcite , the dirac - heisenberg - wigner formalism @xcite or methods from real - time lattice gauge theory @xcite have been employed .
it has been proposed to lower the threshold for pair production by superimposing electric fields with different frequencies ( dynamically - assisted schwinger effect ) @xcite and the resulting momentum spectra have been investigated .
several studies revealed a very rich structure of the momentum distribution , owing to the fact that the pair - production problem depends non - linearly on the applied field profile .
in fact , certain characteristic features of the momentum distribution were interpreted in terms of symmetries of the electric field and resonance conditions @xcite .
however , it is almost impossible to predict the variations in the momentum spectrum upon changing the field parameters _ a priori _ ,
i.e. without actually solving the dynamic equations . a rare exception are solitonic fields , for which particle production can be excluded at one predetermined momentum mode @xcite . from an experimental point of view
, we have to face the challenge that the number of produced electron - positron pairs is exponentially small and therefore hard to discriminate from background noise .
moreover , spectrometers are only sensitive in a certain momentum window and characterized by a finite momentum resolution .
accordingly , it would be most beneficial to generate electric field configurations such that the momentum distribution is peaked in a characteristic momentum window in order to enhance the detection probability .
this situation is reminiscent of the inverse scattering problem in quantum mechanics where asymptotic scattering data is used to reconstruct the scattering potential @xcite . in this manuscript
, we present a method to solve the inverse problem for schwinger pair production , i.e. we determine the field parameters resulting in a predetermined momentum distribution , by combining quantum kinetic theory with optimal control theory .
we employ optimization techniques which have been utilized previously in performing pulse - shape optimization for pair production @xcite , and which have also proven successful in the closely related field of atomic , molecular and optical physics @xcite .
motivated by advances in high harmonic and attosecond pulse generation @xcite , we superimpose a small number of harmonic components and optimize their amplitudes such that a predetermined momentum distribution is well - approximated . in the long run , our results could be used for tailoring superpositions of high harmonics in order to maximize the detection probability in a given momentum window . on the other hand ,
a measured momentum distribution could also serve as a tomograph of the laser field , which is hard to control in an absolute manner at ultra - high intensities .
vacuum electron - positron production in a unidirectional time - dependent electric field @xmath2 can be described in terms of a quasi - particle distribution function @xmath3 , where @xmath4 is the momentum variable with components along and perpendicular to the field direction , and @xmath5 denotes time . given a finite time interval @xmath6 $ ] with @xmath7 ,
@xmath8 can be regarded as the momentum distribution of asymptotic particles @xcite . defining auxiliary functions
@xmath9 and @xmath10 , its dynamics is governed by the coupled system of ordinary first - order differential equations @xcite [ eq : eom ] @xmath11 - 2\omega h \ , \\ & \dot{h}&\ = \ & 2\omega g \ , \end{aligned}\ ] ] with initial values @xmath12 . for @xmath13 and @xmath14
, we have @xmath15 and @xmath16 . in the following ,
we take a superposition of @xmath17 harmonic components on the time interval @xmath6 $ ] [ eq : fields ] @xmath18 \ , \end{aligned}\ ] ] such that @xmath7 and @xmath19 .
we then mean to select the field amplitudes @xmath20 such that a predetermined distribution function @xmath21 is obtained at the final time @xmath22 .
while this search could be achieved by brute force for a very small number of harmonics , this approach becomes practically unfeasible for larger values of @xmath17 .
hence we set up an optimization problem to solve this inverse problem for schwinger pair production by defining the cost functional as the least square deviation of @xmath8 from the objective distribution @xmath21 : @xmath23=\frac{\int{[dq ] \left[f(q , t)-f_0(q)\right]^2}}{\int{[dq ] f_0 ^ 2(q ) } } \ .\ ] ] the global minimum of the cost functional in the space @xmath24 spanned by the field amplitudes is then defined as @xmath25 } \ .\ ] ] for reasons which will become clear later , the cost functional will also be denoted as quality factor in the following .
we emphasize that the distribution function @xmath8 implicitly depends on the field amplitudes @xmath26 via the dynamical system .
the constrained optimization problem can be recast in an unconstrained optimization problem by introducing lagrange multiplier functions @xmath27 @xcite , which need to obey the adjoint equations [ eq : adj ] @xmath28 with final values @xmath29 and @xmath30/\int{[dq]f_0 ^ 2(q)}$ ] .
the stationary condition @xmath31\in\mathbbm{r}^n$ ] is then a necessary condition for being a local extremum , with @xmath32 = e\int{[dq]}\int_{-t}^{t}dt\left[\frac{2p_\parallel}{\omega}\left(\lambda_hg-\lambda_gh\right)\nabla a\right .
\nonumber \\ & \left.+\frac{\epsilon_\perp}{\omega^2}\left(\lambda_g f-\lambda_f g-\lambda_g\right)\left(\nabla e+\frac{2eep_\parallel}{\omega^2}\nabla a\right)\right ] \ . \end{aligned}\ ] ] here , @xmath33 is the gradient with respect to the field amplitudes @xmath34 , and @xmath35 denotes a local minimizer configuration .
moreover , all occurring functions are supposed to be solutions of the dynamical equations and , respectively , and the field gradients @xmath36 are easily determined from . to take full advantage of the gradient information
, we employ a multi - start method along with a local optimization algorithm . to this end
, we generate random trial configurations @xmath37 which are sampled from a probability distribution in field amplitude space .
the corresponding local minimizer configurations @xmath38 are then found iteratively @xmath39 with @xmath40 .
we calculate the search directions @xmath41 according to the broyden - fletcher - goldfarb - shanno ( bfgs ) algorithm , and viable step sizes @xmath42 are found via an inexact line search fulfilling the strong wolfe conditions . for further algorithmic details we refer to , e. g. reference @xcite . as a word of caution
we emphasize that it is only guaranteed for infinite sample sizes that the global minimum @xmath43 $ ] , where @xmath44 denotes the global minimizer configuration , is among the detected stationary points @xcite .
moreover , it is also not guaranteed that an exact solution of the inverse problem , which is characterized by a vanishing cost functional @xmath45 , actually exists . in general , we will rather find @xmath46 for certain momenta , where @xmath47 denotes the momentum distribution corresponding to the global minimum @xmath48 .
this is in fact a sign that the inverse problem for schwinger pair production is in general ill - posed @xcite .
a configuration that approaches the exact solution to the inverse problem is characterized by @xmath49 , and we will study its dependence on the number of harmonic components @xmath17 in the following . 3d - plot of the quality factor landscape for a superposition of @xmath50 harmonics . for display reasons we show @xmath51 $ ] , so that local maxima in the plot actually correspond to local minima of @xmath52 $ ] . in green
we display local minimum configurations @xmath35 , amongst them the global minimum configuration @xmath53 with @xmath54 .
we also display some typical optimization trajectories which lead from randomly chosen initial configurations ( _ red points _ ) to the local extrema ( _ green points _ ) . ]
for simplicity , we study the inverse pair production problem in a @xmath55-dimensional system but emphasize that the general procedure is basically not altered in higher dimensions .
to be specific , we predefine a gaussian model objective distribution @xmath56 and we take as parameters @xmath57 , @xmath58 and @xmath59 .
we could specify any functional form in principle , however , we will restrict ourselves to this simple and instructive example . specifically , we consider the harmonic superposition in the time interval @xmath60 .
we first consider the superposition of @xmath50 harmonics and calculate the landscape @xmath52 $ ] by brute force , cf .
[ fig1:potential ] .
we note that the distribution function @xmath61 is computed for an array of field amplitudes @xmath62 with @xmath63 , resulting in @xmath64 simulated data points .
we therefore conclude that the brute force approach becomes practically unfeasible for large values of @xmath17 . and requiring sufficient momentum resolution , it takes a few minutes on a standard desktop computer to calculate the momentum distribution @xmath61 according to .
consequently , the computation time on the chosen array of field amplitudes is of the order of @xmath65 cpu hours .
increasing the number of harmonics @xmath66 and keeping the resolution @xmath67 unaltered , the computation time grows exponentially @xmath68 , indicating that the brute force approach becomes practically unfeasible for large values of @xmath17 . ]
the landscape shows a diamond - shaped region with quality factors @xmath69=\mathcal{o}(1)$ ] which is rather flat , whereas it increases exponentially outside . on the other hand
, we also find a small number of distinct local minima which are indicated in green in fig .
[ fig1:potential ] , amongst them the global minimum @xmath70 with @xmath71 .
we emphasize that there is only an approximate solution to the inverse problem as the quality factor is non - vanishing .
quality factor @xmath48 for the global minimizer configuration @xmath44 as a function of the number of harmonics @xmath17 ( _ blue dots _ ) .
the exponential fit @xmath72 describes the trend reasonably well ( _ purple line _ ) . ] to illustrate the functionality of the employed optimization method , we first perform the numerical optimization for @xmath50 and we display some of the calculated trajectories in fig .
[ fig1:potential ] . depending on the randomly chosen initial configuration @xmath37
, the various trajectories converge to the diverse stationary points in the landscape @xmath73 $ ] . for @xmath50 , the average number of required iteration steps for sufficient convergence based on @xmath74 randomly chosen initial configuration
is @xmath75 . in order to converge towards the objective momentum distribution @xmath21 ,
i.e. to further decrease the quality factor @xmath48 , we need to increase the number of harmonic components @xmath17 .
most notably , we then observe quick and substantial improvement of @xmath48 .
this assertion is quantified in fig .
[ fig2:quality ] , where we display the quality factor @xmath48 as a function of @xmath17 .
moreover , we display the underlying global minimum distributions @xmath76 for different values of @xmath17 along with the corresponding electric field configurations @xmath77 in fig .
[ fig3:optimization_dist ] . based on our result
, we note that the average number of required iteration steps grows like @xmath78 , indicating that the simulations become more expensive for increasing @xmath17 .
we emphasize , however , that this linear growth of the optimization approach outweighs by far the exponential growth of the brute force approach .
_ left : _ logarithmic plot of the numerically determined global minimum distribution @xmath76 for an increasing number of harmonics @xmath79 .
_ right : _ corresponding electric field configuration @xmath80 .
] for @xmath81 we find an optimal distribution function @xmath82 which exhibits a broad central peak accompanied by two narrow side peaks .
notably , the central peak is located around @xmath83 and is thus substantially shifted from @xmath58 . including one additional harmonic such that @xmath50 , the optimal distribution function @xmath84 becomes peaked around @xmath85 but is still somewhat broader and smaller than @xmath21 .
notably , we still observe two undesired peaks around @xmath86 and @xmath87 , respectively . further increasing the number of harmonics @xmath66
, the optimal distribution @xmath88 approximates the objective distribution @xmath21 better and better . here , we emphasize that the global minimum configuration @xmath88 does not necessarily arise from the global minimum configurations @xmath89 , meaning that @xmath90 does not necessarily lie in the basin of attraction of @xmath91 .
for instance , given the @xmath50 global minimum configuration @xmath92 with @xmath93=0.1592 $ ] ( cf .
fig.[fig1:potential ] ) and taking these field parameters as the initial point for the @xmath94 optimization problem , we converge towards the stationary point @xmath95 with @xmath96=0.1392 $ ] .
on the other hand , starting from the second - best @xmath50 stationary configuration @xmath97 with @xmath98=0.1604 $ ] , we actually converge towards the @xmath94 global minimum configuration @xmath99 with @xmath100=0.1399\cdot 10^{-1}$ ] .
consequently , the shape of the electric field can change quite substantially for different values of @xmath17 ( cf .
[ fig3:optimization_dist ] ) .
the presented results indicate that a rather small number of harmonic components @xmath17 suffices to well - approximate specific predetermined features of the momentum distribution .
for instance , the distribution around @xmath58 is already extremely well - described for @xmath101 whereas there are still comparatively large deviations at other momentum values present . on the other hand ,
the number of harmonics @xmath17 needs to be further increased in order to approach the exact solution of the inverse problem for schwinger pair production .
quantitatively , we find that the quality factor drops from @xmath102 to @xmath103 . based on our results ,
the functional dependence of the quality factor is reasonably well - described by an exponential function @xmath72 .
one may therefore speculate that there is an exact solution of the inverse problem for the chosen objective distribution @xmath21 .
it is , however , not clear whether this solution then exists for finite @xmath17 or only in the limit @xmath104 .
we already mentioned that the employed multi - start method is guaranteed to converge to the global minimum only for infinite sample sizes .
accordingly , fig .
[ fig2:quality ] is to be considered as an upper limit whereas there are , in principle , even better configurations conceivable . in the current study , we have used @xmath74 random initial configurations @xmath37 for all considered values of @xmath17 .
given this ensemble size , all the numerically determined global minimum distribution @xmath88 were encountered several times .
we are therefore confident that the identified configurations actually correspond to the global minimum @xmath48 or lie , at least , in its close neighborhood .
we demonstrated that quantum kinetic theory along with optimal control theory can be utilized to solve the inverse problem for schwinger pair production . based on the instructive example of a gaussian objective distribution @xmath21
, we found that it suffices to superimpose a comparatively small number of harmonics @xmath101 in order to well - approximate specific predetermined features .
these findings could substantially facilitate the observation of schwinger pair production by providing suggestions for tailored field configurations . in the long run
, the presented method could also serve as a tomograph for applied laser pulses by reconstructing the applied field after measuring the asymptotic momentum distribution .
one has to be aware , however , that this procedure might be non - unique due to the ill - posedness of the inverse problem . on the other hand ,
it is not clear at this point whether an exact solution of the inverse problem with @xmath45 actually exists and , if so , a finite number of harmonic components @xmath17 suffices .
this issue is beyond the scope of the current investigation but should be further studied in the future .
the used optimization method is especially well - suited for parameter spaces which are not too high - dimensional .
in fact , we found that the average number of required iterations grows like @xmath78 , indicating that the optimization algorithm becomes less efficient for increasing @xmath17 .
moreover , the ensemble size @xmath105 of random initial configuration @xmath37 needs to be enlarged upon further increasing the number of harmonics @xmath17 , indicating that the multi - start approach becomes less efficient as well .
accordingly , we should also envisage to apply global optimization strategies based on metaheuristic algorithms in the future in order to study the inverse problem for more intricate objective distributions @xmath21 , to investigate the possibility of an exact solution of the inverse problem with @xmath106 , or to perform elaborate pulse shaping investigations .
we like to thank f. fillion - gourdeau for fruitful discussions and valuable comments , as well as r. alkofer , c. kohlfrst , m. mitter and g. von winckel for collaboration on related work . the research leading to these results has received funding from the european research council under the european union s seventh framework programme ( fp7/2007 - 2013)/ erc grant agreement 339220 . | the production of electron - positron pairs in time - dependent electric fields ( schwinger mechanism ) depends non - linearly on the applied field profile .
accordingly , the resulting momentum spectrum is extremely sensitive to small variations of the field parameters . owing to this non - linear dependence
it is so far unpredictable how to choose a field configuration such that a predetermined momentum distribution is generated .
we show that quantum kinetic theory along with optimal control theory can be used to approximately solve this inverse problem for schwinger pair production .
we exemplify this by studying the superposition of a small number of harmonic components resulting in predetermined signatures in the asymptotic momentum spectrum . in the long run
, our results could facilitate the observation of this yet unobserved pair production mechanism in quantum electrodynamics by providing suggestions for tailored field configurations .
dynamically assisted schwinger mechanism , inverse problem , optimal control theory |
the combination of relativity and quantum theory contributes to a fruitful area of research , which in particular includes relativistic quantum information theory and technology @xcite . in information processing ,
particle detectors or observers moving along relativistic trajectories , can send and receive signals which are relevant to internal degrees of freedom of detectors .
the role of detectors is to locate and label events in curved spacetimes .
recently , the use of non - inertial motions to perform quantum computing has been investigated by means of accelerated optical cavities @xcite or atoms in relativistic regimes @xcite .
the relativistic technology contributes to ultra - fast quantum computation which can overcome some obstacles from decoherence . in this letter
, we employ this idea to study the relativistic effects of accelerated qubits on the precision of parameter quantum estimation .
the selection of a physically implementable model is a key to our scenario . for many years , the unruh - dewitt ( @xmath0 ) model @xcite
have been extensively applied to the study of aspects of quantum field theory in curved spacetimes .
this model is referred to as a particle detector , like a two - level atom , which is linearly coupled to a quantum scalar field with simple monopole interactions . in comparison of @xmath1 ,
the model characterizes adequately light - matter interactions @xcite in some specific settings .
one great success of the model lies in the demonstration of unruh effect .
it was shown that a uniformly accelerated detector behaves as an inertial detector in a thermal bath , with a characteristic temperature , i.e. , unruh temperature , proportional to its proper acceleration .
this open quantum system approach is applied to quantum gravity on some spacetimes @xcite .
the phenomenological model can be simulated in trapped ion systems or superconducting circuit experiments @xcite .
our scheme is to explore quantum metrology by the @xmath0 model . as we know
, quantum fisher information ( @xmath2 ) is thought of as one efficient measure for parameter quantum estimation @xcite .
recently , @xmath2 has been widely studied in various fields involving the investigation of uncertainty relations @xcite , the estimation of quantum speedup limit time @xcite , the characterization of quantum phase transition @xcite , and the detection of entanglement @xcite . until now
, some interesting works have considered the unruh effect on quantum metrology in the non - inertial frame @xcite .
the authors have found that fork states can achieve the maximal @xmath2 in the presence of scalar fields in a @xmath3-dimensional minkowski spacetime @xcite .
dirac fields in non - inertial frames were investigated in @xcite .
additionally , relativistic effects were considered as one resource for performing quantum metrology @xcite.the fisher information for population measurement has been evaluated by the use of the open quantum system method @xcite . when a measuring device is in a relativistic motion , observation occurs in a non - inertial frame .
our aim is to determine some common features of relativistic quantum metrology .
we try to find some controllable conditions that contribute to the enhancement of relativistic quantum metrology . in this paper , the dynamics of @xmath2 with respect to phase
is investigated under the condition that a uniformly accelerated atom is weakly coupled to a massless quantum scalar field in the presence of reflecting boundaries .
the paper is organized as follows .
first , the @xmath0 model and one calculation method of @xmath2 are presented .
we choose a pure state with a phase parameter as an initial state , and describe the evolution of the accelerated atom by means of open quantum system approach in the non - inertial frame .
second , in accordance with motions of a measuring device , the calculations of @xmath2 as a function of the proper time in the non - inertial frame are demonstrated in detail .
the effects of reflecting boundaries on phase quantum estimation are considered .
finally , a conclusion depicts key findings about relativistic parameter quantum estimation and some possible physical realizations .
what we are interested in is the dynamical behavior of parameter quantum estimation related to a uniformly accelerated particle detector , such as a two - level atom . in this scenario ,
we present the unruh - dewitt detector model to characterize the atom coupled to a massless quantum scalar field . in the weak coupling limit , atomic transitions with no exchange of angular momentum is reasonably considered @xcite . without loss of generality , the hamiltonian of the model can be expressed as @xmath4 the system hamiltonian of the detector is written as @xmath5 .
@xmath6 represents the energy gap between the ground state and excited state .
@xmath7 is the pauli matrix for the detector .
thus , the quantum states of the detector can be described in terms of a two - dimension hilbert space .
the form is written as a @xmath8 density matrix which is a completely positive and hermitian operator .
@xmath9 is the standard hamiltonian of massless scalar fields , which is determined by the massless klein - gordon equation .
the details of the field need not be specified here .
the interaction hamiltonian @xmath10 between the detector and scalar field is given by @xmath11,\ ] ] where @xmath12 is the strength of the interaction .
the weak coupling condition of @xmath13 is assumed so that we can obtain a completely positive dynamical map for the detector .
the field operator @xmath14 is dependent on the worldline of the accelerated detector . for the description of the dynamics of the system ,
the composed system is initially prepared in a factorized state , with the detector at rest and the scalar field in the vacuum state @xmath15 .
the two sets of spacetime coordinates are involved in this model .
one denotes @xmath16 in a non - inertial frame that is moving together with the atom , and the other one is @xmath17 in the inertial frame .
the relation between two sets of spacetime coordinates holds that @xmath18\end{aligned}\ ] ] where the trajectory of constant rindler position @xmath19 is used to describe the observation of uniformly accelerated atom .
the worldline is just shown by eq(3 ) .
for simplicity , it is assumed that the natural unit @xmath20 and that the atom is in the position @xmath21 at the proper time @xmath22 . under the circumstance of weak couplings
, we use the perturbation method to obtain the dynamics of the accelerated detector .
it is a natural choice to establish a dynamical map in the non - inertial frame . in the interaction picture ,
the dynamics of the reduced density for the detector after tracing over the field degrees of freedom is expressed as @xmath23 \ ; \right ] \ ; \right\},\ ] ] where @xmath24 is the hamiltonian of the total system in the interaction picture .
@xmath25 represents the state of the detector in the non - inertial frame , and @xmath26 denotes the vacuum field in the non - inertial frame . by the approach of @xcite , the analytical solution to the dynamical equation is equivalent to the quantum master equation which has the lindblad form of @xmath27+l[\rho(\tau)].\ ] ] the key to the quantum master equation is the dissipation term given as @xmath28 = \frac 12 \sum_{k=\pm , z } \gamma_k \ ; \left ( 2\sigma_k \rho \sigma^{\dag}_{k}- \sigma^{\dag}_{k}\sigma_k \rho-\rho\sigma^{\dag}_{k}\sigma_k \right),\ ] ] where the operators @xmath29 and the decay rates are also obtained by @xmath30 @xmath31 is determined by the fourier transformation of the field vacuum correlation function @xmath32 , @xmath33.\ ] ] here the correlation function between two different positions for the vacuum field can be calculated as @xmath34 & = \langle 0_f \left| \phi\big(x(\tau ) \big ) \phi \big(x^{\prime}(\tau^{\prime } ) \big ) \right|0_f\rangle
\nonumber \\ & = -\frac { 1}{4\pi^2}\ ; \dfrac { 1}{(ct - ct^{\prime}-i\epsilon)^2-(x - x')^2}\end{aligned}\ ] ] with the proper @xmath35 prescription . in this case
, the initial position of the detector is chosen to be @xmath36 . in the quantum master equation ,
the effective hamiltonian of the detector is expressed as @xmath37 where the effective energy gap @xmath38,\ ] ] where the parameter @xmath39 and @xmath40 denotes the principle value . to mathematically express the density matrix @xmath25 , we equivalently replace the density matrix by the bloch vector , i.e. , @xmath41 .
the components of the bloch vector for the state are obtained by @xmath42 where the unitary matrix @xmath43 for the dynamical map and inhomogeneity vector @xmath44 take the form @xmath45 by means of the above equantion , we can conveniently attain the quantum evolution in the non - inertial referrence frame .
the estimation information with respect to the phase parameter is embodied in the dynamical map .
our aim is to evaluate relativistic effects on dynamical behavior of parameter quantum estimation for an accelerated atom . in experiments ,
the observation of a phase difference between two energy states plays a great role in quantum information technology @xcite . in this scenario , we consider phase estimation for atoms in relativistic motions . for a measuring device in the same motion as a uniformly accelerated atom , the observation of quantum estimation should be presented in the non - inertial frame .
different from quantum estimation in the non - inertial frame , the results are established in the inertial frame when the measuring device is in a motion with uniform velocity .
an selected initial state for the atom at @xmath22 involves a phase parameter and has the form of @xmath46 represent a phase difference between the excited state @xmath47 and ground state @xmath48 .
@xmath49 describes the relative occurrence possibility of the energy states .
it is easily found that the phase quantum estimation is dependent on the evolution of the parameterized states . to evaluate the true value of a phase parameter @xmath50 as precisely as possible
, we need an unbiased estimator @xmath51 whose expectation holds that @xmath52 .
the precision of quantum estimation satisfies the quantum cramr - rao ( @xmath53 ) inequality @xcite , i.e. , @xmath54 , where @xmath55 is the number of independent measurements and @xmath56 is the @xmath2 with respect to phase parameter @xmath50 .
the @xmath53 inequality shows that the high precision of phase estimation is attained when the value of @xmath2 is large .
the standard calculation procedure @xcite starts by the construction of a symmetric logarithmic derivative @xmath57 which is defined as @xmath58 the @xmath2 which is not dependent on the choice of @xmath57 is generally written as @xmath59.\ ] ] the evolution state @xmath60 of the atom in the non - inertial frame is written as @xmath61 for the density matrix of @xmath62 , the elements of @xmath57 are calculated as @xmath63,\ ] ] where the eigenvalues and eigenstates of the density matrix are given by @xmath64 and @xmath65 . as a result , we accomplish the analytical expression of @xmath2 as @xmath66 when the parameter @xmath67 , the value of @xmath2 reaches a maximal one , i.e. , @xmath68 .
it is seen that the dynamical behavior of @xmath2 are determined by the decay rates and proper time . to discover some controllable conditions that contribute to the @xmath2
, we reasonably take into account boundary effects of the scalar field .
boundaries can modify the fluctuations of quantum fields .
it leads to a lot of novel effects , such as casimir effects@xcite , entanglement generation @xcite and the modification for the radiative properties of uniformly accelerated atoms @xcite . in our scheme , two perfectly reflecting plane boundaries that are perpendicular to each other locate at @xmath69 and @xmath70 in the space , respectively . from fig .
( 1 ) , it is shown that @xmath71 denotes the distance between the trajectory of the accelerated atom and the cross of the two boundaries .
the angle @xmath72 represents the relative position of the atom between the two boundaries . in this simple case ,
the boundary conditions holds that @xmath73 and @xmath74 . according to the method of images , the correlation function for the vacuum field
is expressed as @xmath75 } - \dfrac 1{\sinh^2\left [ \frac { a(\tau-\tau^{\prime}}{2 } -i\epsilon \right]-a^2r^2\cos^2\alpha } \right .
\nonumber \\ & \left .
+ \dfrac 1{\sinh^2\left [ \frac { a(\tau-\tau^{\prime}}{2 } -i\epsilon \right]-a^2r^2 } - \dfrac 1{\sinh^2\left [ \frac { a(\tau-\tau^{\prime}}{2 } -i\epsilon \right]-a^2r^2\sin^2\alpha } \right\}.\end{aligned}\ ] ] the fourier transformation of the correlation function @xmath31 is also calculated as @xmath76,\ ] ] where the special function @xmath77 is defined as @xmath78}{2r\sqrt{1+a^2r^2}\omega}$ ] .
the unruh temperature @xmath79 is involved in the above equation .
and in the limit of @xmath80 , the parameter @xmath81 $ ] where the function @xmath82 .
consequently , the analytical expression of the maximal @xmath2 in the non - inertial frame is given by @xmath83 \right .
\nonumber \\
& - \left .
\frac { \lambda^2 a \tau}{\pi^2}\left [ 1-f_2(r\cos\alpha)-f_2(r\sin\alpha)+f_2(r ) \right ] \right\}.\end{aligned}\ ] ] to clearly demonstrate the effects of the relativistic motions and boundaries on the dynamics of the @xmath2 , we carry out the numerical calculation of @xmath2 which are plotted in figs .
( 2)-(4 ) provide the observation results of the maximal @xmath2 in the non - inertial frame .
the observation is achieved by a measuring device in the same motion as the acceleration atom .
it is clearly seen that the values of the maximal @xmath2 are always monotonically decreased with the proper time @xmath84 in fig .
besides it , the values in the low acceleration condition are larger than those in the high acceleration condition .
this means that the relativistic motion of the measuring device can suppress the precision of parameter quantum estimation for the accelerated atom .
the boundary effects on the @xmath2 are shown in fig .
( 3 ) . in the region of @xmath85 ,
the values of @xmath2 can keep high in the short time interval .
when the trajectory of the accelerated atom is far away from the boundaries , i.e. , @xmath86 , the boundary effects are trivial and even neglected .
we also notice that the apparent oscillation of the @xmath2 will happen if the accelerated atom is close to the boundaries .
this phenomena is induced by the relativistic motion . from fig .
( 4 ) , we also see that the values of @xmath2 are symmetrical to the boundaries . in the case of @xmath87 ,
the value of @xmath2 arrives at the minimal one .
meanwhile , the high values of @xmath2 can be reached under the condition of @xmath88 or @xmath89 .
the fact is that the precision of phase quantum estimation is enhanced when the accelerated atom is near to the boundaries .
by means of the @xmath0 model , we study the dynamical behavior of phase quantum estimation for a uniformly accelerated atom weakly coupled to a massless quantum scalar field .
the dynamics of quantum states of the atom can be obtained in the non - inertial frame .
the calculation of the @xmath2 provides an efficient way to estimate the measuring precision .
when a measuring device is accelerated in the same motion as the atom , the monotonic decrease of the @xmath2 is manifest in the high acceleration limit .
this means that the relativistic motion of the measuring device restrains the precision of phase quantum estimation .
it is found out that the obvious oscillation of @xmath2 occur under the condition that the trajectory of the atom is close to the boundaries .
this provides us a possible way to enhance the precision of phase quantum estimation in the relativistic situation .
let us examine the magnitudes involved in some possible experiments .
the natural scale of units is fixed by fixing units for the energy gap @xmath6 of the atom , namely @xmath90
. for atomic gaps of @xmath91 , one natural unit of acceleration corresponds to @xmath92 ( @xmath93 is the earth surface gravitation acceleration ) @xcite .
the acceleration required in our scenario can be reduced by the use of qubits with smaller gaps .
for example , for special qubits defined by nuclear spins , energy gaps can be limited to the order of @xmath94 , which can reduce the acceleration to almost @xmath95 @xcite .
the scale of acceleration can be realized in @xmath96 setups .
in addition , qubits with stark shifted atomic levels or zenner - induced transition nearly have energy gaps of the order of @xmath97 . for these qubits ,
the acceleration can reach @xmath98 @xcite .
those accelerations are indeed experimentally achieved for short time intervals .
current technology of ion trapping and superconducting circuits allows for experiments where relativistic effects can be observed .
the work was supported by the @xmath99 talents project of jiangsu province , by jiangsu government scholarship for study abroad , and by nserc of canada and templeton foundation , grant no .
we would like to thank bill unruh for the discussions on related work .
a. peres and d. r. terno , rev .
phys . * 76 * , 93(2004 ) .
n. friis , m. huber , i. fuentes , d. e. bruschi , phys .
d * 86 * , 105003(2012 ) .
d. e. bruschi , a. dragan , a. r. lee , i. fuentes , j. louko , phys .
lett . * 111 * , 090504(2013 ) .
e. martn - martnez , d. aasen , a. kempf , phys .
lett . * 110 * , 160501(2013 ) .
e. e. martn - martnez and c. sutherland , phys .
b * 739 * , 74(2014 ) .
a. g. s. landulfo and g. e. a. matsas , phys .
a * 80 * , 032315 ( 2009 ) .
l. c. cleri , a. g. s. landulfo , r. m. serra , and g. e. a. matsas , phys .
a * 81 * , 062130 ( 2010 ) .
j. wang , j. jing , and h. fan , phys .
d * 90 * , 025032(2014 ) .
grant salton , robert b mann , and nicolas c menicucci , new j. phys . *
17 * , 035001(2015 ) d. su and t. c. ralph , phys .
d 90 084022 ( 2014 ) w. g. unruh , phys .
d * 14 * , 870(1976 ) .
l. c. b. crispino , a. higuchi , and g. e. a. matsas , rev . mod .
phys . * 80 * , 787(2008 ) . b. s. dewitt , phys . rev . *
160 * , 1113(1967 ) .
e. g. brown , e. martn - martnez , n. c. menicucci , r. b. mann , phys
d * 87*,084062(2013 ) .
m. o. scully , m. s. zubairy , quantum optics , cambridge university press(1997 ) .
h. yu , phys .
106 , 061101(2011 ) .
j. feng , y. z. zhang , m. d. gould , and h. fan , phys .
b * 743 * , 198(2015 ) p. d. nation , m. p. blencowe , a. j. rimberg , and e. buks , phys .
lett . * 103 * , 087004(2009 ) . c. m. wilson , g. johansson , a. pourkabirian , m. simoen , j. r. johansson , t. duty , f. nori , and p. delsing , nature(london ) * 479 * , 376(2011 ) .
s. l. braunstein and c. m. caves , phys .
lett . * 72 * , 3439(1994 ) .
d. petz , linear algebra appl .
* 244 * , 81(1996 ) .
s. luo , lett .
53 * , 243(2000 ) .
p. gibilisco , d. imparato , and t. isola , j. math . phys . * 48 * , 072109(2007 )
. m. m. taddei , b. m. escher , l. davidovich , and r. l. de matos filho , phys .
lett . * 110 * , 050402(2013 ) . c. invernizzi , m. korbman , l. c. venuti , and m. g. a. paris , phys .
a * 78 * , 042106(2008 ) .
s. boixo and a. monras , phys .
100 , 100503(2008 ) .
m. aspachs , g. adesso , and i. fuentes , phys .
lett . * 105 * , 151301(2010 ) . y. yao , x. xiao , l. ge , x. wang , and c. sun , phys .
rev . a * 89 * , 042336(2014 ) .
d. hosler and p. kok , phys .
a * 88 * , 052112(2013 ) .
m. ahmadi , d. e. bruschi , n. friis , c. sabn , g. adesso , and i. fuentes , sci . rep . * 4 * , 4996(2014 ) .
z. tian , j. wang , h. fan and j. jing , sci .
* 5 * , 7946(2015 ) .
f. benatti , r. floreanini , and m. piani , phys .
lett . * 91 * , 070402 ( 2003 ) .
f. benatti and r. floreanini , phys .
a * 70 * , 012112 ( 2004 ) .
v. giovanetti , s. lloyd , and l. maccone , science * 306 * , 1330(2004 ) . c. w. helstrom ,
quantum detection and estimation theory , academic , new york(1976 ) .
a. s. holevo , probabilistic and statistical aspects of quantum theory , north - holland , amsterdam(1982 ) .
h. b. g. casimir , proc .
akad . wet . *
51 * , 793(1948 ) .
j. zhang , h. yu , phys . rev . a * 75 * , 012101(2007 ) .
h. yu and s. lu , phys .
d * 72 * , 064022 ( 2005 ) .
e. arimondo , m. inguscio , and p. violino , rev .
* 49 * , 31(1977 ) .
( atlas collaboration ) , phys .
lett . * 105 * , 161801 ( 2010 ) .
the effects of the acceleration on the dynamics of @xmath2 as a function of the proper time @xmath84 are shown when the measuring device is in the same relativistic motion as the atom .
the observation results are obtained under the conditions that @xmath100 in the non - inertial frame . the boundary effects on the dynamics of @xmath2 as a function of the proper time @xmath84 are shown when the measuring device is in the same relativistic motion as the atom .
the observation results are obtained under the conditions that @xmath101 in the non - inertial frame .
the effects of the relative position of the atom between the boundaries are shown at some proper time when the measuring device is in the same relativistic motion as the atom .
the observation results are obtained under the conditions that @xmath102 in the non - inertial frame . | relativistic effects on the precision of quantum metrology for particle detectors , such as two - level atoms are studied .
the quantum fisher information is used to estimate the phase sensitivity of atoms in non - inertial motions or in gravitational fields .
the unruh - dewitt model is applicable to the investigation of the dynamics of a uniformly accelerated atom weakly coupled to a massless scalar vacuum field .
when a measuring device is in the same relativistic motion as the atom , the dynamical behavior of quantum fisher information as a function of rindler proper time are obtained .
it is found out that monotonic decrease in phase sensitivity is characteristic of dynamics of relativistic quantum estimation . to improve relativistic quantum metrology ,
we reasonably take into account two reflecting plane boundaries perpendicular to each other .
keywords : phase quantum estimation , relativistic effects , unruh - dewitt model , quantum fisher information pacs : 03.67.-a , 03.65.yz , 42.50.ex |
symmetricom test set has become a useful instrument for phase noise and frequency stability measurements that allows one to compare a signal from a device under test ( dut ) against a reference signal and as a result measure the relative phase noise and frequency stability between the two signals @xcite .
the manufacturer supplies a few models but currently the maximum input bandwidth is 400 mhz ( model 5125a ) while the highest performance unit ( model 5120 - 01 ) , with the lowest phase noise measurement noise floor , only has a measurement bandwidth of 30 mhz .
so how would one measure the additive or residual phase noise of a device under test which has an operating frequency much higher than 30 mhz or even 400 mhz ?
one method to achieve this is to take a beat note between two oscillators , and , by engineering a relatively low noise reference through frequency division of a signal derived from one of the oscillators , one can make a measurement on oscillators operating at x - band frequencies @xcite . if the oscillators have similar phase noise 3 db can be subtracted for their individual performance .
however this method is limited to the case where the beat note between the signals from two oscillators being compared falls within the measurement bandwidth of the test set being used .
this paper outlines a method to use the symmetricom test set outside its normal area of application ( oscillator noise measurements ) .
we show that it can be used for measuring phase fluctuations in passive and active microwave components ( circulators , amplifiers , voltage controlled phase shifters etc ) and at frequencies well outside its normal operational bandwidth .
this method is quite different to that normally used for phase noise measurements where a homodyne technique uses baseband measurements and calibration of the mixer is required to calculate the phase noise from voltage noise measurements .
it allows one to simply measure the phase noise and/or frequency stability of practically any device at any frequency , provided the frequency is sufficiently stable .
the dut could be an active device where only phase noise is of interest or alternatively two oscillators where both phase noise and frequency stability are measured .
the essential elements of the measurement technique are shown in fig . 1 .
here the dut is a microwave amplifier operating at frequency @xmath0 .
the microwave signal from a stable source is divided into two arms .
one arm passes through the dut and picks up the intrinsic phase fluctuations due to the amplifier .
an attenuator ( att1 ) is adjusted to give the desired input power and a second attenuator ( att2 ) is used to provide the suitable lo drive level for the first mixing stage . at this point
an auxiliary source is used to provide a signal at a frequency @xmath1 that is within the range of the test set being used .
the important feature of this technique is to impose the dut phase noise onto the low frequency @xmath1 auxiliary signal , which is done via the use of two mixing stages .
if necessary , attenuators ( att3 and att4 ) are used to adjust the lo drive level and the rf power level to the second mixing stage .
alternatively an isolator could be used instead of attenuator att4 .
a band pass filter ( bpf ) is used to reject one of the sidebands , either @xmath2 or @xmath3 , at the output of the first mixing stage .
( an image rejection or single side band mixer could also be used . )
the second mixing stage recovers the frequency ( @xmath1 ) of the auxiliary signal , after low pass filtering ( lpf ) but it carries all the phase fluctuations of the device under test . the signal is then amplified and input to the symmetricom test set .
the reference to the test set is taken directly from the auxiliary oscillator .
this combination of two mixing stages with the test set is integral to this technique and allows one to essentially use the test set to measure the phase noise of any device operating at any frequency within the range of the components used .
considering the configuration of fig .
1(a ) , we model the input signal to the second mixing stage , after filtering , as proportional to , @xmath4,\ ] ] where the upper sideband has been rejected , and where @xmath5 is the phase spectrum of the noise added by the dut to the @xmath0 signal before being mixed with the @xmath1 auxiliary signal .
the parameter @xmath6 represents the initial phase of the signal at lo1 .
after the second mixing stage the resulting signal frequency dependence can be modeled by the product of the latter with the original @xmath0 microwave pump signal as , @xmath7 cos[2 \pi f_0 t + \varphi_{lo2 } ] = \nonumber \\ & \frac{1}{2 } cos [ 2 \pi f_a t + \delta \varphi(t)+ \phi ] , \label{eqn1}\end{aligned}\ ] ] where @xmath8 is a phase constant and the high frequency @xmath9 mixing product has been rejected by the low pass filter .
the @xmath9 signal is well suppressed by the mixer itself if @xmath0 is in the microwave range . here
@xmath10 represents the initial phase of the signal at lo2 .
this and the initial phase at lo1 are a fixed phase with respect to the initial phase of the auxiliary signal at the if port of the first mixing stage .
therefore the last term in eq .
( [ eqn1 ] ) is a constant phase factor and as a result the @xmath1 signal input to the test set contains the phase noise of the dut .
the correlated phase fluctuations of the pump microwave signal ( @xmath0 ) are rejected at the level of the second mixer .
phase fluctuations of the auxiliary oscillator are rejected by the test set because it measures fluctuations of the phase difference between two input signals ( it is essentially a `` digital '' phase bridge ) .
the measurement system white noise floor is then determined primarily by the final low frequency rf amplifier used to provide sufficient signal power to the input of the test set .
the two mixers down converting the microwave signal , that passes through the dut , to the frequency of the auxiliary signal , and the rejection of one of the sidebands , are the main elements in this technique .
all variations on this design contain this important aspect .
ideas similar to this have been developed before under the name of `` transposed gain '' where a signal is down converted from a much higher microwave frequency to a low frequency where the amplifier , used as the important gain stage , has much lower intrinsic noise and after which the signal is up converted again to the higher microwave frequency @xcite .
this technique is quite different as it does not actually use a transpose gain but transfers the noise of the dut to a much lower frequency signal that is accessible to the phase noise test set used to measure the device s phase noise and/or frequency stability .
it is worth noting that the dut could have , alternately , been located on the other branch of the power splitter before the mixer input lo2 but we would get the same result as eq .
( [ eqn1 ] ) .
caution needs to be taken to ensure that the lo power on the mixers is maintained at their optimum drive level and at the same level for all measurements .
one might be measuring the dut as a function of input power , as in fig .
1 , which will change the power at lo1 .
the mixer s contribution to the measurement system noise floor depends on its input power .
it was observed that when the mixer power was allowed to drop well below optimum drive levels the measurement system noise floor was elevated and limited the measurement .
however as indicated in fig .
1(b ) the band - pass filter could also be replaced with a ( mechanical ) phase shifter but it will be shown that this does not result in the correct measurement of the phase noise of the dut .
we model this configuration with a signal at the first mixing stage as proportional to , @xmath11,\ ] ] where both sidebands are present , the term @xmath5 represents the flicker phase fluctuations of the dut .
( the treatment of the white phase noise of the dut is given below ) .
the term @xmath12 represents the added phase from the variable ( mechanical ) phase shifter .
when this is mixed at the second mixing stage the resulting signal frequency dependence can be modeled by the product of the latter with the original @xmath0 microwave pump signal as , @xmath13 \times \nonumber \\ & cos[2 \pi f_0 t + \varphi_{lo2 } ] = \nonumber \\ % & \frac{1}{2}cos[2 \pi ( -f_a ) t + \delta \varphi(t)+\varphi_{lo1}+\varphi_p -\varphi_{lo2 } ] + \nonumber\\ % & \frac{1}{2}cos[2 \pi ( f_a ) t + \delta \varphi(t)+\varphi_{lo1}+\varphi_p -\varphi_{lo2}]= \nonumber\\ % & \frac{1}{2}cos[2 \pi ( f_a ) t - \delta \varphi(t)-\varphi_{lo1}-\varphi_p + \varphi_{lo2 } ] + \nonumber\\ % & \frac{1}{2}cos[2 \pi ( f_a ) t + \delta \varphi(t)+\varphi_{lo1}+\varphi_p -\varphi_{lo2}]=\nonumber\\ & cos[2 \pi f_a t ] cos[\delta \varphi(t)- \phi+\varphi_p ] , \label{eqn2}\end{aligned}\ ] ] where the high frequency @xmath9 mixing product has been low pass filtered out and the phase noise sidebands are added to the phase constants .
in contrast to the previous case , shown by eq .
( 1 ) , the phase noise of the dut is not imposed on the output signal at frequency @xmath1 .
therefore , the measurement system without the band - pass filter ( fig .
1(b ) ) is incapable of measuring _
phase fluctuations of the dut . in this section
we show how the measurement system in fig .
1(b ) treats white phase noise of the dut . in this case the output signal voltage from the dut with effective temperature @xmath14 as given by , @xmath15 + \nu(t),\ ] ] where @xmath16 is the amplitude of the signal input to the dut , @xmath17 is the dut transmission coefficient and @xmath18 is the white noise from the dut with power spectral density , @xmath19 , 1\ ] ] where @xmath20 is boltzmann s constant .
therefore the power spectral density of phase fluctuations due to this thermal noise source is , @xmath21 where @xmath22 is the power of the dut input signal .
after the reference phase shifter is adjusted to maximize the output signal from the second mixer , the input signal to the symmetricom test set , can be modeled by , @xmath23 + \xi(t ) , \label{eqn3}\ ] ] where @xmath24 is the mixer s conversion coefficient . the first term in eq .
( [ eqn3 ] ) is the same as eq .
( [ eqn2 ] ) but with the phase shifter tuned to maximize the @xmath1 signal , and the second term is the broadband white phase noise term , which unlike the first term is not dependent on the chosen reference phase shift .
the power spectral density of @xmath25 can be expressed as , @xmath26 .
1 \label{eqn3 - 2}\ ] ] the effective power spectral density of phase fluctuations of the output signal due to thermal noise is then , @xmath27 this expression results from combining eqs ( [ eqn3 ] ) and ( [ eqn3 - 2 ] ) .
it is power dependent and models what is observed .
equations ( 8) and ( 5 ) coincide .
this proves that the measurement system in fig .
1(b ) , while insensitive to flicker phase noise of the dut , can correctly measure its white phase noise .
hence the technique using the phase shifter , though it _ does not _ allow us to measure the dut narrow band phase noise on the auxiliary signal , it _ does correctly _ measure its broadband white noise .
the same key elements of the configuration of fig .
1(a ) , using the band - pass filter , can be used to measure the phase noise and frequency stability of a pair of oscillators at different frequencies , where we assign @xmath28 and @xmath29 to the frequencies of the two oscillators .
the same aspects apply to this configuration as discussed above , so we ll only consider the situation where a band - pass filter is used .
if the phase shifter is used instead of the band - pass filter it can be shown , like in eq .
( 2 ) , that the relative phase fluctuations of the oscillators are not recovered on the auxiliary signal .
2 describes the basic features for the measurement method for two oscillators where a band - pass filter is used after the first mixing stage .
the main difference with fig .
1 is that two devices are needed .
one may be used as a dut and the other as a reference to obtain the measured signal frequency where , @xmath30 with the @xmath31 depending on the choice of band - pass filter .
this ensures a signal at a frequency measureable by the test set containing the phase fluctuations of the two oscillators .
here we have assumed a 400 mhz bandwidth test set is being used .
therefore eq .
( [ eqn400 ] ) means the frequency difference between the oscillators must be less than twice the bandwidth of the test set otherwise additional mixing stages are needed . in the case where a reference oscillator has much lower phase noise than the dut oscillator
the final result is that of the dut but if they are nominally identical oscillators then we get the relative phase fluctuations of the two devices .
this means 3 db must be subtracted for the phase noise of a single oscillator . from one of the oscillators an intermediate frequency
is generated with an auxiliary oscillator in the first mixing stage , which is filtered to reject one of the sidebands with the band - pass filter , and then mixed in the second stage with the signal from the other oscillator , low pass filtered , amplified and measured on the symmetricom test set .
the phase noise is read directly from the test set , but to find the allan deviation of the signal at frequency @xmath28 or @xmath29 one must scale by the ratio @xmath32 where @xmath0 is chosen as equal to either @xmath28 or @xmath29 .
in the case of fig .
2 we can model the results of the first mixing stage as proportional to , @xmath33,\ ] ] where upper @xmath1 sideband has been filtered out by the band - pass filter . the phase noise of the oscillator is represented by @xmath34 .
when this is mixed at the second mixing stage with the output of the second oscillator with its own phase noise @xmath35 at the frequency @xmath29 , and modeled as proportional to , @xmath36,\ ] ] the resulting signal frequency dependence can be modeled by the product of the latter with the former as , @xmath37 \times \nonumber \\ & cos[2 \pi f_{02 } t + \delta \varphi_2 ( t ) + \varphi_{lo2 } ] = \nonumber \\ & \frac{1}{2}cos[2 \pi ( f_{01}-f_{02}+f_a ) t + \delta \varphi_1(t)+\delta \varphi_2 ( t ) + \phi ] , \label{eqn4}\end{aligned}\ ] ] where @xmath8 is a phase constant and the high frequency mixing product of order @xmath38 has been filtered out .
the power of phase fluctuations of the resulting signal at frequency @xmath39 is equal to the combined power of phase fluctuations of the individual microwave oscillators .
of course the low pass filter must now pass the signal with frequency @xmath39 , and this frequency must be within the bandwidth of the measurement test set .
this then necessarily affects the frequency at which the phase noise is measured and the scaling ratio to calculate the allan deviation becomes @xmath40 . if the oscillators are nominally identical an additional @xmath41 factor must be applied to get the stability of a single oscillator .
the legend indicates the input power to the amplifier and the measurement noise floor ( curve 4 ) which was measured by removing the dut from the circuit .
rf amplifier = minicircuits zfl-500ln+.,width=288 ] = 20 mhz was measured , where a aml amplifier model no .
aml812pna5402 , operating at 11.2 ghz , was used with different input power levels .
the legend indicates the input power to the amplifier and the measurement noise floor ( curve 4 ) which was measured by removing the dut from the circuit .
a smoothing by taking a 50 point average was applied to data of curve 4 .
rf amplifier = minicircuits era-5 + .
see discussion in text.,width=288 ]
using the measurement method of fig .
1(a ) we measured the phase noise of an endwave amplifier ( model number jca812 - 5001 ) as a function of input power .
the band - pass filter is tunable home made cavity filter and the low - pass filter a minicircuits slp-21.4 + .
the microwave mixers are watkins johnson model number m14a .
the function generator is a 20 mhz agilent model 33220a .
the final rf amplifier used was either a minicircuits zfl-500ln+ or an era-5 + amplifier .
the results are shown in fig .
3 . where these are above the measurement noise floor
, they are consistent with those expected for the amplifier and indicate a flicker law for fourier frequencies @xmath42 hz .
the thermal noise floor is dependent on input power to the amplifier as expected .
a carrier frequency of @xmath43 ghz and an auxiliary frequency @xmath44 mhz were used in these measurements .
the measurement system noise floor was measured by removing the dut and it was found that it limits the measurements for fourier frequencies @xmath45 hz .
the phase noise of the final rf amplifier was determined to be the principle limitation of the measurement system especially at fourier frequencies @xmath46 hz . by using either a minicircuits era-5 + or a zfl-500ln+ amplifier the measurement system white noise floor
was limited to about -163 dbc / hz and about -158 dbc / hz ( on the 20 mhz carrier ) respectively .
the former is very close to white noise floor of the symmetricom 5125a test set itself .
the additive phase noise of the rf amplifiers used here were separately measured by splitting the 20 mhz signal of the auxiliary oscillator and passing one signal through the amplifier to the input channel of the test set and the other directly to the reference channel . in both cases , the same input power to the amplifier was used as in the measurement system , and the results matched the thermal noise floor of the measurement system when using these rf amplifiers . finally , when using the era-5 + rf amplifier , a pair of minicircuits zx05 - 153 + mixers were substituted for the watkins johnson m14a mixers and this raised the measurement system white noise floor 3 db to about -160 dbc / hz , indicating that a small contribution also may come from the choice of mixers used .
using the method of fig .
1(b ) and a very low phase noise aml x - band microwave amplifier we measured essentially the phase noise floor of the measurements system as a function of input power to the x - band amplifier . in this configuration where a mechanical phase shifter ( an arra 9428b in this case ) is used instead of
the band - pass filter one does not get the phase noise of the dut ( nor that for the oscillator(s ) if a phase shifter is used in the configuration of fig .
the results are shown in fig .
however the measurements of fig .
4 indicate that as one decreases the input power to the x - band amplifier ( the dut ) one observes rising thermal noise . at input power levels @xmath47 dbm ( curve 1 ) the measured noise is at the measurement system noise floor . at input power levels
less than this we observe an increase in white noise approximately proportional to the change in input power . the measurement system noise floors for both configurations shown in fig .
1 were determined by removing the dut , and ensuring the power to the mixers and the rf amplifier were the same as used when making the dut phase noise measurements .
looking at figs 3 and 4 there is a clear difference in the frequency dependence and level of the measurement system phase noise floor .
the difference is that in the method of fig 1.(a ) , with the band - pass filter , the mixer noise is present for @xmath48 hz ( curve 4 in fig .
3 ) but in the method of fig 1.(b ) , without the band - pass filter , the mixer noise is not present ( curve 4 in fig .
4 ) and we only measure the test set noise floor for @xmath49 hz . at higher offset frequencies it is the rf amplifier that sets the thermal noise floor . as shown in eq .
( [ eqn2 ] ) , the dut phase noise is not added to the auxiliary signal , when the phase shifter was substituted for the band - pass filter , so also the mixers in the circuit are devices under test and their phase noise is not added but shows up as amplitude noise on the signal input to the test set . by contrast , from eq .
( [ eqn1 ] ) , when the band - pass filter is used , the phase noise of the mixers is , in addition to the phase noise of the dut , added to the signal input to the test set .
therefore with the dut removed we measure a system noise floor that includes the phase noise of the two mixers and the rf amplifier ( at the higher offset frequencies ) . using the measurement method of fig .
2 we measured the phase noise of an agilent e8257c synthesizer generating a 1 ghz signal compared to a 1 ghz signal synthesized from an ultra - low phase noise cryogenic sapphire oscillator .
the phase noise spectra are shown in fig . 5 , and the frequency stability in fig .
curve 1 represents the phase noise of the synthesizer without external reference , and curve 2 when it was referenced by a 10 mz signal from the frequency doubled output of an oscilloquartz 8607 quartz oscillator .
the measurement system used two minicircuits zx05 - 10l - s+ mixers , a zvbp-909-s+ band - pass filter , a slp-100 + low pass filter and a zfl-500ln+ amplifier .
its measurement system phase noise floor is shown in curve 3 in fig .
this was determined by using the same low phase noise oscillator on both input ports ( for oscillators 1 and 2 in fig .
the auxiliary oscillator frequency used in these measurements was @xmath50 mhz , derived from the ultra - low phase noise cryogenic sapphire oscillator , with a phase noise of -130 dbc / hz at 1 hz offset @xcite . in this case
it was necessary to scale resulting stability generated by the test set by @xmath51 mhz@xmath52 ghz @xmath53 where @xmath54 ghz .
we also compared the same oscillators where we raised the output signal frequency of the agilent e8257c synthesizer to @xmath55 ghz .
the auxiliary oscillator was still at @xmath50 mhz , which was mixed with the signal at @xmath56 ghz from the ultra - low phase noise cryogenic sapphire oscillator , then we filtered out the upper sideband with the band - pass filter leaving the lower sideband at 900 mhz to be mixed with the 1.2 ghz signal .
after low - pass filtering ( using a minicircuits slp-400 filter ) and amplification this resulted in a 300 mhz signal that was measured by the test set .
curve 4 in fig .
5 is the result where a factor of @xmath57 has been subtracted to compare the results all at 1 ghz .
it is essentially identical with curve 2 as expected . and
in fig 6 we show the allan deviation of the 1.2 ghz synthesizer signal where the correct scaling has been applied to the output of the test set . in this case @xmath58
mhz@xmath59 ghz @xmath60 .
only allan deviation data with a noise equivalent bandwidth ( neq ) of 0.5 hz is shown .
the cryogenic sapphire oscillator s phase noise and frequency stability @xcite are orders of magnitude lower than that of the agilent e8257c synthesizer . at 1 ghz
its phase noise is approximately equal to the measurement system noise floor ( curve 3 or fig .
5 ) , hence it does not contribute to these results .
figs 5 and 6 therefore show only the performance of the e8257c synthesizer . it should be noted that when the frequency difference of the two oscillators falls within the bandwidth of the test set the additive phase noise can be measured by taking the beat note of the two signals and comparing it to a previously characterized low noise reference . to confirm the validity of our transposed gain technique we used this direct comparison method to repeat the measurements in figures 5 and 6 and found no discrepancy in the results .
a technique has been developed that allows one to measure the phase noise of active components operating at frequencies well outside the bandwidth of the phase noise measurement test set being used .
digital measurements of phase fluctuations offer several advantages over the usual homodyne technique . here
is no need for the calibration of the mixer nor is there any requirement to phase lock two oscillators to maintain zero voltage at the output of the mixer during the measurement process .
the latter feature permitted , for the first time , the phase noise measurement of ultra - stable cryogenic sapphire oscillators .
this method is only limited by the operational bandwidth of the components used , and the noise floor imposed on the measurements by the two mixing stages and the final rf amplifier used to provide sufficient signal power to the test set . using the band - pass filter and x - band mixers a measurement system noise floor at low offset frequencies of @xmath61 was obtained , while with 1 ghz mixers @xmath62 was achieved . using a
minicircuits era-5 + rf amplifier at the input to the test set a measurement system white noise floor of -163 dbc / hz was obtained .
hence one may achieve a measurement white noise floor at the limits of the test set itself , near -165 dbc / hz for the 5125a model and -175 dbc / hz for 5120a-01 model with ultra - low - noise option . provided
the final rf amplifier is sufficiently low noise one can reach this white noise floor and potentially measure the phase noise of extremely low noise components .
this work was supported by arc grant lp110200142 . s. stein , `` the allan variance challenges and opportunities , '' _ ieee trans .
57 , no . 3 , 2010 , pp .
hartnett , n.r .
nand and c. lu , `` ultra - low - phase - noise cryocooled microwave dielectric - sapphire - resonator oscillators , '' _ appl .
_ , vol . 100 , no . 18 , 2012 , art .
i d 183501 .
m.m . driscoll and r.w .
weinert , `` spectral performance of sapphire dielectric resonator - controlled oscillators operating in the 80k to 275k temperature range , '' ieee frequency control symposium , 1995 , pp .
401 - 412 .
everard and m.a .
page - jones , `` ultra low noise microwave oscillators with low residual flicker noise , '' ieee international microwave symposium , orlando may 16th -20th 1995 , pp .
693- 696 . j.k.a .
everard and m.a .
page jones `` transposed gain microwave oscillators with low residual flicker noise , '' ieee frequency control symposium , san - fransisco june 1995 , pp .
374 - 378 .
page - jones and j.k.a .
everard . `` enhanced transposed gain microwave oscillators , '' european frequency and time forum , 5th to 7th march , brighton , 1996 , pp .
275 - 278 .
nand , j.g .
hartnett , e.n .
ivanov , g. santarelli , `` ultra - stable very - low - phase - noise signal source for very long baseline interferometry using a cryocooled sapphire oscillator , '' _ ieee trans on micro .
theory & tech .
2978 2986 , 2011 . | the digital signal processing has greatly simplified the process of phase noise measurements , especially in oscillators , but its applications are largely confined to the frequencies below 400 mhz .
we propose a novel transpose frequency technique , which extends the frequency range of digital noise measurements to the microwave domain and beyond .
we discuss two basic variations of the proposed noise measurement technique , one of which enables characterization of phase fluctuations added to the passing signal by the particular microwave component , for example by an amplifier , while another one is intended for measurements of phase fluctuations of the incoming signal itself .
shell : bare demo of ieeetran.cls for journals phase noise , phase measurement , phase spectrum , allan deviation , short - term frequency stability |
one of the longest standing astrophysical problems is the exact relationship between the birth mass of stars and the mass left when they die .
unequal estimates of these masses have led to the planetary nebula `` missing mass''problem .
a typical pn progenitor star of mass 2.5@xmath1 sheds its envelope at the end of the agb phase in a terminal superwind
. however , most pne have an ionized mass of only 0.20.6@xmath1 while the residual white dwarf may be only 0.6@xmath1 . until this mass discrepancy is reconciled , the gaps in our knowledge of the late evolution of intermediate - mass stars impacts on our ability to model the evolution and chemical enrichment of our own galaxy .
a typical agb halo provides a visible fossil record of the agb mass - loss history ( e.g. sandin et al . 2010 ) , but has a mean h@xmath0 brightness @xmath210@xmath3 fainter than the main pn shell , so a deep survey is needed .
other complementary wavebands also provide evidence for the ubiquity of detectable haloes ( see chu , these proceedings ) , at least around young to middle - aged pne . for this optical study
, we classify halos into the following groups : ( i ) agb halos , ( ii ) recombination halos , produced when the luminosity of the ionizing star drops rapidly ( corradi et al . 2003 ) , and ( iii ) ism halos surrounding optically thin pne , a feature not fully appreciated in the literature .
additional examples of the various halo types are required to allow a better statistical study of their distribution and properties .
in order to improve the statistics of agb halos and other external structures in the galactic pn population , we have undertaken an extensive , systematic search around all known pne using the supercosmos h@xmath0 survey ( shs ) , a deep , high - resolution h@xmath0+[nii ] survey of the southern galactic plane ( parker et al .
this study builds on unpublished work undertaken over the last decade from the original photographic films of the aao / ukst h@xmath0 survey ( e.g. parker 2000 ) .
we examined pne from the following catalogues : acker et al .
( 1992 , 1996 ) , kohoutek ( 2001 ) , the mash catalogues ( parker et al . 2006 ;
miszalski et al .
2008 ) , plus a significant number from our own unpublished database ; the only criterion was that the pn was located within the shs coverage .
quotient fits images ( h@xmath0/ broadband @xmath4 ) were generated for each pn in the survey area , of sufficient size to ensure that faint outer structures could be readily found .
the images were generated using the iraf task @xmath5 before converting them to png format using the aplpy library ( the astronomical plotting library in python ) .
each image was linearly stretched using the default minimum and maximum pixel values ( 0.25% and 99.75% ) .
this allowed for rapid examination of the images in common image viewing software .
some examples of newly identified outer features are shown in figure [ fig_halo ] .
we have also noted any external low ionization structures ( lis ; gonalves et al .
2001 ) such as fliers , ansae and jets around pne in the survey zone .
such features are often seen in post - common envelope ( ce ) pne ( de marco 2009 ; miszalski et al . 2009 ) , though it is unclear if lis are also present in pne that have not passed through a ce phase .
additionally , faint extended lobes and irregular halos were found around a high fraction of type i bipolar pne ( frew et al .
2006 ) , which we attribute to the higher initial mass of their progenitor stars .
lastly , we checked the nature of doubtful objects ( frew & parker 2010 ) as part of the search ( see also boissay et al . , these proceedings ) .
once this study is complete , a full catalogue of halos and outer structures will be published .
1992 , _ strasbourg - eso catalogue of galactic planetary nebulae _ , strasbourg 1996 , _ supplement to the strasbourg - eso catalogue _ , strasbourg 2003 ,
_ mnras _ , 340 , 417 2009 , _ pasp _ , 121 , 316 2010 , _ pasa _ , 27 , 129 2006 , _ mnras _ , 372 , 1081 2003 , _ apj _ , 547 , 302 2001 , _ a&a _ , 378 , 843 2008 , _ mnras _ , 384 , 525 2009 , _ a&a _ , 505 , 249 2000 , _ aao newsletter _ , 94 , 9 2005 , _ mnras _ , 362 , 689 2006 , _ mnras _ , 373 , 79 2010 , _ pasa _ , 27 , 214 | we have used the supercosmos h@xmath0 survey to look for faint outer structures such as halos , ansae and jets around known planetary nebulae across 4000 square degrees of the southern milky way .
our search will contribute to a more accurate census of these features in the galactic pn population .
candidate common - envelope pne have also been identified on the basis of their microstructures .
we also intend to determine more reliable distances for these pne , which should allow a much better statistical basis for the post - agb total mass budget .
our survey offers fresh scope to address this important issue . |
the standard approach to cosmological modelling commences with the assumption that our @xmath1dimensional space is homogeneous and isotropic at large scales .
the most general spacetime metric consistent with the existence of a cosmic time @xmath2 and the principle of spatial homogeneity and isotropy is the flrw metric @xmath3,\ ] ] where the function @xmath4 , @xmath5 , @xmath6 depends on the sign of the constant spatial curvature ( @xmath7 ) , @xmath8 is the cosmological scale factor , and we have set the speed of light @xmath9 .
a third assumption in this approach is that the large scale structure of the universe is essentially determined by gravitational interactions , and hence can be described by a metrical theory of gravitation such as general relativity ( gr ) , which we assume in this work . under these very general premises , the total density @xmath10 and the total pressure @xmath11 of the cosmological fluid as a function of scale factor @xmath12
are given by @xmath13\ ; , \label{rho - eq } \\ p & = & - \frac{1}{8\pi g}\left[\ , 2\,\frac{\ddot{a}}{a } + \frac{\dot{a}^2}{a^2 } + \frac{k}{a^2 } \,\right ] \ ; , \label{p - eq}\end{aligned}\ ] ] where @xmath14 is the newton constant .
note that if now one wishes to constrain the physical properties that hold for the matter fields in the universe it is convenient to impose the so - called _ energy conditions _
@xcite that limit the arbitrariness of the energy - momentum tensor @xmath0 of these fields or , equivalently , of the physical behavior of their associated energy density @xmath10 and pressure @xmath11 .
these conditions can be stated in a coordinate - invariant way , in terms of @xmath0 and vector fields of fixed character ( timelike , null and spacelike ) .
the most common energy conditions are is perhaps the best known example of matter fields that violates the strong energy condition , but fulfills the weak energy condition ( @xmath15 and @xmath16 ) .
see , e.g. @xcite for a short historical review . ] : * the null energy condition ( nec ) .
nec states that @xmath17 for null vectors @xmath18 , where @xmath19 is a real 4-dimensional space - time manifold and @xmath20 denotes the tangent space to @xmath19 at a point @xmath21 .
* the weak energy condition ( wec ) .
wec states that @xmath22 for any timelike vector @xmath23 .
this will also imply , by continuity , the nec . *
the strong energy condition ( sec ) .
sec is the assertion that for any timelike vector @xmath24 , where @xmath25 is the trace of @xmath0 . * the dominant energy condition ( dec ) .
dec requires that @xmath22 for any timelike vector @xmath23 and the additional requirement that @xmath26 be a non - spacelike vector . by continuity
this will also hold for any null vector @xmath18 . in terms of the energy - momentum tensor for a perfect fluid , @xmath27
, the above conditions ( see , e.g. , @xcite ) reduce to @xmath28 where the nec restriction ( @xmath29 ) has been incorporated by the wec .
thus , from equations ( [ rho - eq])([p - eq ] ) one can easily rewrite the energy conditions as a set of dynamical constraints involving the scale factor @xmath8 and its derivatives for any spatial curvature , i.e. , @xmath30 \leq \frac{\ddot{a}}{a } \leq \frac{\dot{a}^2}{a^2}+\frac{k}{a^2 } \label{dec - eq}.\end{aligned}\ ] ] clearly , in a expanding flrw universe the sec implies that the expansion of the universe is decelerating irrespective of the sign of the spatial curvature . furthermore , since most of the ordinary forms of matter obey the dec they inevitably fulfill the less restrictive wec
. however , violation of the dec does not necessarily implies violation of wec or even of the sec . in this
_ brief report _ , to shed some light on the energy conditions interrelations from an observational viewpoint , we use the above dynamical formulation of the energy conditions to derive model - independent bounds on the luminosity distance @xmath31 of cosmic sources in the expanding flrw flat universe .
we then concretely confront these bounds with current sne ia observations , as provided by the _ high - z supernovae team _ ( hzst ) @xcite and the _ supernova legacy survey _ ( snls ) collaboration @xcite .
in practice , distance measurements to distant sources are made in terms of _ distance modulus _ @xmath32 , where @xmath33 is the apparent magnitude of the source and @xmath19 its absolute magnitude .
as it is well known , the distance modulus is related to the luminosity distance @xmath31 via @xmath34 where @xmath35 is measured in mpc . in an expanding flrw spatially flat universe @xmath35
is given by @xmath36 where the subscript @xmath37 stands for the present - day quantities . in what follows
we focus our attention only on the flat ( @xmath38 ) case .
model - independent bounds on the distance modulus @xmath39 as a function of the redshift for two different values of the hubble parameter , i.e , @xmath40 ( dashed lines ) and @xmath41 ( solid lines ) . from top
to bottom the curves correspond to the wec , sec and dec @xmath39 predictions .
note that , in the _ hierarchy _ of the predictions for @xmath39 , violation of the wec also means violation of both sec and dec.,width=245,height=302 ] * a. wec * in order to obtain the bounds provided by the wec on @xmath39 we note that eq .
( [ wec - eq ] ) can be written as @xmath42 or , equivalently , @xmath43 where @xmath44 is the velocity of expansion of the universe as a function of scale factor , and @xmath45 is the hubble parameter today .
now , making use of the inequality ( [ wec - ineq ] ) we integrate ( [ dist - lumin ] ) to obtain the following upper bound for the distance modulus : @xmath46 + 25,\ ] ] where we have used that @xmath47 .
clearly , if the wec is obeyed then @xmath39 must take values such that eq .
( [ wec - bound ] ) holds .
note also that the condition @xmath48 gives no further restrictions on the distance modulus , and therefore the wec and nec bounds on @xmath39 coincide [ eq .
( [ wec - bound ] ) ] .
* b. sec * similarly to the sec , eq . ( [ sec - eq ] ) implies @xmath49 the inequality above , along with equations ( [ dist - mod ] ) and ( [ dist - lumin ] ) , furnishes @xmath50 + 25 \;.\ ] ] * c. dec * dec provides an upper and a lower bound on the rate of expansion .
indeed , from ( [ dec - eq ] ) one obtains @xmath51 and @xmath52 @xmath53 @xmath54 .
again , these inequalities , along with ( [ dist - mod ] ) and ( [ dist - lumin ] ) , give the following upper and lower bounds for the distance modulus : @xmath55 + 25 \leq \mu(z ) \nonumber \\ & & \leq 5\,\log_{10}\ , \left[\,h_0^{-1}\,z(1+z)\,\right ] + 25.\end{aligned}\ ] ] as expected [ see eqs .
( [ wec - eq ] ) and ( [ dec - eq ] ) ] , the upper bound of dec coincides with the wec-@xmath39 prediction . in this case , we find , respectively , for wec , sec and dec , @xmath56 , @xmath57 , @xmath58 . ] .
figure 1 shows the distance modulus @xmath39 inequalities [ eqs .
( [ wec - bound]-[dec - bound ] ) ] as a function of the redshift parameter by taking @xmath59 mpc . from top
to bottom the curves correspond , respectively , to the wec , sec and dec predictions for @xmath39 . to plot these curves
we have assumed two different values for @xmath60 , i.e. , @xmath40 ( dashed lines ) and @xmath41 ( solid lines ) , which correspond to @xmath61 of the value provided by current cmb measurements @xcite .
note that , in the _ hierarchy _ of the predictions for @xmath39 , violation of the wec also means violation of both sec and dec , although the opposite does not necessarily hold true .
note also that the curves depend very weakly on the value adopted for the hubble parameter . as an example , at @xmath62 , the difference between the sec-@xmath39 prediction for these two values of @xmath60 is smaller than 1% , which is also maintained for higher values of @xmath63 . in figure 2
we confront the ec predictions for @xmath39 with the current sne ia observations .
the data points appearing in the first row ( panels 2a , 2b and 2c ) correspond to the so - called _ gold _ sample of 157 events distributed over the redshift interval @xmath64 @xcite , whereas in the second row ( panels 2d , 2e and 2f ) the data set corresponds to the first year results of the planned five years snls collaboration @xcite .
the snls sample includes 71 high-@xmath63 sne ia in the redshift range @xmath65 and 44 low-@xmath63 sne ia ( @xmath66 ) . in all the panels above we have adopted @xmath67 , in agreement with third - year wmap results @xcite .
figures 2a and 2d show the upper - bound curves @xmath39 for the wec - fulfillment .
an interesting aspect of these panels is that they seem to suggest that the wec might have been violated by some of the nearby sne ia ( @xmath68 ) in both samples . as an example , let us take the cases of the sne 1992aq , 1996ab and 1997n ( from _ gold _ sample ) which are , respectively , at @xmath69 , @xmath70 and @xmath71 . while their observed distance modulus are @xmath72 , @xmath73 and @xmath74 @xcite , the upper - bound wec predictions for the corresponding redshifts are @xmath75 , @xmath76 and @xmath77 , respectively .
a similar example of wec violation also happens for some sne ia from the slns sample .
besides the low-@xmath63 cases of the sne 1992bs and 1992bh , the high-@xmath63 snls-04d3cp at @xmath78 has @xmath79 @xcite whereas the upper - bound wec prediction is @xmath80 , i.e. , about @xmath81 below the central value observed by the snls collaboration .
note that these same considerations are also applied to the upper - bound dec predictions [ eq .
( [ dec - bound ] ) ] .
note also that the lower - bound of dec is not violated by the current sne ia data ( figs .
2c and 2f ) .
still on wec and dec violations , it is worth mentioning that the observed luminosity distance derived from sne ia observations may be fitted by a dark energy component violating these energy conditions , the so - called phantom fields , first noticed by caldwell @xcite ( see also @xcite and references therein ) .
however , differently from the above indications of wec and dec violations , most of these results are derived in a model - dependent way and under the assumption that the dark energy equation of state @xmath82 is constant .
a similar analysis for the sec - fulfillment is shown in figs .
2b and 2e . for sec , which is responsible to ensure the attractiveness of gravity and , consequently , a decelerating cosmic expansion , the plots clearly indicate a breakdown for essentially all the sne ia in the redshift interval covered by _
gold _ and snls samples .
the interesting aspect here is that the violation of sec happens even at very high-@xmath63 , as indicated by the examples of the sne 2003az ( _ gold _ ) and 04d3dd ( snls ) at @xmath83 and @xmath84 and whose observed distance modulus are , respectively , @xmath85 and @xmath86 . for these redshifts ,
the corresponding upper - bound sec predictions are @xmath87 and @xmath88 .
the most interesting example , however , comes from the sn 2002fw ( _ gold _ ) at @xmath89 .
while its observed magnitude is @xmath90 , the upper - bound sec prediction for the same redshift is @xmath91 , which is about @xmath92 below the central value measured by the hzst collaboration .
the energy conditions play an important role in the understanding of several properties of our universe , including the current and past accelerating expansion phases in the context of flrw models .
they are also necessary in the formulation and proof of various singularity theorems in classical black hole thermodynamics ( e.g. , the proof of the second law of black hole thermodynamics requires the null energy conditions ( @xmath16 ) , whereas the hawking - penrose singularity theorems invoke the strong energy condition , @xmath93 and , by continuity , @xmath16 ) @xcite . in this paper , by using the fact that the classical energy conditions can be translated into differential constraints involving the scale factor @xmath12 and its derivatives [ see eqs.([wec - eq]-[dec - eq ] ) ] , we have derived model - independent bounds on the luminosity distance of extragalactic sources . we have confronted these energy - condition - fulfillment bounds with current sne ia observations from _ gold _ and snls samples and shown that all the energy conditions seem to have been violated in a recent past of the cosmic evolution .
the most surprising fact is that these violations may have happened even at high-@xmath63 [ @xmath94 , when the universe is expected to be dominated by usual matter fields .
finally , we emphasize that in agreement with other recent studies @xcite , the results reported here reinforce the idea that no possible combination of _ normal _ matter is capable of fitting the current observational data .
the authors are grateful to n. pires for valuable discussions .
j. s. acknowledges the support of pronex ( cnpq / fapern ) .
j. s. a. and m. j. r. thank cnpq for the grants under which this work was carried out .
m. visser and c. barcel , _ energy conditions and their cosmological implications _
, cosmo99 proceedings : trieste , sept / oct 1999 .
world scientific ; gr - qc/0001099 ; c. barcel and m. visser , int .
d * 11 * , 1553 ( 2002 ) .
r. r. caldwell , phys .
b * 545 * , 23 ( 2002 ) . v. faraoni , int . j. mod
d11 * , 471 ( 2002 ) ; p. f. gonzalez - diaz , phys .
* d68 * , 021303 ( 2003 ) ; j. s. alcaniz , phys . rev . * d69 * , 083521 ( 2004 ) . ` astro - ph/0312424 ` ; s. nesseris and l. perivolaropoulos , phy
. rev . * d70
* , 123529 ( 2004 ) ; j. santos and j. s. alcaniz , phys . lett .
* b619 * , 11 ( 2005 ) . `
astro - ph/0502031 ` ; f. c. carvalho and a. saa , phys . rev .
* d70 * , 087302 ( 2004 ) . | in general relativity , the energy conditions are invoked to restrict general energy - momentum tensors @xmath0 on physical grounds .
we show that in the standard friedmann lematre
robertson
walker ( flrw ) approach to cosmological modelling where the equation of state of the cosmological fluid is unknown , the energy conditions provide model - independent bounds on the behavior of the distance modulus of cosmic sources as a function of the redshift .
we use both the _ gold _ and the _ legacy _ samples of current type ia supenovae to carry out a model - independent analysis of the energy conditions violation in the context of standard cosmology . |
recently , the _ reconfiguration problems _ attract the attention from the viewpoint of theoretical computer science .
the problem arises when we wish to find a step - by - step transformation between two feasible solutions of a problem such that all intermediate results are also feasible and each step abides by a fixed reconfiguration rule , that is , an adjacency relation defined on feasible solutions of the original problem .
the reconfiguration problems have been studied extensively for several well - known problems , including independent set @xcite , satisfiability @xcite , set cover , clique , matching @xcite , vertex - coloring @xcite , shortest path @xcite , and so on .
the reconfiguration problem can be seen as a natural `` puzzle '' from the viewpoint of recreational mathematics .
the _ 15 puzzle _ is one of the most famous classic puzzles , that had the greatest impact on american and european society of any mechanical puzzle the word has ever known in 1880 ( see @xcite for its rich history ) .
it is well known that the 15 puzzle has a parity ; for any two placements , we can decide whether two placements are reconfigurable or not by checking the parity .
therefore , we can solve the reconfiguration problem in linear time just by checking whether the parity of one placement coincides with the other or not .
moreover , we can say that the distance between any two reconfigurable placements is @xmath4 , that is , we can reconfigure from one to the other in @xmath4 sliding pieces when the size of the board is @xmath5 .
however , surprisingly , for these two reconfigurable placements , finding a shortest path is np - complete in general @xcite .
namely , although we know that it is @xmath4 , finding a shortest one is np - complete .
another interesting property of the 15 puzzle is in another case of generalization . in the 15 puzzle
, every peace has the same unit size of @xmath6 .
we have the other famous classic puzzles that can be seen as a generalization of this viewpoint .
that is , when we allow to have rectangles , we have the other classic puzzles , called `` dad puzzle '' and its variants ( see [ fig : dad ] ) .
gardner said that `` these puzzles are very much in want of a theory '' in 1964 @xcite , and hearn and demaine have gave the theory after 40 years @xcite ; they prove that these puzzles are pspace - complete in general using their nondeterministic constraint logic model @xcite . that is , the reconfiguration of the sliding block puzzle is pspace - complete in general decision problem , and linear time solvable if every block is the unit square .
however , finding a shortest reconfiguration for the latter easy case is np - complete . in other words
, we can characterize these three complexity classes using the model of sliding block puzzle .
from the viewpoint of theoretical computer science , one of the most important problems is the 3sat problem . for this 3sat problem ,
a similar trichotomy for the complexity of finding a shortest sequence has been shown recently ; that is , for the reconfiguration problem of 3sat , finding a shortest sequence between two satisfiable assignments is in p , np - complete , or pspace - complete in certain conditions @xcite . in general ,
the reconfiguration problems tend to be pspace - complete , and some polynomial time algorithms are shown in restricted cases . in the reconfiguration problems , finding a shortest sequence can be a new trend in theoretical computer science because it has a great potential to characterize the class np from a new viewpoint . beside the 3sat problem ,
one of the most important problems in theoretical computer science is the independent set problem . recall that an _ independent set _ of a graph @xmath3 is a vertex - subset of @xmath3 in which no two vertices are adjacent .
( see [ fig : example ] which depicts five different independent sets in the same graph . ) for this notion , the natural reconfiguration problem is called the sliding token problem introduced by hearn and demaine @xcite : suppose that we are given two independent sets @xmath0 and @xmath1 of a graph @xmath7 such that @xmath2 , and imagine that a token ( coin ) is placed on each vertex in @xmath0 .
then , the sliding token problem is to determine whether there exists a sequence @xmath8 of independent sets of @xmath3 such that aaa @xmath9 , @xmath10 , and @xmath11 for all @xmath12 , @xmath13 ; and for each @xmath12 , @xmath14 , there is an edge @xmath15 in @xmath3 such that @xmath16 and @xmath17 , that is , @xmath18 can be obtained from @xmath19 by sliding exactly one token on a vertex @xmath20 to its adjacent vertex @xmath21 along @xmath22 .
figure [ fig : example ] illustrates a sequence @xmath23 of independent sets which transforms @xmath24 into @xmath25 .
hearn and demaine proved that the sliding token problem is pspace - complete for planar graphs . of independent sets of the same graph , where the vertices in independent sets are depicted by small black circles ( tokens).,scaledwidth=80.0% ] ( we note that the reconfiguration problem for independent set have some variants . in @xcite ,
the reconfiguration problem for independent set is studied under three reconfiguration rules called `` token sliding , '' `` token jumping , '' and `` token addition and removal . '' in this paper , we only consider token sliding model , and see @xcite for the other models . ) for the sliding token problem , some polynomial time algorithms are investigated as follows : linear time algorithms have been shown for cographs ( also known as @xmath26-free graphs ) @xcite and trees @xcite .
polynomial time algorithms are shown for bipartite permutation graphs @xcite , and claw - free graphs @xcite . on the other hand , pspace - completeness
is also shown for graphs of bounded tree - width @xcite , and planar graphs @xcite . in this context
, we investigate for finding a shortest sequence of the sliding token problem , which is called the shortest sliding token problem .
that is , our problem is formalized as follows : aaa a graph @xmath27 and two independent sets @xmath28 with @xmath2 . a _ shortest _ reconfiguration sequence @xmath29 , @xmath30 , @xmath31 , @xmath10 such that @xmath18 can be obtained from @xmath19 by sliding exactly one token on a vertex @xmath20 to its adjacent vertex @xmath21 along @xmath22 for each @xmath12 , @xmath14 .
we note that @xmath32 is not necessarily in polynomial of @xmath33 ; this is an issue how we formalize the problem , and if we do not know that @xmath32 is in polynomial or not .
if the length @xmath34 is given as a part of input , we may be able to decide whether @xmath35 in polynomial time even if @xmath32 itself is not in polynomial .
however , if we have to output the sequence itself , it can not be solved in polynomial time if @xmath32 is not in polynomial . in this paper
, we will show that the shortest sliding token problem is solvable in polynomial time for the following graph classes : [ [ proper - interval - graphs ] ] proper interval graphs : + + + + + + + + + + + + + + + + + + + + + + + we first prove that any two independent sets of a proper interval graph can be transformed into each other . in other words ,
every proper interval graph with two independent sets @xmath0 and @xmath1 is a yes - instance of the problem if @xmath2 . furthermore , we can find the ordering of tokens to be slid in a minimum - length sequence in @xmath36 time ( implicitly ) , even though there exists an infinite family of independent sets on paths ( and hence on proper interval graphs ) for which any sequence requires @xmath37 length . [ [ trivially - perfect - graphs ] ] trivially perfect graphs : + + + + + + + + + + + + + + + + + + + + + + + + + we then give an @xmath36-time algorithm for trivially perfect graphs which actually finds a shortest sequence if such a sequence exists .
in contrast to proper interval graphs , any shortest sequence is of length @xmath36 for trivially perfect graphs .
note that trivially perfect graphs form a subclass of cographs , and hence its polynomial time solvability has been known @xcite .
[ [ caterpillars ] ] caterpillars : + + + + + + + + + + + + + we finally give an @xmath38-time algorithm for caterpillars for the shortest sliding token problem . to make self - contained ,
we first show a linear time algorithm for decision problem that asks whether two independent sets can be transformed into each other .
( we note that this problem can be solved in linear time for a tree @xcite . ) for a yes - instance , we next show an algorithm that finds a shortest sequence of token sliding between two independent sets .
we here remark that , since the problem is pspace - complete in general , an instance of the sliding token problem may require the exponential number of independent sets to transform .
in such a case , tokens should make detours to avoid violating to be independent ( as shown in [ fig : example ] ) . as we will see , caterpillars certainly require to make detours to transform .
therefore , it is remarkable that any yes - instance on a caterpillar requires a sequence of token - slides of polynomial length .
this is still open even for a tree .
that is , in a tree , we can determine if two independent sets are reconfigurable in linear time due to @xcite , however , we do not know if the length of the sequence is in polynomial .
as far as the authors know , this is the first polynomial time algorithm for the shortest sliding token problem for a graph class that requires detours of tokens .
in this section , we introduce some basic terms and notations . in the sliding token problem , we may assume without loss of generality that graphs are simple and connected .
for a graph @xmath27 , we let @xmath39 and @xmath40 . for two independent sets @xmath41 and @xmath42 of the same cardinality in a graph @xmath27 ,
if there exists exactly one edge @xmath15 in @xmath3 such that @xmath43 and @xmath44 , then we say that @xmath45 can be obtained from @xmath18 by _ sliding _ a token on the vertex @xmath46 to its adjacent vertex @xmath21 along the edge @xmath15 , and denote it by @xmath47 .
we remark that the tokens are unlabeled , while the vertices in a graph are labeled . a _ reconfiguration sequence _ between two independent sets @xmath48 and @xmath49 of @xmath3 is a sequence @xmath8 of independent sets of @xmath3 such that @xmath50 for @xmath51 .
we denote by @xmath52 if there exists a reconfiguration sequence between @xmath48 and @xmath49 .
we note that a reconfiguration sequence is _
reversible _ , that is , we have @xmath52 if and only if @xmath53 .
thus we say that two independent sets @xmath48 and @xmath49 are _ reconfigurable _ into each other if @xmath52 .
the _ length _ of a reconfiguration sequence @xmath54 is defined as the number of independent sets contained in @xmath54 .
for example , the length of the reconfiguration sequence in [ fig : example ] is @xmath55 .
the sliding token problem is to determine whether two given independent sets @xmath0 and @xmath1 of a graph @xmath3 are reconfigurable into each other .
we may assume without loss of generality that @xmath56 ; otherwise the answer is clearly `` no . ''
note that the sliding token problem is a decision problem asking for the existence of a reconfiguration sequence between @xmath0 and @xmath1 , and hence it does not ask an actual reconfiguration sequence . in this paper , we will consider the shortest sliding token problem that computes a shortest reconfiguration sequence between two independent sets . note that the length of a reconfiguration sequence may not be in polynomial of the size of the graph since the sequence may contain detours of tokens .
we always denote by @xmath0 and @xmath1 the initial and target independent sets of @xmath3 , respectively , as an instance of the ( shortest ) sliding token problem ; we wish to slide tokens on the vertices in @xmath0 to the vertices in @xmath1 .
we sometimes call the vertices in @xmath0 _ blue _ , and the vertices in @xmath1 _ red _ ; each vertex in @xmath57 is blue _ and _ red .
we here give another notation of the sliding token problem , which is useful to explain our algorithm .
let @xmath58 be an initial independent set of a graph @xmath3 .
for the sake of convenience , we label the tokens on the vertices in @xmath0 ; let @xmath59 be the token placed on @xmath60 for each @xmath12 , @xmath61 .
let @xmath54 be a reconfiguration sequence between @xmath0 and an independent set @xmath62 of @xmath3 , and hence @xmath63 .
then , for each token @xmath59 , @xmath61 , we denote by @xmath64 the vertex in @xmath62 on which the token @xmath59 is placed via the reconfiguration sequence @xmath54 . notice that @xmath65 .
let @xmath1 be a target independent set of @xmath3 , which is not necessarily reconfigurable from @xmath0 .
then , we call a mapping @xmath66 a _ target - assignment _ between @xmath0 and @xmath1 .
the target - assignment @xmath67 is said to be _ proper _ if there exists a reconfiguration sequence @xmath54 such that @xmath68 for all @xmath12 , @xmath61 .
note that there is no proper target - assignment between @xmath0 and @xmath1 if @xmath69 .
therefore , the sliding token problem can be seen as the problem of determining whether there exists at least one proper target - assignment between @xmath0 and @xmath1 .
the _ neighborhood _ of a vertex @xmath21 in a graph @xmath27 is the set of all vertices adjacent to @xmath21 , and we denote it by @xmath70 .
let @xmath71 = n(v)\cup\{v\}$ ] .
for any graph @xmath27 , two vertices @xmath72 and @xmath21 are called _ strong twins _ if @xmath73=n[v]$ ] , and _ weak twins _ if @xmath74 . in our problem ,
strong twins have no meaning : when @xmath72 and @xmath21 are strong twins , only one of them can be used by a token .
therefore , in this paper , we only consider the graphs without strong twins .
that is , for any pair of vertices @xmath72 and @xmath21 , we have @xmath73\neq n[v]$ ] .
( we have to take care about weak twins ; see section [ sec : caterpillars ] for the details . ) a graph @xmath7 with @xmath75 is an _ interval graph _ if there exists a set @xmath76 of ( closed ) intervals @xmath77 such that @xmath78 if and only if @xmath79 for each @xmath12 and @xmath80 with @xmath81 .
denotes an `` independent set , '' an italic @xmath82 denotes an `` interval , '' and calligraphy @xmath76 denotes `` a set of intervals . '' ] we call the set @xmath76 of intervals an _ interval representation _ of the graph , and sometimes identify a vertex @xmath83 with its corresponding interval @xmath84 .
we denote by @xmath85 and @xmath86 the left and right endpoints of an interval @xmath87 , respectively .
that is , we always have @xmath88 for any interval @xmath89 $ ] . to specify the bottleneck of the running time of our algorithms
, we suppose that an interval graph @xmath27 is given as an input by its interval representation using @xmath36 space .
( if necessary , an interval representation of @xmath3 can be found in @xmath90 time @xcite . )
more precisely , @xmath3 is given by a string of length @xmath91 over alphabets @xmath92 .
for example , a complete graph @xmath93 with three vertices can be given by an interval representation @xmath94 , and a path of length two is given by an interval representation @xmath95 .
an interval graph is _ proper _ if it has an interval representation such that no interval properly contains another .
the class of proper interval graphs is also known as the class of unit interval graphs @xcite : an interval graph is _ unit _ if it has an interval representation such that every interval has unit length .
hereafter , we assume that each proper interval graph is given in the interval representation of intervals of unit length . in the context of the interval representation ,
an interval graph is proper if and only if @xmath96 if and only if @xmath97 .
an interval graph is _ trivially perfect _ if it has an interval representation such that the relationship between any two intervals is either disjoint or inclusion .
that is , for any two intervals @xmath98 and @xmath99 with @xmath96 , we have either @xmath100 or @xmath101 . a _ caterpillar _ @xmath27 is a tree ( i.e. , a connected acyclic graph ) that consists of two subsets @xmath102 and @xmath103 of @xmath104 as follows .
the vertex set @xmath102 induces a path @xmath105 in @xmath3 , and each vertex @xmath21 in @xmath103 has degree 1 , and its unique neighbor is in @xmath102 .
we call the path @xmath105 _ spine _ , and each vertex in @xmath103 _ leaf_. in this paper , without loss of generality , we assume that @xmath106 , @xmath107 , and @xmath108
. that is , the endpoints @xmath109 and @xmath110 of the spine @xmath105 should have at least one leaf .
it is easy to see that the class of caterpillars is a proper subset of the class of interval graphs , and these three subclasses are incomparable with each other .
we show the main theorem in this section for proper interval graphs , which first says that the answer of sliding token is always `` yes '' for connected proper interval graphs . we give a constructive proof of the claim , and it certainly finds a shortest sequence in linear time .
[ th : proper ] for a connected proper interval graph @xmath27 , any two independent sets @xmath111 and @xmath112 with @xmath56 are reconfigurable into each other , that is , @xmath113 .
moreover , the shortest reconfiguration sequence can be found in polynomial time . we give a constructive proof for theorem [ th : proper ] ,
that is , we give an algorithm which actually finds a shortest reconfiguration sequence between any two independent sets @xmath0 and @xmath1 of a connected proper interval graph @xmath3 . a connected proper interval graph @xmath27 has a unique interval representation ( up to reversal ) , and we can assume that each interval is of unit length in the representation @xcite .
therefore , by renumbering the vertices , we can fix an interval representation @xmath114 of @xmath3 so that @xmath115 ( and @xmath116 ) for each @xmath12 , @xmath117 , and each interval @xmath84 corresponds to the vertex @xmath83 .
let @xmath118 and @xmath119 be any given initial and target independent sets of @xmath3 , respectively . without loss of generality , we assume that the blue vertices @xmath120 are labeled from left to right ( according to the interval representation @xmath76 of @xmath3 ) , that is , @xmath121
if @xmath122 ; similarly , we assume that the red vertices @xmath123 are labeled from left to right .
then , we define a target - assignment @xmath66 , as follows : for each blue vertex @xmath124 @xmath125 to prove theorem [ th : proper ] , it suffices to show that @xmath67 is proper , and each token takes no detours . by traversing the interval representation @xmath76 of a connected proper interval graph @xmath3 from left to right , we can obtain a string @xmath126 which is a superstring of both @xmath127 and @xmath128 , that is , each letter @xmath129 in @xmath102 is one of the vertices in @xmath130 and @xmath129 appears in @xmath102 before @xmath131 if @xmath132 .
we may assume without loss of generality that @xmath133 since the reconfiguration rule is symmetric in sliding token .
if a vertex is contained in both @xmath0 and @xmath1 , as @xmath60 and @xmath134 , then we assume that it appears as @xmath135 in @xmath102 , that is , the blue vertex @xmath60 appears in @xmath102 before the red vertex @xmath134 .
then , for each @xmath12 , @xmath136 , we define the _ height @xmath137 at @xmath12 _ by the number of blue vertices appeared in the substring @xmath138 minus the number of red vertices appeared in @xmath138 . for the sake of notational convenience ,
we define @xmath139
. then @xmath137 can be recursively computed as follows : @xmath140 note that @xmath141 for any string @xmath102 since @xmath56 . using the notion of height
, we split the string @xmath102 into substrings @xmath142 at every point of height @xmath143 , that is , in each substring @xmath144 , we have @xmath145 and @xmath146 for all @xmath12 , @xmath147 .
for example , a string @xmath148 can be split into four substrings @xmath149 , @xmath150 , @xmath151 and @xmath152 .
then , the substrings @xmath142 form a partition of @xmath102 , and each substring @xmath153 contains the same number of blue and red tokens .
we call such a partition the _ partition of @xmath102 at height @xmath143_. [ lem : balance ] let @xmath144 be a substring in the partition of the string @xmath102 at height @xmath143
. then , aaa the blue vertices @xmath154 appear in @xmath153 , and their corresponding red vertices @xmath155 appear in @xmath153 ; if @xmath153 starts with the blue vertex @xmath156 , then each blue vertex @xmath60 , @xmath157 , appears in @xmath153 before its corresponding red vertex @xmath158 ; and if @xmath153 starts with the red vertex @xmath159 , then each blue vertex @xmath60 , @xmath157 , appears in @xmath153 after its corresponding red vertex @xmath158 . by the definitions , the claim ( a ) clearly holds .
we thus show that the claim ( b ) holds .
( the proof for the claim ( c ) is symmetric . )
since @xmath160 and @xmath153 starts with a blue vertex , we have @xmath161 .
we now suppose for a contradiction that there exists a blue vertex @xmath162 which appears in @xmath153 after its corresponding red vertex @xmath163 .
then , @xmath164 . we assume that @xmath165 is the minimum index among such blue vertices in @xmath153 .
then , in the substring @xmath166 of @xmath102 , there are exactly @xmath167 red vertices . on the other hand , since @xmath164 , the substring @xmath166 contains at most @xmath168 blue vertices .
therefore , by the definition of height , we have @xmath169 . since @xmath161 and @xmath169 , by eq .
( [ eq : height ] ) there must exist an index @xmath170 such that @xmath171 and @xmath172 .
this contradicts the fact that @xmath153 is a substring in the partition of @xmath102 at height @xmath143 .
recall that we have fixed the unique interval representation @xmath114 of a connected proper interval graph @xmath3 so that @xmath115 for each @xmath12 , @xmath117 , and each interval @xmath84 corresponds to the vertex @xmath83 .
since all intervals in @xmath76 have unit length , the following proposition clearly holds .
[ prop : optimalway ] for two vertices @xmath173 and @xmath174 in @xmath3 such that @xmath122 , there is a path @xmath175 in @xmath3 which passes through only intervals ( vertices ) contained in @xmath176 $ ] .
furthermore , if @xmath177 for some index @xmath167 with @xmath178 , no vertex in @xmath179 is adjacent to any vertex in @xmath175 . if @xmath180 for some index @xmath181 with @xmath182 , no vertex in @xmath183 is adjacent to any vertex in @xmath175 .
let @xmath102 be the string of length @xmath184 obtained from two given independent sets @xmath0 and @xmath1 of a connected proper interval graph @xmath3 , where @xmath185 .
let @xmath142 be the partition of @xmath102 at height @xmath143 .
the following lemma shows that the tokens in each substring @xmath153 can always reach their corresponding red vertices .
( note that we sometimes denote simply by @xmath153 the set of all vertices appeared in the substring @xmath153 , @xmath186 . )
[ lem : proper - simple ] let @xmath144 be a substring in the partition of @xmath102 at height @xmath143 .
then , there exists a reconfiguration sequence between @xmath187 and @xmath188 such that tokens are slid along edges only in the subgraph of @xmath3 induced by the vertices contained in @xmath189 $ ] .
we first consider the case where @xmath153 starts with the blue vertex @xmath156 , that is , @xmath190 .
then , by lemma [ lem : balance](b ) each blue vertex @xmath60 , @xmath157 , appears in @xmath153 before the corresponding red vertex @xmath158 .
therefore , we know that @xmath191 , and hence it is red .
suppose that @xmath192 , then all vertices appeared in @xmath193 are red .
roughly speaking , we slide the tokens @xmath194 from left to right in this order .
we first claim that the token @xmath195 can be slid from @xmath196 @xmath197 to @xmath198 @xmath199 . by proposition [ prop : optimalway ]
there is a path @xmath175 between @xmath196 and @xmath198 which passes through only intervals contained in @xmath200 $ ] .
since @xmath0 is an independent set of @xmath3 , the vertex @xmath196 is not adjacent to any other vertices @xmath201 in @xmath187 .
since @xmath202 , by proposition [ prop : optimalway ] all vertices in @xmath175 are not adjacent to any of tokens @xmath203 that are now placed on @xmath201 , respectively .
therefore , we can slide the token @xmath195 from @xmath196 to @xmath198 .
we fix the token @xmath195 on @xmath204 , and will not slide it anymore .
we then slide the next token @xmath205 on @xmath206 to @xmath207 along a path @xmath208 which passes through only intervals contained in @xmath209 $ ] . since @xmath1 is an independent set of @xmath3 ,
the corresponding red vertex @xmath207 is not adjacent to @xmath198 on which the token @xmath195 is now placed .
recall that @xmath210 , and hence by proposition [ prop : optimalway ] , @xmath198 is not adjacent to any vertex in @xmath208 .
similarly as above , the tokens @xmath211 are not adjacent to any vertex in @xmath208 .
therefore , we can slide the token @xmath205 from @xmath206 to @xmath207 .
repeat this process until the token @xmath212 on @xmath156 is slid to @xmath159 . in this way
, there is a reconfiguration sequence between @xmath187 and @xmath188 such that tokens are slid along edges only in the subgraph of @xmath3 induced by the vertices contained in @xmath213 $ ] .
the symmetric arguments prove the case where @xmath153 starts with the red vertex @xmath159 .
note that , in this case , we slide the tokens @xmath214 from right to left in this order .
* proof of theorem [ th : proper ] .
* we now give an algorithm which slides all tokens on the vertices in @xmath111 to the vertices in @xmath1 .
recall that @xmath142 are the substrings in the partition of @xmath102 at height @xmath143 .
intuitively , the algorithm repeatedly picks up one substring @xmath153 , and slides all tokens in @xmath187 to @xmath188 . by lemma [ lem : proper - simple ]
it works locally in each substring @xmath153 , but it should be noted that a token in @xmath153 may be adjacent to another token in @xmath215 or @xmath216 at the boundary of the substrings . to avoid this ,
we define a partial order over the substrings @xmath142 , as follows .
consider any two consecutive substrings @xmath153 and @xmath216 , and let @xmath217 .
then , the first letter of @xmath216 is @xmath218 .
we first consider the case where both @xmath219 and @xmath218 are the same color . then , since @xmath219 and @xmath218 are both in the same independent set of @xmath3 , they are not adjacent . therefore , by proposition [ prop : optimalway ] and lemma [ lem : proper - simple ] , we can deal with @xmath153 and @xmath216 independently . in this case , we thus do not define the ordering between @xmath153 and @xmath216 .
we then consider the case where @xmath219 and @xmath218 have different colors ; in this case , we have to define their ordering .
suppose that @xmath219 is blue and @xmath218 is red ; then we have @xmath220 and @xmath221 . by lemma [ lem : proper - simple ] the token @xmath195 on @xmath219
is slid to left , and the token @xmath222 will reach @xmath223 from right . therefore , the algorithm has to deal with @xmath153 before @xmath216 .
note that , after sliding all tokens @xmath214 in @xmath153 , they are on the red vertices @xmath224 , respectively , and hence the tokens in @xmath216 are not adjacent to any of them . by the symmetric argument ,
if @xmath219 is red and @xmath218 is blue , @xmath216 should be dealt with before @xmath153 .
notice that such an ordering is defined only for two consecutive substrings @xmath153 and @xmath216 , @xmath225 .
therefore , the partial order over the substrings @xmath142 is acyclic , and hence there exists a total order which is consistent with the partial order defined above .
the algorithm certainly slides all tokens from @xmath0 to @xmath1 according to the total order .
therefore , the target - assignment @xmath67 defined in eq .
( [ eq : map_proper ] ) is proper , and hence @xmath113 .
therefore , there always exists a reconfiguration sequence between two independent sets @xmath0 and @xmath1 of a connected proper interval graph @xmath3 .
we now discuss the length of reconfiguration sequences between @xmath0 and @xmath1 , together with the running time of our algorithm .
[ prop : proper ] for two given independent sets @xmath0 and @xmath1 of a connected proper interval graph @xmath3 with @xmath226 vertices , aaa the ordering of tokens to be slid in a shortest reconfiguration sequence between them can be computed in @xmath36 time and @xmath36 space ; and a shortest reconfiguration sequence between them can be output in @xmath38 time and @xmath36 space .
we first modify our algorithm so that it finds a shortest reconfiguration sequence between @xmath0 and @xmath1 . to do that
, it suffices to slide each token @xmath59 , @xmath61 , from the blue vertex @xmath60 to its corresponding red vertex @xmath158 along the shortest path between @xmath60 and @xmath134 .
we may assume without loss of generality that @xmath227 , that is , the token @xmath59 will be slid from left to right .
then , for the interval @xmath60 , we choose an interval @xmath228 such that @xmath229 and @xmath230 is the maximum among all @xmath231 .
if @xmath232 , we can slide @xmath59 from @xmath60 to @xmath158 directly ; otherwise we slide @xmath59 to the vertex @xmath99 , and repeat .
we then prove the claim ( 1 ) .
if we simply want to compute the ordering of tokens to be slid in a shortest reconfiguration sequence , it suffices to compute the partial order over the substrings @xmath142 in the partition of the string @xmath102 at height @xmath143 .
it is not difficult to implement our algorithm in section [ subsec : algo_proper ] to run in @xmath36 time and @xmath36 space .
therefore , the claim ( 1 ) holds .
we finally prove the claim ( 2 ) . remember that each token @xmath59 , @xmath61 , is slid along the shortest path from @xmath60 to @xmath158 .
furthermore , once the token @xmath59 reaches @xmath158 , it is not slid anymore .
therefore , the length of a shortest reconfiguration sequence between @xmath111 and @xmath112 is given by the sum of all lengths of the shortest paths between @xmath60 and @xmath158 .
it is clear that this sum is @xmath233 .
we output only the shortest paths between @xmath60 and @xmath158 , together with the ordering of the tokens to be slid .
therefore , the claim ( 2 ) holds .
this proposition also completes the proof of theorem [ th : proper ] .
it is remarkable that there exists an infinite family of instances for which any reconfiguration sequence requires @xmath37 length . to show this
, we give an instance such that each shortest path between @xmath60 and @xmath158 is @xmath234 .
simple example is : @xmath3 is a path @xmath235 of length @xmath236 for any positive integer @xmath34 , @xmath237 , and @xmath238 . in this instance , each token @xmath59 must be slid @xmath234 times , and hence it requires @xmath239 time to output all of them .
we note that a path is not only a proper interval graph , but also a caterpillar .
thus this simple example also works as a caterpillar .
the main result of this section is the following theorem . [
th : trivialperfect ] the sliding token problem for a trivially perfect graph @xmath7 can be solved in @xmath36 time and @xmath36 space .
furthermore , one can find a shortest reconfiguration sequence between two given independent sets @xmath0 and @xmath1 in @xmath36 time and @xmath36 space if there exists . in this section
, we explicitly give such an algorithm as a proof of theorem [ th : trivialperfect ] .
note that there are no - instances for trivially perfect graphs .
however , for trivially perfect graphs , we construct a proper target - assignment between @xmath0 and @xmath1 efficiently if it exists . the @xmath240-tree of an interval graph @xmath3 is a kind of decomposition tree , developed by korte and mhring @xcite , which represents the set of all feasible interval representations of @xmath3 . for an interval graph @xmath3 , although there are exponentially many interval representations for @xmath3 , its corresponding @xmath240-tree is unique up to isomorphism . for the notion of @xmath240-trees , the following theorem is known : [ th : mpq ] for any interval graph @xmath27 , its corresponding @xmath240-tree can be constructed in @xmath90 time .
since it is involved to define @xmath240-tree for general interval graphs , we here give a simplified definition of @xmath240-tree only for the class of trivially perfect graphs .
( see @xcite for the detailed definition of the @xmath240-tree for a general interval graph . )
let @xmath27 be a trivially perfect graph . recall that a trivially perfect graph has an interval representation such that the relationship between any two intervals is either disjoint or inclusion .
then , the _ @xmath240-tree _ @xmath241 of @xmath3 is a rooted tree such that each node , called a _
@xmath242-node _ , in @xmath241 is associated with a non - empty set of vertices in @xmath3 such that ( a ) each vertex @xmath243 appears in exactly one @xmath242-node in @xmath241 , and ( b ) if a vertex @xmath83 is in an ancestor node of another node that contains @xmath244 , then @xmath245 in any interval representation of @xmath3 , where @xmath173 and @xmath174 correspond to the intervals @xmath98 and @xmath99 , respectively ( see [ fig : mpq ] as an example ) . by the property
( b ) , the ancestor / descendant relationship on @xmath241 corresponds to the inclusion relationship in the interval representation of @xmath3 .
thus , @xmath246\subseteq n[v_i]$ ] if @xmath173 is in an ancestor node of another node that contains @xmath174 in the @xmath240-tree .
let @xmath241 be the ( unique ) @xmath240-tree of a connected trivially perfect graph @xmath7 . for two vertices @xmath72 and @xmath247 in @xmath3 , we denote by @xmath248 the least common ancestor in @xmath241 for the nodes containing @xmath72 and @xmath247 . by the property ( a ) the node @xmath248 can be uniquely defined .
-tree.,scaledwidth=60.0% ] let @xmath241 be the ( unique ) @xmath240-tree of a connected trivially perfect graph @xmath7 .
recall that the interval representation of a trivially perfect graph has just disjoint or inclusion relationship .
this fact implies the following observation .
[ obs : twosteps ] every pair of vertices @xmath72 and @xmath247 in a connected trivially perfect graph @xmath3 has a path of length at most two via a vertex in @xmath248 .
we first observe that every @xmath242-node of @xmath241 is non - empty .
if the root is empty , the graph is disconnected , which is a contradiction .
if some non - root @xmath242-node @xmath175 is empty , joining all children of @xmath175 to the parent of @xmath175 , we obtain a simpler @xmath240-tree than @xmath241 , which contradicts the construction of the unique @xmath240-tree in @xcite .
thus every @xmath242-node is non - empty .
therefore , there is at least one vertex @xmath21 in @xmath248 .
then , the property ( b ) implies that @xmath73\subseteq n[v]$ ] and @xmath249\subseteq n[v]$ ] , and hence @xmath15 and @xmath250 are both in @xmath251 .
therefore , there is a path @xmath252 of length at most two between @xmath72 and @xmath247 via @xmath21 .
when we have @xmath253 or @xmath254 , the path degenerates to the edge @xmath255 .
let @xmath256 be the set of vertices in @xmath104 appearing in the @xmath242-nodes on the ( unique ) path from @xmath248 to the root of the @xmath240-tree . by the definition of @xmath240-tree , we clearly have the following observation .
( recall also that each token must be slid along an edge of @xmath3 . )
[ obs : lca_onetoken ] consider an arbitrary reconfiguration sequence @xmath54 which slides a token @xmath59 from @xmath124 to some vertex @xmath158 .
then , @xmath59 must pass through at least one vertex in @xmath257 , that is , there exists at least one independent set @xmath258 in @xmath54 such that @xmath259 .
we are now ready to give the key lemma for trivially perfect graphs .
[ lem : lca ] let @xmath66 be a target - assignment between @xmath0 and @xmath1 . then
, @xmath67 is proper if and only if the nodes @xmath260 and @xmath261 are not in the ancestor / descendant relationship on @xmath241 for every pair of vertices @xmath262 .
we first show the sufficiency . for a target - assignment @xmath67 between @xmath0 and @xmath1 ,
suppose that the nodes @xmath260 and @xmath261 are not in the ancestor / descendant relationship on @xmath241 for every pair of vertices @xmath262 .
then , we can simply slide the tokens one by one in an arbitrary order ; by observation [ obs : twosteps ] each token @xmath59 , @xmath61 , can be slid along a path from @xmath60 to @xmath263 via a vertex @xmath264 in @xmath260 .
note that there is no token @xmath265 adjacent to @xmath264 , because the nodes @xmath260 and @xmath261 are not in the ancestor / descendant relationship on @xmath241 .
thus , there is a reconfiguration sequence between @xmath0 and @xmath1 according to @xmath67 , and hence @xmath67 is proper .
we then show the necessity .
suppose that @xmath67 is proper , and suppose for a contradiction that there exists a pair of vertices @xmath262 such that the nodes @xmath260 and @xmath261 are in the ancestor / descendant relationship on @xmath241 ; without loss of generality , we assume that @xmath260 is an ancestor of @xmath261 . since @xmath67 is proper , there exists a reconfiguration sequence @xmath54 between @xmath0 and @xmath1 which slides the token @xmath59 from @xmath60 to @xmath263 and also slides the token @xmath265 from @xmath266 to @xmath267 .
by observation [ obs : lca_onetoken ] there is at least one vertex @xmath264 in @xmath268 which is passed through by @xmath59 .
similarly , there is at least one vertex @xmath269 in @xmath270 which is passed through by @xmath265 .
let @xmath271 and @xmath272 be the @xmath242-nodes that contains @xmath264 and @xmath269 , respectively .
since @xmath260 and @xmath261 are in the ancestor / descendant relationship on @xmath241 , so are @xmath271 and @xmath272 .
first suppose @xmath272 is an ancestor of @xmath271 .
then we have @xmath273 \subseteq n[v_{i'}]$ ] , @xmath274\subseteq n[v_{i'}]$ ] and @xmath275\subseteq n[v_{i'}]$ ] . therefore ,
if we slide @xmath59 via @xmath264 , then @xmath59 would be adjacent to the other token @xmath265 which is on one of the three vertices @xmath276 , @xmath267 and @xmath269 .
thus , the token @xmath265 should `` escape '' from @xmath266 before sliding @xmath59 .
however , we can establish the same argument for any descendant of @xmath271 , and hence @xmath265 must escape to some vertex @xmath72 that is contained in an ancestor of @xmath271 at first .
however , the vertex @xmath72 is adjacent to all of @xmath277 , and hence @xmath265 can not escape before sliding @xmath59 .
this contradicts the assumption that @xmath54 slides the token @xmath59 from @xmath60 to @xmath263 and also slides the token @xmath265 from @xmath266 to @xmath267 .
the other case , @xmath271 is an ancestor of @xmath271 , is symmetric .
we now describe our linear - time algorithm for a trivially perfect graph .
let @xmath241 be the @xmath240-tree of a connected trivially perfect graph @xmath7 .
let @xmath58 and @xmath278 be given initial and target independent sets of @xmath3 , respectively .
then , we determine whether @xmath279 as follows : aaa construct some particular target - assignment @xmath280 between @xmath0 and @xmath1 ; and check whether @xmath280 is proper or not , using lemma [ lem : lca ]
. we will show later in lemma [ lem : no_case ] that it suffices to check only @xmath280 in order to determine whether @xmath279 or not . indeed , our linear - time algorithm executes ( a ) and ( b ) above at the same time , in the bottom - up manner based on @xmath241 .
[ [ description - of - the - algorithm ] ] description of the algorithm + + + + + + + + + + + + + + + + + + + + + + + + + + + + remember that the vertex - set associated to each @xmath242-node in @xmath241 induces a clique in @xmath3 .
therefore , for any independent set @xmath62 of @xmath3 , each @xmath242-node contains at most one vertex in @xmath62 , and hence contains at most one token .
we put a `` blue token '' for each @xmath242-node containing a blue vertex in @xmath0 , and also put a `` red token '' for each @xmath242-node containing a red vertex in @xmath1 .
note that a @xmath242-node may contain a pair of blue and red tokens .
our algorithm lifts up the tokens from the leaves to the root of @xmath241 , and if a blue token @xmath281 meets a red token @xmath282 at their least common ancestor @xmath283 in @xmath241 , then we replace them by a single `` green token . ''
this corresponds to setting @xmath284 .
more precisely , at initialization step , the algorithm first collects all leaves of @xmath241 in a queue , which is called _
the algorithm marks the nodes in the frontier , and lifts up each token to its parent @xmath242-node .
each @xmath242-node @xmath175 is put into the frontier if its all children are marked , and then , all children of @xmath175 are removed from the frontier after the following procedure at @xmath175 : aaa @xmath175 contains at most one token : the algorithm has nothing to do .
@xmath175 contains only one pair of blue token @xmath281 and red token @xmath282 : the algorithm replaces them by a single green token , and let @xmath284 .
@xmath175 contains only green tokens : the algorithm replaces them by a single green token .
@xmath175 contains two or more blue tokens , or two or more red tokens : the algorithm outputs `` no '' and halts ( that is , @xmath285 in this case ) .
@xmath175 contains at least one green token and at least one blue or red token : the algorithm outputs `` no '' and halts ( that is , @xmath285 in this case ) . repeating this process , and the algorithm outputs `` yes '' if and only when the frontier contains only the root @xmath242-node @xmath282 of @xmath241 which is in one of cases ( 1)(3 ) above .
[ [ correctness - of - the - algorithm ] ] correctness of the algorithm + + + + + + + + + + + + + + + + + + + + + + + + + + + + it is not difficult to implement our algorithm to run in @xmath36 time and @xmath36 space .
therefore , we here prove the correctness of the algorithm .
we first show that @xmath279 if the algorithm outputs `` yes . '' in this case , the algorithm is in cases ( 1 ) , ( 2 ) , or ( 3 ) at each @xmath242-nodes in @xmath241 ( including the root @xmath282 ) .
then , the target - assignment @xmath280 has been ( completely ) constructed : for each blue vertex @xmath124 , @xmath286 is the red vertex in @xmath1 such that @xmath287 has the minimum height in @xmath241 among all vertices @xmath288 .
then , we have the following lemma . [
lem : yes_case ] if the algorithm outputs `` yes , '' then @xmath279 . by lemma [ lem : lca ] it suffices to show that the target - assignment @xmath280 constructed by the algorithm satisfies that the nodes @xmath289 and @xmath290 are not in the ancestor / descendant relationship on @xmath241 for every pair of vertices @xmath262 .
we first consider the case where a @xmath242-node @xmath175 is in case ( 2 ) .
then , there is exactly one pair of a blue token @xmath281 and a red token @xmath291 , and @xmath292 .
since @xmath281 and @xmath282 did not meet any other tokens before @xmath175 , the subtree @xmath293 of @xmath241 contains only the two tokens @xmath281 and @xmath282 .
therefore , the lemma clearly holds for @xmath293 .
we then consider the case where a @xmath242-node @xmath175 is in case ( 3 ) .
then , two or more least common ancestors of pairs of blue and red tokens meet at this node @xmath175 .
notice that the green tokens were placed on children s node of @xmath175 in the previous step of the algorithm , and hence they were sibling in @xmath241 .
therefore , their corresponding least common ancestors are not in the ancestor / descendant relationship on @xmath241 .
the following lemma completes the correctness proof of our algorithm .
[ lem : no_case ] if the algorithm outputs `` no , '' then @xmath69 .
we assume that the algorithm outputs `` no . ''
then , by lemma [ lem : lca ] , it suffices to show that there is no target - assignment @xmath67 between @xmath0 and @xmath1 such that @xmath260 and @xmath261 are not in the ancestor / descendant relationship on @xmath241 for every pair of vertices @xmath262 .
suppose for a contradiction that the algorithm outputs `` no , '' but there exists a target - assignment @xmath294 between @xmath0 and @xmath1 such that @xmath295 and @xmath296 are not in the ancestor / descendant relationship on @xmath241 for every pair of vertices @xmath262 . then , by lemma [ lem : lca ] , @xmath294 is proper and hence @xmath297 .
since the algorithm outputs `` no , '' there is a @xmath242-node @xmath175 which is in either case ( 4 ) or ( 5 ) .
we first assume that the @xmath242-node @xmath175 is in case ( 4 ) . without loss of generality , at least two blue tokens @xmath298 and @xmath299 meet at this node @xmath175 .
then , the @xmath240-tree @xmath241 contains two red tokens @xmath300 and @xmath301 placed on @xmath302 and @xmath303 , respectively .
notice that , since @xmath298 and @xmath299 did not meet any red token before at the node @xmath175 , both @xmath300 and @xmath301 must be placed on either @xmath175 or nodes in @xmath304 . then , the least common ancestor @xmath305 must be an ancestor of @xmath175 , and so is the least common ancestor @xmath306 .
the nodes @xmath305 and @xmath306 are in the ancestor / descendant relationship , a contradiction .
thus , the algorithm outputs `` no '' because the @xmath242-node @xmath175 is in case ( 5 ) . in this case , without loss of generality , at least one blue token @xmath298 and at least one green token @xmath307 meet at this node @xmath175 . then
, the red token @xmath300 corresponding to @xmath302 must be placed on either @xmath175 or some node in @xmath304 .
therefore , the least common ancestor @xmath305 is an ancestor of @xmath307 .
note that @xmath307 corresponds to the least common ancestor of some pair of blue and red tokens , say @xmath266 and @xmath308 , and @xmath309 is an ancestor of it .
therefore , the nodes @xmath305 and @xmath296 are in the ancestor / descendant relationship on @xmath241 , a contradiction . to complete the proof of theorem [ th :
trivialperfect ] , we finally show that our algorithm in section [ subsec : algo_trivially ] can be modified so that it actually finds a shortest reconfiguration sequence between @xmath0 and @xmath1 .
once we know that @xmath279 holds by the @xmath36-time algorithm in section [ subsec : algo_trivially ] , we run it again with modification that `` green '' tokens are left at the corresponding least common ancestors . as in the proof of lemma [ lem : lca ]
, we can now obtain a reconfiguration sequence @xmath310 between @xmath24 and @xmath311 such that each token @xmath59 , @xmath61 , is slid at most twice .
it is sufficient to output @xmath312 and @xmath313 , and hence the running time of the modified algorithm is proportional to @xmath32 , the number of independent sets in @xmath54 . since @xmath314 and each token @xmath59 , @xmath61 ,
is slid at most twice in @xmath54 , we have @xmath315 , that is , the length @xmath32 of @xmath54 is @xmath36 .
therefore , the modified algorithm also runs in @xmath36 time and @xmath36 space .
notice that each token @xmath59 is slid to its target vertex @xmath286 along a shortest path ( of length at most two ) between @xmath60 and @xmath286 without detour , and hence @xmath54 has the minimum length .
this completes the proof of theorem [ th : trivialperfect ] .
the main result of this section is the following theorem . [
th : caterpillar ] the sliding token problem for a connected caterpillar @xmath7 and two independent sets @xmath0 and @xmath1 of @xmath3 can be solved in @xmath36 time and @xmath36 space . moreover , for a yes - instance , a shortest reconfiguration sequence between them can be output in @xmath38 time and @xmath36 space .
let @xmath316 be a caterpillar with spine @xmath102 which induces the path @xmath105 , and leaf set @xmath103 .
we assume that @xmath317 , @xmath107 , and @xmath108 .
first we show that we can assume that each spine vertex has at most one leaf without loss of generality .
[ lem : cat ] for any given caterpillar @xmath316 and two independent sets @xmath0 and @xmath1 on @xmath3 , there is a linear time reduction from them to another caterpillar @xmath318 and two independent sets @xmath319 and @xmath320 such that ( 1 ) @xmath3 , @xmath0 , and @xmath1 are a yes - instance of the sliding token problem if and only if @xmath321 , @xmath319 , and @xmath1 are a yes - instance of the sliding token problem , ( 2 ) the maximum degree of @xmath321 is at most 3 , and ( 3 ) @xmath322 , where @xmath323 . in other words ,
the sliding token problem on a caterpillar is sufficient to consider only caterpillars of maximum degree 3 .
on @xmath3 , let @xmath129 be any vertex in @xmath102 with @xmath324 .
then there exist at least two leaves @xmath325 and @xmath326 attached to @xmath129 ( note that they are weak twins ) .
now we consider the case that two tokens in @xmath0 are on @xmath325 and @xmath326 .
then , we can not slide these two tokens at all , and any other token can not pass through @xmath129 since it is blocked by them . if @xmath1 contains these two tokens also , we can split the problem into two subproblems by removing @xmath129 and its leaves from @xmath3 , and solve it separately . otherwise , the answer is `` no '' ( remind that the problem is reversible ; that is , if tokens can not be slid , there are no other tokens which slide into the situation ) . therefore ,
if at least two tokens are placed on the leaves of a vertex of the original graph , we can reduce the case in linear time .
thus we assume that every spine vertex with its leaves contains at most one token in @xmath0 and @xmath1 , respectively .
then , by the same reason , we can remove all leaves but one of each spine vertex . more precisely , regardless whether @xmath279 or @xmath69 , at most one leaf for each spine vertex is used for the transitions .
therefore , we can remove all other useless leaves but one from each spine vertex .
especially , removing all useless leaves , we have @xmath322 .
hereafter , we only consider the caterpillars stated in lemma [ lem : cat ] .
that is , for any given caterpillar @xmath316 with spine @xmath105 , we assume that @xmath322 and @xmath327 for each @xmath328 . then , we denote the unique leaf of @xmath129 by @xmath325 if it exists .
we here introduce a key notion of the problem on these caterpillars that is named _ locked _ path .
let @xmath3 and @xmath62 be a caterpillar and an independent set of @xmath3 , respectively .
a path @xmath329 on @xmath3 is _ locked _ by @xmath62 if and only if aaa @xmath34 is odd and greater than 2 , @xmath330 , @xmath331 ( in other words , they are leaves ) , and @xmath332 .
this notion is simplified version of a _ locked _
tree used in @xcite . using the discussion in @xcite
, we obtain the condition for the immovable independent set on a caterpillar : [ th : cat - fix ] let @xmath3 and @xmath62 be a caterpillar and an independent set of @xmath3 , respectively
. then we can not slide any token in @xmath62 on @xmath3 at all if and only if there exist a set of locked paths @xmath333 for some @xmath334 such that @xmath62 is a union of them .
the proof can be found in @xcite , and omitted here .
intuitively , for any caterpillar @xmath3 and its independent set @xmath62 , if @xmath62 contains a locked path @xmath175 , we can not slide any token through the vertices in @xmath175 .
therefore , @xmath175 splits @xmath3 into two subgraphs , and we obtain two completely separated subproblems .
( we note that the endpoints of @xmath175 are leaves with tokens , and their neighbors are spine vertices without tokens .
this property admits us to cut the graph at the spine vertices on the locked path . )
therefore , we obtain the following lemma : [ lem : cat2 ] for any given caterpillar @xmath316 and two independent sets @xmath0 and @xmath1 on @xmath3 , there is a linear time reduction from them to another caterpillar @xmath318 and two independent sets @xmath319 and @xmath320 such that ( 1 ) @xmath3 , @xmath0 , and @xmath1 are a yes - instance of the sliding token problem if and only if @xmath321 , @xmath319 , and @xmath1 are a yes - instance of the sliding token problem , and ( 2 ) both of @xmath319 and @xmath320 contain no locked path . in @xmath3 ,
when @xmath0 contains a locked path @xmath175 , it should be appear in @xmath1 ; otherwise , the answer is no .
therefore , we can remove all vertices in @xmath175 and obtain the new graph @xmath335 with two independent sets @xmath336 and @xmath337 such that @xmath3 with @xmath0 and @xmath1 is a yes - instance if and only if @xmath335 with @xmath338 and @xmath339 is a yes - instance . repeating this process , we obtain disconnected caterpillar @xmath340 and two independent sets @xmath341 and @xmath342 such that both of @xmath341 and @xmath342 contain no locked paths . on a disconnected graph , we can solve the problem separately for each connected component .
therefore , we can assume that the graph is connected , which completes the proof .
hereafter , without loss of generality , we assume that the caterpillar @xmath3 with two independent sets @xmath0 and @xmath1 satisfies the conditions in lemmas [ lem : cat ] and [ lem : cat2 ] . that is , each spine vertex @xmath129 has at most one leaf @xmath325 , @xmath109 and @xmath110 have one leaf @xmath343 and @xmath344 , respectively , both of @xmath0 and @xmath1 contain no locked path , and @xmath2 . by the result in @xcite ,
this is a yes - instance .
thus , it is sufficient to show an @xmath38 time algorithm that computes a shortest reconfiguration sequence between @xmath0 and @xmath1 .
each pair @xmath345 can have at most one token .
therefore , without loss of generality , we can assume that the blue vertices @xmath120 in @xmath0 are labeled from left to right ( according to the order @xmath346 , @xmath347 , @xmath31 , @xmath348 of @xmath3 ) , that is , @xmath121 if @xmath122 ; similarly , the red vertices @xmath123 are also labeled from left to right .
then , we define a target - assignment @xmath66 , as follows : for each blue vertex @xmath124 @xmath349 to prove theorem [ th : caterpillar ] , it suffices to show that @xmath67 is proper , and we can slide tokens with fewest detours . here , any token can not bypass the other token since each token is on a leaf or spine vertex .
thus , by the results in @xcite , it has been shown that @xmath67 is proper .
we show that we can compute a shortest reconfiguration in case analysis .
now we introduce _ direction _ of a token @xmath350 denoted by @xmath351 as follows : when @xmath350 slides from @xmath352 in @xmath0 to @xmath353 in @xmath1 with @xmath354 , the direction of @xmath350 is said to be _
r _ and denoted by @xmath355 . if @xmath356 , it is said to be _ l _ and denoted by @xmath357 . if @xmath358 , the direction of @xmath350 is said to be _
c _ and denoted by @xmath359 . has to precede the most left l token @xmath360.,scaledwidth=80.0% ] we first consider a simple case : all directions are either r or l. in this case , we can use the same idea appearing in the algorithm for a proper interval graph in section [ sec : proper ] .
we can introduce a partial order over the tokens , and slide them straightforwardly using the same idea in section [ subsec : algo_proper ] .
intuitively , a sequence of r tokens are slid from left to right , and a sequence of l tokens are slid from right to left , and we can define a partial order over the sequences of different directions . the only additional considerable case is shown in [ fig : rl ] .
that is , when the token @xmath361 slides to @xmath325 from left and the other token @xmath360 slides to @xmath362 from right , @xmath361 should precede @xmath360 .
it is not difficult to see that this ( and its symmetric case ) is the only exception than the algorithm in section [ subsec : algo_proper ] when all tokens slide to right or left . in other words , in this case , detour is required , and unavoidable .
we next suppose that @xmath0 ( and hence @xmath1 ) contains some token @xmath350 with @xmath359 . in other words ,
@xmath350 is put on @xmath129 or @xmath325 for some @xmath12 in both of @xmath0 and @xmath1 .
we have five cases .
bbb @xmath350 is put on @xmath325 in @xmath0 and @xmath1 . in this case
, we have nothing to do ; @xmath350 does not need to be slid .
@xmath350 is put on @xmath129 in @xmath0 and slid to @xmath325 in @xmath1 . in this case , we first slide it from @xmath129 to @xmath325 , and do nothing any more . then no detour is needed for @xmath350 .
@xmath350 is put on @xmath325 in @xmath0 and slid to @xmath129 in @xmath1 . in this case , we lastly slide it from @xmath325 to @xmath129 , and no detour is needed for @xmath350 again . @xmath350 is put on @xmath129 in @xmath0 and @xmath1 , and @xmath325 exists . using a simple induction by the number of tokens , we can determine if @xmath350 should make a detour or not in linear time .
if not , we never slide @xmath350 . otherwise , we first slide @xmath350 to @xmath325 , and lastly slide back from @xmath325 to @xmath129 .
it is clear that the length of detour with respect to @xmath350 is as few as possible .
@xmath350 is put on @xmath129 in @xmath0 and @xmath1 , and @xmath325 does not exist . by assumption , @xmath363 ( since @xmath343 and @xmath344 exist ) . without loss of generality ,
we suppose @xmath350 is the leftmost spine vertex having the condition .
we first observe that @xmath364 is at most 1 .
clearly , we have no token on @xmath365 and @xmath362 .
when we have two tokens on @xmath366 and @xmath367 , the path @xmath368 is a locked path , which contradicts the assumption .
we also have @xmath369 by the same argument .
now we consider the most serious case since the other cases are simpler and easier than this case .
the most serious case is that a blue token on @xmath366 and a red token on @xmath367 .
since any token can not bypass the other , @xmath0 contains an l token on @xmath366 , and @xmath1 contains an l token on @xmath367 . in this case , by the l token on @xmath366 , first , @xmath350 should make a detour to right , and by the l token in @xmath1 , @xmath350 next should make a detour to left twice after the first detour .
it is clear that this three slides should not be avoided , and this ordering of three slides can not be violated .
therefore , @xmath350 itself should slide at least four times to return to the original position , and @xmath350 can done it in four slides . during this slides , since @xmath350 is the leftmost spine with this condition , the tokens on @xmath370 do not make any detours .
thus we focus on the tokens on @xmath371 .
let @xmath372 be the token that should be on @xmath367 in @xmath1 .
since @xmath350 is on @xmath129 , @xmath372 is not on @xmath373 .
if @xmath372 is on one of @xmath374 in @xmath0 , we have nothing to do ; just make a detour for only @xmath350 .
the problem occurs when @xmath372 is on @xmath375 in @xmath0 .
if there exists @xmath376 , we first slide @xmath372 to it , and this detour for @xmath372 is unavoidable . if @xmath376 does not exist , we have to slide @xmath372 to @xmath377 before slide of @xmath350 .
this can be done immediately except the only considerable case ; when we have another l or s token @xmath378 on @xmath377 .
we can repeat this process recursively and confirm that each detour is unavoidable . since @xmath3 with @xmath0 and @xmath1 contains no locked path , this process will halts .
( more precisely , this process will be stuck if and only if this sequence of tokens forms a locked path on @xmath3 , which contradicts the assumption . ) therefore , traversing this process , we can construct the shortest reconfiguration sequence .
* proof of theorem [ th : caterpillar ] .
* for a given independent set @xmath0 on a caterpillar @xmath27 , we can check if each vertex is a part of locked path as follows in @xmath36 time ( which is much simpler than the algorithm in @xcite ) : aaa initialize a state @xmath102 by `` not locked path '' .
for @xmath379 , check @xmath129 and @xmath325 .
we here denote their states by @xmath380 , where @xmath381 , @xmath382 such that @xmath383 means `` token is placed on the vertex '' , @xmath143 means `` no token is placed on the vertex '' , and @xmath384 means `` the leaf does not exist . '' in each case , update the state @xmath102 as follows : bbb if @xmath102 is `` not locked path , '' set @xmath102 by `` locked path ? , '' and remember @xmath12 as a potential left endpoint of a locked path . if @xmath102 is `` locked path '' , @xmath345 is a part of locked path .
therefore , mark all vertices between the previously remembered left endpoint to this endpoint as `` locked path '' . after that , set @xmath102 by `` locked path ? '' again , and remember @xmath12 as a potential left endpoint of the next locked path .
if there is a token on @xmath365 , nothing to do . if there is no token on @xmath365 , reset @xmath102 by `` not locked path '' ( regardless of the previous state of @xmath102 ) .
reset @xmath102 by `` not locked path '' ( regardless of the previous state of @xmath102 ) .
nothing to do .
simple case analysis shows that after the procedure above , every vertex in a locked path is marked in @xmath36 time .
thus , we first run this procedure twice for @xmath385 and @xmath386 in @xmath36 time , and check whether the marked vertices coincide with each other . if not , the algorithm outputs `` no '' .
otherwise , the algorithm splits the caterpillar @xmath3 into subgraphs @xmath387 induced by only unmarked vertices .
then we can solve the problem for each subgraph ; we note that two endpoints of which tokens are placed of a locked path @xmath175 are leaves .
that is , for example , when a locked path @xmath329 splits @xmath3 into @xmath388 and @xmath389 , the neighbors of @xmath388 and @xmath389 in @xmath175 are @xmath390 and @xmath391 , and there are no token on them .
thus , in the case , we can solve the problem on @xmath388 and @xmath389 separately , and we do not need to consider their neighbors . for each subgraph @xmath392 , the algorithm next checks whether each subgraph contains the same number of blue and red tokens .
if they do not coincide with each other , the algorithm output `` no.''otherwise , we have a yes - instance .
the correctness of the algorithm so far follows from theorem [ th : cat - fix ] with results in @xcite immediately .
it is also easy to implement the algorithm to run in @xmath36 time and space .
it is not difficult to modify the algorithm to output the sequence itself based on the previous case analysis . for each token
, the number of detours made by the token is bounded above by @xmath36 , the number of slides of the token itself is also bounded above by @xmath36 , and the computation for the token can be done in @xmath36 time .
therefore , the algorithm runs in @xmath38 time , and the length of the sequence is @xmath38 .
( as shownn in the last paragraph in section [ sec : proper ] , there exist instances that require a shortest sequence of length @xmath239 . )
in this paper , we showed that the shortest sliding token problem can be solved in polynomial time for three subclasses of interval graphs .
the computational complexity of the problem for chordal graphs , interval graphs , and trees are still open .
especially , tree seems to be the next target .
we can decide if two independent sets are reconfigurable in linear time @xcite , then can we find a shortest sequence for a yes - instance in polynomial time ? as in the 15-puzzle ,
finding a shortest sequence can be np - hard . for a tree
, we do not know that the length can be bounded by any polynomial or not .
it is an interesting open question whether there is any instance on some graph classes whose reconfiguration sequence requires super - polynomial length .
demaine , e.d . ,
demaine , m.l . , fox - epstein , e. , hoang , d.a . , ito t. , ono , h. , otachi , y. , uehara , r. , yamada , t. : linear - time algorithm for sliding tokens on trees .
theoretical computer science 600 , pp .
132142 ( 2015 ) hearn , r.a .
, demaine , e.d .
: pspace - completeness of sliding - block puzzles and other problems through the nondeterministic constraint logic model of computation .
theoretical computer science 343 , pp .
7296 ( 2005 ) mouawad , a.e . , nishimura , n. , pathak , v. , raman , v. : shortest reconfiguration paths in the solution space of boolean formulas . in proc . of icalp 2015 , lncs 9134 , pp . 985996 ( 2015 ) mouawad , a.e .
, nishimura , n. , raman , v. , simjour , n. , suzuki , a. : on the parameterized complexity of reconfiguration problems . in proc . of ipec 2013 , lncs 8296 , pp . 281294 ( 2013 ) mouawad , a.e . , nishimura , n. , raman , v. , wrochna , m. : reconfiguration over tree decompositions . in proc . of ipec 2014 , lncs 8894 , pp . 246257 ( 2014 ) | suppose that we are given two independent sets @xmath0 and @xmath1 of a graph such that @xmath2 , and imagine that a token is placed on each vertex in @xmath0 .
then , the sliding token problem is to determine whether there exists a sequence of independent sets which transforms @xmath0 into @xmath1 so that each independent set in the sequence results from the previous one by sliding exactly one token along an edge in the graph .
the sliding token problem is one of the reconfiguration problems that attract the attention from the viewpoint of theoretical computer science .
the reconfiguration problems tend to be pspace - complete in general , and some polynomial time algorithms are shown in restricted cases .
recently , the problems that aim at finding a shortest reconfiguration sequence are investigated . for the 3sat problem
, a trichotomy for the complexity of finding the shortest sequence has been shown ; that is , it is in p , np - complete , or pspace - complete in certain conditions . in general , even if it is polynomial time solvable to decide whether two instances are reconfigured with each other , it can be np - complete to find a shortest sequence between them .
namely , finding a shortest sequence between two independent sets can be more difficult than the decision problem of reconfigurability between them . in this paper
, we show that the problem for finding a shortest sequence between two independent sets is polynomial time solvable for some graph classes which are subclasses of the class of interval graphs . more precisely
, we can find a shortest sequence between two independent sets on a graph @xmath3 in polynomial time if either @xmath3 is a proper interval graph , a trivially perfect graph , or a caterpillar .
as far as the authors know , this is the first polynomial time algorithm for the shortest sliding token problem for a graph class that requires detours . |
much of our current understanding of incompressible magnetohydrodynamic ( mhd ) turbulence has its roots in the pioneering work of iroshnikov ( 1963 ) and kraichnan ( 1965 ) .
these studies emphasized the important fact that alfvn waves travelling in the same direction along a background magnetic field do not interact with one another and explained how one can think of the cascade of energy to small scales as resulting from collisions between oppositely directed alfvn wave packets .
they also argued that in the absence of a mean magnetic field , the magnetic field of the energy - containing eddies at scale @xmath7 affects fluctuations on scales @xmath8 much in the same way as would a truly uniform mean magnetic field .
another foundation of our current understanding is the finding that mhd turbulence is inherently anisotropic .
montgomery & turner ( 1981 ) and shebalin , matthaeus , & montgomery ( 1983 ) showed that a strong uniform mean magnetic field @xmath0 inhibits the cascade of energy to small scales measured in the direction parallel to @xmath0 .
this early work was substantially elaborated upon by higdon ( 1984 ) , goldreich & sridhar ( 1995 , 1997 ) , montgomery & matthaeus ( 1995 ) , ng & bhattacharjee ( 1996 , 1997 ) , galtier et al ( 2000 ) , cho & lazarian ( 2003 ) , oughton et al ( 2006 ) , and many others .
for example , cho & vishniac ( 2000 ) used numerical simulations to show that when the fluctuating magnetic field @xmath9 is @xmath10 the small - scale turbulent eddies become elongated along the local magnetic field direction .
goldreich & sridhar ( 1995 ) introduced the important and influential idea of `` critical balance , '' which holds that at each scale the linear wave period for the bulk of the fluctuation energy is comparable to the time for the fluctuation energy to cascade to smaller scales .
goldreich & sridhar ( 1995 , 1997 ) , maron & goldreich ( 2001 ) , and lithwick & goldreich ( 2001 ) clarified a number of important physical processes in anisotropic mhd turbulence and used the concept of critical balance to determine the ratio of the dimensions of turbulent eddies in the directions parallel and perpendicular to the local magnetic field .
over the last several years , research on mhd turbulence has been proceeding along several different lines .
for example , one group of studies has attempted to determine the power spectrum , intermittency , and anisotropy of strong incompressible mhd turbulence using direct numerical simulations .
( see , e.g. , cho & vishniac 2000 , mller & biskamp 2000 , maron & goldreich 2001 , cho et al 2002 , haugen et al 2004 , muller & grappin 2005 , mininni & pouquet 2007 , perez & boldyrev 2008 ) .
another series of papers has explored the properties of anisotropic turbulence in weakly collisional magnetized plasmas using gyrokinetics , a low - frequency expansion of the vlasov equation that averages over the gyromotion of the particles .
( howes et al 2006 , 2007a , 2007b ; schekochihin et al 2007 ) .
these authors investigated the transition between the alfvn - wave cascade and a kinetic - alfvn - wave cascade at length scales of order the proton gyroradius @xmath11 , as well as the physics of energy dissipation and entropy production in the low - collisionality regime .
turbulence at scales @xmath12 has also been examined both numerically and analytically within the framework of fluid models , in particular hall mhd and electron mhd .
( biskamp , schwarz , & drake 1996 , biskamp et al 1999 , matthaeus et al 2003 ; galtier & bhattacharjee 2003 , 2005 ; cho & lazarian 2004 ; brodin et al 2006 , shukla et al 2006 ) .
another group of studies has investigated the power spectrum , intermittency , and decay time of compressible mhd turbulence .
( oughton et al 1995 , stone et al 1998 , lithwick & goldreich 2001 , boldyrev et al 2002 , padoan et al 2004 , elmegreen & scalo 2004 ) . additional work by kuznetsov ( 2001 ) , cho & lazarian ( 2002 , 2003 ) , chandran ( 2005 ) , and luo & melrose ( 2006 ) has begun to address the way in which alfvn waves , fast magnetosonic waves , and slow magnetosonic waves interact in compressible weak mhd turbulence .
another recent development is the finding that strong incompressible mhd turbulence leads to alternating patches of alignment and anti - alignment between the velocity and magnetic - field fluctuations .
( boldyrev 2005 , 2006 ; beresnyak & lazarian 2006 ; mason , cattaneo , & boldyrev 2006 ; matthaeus et al 2007 ) these studies examined how the degree of local alignment ( and anti - alignment ) depends upon length scale , as well as the effects of alignment upon the energy cascade time and the power spectrum of the turbulence .
the topic addressed in this paper is the role of cross helicity in incompressible mhd turbulence .
the cross helicity is defined as @xmath13 where @xmath14 is the velocity and @xmath15 is the magnetic field .
the cross helicity is conserved in the absence of dissipation and can be thought of as a measure of the linkages between lines of vorticity and magnetic field lines , both of which are frozen to the fluid flow in the absence of dissipation ( moffatt 1978 ) . in the presence of a background magnetic field , @xmath16 ,
the cross helicity is also a measure of the difference between the energy of fluctuations travelling in the @xmath17 and @xmath18 directions .
dobrowolny , mangeney , & veltri ( 1980 ) showed that mhd turbulence with cross helicity decays to a maximally aligned state , with @xmath19 , where @xmath20 and @xmath21 are the fluctuating velocity and magnetic field and @xmath22 is the mass density .
different decay rates for the energy and cross helicity were also demonstrated by matthaeus & montgomery ( 1980 ) . in another early study , grappin , pouquet , &
lorat ( 1983 ) used a statistical closure , the eddy - damped quasi - normal markovian ( edqnm ) approximation , to study strong 3d incompressible mhd turbulence with cross helicity , assuming isotropic power spectra .
they found that when @xmath23 , the total energy spectrum is steeper than the isotropic iroshnikov - kraichnan @xmath24 spectrum .
pouquet et al ( 1988 ) found a similar result in direct numerical simulations of 2d incompressible mhd turbulence .
more recently , lithwick , goldreich , & sridhar ( 2007 ) and beresnyak & lazarian ( 2007 ) addressed the role of cross helicity in strong mhd turbulence taking into account the effects of anisotropy .
this paper presents a new phenomenology for strong , anisotropic , incompressible mhd turbulence with nonzero cross helicity , and is organized as follows .
section [ sec : wpc ] presents some relevant theoretical background .
section [ sec : theory ] introduces the new phenomenology as well as two nonlinear advection - diffusion equations that model the time evolution of the power spectra . analytic and numerical solutions to this equation in the
weak - turbulence and strong - turbulence regimes are presented in sections [ sec : weak ] and [ sec : strong ] .
section [ sec : strong ] also presents a simple phenomenological derivation of the power spectra and anisotropy in strong mhd turbulence .
section [ sec : trans ] presents a numerical solution to the advection - diffusion equation that shows the smooth transition between the weak and strong turbulence regimes .
section [ sec : unequal ] addresses the case in which the parallel correlation lengths of waves propagating in opposite directions along the background magnetic field are unequal at the outer scale . in section [ sec : solarwind ] , the proposed phenomenology is applied to turbulence in the solar wind and solar corona , and in section [ sec : comp ] the results of this work are compared to the recent studies of lithwick , goldreich , & sridhar ( 2007 ) and beresnyak & lazarian ( 2007 ) .
the equations of ideal incompressible mhd can be written @xmath25 where @xmath26 are the elsasser variables , @xmath14 is the fluid velocity , @xmath21 is the magnetic field fluctuation , @xmath22 is the mass density , which is taken to be uniform and constant , @xmath27 is the alfvn speed , @xmath16 is the mean magnetic field , and @xmath28 , which is determined by the incompressibility condition , @xmath29 . throughout this paper
it is assumed that @xmath30 and @xmath31 . in the limit of small - amplitude fluctuations ( @xmath32 ) ,
the nonlinear term @xmath33 in equation ( [ eq : mhd ] ) can be neglected to a first approximation , and the curl of equation ( [ eq : mhd ] ) becomes @xmath34 which is solved by setting @xmath35 equal to an arbitrary function of @xmath36 .
thus , @xmath37 represents fluctuations with @xmath38 that propagate in @xmath39 direction at speed @xmath40 in the absence of nonlinear interactions . in the absence of an average velocity ,
the cross helicity defined in equation ( [ eq : defc ] ) can be rewritten as @xmath41 .
\label{eq : defc2}\ ] ] the cross helicity is thus proportional to the difference in energy between fluctuations propagating in the @xmath17 and @xmath42 directions .
equation ( [ eq : mhd ] ) shows that the nonlinear term is nonzero only at those locations where both @xmath43 and @xmath44 are nonzero .
nonlinear interactions can thus be thought of as collisions between oppositely directed wave packets ( kraichnan 1965 ) .
when both @xmath43 and @xmath44 are nonzero , equation ( [ eq : mhd ] ) indicates that the @xmath37 fluctuations are advected not at the uniform velocity @xmath45 , but rather at the non - uniform velocity @xmath46 .
maron & goldreich ( 2001 ) elaborated upon this idea by showing that to lowest order in fluctuation amplitude , if one neglects the pressure term , then @xmath43 wave packets are advected along the hypothetical magnetic field lines corresponding to the sum of @xmath0 and the part of @xmath47 arising from the @xmath44 fluctuations .
this result can be used to construct a geometrical picture for how wave - packet collisions cause energy to cascade to smaller scales , as depicted in figure [ fig : f1 ] . in this figure , two oppositely
directed wave packets of dimension @xmath48 in the plane perpendicular to @xmath0 and length @xmath49 along @xmath0 pass through one another and get sheared . collisions between wavepackets of similar @xmath50 are usually the dominant mechanism for transferring energy from large scales to small scales .
the duration of the collision illustrated in the figure is approximately @xmath51 .
the fluctuating velocity and magnetic field are taken to be in the plane perpendicular to @xmath0 , as is the case for linear shear alfvn waves .
the magnitude of the nonlinear term in equation ( [ eq : mhd ] ) is then @xmath52 , where @xmath53 is the rms amplitude of the @xmath54 wave packet .
the fractional change in the @xmath14 and @xmath55 fields of the @xmath56 wave packet induced by the collision is then roughly @xmath57 , width=288 ] if this fractional change is @xmath58 for both @xmath43 and @xmath44 fluctuations then neither wave packet is altered significantly by a single collision , and the turbulence is weak .
wave packets travel a distance @xmath59 before being significantly distorted , and the fluctuations can thus be viewed as linear waves that are only weakly perturbed by nonlinear interactions with other waves . in the wave - packet collision depicted in figure
[ fig : f1 ] , the right - hand side of the @xmath60 wave packet is altered by the collision in almost the same way as the left - hand side , since both sides encounter essentially the same @xmath43 wave packet , since the @xmath43 packet is changed only slightly during the collision .
changes to the profile of a wave packet along the magnetic field are thus weaker than changes in the profile of a wave packet in the plane perpendicular to @xmath0 ( shebalin et al 1983 , ng & bhattacharjee 1997 , goldreich & sridhar 1997 , bhattacharjee & ng 2001 , perez & boldyrev 2008 ) . as a result , in the weak - turbulence limit , the cascade of energy to small @xmath49 is much less efficient than the cascade of energy to small @xmath50 ( galtier et al 2000 ) . on the other hand , if the fractional change in equation ( [ eq : defchi1 ] ) is of order unity then a @xmath56 wave packet is distorted substantially during a single collision , and the turbulence is said to be `` strong . '' in the case that the fractional change in equation ( [ eq : defchi1 ] ) is @xmath61 for one fluctuation type , ( e.g. , @xmath44 ) but @xmath58 for the other ( @xmath43 ) , the turbulence is still referred to as strong . it should be noted that strong turbulence can arise when @xmath62 , provided that @xmath63 . in strong turbulence energy cascades to smaller @xmath49 to a greater extent than in weak turbulence , but the primary direction of energy flow in @xmath64-space is still to larger @xmath5 , as discussed in the next section .
in order to develop an analytical model , it is convenient to work in terms of the fourier transforms of the fluctuating @xmath54 fields , given by @xmath65 the three - dimensional power spectrum @xmath66 is defined by the equation @xmath67 where @xmath68 denotes an ensemble average .
cylindrical symmetry about @xmath0 is assumed , so that @xmath69 , where @xmath5 and @xmath70 are the components of @xmath71 perpendicular and parallel to @xmath0 .
the mean - square velocity associated with @xmath54 fluctuations is then @xmath72 it is assumed that at each value of @xmath5 there is a parallel wave number @xmath73 such that ( 1 ) the bulk of the @xmath54 fluctuation energy is at @xmath74 and ( 2 ) @xmath75 depends only weakly on @xmath70 for @xmath74 .
a @xmath54 wavepacket at perpendicular scale @xmath76 then has a correlation length in the direction of the mean field of @xmath77 .
the rms amplitude of the fluctuating elsasser fields at a perpendicular scale @xmath76 , denoted @xmath78 , is given by @xmath79 as described in the section [ sec : wpc ] , when a @xmath56 wave packet at scale @xmath76 collides with a @xmath54 wave packet at scale @xmath76 , the fractional change in the @xmath56 packet resulting from the collision is approximately @xmath80 the wave number @xmath81 is defined to be the value of @xmath82 for which @xmath83 .
thus , @xmath84 when @xmath85 , the value of @xmath86 is @xmath58 and a @xmath43 is only weakly affected by a single collision with a @xmath44 wave packet .
each such collision requires a time @xmath87 .
the effects of successive collisions add incoherently , and thus @xmath88 collisions are required for the @xmath43 wave packet to be strongly distorted , and for its energy to pass to smaller scales .
the cascade time @xmath89 for a @xmath43 wave packet at perpendicular scale @xmath76 is thus roughly @xmath90 similarly , if @xmath91 , then @xmath92 .
when @xmath93 , the value of @xmath94 is @xmath61 , a @xmath43 is strongly distorted during a single wave packet collision , and the turbulence is strong .
each such collision takes a time @xmath95 . since @xmath96 , @xmath97 similarly , if @xmath98 , then @xmath99 . the case
@xmath100 ( i.e. , @xmath101 ) is explicitly excluded from the discussion .
initial conditions could in principle be set up in which @xmath100 .
however , the cascade mechanisms described in section [ sec : parcasc ] will not produce the condition @xmath100 if it is not initially present .
it should be emphasized that in both weak turbulence and strong turbulence , the cascade time is given by the same formula , @xmath102 , which involves the @xmath103 spectrum evaluated only at @xmath104 . the two basic mechanisms for transferring fluctuation energy to larger @xmath70 were identified by lithwick , goldreich , & sridhar ( 2007 ) .
the first of these can be called `` propagation with distortion . ''
suppose a @xmath43 wave packet of perpendicular scale @xmath76 and arbitrarily large initial parallel correlation length begins colliding at @xmath105 with a stream of @xmath44 wave packets of similar perpendicular scale . at time @xmath106 , the leading edge of the @xmath43 wave packet has been distorted substantially by the stream of @xmath44 wave packets , but the trailing portion of the @xmath43 wave packet at distances @xmath107 behind the leading edge has not yet encountered the stream of @xmath44 wave packets . if the parallel correlation length of the @xmath43 wave packet is initially @xmath108 , then during a time @xmath89 the @xmath43 wave packet acquires a spatial variation in the direction of the background magnetic field of length scale @xmath109 .
this process is modeled as diffusion of @xmath54 fluctuation energy in the @xmath70 direction with diffusion coefficient @xmath110 , where @xmath111 and @xmath112 . ``
propagation with distortion '' then leads to a value of @xmath113 of @xmath114 .
the second mechanism identified by lithwick , goldreich , & sridhar ( 2007 ) can be called `` uncorrelated cascade . ''
consider a @xmath43 wave packet of perpendicular scale @xmath76 and arbitrarily large parallel correlation length , and consider two points within the wave packet , @xmath115 and @xmath116 , that move with the wave packet at velocity @xmath117 and are separated by a distance along @xmath0 of @xmath118 .
the @xmath44 wave packets at perpendicular scale @xmath76 encountered by the portions of the @xmath43 wave packet at @xmath115 and @xmath116 are then uncorrelated , because @xmath44 wave packets are substantially distorted while propagating between @xmath115 and @xmath116 .
thus , the way in which the @xmath43 wave packet cascades at location @xmath115 is not correlated with the way in which the @xmath43 wave packet cascades at location @xmath116 .
if the parallel correlation length of the @xmath43 wave packet is initially @xmath119 , then wave - packet collisions introduce a spatial variation along @xmath0 into the @xmath43 wave packet of length scale @xmath120 during a time @xmath89 .
again , we model this as diffusion of @xmath54 fluctuation energy in the @xmath70 direction with @xmath110 and @xmath121 , but now @xmath122
. `` uncorrelated cascade '' thus leads to a @xmath70-diffusion coefficient of @xmath123 .
accounting for both mechanisms , one can write @xmath124 where @xmath125 is the larger of @xmath126 and @xmath127 . if @xmath128 , then @xmath43 energy diffuses in @xmath70 primarily through the `` uncorrelated cascade '' mechanism , while @xmath44 energy diffuses in @xmath70 primarily through the `` propagation with distortion cascade '' mechanism .
the phenomenology described in the preceding sections is encapsulated by the following nonlinear advection - diffusion equation , @xmath129 where @xmath130 is shorthand for @xmath131 , @xmath132 and @xmath133 are dimensionless constants of order unity , and @xmath134 and @xmath135 are forcing and damping terms , respectively . the first term on the right - hand side of equation ( [ eq : fppm ] ) represents advection of fluctuation energy to larger @xmath5 , while the second term represents diffusion of fluctuation energy to larger @xmath136 .
the quantity @xmath137 is an effective cascade time at perpendicular scale @xmath76 .
usually , the transfer of energy to small scales is dominated by local interactions in @xmath64-space , and the cascade time for a @xmath43 wave packet is @xmath138 . in some cases , however , the shearing of small - scale wave packets by much larger - scale wave packets can become important . to account for such cases , the effective cascade time is taken to be @xmath139 \mbox { \hspace{0.3 cm } for $ 0<q_\perp < k_\perp$ , } \label{eq : taueff}\ ] ] i.e.
, @xmath140 is the maximum value of @xmath141 for all perpendicular wave numbers between zero and @xmath5 .
the flux of @xmath54 energy to larger @xmath5 is @xmath142 the term @xmath143 is given by @xmath144 , \label{eq : defh}\ ] ] and is included so that @xmath145 increases as the @xmath146 spectrum becomes a more steeply declining function of @xmath5 , in accordance with weak turbulence theory ( galtier et al 2000 , lithwick & goldreich 2003 ) . to match the energy flux in weak turbulence theory in the limit of zero cross helicity , one must set in equation ( [ eq : valc1 ] ) is a factor of 2 larger than the value that follows from the results of galtier et al ( 2000 ) .
it appears that this discrepancy results from the omission of a factor of 2 in equation ( 54 ) of galtier et al ( 2000 ) .
this can be seen by starting from equation ( 46 ) of galtier et al ( 2000 ) and using the expression on page 1045 of leith & kraichnan ( 1972 ) to simplify polar integrals of the form @xmath147 for two - dimensional wave vectors @xmath71 , @xmath148 , and @xmath149 , where @xmath150 is a function only of the wave - vector magnitudes and the integral is over all values of @xmath148 and @xmath149 . ] @xmath151 where @xmath152^{1/2 } ( 1+xy)^2 [ 8-(x+y)^3 ] \,\ln[(x+y)/2 ] } { ( x^2 - y^2)^4 } \simeq -1.87 .
\label{eq : defj}\ ] ] for simplicity , @xmath153
this section addresses weak turbulence in which @xmath154 at the outer scale .
the weak - turbulence condition , @xmath155 , is equivalent to the condition @xmath156 . because @xmath70-diffusion involves a @xmath157 during a time @xmath158 ,
the @xmath70-increment over which energy diffuses while cascading to larger @xmath5 is much less than the breadth of the spectrum in the @xmath70 direction ( @xmath159 ) , so the @xmath70-diffusion terms can be ignored to a good approximation . in this case , equation ( [ eq : fppm ] ) possesses a steady - state solution in which @xmath160 and @xmath161 are constant , and in which @xmath162 where @xmath163 and @xmath164 are arbitrary functions of @xmath70 , and where @xmath165 with @xmath166 .
equations ( [ eq : ssw1 ] ) and ( [ eq : npnm ] ) match the results of weak turbulence theory for incompressible mhd turbulence if one allows only for three - wave interactions among shear alfvn waves ( galtier & chandran 2006 ) , or if one considers only the limit that @xmath167 ( galtier et al 2002 ) .
if one writes @xmath168 with @xmath169 and sets @xmath170 , then equation ( [ eq : epsilon ] ) gives @xmath171 in the limit @xmath172 , @xmath173 , in agreement with the weak - turbulence - theory result for @xmath174 ( lithwick & goldreich 2003 ) , as in the weak - turbulence advection - diffusion model of lithwick & goldreich ( 2003 ) . in steady state , @xmath175 and @xmath176
are forced to be equal at the dissipation scale so that @xmath177 .
this phenomenon of `` pinning '' was discovered by grappin et al ( 1983 ) for strong mhd turbulence , and further elaborated upon by lithwick & goldreich ( 2003 ) for the case of weak turbulence .
the dominant fluctuation type then has the steeper spectrum .
if @xmath178 is fixed , then the ratio @xmath179 of the rms amplitudes of the two fluctuation types at the outer scale @xmath180 increases as @xmath181 increases , where @xmath182 is the dissipation wave number .
alternatively , if @xmath179 is fixed , then @xmath178 approaches unity as @xmath183 .
several of these results are illustrated by the numerical solution to equation ( [ eq : fppm ] ) shown in figure [ fig : f2 ] .
this solution is obtained using a logarithmic grid for @xmath5 , with @xmath184 for @xmath185 .
similarly , @xmath186 for @xmath187 , but @xmath188 for @xmath189 .
@xmath130 is advanced forward in time using a semi - implicit algorithm , in which the terms @xmath190 , @xmath191 , and @xmath81 on the right - hand side of equation ( [ eq : fppm ] ) are evaluated at the beginning of the time step , and the @xmath146 terms on the right - hand side of equation ( [ eq : fppm ] ) are evaluated at the end of the time step .
the algorithm employs operator splitting , treating the @xmath5-advection , forcing , and damping in one stage , and the @xmath70-diffusion is a second stage . in this approach ,
the matrix that has to be inverted to execute each semi - implicit time step is tri - diagonal .
an advantage of this procedure over a fully explicit method is that the time step is not limited by the @xmath70-diffusion time at large @xmath5 and small @xmath70 .
the discretized equations are written in terms of the energy fluxes between neighboring cells , so that in the absence of forcing and dissipation the algorithm conserves fluctuation energy to machine accuracy .
for the numerical solution plotted in figure [ fig : f2 ] , @xmath192 , @xmath193 , @xmath194 , @xmath195 , @xmath196 , @xmath197 , and @xmath198 , where @xmath199 is an effective viscosity .
the initial spectra are set equal to zero , and the equations are integrated forward in time until a steady state is reached . in steady state , @xmath200 and @xmath201 . ) in the weak - turbulence limit .
_ left panel : _ the dimensionless 1d power spectrum defined in equation ( [ eq : defe ] ) . _
middle panel : _ the spectral slopes at @xmath104 .
_ right panel : _ the weighted value of @xmath70 defined in equation ( [ eq : kparw ] ) . in all panels ,
the solid lines refer to @xmath43 and the dashed lines refer to @xmath44 . in the right - hand panel ,
the two lines are almost on top of each other .
[ fig : f2],title="fig:",width=201 ] ) in the weak - turbulence limit .
_ left panel : _ the dimensionless 1d power spectrum defined in equation ( [ eq : defe ] ) . _
middle panel : _ the spectral slopes at @xmath104 .
_ right panel : _ the weighted value of @xmath70 defined in equation ( [ eq : kparw ] ) . in all panels ,
the solid lines refer to @xmath43 and the dashed lines refer to @xmath44 . in the right - hand panel ,
the two lines are almost on top of each other .
[ fig : f2],title="fig:",width=201 ] ) in the weak - turbulence limit .
_ left panel : _ the dimensionless 1d power spectrum defined in equation ( [ eq : defe ] ) . _
middle panel : _ the spectral slopes at @xmath104 .
_ right panel : _ the weighted value of @xmath70 defined in equation ( [ eq : kparw ] ) . in all panels ,
the solid lines refer to @xmath43 and the dashed lines refer to @xmath44 . in the right - hand panel ,
the two lines are almost on top of each other .
[ fig : f2],title="fig:",width=201 ] the left - hand panel of figure [ fig : f2 ] is a plot of the dimensionless one - dimensional power spectrum , @xmath202 which is proportional to the energy per unit @xmath5 in @xmath54 fluctuations .
the middle panel of figure [ fig : f2 ] shows that in the inertial range , @xmath203 and @xmath204 , as expected for @xmath205 .
the right - hand panel shows that the weighted value of @xmath70 , @xmath206 is roughly constant in the inertial range .
this section addresses strong turbulence with @xmath207 , @xmath208 , @xmath209 , and @xmath210 at the outer scale @xmath180 .
the discussion allows for the possibility that @xmath211 . as fluctuation energy cascades to larger @xmath5
, it diffuses to larger @xmath136 , so that @xmath212 and @xmath213 increase with increasing @xmath5 .
moreover , for both @xmath43 and @xmath44 , the fluctuation energy diffuses over a @xmath70-increment of @xmath214 during one cascade time . for the steady - state solutions of interest , @xmath215 is an increasing function of @xmath5 , and thus at each @xmath5 we will have that @xmath216 .
one can thus define a single parallel - wavenumber , @xmath217 , to describe the spectra , with @xmath218 at each @xmath5 . since @xmath219 at each scale ,
@xmath220 throughout the inertial range . on the other hand , since @xmath221 can be much less than @xmath222 , @xmath223 can be @xmath58 .
the cascade time for the @xmath44 fluctuations is given by the strong - turbulence phenomenology of equation ( [ eq : tau2 ] ) , so that the energy flux in @xmath44 fluctuations is @xmath224 allowing for the possibility that @xmath211 , the cascade time of the @xmath43 fluctuations follows the weak - turbulence phenomenology of equation ( [ eq : tau1 ] ) .
this formula is also accurate for @xmath223 as large as 1 ( in which case @xmath225 ) .
the energy flux in @xmath43 fluctuations is then @xmath226 which is roughly the same as @xmath161 .
it is assumed that the energy flux depends on the spectral slope as in weak turbulence , so that the fluctuation type with the steeper spectrum has the larger energy flux .
if @xmath227 then equations ( [ eq : epsms ] ) and ( [ eq : epsps ] ) imply that when @xmath160 and @xmath161 are independent of @xmath5 , @xmath228 the condition that @xmath98 throughout the inertial range then implies that @xmath229 as discussed in earlier studies ( grappin et al 1983 , lithwick & goldreich 2003 ) , the spectra are pinned at the dissipation scale , so that the dominant fluctuation type will have the steeper spectrum and a somewhat larger energy flux .
for the zero - cross - helicity case , equations ( [ eq : apam1 ] ) and ( [ eq : olkp1 ] ) give @xmath230 and @xmath231 , as in the work of goldreich & sridhar ( 1995 ) [ see also higdon ( 1984 ) ] .
when @xmath232 , equation ( [ eq : fppm ] ) possesses an analytical solution that reproduces the above scalings .
this solution can be obtained by starting with the assumptions that @xmath233 that the energy cascade is dominated by local interactions , and that @xmath234 for all @xmath5 .
equation ( [ eq : taueff ] ) then becomes @xmath235 , and @xmath236 . upon defining @xmath237 and
@xmath238 one can rewrite equation ( [ eq : fppm ] ) as @xmath239 with @xmath240 equation ( [ eq : eqnf ] ) is solved by taking @xmath241 requiring that equation ( [ eq : ansatz ] ) be satisfied , one finds that @xmath242 and @xmath243 the dominance of local interactions requires that @xmath244 , and thus @xmath245 .
when forcing and dissipation are taken into account , the exact solution becomes an approximate solution that is valid only within the inertial range . in this case , @xmath246 because the spectra are pinned at the dissipation scale whereas @xmath175 is larger than @xmath176 within the inertial range .
equation ( [ eq : fsolve ] ) implies that @xmath247 where the dimensionless constants in the expression for @xmath248 have been dropped , but @xmath249 , which has dimensions , has been kept .
equations ( [ eq : defkc ] ) , ( [ eq : ansatz ] ) , and ( [ eq : kpkm0 ] ) show that @xmath250 for all @xmath5 , so that @xmath98 for all @xmath5 . equation ( [ eq : defe ] ) gives @xmath251 from which it follows that @xmath252 and @xmath253 this solution reduces to the critical - balance solution of goldreich & sridhar ( 1995 ) when @xmath254 , in which case @xmath255 , and @xmath256 . comparing equations ( [ eq : wlps ] ) and ( [ eq : wlms ] ) with equation ( [ eq : defapam ] ) , it can be seen that @xmath257 corresponds to @xmath258 and @xmath259 corresponds to @xmath260 .
equation ( [ eq : kpkm0 ] ) is thus equivalent to equation ( [ eq : olkp1 ] ) , and equation ( [ eq : bpbm ] ) is equivalent to equation ( [ eq : apam1 ] ) .
figure [ fig : f3 ] shows the results from a numerical solution of equation ( [ eq : fppm ] ) , obtained by integrating forward in time as described in section [ sec : weak ] with the spectra initially equal to zero .
the numerical solution was obtained by setting @xmath261 with @xmath262 and using the parameters ( defined in section [ sec : weak ] ) @xmath192 , @xmath263 , @xmath194 , @xmath264 , and @xmath265 , where the constant @xmath199 is an effective viscosity .
the rms velocities at steady state are @xmath266 and @xmath267 .
the left - hand panel shows that the one - dimensional energy spectrum @xmath268 is @xmath269 in the inertial range , which corresponds to @xmath270 in equation ( [ eq : eb ] ) .
equation ( [ eq : bpbm ] ) then gives @xmath271 .
the dotted lines in the middle panel of figure [ fig : f3 ] correspond to the values of @xmath270 and @xmath271 , which are reasonably close to the values of @xmath272 in the numerical solution , although these latter values vary throughout the inertial range in the numerical solution . for @xmath270 ,
equation ( [ eq : kpkm0 ] ) gives @xmath273 , which is a close match to the numerical solution , as shown in the right - hand panel of figure [ fig : f3 ] .
the left - hand panel of figure [ fig : f3 ] shows that the steady - state solutions for @xmath274 and @xmath275 are `` pinned '' at the dissipation scale , as expected . ) for strong turbulence with @xmath276 .
_ left panel : _ the dimensionless 1d power spectrum defined in equation ( [ eq : defe ] ) . _
middle panel : _ the spectral slopes at @xmath104 .
_ right panel : _ the weighted value of @xmath70 defined in equation ( [ eq : kparw ] ) . in all panels ,
the solid lines refer to @xmath43 and the dashed lines refer to @xmath44 .
[ fig : f3],title="fig:",width=201 ] ) for strong turbulence with @xmath276 .
_ left panel : _ the dimensionless 1d power spectrum defined in equation ( [ eq : defe ] ) . _
middle panel : _ the spectral slopes at @xmath104 .
_ right panel : _ the weighted value of @xmath70 defined in equation ( [ eq : kparw ] ) . in all panels ,
the solid lines refer to @xmath43 and the dashed lines refer to @xmath44 .
[ fig : f3],title="fig:",width=201 ] ) for strong turbulence with @xmath276 .
_ left panel : _ the dimensionless 1d power spectrum defined in equation ( [ eq : defe ] ) . _
middle panel : _ the spectral slopes at @xmath104 .
_ right panel : _ the weighted value of @xmath70 defined in equation ( [ eq : kparw ] ) . in all panels ,
the solid lines refer to @xmath43 and the dashed lines refer to @xmath44 .
[ fig : f3],title="fig:",width=201 ] it should be noted that when @xmath98 and @xmath211 , the dominant @xmath43 fluctuations are only weakly damped by nonlinear interactions with @xmath44 waves , in the sense that @xmath89 is much larger than the linear wave period . on the other hand , for the smaller - amplitude @xmath44 fluctuations , the linear wave period and cascade time are comparable .
thus , paradoxically , the larger - amplitude @xmath43 fluctuations can be described as waves , or , more precisely , a non - sinusoidal wave train , whereas the smaller - amplitude @xmath44 fluctuations can not be accurately described as waves .
this section again addresses turbulence in which @xmath277 at the outer - scale wavenumber , @xmath278 . in the weak - turbulence limit ,
@xmath279 and @xmath223 increase with increasing @xmath5 .
if the dissipation wavenumber @xmath182 is sufficiently large , then @xmath279 and/or @xmath223 will increase to a value of order unity at some @xmath5 within the inertial range .
this perpendicular wavenumber is denoted @xmath280 .
the turbulence will then be described by the weak - turbulence scalings of section [ sec : weak ] for @xmath281 , and by the strong - turbulence scalings of section [ sec : strong ] for @xmath282 .
figure [ fig : f4 ] shows a numerical solution of equation ( [ eq : fppm ] ) that illustrates how the turbulence makes this transition in a smooth manner . at small wavenumbers ,
this solution is similar to the weak - turbulence solution plotted in figure [ fig : f2 ] , and at large wavenumbers it is similar to the strong - turbulence solution plotted in figure [ fig : f3 ] .
the solution shown in figure [ fig : f4 ] was obtained by integrating equation ( [ eq : fppm ] ) forward in time to steady state using the numerical method described in section [ sec : weak ] .
the spectra were initially set equal to zero .
the numerical solution was obtained by setting @xmath283 with @xmath284 and using the parameters @xmath285 , @xmath286 , @xmath194 , @xmath264 , and @xmath265 , where the constant @xmath199 is an effective viscosity . the rms velocities at steady state are @xmath287 and @xmath288 . ) showing a smooth transition from the weak - turbulence limit at small @xmath5 to the strong - turbulence limit at large @xmath5 .
_ left panel : _ the dimensionless 1d power spectrum defined in equation ( [ eq : defe ] ) . _
middle panel : _ the spectral slopes at @xmath104 .
_ right panel : _ the weighted value of @xmath70 defined in equation ( [ eq : kparw ] ) . in all panels ,
the solid lines refer to @xmath43 and the dashed lines refer to @xmath44 .
[ fig : f4],title="fig:",width=201 ] ) showing a smooth transition from the weak - turbulence limit at small @xmath5 to the strong - turbulence limit at large @xmath5 . _ left panel : _ the dimensionless 1d power spectrum defined in equation ( [ eq : defe ] ) . _
middle panel : _ the spectral slopes at @xmath104 .
_ right panel : _ the weighted value of @xmath70 defined in equation ( [ eq : kparw ] ) . in all panels ,
the solid lines refer to @xmath43 and the dashed lines refer to @xmath44 .
[ fig : f4],title="fig:",width=201 ] ) showing a smooth transition from the weak - turbulence limit at small @xmath5 to the strong - turbulence limit at large @xmath5 .
_ left panel : _ the dimensionless 1d power spectrum defined in equation ( [ eq : defe ] ) .
_ middle panel : _ the spectral slopes at @xmath104 .
_ right panel : _ the weighted value of @xmath70 defined in equation ( [ eq : kparw ] ) . in all panels ,
the solid lines refer to @xmath43 and the dashed lines refer to @xmath44 .
[ fig : f4],title="fig:",width=201 ]
in sections [ sec : weak ] through [ sec : trans ] , it was assumed that @xmath289 at the outer scale .
this assumption is applicable to many settings .
for example , in a plasma of dimension @xmath7 that is stirred by a force that has a correlation length @xmath290 , the velocity fluctuations that are excited have a correlation length @xmath291 , and this correlation length is imprinted on both the @xmath43 and @xmath44 fluctuations . on the other hand ,
if waves are launched along the magnetic field into a bounded plasma from opposite sides of the plasma , and the waves from one side have a much larger parallel correlation length than the waves from the other side , it is possible to set up turbulence in which the two wave types have very different parallel correlation lengths at the outer scale .
this situation is discussed briefly in this section . for strong turbulence ,
if both @xmath279 and @xmath223 are @xmath61 at some perpendicular scale @xmath76 , but one fluctuation type , say @xmath43 , has a much smaller parallel correlation length than the other ( and thus a much larger amplitude ) , then during a time @xmath292 the `` propagation with distortion '' mechanism discussed in section [ sec : parcasc ] will increase @xmath293 until it equals @xmath212 , which will cause @xmath223 to become @xmath58 at scale @xmath76 . at smaller scales ,
the solution can be described by the scalings presented in section [ sec : strong ] , in which @xmath294 .
similarly , if @xmath295 but @xmath211 at some scale @xmath76 and if @xmath296 at that scale , then during a time @xmath292 the `` propagation with distortion '' mechanism discussed in section [ sec : parcasc ] will again increase @xmath213 until it equals @xmath212 , the parallel scales will remain comparable at smaller perpendicular scales , and the solution can be described by the scalings in section [ sec : strong ] .
the case in which @xmath297 , @xmath211 , and @xmath298 is not addressed in this paper .
in this section , the preceding analysis of incompressible mhd turbulence is applied to the solar wind and solar corona . it should be noted at the outset , however , that the solar wind and solar corona ( beyond roughly @xmath299 , where @xmath300 is distance from the sun s center ) are in the collisionless regime , and the pressure tensor is not isotropic as assumed in ideal mhd .
moreover , the value of @xmath301 is @xmath58 in the corona and typically @xmath61 in the solar wind at 1 au , whereas incompressible mhd corresponds to the limit @xmath302 .
a preliminary question that needs to be addressed is thus the extent to which incompressible mhd is an accurate model for these plasmas .
schekochihin et al ( 2007 ) have recently carried out extensive calculations based on kinetic theory that provide a detailed answer to this question .
these authors examined anisotropic turbulence in weakly collisional magnetized plasmas using gyrokinetics , a low - frequency expansion of the vlasov equation that averages over the gyromotion of the particles . by applying the form of the gyrokinetic expansion derived by howes et al ( 2006 ) , schekochihin et
al ( 2007 ) showed analytically that non - compressive alfvnic turbulence in the quasi-2d regime ( i.e. , @xmath167 ) can be accurately described using reduced mhd in both the collisional and collisionless limits , regardless of @xmath303 , provided that the length scales of the fluctuations are much larger than the proton gyroradius and the frequencies are much less than the proton cyclotron frequency . since non - compressive quasi-2d fluctuations
are thought to be the dominant component of the turbulence in the solar wind ( see , e.g. , bieber et al 1994 ) and the solar corona ( dmitruk & matthaeus 2003 , cranmer & van ballegooijen 2005 ) , incompressible mhd is a useful approximation for modeling turbulence in these settings .
cross helicity in the solar wind has been measured _ in situ _ by several different spacecraft . in terms of the elsasser variables
@xmath54 , there is a substantial excess of outward propagating fluctuations ( taken to be @xmath43 throughout this section ) over inward propagating fluctuations ( taken to be @xmath44 ) in the inner heliosphere , although this imbalance decreases with increasing @xmath300 , as seen , for example , in voyager data for low heliographic latitude ( matthaeus & goldstein 1982 , roberts et al 1987 ) and ulysses data at high latitude ( goldstein et al 1995 ) . in a study of ulysses and helios data ,
bavassano et al ( 2000 ) found that @xmath304 for @xmath305 au and @xmath306 for @xmath307 , where @xmath308 is the energy per unit mass associated with @xmath54 fluctuations .
these numbers are intended as illustrative average values , as individual measurements of @xmath309 in the solar wind vary significantly .
although it has not been directly measured , the ratio @xmath309 is likely very large in open - field - line regions of the solar corona .
this can be seen from the work of cranmer & van ballegooijen ( 2005 ) , who modeled the generation of alfvn waves by the observed motions of field - line footpoints in the photosphere , and the propagation and reflection of these waves as they travel along open field lines from the photosphere out into the interplanetary medium .
they found that the ratio of the frequency - integrated rms elsasser variables ( @xmath43 and @xmath44 ) is @xmath310 at @xmath311 ( i.e. , @xmath312 ) .
verdini & velli ( 2007 ) developed a different model for the propagation , reflection , and turbulent dissipation of alfvn waves in the solar atmosphere and solar wind and found that @xmath313 at @xmath314 .
based on these results , one can make the rough estimate that @xmath315 where @xmath278 is the perpendicular wavenumber at the outer scale . in much of the solar wind
, @xmath9 is comparable to @xmath316 , and the turbulence is in the strong - turbulence regime with @xmath317 . for the corona , cranmer & van ballegooijen ( 2005 ) found that the outer - scale fluctuations in open - field - line regions have periods of @xmath318 minutes , @xmath319 km / s , and perpendicular correlation lengths of @xmath320 km .
the alfvn speed in their model corona is between 2000 and 3000 km / s at @xmath314 , and thus @xmath321 .
however , the parallel correlation length @xmath322 of the outer - scale fluctuations is @xmath323 km , which is @xmath324 . because @xmath325 , @xmath326 and the low - frequency fluctuations launched into the corona by footpoint motions are in the strong - turbulence regime
. there may be an additional population of higher - frequency waves in the weak - turbulence regime , but these are not discussed here . in open - field - line regions of the corona ,
when @xmath43 waves are reflected , the resulting @xmath44 waves have the same frequencies as the @xmath43 waves . on the other hand ,
wave - reflection is more efficient at lower frequencies ( velli 1993 ) , so if there is a range of wave frequencies at each @xmath5 , the energy - weighted average frequency of inward waves would tend to be somewhat lower than that of the outward waves .
this suggests that at the outer scale the parallel correlation length @xmath322 of the @xmath44 fluctuations is somewhat larger than the value of @xmath322 for the @xmath43 fluctuations . however , given equation ( [ eq : chicorona ] ) , the @xmath43 fluctuations imprint their parallel correlation length on the @xmath44 fluctuations during a single turnover time @xmath292 , as argued in section [ sec : unequal ] .
the parallel correlation lengths of the @xmath43 and @xmath44 fluctuations in the corona can thus be taken to be approximately equal at the outer scale , and hence also at smaller scales .
the same approximation is reasonable for turbulence in the solar wind .
if we take @xmath278 to be the perpendicular wave number at the outer scale , @xmath327 to be the rms amplitude of the outward - propagating fluctuations at the outer scale , and @xmath328 to be the rms amplitude of the sunward - propagating fluctuations at the outer scale , then equations ( [ eq : epsms ] ) , ( [ eq : epsps ] ) , and ( [ eq : chicorona ] ) imply that @xmath329 this estimate is a factor of @xmath330 smaller than the standard strong - turbulence estimate of @xmath331 that appears in many studies ( e.g. , zhou & matthaeus 1990 , cranmer & van ballegooijen 2005 , lithwick , goldreich , & sridhar 2007 , verdini & velli 2007 ) .
this difference has important implications for turbulent heating of the solar corona and solar wind . for the energetically dominant @xmath43 fluctuations , the cascade time @xmath89
is much longer than the linear wave period , at least at scales much larger than the dissipation scale .
this result is important for determining the conditions under which turbulence can be a viable mechanism to explain the heating of the solar corona .
observations taken with the _ ultraviolet coronagraph spectrometer _ ( uvcs ) indicate that there is strong heating of coronal plasma at @xmath332 ( kohl et al 1998 , antonucci et al 2000 ) . an appealing model to explain
this heating is that low - frequency alfvn waves are launched by turbulent motions of field - line footpoints in the photosphere , that some of these waves are reflected , and that interactions between oppositely directed alfvn wave packets in the corona causes the wave energy to cascade to small scales and dissipate ( matthaeus et al 1999 , 2002 ; dmitruk et al 2001 , 2002 ; cranmer & van ballegooijen 2005 , 2007 ; verdini & velli 2007 ) . in one version of this model ,
the waves that cross the transition region and enter the corona have not yet undergone a turbulent cascade , and their energy is concentrated at the fairly long periods ( @xmath333 minute ) characteristic of the observed footpoint motions that are believed to make the dominant contribution to the outward directed wave flux . in order for this scenario to explain the uvcs measurements , there needs to be time for the outer - scale waves to cascade within the corona before they travel beyond @xmath334 . if , as above , we take @xmath278 to be the value of @xmath5 at the outer scale , @xmath327 to be the rms amplitude of the outward waves at the outer scale , and @xmath322 to be the parallel correlation length of the fluctuations at the outer scale , then equation ( [ eq : tau1 ] ) can be used to express the cascade time for the outward waves at the outer scale as @xmath335 where @xmath336 .
thus , for waves with a period @xmath337 minute , equations ( [ eq : wpwmra ] ) , ( [ eq : chicorona ] ) , and ( [ eq : taucorona ] ) give @xmath338 minutes . on the other hand ,
the alfvn speed in a coronal hole at @xmath339 is @xmath340 km / s ( cranmer & van ballegooijen 2005 ) , and the time for an alfvn wave to travel from the coronal base out to @xmath314 is 4 - 6 minutes .
there is thus not enough time for the energy of waves with periods @xmath341 minute to cascade and dissipate within a few solar radii of the sun .
dmitruk & matthaeus ( 2003 ) and verdini & velli ( 2007 ) avoid this difficulty by postulating that a broad frequency spectrum of waves is launched upwards from the photosphere . another possible way around
this difficulty is the development of a broad frequency spectrum of fluctuations from wave - packet collisions in the chromosphere , in which the energies of inward and outward propagating waves are comparable due to strong wave reflection at the transition region .
much of the discussion of the inertial - range power spectrum of solar - wind turbulence has focused on the question of whether the spectral index is closer to the kolmogorov ( 1941 ) value of @xmath342 or the iroshnikov - kraichnan value of @xmath343 ( iroshnikov 1963 , kraichnan 1965 ) .
a value of @xmath342 is supported by a number of theoretical studies ( e.g. , montgomery & turner 1981 , higdon 1984 , goldreich & sridhar 1995 ) and numerical simulations ( cho & vishniac 2000 , mller & biskamp 2000 , cho et al 2002 , haugen et al 2004 ) .
a value value of @xmath343 is supported by a second group of theoretical studies ( boldyrev 2005 , 2006 ; mason et al 2006 ; see also beresnyak & lazarian 2006 ) and numerical simulations ( maron & goldreich 2001 , mller et al 2003 , mller & grappin 2005 , mininni & pouquet 2007 ) .
it should be noted that all of the above - mentioned studies address mhd turbulence with negligible cross helicity .
spacecraft measurements yield frequency spectra for the magnetic field and velocity fluctuations , where the frequency @xmath344 is approximately @xmath345 , where @xmath346 is the radial component of the wave - vector and @xmath347 is the solar - wind speed [ taylor s ( 1938 ) `` frozen - in flow hypothesis '' ] . below a spectral - break frequency
@xmath348 , the spectra are typically fairly flat , being approximately proportional to @xmath349 ( matthaeus & goldstein 1986 ) . at @xmath350 , the spectra steepen .
the time scale corresponding to the spectral break , @xmath351 , increases with increasing @xmath300 .
for example , bruno & carbone ( 2005 ) found that @xmath351 was @xmath352 hours at 0.3 au , 0.16 hours at 0.7 au , and 0.4 hours at 0.9 au in a sample of helios 2 data . in two other studies based on data from several spacecraft ,
matthaeus & goldstein ( 1986 ) found that @xmath353 hours at 1 au , while klein et al ( 1992 ) found @xmath354 hours at 4 au .
the inertial range roughly corresponds to frequencies in the interval @xmath355 , where @xmath356 is the frequency corresponding to the dissipation scale . at 1 au , @xmath357 ( smith et al 2006 ) .
a large number of inertial - range spectral indices have been reported in the literature .
for example , matthaeus & goldstein ( 1982 ) found a spectral index of @xmath358 for the magnetic field in voyager data at @xmath359 au , and a spectral index of @xmath360 for the total energy .
goldstein et al ( 1995 , fig .
1 ) found that the spectral index for the @xmath43 fluctuations was slightly steeper than @xmath342 in ulysses data at 2 au and 4 au .
their results also suggest a shallower @xmath44 spectrum , consistent with the idea that the spectra are pinned at the dissipation wavenumber @xmath182 . in a study of helios 2 magnetic - field data , bruno &
carbone ( 2005 , figure 23 ) found spectral indices of @xmath361 at 0.3 au , @xmath362 at 0.7 au , and @xmath363 at 0.9 au .
using data from the wind spacecraft at 1 au , podesta et al ( 2007 ) found a total - energy spectral index of @xmath364 , with the velocity spectrum flatter than the magnetic spectrum .
marsch & tu ( 1996 ) found a spectral index of @xmath365 for the magnetic field at 1 au in helios 2 data .
horbury & balogh ( 1995 ) found a spectral index close to @xmath342 for the magnetic field in ulysses data at 2.5 au . using magnetic - field data from the ace spacecraft at 1 au , smith et al ( 2006 ) found a spectral index of @xmath366 in open - field - line regions and @xmath367 in magnetic clouds .
smith ( 2003 ) found spectral indices between @xmath368 and @xmath369 in a study of ulysses magnetic - field data covering a range of heliographic latitudes and radii .
overall , the spectra are more consistent with a kolmogorov scaling than an iroshnikov - kraichnan scaling
. it should be emphasized , however , that the observations in several cases are consistent with inertial range spectra that are steeper than a kolmogorov spectrum .
spectral indices @xmath370 have been found in previous theoretical studies of weak incompressible mhd turbulence ( bhattacharjee & ng 1997 , goldreich & sridhar 1997 , galtier et al 2000 , bhattacharjee & ng 2001 , perez & boldyrev 2008 ) , as well as strong isotropic mhd turbulence with cross helicity ( grappin et al 1983 ) . in this paper , it is argued that spectral indices @xmath370 are a consequence of cross helicity in strong anisotropic mhd turbulence . the simplest way to apply this paper to solar wind turbulence is to model the solar wind fluctuations at some location as steady - state , forced , homogeneous turbulence with the same average value of @xmath179 , where @xmath327 and @xmath328 are the rms amplitudes of the @xmath54 fluctuations at the outer scale @xmath180 . upon setting @xmath371 ,
one can write @xmath372 , where it is assumed that the spectra are equal at the dissipation wave number @xmath182 . equation ( [ eq : apam1 ] ) then gives @xmath373 , and the value of @xmath374 can be obtained from the equation @xmath375 .
the total - energy spectrum , @xmath376 , is approximately @xmath377 , although it is flatter than @xmath377 near the dissipation scale where the flatter spectrum of the @xmath44 fluctuations is important . thus , at scales much larger than the dissipation scale , @xmath378 where @xmath379}{3\log_{10}(k_d / k_f)}. \label{eq : defc3}\ ] ] upon defining the outer - scale fractional cross helicity as @xmath380 one can rewrite equation ( [ eq : defc3 ] ) as @xmath381}{3\log_{10}(k_d / k_f)}. \label{eq : defc32}\ ] ] the spectral index from equation ( [ eq : defc32 ] ) is plotted in figure [ fig : f5 ] , assuming that @xmath382 , where @xmath383 is the break frequency at 1 au discussed above ( matthaeus & goldstein 1986 ) and @xmath384 is the frequency at the dissipation scale .
( smith et al 2006 ) with this choice , @xmath385 for @xmath386 and @xmath387 for @xmath388 .
when equation ( [ eq : defc32 ] ) is applied to the solar wind , @xmath389 should be interpreted as the cross helicity at the outer scale @xmath180 averaged over at least a few outer - scale fluctuations .
, on the outer - scale fractional cross helicity @xmath389 .
the ratio of the dissipation wavenumber @xmath182 to the perpendicular wavenumber at the outer - scale @xmath278 in equation ( [ eq : defc32 ] ) is taken to be @xmath390 .
[ fig : f5],width=336 ] some caution , however , is warranted when applying equation ( [ eq : defc32 ] ) to the solar wind because solar wind conditions vary with @xmath300 , while equation ( [ eq : defc32 ] ) is based on results for homogeneous turbulence .
the condition under which it is valid to treat the solar - wind fluctuations at a some @xmath300 with an observed @xmath389 as homogeneous turbulence with the same @xmath389 is that the cascade time at the outer scale @xmath391 be much smaller than the time @xmath392 for turbulent structures to be advected a distance @xmath300 , where @xmath347 is the solar wind speed .
however , this condition is often either violated or only marginally satisfied .
this is illustrated by the results of matthaeus & goldstein ( 1982 ) based on four days of voyager data ( in their table 1 ) .
for the fluctuations at 1 au , these authors found an rms velocity fluctuation of @xmath393 km / s [ which is comparable to @xmath40 at 1 au ] , a velocity correlation length ( outer scale ) of @xmath394 cm , and an average solar - wind speed of @xmath395 km / s . in this particular data set , @xmath396 , so that @xmath397 , with @xmath398 s. on the other hand , @xmath399 s , which marginally satisfies the condition @xmath400 .
however , if @xmath401 were instead @xmath402 at 1 au , as in the results of bavassano et al ( 2000 ) , then @xmath403 would be somewhat larger than @xmath404 .
thus , the inhomogeneity of the solar wind may influence the effects of cross helicity on the spectral indices .
however , more detailed modeling of inhomogeneous solar - wind turbulence is beyond the scope of this paper .
in this section , the results of this paper are compared to two recent studies of strong anisotropic incompressible mhd turbulence with cross helicity . the model for the cascade of energy to larger @xmath70 used in this paper
is based on the results of lithwick , goldreich , & sridhar ( 2007 ) ( hereafter lgs07 ) . as a result , in both studies , if @xmath43 and @xmath44 have comparable correlation lengths in the direction of @xmath0 at the outer scale , then @xmath405 at all smaller scales .
the principal difference between this paper and lgs07 lies in our treatment of the cascade time for the dominant fluctuation type , @xmath43 . lgs07
argue that if @xmath406 , @xmath98 , and @xmath211 , then the shearing applied by @xmath44 wave packets on a @xmath43 wave packet at perpendicular scale @xmath76 is coherent over a time @xmath407 , which greatly exceeds the time @xmath408 required for a @xmath43 and @xmath44 wave packet at perpendicular scale @xmath76 to pass through each other .
in contrast , in this paper , it is argued that the coherence time for the straining of the @xmath43 wave packet is of order the `` crossing time '' @xmath409 . as a result ,
the results obtained in this paper for the inertial - range power spectra , degree of anisotropy , cascade time , and energy fluxes are different from those of lgs07 . the approach taken in this paper is motivated by the following argument . as argued by lgs07 and maron & goldreich ( 2001 ) , the @xmath44 fluctuations propagate approximately along the hypothetical magnetic field lines obtained from the sum of @xmath0 and the magnetic field of the @xmath43 fluctuations .
let us call these hypothetical magnetic field lines the `` @xmath43 field lines , '' and let us consider a @xmath43 wave packet of perpendicular scale @xmath76 and parallel scale @xmath410 , where @xmath411 .
let us work in a frame of reference that moves at speed @xmath40 in the @xmath17 direction along with the @xmath43 fluctuations .
let us also take an initial snapshot of the turbulence at @xmath105 and trace out all of the `` @xmath43 field lines '' that pass through our wave packet .
the volume filled by these @xmath43 field lines is the `` source region '' from which the @xmath44 wavepackets encountered by our @xmath43 wave packet originate .
if we wait one crossing time @xmath412 and take a new snapshot of the turbulence , then at any given location the @xmath43 fluctuations will not have changed very much , since @xmath413 . however , if we trace out the new @xmath43 field lines passing through our wave packet , the volume that is filled by these new @xmath43 field lines will differ substantially from the initial source region at distances @xmath414 from our wave packet due to the rapid divergence of neighboring field lines in mhd turbulence .
in other words , small local changes in @xmath43 lead to large changes in the connectivity of the @xmath43 field lines . to see this ,
let the @xmath43 field line that passes through some point @xmath415 in our wave packet at @xmath105 be called `` field line @xmath416 . ''
let the @xmath43 field line that passes through point @xmath415 at @xmath417 be called `` field line @xmath418 . '' as before
, let us work in a frame of reference that moves at speed @xmath40 in the @xmath17 direction . field lines @xmath416 and @xmath418 are fixed curves , since they are traced out within two snapshots of the turbulence . if we follow field line @xmath418 for a distance @xmath419 , it will separate from field line @xmath416 by some small distance @xmath420 that is @xmath421 . if we continue to follow field line @xmath418 ,
its separation from field line @xmath416 is analogous to the separation of two neighboring field lines within a single snapshot of the turbulence .
as shown by narayan & medvedev ( 2001 ) , chandran & maron ( 2004 ) , and maron , chandran , & blackman ( 2004 ) , if a pair of field lines is separated by a distance @xmath422 at one location , then the distance the field - line pair must be followed before it separates by a distance @xmath76 is a few times the parallel size of an eddy of perpendicular size @xmath76 - i.e. , a few times @xmath423 .
it turns out that the particular value of @xmath420 has little effect unless one considers the ( irrelevant ) case in which @xmath424 where @xmath425 is large and @xmath426 is the perpendicular dissipation scale .
( chandran & maron 2004 ) this is because within the inertial range the amount of magnetic shear increases towards small scales ; therefore , if @xmath420 is made very small , then the distance one has to follow the field - line pair in order for @xmath420 to double becomes very small .
thus , as a result of the rapid divergence of neighboring field lines in mhd turbulence , the `` source region '' of our @xmath43 wave packet at @xmath427 differs substantially from the source region at @xmath105 at distances @xmath428 from our wave packet . because the @xmath44 fluctuations vary rapidly in the direction perpendicular to the magnetic field , the @xmath44 wave packets encountered by our @xmath43 wave packet will decorrelate on a time scale of order the crossing time @xmath429 due to the time evolution of the source region .
it should noted that there are two unexplained aspects of lgs07 s model , as pointed out by beresnyak & lazarian ( 2007 ) .
the first concerns the nature of the transition from the weak - turbulence regime ( @xmath91 and @xmath211 ) to the strong - turbulence regime ( @xmath276 and @xmath211 ) . in lgs07
s analysis , as one passes from the weak regime to the strong regime , the coherence time for the straining of the @xmath43 wave packets by @xmath44 wave packets increases by a factor of @xmath430 and the energy cascade time @xmath89 decreases by a factor of @xmath223 .
it is not clear why these large changes should occur across the transition scale .
the second issue is that lgs07 find that @xmath431 and @xmath432 regardless of the fractional cross helicity . since the ratio @xmath433 is independent of wavenumber , it is not clear how pinning could occur in their model , or how the spectra would behave near the dissipation scale .
beresnyak & lazarian ( 2007 ) ( hereafter bl07 ) have published an online article on strong mhd turbulence with cross helicity .
the following discussion refers to the version of their article that is available electronically as of the writing of this paper .
the work of bl07 is similar to this paper in that both studies take the dominant fluctuation type , @xmath43 , to undergo a weak cascade .
also , equations ( [ eq : defapam ] ) through ( [ eq : olkp1 ] ) of this paper are equivalent to their equation ( 5 ) , except for the fact that they take the parallel correlation lengths of @xmath43 and @xmath44 to differ by a constant multiplicative factor when a power - law solution for the spectra is assumed .
( the possibility of more general solutions is claimed by bl07 . ) on the other hand , there are a number of significant differences between this paper and beresnyak & lazarian s ( 2007 ) work .
they argue that for the @xmath43 fluctuations , the dominant nonlinear interactions are between fluctuations with comparable parallel correlation lengths and different perpendicular correlation lengths , whereas for the @xmath44 fluctuations the dominant interactions are between fluctuations with comparable perpendicular scales . here , it is argued that for both @xmath43 and @xmath44 the dominant interactions are between fluctuations with similar perpendicular scales .
when @xmath434 and @xmath98 , their figure 1 suggests that the parallel correlation length of @xmath43 fluctuations can be less than the parallel correlation length of @xmath44 fluctuations .
it is argued in section [ sec : unequal ] of this paper that this can not be the case , because the @xmath43 fluctuations will imprint their parallel correlation length onto the @xmath44 fluctuations .
they argue that the scalings given by equations ( [ eq : defapam ] ) and ( [ eq : olkp1 ] ) [ equivalently , their equation ( 5 ) ] can not apply if the parallel correlation lengths of the @xmath43 and @xmath44 fluctuations are equal , arguing that this would require @xmath435 , whereas in this paper the ratio @xmath178 depends upon the slopes of the power spectra , as in weak turbulence .
they argue that if the @xmath43 and @xmath44 fluctuations are driven with the same parallel correlation length at the outer scale , there will be a non - power law part of the solution at large scales that will transition at smaller scales to a power law solution with @xmath222 and @xmath436 both @xmath437 and @xmath438 .
in contrast , in this paper a power - law solution starting at the outer scale is obtained with different scalings for @xmath439 and @xmath436 , and with @xmath440 growing more slowly than @xmath441 .
this paper proposes a new phenomenology for strong , anisotropic , incompressible mhd turbulence with cross helicity and introduces a nonlinear advection - diffusion equation [ equation ( [ eq : fppm ] ) ] to describe the time evolution of the anisotropic power spectra of the @xmath43 and @xmath44 fluctuations .
it is found that in steady state the one - dimensional power spectra of the energetically dominant @xmath43 fluctuations , @xmath442 , is steeper than @xmath1 , and that @xmath442 becomes increasingly steep as the fractional cross helicity @xmath389 increases .
increasing @xmath389 also increases the energy cascade time of the @xmath43 fluctuations , reduces the turbulent heating power for a fixed fluctuation energy , and increases the anisotropy of the fluctuations at small scales .
although most of the discussion has focused on forced , steady - state turbulence , the results of this paper can also be applied to decaying turbulence .
for example , equations ( [ eq : tau1 ] ) and ( [ eq : tau2 ] ) can be used to estimate the time scale for turbulence to decay .
the resulting prediction is that if the fluctuations are initially excited with @xmath443 and with comparable parallel correlation lengths at the outer scale , then the turbulence will decay into a state in which @xmath44 fluctuations are absent , as in the earlier work of dobrowolny , mangeney , & veltri ( 1980 ) , grappin et al ( 1983 ) , and lithwick & goldreich ( 2003 ) .
this `` maximally aligned '' state will then be free from nonlinear interactions , and will persist for long times until it damps via linear dissipation .
i thank chuck smith , sebastien galtier , bernie vasquez , yoram lithwick , alex lazarian , and andrey beresnyak for helpful comments and suggestions .
this work was supported in part by the nsf / doe partnership in basic plasma science and engineering under grant number ast-0613622 , by nasa under grant numbers nnx07ap65 g and nnh06zda001n - shp06 - 0071 , and by doe under grant number de - fg02 - 07-er46372 . | this paper proposes a new phenomenology for strong incompressible mhd turbulence with nonzero cross helicity .
this phenomenology is then developed into a quantitative fokker - planck model that describes the time evolution of the anisotropic power spectra of the fluctuations propagating parallel and anti - parallel to the background magnetic field @xmath0 .
it is found that in steady state the power spectra of the magnetic field and total energy are steeper than @xmath1 and become increasingly steep as @xmath2 increases , where @xmath3 is the cross helicity , @xmath4 is the fluctuation energy , and @xmath5 is the wavevector component perpendicular to @xmath0 .
increasing @xmath6 with fixed @xmath4 increases the time required for energy to cascade to smaller scales , reduces the cascade power , and increases the anisotropy of the small - scale fluctuations .
the implications of these results for the solar wind and solar corona are discussed in some detail . |
in theoretical statistics the marginal polytope plays an important role . it is the polytope of possible values that a sufficient statistics can take .
it encodes in its face lattice the combinatorial structure of the boundary of the exponential family defined by the statistics . for a model on discrete random variables
it can be represented with vertices that have only components 0 or 1 , commonly called a 0/1 polytope . in coding theory when decoding binary linear codes one can apply techniques from linear programming and optimize a linear function over the convex hull of the code words , known as the codeword polytope @xcite . observing that for certain choices of sufficient statistics on binary random variables these two notions coincide ,
our main contribution is a characterization of the corresponding polytopes .
we do not address problems that are directly linked to coding theory .
however , we do hope that our result will contribute to a better understanding of the closure of exponential families , which is an important problem in statistics .
the paper is organized as follows : in section [ sec : prelim ] , we introduce the necessary notions to define hierarchical models and fix the notation .
we review different descriptions of so called interaction spaces in section [ sec : interactionspacesgeneratingsets ] . in section [ sec : codingtheory ] , we establish the link to coding theory .
finally , in section [ sec : classification ] we give our main result , the classification of all such full dimensional polytopes whose vertices form a linear code and give a recursive formula for their number .
given a non - empty finite set @xmath0 , we denote the set of probability distributions on @xmath0 by @xmath1 . the _ support _ of @xmath2 is defined as @xmath3 .
the set of distributions with full support is denoted as @xmath4 .
the set @xmath5 has the geometrical structure of a @xmath6-dimensional simplex lying in an affine hyperplane of @xmath7 the vector space of real valued functions on @xmath8 .
statistical models , such as hierarchical models are subsets of @xmath5 . in this paper , we will only consider so called exponential families which are smooth manifolds .
the map @xmath9 is called the _ exponential map_.
it acts component wise by exponentiating and normalizing .
then , an _ exponential family _
( _ in @xmath4 _ ) is defined as the image @xmath10 of a linear subspace @xmath11 of @xmath12 .
an exponential family @xmath13 naturally has full support and is therefore contained in the open simplex @xmath4 .
however , to get probability distributions with reduced support one has to pass to the closure @xmath14 with respect to the standard topology of @xmath15 .
now we consider a compositional structure of @xmath16 induced by the set @xmath17{\mathrel{\mathop{:}}=}{\left\lbrace 1,\ldots , n \right\rbrace}$ ] . given a subset @xmath18 $ ] , we define @xmath19 and the natural projection @xmath20 } } \to { \mathcal x}_a , \qquad ( x_i)_{i \in [ n ] } \mapsto ( x_i)_{i \in a } \ ; .\ ] ] in the following , we will abbreviate @xmath21}$ ] .
one can view @xmath5 as the set of joint probability distributions of the binary random variables @xmath22 \right\rbrace}$ ] .
we now use the compositional structure of @xmath16 in order to define exponential families in @xmath23 given by interaction spaces .
now , decompose @xmath24 in the form @xmath25\setminus a})$ ] with @xmath26 , @xmath27 \setminus a}\in{\mathcal x}_{[n ] \setminus a}$ ] , and define @xmath28 to be the subspace of functions that do not depend on the configurations @xmath27 \setminus a}$ ] : @xmath29 \setminus a } ) = f(x_a , x_{[n ] \setminus a}')\right .
\\ & & \qquad \qquad \quad \left .
\mbox{for all $ x_a \in { \mathcal x}_a$ , and all $ x_{[n ] \setminus a } , x_{[n ] \setminus a } ' \in { \mathcal x}_{[n ] \setminus a}$}\right\ } \ ; .\end{aligned}\ ] ] in the following , we apply these interaction spaces as building blocks for more general interaction spaces and associated exponential families @xcite .
the definition of a hierarchical model is based on the notion of a hypergraph @xcite : a _ pre - hypergraph _
@xmath30 is a non - empty subset of @xmath31 } \setminus{{\left\lbrace \emptyset \right\rbrace}}$ ] that contains all atoms @xmath32 for @xmath33 $ ] .
a _ hypergraph _ is a pre - hypergraph that is ( inclusion ) complete in the following sense : if @xmath34 and @xmath35 it follows that @xmath36 . for technical convenience ,
we have defined hypergraphs to be complete . in this way , it is easy to define a hierarchical model for each hypergraph .
however , the notion of a pre - hypergraph turns out to be more natural in the context of the polytopes and linear codes that we consider below . given a hypergraph , we define the associated interaction space by @xmath37 note that , since a function that depends only on its arguments in @xmath38 , only depends on its arguments in @xmath39 , it suffices to consider the inclusion maximal elements in @xmath30 . we denote them by @xmath40 and have @xmath41 we consider the corresponding exponential family : the _ hierarchical model _ assigned to the hypergraph @xmath30 is the exponential family @xmath42 we give two examples for hypergraphs : [ graphmod]@xmath43 + * ( 1 ) graphical models : * let @xmath44 be an undirected graph , and define @xmath45 here , a _ clique _ is a set @xmath46 that satisfies the following property : @xmath47 the exponential family @xmath48 is characterized by markov properties with respect to @xmath49 ( see @xcite ) . +
* ( 2 ) interaction order : * the hypergraph associated with a given interaction order @xmath50 is defined as @xmath51 : { \left\lverta\right\rvert } \leq k \right\rbrace}.\ ] ] if appropriate , we will sometimes drop the @xmath52 and write @xmath53 .
we have defined a corresponding hierarchy of exponential families studied in @xcite : @xmath54 the elements of this hierarchy have nice interpretations .
it can be seen that the closure of the family @xmath55 contains exactly all probability distributions that factor .
this means that @xmath56 } p_i(x_i),\ ] ] where @xmath57 are the marginal distributions of @xmath58 .
generally , an element @xmath59 will allow a factorization as @xmath60 , { \left\lverta\right\rvert } = k } \phi_a(x_a),\ ] ] where @xmath61 depends only on @xmath62 .
however , the @xmath63 are not necessarily probability distributions and not unique .
note that @xmath64 , does not necessarily admit such a product structure .
we will clarify these definitions in the following simple consider the case @xmath65 .
the configuration space is given as @xmath66 the vector space of real valued functions @xmath15 is 4-dimensional and the probability measures form a 3-dimensional tetrahedron .
considering the hypergraphs of fixed interaction order and their exponential families , one has only two examples here : @xmath67 and @xmath68 , only the first being nontrivial .
( 7,7 ) ( 0,0 ) in the simplex of probability distributions.,title="fig : " ] ( 3.4,6.28)@xmath69 ( -0.75,3.7 ) @xmath70 ( 5.5,0.25 ) @xmath71 ( 6.9,1.25)@xmath72 ( 4.1,-0.23)@xmath73 ( -0.8,1.2)@xmath74 ( 5.5,1.6)@xmath75 figure [ fig : simplex ] shows the situation .
the exponential family @xmath76 is a two - dimensional manifold lying inside the simplex .
one should already think about this as a square ( the two dimensional cube ) molded into the simplex . in the following , we will study the interaction spaces more thoroughly by comparing different generating systems .
in this section , let @xmath30 be fixed . in statistics ,
different representations of exponential families have been considered , each of which has its own benefits and highlights different aspects .
we will review a number of these representations .
the particular choice of parity functions will allow us to make a link to coding theory . as we have introduced exponential families ,
the key concept is the interaction space @xmath77 which is sometimes also called tangent space to the exponential family .
this space completely characterizes the exponential family .
however , there is a choice of the parameterization of this space , which has been made differently in different fields . speaking in terms of linear algebra
, one has to choose a generating system of a linear space .
let @xmath78 be any finite generating system of @xmath77 .
each such choice gives a different parameterization of the exponential family and a different sufficient statistics .
the parameterization is identifiable if @xmath79 is a basis .
the exponential family is parameterized as @xmath80 where again @xmath81 is the normalization and @xmath82 equals the number of parameters . in statistical physics
the exponent is commonly called the energy . to each choice of @xmath79 there is a polytope constructed as follows .
consider the vectors @xmath83 each such vector has as its components the evaluation of every element in @xmath79 at @xmath84 .
the polytope is @xmath85 since @xmath30 contains all atoms , it can be seen that the polytope has @xmath86 vertices and the dimension equals the dimension of the exponential family . by applying some classical theorems from statistics , such as the existence and uniqueness of maximum likelihood estimates @xcite
, it can be seen that the points of the polytope are in one to one correspondence with points in the closure of the exponential family . as we have introduced it here , it is clear that the different choices of @xmath79 yield different representation of the same polytope in the sense that they are all affinely equivalent .
in particular , they have the same face lattice . the polytope @xmath87 encodes in its face lattice the combinatorial structure of the exponential family in the sense that a knowledge of the face lattice gives precise knowledge about the supports of elements in the closure of the exponential family .
however , direct computation is infeasible for real world problems . in statistical physics , and also for various inference methods
it is of interest to compute the free energy , given as the logarithm of the partition function .
there , variational principles and the techniques of legendre transform are applied . in this setting the points in the polytope are then the so called dual parameters .
see for instance @xcite .
we will review a number of choices for @xmath79 :
in statistical physics one considers so called potentials @xcite . a potential is a collection of functions @xmath88 , where @xmath89 and @xmath90 , such that the energy can be written as a linear combination hereof .
typically one has a distinguished state @xmath91 called the vacuum .
a potential is called _ normalized _ if @xmath92 as soon @xmath93 for some @xmath94 .
given a strictly positive distribution , a corresponding normalized potential exists and is unique . in our binary setting , choosing @xmath95 as the vacuum state , the normalized potential is given by the functions @xmath96 , where @xmath97 .
one has @xmath98 , and a basis of the interaction space is given by @xmath79 together with the constant function @xmath99 . expanding a function @xmath100 in terms of this basis
was called the @xmath101-expansion in the works of caianiello @xcite . in the case of pair interactions where the hypergraph is given by @xmath102
, the polytope @xmath87 coincides with the so called _ correlation polytope _ @xcite . extending the terminology to an arbitrary hypergraph @xmath30 ,
we call @xmath87 the _ moment polytope _ , as each point in it is the vector of moments of some distribution .
[ [ sec : marg - chi - repr ] ] marginals + + + + + + + + + one representation of an exponential family is given via the linear map that computes the marginals .
denote @xmath103 the set of inclusion maximal sets in @xmath30 .
consider the linear map @xmath104 that , for a given vector @xmath105 computes the set of its maximal marginals defined as @xmath106 when represented as a matrix with respect to the canonical basis , @xmath107 has rows indexed by pairs @xmath108 of a set @xmath109 and a configuration @xmath110 .
the columns are indexed by configurations @xmath111 .
each component then contains the value of the indicator @xmath112 .
denote the @xmath84-th column of this matrix as @xmath113 then , the exponential family is parameterized as @xmath114 in terms of these vectors , the polytope is commonly called the marginal polytope .
it is represented as 0/1 polytope embedded in a high dimensional space .
[ [ sec : basis - characters_desc ] ] an orthogonal basis of characters + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in the binary case @xmath115}$ ] , a natural basis for @xmath15 is given by the characters of @xmath8 .
here , we assume pointwise addition modulo 2 as the group operation . for
every subset @xmath116 define the function @xmath117 by @xmath118 where @xmath119 .
it can be seen that , if @xmath30 is a hypergraph , @xmath120 together with the constant function @xmath121 is an orthogonal basis of the interaction space @xmath77 .
this approach was followed in @xcite .
various people , starting with caianiello @xcite have called this the @xmath122-expansion .
note that if one considers random variables taking values in @xmath123 this basis equals the monomial basis @xmath124 \right\rbrace}$ ] considered above .
[ [ sec : basis - parity - funct ] ] a basis of parity functions + + + + + + + + + + + + + + + + + + + + + + + + + + + finally , we will introduce yet another basis of @xmath77 which is derived from the basis of characters .
to each @xmath125 $ ] , we define a vector in @xmath15 . @xmath126
the following proposition is easily checked : [ sec : fa_basis_prop ] let @xmath127 be the constant function @xmath128 .
the set @xmath129 is a basis of @xmath130 .
one crucial point about choosing this representation is that it gives , if the constant function is omitted , a full dimensional 0/1 polytope , the vertices of which form an additive group and thereby a linear code ( see proposition [ sec : parity - code ] ) . for all other choices of @xmath79 discussed in this section the image of @xmath131 is not a subgroup of @xmath132 or the multiplicative group @xmath133 . while in the construction of a hierarchical model we assumed a hypergraph
, the following polytope is an interesting object of study also in the general case of a pre - hypergraph : let @xmath30 be a pre - hypergraph .
we define @xmath134 if @xmath30 is a hypergraph , then this is affinely equivalent to the marginal polytope of the corresponding exponential family . in the case of the hypergraphs @xmath135
we write @xmath136 .
the rest of the paper is devoted to the study of this class of polytopes .
there is a well known @xcite affine equivalence between cut polytopes of graphs @xcite and binary marginal polytopes : namely , to each graph @xmath49 we can associate the hypergraph @xmath137 .
this is distinct from what was called a graphical model above , as not the cliques are considered .
some authors refer to the corresponding statistical model as a _ graph model_. from @xmath49 construct the coned graph @xmath138 with an additional vertex : @xmath139 and edges @xmath140 then , denoting the cut polytope of @xmath138 as @xmath141 one has @xmath142 using the representation in terms of the vectors @xmath143 , @xmath144 , the proof of this equivalence becomes a simple renaming of coordinates .
as remarked above , in the representation with monomials @xmath145 one finds the correlation polytope @xmath146 as a special case . from the last remark it follows that the cut - polytope of the complete graph @xmath147 is equal to @xmath148 .
there exists an affine equivalence between @xmath146 and @xmath149 called the _ covariance mapping _
it can be seen that this mapping generalizes to a mapping between binary marginal polytopes and the corresponding moment polytopes .
it therefore might be suitable to consider the parity representations @xmath150 of binary marginal polytopes for a generalization of cut - polytopes to arbitrary ( pre)-hypergraphs . using the geometry software ` polymake ` @xcite , one can compute linear descriptions of polytopes . as an example , we give here the f - vectors of @xmath151 for the cases @xmath152 . for @xmath153 ,
the f - vector is too complicated to be computed by the brute force approach of ` polymake ` .
however , waiting sufficiently long , one can get the 6800 facet defining inequalities of @xmath154 and the 3835488 facets of @xmath155 . in tables [ tab:3node ] and [ tab:4node ] , we give the f - vectors of @xmath151 for @xmath152 , computed using ` polymake ` .
the rows label the dimension of the faces , the columns the value of @xmath156
. the reader might wonder about the fact that the face lattices of @xmath157 are up to a certain dimension isomorphic the face lattice of the simplex .
this property , commonly called neighborliness , follows from a general result in @xcite .
the last row refers to whether the polytope is simple or not .
.face structure of @xmath158 [ cols="<,^,^,^",options="header " , ] when the generator matrix is chosen , one can apply the method given in section [ sec : codingtheory ] to construct a pre - hypergraph . for our first generator matrix
we get back the hypergraph we started with , for the second choice we read of the pre - hypergraph @xmath159 where we have introduced new units @xmath160 corresponding to the first three columns .
this shows the ambiguity due to the choice of a particular generator matrix if only the code is given .
thomas kahle is supported by the volkswagen foundation , nihat ay is supported by the santa fe institute .
the authors thank bernd sturmfels for many valuable comments on the subject .
we are grateful to the anonymous referees for their reports which led to significant improvement of the paper . | in this paper , we explore a connection between binary hierarchical models , their marginal polytopes and codeword polytopes , the convex hulls of linear codes . the class of linear codes that are realizable by hierarchical models is determined .
we classify all full dimensional polytopes with the property that their vertices form a linear code and give an algorithm that determines them . |
as follows from the relativistic mean field model @xcite strong time
dependent meson fields are generated in the course of relativistic heavy
ion collisions . within the framework of this model several new collective phenomena were predicted : the filamentation instability of interpenetrating nuclei @xcite and the spontaneous creation of the baryon antibaryon ( @xmath10 ) pairs in a superdense baryon
rich matter @xcite . using the approach developed in papers on the pion @xcite and photon @xcite bremsstrahlung
we suggested recently @xcite a new mechanism of the @xmath10 pair production by the collective bremsstrahlung of meson fields in relativistic heavy ion collisions .
these pairs may be produced at sufficiently high bombarding energies when characteristic fourier frequencies of meson fields exceed the energy gap between the positive and negative energy levels of baryons . in the lowest order approximation
the production of the @xmath10 pair may be considered as a two step process @xmath11 . here
@xmath12 denotes the projectile ( target ) nucleus and @xmath13 indicates the off mass shell vector meson . the first step in
the above reaction is the virtual bremsstrahlung producing virtual mesons with masses @xmath14 ( @xmath15 is the baryon mass ) .
the second step is the conversion of a vector meson into @xmath10 pairs .
such a process is suppressed for a
@xmath0 meson which has the mass @xmath16 gev and the relatively small width @xmath17 mev .
it is clear that analogous bremsstrahlung mechanism may produce also pions ( by decays of quasireal mesons @xmath18 ) and low mass dileptons ( @xmath19 ) . in this work
we study the bremsstrahlung of vector meson fields originated from the collective deceleration of the projectile and target nuclei at the initial stage of a heavy ion collision .
the various channels of the bremsstrahlung conversion , including the production of the @xmath20 pairs , pions and dileptons are considered with emphasize to their observable signals .
by the analogy to the walecka model we introduce the vector meson field @xmath21 coupled to the 4current @xmath22 of nucleons participating in a heavy ion collision at a given impact parameter .
the equation of motion defining the space time behavior of @xmath21 may be written as ( @xmath23 ) ( ^_+m_^2)^(x)=g_v j^(x ) , where @xmath24 is the @xmath25 coupling constant . in the mean field approximation
the quantum fluctuation of @xmath26 are disregarded and the vector meson field is purely classical . from one
can see that excitation of propagating waves in a vacuum ( bremsstrahlung ) is possible if the fourier transformed baryonic current j^(p)=^4x j^(x ) e^ipx is nonzero in the time like region @xmath27 . in the following we study the bremsstrahlung process in the lowest order approximation neglecting the back reaction and reabsorption of the emitted vector mesons , i.e. treating @xmath26 as an external current
. from one can calculate the energy flux of the vector field at a large distance from the collision region @xcite .
this leads to the following formulae for the momentum distribution of real @xmath0mesons emitted in a heavy ion collision @xcite e_=s(e _ , ) , where @xmath28 and s(p)=|j_^*(p)j^(p)| is a source function . in our model the latter
is fully determined by the collective motion of the projectile and target nucleons . to take into account the off mass shell effects we characterize virtual @xmath0 mesons by their mass @xmath29 and total width @xmath30 .
the spectral function describing the deviation from the on mass shell may be written as ( m)=. to calculate the distribution of virtual mesons in their 4momentum @xmath31 we use the formulae @xcite = ( m ) s(p ) , where @xmath32 . in the limit
@xmath33 one can replace @xmath34 by @xmath35 . in this case
becomes equivalent to the formulae ( [ rvms ] ) for the spectrum of the on mass shell vector mesons .
below we consider the most important channels of the virtual @xmath0 decay : @xmath36 , @xmath37 , @xmath38 .
the total width @xmath39 can be decomposed into the sum over the partial decay widths @xmath40 : = _ i .
the distribution over the total 4momentum of particles in a given decay channel may be written as = b(_i ) , where @xmath41 is the branching ratio of the @xmath42th decay channel .
the latter is a function of the invariant mass of the decay particles @xmath29 . to calculate the 4vectors @xmath43 defining the source function @xmath44 we assume the simple picture of a high energy heavy ion collision suggested in ref .
@xcite below we consider collisions of identical nuclei ( @xmath45 ) at zero impact parameter . in the equal velocity frame
the projectile and target nuclei initially move to each other with the velocity @xmath46 , where @xmath47 is the c.m . bombarding energy per nucleon . in the frozen density
approximation @xcite the internal compression and transverse motion of nuclear matter are disregarded at the early ( interpenetration ) stage of the reaction . within this approximation
the colliding nuclei move as a whole along the beam axis with instantaneous velocities @xmath48 .
the projectile velocity @xmath49 is a decreasing function of time , chosen in the form @xcite ( t)=v_f+ , where @xmath50 is the effective deceleration time and @xmath51 is the final velocity of nuclei ( at @xmath52 ) . in our approximation
the fourier transforms @xmath43 are totally determined by the projectile trajectory @xmath53
@xcite : j^0(p)=j^3(p)= 2a_-^t e^ip_0tf ( ) , where @xmath54 and @xmath55 are , respectively , the longitudinal and transverse components of the three momentum @xmath56 , @xmath57 is the density form factor of the initial nuclei f(q)^3r(r ) e^-i .
the time integrals in were calculated numerically assuming the woods
saxon distribution of the nuclear density @xmath58 . according to eqs .
( [ sfu ] ) , ( [ ftre ] ) the source function @xmath44 vanishes at @xmath59 . as a result , at high bombarding energies single particle distributions have a dip in a central rapidity region ( see figs .
23 and ref .
the two hump structure of the rapidity spectra is a consequence of the incomplete mutual stopping of nuclei at the initial stage of a heavy ion collision . on the other hand , the conventional mechanism of particle production in incoherent hadron
hadron collisions results in rapidity distributions of pions and antiprotons with a single central maximum even at high bombarding energies @xcite . in this work
we use the same choice of the coupling constant @xmath24 and the stopping parameters @xmath60 as in ref .
@xcite in particular , it is assumed that @xmath50 equals one half of the nuclear passage time = r/ , where @xmath61 is the geometrical radius of initial nuclei . instead of @xmath51
we introduce the c.m .
rapidity loss @xmath62 : v_f=. in the case of a central au+au collision we assume the energy independent value @xcite @xmath63 for @xmath64 gev and full stopping ( @xmath65 ) for lower bombarding energies .
the partial width of the @xmath66 decay channel is calculated assuming that @xmath67 is proportional to the three body phase space volume @xcite .
the normalization constant is determined from the condition that @xmath68 equals the observable value @xmath69 at @xmath70 .
calculation of the @xmath71 matrix element in the lowest order approximation in @xmath24 gives the result ( _ n)= ( 1+)(m-2m_n ) , where @xmath72 .
after substituting ( [ wnnb ] ) into and omitting the second term in the denominator of @xmath34 one arrives at the distribution over the pair 4momentum obtained earlier in ref .
@xcite the dilepton production is studied by calculatingthe matrixelements of the process @xmath73 where @xmath74 is a virtual photon .
this calculation leads to the result ( @xmath75 ) @xcite = ( ) ^6 ( m-2m_l ) , where @xmath76 is the lepton mass .
below we present the results of numerical calculations obtained within the model described in the preceding section .
some of the model predictions concerning the @xmath10 production have been already published in ref .
@xcite + + fig .
1 shows the @xmath77 and @xmath78 multiplicities as functions of the bombarding energy in the case of central au+au collisions . for comparison ,
we show experimental data on the @xmath77 multiplicities in the 11.6 agev / c au+au ( circle ) @xcite and 160 agev pb+pb ( triangle ) @xcite central collisions .
one can see that the multiplicity of pions produced by bremsstrahlung exhibits a rapid growth between the ags ( @xmath79 agev ) and the sps ( @xmath80 agev ) energies and saturates above the rhic ( @xmath81 agev ) energy region .
it is interesting that the bremsstrahlung component of pion yield becomes comparable with pion production in incoherent hadron
hadron collisions @xcite already at the sps bombarding energies .
note , however , that actual pion and antiproton yields , especially for heavy combinations of nuclei , may be reduced due to the absorption and annihilation neglected in the present model .
the results on the @xmath77 rapidity spectra are represented in figs .
the spectra are calculated in the limit of the on mass shell @xmath0 mesons , i.e. assuming @xmath82 .
here we use the kinematic formulae connecting the pion spectrum and the primordial distribution of vector mesons , , suggested in ref .
@xcite similarly to the case of antiprotons @xcite , the pion rapidity spectrum has a pronounced dip at @xmath83 .
the two hump structure of the @xmath84 and @xmath78 spectra may serve as a signature of the bremsstrahlung mechanism . according to fig .
3 this structure can be seen only at high enough bombarding energies .
4 shows the distributions over invariant masses of particles produced in different channels of bremsstrahlung conversion in the case of a central au+au collision at the sps energy .
note that the mass spectrum of e@xmath1e@xmath2 pairs created by the bremsstrahlung mechanism has a strong peak at invariant masses below the @xmath0 meson mass . on the other hand ,
attempts to explain the low mass dilepton yield by the conventional incoherent mechanisms ( e.g. due to the @xmath85 processes ) strongly underestimate the observable data @xcite .
in this work we have shown that the coherent bremsstrahlung of the vector meson field may be an important source of particle production already at the sps bombarding energies .
the observable signals of this mechanism may be the two hump structure of pion and antibaryon rapidity spectra as well as the enhanced yield of low mass dileptons .
the sharp energy and a dependence of pion , dilepton and antibaryon excitation functions can be also a signature of the considered mechanism .
the latter may be responsible , at least partly , for a rapid increase of the pion multiplicity observed in transition from the ags to sps energy @xcite .
the authors thank yu.b .
ivanov , s. schramm and l.a .
winckelmann for valuable discussions .
this work has been supported in part by the eu intas grant no .
we acknowledge also the financial support from gsi , bmft and dfg . | according to the walecka mean field theory of nuclear interaction the collective mutual deceleration of the colliding nuclei gives rise to the bremsstrahlung of real and virtual @xmath0mesons .
it is shown that decays of these mesons may give a noticeable contribution to the observed yields of the baryon
antibaryon pairs , dileptons and pions .
excitation functions and rapidity distributions of particles produced by this mechanism are calculated under some simplifying assumptions about the space
time variation of meson fields in nuclear collisions .
the calculated multiplicities of coherently produced particles grow fast with the bombarding energy , reaching a saturation above the rhic bombarding energy . in the case of central au+au collisions
the bremsstrahlung mechanism becomes comparable with particle production in incoherent hadron
hadron collisions above the ags energies .
the rapidity spectra of antibaryons and pions exhibit a characteristic two hump structure which is a consequence of incomplete projectile target stopping at the initial stage of the reaction .
the predicted distribution of e@xmath1e@xmath2 pairs has a strong peak at invariant masses gev
. 0.5 cm -1 cm # 1@xmath3 # 1eq .
( [ # 1 ] ) # 1 # 1#2 + # 2 # 1#2#3#4#1 * # 2 * , # 3 ( # 4 ) * particle production by time
dependent meson fields + in relativistic heavy ion collisions * + * i.n .
mishustin@xmath4 , l.m .
satarov@xmath5 , h. stcker@xmath6 and w. greiner@xmath6 * = @xmath7_the kurchatov institute , 123182 moscow ,
_ + @xmath8 + @xmath9 + |
the mechanism of diffusive shock acceleration ( dsa ) of relativistic particles ( krymsky @xcite , axford et al .
@xcite , bell @xcite , blandford & ostriker @xcite ) is generally accepted as the most likely paradigm for production of galactic cosmic rays ( cr ) in supernova remnants ( snrs ) . over the last 30 years a significant progress has been achieved in the development of theoretical models and understanding of the basic features of dsa ( see e.g. malkov & drury @xcite , schure et al .
@xcite for a review ) . on the other hand
, the recent detailed studies of spectral and morphological features of young snrs , first of all in the x - ray and very - high - energy ( or tev ) gamma - ray band , provide excellent observational material for development of detailed numerical models of acceleration and radiation of relativistic electrons and protons in young snrs .
these observations generally confirm in general terms the predictions of dsa .
in particular the synchrotron x - radiation observed from several young snrs implies an existence of multi - tev electrons which is naturally explained by dsa .
the detection of tev gamma - rays from snrs , like cas a , rx j1713.7 - 3946 , vela jr .
, rcw 86 ( see hinton & hofmann @xcite ; rieger et al .
@xcite for a recent review ) give a more direct and unambiguous information about the effective acceleration of particles , electrons and/or protons , in snrs to energies exceeding 100 tev .
in this paper we conduct detailed study of acceleration of electrons and protons with an emphasis on the spectral and morphological features of high energy radiation produced by these particles in the young supernova remnant cas a. for that purpose we use a new numerical code of nonlinear diffusive shock acceleration developed by one of us in collaboration with v. ptuskin ( zirakashvili & ptuskin @xcite ) .
this model can be considered as a natural development of existing numerical codes ( see e.g. berezhko et al.@xcite , kang et al @xcite ) , with new additional elements which despite their strong impact on the overall picture of acceleration in general , and on the properties of high energy radiation of snrs in particular , have been ignored in the past .
namely , in our treatment , the solution of spherically symmetric hydrodynamic equations is combined with the transport and acceleration of relativistic particles by the _
forward _ shock and _ reverse _ shocks ( fs and rs , respectively ) .
the nonlinear response of energetic particles via their pressure gradient results in a self - regulation of acceleration efficiency .
the detailed calculations of radio , x - ray , gamma - ray emission components conducted within a self - consistent treatment of the particle acceleration by both forward and reverse shocks should allow a direct comparison of the observed spectral and morphological features with model predictions .
the inclusion of the radiation components related to the reverse shock seems to be a rather obligatory condition , at least for the specific case of cas a. in this regard we note that the parameters that characterize the reverse shock can be significantly different compared to the parameters of the forward shock .
therefore , the properties of radiation components from the reverse and forward shocks are also expected to be significantly different . in particular
, the magnetic field in the reverse shock can be very small which would dramatically increase the contribution of the ic component compared to the hadronic ( @xmath3-decay ) component of gamma - rays . because of the stronger magnetic field , in the forward shock just opposite relation is expected between the contributions of electrons and protons to production of high energy gamma - rays . an interesting feature of cas a is the non - negligible contribution of the decay of radioactive elements in the ejecta to the production of suprathermal electrons and positrons as a potentially important `` injection material '' .
the acceleration of these particles by the reverse shock ( zirakashvili & aharonian @xcite , ellison et al .
@xcite ) can result in a rather high energy content of leptonic component in snrs .
in particular this can explain the high electron to proton ratio found in cas a ( atoyan et al .
@xcite ) . the results and conclusion of this study have a rather general character and can be applied to different snrs . in this paper
the model is used to interpret the multi - wavelength properties of cas a - one of the youngest supernova remnants in our galaxy .
the high quality x - ray images and energy spectra , as well as the coverage of gamma - ray observations from low to very high energies , provide adequate observational material to conduct detailed theoretical studies of acceleration and radiation processes in this unique source .
the paper is organized as follows . in section 2
we briefly summarize multi - wavelength observations of cas a. the short description of the model is given in section 3 .
the results of modeling of the broad - band emission are presented and discussed in sections 4 , 5 , 6 and 7 .
finally , in sections 8 and 9 we discuss and summarize the obtained results .
the supernova explosion related to cas a is likely to be linked to the event observed by flamsteed in 1680 ( ashworth @xcite ) .
recently krause et al .
@xcite reported the detection of its light echo with a spectrum similar to the spectra of iib supernova 1993j .
therefore one may assume that snr cas a was produced in a iib type supernova explosion . generally , the progenitors of such explosions are red super giants ( rsg ) which already have lost their hydrogen envelope via a powerful stellar wind ( chevalier @xcite ) .
cas a has been extensively observed at radio wavelengths ( bell et al .
@xcite ; tuffs @xcite ; braun et al .
@xcite ; anderson et al .
@xcite ; kassim et al .
@xcite ; etc . ) .
the radio spectrum is close to power - law @xmath4 with a spectral index @xmath5 ( baars et al .
the main fraction of radio - emission comes from the bright radio - ring with an angular radius close to @xmath6 and from the faint outer radio - plateau with an angular radius @xmath7 . for a distance to the source @xmath8 kpc ( reed et al .
@xcite ) , the corresponding spatial radii are 1.7 pc and 2.5 pc , respectively .
in addition to the large - scale structures , several hundreds of very compact and bright radio - knots with steeper spectra ( anderson et al .
@xcite ) are present in the radio - shell .
it is believed that the bright radio - ring is related to the reverse shock propagating into the supernova ejecta .
a large number of fast - moving knots ( fmk ) are observed also in optics ( e.g. fesen et al .
@xcite ) are attributed to densest clumps of ejecta which are not strongly decelerated after the supernova explosion .
this hypothesis agrees with x - ray observations which show that the x - ray line emitting shell roughly coincides with the bright radio ring .
the x - ray emitting plasma is rich in o , si , ar , ca and fe dominated by the contribution from oxygen ( fabian et al . @xcite ; markert et al .
@xcite ; vink et al .
@xcite , hughes et al . @xcite ; willingale et al .
@xcite ; hwang & laming @xcite ; laming & hwang @xcite ) .
besides the line x - ray emission , cas a shows a hard x - ray continuum extending up to 100 kev with a photon index @xmath9 as measured by _ bepposax _ ( vink et al .
@xcite ) , _ integral _ ( renaud et al .
@xcite ) and _ suzaku _ observations ( maeda et al .
@xcite ) satellites . unfortunately , because of limited angular resolution of these instruments , the production region(s ) of the component of radiation can not be localized . at low energies ,
thin and faint nonthermal x - ray filamentary structures have been found at the periphery of the remnant close to the boundary of the radio plateau ( gotthelf et al .
these filaments correspond to the position of the forward shock propagating in the circumstellar medium .
the proper motion of these filaments has been measured by the _
chandra _ observations allowing the estimate of the forward shock velocity of about 4900 km s@xmath10 ( patnaude & fesen @xcite ) .
on the other hand , the inner shell in the radio and x - ray images can be naturally attributed to the reverse shock propagating in the supernova ejecta , assuming that the electrons are accelerated also at the reverse shock ( uchiyama & aharonian @xcite , helder & vink @xcite ) .
generally , the reverse shock is not treated as an efficient accelerator , because the magnetic field of ejecta might be very weak due to the large expansion factor of the exploded star .
however , similar to the case of the forward shock , the magnetic field can be significantly amplified also at the reverse shock ( see ellison et al .
@xcite ) , for example , in the course of the nonresonant streaming instability as suggested by bell @xcite .
this implies that we should expect gamma - rays from both forward and reverse shocks the contributions of which however can not be separated by current gamma - ray telescopes both in gev and tev energy bands
. very high energy gamma - ray emission from cas a has been discovered by the hegra system of atmospheric cherenkov telescopes ( aharonian et al .
@xcite ) , and later confirmed by the magic ( albert et al .
@xcite ) and veritas ( acciari et al .
@xcite ) collaborations .
the fluxes published by three groups are in a reasonably good agreement with each other , and indicate on a not - very - hard and not - very - soft energy spectrum with a photon index @xmath11 and a maximal energy of detected photons 5 tev . at low energies , a weak gamma - rays signal from cas
a has been discovered after the first year of observations with the _ fermi _ gamma - ray space telescope ( abdo et al .
while the flux and the energy spectrum reported at energies above 0.5 gev are quite informative for constraining certain model parameters , they are not sufficient for a robust conclusions concerning , in particular , the origin of radiation . in this regard , the derivation of the energy spectrum at energies below 1 gev seems to be crucial for identification of the radiation mechanism(s ) . in particular , a sharp decline of the spectrum below 1 gev would indicate the dominance of hadronic interactions , at least in this energy band .
motivated by the importance of such spectral measurements , we attempted to reanalyze the _ fermi _ lat data based on much larger statistics accumulated over 4 yr observations , and using the recent _ fermi _ lat software package which allows the extension of the analysis down to 100 mev .
the approach used in this paper for analysis of the _ fermi _ lat data is described in @xcite .
we have selected the events with energy between 60 mev and 100gev and applied the usual filters recommended by the _ fermi _
lat collaboration ( removing the intervals when the rocking angle of the lat was greater than 52 deg or when parts of the region - of - interest ( roi ) were observed at zenith angles larger than 100 deg ) . to derive the energy spectrum we applied the maximum likelihood method in 13 independent energy bins from 100 mev to 100 gev
; the galactic and isotropic diffuse emission was modeled using the tools _
gal-2years7v6-v0 _ and _ iso - p7v6source_. during the broad - band fit , all sources in the second fermi catalog ( 2fgl ) in the field of view have been included in likelihood model .
points with significance of more than 3@xmath12 are shown in fig .
6 , 7,8 and 14 .
it should be noted that below 200 mev , the spectrum determination suffers heavily from the background subtraction due to the uncertainties in the model of the galactic diffuse emission .
applying different models generated by the galprop simulator , we obtained a 15 % larger flux that can be account by systematic errors .
the obtained spectral energy distribution ( sed ) shows a significant decrease of the flux below 1 gev which is expected in the sed of gamma - rays from decays of neutral pions produced at interactions of accelerated protons and nuclei with the ambient gas .
throughout this paper we use the approach which lies on a nonlinear treatment of particle acceleration by strong shock waves proposed by zirakashvili & ptuskin @xcite .
non - steady hydrodynamical equations for the gas density @xmath13 , gas velocity @xmath14 , gas pressure @xmath15 , and the equation for the quasi - isotropic cr momentum distribution @xmath16 in the spherically symmetrical case are solved numerically with the use of a finite - difference method .
thermal protons are injected at the forward shock with a radius @xmath17 . since it is believed that the ejecta material in cas a consists mainly on heavy elements , the oxygen ions are injected at the reverse shock with a radius @xmath18 .
the injection efficiency that is the fraction of particles injected into dsa at the reverse and forward shocks @xmath19 and @xmath20 is taken to be independent of time , and the injection momenta are @xmath21 , @xmath22 . here
@xmath23 and @xmath24 are the mass of the proton and oxygen ion respectively .
we normalize the momentum distribution of oxygen ions @xmath25 to the number density of nucleons .
thus the number density of oxygen ions with atomic mass @xmath26 is given by @xmath27 . to obtain the sub - shock compression ratio close to 2.5 - 2.7 , for simulations of cosmic ray modified shocks
we adopt the injection efficiency @xmath28 .
this is in agreement with calculations of quasi - parallel collisionless shocks ( zirakashvili @xcite ) and is supported by radio - observations of young extragalactic ib / c snrs ( chevalier & fransson @xcite ) . the amplified magnetic field is rather large in these snrs and the synchrotron radio - emission is produced by electrons with energies below 1 gev . the observed radio - flux @xmath29 has a power - law dependence on the frequency @xmath30 .
note that since the radiative losses of radio electrons are negligible , this corresponds to a @xmath31 type energy spectrum of accelerated electrons .
such a spectrum can be formed at the sub - shock with a compression ratio 2.5 . in older remnants ,
the magnetic field strength drops and the energy spectra of gev electrons become flatter due to the higher compression ratio `` seen '' by these particles .
this change of the electron energy distribution is reflected in the spectrum of radio emission in snr shells the ion injection may be suppressed at oblique shocks ( see vlk et al .
@xcite for the discussion of this topic ) .
this can be realized in cas a where the forward shock propagates in the stellar wind with an azimuthal magnetic field .
that is why below we consider two extreme cases with @xmath32 ( hadronic model ) and @xmath33 ( leptonic model ) .
usually the pressure of energetic electrons is neglected , i.e. the electrons are treated as test particles . however , recently it has been proposed that gamma - rays from the radioactive decay of @xmath34ni may produce a large number of energetic electrons via compton scattering in the circumstellar medium .
in addition the decays of @xmath2ti can directly supply energetic electrons and positrons in supernova ejecta ( see zirakashvili & aharonian @xcite for details ) .
these energetic particles are further accelerated by the forward and reverse shocks . under these conditions , the reverse shock may be modified by the pressure of electrons ( positrons ) . for this reason in calculations
we include the contribution of relativistic electrons to the overall pressure .
the evolution of electron distribution is described by an equation similar to transport equation for protons , but with additional terms describing synchrotron , coulomb , bremsstrahlung and ic losses . in the scenario of electron injection through the decays of radioactive nuclei @xmath34ni and @xmath2ti ,
the amount of positrons and electrons injected at the forward and reverse shock is described by the injection efficiencies @xmath35 and @xmath36 respectively . for the forward shock
the injection efficiency is determined by the probability of appearance of an energetic electron due to the compton scattering of gamma - rays from @xmath34co decay and is given by ( zirakashvili & aharonian @xcite ) @xmath37 here @xmath38 is the mass of @xmath34ni just after the explosion and @xmath39 is the distance from the center of the remnant expressed in parsecs .
@xmath34co is produced via decay of @xmath34ni . for the reverse shock
the injection efficiency of electrons and positrons produced directly from the decay of @xmath2ti can be written as ( zirakashvili & aharonian @xcite ) @xmath40 here @xmath41 is the mass of ejecta , @xmath42 is the mass of @xmath2ti just after the explosion , and @xmath43 yr is the half - time of @xmath2ti .
it was taken into account that one electron and 0.94 positrons appear per a decay of @xmath2ti .
cr diffusion is determined by magnetic inhomogeneities .
strong streaming of accelerated particles changes medium properties in the shock vicinity .
in particular , the cr streaming instability in young snrs results in a high level of magnetohydrodynamical ( mhd ) turbulence ( bell @xcite ) and even in the amplification of magnetic field ( bell @xcite ) . due to this effect protons can be accelerated to energies beyond the the so - called lagage and cesarsky upper limit , @xmath44 tev ( lagage & cesarsky @xcite ) .
the magnetic field amplified upstream the shock is enhanced further via compression in the shock transition region .
it can even play a dynamical role downstream of the shock .
we take magnetic pressure and magnetic energy flux into account downstream of the shock .
this is a new element in comparison with the approach of zirakashvili & ptuskin @xcite where the magnetic field spatial distribution was prescribed .
the magnetic field was transported in the downstream region as the gas with adiabatic index @xmath45 .
its impact on the shock dynamics was also taken into account via hugoniot conditions .
upstream of the forward shock where dynamical effects of the magnetic fields are small , the coordinate dependence of the magnetic field @xmath46 can be prescribed as : @xmath47 here @xmath48 and @xmath49 are the undisturbed gas density at the shock position and the density of the medium where the shock propagates respectively , @xmath50 is the speed of the forward shock .
the parameter @xmath51 is similar to the alfvn much number of the shock and determines the value of the amplified magnetic field strength far upstream of the shock . in the shock transition region
the magnetic field strength is increased by a factor of @xmath52 , where @xmath53 is the shock compression ratio .
an expression similar to eq .
( 3 ) is used also in the upstream region of the reverse shock .
below we use the adiabatic index of isotropic random magnetic field @xmath54 . for this value of the adiabatic index
the magnetic pressure @xmath55 is three times smaller than the magnetic energy density ( @xmath56 ) .
cr advective velocity may differ from the gas velocity by the value of the radial component of the alfvn velocity @xmath57 calculated in the isotropic random magnetic field : @xmath58 .
the factor @xmath59 describes the possible deviation of the cosmic ray drift velocity from the gas velocity .
we use values @xmath60 and @xmath61 upstream of the forward and reverse shocks respectively , where alfvn waves are generated by the cosmic ray streaming instability and propagate in the corresponding directions .
the damping of these waves heats the gas upstream of the shocks ( see mckenzie & vlk @xcite ) and limits the total compression ratios of cr modified shocks . in the downstream region of the forward and reverse shock at @xmath62 we put @xmath63 and therefore @xmath64 .
we use the diffusion coefficients @xmath65 and @xmath66 at the forward and reverse shocks in the following form @xmath67 were @xmath68 is the bohm diffusion coefficient . since it is expected that the highest energy particles are scattered by small - scale magnetic fields generated in the course of the non - resonant streaming instability , their diffusion coefficient is larger than the bohm diffusion coefficient and is proportional to @xmath69 ( dolginov & toptygin @xcite ) .
this effect is described by the second term in parenthesis where @xmath70 and @xmath71 are the momenta which separate these two different regimes of diffusion .
the parameter @xmath72 describes the possible deviations of diffusion coefficient from the one achieved in the regime of bohm diffusion at small energies .
it is expected that @xmath73 because small energy particles can be resonantly scattered only by a fraction of the magnetic spectrum . throughout the paper
we use the value @xmath74 . note that at high energies the particle diffusion is determined by the small - scale magnetic fields , and the values of @xmath75 in eq .
( 4 ) are adjusted to reproduce the gamma and x - ray observations ( see below ) .
therefore the results of calculations do not strongly depend on the chosen value of @xmath74 . in real situations
the level of mhd turbulence drops with distance upstream the shock , so the diffusion could be quite fast there .
the characteristic diffusive scale of highest energy particles is a small fraction ( @xmath76 ) of the shock radius ( see zirakashvili & ptuskin @xcite ) , and is determined by the generation and transport of mhd turbulence in the upstream region . the value @xmath77 is determined by the ratio of diffusion coefficient @xmath78 in the circumstellar medium and diffusion coefficient @xmath79 in the vicinity of the shock .
the mhd turbulence is amplified exponentially in time before the shock arrival from the background level by the cosmic ray streaming instability .
this process is not modeled in the present study ; we simply multiply the cr diffusion coefficient @xmath80 to the additional factor @xmath81 upstream the forward shock and to a similar factor @xmath82 upstream the reverse shock .
the characteristic range of variation of @xmath83 is @xmath84 ( zirakashvili & ptuskin @xcite ) .
below we use the value @xmath85 . in supernova remnants ,
the plasma is heated to kev temperatures .
generally , in young snrs the thermal electrons are not in equilibrium with ions . here
we assume that the exchange between electrons and ions proceeds at the minimum rate , i.e. through the coulomb collisions ( spitzer @xcite ) .
( thick solid line ) , the reverse shock radius @xmath86 ( thick dashed line ) , the forward shock speed @xmath50 ( thin solid line ) , the reverse shock speed @xmath87 ( thin dashed line ) , the reverse shock speed in the ejecta frame @xmath88 ( thin dotted line ) .
the ratio of the energy released in cosmic rays to the total energy of the supernova explosion @xmath89 ( dotted line ) is also shown .
, width=283 ] yr . in the same figure
the positions of the forward and reverse shocks , ( fs and rs , respectively ) as well as of the contact discontinuity ( cd ) are shown . ,
it is expected that a circumstellar medium around progenitors of iib supernova is strongly nonuniform . at the main sequence ( ms )
phase , the stellar wind of progenitor creates a rarefied bubble in the surrounding medium .
later the part of this bubble is filled with a dense gas ejected by progenitor at the red super giant ( rsg ) phase of the stellar evolution ( see chevalier @xcite for a review ) .
typically , the progenitors of iib supernova are red - super giants ( rsg ) which have lost almost all hydrogen envelope vie powerful stellar wind with mass - loss rate @xmath90 yr@xmath10 .
it is likely , however , that the progenitor of cas a prior the explosion had a short wolf - rayet or blue supergiant stage . at this stage
the progenitor emits a tenuous and fast stellar wind that creates a low density bubble around the progenitor .
the fast wind also can compress an inner part of the external slow rsg wind . as a result
, a narrow high - density rsg shell is formed at a distance @xmath91 . for cas
a such a model has been proposed by chevalier & liang @xcite .
it is possible that almost all progenitors of iib supernova pass through this stage .
since the progenitor of the famous type iip supernova 1987a also has been in the blue super giant stage , we perhaps can extend this conclusion to other core - collapse supernova progenitors . the hydrodynamical model with the circumstellar bubble was also considered for cas a snr by borkowsky et al .
@xcite , by hwang & laming @xcite and by schure et al .
@xcite .
stellar evolution models for low mass iib supernova progenitors ( gregory @xcite ) do indicate presence of the blue supergiant stage .
the blast wave of iib snr 1993j has been propagating in the dense rsg wind one decade after explosion and later entered into a low density region ( weiler et al @xcite ) .
the main part of the mass lost by the progenitor of 1993j supernova is still outside the snr shell . in any case
the forward shock of cas a propagates into the stellar wind .
a steady stellar wind has a @xmath92 density profile .
the mass density of the stellar wind is given by @xmath93 here @xmath94 and @xmath95 are the speed and mass - loss rate of the stellar wind respectively . for the rsg wind speed we will use the characteristic value of @xmath96 km s@xmath10 which agrees with the measured motion of hi absorption features observed in the direction of cas a ( reynoso et al .
it should be noted that most of measurements show that most of rsgs have lower speeds .
however the wind speeds are higher for rsgs with high mass - loss ( see tables 1 , 2 of mauron & josselin @xcite ) .
since the forward shock speed is much larger compared to the wind speed , the results depend only on the ratio @xmath97 . for simplicity
we assume that inside the shell at @xmath98 the gas density is given by the same eq .
( 5 ) with the ratio @xmath97 a factor of 100 lower in comparison with the ratio in the outlaying rsg wind .
two regions are separated by the thin gas shell at @xmath99 with the mass equal to the mass of the swept up rsg wind . for the numerical modeling
we assume that the shell and the transition region have the thickness 0.1@xmath91 .
the actual density distribution is more complex inside the shell ( see schure et al .
however it influences only on the early evolution of snr when the forward shock propagates in this region .
the results obtained at later epochs depend only on the shell radius @xmath91 and the rsg wind density because of the low gas density in the bubble region . &
@xmath100&@xmath101&@xmath102&@xmath103&@xmath104&@xmath105&@xmath106&@xmath107&@xmath108&@xmath109&@xmath110&@xmath111&@xmath112&@xmath113&@xmath114&@xmath115&@xmath116 + h1 & 2.0 & 9 & 2.5 & 3.4 & 1.5 & 4.5 & 8 & 0.40 & 1.2 & 0.69 & 1.68 & 2.07 & 5.8 & 0.77 & 4.2 & 1.16 & 0.57 + h2 & 2.0 & 9 & 2.5 & 3.4 & 0 & 5 & 5 & 0.36 & 1.2 & 0.67 & 1.82 & 1.90 & 6.0 & 4.0 & 1.4 & 1.0 & 0.27 + l1 & 2.0 & 9 & 2.5 & 3.4 & 0 & 10 & 10 & 1.0 & 2.0 & 0.30 & 1.64 & 1.74 & 5.7 & 3.1 & 1.8 & 0.78 & 0.12 + @xmath117mass of ejecta , solar masses + @xmath118 power - law index of ejecta density distribution + @xmath119forward shock radius , pc + @xmath120distance to snr , kpc + @xmath121rsg shell radius , pc + @xmath122parameter of magnetic amplification at the forward shock + @xmath123parameter of magnetic amplification at the reverse shock + @xmath124undisturbed hydrogen number density at the forward shock position , @xmath125 + @xmath126explosion energy , 10@xmath127 erg + @xmath128number density of unshocked ejecta , nucleons @xmath125
+ @xmath129reverse shock radius , pc + @xmath130radius of the contact discontinuity , pc + @xmath131forward shock speed , 10@xmath132 km s@xmath10 + @xmath133reverse shock speed in the laboratory frame , 10@xmath132 km s@xmath10 + @xmath134reverse shock speed in the ejecta frame , 10@xmath132 km s@xmath10 + @xmath135magnetic field just downstream of the forward shock , mg + @xmath136magnetic field just downstream of the reverse shock , mg & @xmath137&@xmath138&@xmath139 & @xmath140&@xmath141 & @xmath142&@xmath143&@xmath144 & @xmath145 & @xmath146&@xmath147&@xmath148&@xmath149 + h1 & 0.01 & 0.01 & @xmath150 & @xmath151 & 0.94 & 9.4 & @xmath152 & 40 & @xmath153 & 0.004 & 0.25 & 4.3 & 4.9 + h2 & 0.01 & 0.01 & @xmath150 & @xmath154 & @xmath155 & @xmath155 & @xmath152 & 90 & @xmath156 & 0.18 & 0.18 & 4.4 & 4.4 + l1 & @xmath157 & 0.01 & @xmath150 & @xmath158 & @xmath155 & @xmath155 & 200 & 200 & 0.25 & 0.73 & @xmath159 & 4.0 & 5.5 + @xmath117injection rate of protons at the forward shock + @xmath118injection rate of oxygen ions at the reverse shock + @xmath119injection rate of electrons ( positrons ) at the reverse shock at present epoch , see eq .
( 2 ) + @xmath120injection rate of electrons at the forward shock , for the model l1 see eq .
( 1 ) + @xmath121momentum that separates two regimes of diffusion at the forward shock , tev / c , see eq .
( 4 ) + @xmath122momentum that separates two regimes of diffusion at the reverse shock , tev / c , see eq .
( 4 ) + @xmath123injection momentum of electrons at the forward shock , mev / c + @xmath124injection momentum of electrons at the reverse shock , mev / c + @xmath126electron to proton ratio at the forward shock calculated for the energy 10 gev + @xmath128electron to nucleon ratio at the reverse shock calculated for the energy 10 gev + @xmath129ratio of cosmic ray energy @xmath160 to the energy of supernova explosion @xmath161 + @xmath130total compression ratio of the forward shock + @xmath131total compression ratio of the reverse shock the rsg wind density is poorly constrained by x - ray observations of cas a.
the matter is that most of thermal x - rays are produced in the shocked ejecta gas .
in addition the forward shock region shows a significant non - thermal x - ray component . that is why the flux of thermal x - rays and the corresponding amount of the stellar wind gas swept up by the forward shock are rather uncertain in this region . on the other hand , the ejecta mass estimate @xmath162 @xmath163 ( vink et al . @xcite , willingale et al .
@xcite ; hwang & laming @xcite ; laming & hwang @xcite ) are rather robust .
this range of ejecta masses is also in agreement with stellar evolution models ( e.g. gregory @xcite ) and the predicted light curves ( e.g. iwamoto @xcite ) of iib supernova .
we shall use the lower limit of ejecta mass @xmath164 below . in this study
we use the solar composition for the stellar wind material .
the corresponding number density of helium nuclei is @xmath165 where @xmath166 is the undisturbed number density of hydrogen ions at the current forward shock position .
so the total nucleon density is @xmath167 .
, thin lines ) and of electrons at the reverse shock ( thin dashed line ) calculated for the model h1 at the epoch @xmath168 yr . particle momenta are normalized to the proton mass @xmath23 and the mass of oxygen ion @xmath24 .
the oxygen spectrum is normalized to the nucleon number density .
, width=302 ] ( thin solid line ) at @xmath168 yr obtained in the model h1
. spectrum of run - away protons which have left the remnant is also shown ( dotted line ) .
, width=302 ] the principal parameters used in calculations are summarized in table 1 and table 2 .
both tables consist of three parts ( separated by vertical lines ) .
the first parts contain general parameters that are fixed for all model calculations .
the second parts contain free parameters which are adjusted to match the key observations .
the third parts contain the results derived from modeling .
( thin lines ) and of electrons at the reverse shock ( thin dashed line ) calculated for the model h2 at the epoch @xmath168 yr . particle momenta are normalized to the proton mass @xmath23 and the mass of oxygen ion @xmath24 .
the oxygen spectrum is normalized to the nucleon number density .
, width=302 ] for hadronic model h1 we use high injection efficiencies @xmath169 at forward and reverse shocks .
the stellar wind density @xmath166 is then adjusted to reproduce the fluxes of gamma - emission .
the explosion energy is adjusted to reproduce the forward shock radius @xmath170 pc at the remnant age @xmath168 yr .
the value of the amplified magnetic field is parameterized by the parameters @xmath171 and @xmath172 at forward and reverse shocks respectively .
these numbers regulates the shape of particle spectra and the corresponding spectra of gamma - emission .
we found that in this model with a rather low stellar wind density , @xmath173 @xmath125 , the amount of suprathermal electrons produced via compton scattering of gamma - rays from @xmath34co decay is not sufficient for explanation of the observed radio fluxes .
therefore , in this model it is assumed that the injection of electrons at the forward shock is similar to the injection of protons . on the other hand
the amount of electrons ( positrons ) from the decays of @xmath2ti is sufficient for production of radio flux of the reverse shock . in calculations we adopted the mass of @xmath2ti found in observations of cas a @xmath174 ( renaud et al .
@xcite ) . the electron injection efficiency at the forward shock @xmath175 and the injection momentum 40 mev / c of electrons ( positrons ) at the reverse shock
were adjusted to reproduce the radiofluxes of the shocks .
figures 1 - 7 and 9 - 13 illustrate the numerical results obtained in the hadronic scenario .
the time evolution of the shock radii @xmath176 and @xmath86 , the forward and reverse shock speeds @xmath177 and @xmath178 , the reverse shock speed in the ejecta frame @xmath179 and the ratio of cr energy to the energy of supernova explosion @xmath89 are shown in fig.1 . at early times after
sn explosion the ratio of forward and reverse shock radii is independent on time .
approximately after 40 years after the explosion , the forward shock starts to interact with the dense gas of the rsg shell , the speeds of both shocks drop and the distance between shocks decreases . at 150 years after explosion the forward shock leaves the shell and enters into undisturbed rsg wind . at this time the forward shock swept up the gas mass @xmath180 comparable to the ejecta mass and the transition to the sedov phase began . at present the forward shock speed is @xmath181 km s@xmath10 that is @xmath182 higher than the measured proper motion @xmath183 km s@xmath10 of the x - ray filaments ( patnaude & fesen @xcite ) .
the reverse shock position @xmath184 pc is at the upper boundary of observed values @xmath185 pc . for higher ejecta mass @xmath186 the reverse shock
is shifted to larger radii in contradiction with observations .
the reverse shock speed @xmath187 km s@xmath10 is in agreement with measured in radio speed @xmath188 km s@xmath10 ( delaney & rudnick @xcite ) and is lower than measured in optics speed @xmath189 km s@xmath10 ( morse et al .
@xcite ) .
radial dependencies of several key parameters at the present epoch @xmath168 yr are shown in fig.2 . in the same figure
we show the positions of the contact discontinuity and the forward and reverse shocks .
cr , gas and magnetic pressures at the forward shock are 30@xmath190 , 42@xmath190 and 6@xmath190 of the ram pressure @xmath191 , respectively . at present the forward shock have swept up 2.7@xmath192 of the stellar wind material .
the reverse shock have swept up 1.66@xmath192 of ejecta while 0.34@xmath192 of ejecta is not shocked yet .
the minimal electron heating by coulomb collisions still results in rather high electron temperatures .
the shocked ejecta and the shocked stellar wind plasma are heated up to 2.0 kev and 1.4 kev respectively .
it should be noted that our one - dimensional calculations can not adequately describe the development of the rayleigh - taylor instability of the contact discontinuity . in real situations
the supernova ejecta and the circumstellar gas are mixed by turbulent motions in this region ( see e.g. mhd modeling of jun & norman @xcite ) .
spectra of accelerated protons , the oxygen ions and electrons are shown in fig.3 . at the present epoch
the maximum energy of protons accelerated in this snr and still confined in the shell , is about 40 tev ; the higher energy protons have already left the remnant .
spatially integrated proton and electron spectra at the present epoch @xmath168 yr are shown in fig.4 .
we also show the spectrum of run - away particles .
these particles have already left the acceleration site through an absorbing boundary at @xmath193 .
the sum of the proton spectra shown is the total cosmic ray spectrum produced in this snr over the last 330 years after sn explosion . for this snr ,
the spectrum of cosmic ray protons have a maximum energy about 60 tev that is lower than the knee in the observable cosmic ray spectrum . to demonstrate the influence of the enhanced diffusion and rsg shell we consider also the hadronic model h2 with the parameters similar to the parameters of the model h1 but without rsg shell and with bohm diffusion ( @xmath194 , see tables 1,2 ) .
spectra of accelerated protons , oxygen ions and electrons are shown in fig.5 .
in this model with slower diffusion the maximum energy of the protons accelerated is close to 1 pev . in leptonic models of gamma - ray emission , protons play negligible role in the forward shock dynamics and in the production of electromagnetic emission .
therefore we use rather low injection efficiency @xmath195 of protons at the forward shock .
this very low injection efficiency is possible because the forward shock propagates in the stellar wind with the azimuthal magnetic field .
the magnetic field amplification here ca nt be produced by accelerated particles .
it is probably due to vortex motions which are created downstream of the shock in the course of an interaction of stellar wind density fluctuations with the shock front ( see giacalone & jokipii @xcite ) . in this model with a rather high stellar wind density @xmath196 @xmath125 the amount of suprathermal electrons generated in the stellar wind gas by the compton scattering of gamma - rays from @xmath34co decay is enough for the electron injection at the forward shock .
the electron injection efficiencies correspond to the realistic mass of @xmath34ni @xmath197 and to the @xmath2ti mass @xmath174 revealed by observations of cas a ( renaud et al .
the injection momenta 200 mev / c of electrons from ic scattered gamma - rays at the forward shock and directly from @xmath2ti decays at the reverse shock are adjusted to reproduce radio , x - ray and gamma - ray fluxes . since the electrons and gamma - rays from radioactive decays are produced with small energies , a significant increase of their energy up to @xmath198 mev is required .
it can be realized through a stochastic mechanism of acceleration in highly turbulent upstream regions of the reverse and forward shocks ( zirakashvili & aharonian @xcite ) .
the pressure of electrons and positrons at the reverse shock could be comparable to the pressure of ions .
therefore the reverse shock can be significantly modified due to the total pressure of ions and relativistic electrons .
-decay gamma - rays ( solid line ) , ic gamma - rays ( dashed line ) .
gamma - emission from nonthermal bremsstrahlung is very small ; its contribution is not shown.,width=302 ] -decay gamma - rays ( solid line ) , gamma - rays from nonthermal bremsstrahlung ( dotted line ) . the contribution of ic gamma - ray emission is negligible.,width=302 ]
in this section we present the results of numerical calculations of radiation produced in interactions of accelerated electrons and protons .
the gamma - ray spectra from proton - proton interactions are calculated using the formalism of kelner et al .
since @xmath3-decay gamma rays from the reverse shock are produced mainly in oxygen - oxygen collisions , we use the following approximation . the gamma - ray flux produced by energetic nuclei
is given by the gamma - ray flux produced by energetic protons with a number density equal to the nucleon number density of energetic nuclei .
the target medium with a mixed composition is substituted by a pure hydrogen plasma with a number density equal to the number density of nucleons in plasma with a mixed composition . for heavy nuclei
this approximation slightly overestimate the flux of gamma - emission .
this was taking into account using the calculations of the nuclear enhancement factor of mori @xcite . for oxygen - oxygen collisions
this gives the correction factor 0.6 . for ic gamma - rays we use the standard expressions of cross - sections (
see e.g. blumenthal & gould @xcite ) .
the main the target photons contributing to the gamma - ray production are the microwave background radiation and the infra - red emission of cas a itself , with the energy density 2 ev @xmath125 and temperature @xmath199 k. the contributions of other diffuse radiation fields can be neglected . for calculations of synchrotron radiation
we take into account that in highly turbulent regions close to the inverse and forward shocks , the magnetic field has some probability distribution @xmath200 .
this makes the cut - off in the spectrum of synchrotron radiation somewhat smoother ( zirakashvili & aharonian @xcite , see also bykov et al .
@xcite ) . the results of calculations of the broad - band emission for the hadronic scenario h1 are shown in fig.6 .
the principal model parameters used in calculations are described in table 1 and table 2 .
note that at the present epoch , 25 % of the explosion energy ( @xmath201 erg ) has been transferred to accelerated particles ( see fig.1 ) , most of which are still confined in the shell of the remnant ( see fig.4 ) . at low energies ,
the spectra of accelerated particles at cr modified shocks become softer which may results in a significant spectral steepening of synchrotron radio emission .
the contribution of electrons directly accelerated by the reverse shock to synchrotron radiation and ic gamma - rays is significant . in particular
, the reverse shock produces 50@xmath202 of radio and 90@xmath202 of 5 kev synchrotron x - rays . in this model gamma - rays
are produced mainly via pion decay downstream of the forward and reverse shock ( equal contributions at 1 tev ) .
the modeled gamma - ray flux is slightly above the data at high energies beyond 1 tev .
a better agreement with data can be achieved for smaller values of the parameter @xmath203 . in fig.6
we show also the energy flux of the thermal bremsstrahlung .
it is known that the thermal x - rays of cas a are produced mainly in the dense knots of supernova ejecta . to take this effect into account and to explain soft x - ray data we multiply the x - ray thermal bremsstrahlung flux on the factor of @xmath204 .
this means that the volume factor of the x - ray emitting plasma is below 0.2 .
the flux of non - thermal x - rays is about 70@xmath202 of the total x - ray flux at 5 kev .
the results of calculations of the broad - band emission for hadronic model h2 are shown in fig.7 .
in this model the ejecta gas is compressed into the thin shell downstream of the reverse shock .
the thermal bremsstrahlung from this shell makes the main x - ray contribution at energies below 10 kev . at higher energies
the synchrotron x - rays from the forward shock dominate in x - ray spectrum . the reverse shock produces only 12@xmath202 of non - thermal x - rays .
the calculated gamma - ray emission is produced mainly at the forward shock and is above the data points at high energies above 1 tev .
the same is true at low energies below 1 gev where the non - thermal bremsstrahlung of electrons accelerated at the reverse shock gives a non - negligible contribution .
the shocks produce equal amount of radio - emission .
the results of calculations of the broad - band emission for leptonic scenario are shown in fig.8 .
the reverse shock produces 40@xmath202 of radio and 50@xmath202 of 5 kev x - rays .
the radio - spectrum of the non - modified forward shock is rather steep .
this is because the electrons were mainly injected and accelerated at early stages of the remnant expansion ( see eq.(1 ) ) when the synchrotron losses result in the steepening of electron energy distribution . in this model gamma - rays
are produced mainly by non - thermal bremsstrahlung and ic scattering at the reverse shock .
the modeled gamma - ray flux is in good agreement with observations at high energies but is significantly higher than fermi lat data below 1 gev .
the radial profiles of brightness distributions of x - ray , gamma - ray and radio- emissions calculated for the hadronic model h1 are shown in fig.9 - 11 .
the projection effect is taken into account .
all three components of radiation in the radio , x - ray and gamma - ray bands peak at the contact discontinuity and at the reverse shock ( approximately @xmath205 of the angular radius ) .
the narrow peak of gamma - emission at angular radii @xmath206 ( see figures 9 and 10 ) is produced by the forward shock protons in the gas of the rsg shell swept up by the forward shock .
the radio and x - ray profiles in fig.11 are in agreement with available observations ( anderson et al .
@xcite , gotthelf et al .
@xcite , helder & vink @xcite ) .
the sharpest features in the radial profiles appear in x - rays ( see fig.11 ) .
such sharp features imply filamentary structures at forward and reverse shocks .
note that the projection effect works differently for reverse shock in comparison with the forward shock ( see zirakashvili & aharonian @xcite ) . as a result
, the filaments look similar in spite of a relatively weak magnetic field at the reverse shock .
the calculated width of filaments shown in fig.11 is close to @xmath207 which is slightly less than the thickness measured by _
( bamba et al .
it has been argued by vlk et al .
@xcite that the observed thickness corresponds to the downstream magnetic field of about 500 @xmath208 g .
the larger fields demanded by our model results in narrower filaments .
the observed broader filaments can be explained by the limited angular resolution and/or by inhomogeneous circumstellar medium . in realistic scenarios , the corrugated shock surfaces and
the development of rayleigh - taylor instability near the contact discontinuity will destroy a picture of ideal circle filaments .
nevertheless , non - thermal x - ray filaments must be observed in the reverse shock region in cas a. the dominance of ejecta in thermal x - rays is not surprising since cas a is at transition from the ejecta dominated stage to sedov stage now .
it is well established that thermal x - rays from very young extragalactic supernova are produced by a shocked ejecta gas ( see e.g. chevalier & fransson @xcite ) .
the calculated temporal dependencies of radiation fluxes in radio , x - ray , and gamma - ray bands in the hadronic models h1 and h2 is shown in fig.12 and 13 .
results for epochs @xmath209 yr shown in fig .
12 can be used also for ib / c snrs with tenuous winds of wolf - rayet stars .
results for h2 model ( fig .
13 ) are more appropriate for iib snrs with dense rsg stellar wind .
the rate of the secular decrease of electromagnetic radiation calculated for these two hadronic models , as well as for the leptonic model , are summarized in table 3 .
the rates of decrease of gev and tev gamma - ray flux predicted by hadronic models are significantly larger than the one expected in the leptonic model .
it should be noted that since the radiation field of cas a is dominated by its own infra - red radiation , the secular decrease of inverse compton gamma - rays is affected by the decrease of the density of the target field . for simplicity
, the decrease rate of ic gamma - rays 0.12@xmath202 yr@xmath10 in table 3 has been calculated assuming a time - independent radiation field .
therefore the secular decrease of ic gamma - rays should be somewhat faster given the secular evolution of the infra - red radiation , but still it remains significantly slower than the decrease of gamma - ray flux in hadronic models .
the current secular decrease of radio - intensity 0.85@xmath202 yr@xmath10 is close to measurements of baars et al .
@xcite ( [email protected]@xmath202 yr@xmath10 at 1.4 ghz ) and higher than measurements of osullivan & green @xcite ( 0.6@xmath202 yr@xmath10 at 15 ghz ) . at the same time , the calculated rates of decrease of 5 kev x - ray fluxes predicted by hadronic and leptonic models close to 1% per year is slower than the measured value @xmath211 yr@xmath10 ( patnaude et al .
@xcite ) . the fast decrease of radio and x - ray emissions in the models h1 and
l1 is explained by the significant contribution of the reverse shock to the radio and x - ray production . presently , it propagates through the flat part of the ejecta spatial density distribution .
the ejecta density @xmath212 drops proportional to @xmath213 , i.e , 0.9% per year .
the fast drop is the main factor which determines the decrease of thermal and nonthermal x - rays .
since the synchrotron x - rays are produced in the loss - dominated regime the x - ray flux @xmath214 of the reverse shock can be written as @xmath215 where @xmath216 is the maximum energy of synchrotron x - rays and the function @xmath217 describes the shape of the spectral synchrotron cut - off ( see also patnaude et al .
it is also assumed that the the electron energy density is proportional to the ram pressure of the shock @xmath218 .
for the energy @xmath216 which is proportional to the square of the shock speed @xmath88 , we can estimate the secular decrease of x - ray flux as @xmath219 here @xmath220 is the spectral slope @xmath221 .
using dependencies @xmath222 , @xmath223 ( see fig.1 ) and the observational value @xmath224 we obtain @xmath225 that is the secular decrease @xmath226 yr@xmath10 .
the total secular decrease of x - rays is lower because the thermal x - rays and the forward shock give an additional contribution .
the x - ray flux of the forward shock decreases slower because in the stellar wind the density at the forward shock @xmath227 .
that is why the decrease of synchrotron x - rays at the forward shock depends only on the decrease of the forward shock speed . then the equation similar to eq .
( 7 ) gives the synchrotron flux decrease proportional to @xmath228 .
for the measured dependence @xmath229 , we obtain the secular decrease at the forward shock of order of @xmath230 yr@xmath10 .
this is demonstrated in fig .
13 for the h2 model in which the non - thermal x - rays are produced at the forward shock . in this regard the fast decrease ( @xmath231 yr@xmath10 ) of hard x - ray flux reported by patnaude et al
@xcite is in favor of the interaction with the rsg shell in past ( model h1 ) . during this interaction
the reverse shock speed @xmath88 increases significantly , but presently this speed decreases ( see fig.1 ) .
this results in the secular decrease of the non - thermal x - ray flux from the reverse shock that is faster than secular decrease of the ejecta density @xmath232 yr@xmath10 .
the situation is different in the models without rsg shell ( h2 , l1 ) . in these models
, the speed @xmath88 presently is almost constant .
correspondingly , the secular decrease of the nonthermal x - ray flux can not be faster than @xmath232 yr@xmath10 .
we conclude that the model h1 is in better agreement with the measured secular decrease , although for x - rays even the model h1 predicts slower decrease than the observations show .
we may directly compare the results of our calculations with radio - observations of well - studied extragalactic iib supernova 1993j ( weiler et al .
its radio flux at @xmath233 cm was @xmath234 30 mjy at day 3000 .
this corresponds to @xmath235 30 kjy recalculated for cas a distance which is a factor of 3 smaller than the value corresponding to the radio - curve in hadronic model h2 ( see fig.13 ) .
the reason could be that in cas a the electron injection is more efficient in cas a than in snr 1993j or it was lower in past .
the model h2 predicts gamma - ray flux one year after the explosion at the level of @xmath236 erg @xmath237s@xmath10 .
this corresponds to the flux @xmath238 erg @xmath237s@xmath10 recalculated for snr 1993j distance .
this estimate agrees with the results of tatischeff @xcite , but is below by two orders of magnitude the flux predicted by kirk et al . @xcite .
the reason of such a large discrepancy is the higher wind density assumed by kirk et al .
@xcite .
h1 & 0.85 & 1.14 & 0.70 & 0.82 + h2 & 0.53 & 0.77 & 0.35 & 0.47 + l1 & 0.73 & 0.91&0.14 & 0.12 +
according to the recent proper motion x - ray measurements ( patnaude & fesen @xcite ) , the expansion parameter of cas a is @xmath239 this is close to the expansion parameter @xmath240 at the sedov phase for a supernova shock moving in the medium with the @xmath92 density profile . at first glance
this can be interpreted that cas a is already in the sedov phase .
however this contradicts to the current radius of the reverse shock @xmath241 .
detailed hydrodynamical simulations confirm this qualitative argument . in order to reproduce the current positions of the forward and reverse shocks in cas a , one has to assume for the forward shock velocity @xmath242 km s@xmath10 ( e.g. chevalier & oshi @xcite ; patnaude & fesen @xcite ) .
this is close to the value @xmath243 km s@xmath10 found in our leptonic model calculations and is 15@xmath190 higher than the proper motion @xmath244 km s@xmath10 of x - ray filaments .
patnaude and fesen @xcite proposed an explanation of this discrepancy assuming that the forward shock is modified by the cosmic ray pressure .
then the highest energy particles leave the remnant and take away some part of energy .
this results in a smaller shock velocity .
however , this scenario is possible only if the number density of the circumstellar medium and the ejecta mass are several times lower compared to the parameters @xmath245 cm @xmath246 and @xmath247 derived from x - ray observations . for number density @xmath245 cm @xmath246 assumed by patnaude & fesen , the gamma - ray flux form the pion decay would be a factor of 10 higher in comparison to the one measured by _
lat at gev energies and a factor of 30 higher than the flux of tev gamma - rays measured by hegra , magic and veritas .
the lower number densities are indeed used in our models h1 and h2 with an efficient cosmic ray acceleration .
however strong magnetic fields assumed results in the steep spectra of accelerated particles .
that is why the energy flux of escaped particles is low and does not produce any effect on the forward shock velocity .
an alternative way to resolve the problem would be a deviation from the @xmath92 density profile .
the cas a progenitor might emit fast and tenuous wind prior the explosion and produce a bubble of rarefied gas and a shell of compressed rsg wind in its vicinity . in our hadronic scenario , the collision of the forward shock with this shell results in a significant transfer of the explosion energy to the reverse shock .
correspondingly , the energy related to the forward shock will be reduced leading to a lower speed of the forward shock .
our calculations show that in this way one can bring the forward shock speed in accordance with measurements for the shell radius @xmath248 pc . on the other hand , for this shell radius the reverse shock radius appears too large compared to observations .
in our modeling we adopt an ad - hoc ( smaller ) shell radius , @xmath249 pc . in this
regard we should note that it is possible to reproduce both the measured shock speed and the reverse shock radius for higher stellar wind densities
. a similar conclusion has been made by van veelen et al .
@xcite who investigated the impact of a possible wolf - rayet phase for the cas a progenitor on the snr evolution .
severe constrains on the radius of the rsg shell have been found by schure et al .
they showed that the jet observed in cas a can not pass through the rsg shell if its radius @xmath250 pc .
however the stellar wind density in our model h1 is a factor of 3 lower than the stellar wind density assumed by schure et al .
therefore the jet can survive even for larger size of the bubble . the earlier proper motion measurements by delaney and rudnick @xcite is in a good agreement with the forward shock speed derived from our model calculations .
however one should note that the results of delaney and rudnick @xcite imply @xmath251 larger speeds of x - ray filaments than the ones reported by patnaude and fesen @xcite ( see table 1 of patnaude and fesen @xcite ) .
the reason of the discrepancy remains unclear .
finally , the low measured velocity of x - ray filaments may be related with density disturbances in the circumstellar medium when we observe the brightest x - ray filaments which are formed in the regions with a higher stellar wind density and a correspondingly lower local shock speed .
this scenario can be checked easily .
patnaude & fesen @xcite reported the increase of the brightness of the outer x - ray filament in the north - east periphery of cas a. the measured value of its speed is @xmath234 5000 km s@xmath10 .
probably the forward shock recently has entered a denser region in this position and initiated an increase of the x - ray brightness . after the shock exits this high density region and enters a lower density region , in future we will observe a decrease of the x - ray brightness and increase of the filament speed . taking into account that the speeds of different forward shock filaments show a significant scatter @xmath252 km s@xmath10 ( see the earlier measurements of delaney & rudnick @xcite ) this scenario seems rather plausible .
we should note that our estimate of the stellar wind mass @xmath253 swept up by the forward shock in the hadronic model h1 is several times smaller than the estimate @xmath254 of vink et al .
however , as indicated in sect .
4 , the fraction of thermal x - rays produced in the shocked circumstellar medium of cas a is rather uncertain .
for example favata et al .
@xcite , after subtraction of the non - thermal fraction of x - ray emission , derived @xmath255 of swept up material in cas a which is closer to our estimate .
lee et al .
@xcite recently reported the observation of thermal x - rays produced in the shocked circumstellar gas beyond the non - thermal x - ray filaments of cas a. according to their explanation this x - ray emission is produced at the outlaying parts of the forward shock where cr acceleration is not effective .
they also estimated the stellar wind density and derived @xmath256 of the wind material swept up by the forward shock . in our opinion , it is more natural to relate the thermal x - ray emission to the strong gas heating in the upstream regions of the forward shock .
the alfvn velocity is of the order of @xmath257 km s@xmath10 in this region ( see section 8.3 below ) .
the corresponding magnetic amplification is produced by the plasma motions with similar velocities of the gas .
these motions unavoidably produce multiple small - scale shocks and the gas heating in the upstream region .
it is known that the development of the non - resonant streaming instability results in the magnetic amplification and in the very complex density structure when the main part of the volume is filled with almost empty gas cavities .
the cavities are separated by the walls with a denser gas ( see e.g. bell @xcite )
. therefore mean plasma density derived from x - ray observations can be significantly overestimated . as for the thermal x - rays produced by ejecta ,
the flux of thermal x - rays is underproduced in our model h1 ( see section 5 ) .
the enhancement factor @xmath204 assumed to reproduce the soft x - ray data may be used to estimate the properties of x - ray emitting knots .
the number density of nucleons @xmath258 in the x - ray emitting plasma is estimated as @xmath259 , where @xmath260 is the shock speed in the knot , @xmath88 is the reverse shock speed in the frame of ejecta , and @xmath261 is the mean density of the shocked ejecta . in the model h1
we have @xmath262 km s@xmath10 , while the shock speeds in the x - ray emitting knots are of the order of 800 km s@xmath10 .
this gives @xmath263 .
then the volume factor of the x - ray emitting plasma must be @xmath264 in order to reproduce the enhancement factor @xmath204 .
the x - ray emitting plasma contains @xmath265 of the shocked ejecta mass . using @xmath266 @xmath125 in the model h1
, we estimate the nucleon number density of the x - ray emitting plasma @xmath267 @xmath125 .
the corresponding electron number density @xmath268 @xmath125 appears to be in agreement with the results of lazendic et al . @xcite .
the measured mean ionization age @xmath269 @xmath125 s ( lazendic et al .
@xcite , hwang & laming @xcite ) then corresponds to the age of the x - ray emitting plasma @xmath270 years . in the model h1 ,
the ram pressure of the stellar wind is @xmath271 erg @xmath125 .
this value is comparable with the pressure of the interstellar medium .
it is expected that the rsg wind is embedded in the rarefied bubble created by the fast wind of cas a progenitor at the main sequence stage .
the gas in the bubble is probably in the pressure equilibrium with the interstellar medium .
then the rsg wind is terminated somewhere not far from the current forward shock position .
this occurs at distances @xmath272 pc for a range of interstellar pressures @xmath273 erg @xmath125 .
the stellar wind gas is compressed and forms a dense outer shell ( see modeling of prez - rendn et al .
the forward shock of cas a will encounter this shell in future .
the expected mass of rsg wind material beyond the current forward shock position including the outer rsg shell can not be less than @xmath274 .
another high density shell with mass @xmath275 forms the walls of the rarefied bubble at distances several tens of parsecs ( prez - rendn et al .
some gas shells are indeed observed in infrared in the vicinity of cas a ( barlow et al .
@xcite ) .
both shells are considered as perspective targets for gamma - ray production by the highest energy protons escaping snr cas a. these protons had no time to travel far from the snr because even a rectangular propagation at 300 years corresponds to 100 pc distance .
probably they are captured by tangled magnetic fields in the bubble walls .
the gamma - emission calculated in the hadronic model h1 is in reasonably good agreement with measurements ( see figure 6 ) .
the steepening of the gamma - ray spectrum at tev energies is achieved by assuming faster diffusion of highest energy particles in accordance with eq.(4 ) .
this assumption affects only the diffusion of highest energy protons and nuclei , but does not have a strong impact on electrons ; below the maximum energy of electrons around 10 - 20 tev determined by synchrotron losses ( see figure 3 ) the diffusion is still close to the bohm limit .
the adjusted values of momenta @xmath276 , @xmath277 tev / c which separate two regimes of diffusion can be neither increased nor decreased .
the increase of @xmath278 will lead to contradiction with tev gamma - ray data , while the decrease of @xmath278 will contradict to hard x - ray data .
thus we can conclude that the tev gamma - ray and hard x - ray emission components are produced by particles scattered by small - scale magnetic inhomogeneities ( see eq .
diffusion coefficient corresponding to the scattering by small - scale magnetic inhomogeneities is estimated as @xmath279 . here
@xmath280 is the gyroradius of a charged particle calculated in the small - scale random magnetic field with the square root of the mean square @xmath46 , @xmath281 is the inverse scale of the random field . if the magnetic field is amplified in the course of the non - resonant streaming instability , then @xmath282 ( bell @xcite ) . here
@xmath283 is the increment of the instability and @xmath284 is the alfvn velocity calculated in the non - amplified magnetic field . comparing the diffusion coefficient with eq.(4 )
, one finds @xmath285 .
we may estimate the increment of the instability as @xmath286 where @xmath287 is the age of the snr .
the value of the amplified field at the forward shock is @xmath288 ( cf .
finally we obtain @xmath289 @xmath290 here @xmath291 is the magnetic field in the stellar wind . for the forward shock @xmath292 , @xmath181 km s@xmath10 and the adjusted value of @xmath70 corresponds to the magnetic field strength in the rsg stellar wind @xmath293 g .
it corresponds to the magnetosonic mach number of the wind @xmath294 , while the upper limit @xmath295 g corresponds to the condition @xmath296 . in reality
, the stellar wind with an azimuthal magnetic field and @xmath297 is collimated close to the parent star .
since the forward shock in cas a is rather circular , we conclude that the wind collimation does not occur and the magnetosonic mach number @xmath298
. then our estimate of the wind magnetic field seems reasonable .
the scale of the amplified magnetic field is only @xmath299 cm , i.e. @xmath300 fraction of the remnant radius .
it is comparable with the gyroradius of tev protons calculated in the amplified magnetic field . using the estimate of the obtained magnetic field , we can estimate the magnetic field of the cas a progenitor @xmath301 where @xmath302 is the rate of stellar rotation and @xmath303 is the radius of the progenitor . for characteristic values
@xmath304 cm and @xmath305 s@xmath10 we obtain @xmath306 g. this number is significantly below than the value @xmath307 kg inferred from observations of h@xmath308o masers at distances @xmath309 cm from several rsgs with powerful stellar winds ( vlemmings et al .
however , it is not excluded that the strong magnetic fields of rsg progenitor dissipate at large distances in the stellar wind e.g. via magnetic reconnection .
we can use the expression similar to eq .
( 8) also for the reverse shock . substituting the reverse shock speed in the ejecta frame , @xmath310 km s@xmath10 , and @xmath311
we find the magnetic field in ejecta @xmath312 g .
the maximum energy of protons @xmath234 40 tev in the hadronic model h1 is in agreement with previous estimates of maximum energies attainable at fast shocks with non - resonant streaming instability ( zirakashvili & ptuskin @xcite , bell et al .
probably the obtained low value of the strength of the magnetic field in the stellar wind implies that the stellar wind is not a good place for acceleration of cosmic rays up to pev energies .
if the forward shock of cas a would propagate in the interstellar medium with the magnetic field strength @xmath313 @xmath208 g , the current value of the maximum energy would be a factor of @xmath314 higher , namely 800 tev .
generally , the high - energy part of the observed spectrum is well reproduced in the leptonic model l1 ( see figure 8)
. however the low energy part of the spectrum below 1 gev contributed by non - thermal bremsstrahlung of electrons , does not agree with _ fermi _ lat data . to avoid overproduction of the non - thermal bremsstrahlung , in the hadronic model h1
we assume that the magnetic field at the reverse shock is rather large .
it is interesting that this is possible only due to the presence of the rsg shell .
as mentioned in section 8.1 , this shell results in an effective transfer of the explosion energy in the reverse shock region and in a stronger magnetic field . because of the lack of such a shell in the hadronic model h2 , the energetics of the reverse shock is low , and the magnetic field is rather weak .
therefore , the number of relativistic electrons must be very high to reproduce the radio flux observed from the reverse shock .
then , the non - thermal bremsstrahlung produced by same electrons might significantly exceed the gamma - ray flux observed at sub - gev energies ( see figure 7 ) .
the hadronic models demand strong amplification of the magnetic field in order to reproduce the shape of the observed broad - band energy spectra of gamma - rays . for the model h1 ,
the required upstream values are 0.4 mg and 0.2 mg for the forward and reverse shocks , respectively .
the alfvn drift in such fields may lead to steepening of the energy spectra of accelerated particles .
this effect has been earlier recognized and explored by ellison et al .
( @xcite ) for cas a , and recently also by morlino and caprioli ( @xcite ) for tycho .
the comparison of the magnetic fields in the stellar wind and ejecta implies very large amplification factors : @xmath315 and 700 for the forward and reverse shocks , respectively .
the magnetic field of the ejecta is the field of the exploded supernova progenitor . presumably , the field was additionally amplified during the propagation of the radiative shock inside the exploded star when vortex motions downstream of the shock result in the efficient gas mixing ( see the modeling of iwamoto et al .
these motions can also amplify random magnetic fields in the star interior .
the random magnetic field may be amplified downstream of the shock moving in the medium with density perturbations ( giacalone & jokipii @xcite ) .
eventually the shock crosses the outer layers of the exploded star and enters into a circumstellar medium .
the downstream material forms a supernova ejecta .
in order to estimate the ejecta random magnetic field in this scenario , let us assume that a small fraction @xmath316 of a supernova mechanical energy @xmath161 is transformed into a magnetic energy .
at later epochs the magnetic field strength drops proportional to @xmath317 , where @xmath303 is the radius of a progenitor and @xmath318 is the characteristic velocity of ejecta .
hence the magnetic field of ejecta at time @xmath287 is determined as @xmath319 @xmath320 even for the modest value of @xmath321 the magnetic field in the ejecta of a core collapse supernovae at @xmath322 yr achieves @xmath323 level .
lower values @xmath324 g are expected for ia supernovae originating from a thermonuclear explosion of a white dwarf with a characteristic radius @xmath325 cm . in the case of cas
a , we obtain the magnetic field of ejecta @xmath234 0.5 @xmath208 g which is close to the value found in the previous section .
although this field is random , it has a rather large scales , namely an order of 1/100 of the radius of the reverse shock .
it can be considered as a large scale field because the non - resonant streaming instability amplifies magnetic fields with even smaller scales .
we should note that the above amplification factors are significantly larger compared to the results obtained from the numerical modeling of nonresonant streaming instability ( zirakashvili & ptuskin @xcite ) . to a large extent
this is not a surprise ; it is difficult to amplify the magnetic field by three orders of magnitude in mhd modeling with its limited spatial resolution .
it is likely that the amplified magnetic fields are strongly underestimated in the available numerical simulations because of the numerical viscosity .
another possibility can be related to a possible additional amplification that takes place in the upstream regions of cr modified shocks moving in the medium with strong density perturbations .
the deceleration of gas clumps by the cr pressure is slower compared to a less dense plasma .
they will stretch and additionally amplify the random magnetic field pre - amplified far upstream of the shock by the non - resonant streaming instability .
the possibility of this kind of magnetic amplification was demonstrated recently by drury and downes ( @xcite ) .
a similar idea was also proposed earlier by beresnyak et al .
( @xcite ) . given that the stellar winds and ejecta are very clumpy , so this scenario of additional amplification seems quite plausible .
it should be noted that the high magnetic fields found in our modeling are not rare in young supernova remnants .
for example chandra et al .
( @xcite ) , based on the observed effect of synchrotron aging of radio electrons in the extragalactic iib snr 1993j , derived the downstream magnetic field strength between 0.19 g and 0.33 g after 3200 days of the explosion .
this value is also in agreement with the synchrotron self - absorption model of fransson and bjrnsson ( @xcite ) . for comparison , in our hadronic model h2 for cas
a the downstream magnetic field strength is only 0.04 g at the same 3200th day after the explosion .
the upper limit of the maximum energy of accelerated protons at the forward shock is estimated as ( see zirakashvili & ptuskin @xcite ) @xmath326 @xmath327 here @xmath328 is the ratio of the energy flux of run - away particles @xmath329 to the flux of kinetic energy @xmath330 , @xmath331 is the strength of the random magnetic field in the circumstellar medium .
this limit corresponds to the situation when the random magnetic field is amplified via the nonresonant streaming instability exponentially in time ( see also bell et al .
@xcite , schure & bell @xcite ) .
the instability is driven by the electric current of the highest energy particles . for estimates ,
below we use the value of logarithm @xmath332 in the denominator of eq .
( 10 ) , and fix the expansion parameter @xmath333 .
the parameter @xmath334 is determined by the energy density of particles @xmath335 at the end of their spectrum , just before the cut - off . for shocks with compression ratio @xmath336 the parameter @xmath334
is estimated as @xmath337 .
using fig.3 , we find @xmath338 .
then one obtains @xmath339 tev for the forward shock . for the reverse shock we can use an expression similar to eq .
( 10 ) . using fig .
3 we estimate @xmath340 and obtain @xmath341 tev for the reverse shock . from comparison of these numbers with fig.3
we see that the maximum energies found in the modeling are significantly larger : @xmath342 tev and @xmath343 tev for the forward and reverse shocks , respectively ( the maximum energy of oxygen ions 40 tev per nucleon corresponds to the maximum energy 80 tev for protons ) .
since the parameters of our modeling are adjusted to reproduce the multiwavelength observations of cas a , we conclude that both magnetic amplification and maximum energy of accelerated particles are underestimated by the theory , despite the rather modest maximum energy of about 40 tev required to explain the spectral points at very high energy gamma - rays .
the reason is the steep energy distribution of accelerated particles derived from the gamma - ray data .
this results in a small value of @xmath344 at the forward shock .
the steep particle spectrum itself is explained by the strong magnetic field and correspondingly by the high alfvn speed upstream the shock .
this result found for cas a has a more general implication ; it shows that the strong magnetic field does not necessarily provide higher maximum energy of accelerated particles .
we should note that in the more realistic case of oblique shocks the maximum energy is higher than the one given by eq .
this equation was derived for quasiparallel shocks . taking this into account
we can conclude that maximum energy of particles accelerated in cas a corresponds to the number density of the highest energy particles ( see eq.10 ) .
both quantities are in agreement with avalable tev gamma - ray data . apparently , despite certain achievements in the treatment of amplification of the magnetic field in young snrs in general , further development of theory is needed for better understanding this complex phenomenon regarding , in particular , the demand of very large field amplification by a factor of @xmath345 .
the problem of electron injection remains one of the poorly understood aspects of the theory of diffusive shock acceleration of particles in snrs .
this general problem becomes especially important in the models of cas a without a rsg shell ( models h2 and l1 ) which require very high electron - to - nucleon ratio @xmath346 at the reverse shock .
this exceeds significantly the ratio @xmath347 observed in crs .
for this reason we invoke the electron ( positron ) injection from @xmath2ti decays ( zirakashvili & aharonian @xcite ) . to our knowledge
, no other ideas have been proposed so far which would provide such a high electron - to - nucleon ratio .
note that the electron injection from @xmath2ti decay is not required in the model h1 in which @xmath348 .
thus , any other electron injection mechanism which can provide this ratio at the reverse shock can be used in the model h1 .
however , the positrons from @xmath2ti decays accelerated later by the reverse shock are of certain interest from the point of view of origin of galactic cr positrons .
the reason of high values of @xmath349 in the models h2 and l1 is the low energetics of the reverse shock . correspondingly , the amplified magnetic fields are weak . thus , to reproduce the radio flux from the reverse shock , one has to assume large number of injected nonthermal electrons . the forward shock of cas a propagates in the stellar wind with a @xmath92 density profile . in this environment
the energy of the shocked ejecta consists only 1@xmath202 of the total energy in shocked plasma .
the energetics of reverse shocks is higher and is of the order of @xmath350 for snrs expanding in the uniform medium .
it can reach up to @xmath351 for snrs expanding in the bubbles where plasma density increases towards the walls of the bubble .
this is exactly what happened 40 years after explosion in the model h1 .
the collision of the shock with the rsg shell transfers a significant fraction of the explosion energy into the region of the reverse shock . as for the electron injection at the forward shock of cas
a , we used a standard approach and prescribed some injection efficiency in models h1 and h2 . because of the low wind density , the amount of suprathermal electrons produced via compton scattering of gamma - rays from the decay of @xmath34co appears to be not sufficient for explanation of the radio and x - ray production by the forward shock of cas a. at first glance one may increase the injection momentum of electrons to reproduce observations .
however , any assumption on injection momenta higher than 200 - 300 mev / c leads to contradiction with the radio spectrum .
while the acceleration of particles by the reverse shock in cas a seems to be a natural and well justified component in the general picture , it is fair to ask a question whether can we explain the multiwavelength data of cas a without invoking the contribution of the reverse shock .
formally , the available data do not allow us to give a certain answer to this question .
note that a nonlinear dsa model without the consideration of the reverse shock has been discussed by berezhko et al .
their model contains many components of our models h1 and h2 , including the presence of the rsg shell . in the model
h2 almost all non - thermal x - rays are produced at the forward shock while the thermal bremsstrahlung from the reverse shock contributes to x - rays at energies below 10 kev .
the production of non - thermal x - ray emission at the reverse shock is a favored option ( uchiyama & aharonian @xcite , helder & vink @xcite ) , however it is not yet firmly established . in particular , it has been argued that the non - thermal x - ray filaments in the reverse shock region are in fact the projected filaments and belong to the forward shock ( delaney et al .
@xcite ) .
the observed bright radio - ring can also be explained without the electron acceleration at the reverse shock .
the development of the reighley - taylor instability at the contact discontinuity can result in an additional amplification of magnetic field in the snr interior ( see e.g. the mhd modeling of jun and norman @xcite ) , thus the electrons accelerated at the forward shock may produce enhanced synchrotron radio - emission inside the shell of snr .
so , perhaps one may argue that the major morphological and spectral features of cas a reported so far , can be addressed , in principle , by the forward shock .
on the other hand one may ask another question about the fundamental difference between the reverse and forward shocks in cas a. both shocks propagate in the media having a common origin .
they are the same stellar material ejected prior to the explosion ( the stellar wind ) and during the explosion ( the supernova ejecta ) .
as argued in section 8.2 , presently the magnetic fields in the wind and ejecta are comparable .
the speeds of both shocks also are similar .
therefore , there is no reason for dsa to operate effectively in the forward shock and be suppressed in the reverse shock
. the situation could be quite different for snrs produced in ia supernova explosions .
the magnetic field of ejecta in these snrs is significantly lower compared to the interstellar magnetic field ( see section 8.2 ) .
thus only the forward shock can contribute to the very high energy radiation .
however , even at such small fields , the reverse shock can accelerate electrons to gev energies , and thus be a source of radio emission and gamma - ray bremsstrahlung .
such a scenario presumably is realized in the tycho where non - thermal x - ray emission is not observed in the reverse shock region while we see radio - emission related with the supernova ejecta ( dickel et al .
@xcite ) .
one of the distinct features of the reverse shock is the high electron - to - nucleon ratio at the reverse shock , @xmath352 .
since the reverse shock is modified by the pressure of accelerated ions , a significant part of energy goes into accelerated electrons at the reverse shock . in this
regard a plasma of the reverse shock have properties similar to properties of bright radio - knots .
atoyan et al .
@xcite found that the energy of relativistic electrons in the knots is close to @xmath353 fraction of the explosion energy . since
a volume factor of the knots is of the order of @xmath354 it is easy to estimate that the energy density of accelerated electrons in the radio - knots may be comparable with the mean energy density inside the supernova shell . this may be considered as an independent evidence for a common origin of radio - knots and supernova ejecta . in our model
the high energy density is explained by the `` radioactive '' origin of electrons and positrons injected at the reverse shock ( zirakashvili & aharonian @xcite ) . in the inner region of cas
a , the sheet - like structures and filaments are observed in the infrared band ( isensee et al .
this material has not yet encountered the reverse shock .
these clumpy structures can produce the x - ray , optical and radio - knots knots after the reverse shock passage .
it is expected that the ejecta clumps have a broad range of densities .
the clumps with the lowest density contrast move with a velocity close to the the gas velocity just downstream of the reverse shock .
this velocity @xmath355 km s@xmath10 in the model h1 ( see fig.2 ) .
the corresponding expansion age @xmath356 years is close to the measured value of the radio - knots of the bright radio - ring ( tuffs @xcite , anderson & rudnick @xcite ) .
the large - scale expansion of the radio - ring is even slower and is treated as the expansion of the reverse shock with the speed @xmath357 km s@xmath10 ( delaney & rudnick @xcite ) which is comparable with the reverse shock velocity @xmath358 km s@xmath10 in the model h1 .
denser ejecta clumps are less decelerated after the reverse shock crossing and are initially observed as x - ray knots , while the densest clumps are initially observed as optical knots . at later epochs they have entered the region downstream the forward shock .
the radio - electrons accelerated during the transition through the reverse shock are still inside the clumps .
alternatively they can be accelerated by internal shocks in the clump .
the clumps are eventually destroyed by the shear kelvin - helmholz instability .
a material and accelerated electrons of clumps are mixed with the plasma of radio - plateau ( see discussion and modeling of anderson et al .
the electrons emit a radio synchrotron radiation in the strong magnetic field downstream of the forward shock .
this field probably is additionally amplified by the shear instability .
since the reverse shock is strongly modified by cr pressure , the electron spectra at the reverse shock and in the clumps have concave shape and are steeper at low energies .
this results in steep radio - spectra of the reverse shock and knots .
the electrons that escape the knots may produce even steeper radio - spectra because of higher magnetic fields and because of an energy dependent propagation as discussed by atoyan et al . @xcite .
in addition , the radio - knots near the forward shock are observed with steeper spectra ( anderson & rudnick @xcite ) . a possible interpretation could be that closer to the forward shock the magnetic field is stronger . here , the electrons emitting at the given radio frequency have lower energies
. then their concave spectrum results in the steeper spectra of the knots near the forward shock .
this qualitative picture seems in agreement both with theoretical ( berezhko et al .
@xcite ; atoyan et al .
@xcite ) and observational conclusions ( anderson & rudnick @xcite ; delaney & rudnick @xcite ) .
below we summarize the main results of study of the present work on the nonlinear diffusive shock acceleration in cas a , and the related broad - band electromagnetic radiation .
the available observational properties of cas a are better reproduced by the hadronic model h1 when both the forward and reverse shocks significantly contribute to the broad - band emission .
the gamma - ray production is strongly dominated by decays of @xmath3-mesons from interactions of protons ( in the forward shock ) and ions ( in the reverse shock ) .
the contributions of the forward and reverse shocks around 1 tev are comparable .
the reverse shock produces also approximately 50% of the radio , and 90% of the x - ray synchrotron emission of cas a. \2 .
the model requires ejected mass of about @xmath359 and the explosion energy @xmath360erg .
the forward shock propagates into the stellar wind with the mass - loss rate @xmath361 km s@xmath10 ) yr@xmath10 .
magnetic fields required by the model significantly exceeds the strength of the magnetic field allowed by the mhd treatment of the nonresonant streaming instability .
either the current mhd simulations strongly underestimates the strength of the amplified magnetic field , or an additional amplification is needed .
a possible mechanism could be related to the clumpiness of ejecta and the circumstellar medium .
. another important ingredient of the model h1 is the presence of the dense shell of the compressed red supergiant stellar wind at the distance @xmath249 pc from the supernova progenitor .
contrary to the pure stellar wind models h2 and l1 , this permits to explain several observational features of cas a like the fast decrease of the flux of non - thermal x - rays , the dominance of the reverse shock in the production of non - thermal x - rays and the slow expansion rate of the bright radio - ring and radio - knots .
the acceleration efficiency in cas a should be very high . at present 25@xmath190 of the energy of supernova explosion
is transferred into accelerated particles .
the previous claims of the low acceleration efficiency have been based on the high gas density derived from x - ray observations .
the plasma density has been strongly overestimated because the non - thermal x - ray component was not properly taken into account .
\5 . the reverse shock of cas a is modified by the pressure of energetic ions , positrons and electrons .
the energy content in accelerated electrons and positrons from radioactive decay of @xmath2ti is very close to @xmath1 erg or 0.1% of the explosion energy .
the maximum energy of accelerated particles in cas a can not exceed 100 tev .
this can be related to the weak magnetic field of about @xmath362 in the red supergiant stellar wind where the forward shock propagates .
because of this , the scale of the random magnetic field generated by the nonresonant streaming instability is rather small @xmath363 cm .
the corresponding to the small - scale scattering @xmath69 type dependence of the diffusion coefficient is favored by hard x - ray and tev gamma - ray data .
more accurate spectroscopic measurements of the gamma - ray spectrum of cas a are highly desirable .
the measurements at sub gev and multi - tev energies will help to better distinguish between the hadronic and leptonic scenarios . in the case of hadronic origin of gamma - ray emission ,
the spectral shape at multi tev energies will give an important information about the maximum energies of accelerated particles .
it will be also interesting to search gamma - rays from the vicinity of cas a. the dense gas shells formed by the winds of the cas a progenitor seems to be ideal targets for production of gamma - rays by highest energy particles which already have left the remnant .
these shells , presumably situated at 5 - 30@xmath364 from the center of cas a , can be considered as potential sources of multi - tev gamma - rays - presently or in the near future .
it depends on whether or not the high energy particles have already reached the shells . | the processes responsible for the broad - band radiation of the young supernova remnant cas a are explored using a new code which is designed for a detailed treatment of the diffusive shock acceleration of particles in nonlinear regime .
the model is based on spherically symmetric hydrodynamic equations complemented with transport equations for relativistic particles .
electrons , protons and the oxygen ions accelerated by forward and reverse shocks are included in the numerical calculations .
we show that the available multi - wavelength observations in the radio , x - ray and gamma - ray bands can be best explained by invoking particle acceleration by both forward and reversed shocks .
although the tev gamma - ray observations can be interpreted by interactions of both accelerated electrons and protons / ions , the measurements by fermi lat at energies below 1 gev give a tentative preference to the hadronic origin of gamma - rays .
then , the acceleration efficiency in this source , despite the previous claims , should be very high ; 25 % of the explosion energy ( or approximately @xmath0 erg ) should already be converted to cosmic rays , mainly by the forward shock . at the same time
, the model calculations do not provide extension of the maximum energy of accelerated protons beyond 100 tev . in this model ,
the acceleration of electrons is dominated by the reverse shock ; the required @xmath1 erg can be achieved under the assumption that the injection of electrons ( positrons ) is supported by the radioactive decay of @xmath2ti . |
phase behavior of non - spherical hard bodies with their centers of mass confined in planar geometry is receiving considerable attention due to the recent development of the preparation of colloidal particles with various shapes and new experimental techniques .
nowadays it is possible to prepare colloids with several geometrical shapes such as cubes , polyhedrons , octopods , ellipsoids and helices [ 1 - 6 ] .
the anisotropic colloids can be confined at the interfaces [ 7 - 11 ] , between two parallel solid walls [ 12,13 ] , at the bottom of the sample holder [ 14 ] , at a substrate surface [ 15 ] and into a lamellar matrix of surfactants [ 16,17 ] .
the confinement can be so strong that even colloidal monolayers can be realized experimentally .
ordering properties of two - dimensional and quasi two - dimensional ( q2d ) non- spherical colloids has been the subject of several experimental and theoretical studies [ 17 - 36 ] .
the reason for this is that the nature of two - dimensional ( 2d ) nematic ordering is quite different from the three - dimensional ( 3d ) one .
it shows only quasi - long - range orientational order with algebraically decaying orientational correlations and the ordering transitions between isotropic and nematic phases are first order or continuous through a kosterlitz - thouless disclination unbinding type mechanism [ 18 , 37 ] .
strictly 2d colloidal systems can not be examined experimentally , because the out - of- plane orientational and positional freedoms are always present to some extent .
therefore it is desirable to extend the theoretical studies in such directions , where the extra orientational and positional freedoms are present . along this line
, the ordering properties of microtubules confined in a thin slit have been modeled as hard spherocylinders placed between two planar walls in [ 28 ] . in agreement with the experiment
it has been found that the isotropic - nematic transition density increases with the wall separation [ 28 ] .
ellipsoidal shaped colloidal particles are gaining widespread applications due to development of the stretching techniques in preparation of monodisperse prolate and oblate ellipsoids from spherical latex particles [ 4,38 ] .
they can be also confined into planar geometry and create a monolayer to study ordering and glassy behavior of the q2d ellipsoid systems [ 12,13 ] . in this paper
we examine the orientational ordering properties of q2d hard ellipsoid systems , where particles are allowed to rotate out the confinement plane to some extent , while it is assumed in first approximation that the centers of the particles are always in the same plane . switching on the out - of - plane orientational freedom by gradual increase ( decrease ) of the limiting polar angle ( @xmath0 ) for oblate ( prolate ) shapes ,
it is possible to make a link between strictly 2d hard ellipse ( hard disk ) systems for prolate ( oblate ) shapes and q2d freely rotating prolate ( oblate ) ellipsoids systems .
we show that the additional out - of - plane orientational freedom changes substantially the orientational ordering and the transition properties of both oblate and prolate shaped ellipsoids . to maximize the available free area on the confining surface ,
the axis of revolution of the prolate particle leans out from the plane , while the oblate particle leans into the plane . as a consequence , the freely rotating prolate ellipsoids resemble hard disks at high densities , while those of oblate ellipsoids behave similarly to hard ellipses .
the isotropic - nematic phase transition of hard ellipses corresponds to a planar nematic - biaxial nematic phase transition of the oblate ellipsoids . here
we note that our present work can be considered as an extension of our previous studies done for monolayers of uniaxial and biaxial hard particles using restricted - orientation approximation [ 39,40 ] .
the paper is organized as follows .
the molecular model and the details of the confinement are presented in sec .
iii is devoted to the parsons - lee theory of q2d hard ellipsoid fluids , where we show how to determine the equilibrium free energy , surface coverage , and order parameters of the uniaxial and biaxial nematic phases .
technical details of the replica exchange monte carlo simulation method are given in sec .
iv . the order parameters , the surface coverage , the equation of state and the phase diagram of the system hard ellipsoids are presented in sec .
v. finally , some conclusions are drawn in sec . vi .
) and is restricted by the polar angle ( @xmath0 ) .
the arrows indicate the direction of the main symmetry axis of the ellipsoids . ] in this work we examine the q2d system of hard ellipsoids , where the particles are allowed to rotate freely in a restricted region of the solid angle , while the centers of particles can move only on a two - dimensional xy plane .
the shape of the particles can be both prolate and oblate , i.e. the aspect ratio or shape anisotropy ( @xmath1 , where @xmath2 and @xmath3 are the lengths along and perpendicular to the axis of revolution of the ellipsoid , respectively ) can be either larger than one ( @xmath4 , prolate shape ) or between zero and one ( @xmath5 , oblate shape ) .
the orientational restriction takes place only in the polar angle ( @xmath6 ) , which is measured from the axis of revolution of each particle to the normal of the confining xy plane .
while the particles can rotate freely in the azimuthal angle ( @xmath7 ) , the allowed range of the polar angle is different for prolate and oblate ellipsoids , namely , the range of the polar angle is given by @xmath8 for prolate shapes , while @xmath9 and @xmath10 intervals are allowed for oblate ones ( see fig .
. we can tune @xmath0 between 0 and @xmath11 . in the case of prolate ellipsoids
( @xmath4 ) @xmath12 corresponds to the confined system of freely rotating particles , while @xmath13 gives the two - dimensional system of hard ellipses .
the situation is different for oblate ellipsoids ( @xmath5 ) , because the hard disk limit is given by @xmath12 , while the freely rotating system of confined oblate ellipsoids corresponds to @xmath13 .
the interaction between ellipsoids is purely hard , i.e. the pair potential between two particles is given by @xmath14 where @xmath15 is the center - to - center distance , @xmath16 is the unit vector connecting the centers of the two ellipsoids , @xmath17 is the orientational unit vector of particle @xmath18 ( @xmath19 ) and @xmath20 is the distance of closest approach . note that the third component of @xmath21 is always zero , which ensures that the centers of the particles are always in the xy plane .
the distance of closest approach between two ellipsoids [ @xmath20 ] is approximated in the manner proposed in our previous works [ 41,42 ] .
to describe theoretically the orientational ordering properties of the monolayer of hard ellipsoids we derive our working equations from the well - known parsons - lee ( pl ) theory of hard bodies [ 43,44 ] , which proved very successful in determination of the equation of state and the transition properties of in phase coexistence of non - spherical hard body fluids both in two [ 23,24 ] and three dimensions [ 45 - 47 ] . here
we present only the important ingredients of the theory , specific equations for the ellipsoid monolayer and some technical details .
it is customary to deal with the free energy of the system , which is the sum of ideal and excess free energy terms , i.e. @xmath23 . the ideal term can be determined exactly from @xmath24 where @xmath25 is the inverse temperature , @xmath26 is the number of particles , @xmath27 , @xmath28 is the area of the plane , @xmath29 is the collection azimuthal and polar angles , @xmath30 and @xmath31 is the normalized orientational distribution function ( @xmath32 ) .
the ranges of the azimuthal and polar angles in the integrations have been already given in sec .
the excess free energy contribution can be obtained approximately with the mapping procedure from the actual system to a reference one , where the second virial coefficient and the excess free energy can be obtained with good accuracy . in our case
we choose the system of 2d hard disks as a reference system , because our confined ellipsoid system is q2d .
the second virial coefficient of the hard disks is given by @xmath33 , where @xmath34 is the diameter of the hard disk .
a luckily simple and accurate expression can be derived for the excess free energy of hard disks , too , using the scaled particle theory [ 48 ] @xmath35 where @xmath36 is the packing fraction of the hard disk and @xmath37 is the area of the hard disk . to perform the mapping procedure from the hard ellipsoids into hard disks we introduce the surface coverage ( or packing fraction ) of the hard ellipsoids , because the intersection of the ellipsoid with the xy plane is an ellipse with characteristic lengths depending on the orientation of the ellipsoid .
these characteristic lengths of the ellipse can be calculated easily using simple geometry .
one can derive that @xmath38 and @xmath39 , where @xmath6 is the polar angle of the ellipsoid .
this means that the monolayer of the hard ellipsoids can be visualized as a multicomponent mixture of hard ellipses on the xy plane . using the dimensions of the hard ellipse ,
the intersected area of the hard ellipsoid with the xy plane can be obtained from @xmath40 this polar angle dependent area together with the orientational distribution function allows us to determine the surface coverage of the plane by ellipses , which is given by @xmath41 , where the average area of the intersected ellipse can be determined from @xmath42 .
note that the surface coverage gives back the packing fraction of hard disks in the hard - sphere limit , because @xmath43 and the distribution function is a constant . in pl approach the area of hard disks and the average area of hard ellipses are assumed to be the same ( @xmath44 ) , which implies also that @xmath45 due to one - to - one correspondence between the two systems .
based on the concept of parsons and lee we can now calculate the excess free energy of the monolayer of hard ellipsoids using the thermodynamic properties of hard disks as follows @xmath46 where the second virial coefficient of hard ellipsoids restricted to xy plane is given by @xmath47 this equation contains the excluded area between two ellipsoids , which can be obtained from the distance of closest approach ( @xmath48 ) as follows @xmath49 using eqs .
( 3)-(6 ) and @xmath45 condition the excess free energy of the system becomes @xmath50\nonumber\\ & & \times\frac{\int d\omega_1f(\omega_1)\int d\omega_2f(\omega_2)a_{\rm excl}(\omega_1,\omega_2 ) } { 4\int d\omega a(\theta)f(\omega)}.\end{aligned}\ ] ] the sum of eqs .
( 2 ) and ( 8) constitutes our density functional equation to determine the equilibrium orientational distribution function [ @xmath31 ] and the total free energy density at a given density ( @xmath27 ) .
the minimization of the free energy functional with respect to @xmath31 gives us the euler - lagrange equation for the equilibrium @xmath31 .
note that the minimization must be carried out by maintaining the normalization condition ( @xmath32 ) .
although the minimization procedure is very simple and straightforward , we do not present the equation of @xmath31 , since the resulting equation is too long .
once the equilibrium @xmath31 is determined with a standard iterative method , the free energy density can be obtained by the substitution of the resulting @xmath31 into the free energy functional ( sum of eq .
( 2 ) and ( 8) ) .
we determine the pressure from the free energy using @xmath52 and examine the orientational ordering properties of the monolayer by the standard uniaxial and biaxial order parameters .
these are defined as @xmath53 and @xmath54 where @xmath55 is the second order legendre polynomial .
the uniaxial order parameter ( @xmath56 ) is positive for out - of - plane ordering ( @xmath57 ) , while it is negative for in- plane ordering ( @xmath58 ) .
the biaxial order parameter ( @xmath59 ) can be very useful in finding in - plane orientational ordering transitions , because @xmath59 is zero for in - plane complete disorder , while it is nonzero for in - plane order .
@xmath56 is practically a three - dimensional order parameter of bulk ellipsoids , while @xmath59 is the corresponding two - dimensional one of bulk ellipses when @xmath13 .
since our system is q2d , we need both of them . to see the effect of out - of - plane orientational freedom and to which extent the particles are tilted off from the confining plane , we determine the average aspect ratio of the effective ellipse system , which is defined as @xmath60 . in sec .
v we plot the density and the pressure in dimensionless units , which are defined as @xmath61 and @xmath62 .
as for the 2d hard ellipse study [ 35 ] , we are implementing the replica exchange monte carlo technique [ 49 - 51 ] .
this is done to avoid , as far as possible , the inherent hysteresis associated to transitions [ 52 ] .
the method is based on the definition of an extended ensemble with partition function @xmath63 , where @xmath64 is the partition function of ensemble @xmath18 .
@xmath65 ensembles are considered , and @xmath65 replicas are employed to sample the extended ensemble , each one at each ensemble .
defining @xmath66 allows introducing swap trial moves between any two replicas , whenever the detailed balance condition is satisfied . in our case
it is convenient to expand isobaric - isothermal ensembles in pressure [ 53 ] .
this is so since we are studying hard particles .
hence , the partition function of the extended ensemble reads [ 53,54 ] @xmath67 where @xmath68 is the partition function of the isobaric - isothermal ensemble of the system at pressure @xmath69 and temperature @xmath70 .
@xmath26 particles are taken into account at each ensemble .
a standard implementation is used to sample the ntp i ensembles .
this implies independent trial 2d displacements , 3d rotations of single ellipsoids , and area changes of the simulation cell . in case of having confining planes , 3d rotations of single ellipsoids
are constrained by their presence .
we are also accounting for non - orthogonal parallelogram cells and so , additional trial changes of the angles and relative length sides of the cell lattice vectors are included .
the following acceptance rule is set [ 53 ] @xmath71 \right\},\end{aligned}\ ] ] where @xmath72 is the area difference between replicas @xmath18 and @xmath73 .
adjacent pressures must be close to provide swap acceptance rates over 0.1 .
simulations are started from a packed triangular arrangement of spheres which are elongated in the direction normal to the plane by a factor @xmath74 . in case of oblates , a stretching factor @xmath74
is also applied in a certain in - plane direction .
conversely to the stretching of spheres in a 3d cell , this procedure leads to the largest packed arrangement of spheroids in a plane [ 55 ] .
a stationary state is reached faster by decompressing packed cells than by compressing lose random configurations [ 52 ] .
we perform the necessary trial moves to observe a stationary state . at this stage
we adjust maximum displacements to produce acceptance rates close to 0.3 .
we also relocate all pressures , initially set by following a geometric progression with the replica index , to obtain similar swap acceptance rates for all pairs of adjacent ensembles .
next , we perform @xmath75 sampling trials for fixed maximum particle displacements , maximum rotational displacements , maximum area changes , and maximum changes of the lattice vectors
. verlet neighbor lists [ 56 ] are used to improve performance .
we set @xmath76 ellipsoids and @xmath65 as a function of the pressure range to be covered .
@xmath76 is sufficiently large in view of xu et .
analysis of system size effects [ 33 ] .
more details on the employed methods are given in previous works [ 52 ] .
, 4 , 5 , 10 and 20 from left to right .
the curves are the results of pl theory , while the diamond symbols correspond to mc simulation results .
the hard ellipse system is obtained from oblate ellipsoids by tilting their orientations into the xy plane .
, title="fig : " ] , 4 , 5 , 10 and 20 from left to right .
the curves are the results of pl theory , while the diamond symbols correspond to mc simulation results .
the hard ellipse system is obtained from oblate ellipsoids by tilting their orientations into the xy plane .
, title="fig : " ] in this section we present our theoretical and simulation results for hard ellipses , freely rotating and orientationally restricted monolayers of hard ellipsoids .
the equation of state , the surface coverage , the in - plane and out - of - plane order parameters and the isotropic - nematic ( in ) transition densities are determined for various values of shape anisotropy .
we start with the system of hard ellipses , which is obtained by setting the polar angles of all ellipsoidal particles ( @xmath6 ) to be @xmath11 .
this condition is accompanied by @xmath77 and that the uniaxial order parameter ( @xmath56 ) is always -1/2 and the biaxial order parameter can be obtained from @xmath78 .
note that @xmath59 now serves as a 2d orientational order parameter .
the theoretical calculations and simulations can also be carried out with both oblate - shaped ( @xmath5 ) and prolate - shaped hard ellipsoids ( @xmath4 ) , where @xmath79 for all particles .
here we perform the hard ellipse study using oblate - shaped ellipsoids .
[ fig2 ] shows the pressure and the 2d order parameter ( @xmath59 ) as a function of density for @xmath80 , 4 , 5 , 10 and 20 .
one can see that more particles on the surface ( higher densities ) are required to maintain the same value of the pressure with increasing shape anisotropy .
this is due to the fact that the xy plane becomes more spacious with increasing @xmath81 , i.e. @xmath2 must decrease at fixed @xmath3 . .
] it can be also seen in fig .
[ fig2 ] that the simulation data are well reproduced by the approximate pl theory except the values of the order parameters of the n phases corresponding to extremely anisotropic particles ( @xmath82 and 20 ) .
these data well agree with those given in [ 33 ] .
the order parameter curves reveal for the occurrence of isotropic - nematic phase transition , because the phase is isotropic at low densities ( @xmath83 ) , while it is nematic at high ones ( @xmath84 ) .
the phase transition is second order in the theory , while it is higher order ( kosterlitz - thouless type continuous transition ) in the simulation .
both the theory and the simulation show the same tendencies for in transition densities and packing fractions ( see fig .
[ fig3 ] ) . making the ellipsoids more anisotropic
one can see that the in transition density increases , while the in packing fraction decreases .
this apparent contradiction is due to the fact that the increasing shape anisotropy makes the system more spacious ( @xmath3 is constant , while @xmath2 is decreasing ) , which requires more particles for the initiation of the phase transition , while the xy plane can be less occupied at the same time . it can be also seen that the theory underestimates in transition densities and packing fractions in higher extent with increasing shape anisotropy , which is due to the fact that the contribution of higher order virial coefficients are not negligible with increasing @xmath81 . .
the effective aspect ratio of the corresponding hard ellipse system ( @xmath85 ) is shown in the inset of the upper panel , while the out - of - plane order parameter ( @xmath56 ) is presented in the inset of the lower panel .
the curves are the results of pl theory , while the diamond symbols correspond to mc simulation results .
the snapshot is the result of replica exchange monte carlo simulation method .
, title="fig : " ] . the effective aspect ratio of the corresponding hard ellipse system ( @xmath85 ) is shown in the inset of the upper panel , while the out - of - plane order parameter ( @xmath56 ) is presented in the inset of the lower panel .
the curves are the results of pl theory , while the diamond symbols correspond to mc simulation results .
the snapshot is the result of replica exchange monte carlo simulation method .
, title="fig : " ] the monolayer of freely rotating hard ellipsoids can be achieved by using @xmath13 and @xmath12 limiting angles for oblate and prolate shapes , respectively .
the resulting surface coverage , effective aspect ratio , pressure and order parameters are shown for prolate and oblate shaped ellipsoids in figs . [ fig4 ] and [ fig5 ] , respectively . in these figures
the effective aspect ratio of the corresponding hard ellipse system is given by @xmath86 , which is the average aspect ratio of the ellipses obtained by intersecting the ellipsoids by the xy plane . in turn
, the surface coverage can be obtained with the help of @xmath87 as follows : @xmath88 .
starting with prolates ellipsoids , one can see that the particles do not form an isotropic phase , but they align along the normal of the xy plane ( see the simulation snapshot of fig . [ fig4 ] ) even at very low densities .
in fact , an isotropic phase would appear only at the limit of infinite dilution .
the density dependence of the surface coverage and the effective aspect ratio shows that the particles have less and less chance to lie into the xy surface with increasing density because of packing effects .
the intersection of the ellipsoid with the xy surface is practically a hard disk for @xmath89 , where the effective ellipse aspect ratio is almost one .
this makes the monolayer of hard ellipsoids very similar to the system of hard disks at high densities despite the presence of orientational fluctuations . the in - plane order parameter ( eq .
10 ) is always zero , while the out - of - plane order parameter ( eq .
9 ) is positive .
this shows that the out - of - plane orientational ordering is uniaxial with nematic director parallel to the layer normal .
it can be also seen that @xmath56 is very close to its maximum value ( @xmath90 ) at @xmath91 , which corresponds to the case where all prolate ellipsoids are parallel and align along the layer normal .
the reason why prolate ellipsoids prefer the out - of - plane ordering is that they can maximize the free area available on the surface with the penalty of orientational entropy loss .
moreover the close packing structure of prolate ellipsoids is identical to that of 2d hard disks .
one can also see that the theory reproduces quite well the simulation data for all properties except the equation of state at high densities .
regarding the possible freezing transition at high densities , it has not been examined here , because it is not the scope of our present study . .
the effective aspect ratio of the corresponding hard ellipse is shown in the inset of the upper panel , while the out - of - plane ( @xmath56 ) and the in - plane ( @xmath59 ) order parameters are presented in the inset of the lower panel .
the curves are the results of pl theory , while the diamond symbols correspond to mc simulation results .
the snapshot is the result of replica exchange monte carlo simulation method .
, title="fig : " ] . the effective aspect ratio of the corresponding hard ellipse is shown in the inset of the upper panel , while the out - of - plane ( @xmath56 ) and the in - plane ( @xmath59 ) order parameters are presented in the inset of the lower panel .
the curves are the results of pl theory , while the diamond symbols correspond to mc simulation results .
the snapshot is the result of replica exchange monte carlo simulation method .
, title="fig : " ] the principle of minimizing the intersected area with the xy plane applies also for oblate ellipsoids ( see fig .
[ fig5 ] ) . the minimal intersected area , which is actually an ellipse with @xmath92 and @xmath38 dimensions , can be achieved with ordering into the xy plane , while the out - plane ordering result in higher intersected area , because @xmath93 for oblate ellipsoids .
the resulting in - plane ordering can be seen in the simulation snapshot and from the high density behavior of the effective aspect ratio ( @xmath94 ) .
the out of plane order parameter goes to -1/2 , which corresponds to complete in - plane ordering .
this in - plane order is isotropic at low densities ( @xmath83 ) , while it is nematic at high densities ( @xmath84 ) .
this means that the system undergoes a phase transition from a planar uniaxial nematic order ( @xmath95 , @xmath83 ) to a biaxial nematic one ( @xmath95 , @xmath96 ) , which can be considered as a 2d in phase transition .
the order of the uniaxial nematic - biaxial nematic ( n - bn ) phase transition is proved to be second order in the approximate pl theory , while the transition is higher order continuous in the simulation .
this transition is the result of the competition between in - plane orientational entropy ( favoring disorder ) and the in - plane packing entropy ( favoring order ) . the high density structure of the hard ellipsoid monolayer resembles the nematic phase of the 2d hard ellipses , while at low densities the structure more or less corresponds to an isotropic phase of a polydisperse mixture of hard ellipses .
one can also see that the agreement between the theory and the simulation is quite good for oblate ellipsoids , especially with respect to the equation of state .
an interesting feature is that the theory overestimates the out - of - plane ordering ; this could stem from the effect of higher virial coefficients , which are not included in the theory .
[ fig6 ] shows the in - plane isotropic - nematic ( or n - bn ) transition densities as a function of aspect ratio resulting from the pl theory and simulation . on the other hand ,
the transition density increases with increasing shape anisotropy , while the packing fraction decreases .
this is due to the fact that we go to the `` volumeless '' hard needle limit with decreasing @xmath74 , where the transition density saturates at finite value , while the packing fraction vanishes ( note that @xmath97 and @xmath88 ) .
the agreement for the transition densities is again very good between the theory and simulation . to see the effect of out - of - plane orientational freedom it is worth plotting the in densities of 2d hard ellipse system and those of freely rotating hard oblate monolayer together ( see fig .
[ fig7 ] ) .
it can bee seen that the simulation does not show a substantial difference between the two systems , while the theory predicts that the out - of - plane freedom destabilizes the nematic order , i.e. the in density curve is shifted into the direction of higher densities . in the simulation
the nematic phase evolves from a very ordered planar phase ( isotropic in the xy plane ) , where the effective aspect ratio is almost identical with the aspect ratio of the ellipsoids ( see fig .
[ fig5 ] ) , i.e. the corresponding hard ellipse system is almost monodisperse .
this is not the case in the theory , where the transition happens at such densities , where the corresponding hard ellipse system is still polydisperse with larger shape anisotropy ( @xmath87 ) than @xmath74 . as a result
the effect of out - of - plane orientational freedom is more pronounced in the theory than in the simulation .
this shows that the theory exaggerates the effect of out - of - plane orientational freedom on the n - bn phase transition .
finally we show how the in transition properties change with switching on the out of plane orientational freedom through the varying limiting polar angle ( @xmath0 ) , which is between 0 and @xmath11 for both prolate and oblate shapes .
if @xmath13 ( @xmath98 ) the prolate ellipsoids on the xy plane behave like the system of hard ellipses , i.e. they undergo a 2d in phase transition with increasing density ( see fig .
[ fig8 ] ) .
however this phase transition is destabilized with decreasing the limiting angle ( @xmath0 ) , which corresponds to increasing @xmath99 . as the orientational window
is widened ( decreasing @xmath0 ) , the prolate ellipsoids tend to minimize their occupied area on the xy surface , which results in less anisotropic in - plane ellipses and higher transition densities for all studied aspect ratios .
in addition to this , the in - plane order transforms continuously into out - of - plane order , i.e. s becomes positive with @xmath99 ( see fig .
[ fig8 ] ) .
hence , the 2d in transitions turn into n - bn transitions for confined prolates which are lost for the free rotating case . the surface coverage ( packing fraction )
curves show very clearly that the occupied area on the xy surface increases enormously with decreasing the polar angle restriction , i.e. the 2d in phase transition is destabilized with decreasing @xmath0 .
the value of @xmath0 where the orientational ordering transition is preempted by the positional one ( freezing transition ) can not be determined with the present pl theory .
interestingly , mc simulation shows that the in density is practically not affected by the value of limiting polar angle , but the transition disappears at values of @xmath0 close to zero .
note that the in transition density curves move into the direction of lower densities with increasing shape anisotropy because the occupied area of the ellipsoids becomes larger with increasing @xmath74 as @xmath100 .
the case of oblate ellipsoids is different because @xmath13 corresponds to the freely rotating limit , while @xmath12 is the hard disk limit .
starting from the freely rotating case one can see that the in transition density shows a maximum at an intermediate value of @xmath0 , which can be attributed to the combined effect of decreasing in - plane shape anisotropy and the increasing intersected area of the ellipsoids with decreasing @xmath0 .
the n - bn transition packing fraction behaves simply since it is always an increasing function of @xmath99 , i.e the nematic phase is destabilized .
this means that the nematic - biaxial nematic curve and the isotropic - solid curves must cross each other at a threshold value of @xmath99 making the nematic order metastable and producing a plastic solid ( an orientationally disordered solid ) .
this plastic solid would eventually turn into an ordered solid at higher densities .
however , this value of @xmath99 has not been searched by the simulation and theory , because our study focuses only on the orientational ordering .
finally , the @xmath56 order parameter shows that the system undergoes a structural change from in - plane order into out - of - plane order at the n - bn transition .
our results show that the effect of out - of - plane orientational freedom is to minimize the occupied area on the surface which allows the most efficient packing of the ellipsoids on the xy plane . .
the upper panel is for prolate shapes while the lower one for oblate shapes .
insets show the @xmath99 dependence of the surface coverage and the out - of - plane order parameter at the n - bn ordering transition .
the curves are the results of pl theory .
, title="fig : " ] .
the upper panel is for prolate shapes while the lower one for oblate shapes .
insets show the @xmath99 dependence of the surface coverage and the out - of - plane order parameter at the n - bn ordering transition .
the curves are the results of pl theory .
, title="fig : " ]
we have examined the orientational ordering properties of the monolayer of hard ellipsoids using replica exchange monte carlo simulation method and the parsons - lee density functional theory .
we have found that both the shape anisotropy and the orientational restriction affect substantially the orientational order on the plane . the subtle interplay between orientational and packing entropies results in different structures and phase behaviors for prolate and
oblate shaped ellipsoids .
the main driving force of the ordering behavior is to minimize the intersected area between the particle and the surface to realize the close packing structure with increasing surface density .
the minimal intersected area is a disk for prolate , while it is an ellipse for oblate ellipsoids .
highly packed structures can be achieved with only out - of - plane ( in - plane ) ordering for prolate ( oblate ) ellipsoids , i.e. prolate ellipsoids prefer to order along the normal of the plane , while oblate ellipsoids like to order in the plane .
the monolayer of prolate ellipsoids with very small out - of - plane orientational freedom behaves almost identically to the system of 2d hard ellipses , i.e. it form 2d isotropic and nematic phases .
the gradual rise of the out - of - plane freedom , through decreasing @xmath0 , allows the ellipsoid particle to lean out from the plane and to decrease the intersected area with the plane .
this involves less anisotropic shape in the interactions and destabilization of the biaxial nematic phase ( 2d nematic ) with respect to uniaxial nematic one ( 2d isotropic ) . at full orientational freedom ( @xmath101 ) prolate ellipsoids
do not form an isotropic phase , but they are ordered along the normal of the confining plane even at very low densities and behave similarly to the 2d system of hard disks at high densities .
the monolayer of oblate ellipsoids without out - of - plane freedom is identical to the 2d system of hard disks .
the gradually increasing freedom in polar angle ( @xmath6 ) allows the oblate ellipsoid to decrease its intersected area with the confining plane through leaning out from the plane .
this makes the ordering planar for the main symmetry axis of the ellipsoid .
however the anisotropic interactions between the tilted ellipsoids give rise to an additional in - plane order , i.e. the phase is biaxial nematic , which is due to the excluded area gain coming from the in - plane ordering of elliptical intersections . in the freely rotating case the oblate ellipsoids form a planar nematic phase at low densities and a biaxial nematic one at high densities .
interestingly , the in density of 2d ellipses is almost identical to that of freely rotating ellipsoids confined to a plane . in summary the out - of - plane orientational freedom
stabilizes the in - plane nematic ordering in the monolayer of oblate ellipsoids , while the opposite happens in the monolayer of prolate ellipsoids .
this is congruent with the 2d ellipse and 2d disk phase diagrams , which are obtained as limiting cases in this study . in our simple model
we have neglected the effect of out - of - plane positional freedom which we expect to have a relatively small effect on the system behavior . on the other hand ,
not including a soft wall - particle interaction may lead to deviations from experimental set - ups where capillary phenomena are frequently important .
no doubt , the inclusion of this interaction would substantially increase the computation burden of the problem .
we leave this issue for future studies .
sv acknowledges the financial support of the hungarian state and the european union under the tamop-4.2.2.a-11/1/konv-2012 - 0071 .
grant fis2013 - 47350-c5 - 1-r from ministerio de educacin y ciencia of spain is also acknowledged .
56 s. sacanna , d. j. pine and gi - ra yi , soft matter , * 9 * , 8096 ( 2013 ) .
w. qi , j. de graaf , f. qiao , s. marras , l. manna , and m. dijkstra , nano lett .
* 12 * , 5299 ( 2012 )
. k. j. lee , j. yoon and j. lahann , current opinion in colloid and interface science * 16 * , 195 ( 2011 ) .
d. florea and h. m. wyss , j. coll .
interface sci . * 416 * , 30 ( 2014 ) .
n. f. bouxsein , c. leal , c. s. mcallister , k. k. ewert , y. li , c. e. samuel , and c. r. safinya , , j. am .
soc . * 133 * , 7585 ( 2011 ) .
ravi kumar pujala , nidhi joshi , h. b. bohidar , colloid polym .
3651 ( 2015 ) , doi 10.1007/s00396 - 015 - 3651 - 3 . t. e. herod and r. s. duran , langmuir * 14 * , 6606 ( 1998 ) .
f. kim , s. kwan , j. akana , and p. d. yang , j. am .
soc . * 123 * , 4360 ( 2001 ) .
h. razafindralambo , a. richel , m. paquot , l. lins and c. blecker , j. phys .
b * 116 * , 3998 ( 2012 ) .
g. b. davies , t. krger , p. v. coveney , j. harting and f. bremse , adv .
* 26 * , 6715 ( 2014 )
. g. b. davies , t. krger , p. v. coveney , j. harting and f. bremse , soft matter * 10 * , 6742 ( 2014 ) .
z. zheng , f. wang , and y. han , phys .
107 * , 065702 ( 2011 ) . c. k. mishra , a. rangarajan , and r. ganapathy , phys .
110 * , 188301 ( 2013 ) .
e. l. hansen , s. jabbari - farouji , h. mauroy , t. s. plivelic , d. bonne and j. o. fossuma , soft matter * 9*,9999 ( 2013 ) .
l. s. li and a. p. alivisatos , adv .
* 15 * , 408 ( 2003 ) .
k. bneut , d. constantin , p. davidson , a. dessombz , and c. chanac , langmuir * 24 * , 8205 ( 2008 ) .
d. constantin , p. davidson , and c. chanac , langmuir * 26 * , 4586 ( 2010 ) .
j. a. cuesta and d. frenkel , phys .
a * 42 * , 2126 ( 1990 ) .
a. m. somoza and r. c. desai , j. phys .
96 * , 1401 ( 1992 ) .
d. kramer , a. ben - shaul , z .- y . chen and w. m. gelbart , j. chem .
* 96 * , 2236 ( 1992 ) .
m. r. fisch and c. rosenblatt , j. phys .
ii * 4 * , 103 ( 1994 ) .
h. schlacken , h .- j .
mogel , and p. schiller , mol . phys . * 93 * , 777 ( 1998 ) .
s. varga and i. szalai , mol . phys . * 95 * , 515 ( 1998 ) . s. varga and i. szalai , j. mol . liq .
* 85 * , 11 ( 2000 ) .
r. van roij , m. dijkstra and r. evans , j. chem .
113 * , 7689 ( 2000 ) .
r. van roij , m. dijkstra and r. evans , europhys . lett . * 49 * , 350 ( 2000 ) .
a. c. d. van enter and s. b. shlosman , phys .
lett . * 89 * , 285702 ( 2002 ) .
m. cosentino lagomarsino , m. dogterom , and m. dijkstra , j. chem .
phys . * 119 * , 3535 ( 2003 ) .
y. martnez - ratn , e. velasco , and l. mederos , j. chem . phys .
* 122 * , 064903 ( 2005 ) . y.
martnez - ratn , e. velasco , and l. mederos , j. chem . phys . *
125 * , 014501 ( 2006 ) .
r. l. c. vink , phys .
lett . * 98 * , 217801 ( 2007 ) . y. martnez - ratn , liq
* 38 * , 697 ( 2011 ) .
wen - sheng xu , yan - wei li , zhao - yan sun , and li - jia an , j. chem . phys . * 139 * , 024501 ( 2013 ) . ravi kumar pujala and h. b. bohidar , phys .
e * 88 * , 052310 ( 2013 ) .
g. bautista - carbajal and g. odriozola , j. chem . phys .
* 140 * , 204502 ( 2014 ) .
wen - sheng xu , xiaozheng duan , zhao - yan sun , and li - jia an , j. chem .
142 , 224506 ( 2015 ) .
d. frenkel and r. eppenga , phys .
a * 31 * , 1776 ( 1985 ) .
z. zhang , p. pfleiderer , a. b. schofield , c. clasen , and j. vermant , j. am .
soc . * 133 * , 392 ( 2011 ) .
m. gonzlez - pinto , y. martnez - ratn , e. velasco and s. varga , pccp * 17 * , 6389 ( 2015 ) .
y. martnez - ratn , s. varga and e. velasco , j. chem . phys . *
140 * , 204906 ( 2014 ) .
e. meneses - juarez , s. varga , and g. odriozola , soft matter * 9 * , 5277 ( 2013 ) .
s. varga , e. meneses - juarez , and g. odriozola , soft matter * 9 * , 11178 ( 2013 ) .
j. d. parsons , phys . rev .
a * 19 * , 1225 ( 1979 )
. s. d. lee , j. chem .
phys . * 87 * , 4972 ( 1987 ) ; s. d. lee , j. chem .
phys . * 89i * , 7036 ( 1988 ) .
s. c. mcgrother , d. c. williamson , and g. jackson , j. chem .
phys . * 104 * , 6755 ( 1996 ) . p. j. camp , c. p. mason , m. p. allen , a. a. khare and d. a. kofke , j. chem .
* 105 * , 2837 ( 1996 ) .
p. padilla and e. velasco , j. chem .
phys . * 106 * , 10299 ( 1997 ) .
h. reiss , h. l. frisch , and j. l. lebowitz , j. chem .
* 31 * , 369 ( 1959 ) .
e. marinari and g. parisi , europhys
* 19 * , 451 ( 1992 ) .
a. p. lyubartsev , a. a. martinovski , s. v. shevkunov , and p. n. vorontsov - velyaminov , j. chem . phys . * 96 * , 1776 ( 1992 ) . k. hukushima and k. nemoto , j. phys .
jpn . * 65 * , 1604 ( 1996 ) .
g. bautista - carbajal , a. moncho - jord and g. odriozola , j. chem .
* 138 * , 064501 ( 2013 ) .
g. odriozola , j. chem .
phys . * 131 * , 144107 ( 2009 ) .
t. okabe , m. kawata , y. okamoto and m. mikami , chem .
* 335 * , 435 ( 2001 ) .
a. donev , f. h. stillinger , p. m. chaikin s. and torquato , phys .
rev . lett . * 92 * , 255506 ( 2004 ) .
a. donev , s. torquato and f. h. stillinger , j. comp . phys . * 202 * , 737 ( 2005 ) . | the effect of out - of - plane orientational freedom on the orientational ordering properties of a monolayer of hard ellipsoids is studied using parsons - lee scaling approach and replica exchange monte carlo computer simulation .
prolate and oblate ellipsoids exhibit very different ordering properties , namely , the axes of revolution of prolate particles tend to lean out , while those of oblate ones prefer to lean into the confining plane .
the driving mechanism of this is that the particles try to maximize the available free area on the confining surface , which can be achieved by minimizing the cross section areas of the particles with the plane . in the lack out - of - plane orientational freedom
the monolayer of prolate particles is identical to a two - dimensional hard ellipse system , which undergoes an isotropic - nematic ordering transition with increasing density . with gradually switching on the out - of - plane orientational freedom
the prolate particles lean out from the confining plane and a destabilisation of the in - plane isotropic - nematic phase transition is observed .
the system of oblate particles behaves oppositely to that of prolates .
it corresponds to a two - dimensional system of hard disks in the lack of out - of - plane freedom , while it behaves similar to that of hard ellipses in the freely rotating case . |
the standard cold dark matter ( scdm ) model @xcite was in good health until around the last decade of the previous century when it became apparent that the fractional density of matter falls well below the einstein - de sitter value , @xmath0 -see e.g. @xcite .
the death blow came at the close of the century with the discovery of the current cosmic acceleration @xcite , something the said model can not accommodate by any means .
however , to account for the acceleration in homogeneous and isotropic models one must either introduce some exotic energy component with a huge negative pressure ( dubbed dark energy ) or , more drastically , devise some theory of gravity more general than einstein relativity @xcite .
thus , both solutions appear somewhat forced and not very aesthetical . here
we argue that dark energy ( or something equivalent ) is demanded on thermodynamic grounds .
in other words , we provide what we believe is a sound thermodynamic motivation for the existence of dark energy .
our argument is based on that the natural tendency of systems to evolve toward thermodynamical equilibrium is characterized by two properties of its entropy function , @xmath1 , namely , it never decreases , @xmath2 , and is convex , @xmath3 @xcite . in the context of an ever expanding friedmann - roberson - walker ( frw ) cosmology
this translates in that the entropy of the apparent horizon plus that of matter and fields enclosed by it must fulfill @xmath4 at any @xmath5 -the generalized second law ( gsl)- as well as @xmath6 as @xmath7 , where @xmath5 is the scale factor of the frw metric and the prime means @xmath8 .
the apparent horizon in frw universes always exists ( which is not generally true for the particle horizon and the future event horizon ) and is known to posses not only an entropy proportional to its area @xcite but also a temperature @xcite . besides
, it appears to be the appropriate thermodynamic boundary @xcite .
before we proceed , it is fair to recall the ( at least theoretical ) existence of systems lacking any global maximum entropy state , such as antonov s sphere @xcite .
the latter system consists in a sphere enclosing a number of particles that share some total energy .
if the sphere radius happens to increase beyond some critical value , the system becomes unstable and the entropy function ceases to have a global maximum .
we shall apply our argument under the assumption that the universe tends to a state of maximum entropy irrespective of whether it may eventually reach it or not .
it would be odd and frustrating that such a universal principle as the second law of thermodynamics could not be applied to the universe as a whole , especially given the close connection between thermodynamics and gravity @xcite .
section ii illustrates why dark energy ( or some or other modified gravity model ) is required on thermodynamic basis and study some dark energy models to see whether they fulfill the thermodynamic criteria .
section iii applies the said criteria to some representative modified gravity models .
finally , section iv summarizes and discusses our findings .
as is customary , a naught subscript stands for the present value of the corresponding quantity .
let us consider a frw universe .
its entropy is contributed by two terms : the entropy of the apparent horizon which is proportional to its area , @xmath9 , and the entropy of the fluids enclosed by the horizon . here
@xmath10 denotes the radius of the horizon and @xmath11 the hubble factor of the frw metric @xcite . as is well - known , @xmath12 where @xmath13 and @xmath14 stand for planck s length and boltzmann s constant , respectively . in virtue of the first friedmann equation @xmath15 where @xmath16 is the total energy density
, we can write @xmath17 as well as @xmath18 where we have used the conservation equation @xmath19 , where @xmath20 denotes the overall equation of state parameter , i.e. , not just of dark energy .
from ( [ primea1 ] ) the area will augment in expanding universes if @xmath21 and decrease otherwise . a further derivation with @xmath22 constant yields @xmath23 accordingly , @xmath24 for @xmath25 , and @xmath26 otherwise .
thus , the above criterion disfavor the dominance of fluids at late times such that the overall equation of state is either of phantom type or larger than @xmath27 .
we note in passing that when @xmath28 constant last equation generalizes to @xmath29\ , .
\label{2primeawvar}\ ] ] let us consider the entropy associated to the fluid enclosed by the apparent horizon . if the fluid is just dust ( pressureless matter , subscript @xmath30 ) , i.e. @xmath31 and @xmath32 , the unphysical result that the temperature vanishes follows whence euler s equation , @xmath33 , can not be used to determine the entropy of cold matter ( i.e. , dust ) .
we proceed instead as follows .
every dust particle contributes a given bit -say @xmath14- to the fluid entropy .
so , within the apparent horizon we will have @xmath34 being @xmath35 the number of particles there , and @xmath36 the ( conserved ) number density of dust particles .
then , @xmath37 hence @xmath38 we note parenthetically that the entropy augments in the volume enclosed by the horizon because the latter encompasses more and more particles as the universe expands ( i.e. , as @xmath11 diminishes ) .
thus , @xmath39 .
we may now readily understand why sooner or later the universe should accelerate , i.e. , why it must be endowed with dark energy ( subscript @xmath40 , equation of state @xmath41 ) , or something dynamically equivalent at the background level ( as a suitably modified gravity ) .
we note parenthetically that , at late times , dark energy is to dominate over all other energy components whereby @xmath42 as @xmath43 . were the universe dominated by radiation and/or matter for ever , @xmath44 could never become negative . and
if the horizon entropy dominated the total entropy , then the universe would never tend to a state of maximum entropy ( compatible with the constraints of the system ) .
figure [ fig : threegraphs ] sketches the evolution of the horizon area for a spatially flat frw universe dominated by a mixture of cold matter and dark energy , i.e. , @xmath45^{-1}$ ] for @xmath46 ( solid line ) , @xmath47 ( dashed ) , and @xmath48 ( dot - dashed ) -in the three cases we have assumed @xmath49 .
it is readily seen that for @xmath50 ( solid and dashed lines ) the curvature evolves from positive to negative values and stays negative forever .
by contrast , when the dark energy component is phantom , @xmath51 ( dot - dashed line ) the curvature goes from positive to positive values with an intermediate , transient , phase of negative values about the maximum . clearly , while in the two first instances the universe tends to thermodynamic equilibrium at late times in the case of phantom it does not .
needless to say , in the absence of dark energy ( not shown ) we would have @xmath52 for whatever value of the scale factor .
( solid line ) , @xmath47 ( dashed ) , and @xmath53 ( dot - dashed ) . in plotting the graphs we have taken @xmath49.,width=377 ] for completeness , fig .
[ fig : lambda - omegak ] illustrates the influence of the spatial curvature on the evolution of @xmath54 when @xmath55 for three @xmath56cdm models ( @xmath47 ) .
as can be seen , the curvature of the graph for positively curved models remains positive ( recall that @xmath57 ) . ) . from top to bottom , @xmath58 , and @xmath59 , respectively , -see @xcite .
for @xmath5 values of the order unity and below ( not shown ) the graphs practically overlap each other . in plotting the graphs we have chosen @xmath60.,width=377 ] we believe that the gsl alongside the criterion that the total entropy @xmath61 must fulfill @xmath62 at late times may serve to discard some cosmological models and to set limits on the evolution of @xmath63 .
it will prove useful to express @xmath64 and its derivatives in terms of the deceleration parameter , @xmath65 .
the latter can be written as @xmath66 in spatially flat frw universes @xmath67 , whence @xmath68 and @xmath69 \ , .
\label{2primea2}\ ] ] it follows that for @xmath70 the second derivative of the horizon area , @xmath71 , can not go from positive to negative values unless @xmath72 .
this excludes evolutions of @xmath73 as in fig .
[ fig : q(t ) ] .
the latter corresponds to cosmological models in which the current accelerated expansion is just transitory -after a period of dark energy dominance , pressureless matter is assumed to take over @xcite . , given by eq .
( [ decp ] ) , in a frw universe where the present acceleration stage is transitory and reverts for ever to matter domination.,width=377 ] at this stage it seems fitting to ask ourselves whether the second derivative of the entropy of the energy components -that for simplicity we will call fluids , subscript @xmath74- enclosed within the apparent horizon will be positive enough so that the sum @xmath75 be positive .
however , before proceeding we wish to remark that if the energy component is a scalar field in a pure quantum state , or the cosmological constant , it will have no entropy at all .
nevertheless , we wish to consider the possibility of the scalar field being in a mixture state and accordingly entitled to have entropy ( obviously , this excludes the case @xmath76 ) . for simplicity we will take @xmath77 and a single fluid component with @xmath78 constant .
the entropy of the fluid filling the volume enclosed by the apparent horizon can be estimated by virtue of gibbs equation , @xmath79 with help of ( [ eq : friedmann1 ] ) and noting that @xmath80 @xmath81 using the second friedmann equation , @xmath82 and eq .
( [ eq : friedmann1 ] ) once more , we get @xmath83 thus , @xmath84 will be negative or nil for @xmath85 and positive otherwise . from last equation
we can write , @xmath86 the evolution of the temperature of a perfect fluid , @xmath87 , readily follows from gibbs equation and the condition for @xmath88 to be a differential expression -see e.g. @xcite . recalling that @xmath78 constant
, it leads to @xmath89 .
using now @xmath90 we arrive to @xmath91 for @xmath92 ( dark energy ) as well as for @xmath93 one follows that @xmath94 .
however , @xmath95 decreases with expansion for @xmath96 .
the question arises whether the sum @xmath75 will be positive or negative in the long run . to answer it
we inspect the ratio between the right hand sides of eqs .
( [ d2sfluid ] ) and ( [ 2primea1 ] ) and note that @xmath97 , @xmath98 this expression vanishes for @xmath99 as @xmath100 , whereby @xmath101 in the long run provided @xmath102 , i.e. , if the fluid is dark energy .
it remains to be seen whether the gsl , @xmath103 , is fulfilled .
note that in virtue of eqs .
( [ primea1 ] ) and ( [ 1primesf ] ) , with @xmath104 , the ratio @xmath105 vanishes in the long run provided @xmath99 . thus albeit @xmath106 is negative for @xmath107 , the gsl is satisfied for @xmath108 when @xmath109 .
altogether , dark energy with constant equation of state in the interval @xmath110 respects the gsl as well as the criterion that @xmath111 when @xmath43 .
this result is consistent with the tightest observational constraints available -cfr .
table in iv in @xcite . in our view , cosmological models that meet both criteria should be preferred to those failing any of the two . in this subsection
we consider the model of barboza and alcaniz in which the dark energy equation of state is parameterized in terms of redshift as @xcite @xmath112 it has the advantage over other parameterizations of not diverging at any time .
we recall , parenthetically , that the redshift is related to the scale factor by @xmath113 .
the pair of free , constants parameters @xmath114 and @xmath115 in ( [ eq : barboza ] ) ought to be restricted by physical requirements and observation .
a first constraint @xmath116 follows from demanding that dark energy be subdominant at early times ( when @xmath117 ) , otherwise cosmic structure would never had formed .
next , we will use the gsl together with the condition that the sum @xmath118 be negative or nil to set further constraints on the said pair .
we begin by writing the hubble function of a spatially - flat frw universe dominated by cold matter and dark energy , with equation of state ( [ eq : barboza ] ) , @xmath119 \ , .
\label{barboza - hubble}\ ] ] we next consider the entropy of the apparent horizon , @xmath120 .
two separate cases arise : ( i ) @xmath121 and @xmath122 as well as ( ii ) @xmath123 and @xmath124 . in the first case
, @xmath125 grows without bound and we do not consider it any further . in the second one @xmath125 approaches zero asymptotically from above . as @xmath126
, we see that @xmath127 , @xmath128 , @xmath129 and @xmath130 whence all the terms but the last one , that decays as @xmath131 , in the second derivative of the horizon entropy , @xmath132 \ , , \ ] ] present a like behavior . in consequence ,
the quantity in the square brackets tends to a constant , and @xmath133 as @xmath134 . for @xmath135 to be negative that constant must be negative as well , then @xmath136 .
finally , this constraint and the previous one set the accessible range to @xmath137 as shown in fig .
( [ fig : planew0w1 ] ) .
plane for the parameterized-@xmath63 model of barboza and alcaniz @xcite.,width=302 ] obviously , when the entropy of matter and dark energy are taken into account the previous result may vary . with the help of the equation for the temperature of dark energy @xmath138
it can be checked that @xmath139 where @xmath140 and @xmath141 are constants .
hence we focus on matter contribution with respect to the horizon .
when @xmath142 it follows that @xmath143 whereby the horizon contribution prevails in the long run .
however , we have already discarded this solution because @xmath144 diverges . in a sense , this is a consistency check on our criteria because observationally @xmath114 is known to be negative .
in the opposite case , @xmath145 , one obtains @xmath146 again , the horizon contribution prevails , and the result ( [ hor1 ] ) stands .
the frw universe model dominated by cold matter plus dark energy , with equation of state parameterized according to eq .
( [ eq : barboza ] ) , has been observationally constrained in @xcite by using data from supernovae type ia , baryon acoustic oscillations , and cosmic background radiation .
the best fit values of the parameters at @xmath147 confidence level were found to lie in the ranges @xmath148 in full consistency with the above result , eq .
( [ hor1 ] ) . the original chaplygin gas model unifies matter and dark energy in the sense that they are no longer two separate components but a unique entity that mimics cold matter at early times and a cosmological constant at late times @xcite .
its equation of state @xmath149 , with @xmath150 a positive constant and @xmath16 the energy density of the gas ( i.e. , the total energy density ) , is obtainable from the nambu - goto action for a d - brane moving in the @xmath151 dimensional bulk @xcite . in a frw universe , the dependence of the energy density on the scale factor reads @xmath152 , where @xmath153 is nonnegative integration constant ; or equivalently , @xmath154^{1/2 } \label{eq : hubble - chaplygin}\ ] ] with @xmath155 .
the evolution of the horizon area ( @xmath156 ) is akin to the one depicted by the solid line in fig .
[ fig : threegraphs ] ; that is to say , it grows monotonously and presents negative curvature at sufficiently large scale factor .
by contrast , since the entropy of the chaplygin gas obeys @xmath157 , where @xmath158^{-1 } \
, , \label{eq : w - chaplygin}\ ] ] refers to the equation of state parameter of the gas ( which evolves from zero to @xmath159 ) , and the temperature is governed by @xmath160 , one follows that @xmath161 and @xmath162 for large scale factor .
however , as it can be checked , @xmath163 and @xmath164 as @xmath43 .
accordingly , the total entropy , @xmath165 , fulfills @xmath166 and @xmath167 in the same limit . in other words
, the gsl is respected in that limit and the universe tends to thermodynamic equilibrium in the long run . here
we first consider a spatially - flat , holographic , interacting model dominated by pressureless matter and dark energy .
the latter component is assumed holographic in the sense that its energy density varies as the area of the apparent horizon and interacts with matter at a given , non - constant , rate .
as a result @xmath63 decreases with expansion and the fractional densities of both components remain constant .
this much alleviates the cosmic coincidence problem @xcite and the model shows compatibility with observation @xcite .
inspection of the hubble factor , given by eq .
( 2.4 ) of @xcite , @xmath168 \ , , \label{eq : h(a)ivan}\ ] ] readily reveals that the graph of the evolution of the area of the apparent horizon , @xmath169 , has positive curvature at the beginning of the expansion and gently evolves to negative curvature values remaining thus for ever , similarly to the solid line in fig .
[ fig : threegraphs ] , for the best fit observational values , @xmath170 and @xmath171 , obtained in @xcite . as can be readily checked ,
the second derivative of the entropy of the dark energy within the horizon , @xmath172 , is negative as well .
effectively , @xmath173 where @xmath174 and @xmath175 when @xmath43 . for @xmath55
one has that @xmath176 , and one obtains @xmath177 since , in the model at hand , @xmath178 .
+ as for the entropy of pressureless matter @xmath179 where @xmath30 stands for the mass of the particles .
( recall that in this model @xmath180 ) . we then have , @xmath181 and @xmath182 . using ( [ eq : h(a)ivan ] )
we find that @xmath183 and @xmath184 where @xmath185 is a short - hand for the dimensionless combination @xmath186 and @xmath187 ; both quantities lie in the range ( 0,1 ) .
for @xmath55 , the second term is the leading one whence , at sufficiently large scale factor , @xmath188 is also negative . in summary , in this model , @xmath189 for @xmath43 .
it would be misleading , however , to believe that for all holographic models @xmath190 when @xmath55 .
for instance , in the holographic model of gao _ et al . _
@xcite , which uses the ricci s length as infrared cutoff , the hubble equation reads @xmath191 where the subscript @xmath192 stands for radiation ; @xmath193 and @xmath194 are dimensionless parameters . as fig .
[ fig : gao - holographic ] shows , the area of the apparent horizon grows to a maximum to monotonously decrease afterwards for ever .
consequently , @xmath195 as @xmath43
. , @xmath196 , @xmath197 , @xmath198 , @xmath199 , and @xmath200.,width=377 ] on the other hand , the entropies of the fluid components ( radiation , matter , and dark energy ) decrease with expansion and their second derivatives are positive for large scale factor .
thus , this model violates the gsl and does not approach thermodynamic equilibrium at late times .
models that depart from einstein gravity may lead to a late acceleration era without the help of any exotic component of negative pressure . here
we examine some of them , namely : the one based in the brane - induced gravity model of dvali _ et al . _
@xcite ( see @xcite ) , the original cardassian model proposed by freese and lewis @xcite , and the torsion model of bengochea and ferraro @xcite , to check whether they fulfill these criteria .
before we proceed , it is fair to say that , as far as we know , it has not been rigorously proven that the entropy associated to the apparent horizon of the two latter models simply obeys @xmath201 , but it seems to us a very reasonable assumption and we shall adopt it .
this model considers our 4-dimensional universe as a brane embedded in a 5-dimensional bulk with flat minkowski metric . as a consequence of the brane - induced term
, the conventional friedmann s equation modifies to @xmath202 where @xmath16 and @xmath203 stand for the total energy density ( matter plus radiation in this model ) and the crossover scale below which gravity appears as four dimensional , respectively .
the entropy of the apparent horizon , as computed by sheykhi _
@xcite , @xmath204 \ , , \label{eq : sa - dgp}\ ] ] is no longer proportional to the horizon area but contributed by two terms .
the first one corresponds to the bekenstein - hawking entropy and accounts for gravity on the brane .
the second one is related to the bulk .
its evolution in terms of the scale factor is sketched in the left panel of fig .
[ fig : dgp ] . from a certain point on
it increases monotonously with negative curvature . and
@xmath205 , obtained in @xcite.,width=302 ] and @xmath205 , obtained in @xcite.,width=302 ] as before , the entropy of dust and radiation are proportional to @xmath206
. both of them decrease monotonously with positive curvature .
it is found that @xmath207 ( subscript @xmath192 for radiation ) , bear in mind that @xmath203 has physical dimensions of time .
thus , in the long run it leads to a negative constant -though the latter does not exactly coincides in the radiation and matter cases .
once again the question arises whether the total entropy , @xmath208 , obeys the gsl and presents negative curvature when @xmath55 . to answer this
consider the limiting expressions when @xmath43 , @xmath209 \rightarrow \ \ 0 \ , , \label{eq : dgplimiting2a}\\ s^{\prime \prime } & = & s_{a}^{\prime \prime } \left [ 1 \ , + \ , \frac{s_{m}^{\prime \prime } } { s_{a}^{\prime \prime } } \ , + \ , \frac{s_{r}^{\prime \prime } } { % s_{a}^{\prime \prime } } \right ] \rightarrow \ \ 0 \ , .
\label{eq : dgplimiting2b}\end{aligned}\ ] ] the first one tends to zero from above , the second one from below ( following the healthy behavior of the apparent horizon entropy ) .
both relations when taken together imply the inequality @xmath210 \ , , \label{eq : dgpinequality1}\ ] ] which translates into an upper bound on the current number density of dust particles , @xmath211 \sim 10^{38 } \ ; { \rm cm}^{-3 } \ , .
\label{eq : dgpinequality2}\ ] ] since it is fulfilled by a huge margin the gsl is satisfied and @xmath212 results negative in the long run . in this spatially - flat frw model
the first friedmann equation acquires an extra term that accounts for acceleration at sufficiently high redshifts , @xmath213 here two new non - negative constants , @xmath153 and @xmath214 , appear while @xmath16 stands for the energy density of cold matter -the only energy component . for @xmath215
the universe features a transition from @xmath216 to @xmath217 at redshift @xmath218^{1/3(1-\alpha ) } \ , , \ ] ] without need whatsoever of dark energy .
as can be checked , @xmath219 evolves in such a way that while @xmath220 at all redshifts , @xmath221 changes from positive to negative values at some point and stays thus for ever .
this was to be expected since the extra term on the right had side of ( [ eq : cardassian ] ) dynamically amounts to the presence of some dark energy component .
thus , the overall behavior of @xmath54 should be qualitatively similar to that of a model dominated by a mixture of pressureless matter and dark energy . to be more specific ,
comparison of the right hand sides of ( [ eq : cardassian ] ) and the first friedmann equation for a mixture of cold dark matter and dark energy with constant @xmath63 and @xmath77 , namely , @xmath222 $ ] , shows that , at the background level , every cardassian model can be mapped to a spatially - flat dark energy one satisfying @xmath223 and @xmath224 . as a consequence @xmath225 for sensible @xmath226 values ( i.e. , @xmath227 ) . although the entropy of the matter enclosed by the apparent horizon varies as @xmath228 , whence @xmath229 and @xmath230 for @xmath55 , the ratios @xmath231 and @xmath232 tend to zero in the same limit .
it implies that @xmath233 and @xmath234 as @xmath43 .
the torsion model of bengochea and ferraro @xcite falls into a class of models that describes gravitation in terms of the torsion scalar @xmath235 rather than curvature , @xmath236 .
thus , the action takes the form @xmath237 where @xmath238 is a free function to be constrained by observation and experiments .
an important advantage of this set of theories is that the field equations are second order as opposed to the four order equations of @xmath239 gravity .
however , up to now , specific models have been proposed in the cosmological context only .
works investigating the symmetries and dynamics of the theory show that it can exhibit extra degrees of freedom since the theory is not local lorentz invariant @xcite .
bearing in mind that for the frw metric @xmath240 , from ( [ eq : itorsion ] ) the generalized friedmann equations @xmath241 follow . here ,
@xmath16 and @xmath242 denote the energy density and pressure of the fluid component -which , we will assume , just cold matter . in the model of ref .
@xcite @xmath243 where @xmath244 .
we evaluate the entropy of the apparent horizon and matter entropy using the best fit values of the parameters , @xmath245 and @xmath197 -cfr .
left panel of fig .
[ fig : af(t ) ] shows the evolution of area of the apparent horizon in terms of the scale factor .
the plot starts with positive curvature and is ever increasing .
later on , the effect of the dark torsion " is felt and the sign of the curvature changes to remain negative for ever .
the right panel qualitatively depicts the evolution of the entropy of cold matter ( proportional to @xmath206 ) .
it begins increasing to decrease after the maximum .
the curvature is negative about the maximum only .
thus , the question arises whether for @xmath43 the gsl will be respected and the sum @xmath246 will be negative .
a numerical study shows that the ratios @xmath247 and @xmath248 increase with the scale factor when @xmath55 .
therefore the answer to both questions appears to be no .
model of ref .
@xcite ( left panel ) , and the entropy of the cold matter enclosed by the apparent horizon ( right panel ) . in plotting the graphs we employed the best fit values of the said model , @xmath249 and @xmath250.,width=302 ] model of ref .
@xcite ( left panel ) , and the entropy of the cold matter enclosed by the apparent horizon ( right panel ) . in plotting the graphs we employed the best fit values of the said model , @xmath249 and @xmath250.,width=302 ] before closing this subsection it is fair to recall that it is still unknown whether @xmath238 models admit black holes solutions . if they do nt , it will become unclear that some connection entropy - horizon area really holds in these theories .
as we have argued , neither a radiation nor a cold matter dominated universe can tend to thermodynamic equilibrium in the long run .
by contrast , dark energy dominated universes may ; this holds true irrespective of whether the dark energy component has entropy or not .
accordingly , dark energy ( or something dynamically equivalent at the background level , such as a suitably modified gravity theory ) appears thermodynamically motivated . in other words ,
any of these two kind of ingredients was to be expected on thermodynamic grounds .
we , therefore , should not wonder that the universe is accelerating .
however , it does not mean that every accelerating universe is thermodynamically motivated ; that is the case , for instance , of any phantom dominated expansion with @xmath251 , and some modified gravity models .
one may object that if our reasoning were valid , the universe would have never ceased to inflate as it would mean a transition from acceleration , @xmath252 , to deceleration @xmath253 . clearly if the primordial inflation would have lasted for ever , big bang nucleo - synthesis
would never have occurred , galaxies could nt have come into existence , and so on -something in stark contrast with observation . however , this reasoning is rather incomplete as it leaves aside the huge entropy generated during the reheating process at the end of inflation . in this explosive and quasi - instantaneous event
the inflaton field relinquishes all its energy in the form of matter and radiation and , as a consequence , the universe sees its temperature enormously increased @xcite .
since this huge amount of matter and radiation thermalizes ( a necessary condition for primordial nucleosynthesis ) both second derivatives , @xmath188 and @xmath254 , are negative and , as a result , we may well have that @xmath255 .
obviously , this will depend on the specific inflationary model and the particular reheating process involved , but we are not aware of any general argument against the fulfillment of this inequality .
specific calculations in this connection will be the subject of a future research .
interestingly enough , cosmological models complying both with the gsl and the thermodynamic criterion that , at late times , the universe should approach thermodynamic equilibrium appear to be compatible with observational data .
nevertheless , some models that do not comply with the said criteria ( as some phantom models and some modified gravity models ) seem also consistent with the said data . our argument could be falsified if it were discovered that in the past ( but after the radiation - dominated period set in ) the universe experienced one ( or more ) transitions from deceleration to acceleration and back .
cosmologies of the type have been proposed to explain the seemingly periodic distribution of galaxies with redshift -see , e.g. @xcite , @xcite- and as an expedient to solve the cosmic coincidence problem @xcite .
however , recent studies on the impact of hypothetical transient periods of acceleration - deceleration on the matter growth @xcite and on the radiation power spectrum @xcite from the decoupling era , @xmath256 , to @xmath257 practically discard these periods in the redshift intervals considered .
if eventually it gets confirmed that the present phase of acceleration is to last forever , it may be seen as an indication that the universe as a whole obeys the laws of thermodynamics ( with the reservation that not all models that accelerate at late times comply with them ) .
if , on the contrary , the universe resumed a decelerated stage one should either call into question the validity of applying the said laws in a cosmological setting or wait for a later , and definitive , accelerating era .
altogether , it is for the reader to decide which possibility looks the less queer : dark energy ( or modified gravity ) or a universe that will never approach thermodynamic equilibrium .
the remaining possibility , either dark energy or modified gravity combined with an increasing departure from equilibrium , involves two oddities rather than one . 99 p.j.e .
peebles , _ principles of physical cosmology _ ( princeton university press , princeton , nj , 1993 ) .
maddox et al . , _ galaxy correlations on large scales _ , mon . not .
r. astron .
soc . , 43 ( 1990 ) .
g. efstathiou , w.j .
sutherland , and s.j .
maddox , _ the cosmological constant and cold dark matter _ , nature , 705 ( 1990 ) .
_ , _ observational evidence from supernovae for an accelerating universe and a cosmological constant _ , astron .
j. , 1009 ( 1998 ) ; s. perlmutter _
_ , _ measurements of omega and lambda from 42 high - redshift supernovae _ , astrophys .
j. , 565 ( 1999 ) .
r. durrer and r. maartens , _ dark energy and dark gravity : theory overview _ , gen .
grav . , 301 ( 2008 ) ; l. perivolaropoulos , _ accelerating universe : observational status and theoretical implications _ , aip conf .
proc . , 698 ( 2006 ) ; j. frieman , m. turner , and d. huterer , _ dark energy and the accelerating universe _ , ann
astrophys . , 385 ( 2008 )
caldwell and m. kamionkowski , _ the physics of cosmic acceleration _ , ann
, 397 ( 2009 ) ; l. amendola and s. tsujikawa , _ dark energy _ , ( cup , cambridge , 2010 ) .
( j. wiley , ny,1960 ) .
d. bak , s .- j .
rey , _ cosmic holography _ , class .
quantum grav . , l83 ( 2000 ) .
thermodynamics of the apparent horizon in brane world scenarios _ , progr .
suppl . , 100 ( 2008 ) .
g . cai , l .- m .
cao , and y .- p .
hu , _ hawking radiation of an apparent horizon in a frw universe _ , class .
quantum grav .
, 155018 ( 2009 ) .
b. wang , y. gong , and e. abdalla , _ thermodynamics of an accelerated expanding universe _ ,
d. , 083520 ( 2006 ) .
d. lynden - bell and r. wood , mon . not .
r. astron . soc . , 495 ( 1968 ) .
d. lynden - bell , _ negative specific heat in astronomy , physics and chemistry _ , physica a , 293 ( 1999 ) .
t. jacobson , _ thermodynamics of spacetime .
the einstein equation of state _ , phys .
, 1260 ( 1995 ) .
t. padmanbhan , _ gravity and the thermodynamics of horizons _ ,
phys . rep .
, 49 ( 2005 ) ; t. padmanabhan , _ gravity : the inside story _ , j.c .
fabris , _ et al .
_ , _ transient cosmic acceleration from interacting fluids _ , jcap 04(2010)008 .
n. bose and a.s .
majumdar , _ future deceleration due to effect of event horizon on cosmic back reaction _ , arxiv:1010.5071v1 [ astro-ph.co ] . m.o .
calvo , j.a.s .
lima and i. waga , _ on the thermodynamics of matter creation in cosmology _ , phys .
a , 223 ( 1992 ) .
d. pavn and b. wang , _ le chtelier - braun principle and cosmological physics _ , gen .
grav . , 1 ( 2009 ) .
e. komatsu _ et al .
_ , _ seven - year wilkinson microwave ansotropy probe ( wmap)observations : cosmological interpretation _ ,
j. supplement ( in the press ) , arxiv:1001.4538v3 .
e.m . barboza jr . , and j.s .
alcaniz , _ a parametric model for dark energy _ , phys .
b , 415 ( 2008 ) .
kamenschchik , u. moschella , and v. pasquier , _ an alternative to quintessence _ , phys .
b , 265 ( 2001 ) .
r. jackiw , _ a particle field theorist s lectures on supersymmetric , non - abelian fluid mechanics ( and d - branes ) _ , arxiv : physics/0010042 .
w. zimdahl and d. pavn , _ interacting holographic dark energy _ , class .
quantum grav . , 5461 ( 2007 ) .
i. durn , d. pavn , and w. zimdahl , _ observational constraints on a holographic , interacting dark energy model _ , jcap 07(2010)018 .
c. gao , f. wu , x. chen , and y.g .
shen , _ holographic dark energy model from ricci scalar curvature _ , phys .
d , 043511 ( 2009 ) .
g. dvali , g. gabadadze and m. porrati , _
4d gravity on a brane in 5d minkowski space _ ,
b , 208 ( 2000 ) .
c. deffayet , _ cosmology on a brane in minkowski bulk _ ,
b , 199 ( 2001 ) .
c. deffayet , g. dvali , and g. gabadadze , _ accelerated universe from gravity leaking to extra dimensions _
d , , 044023 ( 2002 ) .
k. freese and m. lewis , _ cardassian expansion : a model in which the universe is flat , matter dominated and accelerating _ , phys .
b , 1 ( 2002 ) .
bengochea and r. ferraro , _ dark torsion as the cosmic speed - up _ , phys .
d , 124019 ( 2009 ) .
a. sheykhi , b. wang and r .-
cai , _ deep connection between thermodynamics and gravity in gauss - bonet braneworld _ , nucl .
b , 1 ( 2007 ) .
c. deffayet , s.j .
landau , j. raus , m. zaldarriaga and p. astier , _ supernovae , cmb , and gravitational leakage into extra dimensions _ ,
d , 024019 ( 2002 ) .
b. li , t. sotirou , and j.d .
@xmath258 gravity and local lorentz invariance _ ,
d.h . lyth and a.r .
liddle , _ the primordial density perturbation _ ( cup , cambridge , 2009 ) .
m. morikawa , _ oscillating universe : the periodic distribution of galaxies with a scale @xmath259 megaparsecs at the galactic poles _
, astrophys .
j. , l37 ( 1990 ) .
n. banerjee , d. pavn , and s. sen , _ peridoc distribution of galaxies in generalized scalar tensor theory _ ,
, 851 ( 2003 ) .
r. lazkoz , v. salzano , and i. sendra , _ oscillations in the dark energy equation of state : new mcmc lessons _ , phys .
b , 198 ( 2010 ) .
linder , _ uniqueness of current cosmic acceleration _
d , 063514 ( 2010 ) .
e.v . linder and t.l .
smith , _ dark before light : testing the cosmic expansion history through the cosmic microwave background _
, jcap 04(2011)001 . | it is argued that the discovery of cosmic acceleration could have been anticipated on thermodynamic grounds , namely , the generalized second law and the approach to equilibrium at large scale factor .
therefore , the existence of dark energy -or equivalently , some modified gravity theory- should have been expected . in general , cosmological models that satisfy the above criteria
show compatibility with observational data . |
consider @xmath7 with coordinates @xmath8 . a _
diagonal subspace _ @xmath9 is a linear subspace of the form @xmath10 .
a _ diagonal arrangement _ ( or a _ hypergraph arrangement _ ) @xmath3 is a finite set of diagonal subspaces of @xmath7 . throughout this paper
, we assume that for any @xmath11 , @xmath12 is not included in @xmath13 . for a simplicial complex @xmath0 on a vertex set @xmath14 :
= \{1 , 2 , \dots , n \}$ ] such that @xmath15 , one can associate a diagonal arrangement @xmath2 in @xmath7 as follows . for a facet @xmath16 of @xmath0 , let @xmath17 be the diagonal subspace @xmath10 where @xmath18-f = \ { i_1 , \dots , i_r \}$ ] .
define @xmath19 if every subspace in a diagonal arrangement @xmath3 in @xmath7 has the form @xmath9 with @xmath20 , one can find a simplicial complex @xmath0 on @xmath14 $ ] satisfying @xmath21 .
two important spaces associated with an arrangement @xmath3 of linear subspaces in @xmath7 are @xmath22 called the _ complement _ and the _ singularity link_. we are interested in the topology of @xmath4 and @xmath23 for a diagonal arrangement @xmath3 .
diagonal arrangements arise in connection with important questions in many different fields . in computer science , bjrner , lovsz and
yao @xcite found lower bounds on the complexity of @xmath24-equal problems using the topology of diagonal arrangements ( see also @xcite ) . in group cohomology , it is well known that @xmath25 for the braid arrangement @xmath26 in @xmath27 is a @xmath28 space with the fundamental group isomorphic to the pure braid group @xcite .
khovanov @xcite showed that @xmath29 for the @xmath6-equal arrangement @xmath30 in @xmath7 is also a @xmath28 space .
note that @xmath4 and @xmath23 are related by alexander duality as follows : @xmath31 in the mid-1980 s goresky and macpherson @xcite found a formula for the betti numbers of @xmath4 , while the homotopy type of @xmath23 was computed by ziegler and ivaljevi @xcite ( see section [ set - homotopy ] ) .
the answers are phrased in terms of the lower intervals in the _ intersection lattice _
@xmath32 of the subspace arrangement @xmath3 , that is the collection of all nonempty intersections of subspaces of @xmath3 ordered by reverse inclusion . for general subspace arrangements
, these lower intervals in @xmath33 can have arbitrary homotopy type ( see ) .
our goal is to find a general sufficient condition for the intersection lattice @xmath32 of a diagonal arrangement @xmath3 to be well behaved .
bjrner and welker @xcite showed that @xmath34 has the homotopy type of a wedge of spheres , where @xmath35 is the @xmath24-equal arrangement consisting of all @xmath36 for all @xmath37 ( see section [ sec - earlier - results ] ) , and bjrner and wachs @xcite showed that @xmath34 is shellable .
more generally , kozlov @xcite showed that @xmath32 is shellable if @xmath3 satisfies certain technical conditions ( see section [ sec - earlier - results ] ) .
suggested by a homological calculation ( theorem [ nonpure - by - walls ] below ) , we prove the following main result , capturing the homotopy type assertion from @xcite ( see section [ sec - homotopy ] ) .
[ main - theorem ] let @xmath0 be a shellable simplicial complex .
then the intersection lattice @xmath1 for the diagonal arrangement @xmath2 is homotopy equivalent to a wedge of spheres .
furthermore , one can describe precisely the spheres in the wedge , based on the shelling data .
let @xmath0 be a simplicial complex on @xmath14 $ ] with a shelling order @xmath38 on its facets .
let @xmath39 be the intersection of all facets , and @xmath40 its complement .
let @xmath41 and for each @xmath42 , let @xmath43 be the face of @xmath44 obtained by intersecting the walls of @xmath44 that lie in the subcomplex generated by @xmath45 , where a _ wall of _ @xmath44 is a codimension @xmath46 face of @xmath44 .
an ( _ unordered _ ) _ shelling - trapped decomposition _
( _ of @xmath40 over @xmath0 _ ) is defined to be a family @xmath47 such that @xmath48 is a decomposition of @xmath40 as a disjoint union @xmath49 and @xmath50 are facets of @xmath0 such that @xmath51 for all @xmath52
. then the wedge of spheres in theorem [ main - theorem ] consists of @xmath53 copies of spheres of dimension @xmath54 for each shelling - trapped decomposition @xmath55 of @xmath40 .
moreover , for each shelling - trapped decomposition @xmath56 of @xmath40 and a permutation @xmath57 of @xmath58 $ ] , there exists a saturated chain @xmath59 ( see section [ sec - chain ] ) such that removing the simplices corresponding to these chains leaves a contractible simplicial complex .
the following example shows that the intersection lattice in theorem [ main - theorem ] is not shellable in general , even though it has the homotopy type of a wedge of spheres .
[ ex - not - shellable ] let @xmath0 be a simplicial complex on @xmath60 $ ] with a shelling @xmath61 ( short for @xmath62 ) , @xmath63 .
then the order complex of the upper interval @xmath64 is a disjoint union of two circles , hence is not shellable .
therefore , the intersection lattice @xmath1 for the diagonal arrangement @xmath2 is also not shellable .
the intersection lattice @xmath1 is shown in figure [ nonshellable ] ( thick lines represent the open interval @xmath64 ) . in figures ,
the subspace @xmath65 is labeled by @xmath66 . also note that a facet @xmath16 of @xmath0 corresponds to the subspace @xmath67 - f}$ ] .
for example , the facet @xmath68 corresponds to @xmath69 .
the intersection lattice @xmath1 for example [ ex - not - shellable ] ( @xmath65 is labeled by @xmath66 . ) [ nonshellable ] the next example shows that there are nonshellable simplicial complexes whose intersection lattices are shellable .
let @xmath0 be a simplicial complex on @xmath70 $ ] whose facets are @xmath71 and @xmath72 . then @xmath0 is not shellable .
but the order complex of @xmath73 consists of two vertices , hence is shellable . in section [ sec - earlier - results ] , we give kozlov s result and show that its homotopy type consequence is a special case of theorem [ main - theorem ] . also , we give a new proof of bjrner and welker s result using theorem [ main - theorem ] . in section [ sec - homotopy ] , we prove theorem [ main - theorem ] . in section [ set - homotopy ]
, we deduce from it the homotopy type and the homology of the singularity link ( and hence the homology of the complement ) of a diagonal arrangement @xmath74 for a shellable simplicial complex @xmath0 . in section [ sec - examples ] , we give some examples in which @xmath4 are @xmath5 , coming from matroids .
in this section , we give a few definitions used throughout this paper . first , we start with the definition of ( nonpure ) shellability of simplicial complexes ; see @xcite , @xcite for further background on this notion .
a simplicial complex is _ shellable _ if its facets can be arranged in linear order @xmath75 in such a way that @xmath76 is pure and @xmath77-dimensional for all @xmath78 , where @xmath79 .
such an ordering of facets is called a _ shelling order _ or _
shelling_. there are several equivalent definitions of shellability .
the following restatement of shellability is often useful .
[ restatement - of - shellability ] a linear order @xmath75 of facets of a simplicial complex is a shelling if and only if for every @xmath44 and @xmath80 with @xmath81 , there is a facet @xmath82 such that @xmath83 , where @xmath84 means that @xmath85 has codimension @xmath46 in @xmath16 .
it is well known that a pure @xmath86-dimensional shellable simplicial complex has the homotopy type of a wedge of @xmath86-spheres .
bjrner and wachs @xcite generalized this result to the nonpure case , i.e. , a nonpure shellable simplicial complex has the homotopy type of a wedge of spheres .
however , these spheres need not be equidimensional .
the _ link _ of a face @xmath39 of a simplicial complex @xmath0 is @xmath87 bjrner and wachs @xcite showed that shellability is inherited by all links of faces in a simplicial complex .
[ link - shellable ] if @xmath0 is shellable , then so is @xmath88 for all faces @xmath89 , using the induced order on facets of @xmath88 .
now we give the definition of the order complex of a poset which we will use frequently .
the _ order complex _
@xmath90 of a poset @xmath91 is the simplicial complex whose vertices are the elements of @xmath91 and whose faces are the chains of @xmath91 . for the order complex @xmath92 of an open interval @xmath93
, we will use the notation @xmath94 .
when we say that a finite lattice @xmath95 with bottom element @xmath96 and top element @xmath97 has some topological properties , such as purity , shellability and homotopy type , it means the order complex of @xmath98 has those properties .
in this section , we give kozlov s theorem and show how its consequence for homotopy type follows from theorem [ main - theorem ] . also , we give a new proof of bjrner and welker s theorem about the intersection lattice of the @xmath24-equal arrangements using theorem [ main - theorem ] .
kozlov @xcite showed that @xmath74 has shellable intersection lattice if @xmath0 satisfies some conditions .
this class includes @xmath24-equal arrangements and all other diagonal arrangements for which the intersection lattice was proved shellable .
* corollary 3.2 ) [ kozlov - corollary ] consider a partition of @xmath99 = e_1 \sqcup \cdots \sqcup e_r $ ] such that @xmath100 for @xmath101 .
let @xmath102 be a nondecreasing map .
let @xmath0 be a simplicial complex on @xmath14 $ ] such that @xmath16 is a facet of @xmath0 if and only if 1 .
@xmath103 for @xmath104 ; 2 .
if @xmath105 , then @xmath106 .
then the intersection lattice for @xmath74 is shellable . in particular
, @xmath107 has the homotopy type of a wedge of spheres .
@xmath0 in theorem [ kozlov - corollary ] is shellable .
we claim that a shelling order is @xmath75 such that the words @xmath108 are in lexicographic order , where @xmath109 is the increasing array of elements in @xmath110 .
let @xmath111 be two facets of @xmath0 with @xmath112 .
then @xmath113 . let @xmath114 be the first number in @xmath115 $ ] such that @xmath116 .
construct the word @xmath57 as follows : 1 .
@xmath117 for @xmath118 ; 2 . for @xmath119 ,
@xmath120 where @xmath121 .
note that the length of @xmath57 is @xmath122 and @xmath123 .
let @xmath124 be the set of all elements which do not appear in @xmath57 . since @xmath16 satisfies the two conditions from theorem [ kozlov - corollary ]
, @xmath16 is a facet of @xmath0 .
since @xmath125 and @xmath126 is a subset of @xmath127 of size @xmath46 , @xmath128 has codimension @xmath46 in @xmath127 .
also @xmath129 .
hence @xmath75 is a shelling by lemma [ restatement - of - shellability ] .
[ kozlov - example ] consider the partition of @xmath130 = \ { 1 \ } \sqcup \ { 2 , 3 \ } \sqcup \ { 4 \ } \sqcup \ { 5 , 6 , 7 \}\ ] ] and the function @xmath131 given by @xmath132 , and @xmath133 .
then the facets of the simplicial complex @xmath0 that satisfy the conditions from corollary [ kozlov - corollary ] and the corresponding words can be found in table [ table - kozlov - example ] . [
cols="^,^,^,^,^,^,^,^,^",options="header " , ] shelling - trapped decompositions for @xmath0 [ delta - table ] [ ex - homology - of - link ] let @xmath0 be a simplicial complex from example [ ex - upper - iso ] .
in example [ ex - shelling - trapped - decomposition ] , we show @xmath134 is a shelling of @xmath0 and @xmath135 table [ delta - table ] shows shelling - trapped decompositions @xmath136 of subsets of @xmath137 $ ] with corresponding dimensions @xmath138 therefore corollary [ cor - singularity - link ] shows that the singularity link of @xmath2 is homotopy equivalent to a wedge of three @xmath6-dimensional spheres and eight @xmath139-dimensional spheres .
the following theorem is a homology version of corollary [ cor - singularity - link ] .
[ nonpure - by - walls ] let @xmath0 be a shellable simplicial complex and @xmath140 be a shelling of @xmath0 . then @xmath141 is the number of ordered shelling - trapped decompositions @xmath142 with @xmath143 this can be proven without theorem [ one - set - nonpure ] by combining a result of peeva , reiner and welker with results of herzog , reiner and welker along with the theory of _ golod rings_. it is what motivated us to prove the stronger corollary [ cor - singularity - link ] and eventually theorem [ main - theorem ] .
in this section , we give some examples of diagonal arrangements @xmath3 where the complement @xmath4 is @xmath5 , coming from rank @xmath6 matroids .
one should note that an arrangement having any subspace of real codimension @xmath46 ( hyperplane ) will have @xmath4 disconnected .
so one may assume without loss of generality that all subspaces have real codimension at least @xmath139 .
furthermore , if any maximal subspace @xmath144 in @xmath3 has codimension at least @xmath6 , then it is not hard to see that @xmath4 is not @xmath5 .
hence we may assume without loss of generality that all maximal subspaces have real codimension @xmath139 .
a hyperplane arrangement @xmath145 in @xmath7 is _ simplicial _ if every chamber in @xmath146 is a simplicial cone .
davis , januszkiewicz and scott @xcite showed the following theorem .
[ davis - januszkiewicz - scott - thm ] let @xmath145 be a simplicial real hyperplane arrangement in @xmath7 .
let @xmath3 be any arrangement of codimension-@xmath139 subspaces in @xmath145 which intersects every chamber in a codimension-@xmath139 subcomplex .
then @xmath147 is @xmath28 . in order to apply this to diagonal arrangements
, we need to consider hyperplane arrangements @xmath145 which are subarrangements of the real braid arrangement @xmath26 in @xmath7 and also simplicial .
it turns out ( and we omit the straightforward proof ) that all such arrangements @xmath145 are direct sums of smaller braid arrangements .
so we only consider @xmath148 itself here .
[ davis - januszkiewicz - scott - cor ] let @xmath3 be a subarrangement of the @xmath6-equal arrangement of @xmath7 so that @xmath149 for some collection @xmath150 of @xmath6-element subsets of @xmath14 $ ] .
then @xmath3 satisfies the hypothesis of theorem [ davis - januszkiewicz - scott - thm ] ( and hence @xmath147 is @xmath28 ) if and only if every permutation @xmath151 in @xmath152 has at least one triple in @xmath150 consecutive , i.e. , there exists @xmath153 such that @xmath154 .
it is easy to see that there is a bijection between chambers of the real braid arrangement @xmath26 in @xmath7 and permutations @xmath155 in @xmath152 .
moreover , each chamber has the form @xmath156 with bounding hyperplanes @xmath157 and intersects the @xmath6-equal subspaces of the form @xmath158 for @xmath159 .
we seek shellable simplicial complexes @xmath0 for which @xmath2 satisfies this condition .
if @xmath0 is the independent set complex @xmath160 for some matroid @xmath161 ( see @xcite for the definition of independent sets and further background on matroids ) , then facets of @xmath0 are bases of @xmath161 .
simplicial complexes of this kind are called _ matroid complexes _ , and they are known to be shellable @xcite . for a rank @xmath6 matroid @xmath161 on @xmath14 $ ] , consider @xmath162 - b \text { for some } b \in \mathcal{b}(m^\perp ) \}\\ & = \ { u_{ijk } \mid \ { i , j , k \ } \in \mathcal{b}(m ) \ } , \end{aligned}\ ] ] where @xmath163 is the dual matroid of @xmath161 .
note that adding a loop to @xmath161 does not change the structure of the intersection lattice for @xmath164 .
thus , if @xmath165 on the set of all non - loop elements satisfies the condition of corollary [ davis - januszkiewicz - scott - cor ] , then the diagonal arrangement corresponding to the matroid in which all loops have been deleted has @xmath5 complement , and hence @xmath164 has @xmath5 complement .
let @xmath161 be a rank @xmath6 matroid on @xmath14 $ ] .
say @xmath161 is _ djs _ if @xmath165 on the set of all non - loop elements satisfies the condition of corollary [ davis - januszkiewicz - scott - cor ] .
the following example shows that matroid complexes are not djs in general .
thus we look for some subclasses of matroid complexes which are djs , and hence whose corresponding diagonal arrangements have @xmath28 complements ; for these , theorem [ nonpure - by - walls ] gives us the group cohomology @xmath166 .
let @xmath161 be a matroid on @xmath167 $ ] which has three distinct parallel classes @xmath168 @xmath169 and @xmath170 .
then @xmath161 is self - dual and @xmath171 is a simplicial complex on @xmath167 $ ] whose facets are @xmath172 @xmath173 and @xmath174 . but
@xmath175 is a permutation that does not satisfy the condition of corollary [ davis - januszkiewicz - scott - cor ] .
recall that a matroid is _ simple _ if it has no loops nor parallel elements .
the following proposition shows that rank @xmath6 simple matroids are djs .
let @xmath161 be a matroid of rank @xmath6 on @xmath14 $ ] .
if @xmath161 does not have parallel elements , then @xmath161 is djs .
in particular , rank @xmath6 simple matroids are djs . without loss of generality
, we may assume that @xmath161 is simple .
@xmath161 is not djs if and only if there is a permutation @xmath176 such that every consecutive triple is not in @xmath165 . since @xmath161 is simple
, the latter statement is true if and only if each consecutive triple in @xmath57 forms a circuit , i.e. , all elements lie on a rank @xmath139 flat .
but this is impossible since @xmath161 has rank @xmath6 .
the following two propositions give some subclasses of matroids with parallel elements which are djs .
let @xmath161 be a rank @xmath6 matroid on @xmath14 $ ] with no circuits of size @xmath6 .
let @xmath177 be the distinct parallel classes which have more than one element , and let @xmath178 be the set of all non - loop elements which are not parallel with anything else . then
, @xmath161 is djs if and only if @xmath179 we may assume that @xmath161 does not have loops . since @xmath161 does not have loops nor circuits of size @xmath6 , @xmath161 is not djs if and only if one can construct a permutation @xmath176 such that for each consecutive triple in @xmath57 there are at least two elements which are parallel .
this means if @xmath180 , then @xmath181 and @xmath182 ( if they exist ) must be in the same parallel class . such a @xmath57 can be constructed if and only if @xmath183 .
a simplicial complex @xmath0 on @xmath14 $ ] is _ shifted _ if , for any face @xmath39 of @xmath0 , replacing any vertex @xmath184 by a vertex @xmath185 with @xmath186 gives another face in @xmath0 .
a matroid @xmath161 is _ shifted _ if its independent set complex is shifted .
klivans @xcite showed that a rank @xmath6 shifted matroid on the ground set @xmath14 $ ] is indexed by some set @xmath187 with @xmath188 as follows : @xmath189 it is not hard to check the following .
let @xmath161 be the shifted rank @xmath6 matroid on the ground set @xmath14 $ ] indexed by @xmath190 .
then , @xmath161 is djs if and only if @xmath191 .
we have not yet been able to characterize all rank @xmath6 matroids which are djs .
the author would like to thank his advisor , vic reiner , for introducing the topic of subspace arrangements and encouraging him to work on this problem .
the author also thanks ezra miller , dennis stanton , michelle wachs and volkmar welker for valuable suggestions .
finally , the author thanks the referee for a careful reading of the manuscript , and valuable suggestions for improvement . | we prove that if a simplicial complex @xmath0 is shellable , then the intersection lattice @xmath1 for the corresponding diagonal arrangement @xmath2 is homotopy equivalent to a wedge of spheres .
furthermore , we describe precisely the spheres in the wedge , based on the data of shelling .
also , we give some examples of diagonal arrangements @xmath3 where the complement @xmath4 is @xmath5 , coming from rank @xmath6 matroids . |
an important step for understanding the dynamics and phenomenology of astrophysical jets is the study of their instabilities .
instabilities have a substantial importance , on one hand for the formation and evolution of various observed structures and , on the other hand , for dissipating part of the jet energy and leading to the observed radiation .
there are several possible sources of instabilities , like the velocity shear between the jet and the ambient medium , which drives the kelvin - helmholtz instability , the current flowing along magnetic field lines , which drives the current driven instability ( cdi ) and rotation that can drive several kinds of instabilities . since the most promising models for the acceleration and collimation of jets involve the presence of a magnetic field with footpoints anchored to a rotating object ( an accretion disk or a spinning star or black hole ) , the presence of a toroidal field component and of rotation seems to be a natural consequence and both cdi and rotation driven instabilities may play an important role in the jet propagation .
cdi have been , for example , suggested as being responsible for the conversion from poynting to kinetic energy flux in the first phases of jet propagation @xcite .
khi have been extensively studied in several different configurations both in the newtonian ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and relativistic ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) cases .
similarly , cdi have been widely studied in the newtonian limit ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , while the analysis of the relativistic case has been more limited , most of the studies have considered the force - free condition @xcite and only @xcite studied the full mhd case .
the study of the effects of rotation have been mainly focused on the accretion disk problems , where the main instability considered is the magnetorotational instability @xcite , however , the combination of magnetic field and rotation can give rise to several other instabilities ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and the interplay between the different modes can become quite complex .
our goal is to study these rotation - induced instabilities in the context of magnetized jets . in @xcite ( hereinafter paper i )
we studied the interplay between khi and cdi in a relativistic non - rotating cold jet configuration , characterized by a current distribution concentrated inside the jet and closing at large distances . in this paper , we introduce the effects of rotation which , however , makes the analysis of the unstable modes much more intricate .
therefore , before tackling the full relativistic case , in this paper we limit ourselves to a newtonian analysis , neglecting again thermal pressure compared with the magnetic one .
moreover , since most of the unstable modes that we will consider are concentrated inside the jet radius and therefore the effect of the jet velocity would be to simply doppler shift their frequencies , in this first step we ignored also the presence of the longitudinal velocity component . the main focus of this paper will then be on the effect of rotation on cdi and on the new modes of instability introduced by rotation .
the effect of rotation on cdi was considered by @xcite who analyzed a rigidly rotating jet and found a stabilizing effect for rotation periods shorter than a few alfvn times .
an analysis of the unstable modes introduced by rotation in a configuration and parameter range similar to ours has been performed by @xcite .
they discuss these modes in the cold plasma limit , however , their study is mainly local , whereas we focus more on global analysis of these instabilities , besides they do not discuss the cdi .
another related works are by @xcite and @xcite , who examined the instabilities of axisymmetric perturbations in the presence of rotation and superthermal magnetic fields .
the treatment of the first paper is again local and mostly focuses on an equilibrium configuration typical of accretion discs , while the second one analyses the stability a rotating cylindrical plasma dean flow with only axial field both with local and global approach .
the plan of the paper is the following : in the next section we present the equilibrium configuration , in section [ sec : linequations ] we derive the linearized equations , in section [ modeclassification ] we discuss the wkbj local dispersion relation and present energetic considerations based on the frieman - rotenberg approach @xcite .
the local dispersion relation and the energetic considerations will be useful in understanding the nature of the unstable modes that will be discussed in section [ results ] , where we present our results on global modes . finally in the last section [ summary ] we summarize our findings .
we study the linear stability of a cold magnetized cylindrical jet flow .
although in the following we will consider only the zero thermal pressure case , we keep here the presentation more general .
the relevant equations are the equations of ideal mhd : @xmath0 @xmath1 @xmath2 @xmath3 where @xmath4 is the density , @xmath5 is the pressure , @xmath6 is the sound speed , @xmath7 , @xmath8 , are , respectively , the velocity and magnetic fields .
we remark that a factor of @xmath9 is absorbed in the definition of @xmath8 .
the first step in the stability analysis is to define an equilibrium state satisfying the stationary form of equations ( [ eq : drho / dt]-[eq : dp / dt ] ) and this will be done in the next subsection .
we adopt a cylindrical system of coordinates @xmath10 ( with versors @xmath11 ) and seek for axisymmetric steady - state solutions , i.e. , @xmath12 .
the jet propagates in the vertical ( @xmath13 ) direction , the magnetic field and velocity have no radial component and consist of a vertical ( poloidal ) component @xmath14 , and a toroidal component @xmath15 . the magnetic field configuration can be characterized by the pitch parameter @xmath16 the only non - trivial equation is given by the radial component of the momentum equation ( [ eq : dm / dt ] ) which , in the zero pressure case , simplifies to @xmath17 equation ( [ eq : radial_eqmhd ] ) leaves the freedom of choosing the radial profiles of all flow variables but one and then solve for the remaining profile .
furthermore , we note that the presence of a longitudinal velocity has no effect on the radial equilibrium .
the choice of the radial profiles is somewhat arbitrary since we have no direct information about the magnetic configuration in astrophysical jets .
the choice of the @xmath18 distribution is equivalent to a choice of the distribution of the longitudinal component of the current and also determines the behavior of the pitch parameter @xmath19 , that is important for the stability properties . in principle
, one can then have several equilibria characterized by different forms of the current distribution , that can be more or less concentrated , can peak on the axis or at the jet boundary , and can close in different ways ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
our choice is to consider a general class of constant density equilibria in which the vertical current density is peaked on the axis and is concentrated in a region of radius @xmath20 .
the azimuthal component of magnetic field has therefore to behave linearly with radius close to the origin and decay as @xmath21 at large distances , more precisely we can write it as @xmath22 where the function @xmath23 behaves in the following way at small and large radii : @xmath24 for the rotational frequency @xmath25 we assume the form @xmath26 where @xmath27 is the value of @xmath28 on the axis , i.e. @xmath29 .
we make this assumption for simplicity and for avoiding any possible non - monotonic behaviors of @xmath30 . inserting expressions ( [ eq : bphi ] ) and
( [ eq : omega ] ) in the equilibrium condition ( [ eq : radial_eqmhd ] ) , we can get the profile of @xmath30 as @xmath31 where @xmath32 @xmath33 and @xmath34 is a monotonic function with @xmath35 and @xmath36 . depending on the value of @xmath37 ,
@xmath30 is decreasing ( for @xmath38 ) or increasing ( for @xmath39 ) . from a physical point of view
, if we look at equation ( [ eq : radial_eqmhd ] ) , we can see that the equilibrium is given by the balance of three forces : gradient of @xmath40 , the gradient of @xmath41 and the centrifugal force . when there is no rotation , the gradient of @xmath40 , which always points towards the jet axis , is balanced by the gradient of @xmath41 .
increasing the rotation rate , the gradient of @xmath41 decreases , until , for @xmath42 , @xmath30 becomes constant .
if we still increase rotation beyond this point , the centrifugal term becomes larger than the gradient of @xmath40 , thus @xmath30 has to increase outward for providing an inward force term , needed for having equilibrium .
we can now define a radially averaged alfvn velocity via @xmath43 and inserting the expressions for @xmath18 and @xmath30 from equations ( [ eq : bphi ] ) and ( [ eq : bz ] ) in equation ( [ eq : bav ] ) , we get @xmath44 where @xmath45 from which we get @xmath46 where @xmath47 are , respectively , nondimensional measures of the rotation rate and of the value of the pitch on the axis . for a given choice of the function @xmath23 and the values of these two parameters
, we can derive the value of @xmath48 from equation ( [ eq : hc ] ) and the equilibrium structure is then fully determined .
however not all combinations of @xmath49 and @xmath50 are allowed , because , in order to have a physically meaningful solution , we have to impose the additional constraints that @xmath51 and @xmath52 have to be everywhere positive , which translate as @xmath53 respectively , for equations ( [ eq : hc ] ) and ( [ eq : bz ] ) . in order to exemplify the structure of the equilibrium solution
, we now make a specific choice for the function @xmath23 , @xmath54 which gives the same solution used in paper i. this same solution will also be used in the next sections for computing the instability behavior .
more specifically , we have the following profiles for @xmath51 , @xmath52 and @xmath55 @xmath56,\ ] ] @xmath57 where @xmath58 is the error function , and @xmath59 in fig . 1
, we present the regions in the @xmath60 plane for which the equilibrium is possible . the red curve represents the condition @xmath61 , while the blue curve represents the condition @xmath62 and the green curve represents the combinations of @xmath49 and @xmath50 for which @xmath30 is constant .
therefore the green region represents equilibria for which @xmath63 , while in the red region @xmath64 .
we conclude by summarizing the parameters determining the equilibrium : once the magnetization radius @xmath20 and the average alfvn velocity , @xmath65 , are fixed , the equilibrium structure is fully determined by the two parameters @xmath50 and @xmath49 .
we also observe that an arbitrary profile of the longitudinal velocity @xmath66 can be superposed to the equilibrium , however in this paper we consider only the case @xmath67 .
we consider small perturbations of the form @xmath68 to the equilibrium state described above . by linearizing the ideal mhd equations , we obtain the following system of two first order ordinary differential equations @xmath69 @xmath70 where @xmath71 is the radial component of the lagrangian displacement related to the eulerian perturbation of the velocity field @xmath72 through @xmath73 and @xmath74 is the total pressure perturbation @xmath75 with @xmath76 and @xmath77 being respectively the density and magnetic field perturbations . in equations ( [ eq : lin_system1 ] ) and ( [ eq : lin_system2 ] ) , @xmath78 , @xmath79 , @xmath80 and @xmath81
depend on the equilibrium quantities and on @xmath82 , @xmath83 , @xmath84 and are defined as @xmath85[\tilde{\omega}^2(b_0 ^ 2+\rho_0c_s^2)-c_s^2k_b^2 ] , \end{split}\ ] ] @xmath86 - \frac{2m}{r^2}(k_bb_{0\varphi}+\rho_0v_{0\varphi}\tilde{\omega})[\tilde{\omega}^2(b_0 ^ 2+\rho_0c_s^2)-c_s^2k_b^2 ] , \end{split}\ ] ] @xmath87,\ ] ] @xmath88 -\frac{4[\tilde{\omega}^2(\rho_0c_s^2+b_0 ^ 2)-c_s^2k_b^2](k_bb_{0\varphi } + \rho_0v_{0\varphi}\tilde{\omega})^2}{r^2 } \\
+ \frac{\rho_0[\tilde{\omega}^2(b_{0\varphi}^2-\rho_0v_{0\varphi}^2 ) + ( \tilde{\omega}b_{0\varphi}+k_bv_{0\varphi})^2]^2}{r^2 } , \end{split}\ ] ] where quantities with @xmath89 subscript refer to the equilibrium state and @xmath90 we note that the system ( [ eq : lin_system1 ] ) and ( [ eq : lin_system2 ] ) was derived by @xcite and we kept it in its general form even though in the following we will consider only the case with @xmath91 and @xmath92 .
this system , supplemented with appropriate boundary conditions at @xmath93 and @xmath94 , poses an eigenvalue problem for @xmath82 . on the axis , at @xmath93 , the equations are singular but the solutions have to be regular , while at infinity the solutions have to decay and no incoming wave is allowed ( sommerfeld condition ) .
this asymptotic behaviour of the solutions for small and large radii are used in the numerical integration of the eigenvalue problem . for finding eigenvalues
we use a shooting method with a complex secant root finder , as we did in paper i. the numerical integration can not start at @xmath93 because of the singularity , so we start at a small distance from the origin where the solution is obtained through a series expansion of the equations described in the appendix [ ap : small_r ] .
similarly , we start a backward integration from a sufficiently large radius , where the asymptotic solution is obtained as described in the appendix [ ap : large_r ] and then we match the two numerical solutions at an intermediate radius . equations ( [ eq : lin_system1 ] ) and ( [ eq : lin_system2 ] ) have singularities whenever @xmath95 which give rise to four distinct continua , two alfvn continua for @xmath96 and two slow continua for @xmath97 in the case of zero pressure and zero longitudinal velocity , that we consider in this paper , the slow continua reduce to the single flow continuum defined by the condition @xmath98 since we are interested only in unstable solutions , our integration path will always avoid the singularities , however the presence and position of the continua is fundamental in shaping the overall mhd spectrum @xcite .
to classify the unstable modes present in the jet and understand their physical origin , following @xcite , we employ the wkbj approach and energetic considerations following the frieman - rotenberg formalism @xcite .
the combination of these methods allows us to gain insight into the nature of dominant driving forces inducing the instability of each mode and classify them accordingly .
equations ( [ eq : lin_system1 ] ) and ( [ eq : lin_system2 ] ) can be combined in a single second - order differential equation only for @xmath99 , @xmath100(r\xi_{1r})=0 . \end{split}\ ] ] assuming the radial wavelength of perturbations small compared to the length scale over which there are significant variations in the equilibrium quantities , we can represent the radial dependence of the displacement as @xmath101 , where the radial wavenumber @xmath102 is assumed to be large , @xmath103 . substituting this into equation ( [ eq : secondorder ] ) and neglecting the radial variations of the equilibrium quantities ,
we obtain , to leading order in the large parameter @xmath104 , the following local dispersion relation ( see e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) @xmath105 which after using the expressions for @xmath106 and @xmath78 reduces to a sixth - degree polynomial @xmath107 - 4\rho_0\omega \left ( \frac{m}{r } \rho_0(r\omega^2 ) + 2\omega_b kb_{0z } \right ) \tilde{\omega}^3 \\
+ \tilde{\omega}^2 \bigg [ k_t^2 k_b^2b_0 ^ 2 + \left(k^2+\frac{m^2}{r^2}\right ) \big ( 4\rho_0\omega^2b_{0\varphi}^2 -\rho_0 ^ 2(r\omega^2)^2 + b_0 ^ 2r \frac{d}{dr } \left ( \rho_0\omega^2 -\omega_b^2 \right ) + 4 \rho_0 b_{0\varphi } \omega_b ( r\omega^2 ) \big ) \\ - 4\frac{m}{r } \rho_0 k_b \omega_b ( r\omega^2 ) - 4 b_0 ^ 2 k^2\omega_b^2 \bigg ] + 4\rho_0b_{0\varphi}k_b\omega(r\omega^2)\left(k^2+\frac{m^2}{r^2}\right)\tilde{\omega } + \rho_0(r\omega^2)^2k_b^2\left(k^2+\frac{m^2}{r^2 } \right)=0 , \end{split}\ ] ] where @xmath108 describes the effect of the curvature of toroidal field lines in cylindrical geometry and @xmath109 is the total wavenumber squared .
we intentionally separated out the product @xmath110 , which represents the centrifugal force per unit volume .
this general dispersion relation describes , in principle , all the modes in the local approximation , however , the coefficients are complicated and not physically revealing .
so , we examine various limiting cases to identify these modes .
we start by considering the limits for large and small wavenumbers . at large @xmath83 ,
when @xmath111 and @xmath112 , to leading order from ( [ eq : fullwkb ] ) we get @xmath113 where @xmath114 is the local alfvn velocity .
this dispersion relation resembles that of the parker instability with the driving role of external gravity replaced here by the centrifugal force per unit mass @xmath115 ( see e.g. , * ? ? ?
thus , the unstable mode at large @xmath83 , is driven mainly by the centrifugal force and can be identified with a magnetic buoyancy mode @xcite , which in this limit of large vertical wavenumber operates by bending mostly poloidal field lines .
the growth rate , @xmath116 , for @xmath117 tends asymptoticallly to its maximum value @xmath118 . at small @xmath119 and for @xmath120
, from equation ( [ eq : fullwkb ] ) we get the following dispersion relation @xmath121 where @xmath122\ ] ] and @xmath123 is the epicyclic frequency , @xmath124 .
the solution is given by @xmath125 it is clear that if @xmath126 is positive , which , as we will show below , corresponds to the jet flow being stable against the cold magnetorotational instability ( mri ) , only the last term of equation ( [ eq : tbm ] ) , proportional to the square of the centrifugal acceleration , guarantees the existence of instability . as a result , also in this limit , we have again the centrifugal buoyancy mode , but now , at small @xmath83 , which mainly operates by bending toroidal field lines .
in fact , the dispersion relation ( [ eq : tbm ] ) , in the @xmath127-dominated regime , is similar to that of the non - axisymmetric toroidal buoyancy mode derived in @xcite ( see their equation 51 ) .
note that this approximation can work only for non - axsymmetric modes , since it is based on the condition @xmath128 , that can not be satisfied for @xmath129 . on the other hand , in the axisymmetric case
, it is not possible to bend toroidal field lines .
the instability condition for this mode can be derived from equation ( [ eq : tbmsol ] ) as @xmath130 in principle , the cold differentially rotating jet can also support the mri arising from the combined effect of differential rotation and magnetic fields . to capture this instability , in equation ( [ eq : fullwkb ] )
we ignore centrifugal @xmath131 and curvature @xmath132 terms , which are not its main driving factors , but retain rotation @xmath28
( i.e. , coriolis force ) and shear @xmath133 , which together with azimuthal and vertical magnetic fields cause this instability . as a result
, we obtain a more compact dispersion relation describing the mri in cold differentially rotating cylindrical flows ( see also * ? ? ?
* ; * ? ? ?
* ) @xmath134 \\
+ \left(k_r^2+k^2+\frac{m^2}{r^2}\right ) k_b^2b_0 ^ 2 + \left(k^2+\frac{m^2}{r^2}\right ) \left(4 \rho_0\ omega^2 b_{0\varphi}^2 + b_0 ^ 2 r \frac{d}{dr } ( \rho_0 \omega^2 )
\right ) = 0 \end{split}\ ] ] this expression is a quadratic polynomial for @xmath135 from which a condition for the cold mri can be readily deduced .
that is , an unstable solution @xmath136 can exist whenever @xmath137 the most favourable condition for the instability is when @xmath138 and from this we can derive the necessary condition @xmath139 .
hence , in the cold plasma limit ( @xmath140 ) , the mri vanishes in the case of a purely toroidal field and the presence of a nonzero poloidal / vertical field component is necessary for its operation . in fig .
[ fig : figwkbj ] we show a comparison between representative full numerical solutions to equation ( [ eq : fullwkb ] ) , and the analytical approximations given by equations ( [ eq : pbm ] ) , ( [ eq : tbmsol ] ) and ( [ eq : mri ] ) .
the left panel is for the axisymmetric mode with @xmath141 and @xmath142 . in this case
we have only the poloidal buoyancy mode , the numerical solution is represented by the red curve , while the analytical approximation , given by equation ( [ eq : pbm ] ) , is represented by the green curve and the two curves are indistinguishable .
the mid and right panels are for @xmath143 , the same value of @xmath50 and two different values of the rotation rate , @xmath144 for the mid panel and @xmath145 for the right panel .
again , the numerical solution is represented by the red curve and , in this case , we have the analytical approximations for the poloidal buoyancy mode ( equation [ eq : pbm ] ) , at large wavenumbers , represented by the green curve , and for the toroidal buoyancy mode ( equation [ eq : tbm ] ) at small wavenumbers , represented by the blue curve .
we can furthermore notice that around the value of @xmath83 for which @xmath138 , we have a narrow peak in the growth rate , which corresponds to a very localized mri .
the peak is very narrow for @xmath142 and widens with increasing the rotation rate ( right panel with @xmath146 .
the black curve , representing the solution to equation ( [ eq : mri ] ) , reproduces the behavior of the numerical solution , although the agreement is not as good as for the other two approximations due to the terms neglected in the derivation of equation ( [ eq : mri]).the solutions are taken at a particular radial position , @xmath147 , however , taking different radial positions , the qualitative behavior of the solution remains the same . alternatively , the classification of modes and related instabilities performed above using the wkbj approach can also be made based on energetic considerations that can be derived following the frieman - rotenberg formalism @xcite .
the equation of motion for the lagrangian displacement @xmath148 is @xmath149=0,\ ] ] where the generalized force operator @xmath150 is given by @xmath151 = \rho_0\left(\frac{m}{r}v_{0\varphi}+kv_{0z } \right)^2{\ensuremath{\boldsymbol{\xi}}}_1 - 2i\rho_0\omega\left(\frac{m}{r}v_{0\varphi}+kv_{0z}\right)\left(\xi_{1\varphi}{\hat{\bf r}}-\xi_{1r}\hat{{\ensuremath{\boldsymbol { \varphi}}}}\right)-\\ -2\rho_0r\omega\frac{d\omega}{dr}\xi_{1r}\hat{\bf r}+\rho_1(r\omega^2)\hat{\bf r}+{{\ensuremath{\boldsymbol{j}}}}_0\times { { \ensuremath{\boldsymbol{b}}}}_1+(\nabla\times { { \ensuremath{\boldsymbol{b}}}}_1)\times { { \ensuremath{\boldsymbol{b}}}}_0,\end{gathered}\ ] ] and the @xmath89 subscript indicates again the equilibrium quantities , while the @xmath152 subscript indicates perturbations . substituting @xmath153 in equation ( [ eq : force_op ] ) , we get @xmath154=0.\ ] ] the various terms entering the expression of @xmath150 correspond to different forces acting on the perturbations in the jet flow .
the first term comes from a convective derivative and describes the advection of perturbations by the mean flow . in unmagnetized flows
, this term contributes to the kh instability .
the second term is related to coriolis force due to rotation , the third term is related to shear , or differential rotation of the flow , since it is proportional to the radial derivative of the angular velocity @xmath28 , the fourth term proportional to @xmath76 corresponds to the centrifugal force ( radial buoyancy ) , the fifth and sixth terms are the linearized lorentz force , respectively , due to the equilibrium current @xmath155 and the perturbed magnetic field @xmath156 and due to the perturbed current @xmath157 and the equilibrium magnetic field @xmath158 . one can show that the force operator @xmath150 is self - adjoint @xmath159 d^3 { \ensuremath{\boldsymbol{r}}}= \int { \ensuremath{\boldsymbol{\xi } } } _ 1\cdot { \ensuremath{\boldsymbol{g } } } \left [ { \ensuremath{\boldsymbol{\eta } } } \right ] d^3 { \ensuremath{\boldsymbol{r}}}\ ] ] while the second term in equation ( [ eq : var_disprel ] ) is antisymmetric @xmath160 where @xmath161 is an arbitrary function and integration is performed over an entire fluid volume provided that displacement @xmath148 and @xmath161 vanish at the flow boundaries .
if we take @xmath162 , we can write @xmath163 d^3 { \ensuremath{\boldsymbol{r}}}= \int { \ensuremath{\boldsymbol{\xi}}}_1 \cdot { \ensuremath{\boldsymbol{g } } } \left [ { \ensuremath{\boldsymbol{\xi}}}^{\ast}_1\right ] d^3 { \ensuremath{\boldsymbol{r}}}\ ] ] and @xmath164 therefore @xmath165 d^3 { \ensuremath{\boldsymbol{r}}}$ ] is a real quantity and @xmath166 purely imaginary .
we will see below that these properties are necessary for establishing stability criteria for the flow .
multiplying equation ( [ eq : var_disprel ] ) by @xmath167 , integrating by @xmath168 over the interval @xmath169 $ ] and taking into account that the perturbations vanish for @xmath170 and are regular at @xmath93 , we get @xmath171 where the coefficients @xmath172 and @xmath173 are @xmath174 r d r \nonumber \\ f & = & \int^{\infty}_0 { \ensuremath{\boldsymbol{\xi}}}^{\ast}_1 \cdot { \ensuremath{\boldsymbol{g } } } \left [ { \ensuremath{\boldsymbol{\xi}}}_1 \right ] r d r. \end{aligned}\ ] ] @xmath175 and @xmath176 are real by definition , while @xmath173 is real due to the self - adjointness of the force operator @xmath177 . using the expression of @xmath150 in equation ( [ eq : var1 ] ) ,
we can write @xmath173 in a symmetric form with respect to @xmath148 and @xmath167 : @xmath178 r dr , \end{gathered}\ ] ] where the various terms are grouped according to driving forces they correspond to , as in @xmath179 .
the solution to the quadratic equation ( [ eig3 ] ) is @xmath180 of course this is a formal solution , since the terms @xmath175 , @xmath176 and @xmath173 depend on the eigenfunctions , so they can be computed only after the eigenvalue problem has been solved . if @xmath181 for an eigenmode , this solution comes in complex conjugate pairs that indicates instability of the mode .
therefore , in the expression ( [ eq : force_op_sym ] ) for @xmath173 , negative terms are stabilizing and positive ones destabilizing .
we distinguish four distinct destabilizing contributions : 1 .
the sum of the first two terms @xmath182r dr\ ] ] describe the combined effect of advection by the mean flow and coriolis force .
however , these processes also define @xmath176 and only the sign of @xmath183 actually characterizes stabilizing or destabilizing contribution due to these two effects .
this term is responsible for the velocity shear , or kh instability .
2 . the third term @xmath184 describes the effect of shear , or differential rotation and is destabilazing when @xmath185 , i.e. , @xmath186 somewhere in the flow field .
@xmath187 , together with the combined effect of advection and coriolis force characterized by @xmath188 , determines instability in shear flows . in magnetized shear flows ,
the condition @xmath186 is necessary for the existence of the mri @xcite .
the fourth term @xmath189 which is proportional to the centrifugal acceleration , describes the effect of centrifugal force . if @xmath190 , the centrifugal force can give rise to the magnetic buoyancy instability .
this term depends on the density perturbation , which is expressed via @xmath71 and @xmath74 as @xmath191\ ] ] at large @xmath83 , to leading order the density perturbation becomes @xmath192 indicating that it is produced mainly by bending poloidal field lines .
by contrast , at small @xmath83 , @xmath193 and the density perturbation is determined primarily by bending the toroidal field ( especially at small pitch , when the growth rates are higher ) .
this implies that the density perturbation in this regime arises due to bending of toroidal field lines .
the fifth term @xmath194 is proportional to the equilibrium current and corresponds to the lorentz force .
if @xmath195 , this term is destabilizing , giving rise to current driven instability .
the last term is the magnetic tension force , which is always stabilizing .
based on the above analysis , we classify unstable modes according to which of these four contributions prevails over the net effect of other three ones and results in the destabilization of a given mode .
we then label the mode according to the type of this dominant destabilizing term .
so , for example , if @xmath187 is positive and dominates over the net contribution from all the other terms in the square root in equation ( [ eq : omg_var ] ) , this implies that the instability is caused by differential rotation , which in the case of the considered jet flow threaded by the magnetic field in fact corresponds to the `` cold '' version of mri .
if @xmath196 is positive and dominates , the main destabilizing force is the centrifugal force , which via bending magnetic field lines , gives rise to the magnetic buoyancy instability .
finally , if @xmath197 term is positive and dominates , the destabilization comes from the lorentz force due to the presence of the equilibrium current and hence the resulting instability is current driven .
as discussed in section [ sec : equilibrium ] , the basic equilibrium depends on the two parameters @xmath49 and @xmath50 , wich are defined in equation ( [ eq : param ] ) and represent respectively nondimensional measures of the rotation rate and of the pitch on the vertical axis . instead of @xmath49 , which measures the rotation rate in terms of the average total alfv ' en velocity
, we can alternatively make use of the parameter @xmath37 , defined in equation ( [ eq : alfa ] ) , which measures rotation in terms of the alfvn velocity associated only with the azimuthal magnetic field component .
reference to parameter @xmath37 can be convenient because its value can be related to the sign of the radial gradient of @xmath30 , i.e. for @xmath38 , @xmath30 decreases outward , for @xmath42 , @xmath30 is constant and , for @xmath39 , @xmath30 increases outward . in the following
we will focus our discussion on a number of equilibrium solutions , whose position in the @xmath198 plane is shown in fig .
[ fig : figcases ] .
the green curve corresponds to solutions with @xmath42 , the red dots represents equilibria with @xmath199 , the blue dots are for @xmath42 and , finally , the black dots are for @xmath200 .
we choose these three values of @xmath37 in order to sample the solutions with different gradients of @xmath30 .
for which we computed the behavior of unstable modes .
the red dots represents equilibria with @xmath199 , the blue dots are for @xmath42 and the black dots are for @xmath200 .
the green curve corresponds to solutions with @xmath42.,width=566 ] we start our discussion with the axisymmetric modes , in this case we know that the cdi mode is stable and instabilities can be only due to rotation . in fig . [
fig : complexplanem0 ] we plot in the complex plane the position of unstable modes for a given parameter set @xmath141 , @xmath202 ( @xmath203 ) and @xmath204 .
we observe a sequence of modes clustering to @xmath205 , that is the point where the slow continuum collapses in the present conditions ( @xmath206 , @xmath91 and @xmath129 ) .
the modes in the sequence differ by the number of radial oscillations which increases as the sequence approaches @xmath207 .
, @xmath141 ( corresponding to @xmath208 ) , @xmath129 and @xmath204 .
, width=566 ] we can then investigate the physical origin of this sequence of modes by comparing , in fig .
[ fig : centrifugalm0 ] , the growth rate of the most unstable one as a function of @xmath49 ( solid curve ; the value of @xmath209 is again 1.5 ) with an approximation obtained by equation ( [ eq : omg_var ] ) in which we consider only the centrifugal term ( dashed curve ) , i.e. we approximate the growth rate by @xmath210 where @xmath196 and @xmath175 are given respectively by equations ( [ eq : fc ] ) and ( [ eq : var1 ] ) and can be computed once we have solved the eigenvalue problem and found the eigenfunctions .
we can see that the approximation reproduces very well the behavior of the actual growth rate , so we can regard the centrifugal term as being responsible for the destabilization of these modes and hence identify them as magnetic buoyancy instabilities .
we already discussed these instabilities in section [ modeclassification ] , when we considered the wkbj local dispersion relation .
they have been also already studied by @xcite and @xcite and , as we mentioned above , are analogous to the parker instability , with the centrifugal force replacing gravity and operate by bending the poloidal field lines . for the most unstable mode of the axisymmetric centrifugal buoyancy mode shown in fig .
[ fig : complexplanem0 ] .
the dashed curve represents an approximation to the growth rate given by equation ( [ eq : cengr ] ) .
, width=566 ] from the local dispersion relation ( [ eq : pbm ] ) appropriate for the poloidal buoyancy mode , we expect their growth rates to be proportional to the square of the rotation rate . in fig .
[ fig : m0 ] we then show their growth rates divided by @xmath211 as a function of the wavenumber .
the three panels are for three different values of @xmath37 and the different curves in each panel are for different values of @xmath50 . as already discussed , the centrifugal buoyancy modes represent actually a sequence of unstable modes and , in the panels , we show only those branches of this mode with the maximum growth rate .
this figure demonstrates that the @xmath211 scaling law is quite good and that , as expected from the local dispersion relation , the growth rate increases with the vertical wavenumber and tends to an asymptotic limit as @xmath212 , in fact the behaviour of the growth rate as a function of the wavenumber is the same as in fig .
[ fig : figwkbj ] .
these modes appear to be always unstable , the reason , discussed by @xcite , is related to the fact that , since the plasma has no pressure , it is possible to compress it along the field lines and create a density perturbation without performing any work .
the centrifugal force can then always overcome the magnetic restoring forces .
however , the inclusion of a finite pressure tends to stabilize this mode . , the left panel is for @xmath199 , the mid panel is for @xmath203 and the right panel is for @xmath213 .
the different curves refer to different values of @xmath50 and the corresponding values of @xmath50 and @xmath49 are given in the legend.,width=566 ] we start our analysis of non - axisymmetric instabilities by considering first the cases with @xmath42 and @xmath215 and , in order to get a first indication on the number and the kind of modes that we can find , in fig . [
fig : complexplanem1 ] we plot in the complex plane the position of unstable modes for a given parameter set .
the figure is for @xmath141 and we consider two values of the wavenumber : squares are for @xmath216 and dots are for @xmath217 .
both at large and small wavenumbers , we observe an isolated mode and a series . in fig .
[ fig : grvsalfa ] we consider the behavior of the modes represented in fig .
[ fig : complexplanem1 ] as a function of the rotation parameter @xmath49 , the colors of the curves are in correspondence with the colors in fig .
[ fig : complexplanem1 ] , and , for the series , we have considered only the mode with the largest growth rate . from the figure
we can see that the only mode that survives when we let rotation go to zero is the one corresponding to the green dot in fig .
[ fig : complexplanem1 ] , all the others become stable .
we can then conclude that the mode corresponding to the green dot reduces to the cdi mode in the zero rotation limit , while rotation is at the origin of all the other modes . in this figure , we can further notice a stabilizing effect of rotation on the cdi , however , this is noticeable only at large values of @xmath49 .
this can be compared with the results of @xcite who find also a stabilizing effect of rotation , but at smaller values of the rotation rate , in their case however there is a rigid rotation , while in this case the rotation rate decreases radially . ,
@xmath141 , @xmath143 , the ( green and red ) dots are for @xmath218 , while the ( black and blue ) squares are for @xmath216 .
the orange line shows the frequency range of the slow continuum , which , in the zero pressure case , reduces to the flow continuum .
as it is discussed in the text , the green dot represents the cdi , the black dot mri , while the red and blue series represent centrifugal modes .
, width=566 ] for the modes represented in fig .
[ fig : complexplanem1 ] .
the colors of the curves correspond to the color in fig .
[ fig : complexplanem1 ] .
, width=566 ] we can further investigate the physical origin of the different modes by computing the stabilizing and destabilizing terms based on the energetic considerations discussed in the previous section . starting from the cdi mode , we find that the destabilizing terms for this mode is not only @xmath197 , as it should be expected for the current driven mode , but also @xmath187 . in fig .
[ fig : cdvar ] we plot the fractional contributions of these two terms as a function of the rotation rate @xmath49 .
the fractional contributions for the two terms are defined , respectively , as @xmath219 and @xmath220 we see that at low rotation rates , the dominant term is the current term represented by the green curve , but , as we increase the rotation rate , the contribution by the shear term increases until it becomes dominant for @xmath221 .
( green ) , and of the shear term , @xmath222 ( black ) , as a function of the rotation rate @xmath49 for the cdi mode .
the values of other parameters are the same as in fig .
[ fig : complexplanem1 ] .
, width=566 ] as we did for the axisymmetric modes , we can investigate the physical origin of the two series of modes , represented by blue squares and red dots in fig .
[ fig : complexplanem1 ] , by comparing in fig .
[ fig : centrifugal ] the growth rate as a function of @xmath49 ( solid curves ) with the approximation given by equation ( [ eq : cengr ] ) ( dashed curves ) .
we see that the approximation reproduces fairly well the behavior of the actual growth rate , so we can then regard the centrifugal term as being responsible for the destabilization of these modes and hence identify them as magnetic buoyancy instabilities . as discussed in the previous section and in section
[ sec : wkb ] , with the wkbj analysis , we can distinguish them as a toroidal buoyancy mode at low wavenumbers ( red curve ) and a poloidal buoyancy mode at high wavenumbers ( blue curve ) .
the two sequences of modes cluster to the edge of the flow continuum ( which is what is left of the slow continua in the zero pressure case ) , whose frequency range is represented in fig .
[ fig : complexplanem1 ] by the orange line .
consider now the mode represented by the black square in fig .
[ fig : complexplanem1 ] . the black curve in fig .
[ fig : centrifugal ] traces this mode as @xmath49 varies .
similarly to the buoyancy modes , we can aproximate its growth rate by an expression analogous to equation ( [ eq : cengr ] ) in which @xmath196 is replaced by @xmath187 , @xmath223 the dashed black curve in fig .
[ fig : centrifugal ] shows the growth rate given by equation ( [ eq : growthmri ] ) , which indeed closely follows an actual one represented by the black curve .
so , the main driving force for this mode is related to the shear of the radially decreasing rotation rate and therefore it should be identified with the mri .
however , this occurs only in a very limited parameter range , while in other regions it merges with either the buoyancy modes or with the cdi mode , where the driving force become either the centrifugal term or the current term , respectively ( see below ) .
so , from now on we mostly concentrate on the cdi and centrifugal buoyancy modes . for the centrifugal buoyancy modes ( blue and red ) and the mri branch ( black ) .
the blue and red dashed curves represent an aproximation to the growth rate given by equation ( [ eq : cengr ] ) for the buoyancy modes , whereas the black dashed curve represents an approximation to the growth rate given by equation ( [ eq : growthmri ] ) for the mri .
the values of other parameters are the same as in fig .
[ fig : complexplanem1].,width=566 ] we can now proceed with a more detailed analysis of the dependence of the growth rates on the wavenumber , the pitch and rotation . in fig .
[ fig : gr1 ] , we plot the growth rates as a function of the wavenumber for @xmath224 ( left panel ) and for @xmath225 ( right panel ) for @xmath203 and @xmath143 . in the left panel , we have clearly distinct the cdi and the toroidal and poloidal buoyancy modes . for both values of the pitch ,
the black part of the curve , which corresponds to the mri , is distinct only in the growing part over a relatively narrow range of wavenumbers and is merged with one of the poloidal buoyancy modes in the constant region . as expected and discussed in paper i , the cdi mode ( green curve ) increases its growth rate and the value of its maximum unstable wavenumber as we decrease the pitch . for @xmath226 ,
the increase of the maximum unstable wavenumber brings the cdi mode to a complicated interaction with the other modes , merging first with the mri branch and then with one of the poloidal buoyancy modes .
for @xmath227 the cdi mode becomes unstable for all wavenumbers , while the driving force , increasing the wavenumber , changes nature , becoming first related to the shear of rotation and then centrifugal .
both the toroidal and the poloidal buoyancy modes increase their growth rate as we decrease the pitch , this is because in the equilibrium configuration the rotation rate increases as the pitch decreases .
the toroidal buoyancy mode becomes stable at high wavenumbers , while the poloidal buoyancy mode becomes stable at small wavenumbers .
decreasing @xmath50 , the stable region between them shows a small increase in width and moves towards high wavenumbers . .
the left panel is for @xmath228 while the right panel is for @xmath141 .
the other parameters are @xmath42 and @xmath143 . , width=566 ] for discussing in more detail the behavior of the cdi , it can be useful to examine how the properties of the equilibrium structure are modified in the different parameter ranges .
figures [ fig : pitch ] and [ fig : jpar ] show the radial profiles of the pitch and the equilibrium current component parallel to the magnetic field , @xmath229 , which is the destabilizing factor for the cdi .
in fact , by rearranging expression ( [ eq : fcd ] ) for the current term @xmath197 , one can show that the destabilizing term is due to a contribution proportional to the current component parallel to the magnetic field ( e.g. , see * ? ? ?
* ch . 8) .
therefore , the latter is a central quantity determining the growth rate of the cdi .
the three panels are respectively for @xmath199 ( left panel ) , @xmath42 ( mid panel ) and @xmath200 ( right panel ) and the different curves in each panel refer to different values of @xmath50 , in each panel we also plotted the case with @xmath224 and no rotation for reference ( black curves ) .
we remember that a variation of the pitch leads also to a variation of the rotation rate : for lower values of the pitch we have higher values of @xmath49 and these values can be read in the legend .
the pitch profile is normalized to the value @xmath50 , on the axis , while the parallel current is multiplied by @xmath50 to bring curves for different values of @xmath50 on the same scale , since for large values of @xmath50 the parallel current scales as @xmath230 .
the pitch profile is in general characterized by a flat part up to @xmath231 followed by a steep increase . in the left panel
, we see that a decrease of @xmath50 leads to a slower increase of @xmath232 for @xmath233 .
correspondingly , in the left panel of fig . [ fig : jpar ] , we observe a slight increase of the parallel current . in the case of @xmath42 ( mid panel ) , when we have @xmath30 constant , the pitch profiles remain essentially unchanged when we decrease @xmath50 , while the parallel current shows a substantial decrease for low values of @xmath50 . for @xmath200 ( right panel ) ,
@xmath30 increases with radius and consequently the pitch show increasingly steeper profiles as we decrease @xmath50 and the decrease of the parallel current is larger than in the previous case . .
the three panels refer to three different values of @xmath37 , the left panel is for @xmath199 , the mid panel is for @xmath203 and the right panel is for @xmath213 .
the different curves refer to different values of @xmath50 and the corresponding values of @xmath50 and @xmath49 are given in the legend .
, width=566 ] .
the three panels refer to three different values of @xmath37 , the left panel is for @xmath199 , the mid panel is for @xmath203 and the right panel is for @xmath213 .
the different curves refer to different values of @xmath50 and the corresponding values of @xmath50 and @xmath49 are given in the legend . , width=566 ] in paper i we discussed a scaling law for the growth rate of the cdi of the form @xmath234 and in fig .
[ fig : cd ] we can investigate the effect of rotation by considering the deviation from this scaling law .
the three panels and the curves in each panel correspond to the same cases shown in figs .
[ fig : pitch ] and [ fig : jpar ] . in the left panel
we can observe that a first effect of rotation is to move the cutoff wavenumber to smaller values , from @xmath235 without rotation ( paper i ) to @xmath236 .
apart from that , the scaling provided by eq .
( [ eq : scaling ] ) is quite good , slight deviations can be observed only for the smallest value of @xmath50 ( blue curve ) , partly due to the interaction with other modes ( mri ) and partly ( as already discussed in paper i ) related to the change of the pitch profile and parallel current observed in the corresponding equilibrium solution . comparing the red curves ( largest values of @xmath50 ) in the three panels we see that the increase of the rotation rate leads the cutoff wavenumber to shift towards increasingly lower values , in parallel , however , we have also a slight increase of the growth rate in the unstable range . from the other curves ( green , blue , orange and purple ) , we see that , for lower values of @xmath50 , at @xmath42 , the cutoff disappears because the cdi mode starts to interact and merge with the centrifugal mode and the growth rate decreases as a result of the decrease of the parallel current in the equilibrium configuration . for @xmath200
, we also observe a decrease of the growth rate for increasingly lower values of @xmath50 , corresponding to the decrease of the parallel current . in summary
, rotation has , in general , a stabilizing effect on the cdi mainly because it modifies the equilibrium structure by decreasing the parallel current .
this is consistent with @xcite , who also found stabilization of cdi at high rotation rates in the case of rigid rotation .
however , especially at large values of @xmath50 , there are situations in which , on the contrary , the growth rate of cdi shows a slight increase with the rotation rate .
finally , we recall that the cdi is stable for @xmath237 . for @xmath143 .
the three panels refer to three different values of @xmath37 , the left panel is for @xmath199 , the mid panel is for @xmath203 and the right panel is for @xmath213 .
the different curves refer to different values of @xmath50 and the corresponding values of @xmath50 and @xmath49 are given in the legend .
, width=566 ] we can now turn our attention to the toroidal and poloidal buoyancy modes . from the local dispersion relations , given by equations ( [ eq : pbm ] ) and ( [ eq : tbm ] ) , we expect their growth rates to be proportional to the square of the rotation rate . in figs .
[ fig : tb ] and [ fig : pb ] , we then show their growth rates divided by @xmath211 as a function of the wavenumber .
the three panels are , as before , for three different values of @xmath37 and the different curves in each panel are for different values of @xmath50 . as already discussed , the toroidal and poloidal buoyancy modes represent actually a series of unstable modes and , in these figures , we show only the modes with the maximum growth rate .
the @xmath211 scaling law is quite good for the toroidal buoyancy mode , slightly less valid in the case of the poloidal buoyancy mode . in general , the value of @xmath238 represents a high wavenumber cutoff for the toroidal buoyancy mode and a low wavenumber cutoff for the poloidal buoyancy mode , however , for large values of the rotation rate , as shown in the right panels of figs .
[ fig : tb ] and [ fig : pb ] , we observe merging between modes of the two series and deviations from the @xmath238 cutoff ( note that the merging of poloidal and toroidal modes at @xmath213 shown in the right panels of figs . [
fig : tb ] and [ fig : pb ] , may refer to different mode branches in the two series ) . in fig .
[ fig : buoym-1 ] we show the results for the case with @xmath237 .
overall , the buoyancy modes behave similarly with the wavenumber , although there are some differencies with the @xmath143 case at intermediate @xmath83 .
specifically , we see that the buoyancy mode is unstable at all wavenumbers , in fact the narrow stability region around @xmath239 disappears .
the growth rate , which is independent of the wavenumber at small values of the latter , shows however a variation around @xmath238 , sometimes a decrease but typically an increase going towards larger values of the wavenumber .
of course , this increase of the growth rate eventually asymptotes to a finite value as @xmath117 , as in the @xmath143 case above , but at small @xmath50 , this asymptotic value is reached at higher wavenumbers .
so far we concentrated on the values @xmath240 of the azimuthal wavenumber and now we examine different values .
figure [ fig : diffm ] shows the toroidal ( red curves ) and poloidal ( blue curves ) buoyancy modes as well as mri ( black curves ) at different @xmath84 and fixed @xmath225 and @xmath203 .
first consider the case of positive @xmath84 , when the toroidal and poloidal modes are separated .
the growth rate of the toroidal mode increases with @xmath84 and the instability boundary extends to larger @xmath83 .
however , it is clear from this figure and also from the local dispersion relation ( [ eq : tbmsol ] ) that the growth rate at small @xmath83 converges to a finite value as @xmath84 becomes large .
by contrast , the growth rate of the poloidal mode decreases with @xmath84 and the instability boundary shifts to larger @xmath83 , but the maximum growth achieved in the limit of high @xmath83 is essentially independent of @xmath84 , as it also follows from the local dispersion relation ( [ eq : pbm ] ) .
the behaviour of these modes in the case of negative @xmath84 , where , as we discussed above , they are represented by a single curve with respect to @xmath83 ( fig .
[ fig : buoym-1 ] ) , are similar to that for positive @xmath84 . at @xmath241 , corresponding to the toridal mode , the growth rate increases with the absolute value of @xmath84 and converges to a constant value at a given @xmath83 , while at @xmath242 , corresponding to the poloidal mode , it decreases with the absolute value of @xmath84 , but tends to the same limiting value .
the mri exists only for @xmath243 and 3 in a certain interval of @xmath83 being most unstable at @xmath243 and @xmath244 .
this maximum growth rate of the mri is higher than that at @xmath143 , is comparable to that of the toroidal and poloidal modes for the same @xmath243 and it does not merge with the latter as opposed to the case @xmath143 ( see fig .
the growth rate of the mri at @xmath245 is decreased , its range in @xmath83 is narrower and shifted to larger values .
the mri disappears beyond this azimuthal wavenumber ; we did not find it for larger positive @xmath246 as well as for all negative @xmath84 . for @xmath143 .
the three panels refer to three different values of @xmath37 , the left panel is for @xmath199 , the mid panel is for @xmath203 and the right panel is for @xmath213 .
the different curves refer to different values of @xmath50 and the corresponding values of @xmath50 and @xmath49 are given in the legend .
, width=566 ] for @xmath143 .
the three panels refer to three different values of @xmath37 , the left panel is for @xmath199 , the mid panel is for @xmath203 and the right panel is for @xmath213 .
the different curves refer to different values of @xmath50 and the corresponding values of @xmath50 and @xmath49 are given in the legend .
, width=566 ] for @xmath237 .
the three panels refer to three different values of @xmath37 , the left panel is for @xmath199 , the mid panel is for @xmath203 and the right panel is for @xmath213 .
the different curves refer to different values of @xmath50 and the corresponding values of @xmath50 and @xmath49 are given in the legend .
, width=566 ] ( solid ) , @xmath247 ( dashed ) , @xmath248 ( dash - dot ) .
the pitch parameter @xmath141 and the rotation @xmath42 . ]
, in the different regions of the parameter plane @xmath198 . in the region marked by a green shading , the instability with the largest growth rate is the cdi , while in the region marked by a blue shading the instability with the largest growth rate is the centrifugal buoyancy instability.,width=566 ] we have examined the stability properties of a rotating magnetized jet flow in the approximation of zero thermal pressure .
our study has focused on the effect of rotation on the cdi and on the new modes of instability introduced by rotation . in this spirit , as a first step , we did not consider the presence of a longitudinal flow , whose main effect on the modes concentrated inside the jet radius , as it is for most of the rotationally - induced modes studied in this paper , is only that of doppler shifting the frequency .
the instability behaviour depends , of course , on the chosen equilibrium configuration and our results can be considered representative of an equilibrium configuration characterized by a distribution of current concentrated in the jet , with the return current assumed to be mainly found at very large distances .
similar stability analyses of rotation - induced modes in magnetized flows in the limit of zero thermal pressure are presented in @xcite and @xcite , they however make use only of a local wkb approach and the last two papers consider only axisymmetric perturbations . our local analysis ( see section [ sec : wkb ] ) generally agrees with the results of these papers in the parameter regimes they consider .
however , for our specific jet configuration we found that the mri in the cold limit can be present , but only in a very limited region of the parameter space and is never dominant .
we extended the results of these papers to the global domain , where the wkb approach no longer holds , by solving a boundary value problem and revealed new properties of these modes that are summarized below : + 1 . at small and large axial wavenumbers @xmath83 , the growth rates , respectively , for the toroidal and poloidal buoyancy modes obtained from the global calculations ( figs .
6 , 11 and 15 - 18 ) actually exhibit a dependence on @xmath83 similar to what is obtained by the corresponding local dispersion relations ( see fig .
+ 2 . at intermediate @xmath83
, the behavior can be different from that predicted on the basis of the local dispersion relation , with the presence of stability gaps ( for positive @xmath84 ) and merging of poloidal and toroidal modes ( for negative @xmath84 ) .
the properties of the mri in the cold plasma limit , studied in the above papers based on the local dispersion relation , qualitatively agrees with our global calculations . in the nonaxisymmetric case , for positive @xmath84
, the mri is present only in a limited range of wavenumbers , its growth rate and the width of the unstable range reach a maximum for @xmath249 and then ( for larger @xmath84 ) decrease ( fig .
18 ) consistently with @xcite , however , the mri is absent for @xmath246 and for every negative value of @xmath84 .
in addition , we did not find the mri for @xmath129 in both local and global cases , likely because of the different equilibrium adopted . in general , the mri has always a growth rate smaller than that of the buoyancy modes .
we have shown that two main kinds of instabilities cdi and buoyancy prevail in the considered jet flow . in fig .
[ fig : summary ] we represent , in the parameter plane @xmath198 , with different shadings , the regions where each of them has the largest growth rate .
the figure refers to non - axisymmetric modes with @xmath250 .
we can observe that the cdi is dominant at small rotation rates and that the boundary between the cdi and centrifugal buoyancy instability regions moves towards larger values of @xmath49 as we decrease @xmath50 .
for @xmath251 , the cdi is stable for this value of the wavenumber and the only instability is the centrifugal buoyancy , which is , however , obviously stable at zero rotation .
it is seen that the buoyancy instability occupies quite a large area in this parameter space in comparison with the cdi and hence should be important in jets with rotation .
when we increase the wavenumber , the cdi tends to be stabilized and the centrifugal instability tends to become dominant everywhere in the parameter plane .
comparing now the growth rates of axisymmetric and non - axisymmetric centrifugal buoyancy modes , we see that at high wavenumbers , axisymmetric modes have a larger growth rate , which decreases monotonically with decreasing @xmath83 .
by contrast , non - axisymmetric modes have a growth rate that is almost independent from the wavenumber and , therefore , become dominant at low values of @xmath83 .
summarizing , at low rotation rates , the non - axisymmetric cdi is the instability that grows fastest and has large wavelengths .
increasing the rotation rate , the prevailing instability becomes the centrifugal axisymmetric one , which operates at small wavelengths .
these results are applicable to magnetically and rotationally dominated jets , since , increasing the importance of thermal pressure , centrifugally driven modes tend to be stabilized and other modes , like pressure driven modes @xcite may appear . at the same time , taking into account the shear of longitudinal velocity can give rise to unstable kh modes in the jet .
this first step will be extended by introducing the effects of the longitudinal velocity also in the relativistic regime and these results will be presented in a following paper .
the different behaviour in the explored parameter space may be important for understanding the nonlinear stages since distinct types of instability may evolve differently .
this study is therefore an essential first step for the interpretation of the results of numerical simulations and for their comparison with astrophysical data .
g.m . acknowledges the georg forster postdoctoral research fellowship from the alexander von humboldt foundation .
m. , stegun i. a. , 1972 , handbook of mathematical functions , new york : dover s. , 1996 , , 314 , 995 s. , camenzind m. , 1992 , , 256 , 354 s. , lery t. , baty h. , 2000 , , 355 , 818 s. a. , hawley j. f. , 1992 , , 400 , 610 h. , keppens r. , 2002 , , 580 , 800 m. c. , 1998 , , 493 , 291 m. , 1991 , the stability of jets .
p. 278
j. w. s. , van der swaluw e. , keppens r. , goedbloed j. p. , 2005
, , 444 , 337 g. , mamatsashvili g. , rossi p. , mignone a. , 2013 , , 434 , 3030 g. , rosner r. , ferrari a. , knobloch e. , 1989 , , 341 , 631 g. , rosner r. , ferrari a. , knobloch e. , 1996 , , 470 , 797 a. , urpin v. , 2006 , , 73 , 066301 a. , urpin v. , 2007 , , 662 , 851 a. , urpin v. , 2008 , , 488 , 1 a. , urpin v. , 2011a , , 84 , 056310 a. , urpin v. , 2011b , , 525 , a100 a. , iacono r. , bhattacharjee a. , 1987 , physics of fluids , 30 , 2167 c. s. , sovinec c. r. , 2009 , , 699 , 362 a. , trussoni e. , zaninetti l. , 1978 , , 64 , 43 j. , 1987 , ideal magnetohydrodynamics , plenum press , new york e. , rotenberg m. , 1960 , reviews of modern physics , 32 , 898 w. , lai d. , 2011 , , 410 , 399 j. p. , 2009
, physics of plasmas , 16 , 122110 goedbloed j. p. , keppens r. , poedts s. , 2010 , advanced magnetohydrodynamics : with applications to laboratory and astrophysical plasmas . cambridge university press , cambridge ; new york m. , sol h. , sauty c. , 2000 , , 316 , 494 p. e. , 1979 , , 234 , 47 p. e. , 2006 , in p. a. hughes & j. n. bregman ed . , relativistic jets : the common physics of agn , microquasars , and gamma - ray bursts vol .
856 of american institute of physics conference series , agn jets : a review of stability and structure .
pp 5777 p. e. , cooper m. a. , norman m. l. , stone j. m. , 1992 , , 399 , 478 y .- m . , hassam a. b. , 2003 , physics of plasmas , 10 , 204 y. n. , pariev v. i. , 1994 , , 267 , 629 y. n. , pariev v. i. , 1996 , , 281 , 1 r. , casse f. , goedbloed j. p. , 2002
, , 569 , l121 e. , longaretti p .- y . , pelletier g. , 2000 , , 363 , 1166 j. , balsara d. s. , lyutikov m. , komissarov s. s. , george d. , siddireddy p. k. , 2015 , , 450 , 982 w .- t . , ostriker e. c. , 2000 , , 540 , 372 y. e. , 1999 , , 308 , 1006 y. , hardee p. , nishikawa k .- i . , 2007 , , 662 , 835 r. , li j. , tchekhovskoy a. , 2009 , , 697 , 1681 m. , hanasz m. , mart j. m. , sol h. , 2004 , , 427 , 415 m. , mart j. m. , cela j. m. , hanasz m. , de la cruz r. , rubio f. , 2010 , , 519 , a41 + m. e. , psaltis d. , 2005 , , 628 , 879 m. , begelman m. c. , madejski g. m. , lasota j .-
p . , 2005 , , 625 , 72 a. , matsuoka t. , takahashi m. , 2001 , , 64 , 123003 v. , 2002 , , 385 , 14 p. , tagger m. , 2002 , , 394 , 329
to find a solution of equations ( [ eq : lin_system1 ] ) and ( [ eq : lin_system2 ] ) at small radii , we calculate the coefficients entering these equations at @xmath253 taking into account that in this limit the equilibrium quantities @xmath254 , whilst @xmath255 and @xmath256 tend to constant values . at @xmath257 ,
these coefficients behave differently depending on whether @xmath129 or @xmath258 , so we should distinguish these two cases . in this subsection , we consider the case @xmath259 and in the next one the case @xmath129 . for @xmath259 ,
we have ( primes everywhere denote radial derivative ) @xmath260 @xmath261,\ ] ] @xmath262\ ] ] @xmath263(k_bb'_{0\varphi}+\varrho_0v'_{0\varphi}\tilde{\omega})^2.\ ] ] substituting these coefficients into equations ( [ eq : lin_system1 ] ) and ( [ eq : lin_system2 ] ) and taking solutions with the form @xmath264 , to leading order , we obtain for the power index @xmath37 , @xmath265 but because a solution must be regular at @xmath93 we choose only @xmath266 , and after that the ratio @xmath267\ ] ] this equation together with the choice @xmath268 serves as our boundary condition at small radii . in the axisymmetric case @xmath129 , @xmath78 and the coefficients @xmath106 in equations ( [ eq : lin_system1 ] ) and ( [ eq : lin_system2 ] ) take the form @xmath269,\ ] ] @xmath270,\ ] ] @xmath271,\ ] ] @xmath272(k_bb'_{0\varphi}+\varrho_0v'_{0\varphi}\tilde{\omega})^2,\ ] ] where now @xmath273 , @xmath274 and the primes , as before , denote the radial derivatives of @xmath275 and @xmath127 at @xmath93 .
we see that all these coefficients are regular and finite as @xmath276 and only @xmath79 is proportional to @xmath168 .
we can express @xmath71 through @xmath74 from equation ( [ eq : lin_system2 ] ) @xmath277 and substitute into equation ( [ eq : lin_system1 ] ) . keeping only dominant terms in the limit @xmath276
, we get a single second order equation only for pressure @xmath278p_1=0.\]][pr_eq ] since @xmath279 , the coefficient in front of @xmath74 in this equation is regular and finite and explicitly calculating it at @xmath93 yields @xmath280=\frac{1}{\delta}\left [ 2\frac{dc_1}{dr}+\frac{c_2c_3}{\delta}\right]=\frac{(\tilde{\omega}^2-c_s^2k^2)(\varrho_0\tilde{\omega}^2-k_b^2)^2}{\delta}+\\+\frac{4b^{'2}_{0\varphi}}{\delta}\left[\tilde{\omega}^2(\varrho_0\tilde{\omega}^2-k_b^2)+c_s^2k^2k_b^2\right ] -\frac{4\varrho_0k_b\tilde{\omega}b'_{0\varphi}v'_{0\varphi}}{\delta}\left(\tilde{\omega}^2 - 2c_s^2k^2\right)-\\-\frac{2\varrho_0\tilde{\omega}^2v^{'2}_{0\varphi}}{\delta}\left ( 3\varrho_0\tilde{\omega}^2-k_b^2 - 2\varrho_0c_s^2k^2\right),\end{gathered}\ ] ] where all the quantities in this expression are calculated at @xmath93 .
@xmath175 is nonzero constant and equation ( [ pr_eq ] ) takes the form of bessel equation of zeroth order @xmath281[bessel ] at @xmath282 , this equation has two linearly independent solutions @xmath283j_0(ra^{1/2}),\ ] ] where @xmath284 and @xmath285 are the 0-th order bessel and neumann functions and @xmath286 is the euler - mascheroni constant @xcite .
from these two solutions , we select the first one which is regular at small @xmath168 : @xmath287 and correspondingly for the displacement @xmath71 , to leading order we have @xmath288 | we perform a linear stability analysis of magnetized rotating cylindrical jet flows in the approximation of zero thermal pressure .
we focus our analysis on the effect of rotation on the current driven mode and on the unstable modes introduced by rotation .
we find that rotation has a stabilizing effect on the current driven mode only for rotation velocities of the order of the alfvn velocity .
rotation introduces also a new unstable centrifugal buoyancy mode and the `` cold '' magnetorotational instability .
the first mode is analogous to the parker instability with the centrifugal force playing the role of effective gravity .
the magnetorotational instability can be present , but only in a very limited region of the parameter space and is never dominant .
the current driven mode is characterized by large wavelenghts and is dominant at small values of the rotational velocity , while the buoyancy mode becomes dominant as rotation is increased and is characterized by small wavelenghts . [ firstpage ] galaxies : jets , mhd , instabilities |
a large azimuthal anisotropy is observed for particle production in heavy ion collisions at the relativistic heavy ion collider ( rhic ) and the large hadron collider ( lhc ) .
this anisotropy has been successfully described in terms of pressure - driven collective flow , which converts the spatial asymmetries of the initially created matter into transverse momentum ( @xmath5 ) space @xcite .
such anisotropy can be expressed as fourier series in azimuthal angle ( @xmath9 ) : @xmath10 where @xmath11 and @xmath12 represent the magnitude and direction of @xmath13-order flow respectively .
asymmetries giving rise to non - zero @xmath11 are associated with either an average shape ( for @xmath14 ) or shapes due to the spatial fluctuations of the participating nucleons @xcite ; the @xmath15 term is known as dipole asymmetry . due to the approximate boost invariance of the created matter , the @xmath11
are expected to be even functions in rapidity ( rapidity - even ) @xcite . however ,
a small rapidity - odd @xmath11 component , especially for @xmath15 , may arise from the `` sideward '' deflection of the created matter @xcite .
this rapidity - odd @xmath3 is expected to be very small for the rapidity range and the collision energies considered here @xcite , and is not considered in this paper .
the harmonics in eq .
[ eq : vn1 ] is also reflected in the two - particle correlation ( 2pc ) in @xmath16 : @xmath17 with @xmath18 recent studies at the lhc show that the latter factorization is valid for @xmath19 at low @xmath5 for particle pairs with a large rapidity gap @xcite , but fails for @xmath15 over the entire @xmath5 range .
this is understood to be due to global momentum conservation for a finite multiplicity system , which may induce a negative dipole component in eq .
[ eq : vnnb ] @xcite : @xmath20 where @xmath21 is the total multiplicity in the event and @xmath22 is the fraction of particles that are correlated via momentum conservation , and hence the parameter @xmath23 has a dimention of gev@xmath24 .
recent studies by the atlas collaboration @xcite and retinskaya _ et.al . _
@xcite shows that @xmath25 can indeed be described by eq .
[ eq : v1a ] over broad ranges of @xmath1 .
many models based on ideal / viscous hydrodynamics or transport theory have been developed to describe the @xmath11 and @xmath26 data @xcite .
they show that sizable higher - oder flow ( @xmath19 ) can be generated by including fluctuations of the initial geometry . in some cases ,
reasonable agreement with the @xmath11 data has been achieved . however , due to the complication associated with the global momentum conservation , most of these model attempts have avoided a prediction for @xmath3 .
recently , gardim _ et.al _
@xcite performed the first calculation of the rapidity - even @xmath3 in a hydrodynamic calculations ; a modified event plane ( ep ) method was also proposed @xcite .
the @xmath27 was predicted to change sign at low @xmath5 , due to the constraint that the total transverse momentum of the system arising from dipolar flow should vanish , i.e. @xmath28 @xcite .
this sign - flip can lead to rather complicated dependence of @xmath25 on @xmath29 and @xmath30 ( see eq .
[ eq : v1b ] ) .
this complicated pattern has been observed in the atlas data , and is expected to also exist in transport models .
furthermore , the systems that are related via momentum conservation could be smaller than the whole event ( for example a dijet system ) @xcite .
therefore , the value of @xmath23 ( eq .
[ eq : v1b ] ) may depend on other kinematic variables such as the relative pseudorapidity ( @xmath31 ) of the particle pairs .
hence , it is advantageous to determine @xmath23 via the fit method used by the atlas collaboration , instead of calculating it directly . in this work
, we present a quantitative study of the rapidity - even @xmath32 using a generalization of the atlas fitting method within the ampt ( a multi phase transport ) model @xcite .
the ampt model combines the initial condition from hijing @xcite and final state interaction via a parton and hadron transport model .
the outputs are complete monte - carlo events that naturally contains various flow and non - flow effects , including the global momentum conservation
. we shall demonstrate that the @xmath25 from ampt has similar features as the experimental data , and can be well described by eq .
[ eq : v1a ] , lending support to the procedure used by atlas .
we then make quantitative predictions on how @xmath3 varies with @xmath5 , @xmath6 , centrality , collision energy and parameters in the partonic transport , as well as to compare with the atlas data .
the initial condition from ampt is seeded by a glauber model for nucleon - nucleon scatterings , hence it contains the fluctuations which generate the event - by - event spatial asymmetries .
the nucleon - nucleon scatterings provide both minijet partons for hard scattering and strings for soft coherent interactions ; these strings are then converted into soft partons via a string melting scheme .
the parton and hadron transport is responsible for transforming the initial asymmetries into the momentum anisotropy .
the parton transport include only elastic scattering whose cross - sections are controlled by the values of the strong coupling constant ( @xmath33 ) and the debye screening mass ( @xmath34 ) @xcite : @xmath35 where @xmath36 is the mandelstam variable for four momentum transfer .
the total cross - section can be increased either with a larger @xmath33 or a smaller value of @xmath34 .
however changing @xmath33 is more effective for momentum dissipation than changing @xmath34 , as decreasing @xmath34 includes only softer scattering ( eq . [ eq : cros ] ) .
partons are recombined into hadrons at the freezeout , followed by a hadronic transport .
ampt events are generated for two different @xmath33 values for parton transport : set - a with @xmath37 and set - b with @xmath38 from @xcite . ] .
we vary the total cross - section from 1.5 mb to 10 mb by adjusting @xmath34 according to @xcite , separately for set - a and set - b .
varying @xmath33 and @xmath34 allows us to study how @xmath3 depends on the strength of final state interactions .
these studies are carried out at rhic energy ( au+au at @xmath39 tev ) and/or lhc energy ( pb+pb at @xmath40 tev ) for the impact parameters @xmath41 fm and @xmath42 fm , corresponding to approximately 0 - 10% and 30 - 40% most central events , respectively .
we follow the procedure of extracting @xmath3 as documented in @xcite .
first , the correlation function for a given @xmath29 , @xmath30 and @xmath31 range is constructed from all final state particles , and @xmath43 , @xmath44 values are calculated .
second , a least - square minimization of these @xmath25 values following eq .
[ eq : v1a ] is used to obtain @xmath45 for a given @xmath31 range , where the @xmath45 function is defined by its values at 9 @xmath5 points ( 0.5,0.7,0.9,1.1,1.5,2.0,3.0,3.8,4.6 gev ) via a cubic - spline interpolation procedure from root @xcite .
figure [ fig:1 ] ( a ) and ( c ) show examples of such a study at rhic energy , with the fit overlaid with the @xmath25 from ampt for set - a with a 10 mb partonic cross - section .
a pseudorapidity gap of 1.5 is imposed to suppress the short range correlations .
the @xmath25 values for different @xmath29 selections cross each other and change sign as one increase @xmath1 , a feature qualitatively similar to the atlas data . figure [ fig:1 ] ( b ) and ( d ) show the @xmath25 values obtained from hijing , which has no final state interaction .
they vary nearly linearly with @xmath46 , consistent with behavior expected for global momentum conservation .
this suggests that the complex dependence of the @xmath25 on @xmath1 in the left panels is a natural consequence of final state interactions .
note that the @xmath25 values for @xmath47 gev are similar between ampt and hijing , this is because @xmath48 is very close to zero in this @xmath5 range ( see figure [ fig:2 ] ) , hence @xmath25 values are dominated by momentum conservation term according to eq .
[ eq : v1b ] .
figure [ fig:2 ] shows @xmath45 extracted from the fit .
the @xmath49 function crosses zero at low @xmath5 , with the crossing point of @xmath50 gev at rhic energy and 0.9 gev at lhc energy , respectively .
since the total transverse momentum arising from dipolar flow is expected be zero , i.e. @xmath28 @xcite , the change of @xmath51 mainly reflects the change of the @xmath52 and @xmath53 ( value shown in the legend ) .
one can show that if @xmath27 is a linear function of @xmath5 , the crossing point is @xmath54 .
however , a change in the @xmath52 and @xmath53 should have little influence on the @xmath3 values at high @xmath5 ; there the @xmath3 values are more sensitive to transport properties of the medium .
the @xmath45 increases with @xmath5 to a value @xmath55 at 2 - 3 gev , and then decreases at higher @xmath5 .
the @xmath49 values for @xmath41 fm and @xmath42 fm are comparable , consistent with the lack of centrality dependence in the initial dipole asymmetry @xcite . the @xmath49 value at lhc energy is larger , consistent with a stronger collective flow at lhc .
figure [ fig:3 ] shows @xmath56 for set - a and set - b at different parton cross - sections tuned by changing @xmath34 .
the magnitude of @xmath45 clearly grows with increasing cross - section .
however the @xmath3 values are more sensitive to changes in @xmath33 : a change of @xmath37 for set - a to 0.33 for set - b almost leads to a 60% increase of the peak value of @xmath3 .
results for set - b with a 1.5 mb cross - section agree better with the atlas data at low @xmath5 .
however both sets fail to describe the trend of the data for @xmath57 gev .
this reflects a limitation in the ampt model for predicting the behavior of @xmath11 at high @xmath5 : the jet quenching physics is important at high @xmath5 , but is not modeled correctly in ampt .
similar studies for @xmath58 also found that the set - b 1.5 mb has the best agreement with the data for @xmath59 gev @xcite .
the original fitting method of atlas can be generalized to measure the @xmath6 dependence of @xmath3 in ampt .
we first perform the 2pc analysis in bins of @xmath60 , and @xmath61 .
the resulting 4-d dataset @xmath62 is then parameterized by : @xmath63 to directly obtain @xmath64 .
a minimum @xmath65 separation for the pairs is imposed to suppress the short range non - flow correlations .
therefore the total number of @xmath25 data points used in the fit depends on this gap requirement and the binning in @xmath5 and @xmath6 . in the default fit
, both @xmath49 and @xmath23 functions are assumed to be symmetric in @xmath6 i.e. @xmath66 and @xmath67 .
in addition , we also have @xmath68 .
we divide the @xmath69 into 12 equally spaced @xmath6 bins and use the same 9 bins in @xmath5 as for fig [ fig:1 ] and [ fig:2 ] .
the fit uses a @xmath70 mesh in @xmath71 space with a simple linear interpolation , which has 54 independent mesh nodes due to symmetry in @xmath6 .
this together with 24 parameters for @xmath23 , gives a total of 78 fitting parameters .
the quality of the fit is generally comparable to those shown in fig .
[ fig:1 ] .
figure [ fig:4 ] shows the @xmath6 dependence of the @xmath3 obtained this way for two @xmath5 bins .
the error bands account for the statistical uncertainty as well as the variation of the fit result by changing the @xmath65 gap from 0.5 to 2.5 .
the @xmath72 generally shows a weak @xmath6-dependence at rhic energy , especially in the more central collisions . at lhc energy ,
however , the @xmath72 shows a small dip at mid - rapidity .
the physics behind this dip is currently under investigation , but probably is related to the stronger longitudinal flow at higher collision energy .
figure [ fig:5 ] shows the values of @xmath73 plotted as a function of @xmath65 for @xmath42 fm at rhic and lhc .
they fall approximately on a common curve for @xmath74 , suggesting that @xmath23 nearly depends only on @xmath65 for pairs with large rapidity separation .
the values of @xmath23 at large @xmath65 are not sensitive to the @xmath65 gap used in the global fit ( compare the solid and open symbols ) , and they are three times larger than that calculated via eq .
[ eq : v1b ] assuming @xmath75 at lhc ( right panel ) , but are comparable at rhic ( left panel ) . in the original functional eq .
[ eq : v1a ] , parameter @xmath23 is assumed to be a single number for each centrality class .
figure [ fig:5 ] clearly show that this is not the case in ampt , and the global momentum conservation contribution changes with @xmath65 .
some of this dependence is due to the influence of the short range correlations such as jet fragmentation and resonance decay , which tend to give a negative contribution to @xmath23 at small @xmath65 .
however , the values of @xmath23 still change with @xmath65 up to @xmath76 then seem to flatten out at higher @xmath65 , but at a level much bigger than that predicted by eq .
[ eq : v1b ] .
this may suggest that evaluating @xmath23 with all particles in the event might be incorrect . in the modified ep method of ref .
@xcite , a @xmath5 dependent weight @xmath77 is calculated for each particle used in the ep calculation .
note that if @xmath23 depends on @xmath6 of the two particles , @xmath53 and @xmath52 need to be calculated in bins of @xmath6 , instead of a single number calculated over the full @xmath6 range of the ep .
in other words , the weight is a function of @xmath5 and @xmath6 : @xmath78 .
in summary , we study the @xmath3 associated with the initial dipole asymmetry in the ampt model using @xmath79 obtained from the two - particle correlation technique .
a global fit of the @xmath25 using the procedure from the atlas collaboration @xcite allows the simultaneous extraction of @xmath27 and the global momentum conservation component .
the excellent fit demonstrates that the complex pattern of the @xmath25 data from the model , which are similar to the atlas data , is a natural consequence of the interplay between final state interaction and global momentum conservation .
the extracted @xmath3 function is negative at low @xmath5 , and increases with @xmath5 until 2 - 3 gev , qualitatively agreeing with the behavior predicted by the hydrodynamic models .
the @xmath3 function shows very little centrality dependence but increases with collision energy and parton cross - sections , suggesting that its magnitude increases with the strength of the final state interactions . by choosing the parameters in ampt carefully , reasonable agreement with the atlas data
can be achieved for @xmath59 gev , but not at higher @xmath5 .
the fitting method is extended to directly extract the @xmath6 dependence of @xmath3 and the global momentum conservation component ( eq . [ eq : v1c ] ) .
the extracted @xmath3 shows a weak dependence on @xmath6 .
the coefficient of global momentum conservation component is found to depend on @xmath31 and differs from that given by eq .
[ eq : v1b ] . this extended method should allow the extraction of the @xmath32 and validation of the ep method ( it requires detector acceptance to @xmath80 , which is usually not satisfied in actual experiment ) .
we acknowledge valuable discussions with matthew luzum and jean - yves ollitrault , in particular , the relation between the event plane method and two - particle correlation method .
we thank roy lacey for a careful proofreading of the manuscript .
this research is supported by nsf under grant number phy-1019387 and phy-1305037 .
10 s. a. voloshin , a. m. poskanzer , r. snellings , arxiv:0809.2949 [ nucl - ex ] , and references there in .
m. miller and r. snellings , nucl - ex/0312008 .
phobos collaboration , nucl .
a * 774 * , 523 ( 2006 ) ; phys .
lett . * 98 * , 242302 ( 2007 ) .
b. alver and g. roland , phys .
c * 81 * , 054905 ( 2010 ) [ erratum - ibid .
c * 82 * , 039903 ( 2010 ) ] .
b. h. alver , c. gombeaud , m. luzum , j. -y .
ollitrault , phys . rev .
c * 82 * , 034913 ( 2010 ) .
p. staig and e. shuryak , phys .
c * 84 * , 034908 ( 2011 ) .
d. teaney and l. yan , phys . rev .
c * 83 * , 064904 ( 2011 ) .
f. g. gardim , f. grassi , y. hama , m. luzum and j. -y .
ollitrault , phys . rev .
c * 83 * , 064901 ( 2011 ) .
bozek , w. broniowski and j. moreira , phys .
c * 83 * , 034911 ( 2011 ) .
phobos collaboration , phys .
lett . * 97 * , 012301 ( 2006 ) .
star collaboration , phys .
* 101 * , 252301 ( 2008 ) .
alice collaboration , j. phys .
g * 38 * , 124167 ( 2011 ) .
alice collaboration , phys .
b * 708 * , 249 ( 2012 ) .
cms collaboration , eur .
j. c * 72 * , 2012 ( 2012 ) .
atlas collaboration , phys .
c * 86 * , 014907 ( 2012 ) .
n. borghini , p. m. dinh , j. -y .
ollitrault , phys .
c * 62 * , 034902 ( 2000 ) .
n. borghini , p. m. dinh , j. -y .
ollitrault , a. m. poskanzer , s. a. voloshin , phys .
c * 66 * , 014901 ( 2002 ) .
n. borghini , phys . rev .
c * 75 * , 021904 ( 2007 ) .
m. luzum and j. y. ollitrault , phys .
lett . * 106 * , 102301 ( 2011 ) .
e. retinskaya , m. luzum and j. -y .
ollitrault , phys .
lett . * 108 * , 252302 ( 2012 ) .
j. xu and c. m. ko , phys .
c * 84 * , 044907 ( 2011 ) .
j. xu and c. m. ko , phys .
c * 84 * , 014903 ( 2011 ) , phys .
c * 83 * , 034904 ( 2011 ) .
j. xu and c. m. ko , phys .
c * 83 * , 021903 ( 2011 ) .
g. -l .
ma and x. -n .
wang , phys .
lett . * 106 * , 162301 ( 2011 ) .
b. schenke , s. jeon and c. gale , phys .
c * 85 * , 024901 ( 2012 ) .
z. qiu and u. w. heinz , phys .
c * 84 * , 024911 ( 2011 ) .
z. qiu , c. shen and u. w. heinz , phys .
b * 707 * , 151 ( 2012 ) .
j. takahashi , b. m. tavares , w. l. qian , r. andrade , f. grassi , y. hama , t. kodama and n. xu , phys .
lett . *
103 * , 242301 ( 2009 ) .
z. chajecki and m. lisa , phys .
c * 79 * , 034908 ( 2009 ) .
z. -w .
lin , c. m. ko , b. -a .
li , b. zhang and s. pal , phys .
c * 72 * , 064901 ( 2005 ) .
m. gyulassy , x. -n .
wang , comput .
commun .
* 83 * , 307 ( 1994 ) . | the anisotropy associated with the initial dipole asymmetry in heavy ion collisions is studied via the two - particle relative azimuthal azimuthal angle ( @xmath0 ) correlations , within a multi - phase transport model . for a broad selection of centrality , transverse momenta ( @xmath1 ) and pseudorapidity ( @xmath2 ) ,
a fitting method is used to decompose the anisotropy into a rapidity - even component , characterized by the fourier coefficient @xmath3 , and a global momentum conservation component .
the extracted @xmath3 values are negative for @xmath4 gev , reach a maximum at 2 - 3 gev , and decreases at higher @xmath5 .
the @xmath3 values vary weakly with @xmath6 and centrality , but increases with collision energy and parton cross - section .
the extracted global momentum conservation component is found to depend on @xmath7 for @xmath8 . |
it has turned out that field theories can be described best in the jet bundle formalism . for first order theories , i.e.
when the lagrange density is a function of the fields and their first derivatives only , the description can be reduced to the first jet bundle .
this description is an extension of the lagrangean formulation of mechanics in the following sense . in particle mechanics ,
particles are described by curves in the tangent space @xmath0 , i.e. by sections of the ( trivial ) bundle @xmath1 . in field theory ,
physical fields are sections of some bundle over space - time , i.e. higher dimensional curves .
consequently , the base manifold of a given field theory which quite often is space - time _ should not be confused with the configuration space of mechanics .
_ rather , the latter is represented by the fibres of the given fibre bundle when the base manifold is @xmath2 , i.e. the time axis .
+ the corresponding field equations are second order but for nicely behaving lagrange densities can be reduced to a system of first order equations , the so - called de donder weyl equations ( for a review see @xcite ) .
one might view these equations as a generalisation of the usual hamilton equations in mechanics .
however , a peculiar feature of this approach is that there are as many conjugate momenta corresponding to each field coordinate degree of freedom as there are space - time dimensions .
+ for the study of quantum effects in field theory , at least in the sense of star products ( @xcite ) there is the need of an algebraic formulation of the theory .
there has been recent progress on that , namely the gerstenhaber structure found by kanatchikov ( @xcite ) .
+ in this article , a geometric construction of this gerstenhaber structure will be proposed that works for arbitrary fibre bundles .
particular attention will be paid to the treatment of internal symmetries and to the correspondence to the standard hamiltonian formulation of field theory .
let @xmath3 be a fibre bundle with @xmath4 dimensional fibres over an @xmath5 dimensional orientable manifold @xmath6 .
its first jet bundle , @xmath7 ( cf .
@xcite ) , consists of equivalence classes of local sections of @xmath8 : two sections @xmath9 and @xmath10 are considered equivalent at a point @xmath11 , if they take the same value in @xmath12 , @xmath13 , and their tangent mappings coincide , @xmath14 . the class defined by a section @xmath9
is denoted by @xmath15 .
@xmath16 is a fibre bundle of fibre dimension @xmath17 , while @xmath18 is an affine bundle with associated vector bundle @xmath19 ( cf .
@xcite ) . here , @xmath20 denotes the vertical ( w.r.t .
@xmath21 ) tangent bundle on @xmath8 , while @xmath22 and @xmath23 stand for the respective canonical projections .
note that @xmath24 is in fact pulled back onto @xmath8 this has been suppressed for simplicity .
let the dual jet bundle @xmath25 be the fibre bundle over @xmath8 with fibres the affine , fibre preserving mappings from @xmath7 to @xmath26 , i.e. for every @xmath27 one has @xmath28 @xmath29 inherits a vector space structure from @xmath30 . as a connection provides a map @xmath31 ( see appendix [ conn - facts ] for details ) , there is an isomorphism @xmath32 that sends the image of @xmath33 to the zero section .
this induces a splitting of the dual jet bundle @xmath34 , @xmath35 where we have introduced the notation @xmath36 for the first summand and @xmath37 for the latter .
observe that @xmath37 is a line bundle over @xmath8 which is trivial for orientable manifolds .
@xmath36 will be called the _
polysymplectic phase space _ from now on .
let @xmath38 label directions on the base manifold @xmath6 , i.e. run from @xmath39 to @xmath5 , while capital latin characters @xmath40 are used for fibre coordinates of @xmath8 , i.e. run from @xmath39 to @xmath4 .
] be local coordinates of @xmath8 .
then to every point @xmath15 in @xmath7 one can assign the matrix elements of @xmath41 in those coordinates , hence we have locally a set @xmath42 for each such point .
further , the choice of coordinates in @xmath8 induces a trivialisation of the dual jet bundle : every affine map @xmath43 can be written as @xmath44 hence @xmath45 provides a set of local coordinates .
then , using the isomorphism @xmath46 , ( [ i - gamma - star ] ) , one can show ( @xcite ) that the first three sets of coordinates can @xmath47 can be used to label @xmath36 locally .
furthermore , the arbitrariness of the choice of a connection connection @xmath33 is projected out , hence the projection @xmath48 from @xmath29 onto @xmath36 is canonical .
definition[@xmath49-vertical differential ] + let @xmath50 be the projection of @xmath51 onto its vertical subbundle @xmath52 constructed from a connection @xmath53 , cf .
( [ conn - facts ] ) .
the corresponding @xmath49-vertical differential @xmath54 is the map @xmath55 satisfying the following properties ( @xmath56 denotes the exterior derivative on @xmath36 ) 1 . for @xmath57 , @xmath58 , one has @xmath59 2 .
@xmath54 acts as a graded derivation of degree @xmath39 . for @xmath60 ,
@xmath61 3 .
4 . @xmath54 vanishes on pulled back forms , i.e. for @xmath63 , @xmath64 one can indeed show that for every @xmath49 there is a unique @xmath54 .
note that @xmath56 and @xmath54 do not anticommute .
+ in what follows , we fix the connection to be the map @xmath49 defined in the appendix , ( [ gamma - bar - coord ] ) .
then @xmath54 is locally given by @xmath65 where @xmath57 is a function and the forms @xmath66 and @xmath67 are defined by ( [ gamma - basis ] ) and together with the @xmath68 form a basis of @xmath69 .
as @xmath70 and fibres of @xmath26 are one dimensional , every point of @xmath36 can be written as @xmath71 , where @xmath72 and @xmath73 are elements of @xmath74 and @xmath26 respectively .
@xmath72 can be viewed as a map @xmath75 .
define an @xmath5-form @xmath76 on @xmath36 by @xmath77 as @xmath73 is linear in each argument , the right hand side is well defined , i.e. does not depend on the splitting @xmath71 .
@xmath76 generalises the construction of the canonical @xmath39-form on a cotangent bundle , but due to the fact that @xmath8 is non trivial one has to choose a connection .
the _ polysymplectic @xmath78-form _ is defined through the action of @xmath54 on @xmath76 , @xmath79 using coordinates as above , one finds @xmath80 note that we have used ( [ gamma - spur ] ) and that @xmath81 indeed does depend on @xmath33 rather than on @xmath49 .
moreover , the @xmath54-cohomology class of @xmath82 is independent of @xmath33 .
definition[hamiltonian forms ] + let @xmath54 be the vertical differential to @xmath49 of ( [ gamma - bar - coord ] ) .
a horizontal form @xmath83 is called hamiltonian if there exists a multi - vector @xmath84 , i.e. an element in @xmath85 , that satisfies @xmath86 in this case , @xmath87 is called hamiltonian multi - vector field .
the set of all hamiltonian forms is denoted by @xmath88 .
note that not every horizontal form is hamiltonian .
rather , condition ( [ ham - vf ] ) imposes a severe restriction on such forms .
proposition[momentum - dep ] [ multimomentum dependence of hamiltonian forms , @xcite ] + let @xmath89 be a hamiltonian form of degree @xmath90 . if @xmath91 , then , using coordinates as above , @xmath92 where the functions @xmath93 are pulled back from @xmath8 . for a proof
, we refer to @xcite .
there is no restriction on functions on @xmath36 .
section [ symmetries ] provides ( standard ) examples for hamiltonian @xmath94-forms .
there is an additional operation on @xmath95-horizontal forms , namely the lifted hodge-@xmath96 operation on @xmath6 .
of course , @xmath96 is invertible . definition and theorem[gerstenhaber structure , @xcite ] +
the maps @xmath97 are well defined ( for the definition of @xmath96 see the remark above ) and satisfy ( let @xmath98 be horizontal forms of ( exterior ) degree @xmath99 , @xmath100 , and @xmath90 , respectively ) 1 .
@xmath101 , 2 .
@xmath102 , 3 .
@xmath103 + @xmath104 , 4 .
@xmath105 .
further , the bracket does not depend on the specific choice of @xmath49 .
a detailed proof can be found in @xcite .
note that , in particular , @xmath106 does not depend on the ambiguity in the correspondence of hamiltonians @xmath107 and @xmath108 and their hamiltonian multi - vector fields @xmath109 , @xmath110 .
in this section , we try to establish a link between the multisymplectic formalism and the standard hamiltonian formulation of field theory . the euler lagrange equations of a given field theory are partial differential equations .
typically a well posed initial value problem for such a set of field equations consists of the specification of two sets of functions @xmath111 on some hypersurface @xmath112 of @xmath6 .
@xmath113 is known as the momentum variable to the field @xmath114 .
the corresponding solution is a set of functions @xmath115 ( @xmath116 labels the additional coordinate in @xmath6 ) on @xmath6 .
the initial data ( [ initial - data ] ) span the field theoretical phase space @xmath117 , while the fields @xmath118 span the covariant section space @xmath119 ( the name has been chosen to stress the analogy to the path space of classical mechanics , cf @xcite ) which typically is some subspace of @xmath120 . despite the similarity in the names , @xmath117 should not be confused with the polysymplectic phase space , @xmath36 .
in particular , both the field theoretical phase space and the section space are infinite dimensional spaces , while @xmath36 is of dimension @xmath121 .
the legendre map provides a map from the section space to @xmath117 , which is explicitely given by @xmath122 where @xmath123 is the lagrange density and @xmath124 denotes all derivatives of the fields @xmath118 . on the field theoretical phase space there is a canonical ( weak symplectic ) @xmath125-form that can be pulled back using the tangent map of ( [ legendre ] ) .
thus there is a ( non - canonical ) @xmath125-form on @xmath126 on the section space @xmath119 .
the vector fields which can be plugged into @xmath126 are infinitesimal ( vertical ) variations of the fields @xmath118 .
they form the tangent space of @xmath119 ( we leave problems arising from the infinite dimensionality aside ) , which therefore consists of all sections of the vertical bundle @xmath20 to @xmath8 . the explicit expression for the contraction of @xmath126 with two such vectors @xmath87 , @xmath127 is found to be ( cf .
@xcite for details ) @xmath128 where @xmath129 denotes a covariant derivative on sections of @xmath130 ( and hence on those of @xmath20 ) .
the @xmath131-summation of this formula reduces to the @xmath116-component if the normal direction of @xmath112 happens to be the time coordinate . on the other hand ,
let @xmath132 be the prolongation to the vertical bundle @xmath20 of a given variational field @xmath87 .
explicitely , one finds @xmath133 where we have used the christoffel symbols of @xmath129 for the map @xmath134 , ( [ def - i - gamma ] ) .
now one can use the legendre transformation and the connection @xmath33 to pull back the polysymplectic @xmath135-form onto @xmath20 .
the contraction of the resulting @xmath135-form with two prolonged variational fields @xmath132 , @xmath136 yields a horizontal @xmath94-form , which can be integrated over @xmath112 .
this yields back the expression ( [ def - omega - l ] ) .
indeed , a calculation shows @xmath137 where @xmath138 denotes the legendre map and the repeated use of the map @xmath134 in the pull back has been suppressed in the notation .
note that the last line vanishes on two vertical vectors which establishes the following result .
lemma[symplectic-2-form ] let @xmath139 be field variations on @xmath6 , i.e. @xmath21-vertical tangent vector fields on @xmath8 .
let @xmath140 be their lifts via the legendre mapping ( [ legendre ] ) taken for arbitrary value of @xmath116 to @xmath36 .
then on every section @xmath141 we find @xmath142 where @xmath143 denotes the prolongation of @xmath118 to @xmath36 using ( [ legendre ] ) again .
the lemma shows how the standard hamiltonian formulation of field theory and the polysymplectic approach can be related .
objects ( vector fields in this case ) in polysymplectic fields theory have to be evaluated on ( prolonged ) sections and integrated over the submanifold @xmath112 that has been chosen for the hamiltonian picture .
for instance , the variations @xmath87 and @xmath127 are given for all sections , but when evaluated on a specific @xmath118 , they become variations of that @xmath118 .
from the peculiar momentum dependence of multisymplectic hamiltonian @xmath94-forms as stated in proposition 3 one immediately concludes the following .
lemma let @xmath87 be a hamiltonian @xmath39-vector field in the sense of ( [ ham - vf ] ) .
then @xmath87 is projectable along @xmath144 onto @xmath8 .
conversely , the canonical lift @xmath145 to @xmath36 of a @xmath21-vertical vector field @xmath87 on @xmath8 is hamiltonian .
its hamiltonian form is given by @xmath146 .
* according to proposition 3 , a general hamiltonian @xmath94-form @xmath147 can be written as @xmath148 where @xmath149 are pulled back functions from @xmath8 . thus the corresponding hamiltonian vector @xmath87 field takes the form @xmath150
so @xmath87 is projectable as stated .
+ let @xmath87 be a vertical vector field on @xmath8 .
@xmath87 generates a one - parameter group of fibre diffeomorphisms on @xmath8 which by jet prolongation can be transported to @xmath7 ( see @xcite , sec .
4b ) . as @xmath29 is affine dual to the first jet bundle , one can lift this action further by pull back and using the canonical projection @xmath48 transport it to @xmath36 . for a given vector field
@xmath151 the coordinate expression of the latter is as follows @xmath152 one immediately verifies that such an @xmath145 is hamiltonian with hamiltonian @xmath94-form @xmath153 : @xmath154 further , as the lifting of two such vector fields @xmath139 is a lie - homomorphism , @xmath155}=[\hat x,\hat y]$ ] , and @xmath156 { \mathbin { \hbox { \!\vrule height 0.4pt width 5.0pt } \hbox{\vrule height 6.0pt width 0.4pt}\ ; } } { \ensuremath{{\theta_{\gamma}}}}$ ] , one has @xmath157 { \mathbin { \hbox { \!\vrule height 0.4pt width 5.0pt } \hbox{\vrule height 6.0pt width 0.4pt}\ ; } } { \ensuremath{{\theta_{\gamma}}}}=i([x , y]).\ ] ] hence the map @xmath158 provides a momentum map for vertical symmetries of @xmath8 .
one might wonder what the different ways to lift vertical automorphisms to the jet bundle and to its affine dual have to do with each other ( cf . [ x - hat ] to [ prol - variation ] ) .
the link is provided by the legendre transformation : if an automorphism leaves the lagrange density invariant , then its lift to the first jet bundle is mapped to its lift to the co - jet bundle via the legendre transformation .
most generally , a connection on a fibre bundle @xmath3 is a projection onto the vertical subbundle @xmath20 .
one can show that such projections are in @xmath159 correspondence with maps @xmath31 . in coordinates
, one writes @xmath160 for @xmath3 being a vector bundle and @xmath33 a linear connection ( cf .
@xcite ) , one has in addition @xmath161 , where @xmath162 denote the christoffel symbols of the corresponding covariant derivative .
if we want to construct a connection on @xmath36 , then we are looking for a @xmath144-fibre preserving map @xmath53 , where @xmath163 is defined w.r.t .
the projection @xmath95 onto @xmath6 .
for such a map , one needs both a connection @xmath33 on @xmath8 and a connection @xmath164 on @xmath165 ( cf .
the coordinate expression of the desired map @xmath166 turns out to be ( cf .
@xcite ) @xmath167 since @xmath165 is a vector bundle we can use the christoffel symbols @xmath168 of @xmath164 .
moreover , @xmath169 but the last term vanishes for torsion free connections @xmath164.with the help of this connection , we can define the a basis on @xmath69 at every point @xmath170 by @xmath171 j. dito : _ lett .
* 20 * , 125134 ( 1990 ) .
+ m. dtsch and k. fredenhagen : _ algebraic quantum field theory , perturbation theory , and the loop expansion _ , hep - th/0001129 .
+ h. rmer and c. paufler : _ anomalies and star products _ , in quantum theory and symmetries , h .-
doebner , j .- d .
hennig , w. lcke , and v. k. dobrev eds . ,
world scientific , singapore , 2000 . | field theoretical models with first order lagrangean can be formulated in a covariant hamiltonian formalism . in this article , the geometrical construction of the gerstenhaber structure that encodes the equations of motion is explained for arbitrary fibre bundles .
special emphasis has been put on naturality of the constructions .
further , the treatment of symmetries is explained .
finally , the canonical field theoretical 2-form is obtained by pull back and integration of the polysymplectic form over space like hypersurfaces .
* key words : * geometric field theory , multisymplectic geometry , hamiltonian formulation , gerstenhaber algebra |
hybridization of single - stranded nucleic acids to form a duplex is a reversible chemical reaction , which is at the basis of many processess and techniques currently used in biotechnology , as for instance pcr @xcite . due to its central importance
, hybridization has been intensively studied in experiments ( focusing on the thermodynamics @xcite or kinetics of the process ) and also in computer simulations @xcite .
the thermodynamics of dna hybridization is usually described by the nearest - neighbor ( nn ) model @xcite .
this model assumes that the free energy of a duplex can be expressed as a sum of dinucleotide stability parameters ; it is therefore based on the principle of additivity . from the nn parameters one can
, for instance , estimate melting temperatures , compute melting curves and predict secondary structures in which rna molecules fold @xcite . in the folding problem , many different local conformations arise as single nucleotide mismatches , bulges , stem - loop structures , etc . describing these conformations in the framework of
the nn model is very challenging and requires a large number of parameters @xcite .
however , only a limited number of them have been measured directly in experiments @xcite .
in addition , one may also wonder whether additivity holds in such cases . to investigate a large number of different conformations
, it would be very advantageous to have access to high - throughput measurements , provided that they are sufficiently accurate . in this article
, we quantitatively determine free energy penalties for mismatches using microarray data obtained from a set of optimized and dedicated experiments . in dna microarrays ,
several thousand of different sequences can be spotted at a surface , hence a large number of hybridization reactions takes place simultaneously .
we use two different approaches : the first one is based on a linear regression of a large set of experimental data points ( @xmath1 ) to fit @xmath2 nn dinucleotide parameters .
the second method relies on the computation of the logarithm of the ratios of fluorescent intensities measured from different spots of the arrays .
we show that both methods provide highly correlated set of nn parameters .
in addition , the second approach allows to probe the limitations of the nn model .
it is found that when two mismatches are closer than @xmath0 nt additivity breaks down and the free energy of the duplex is not equal to the sum of the two separate contributions of isolated mismatches .
we also quantify the influence of mismatches close to the edge of the double helix and show that the free energy penalty is much weaker in those cases .
overall , this work provides new insights on dna hybridization thermodynamics and can help to increase the accuracy of hybridization - based technologies . [ cols="^,^",options="header " , ]
the experiments were performed on custom agilent arrays , following a standard protocol , which is discussed in @xcite . in each experiment ,
a single target sequence in solution was hybridized at concentrations ranging typically from @xmath3 picom to @xmath4 nanom . in total , three different sets of experiments were performed using the target sequences shown in table [ tab01 ] .
these sequences were selected from @xmath5-mers human dnas using optimal design methods @xcite . the theory of optimal design provides some criteria of selecting an optimal set of measurements , which minimize the uncertainties in the parameters of a statistical model ( see supplementary data ) . from the targets of table [ tab01 ] ,
three different microarrays were designed and used for hybridization to either @xmath6 , @xmath7 or @xmath8 .
each microarray contains probes with either zero , one or two mismatches with respect to the given target , covering all possible mismatch combinations . in a stretch of @xmath9 nucleotides
there can be @xmath10 single mismatch probes and @xmath11 double mismatch probes . for @xmath12
this gives in total @xmath13 different sequences , which were spotted in the microarray .
the sequences were replicated @xmath14 times to fill up completely a @xmath15k custom agilent array .
another design was also used mismatches have a minimal distance of 4 nt from the border and a minimal relative distance of 5 nt . in this case
the total number of sequences is @xmath16 .
these sequences were replicated @xmath17 times to fill a 15k custom arrays .
we considered hybridizing sequences of 25 nucleotides .
this is because in previous studies @xcite these sequences were found to attain thermodynamic equilibrium after @xmath18 h of hybridization ( in the experiments the hybridization time is of @xmath19 h , hence thermodynamic equilibrium is guaranteed ) .
a hybridization experiment provides a large number of fluorescence intensities : the highest intensity is from spots containing perfect match sequence , whereas the intensity decreases with the number and type of mismatches .
the reduction of the intensity provides an estimate of the hybridization free energy .
we use two different methods to obtain the nn parameters , as discussed in the next sections .
equilibrium thermodynamics predicts that the measured fluorescence intensity from a spot @xmath20 equals to : @xmath21 where @xmath22 is the hybridization free energy between the target sequence and a probe sequence in @xmath20 , @xmath23 is a parameter , which sets the intensity scale , @xmath24 the target concentration , @xmath25 the gas constant and @xmath26 the temperature ( experiments are performed at @xmath27c @xmath28 , which is the value of the temperature used in the rest of the analysis ) .
although the data analyzed are background - subtracted from the agilent scanner , there remains always some small aspecific signals , which we denote by @xmath29 in equation ( [ iivsdg ] ) . in the experiments @xmath30
is obtained from the average over typically approximately @xmath14 replicated spots .
one should note that equation ( [ iivsdg ] ) is valid at sufficiently low target concentrations , i.e. when only a limited fraction of probes is hybridized in a spot , hence far from chemical saturation . on the other hand , at very low concentrations ,
the specific signal , i.e. the second term in equation ( [ iivsdg ] ) , can become comparable to @xmath29 .
therefore , for the analysis of the data we restricted ourselves to intermediate concentrations and intensities for which we explicitly verified that the intensities scale linearly with concentrations , as predicted by equation ( [ iivsdg ] ) ( more details can be found in the supplementary data ) . in the intensity scale of the experiments
@xmath31 , whereas the values used in the analysis are @xmath32 .
in practice , the large majority of the intensities in experiments with target concentration @xmath33 pm or higher are above this threshold value . in the following
, we will consider the logarithm of the intensities measured with respect to the perfect match ( pm ) intensity .
using equation ( [ iivsdg ] ) , for @xmath34 we get : @xmath35 which defines the free energy penalty of probe @xmath20 with respect to the perfectly matching probe .
this penalty can be expressed as a sum of nn dinucleotide parameters .
consider , for instance , the example of a probe @xmath20 with a single mismatch of type a with respect to the target nucleotide g and with neighboring nucleotides g and t. we have : @xmath36 we use the following notation : the target sequence is the bottom strand and the probe sequence , which is oriented from 5@xmath373 , is the top strand .
this example corresponds to target @xmath38 or @xmath39 at position 10 , counting from 3 end ( the triplet of nucleotides are indicated in bold in table [ tab01 ] ) . in equation ( [ yi_eq2 ] )
@xmath40 is defined as the free energy penalty of an isolated mismatch in a dna duplex .
this penalty is expected to be a local effect . in the nn model
this locality is inherent : the dots in equation ( [ yi_eq2 ] ) indicate identical nucleotides in the two sequences , their contribution cancels out and leaves per isolated mismatch only four dinucleotide parameters around the mismatch position .
there are in total only 58 such dinucleotide parameters : @xmath41 perfect match parameters and @xmath42 single mismatch parameters ( taking into account symmetries ) .
the dinucleotide parameters are not directly experimentally accessible and are not unique @xcite , e.g. they can be shifted by some constant value such that the physically accessible @xmath40 remains unchanged ( see supplementary data ) .
pm from the experiments using hybridization of targets @xmath6 , @xmath7 or @xmath8 , as a function of the @xmath40 parameters obtained from least - squared minimization .
the data agree well with hybridization isotherm given in equation ( [ yi_linear ] ) , shown as a straight line in the linear - log scale . ]
equations ( [ yi_linear ] ) and ( [ yi_eq2 ] ) define a linear problem : each measured @xmath43 can be expressed by a linear combination of dinucleotide parameters . in order to extract the parameters from the data we combined the results of the three experiments and performed a least square minimization of equation ( [ yi_linear ] ) .
mismatches closer than five sites from the helix edges were excluded from the analysis , as well as pairs of mismatches with a distance smaller than 5 nt .
the @xmath2 adjustable parameters were fitted on a set of about a thousand of experimental data points above the intensity threshold .
the fitted parameters then applied to produce the plot as shown in figure [ fig_fit_collapse ] for all available intensities of the experiments in which either sequence @xmath6 , @xmath7 or @xmath8 was hybridized on its corresponding microarray at a concentration of @xmath33 pm . the data are plotted as a function of the unique @xmath40 for triplets defined as in equation ( [ yi_eq2 ] ) .
we note that there is very good agreement between the data and the thermodynamic model of equation ( [ iivsdg ] ) .
the experiments follow the equilibrium isotherm ( a straight line with a slope equal to @xmath44 ) for a range of intensities of more than four orders of magnitude
. a previous study @xcite in which hybridizing strands were 30-mers did not provide a single straight line in a @xmath45 versus @xmath40 plot .
deviations due to lack of thermodynamic equilibrium were observed in the high - intensity ranges , as discussed in @xcite .
further it is important to note that we do not only find internally consistent results , but that our microarray - derived free energy parameters also correlate to a fair degree with those reported in literature for hybridization in solution @xcite .
figure [ fig_array_vs_solution ] shows a correlation plot of the free energy penalties ( i.e. the @xmath46 defined as in the example of equation ( [ yi_eq2 ] ) ) obtained from the microarray data analysis and those from santalucia et al . from @xcite .
the spearman correlation coefficient is equal to @xmath47 .
this clearly shows that free energy parameters for dna features measured by the presented microarray approach also apply for thermodynamic properties in solution .
this opens the highly parallelled microarray toolbox for the study of thermodynamics of dna structures .
an example is discussed in the next section . for triplets obtained from the microarray fit versus those from hybridization in solution @xcite .
the central mismatching nucleotides of the triplet ( underlined in equation ( [ yi_eq2 ] ) ) are indicated in the plot . ]
the crucial assumption of the nn model is additivity of local free energy contributions .
we probe here the limits of additivity of free energy penalties as a function of the distance between two mismatches .
we will access the free energy parameters by comparing ratios of intensities measured from different spots in the microarray . and @xmath48 at positions
@xmath49 and @xmath50 respectively , ( d ) hybridization with a probe carrying two mismatches .
we use the notations @xmath51 , @xmath52 and @xmath53 to denote the corresponding intensities measured in the experiment.,title="fig : " ] + hereto , we combine microarray spots that contain probes with zero , one or two mismatches with respect to the target and we denote the location of the mismatch by @xmath49 or @xmath54 as illustrated in figure [ fig_mm_sketch ] .
the associated free energy penalties can then be derived from the intensity measurements as follows @xmath55 in which the superscript @xmath56 and @xmath48 represent the three possible mismatching nucleotides at location @xmath49 and @xmath57 respectively .
if the additivity of the nn model holds , the free energy penalty of equation ( [ eq penalty3 ] ) should equal the sum of the individual penalties of equations ( [ eq penalty1 ] ) and ( [ eq penalty2 ] ) . to test this
, we introduce @xmath58 which measures the relative deviation from additivity .
figure [ fig_interaction_alpha ] shows the experimental results for @xmath59 in which we averaged over @xmath49 , @xmath56 and @xmath48 , leaving @xmath59 as a function of the distance @xmath60 between two mismatches . from this data
, we notice that @xmath59 has a value of about zero when the mismatches are separated by @xmath61 nt , but a clear positive value for smaller @xmath62 .
apparently the free energy penalty of two nearby mismatches is smaller than the sum of the two individual contributions , resulting in a positive @xmath59 .
furthermore , the inset from figure [ fig_interaction_alpha ] shows that the relationship is linear in a semi logarithmic plot , hence @xmath59 decays exponentially with @xmath60 . note that at @xmath63 only one mismatch is present , hence @xmath64 and @xmath59 will be identical to @xmath65 according to equation ( [ eq alpha ] ) .
all these observation result from direct measurement values , containing no fitting parameters and strongly suggest that in double - stranded dna , mismatches have a physical interaction with each other which decays exponentially to zero over a distance of about five nucleotides .
, the relative deviation from additivity , from the experiment of target @xmath38 , averaged over @xmath66 and @xmath48 as a function of the distance @xmath60 between two mismatches .
the inset shows the plot with @xmath59 in log scale.,title="fig : " ] + these results are setting some limitations on the additivity of the nn model .
however , outside this interaction region of @xmath67 nt we expect the nn model to hold i.e. @xmath59 should be zero and mismatches can be considered as isolated
. this can be explicitly checked in a very direct way .
when @xmath68 we get from equation ( [ eq alpha ] ) @xmath69 the free energy penalty @xmath70 of a mismatch @xmath56 at location @xmath49 , which we will call the focus mismatch @xmath71 , can be estimated either directly using equation ( [ eq penalty1 ] ) or via a second mismatch @xmath72 using equations ( [ eq penalty2 ] ) and ( [ eq penalty3 ] ) for any choice of @xmath48 and @xmath73 . hence , the free energy penalty of the focus mismatch can be estimated from measurements in many independent ways and they should provide the same answer if additivity holds . note
that , using equations ( [ eq penalty2 ] ) and ( [ eq penalty3 ] ) @xmath74 drops out in the right hand side of equation ( [ direct ddg ] ) .
figure [ fig_median ] illustrates how equation ( [ direct ddg ] ) can be used to estimate the @xmath40 using different combinations of @xmath48 and @xmath62 . in this specific example
we consider @xmath75 which corresponds both for target @xmath38 and @xmath39 to @xmath76 ( in the supplementary data , we show other examples featuring additivity for different focus mismatches ) . in the pane for target @xmath7 ,
all the estimates of the free energy penalty are close the each other , the @xmath77 estimates tightly lie around a median value , in this case @xmath78 kcal / mol , indicated by the dotted line . the picture in the right pane is a typical one which we observe for any focus mismatch @xmath71 .
this confirms that additivity holds in the regime @xmath79 , i.e. when mismatches are separated by @xmath80 nt .
moreover , it shows that the microarray measurement is internally consistent .
secondly , the left pane , i.e. experiment @xmath6 , provides the same median value for the free energy penalty , showing also the robustness of the microarray approach to estimate free energies of dna structures .
however , this figure was chosen because it is atypical in the sense that one notices two pronounced outlying values .
they correspond to a sequence where both the focus mismatch and the second mismatch are of type @xmath81 .
since they clearly deviate from an otherwise nicely consistent picture , we believe there must a physically underlying reason for it .
we will come back to this point in the section where we discuss thermodynamic outliers .
note that with this second method we accessed values for the free energy penalties of isolated mismatches without using any multiple regression or fitting procedure , but we simply compared the ratios of intensities , equations ( [ eq penalty1])-([eq penalty3 ] ) , to get a consistent set of independent estimates .
the free energy penalties are then obtained from the median over all data points .
we compared the free energy penalties obtained from this method ( median ) with those obtained from linear regression as discussed in the previous section .
the two sets of data are well - correlated with a pearson correlation coefficient equal to @xmath82 ( see supplementary data ) .
this correlation shows the equivalence of the two approaches . in this analysis , we restricted ourselves to mismatches in the bulk of the sequence , i.e. @xmath49 is @xmath83 nt from the border .
closer to the border we observe boundary effects , which are covered in the next section . the previous section ended by showing
the equivalence of both approaches to access free energy penalties of an isolated mismatch , provided the data are restricted to bulk mismatches .
the direct median method of the previous section can also assess penalties of mismatches close to the boundary , whereas on the contrary the fitting method can not by construction .
the latter , however , has the advantage of fitting a full parameter set of the nn model and as such can easily provide bulk values for the free energy penalty of any isolated mismatch .
the combination of both methods now provides an elegant way to assess the effect of boundary proximity on an isolated mismatch .
hereto , we introduce the parameter @xmath84 as the relative reduction of free energy penalty of a mismatch when compared to its bulk value .
@xmath85 , the relative reduction of mismatch free energy penalty , as a function of location for experiment with target @xmath6 .
each point is the average of three estimates , one per possible mismatch .
data are absent for the extremal locations @xmath86 and @xmath87 , since no value can be calculated by the nn model . ] in figure [ fig boundary ] the parameter @xmath84 is shown as a function of @xmath49 after averaging over @xmath56 .
it is clear that , as expected , @xmath84 is approximately equal to one in the bulk , whereas when approaching the boundary , a reduction of free energy penalty occurs which reaches up to 80% .
note that for mismatches at the boundary , @xmath86 and @xmath87 , the nn model is not applicable and no data is presented .
figure [ fig boundary ] show that the range of the boundary effect is @xmath88 nt . as a final result of this article
, we come back to the two outliers observed in figure [ fig_median](a ) ; the same deviations are found in replicated experiments at different concentrations : therefore , they are unlikely due to experimental errors . for these two cases
we find @xmath89 kcal / mol and @xmath90 kcal / mol , strongly deviating from the median value ( @xmath91 kcal / mol ) .
the common feature of these two sequences is that they involve ga mismatches .
the two set of mismatches are arranged in an antiparallel way i.e. one g and one a are on the same strand .
mismatches of ga type in dna and rna helices have been the subject of several studies in the past @xcite . in the rna folding
, it is known that ga pairs contribute substantially to the rna helix stability .
their contribution is comparable to that of a canonical at pair .
as at pairs , ga form two hydrogen bonds , but can also assume four different conformations @xcite .
the microarray data suggest that the antiparallel combination of ga and ag pairs of mismatches have a long range interaction effect , which is probably a signature of some structural conformational change of a double helix containing these pairs .
next - nearest neighbor effects extending up to @xmath67 nt distance for antiparallel ga mismatches have been reported in the case of rna duplexes in @xcite ( longer distances were not considered that case ) .
we investigated antiparallel ga and ag pairs of mismatches also in sequences @xmath7 and @xmath8 , but found no anomalous behavior in those cases .
this suggests that the nucleotide sequences between the two @xmath92 pairs plays an important role in the overall stability of the duplex . as a further proof of the outlying behavior of antiparallel @xmath92 pairs we show in figure [ fig_tandem ] a plot of free energy penalties for tandem mismatches ( neighboring double mismatches )
these are again obtained from equation ( [ eq penalty3 ] ) for different @xmath56 and @xmath48 mismatching nucleotides , where in the case of tandem mismatches , @xmath62 is equal to 1 . on each location of the sequence our data set
contains nine different types of tandem mismatch .
a clear boundary effect is noticeable , but when looking at the bulk data points tandem mismatch of the type @xmath92 are again outlying , they appear to be particularly stable with a free energy penalty @xmath93 kcal / mol below average .
in this article , we have analyzed dna hybridization reactions in microarrays and quantified free energy penalties of single and double mismatches .
we have shown that the experimental data are very precise and reproducible .
the microarray data follow an equilibrium isotherm over a range of four orders of magnitude in the fluorescence intensities and allow the extraction of accurate thermodynamic parameters .
first , the analysis provides a database with a large number of nn parameters for isolated mismatches .
these parameters correlate well with those reported in the literature from hybridization experiments in solution .
second , the experiments contain systematic measurements of hybridization with two mismatches , which allowed us to probe the validity limit of the nn approximation .
we showed that when two mismatches are separated by a distance of @xmath61 nt their effect is additive , allowing a standard approach with the nn model .
however , for shorter distances , the additivity is no longer valid and we found that duplexes with neighboring mismatches are more stable than expected from additivity .
this interaction was shown to decay exponentially as a function of the distance between mismatches .
further , we investigated the behavior of mismatches close to the helix edges , and showed that their free energy penalty is reduced up to @xmath94 when compared to the bulk behavior .
the boundary effect was observable up to @xmath67 nt from the helix edge .
finally , we also found some thermodynamic outliers , sequences involving two antiparallel ga mismatches , in which the mismatch interaction appears to persist beyond @xmath0 nt .
these outliers were not related to experimental error indicating a signature of some structural conformational change of a double helix containing these mismatch pairs .
overall , the analysis of the microarray data reported in this article provides new quantitative insights on the dna hybridization parameters , on the nn model and its present limitations .
our study is in line with a number of recent articles , which have been dedicated to the investigations of fundamental physico - chemical properties of dna arrays @xcite . due to the relevance of hybridization in many technologies ,
going from pcr @xcite to recent developments in biosensors , e.g. @xcite , a good thermodynamic model is also important from the application point of view .
a precise quantification of interaction free energies involved in the hybridization will help to increase the accuracy of microarrays and other hybridization - based technologies , so that these devices could realize their full potential , for instance , for clinical applications @xcite .
for these applications , an increase in specificity and sensitivity is very important and can be achieved through better understanding of fundamental properties of hybridization in these devices .
there has been considerable attention in recent years @xcite in understanding the fundamentals of hybridization in dna microarrays and its impact in data analysis .
here , we have shown that microarrays are a reliable and high - throughput tool to gain insight on dna hybridization thermodynamics .
the same method could be used to screen other types of defects , as bulges .
indeed , it was recently used for understanding loop conformations @xcite .
supplementary data available in appendix .
we thank karen hollanders for expert technical assistance . we acknowledge financial support from research foundation - flanders ( fwo )
g.0311.08 , kuleuven grant no .
strt1/09/042 and vito grant zl39010200 - 401 .
conflict of interest statement : _
none declared_. 34 alberts , b . ,
johnson , a . ,
lewis , j . , raff , m . , roberts , k . and
walter , p .
( 2002 ) _ molecular biology of the cell_. garland science , new york .
breslauer , k.j . ,
frank , r .
, blcker , h . and marky , l.a .
( 1986 ) predicting dna duplex stability from the base sequence .
usa _ , * 83 * , 37463750 .
santalucia , j .
( 1998 ) a unified view of polymer , dumbbell , and oligonucleotide dna nearest - neighbor thermodynamics . _
usa _ , * 95 * , 14601465 .
sambriski , e.j . ,
schwartz , d.c . and de pablo , j.j .
( 2009 ) a mesoscale model of dna and its renaturation .
_ biophys .
j. _ , * 96 * , 16751690 .
bloomfield , v.a .
, crothers , d.m . and
tinoco , i .
( 2000 ) _ nucleic acids structures , properties and functions_. university science books , mill valley .
mathews , d.h . ,
disney , m.d . , childs , j.l . , schroeder , s.j .
, zuker , m . and turner , d.h .
( 2004 ) incorporating chemical modification constraints into a dynamic programming algorithm for prediction of rna secondary structure . _
* 101 * , 72877292 .
andronescu , m . ,
condon , a .
, hoos , h.h . , mathews , d.h . and
murphy , k.p .
( 2010 ) computational approaches for rna energy parameter estimation . _
rna _ , * 16 * , 23042318 .
santalucia , j .
jr . and hicks , d .
( 2004 ) the thermodynamics of dna structural motifs .
_ , * 33 * , 415440 .
hooyberghs , j . , van hummelen , p . and
carlon , e .
( 2009 ) the effects of mismatches on hybridization in dna microarrays : determination of nearest neighbor parameters . _
nucleic acids res .
_ , * 37 * , e53 .
atkinson , a.c . and
donev , a.n .
( 1992 ) _ optimum experimental designs_. clarendon press , oxford .
hooyberghs , j . , baiesi , m . ,
ferrantini , a . and
carlon , e .
( 2010 ) breakdown of thermodynamic equilibrium for dna hybridization in microarrays .
e _ , * 81 * , 012901 .
gray , d.m .
( 1997 ) derivation of nearest - neighbor properties from data on nucleic acid oligomers .
i. simple sets of independent sequences and the influence of absent nearest neighbors . _
_ , * 42 * , 783793 .
walter , j .- c . , kroll , k.m . ,
hooyberghs , j . and
carlon , e .
( 2011 ) nonequilibrium effects in dna microarrays : a multiplatform study
. _ j. phys .
b _ , * 115 * , 67326739 .
li , y . , zon , g . and wilson , w.d .
( 1991 ) thermodynamics of dna duplexes with adjacent g.cntdot.a mismatches . _
_ , * 30 * , 75667572 .
li , y . , zon , g . and wilson , w.d .
( 1991 ) nmr and molecular modeling evidence for a g.a mismatch base pair in a purine - rich dna duplex .
usa _ , * 88 * , 2630 .
santalucia , j .
turner , d.h .
( 1993 ) structure of ( rggcgagcc)2 in solution from nmr and restrained molecular dynamics . _
_ , * 32 * , 1261212623 .
lane , a . ,
ebel , s . and brown , t .
( 1994 ) properties of multiple g @xmath95 a mismatches in stable oligonucleotide duplexes .
j. biochem .
_ , * 220 * , 717727 .
ebel , s . ,
brown , t . and lane , a.n .
( 1994 ) thermodynamic stability and solution conformation of tandem g @xmath95 a mismatches in rna and rna @xmath95 dna hybrid duplexes .
_ , * 220 * , 703715 .
morse , s.e . and
draper , d.e .
( 1995 ) purine - purine mismatches in rna helices : evidence for protonated g.a pairs and next - nearest neighbor effects . _
nucleic acids res .
_ , * 23 * , 302306 .
seela , f. and budow , s .
( 2008 ) mismatch formation in solution and on dna microarrays : how modified nucleosides can overcome shortcomings of imperfect hybridization caused by oligonucleotide composition and base pairing . _
_ , * 4 * , 232245 .
chang , c.y.(j . ) and stellwagen , n.c .
( 2011 ) tandem ga residues on opposite sides of the loop in molecular beacon - like dna hairpins compact the loop and increase hairpin stability . _
_ , * 50 * , 91489157 .
trapp , c .
, schenkelberger , m . and
( 2011 ) stability of double - stranded oligonucleotide dna with a bulged loop : a microarray study .
_ bmc biophys .
_ , * 4 * , 20 .
golovkin , m.v . ,
matveeva , o.v . and nechipurenko , iu.d .
( 2009 ) properties of hybridization isotherms upon binding of ligands on microchips .
_ biofizika _ , * 54 * , 820823 .
pozhitkov , a.e . ,
boube , i . , brouwer , m.h . and noble , p.a .
( 2010 ) beyond affymetrix arrays : expanding the set of known hybridization isotherms and observing pre - wash signal intensities . _
nucleic acids res .
_ , * 38 * , e28 .
irving , d . ,
gong , p . and levicky , r .
( 2010 ) dna surface hybridization : comparison of theory and experiment .
_ j. phys .
b _ , * 114 * , 7631 - 7640 . fuchs , j .
, fiche , j.b . ,
buhot , a . ,
calemczuk , r . and livache , t .
( 2010 ) salt concentration effects on equilibrium melting curves from dna microarrays .
_ biophys .
j. _ , * 99 * , 18861895 .
burden , c.j . and binder , h .
( 2010 ) physico - chemical modelling of target depletion during hybridization on oligonulceotide microarrays .
_ , * 7 * , 016004 .
qamhieh , k . , wong , k .- y . , lynch , g.c . and pettitt , b.m .
( 2009 ) the melting mechanism of dna tethered to a surface .
_ , * 6 * , 474488 .
gharaibeh , r.z . ,
fodor , a.a . and
gibas , c.j .
( 2010 ) accurate estimates of microarray target concentration from a simple sequence - independent langmuir model . _
plos one _ , * 5 * , e14464 .
lee , c .- y . , harbers , g.m . ,
grainger , d.w . , gamble , l.j . and castner , d.g .
( 2007 ) fluorescence , xps , and tof - sims surface chemical state image analysis of dna microarrays .
_ , * 129 * , 94299438 .
binder , h . ,
preibisch , s . and
kirsten , t .
( 2005 ) base pair interactions and hybridization isotherms of matched and mismatched oligonucleotide probes on microarrays .
_ langmuir _ , * 21 * , 92879302 .
van grinsven , b . , vanden bon , n . ,
grieten , l . ,
murib , m . , janssens , s.d . , haenen , k . ,
schneider , e . ,
ingebrandt , s . , schning , m.j . , vermeeren , v . ,
ameloot , m . ,
michiels , l . , thoelen , r . , de ceuninck , w . and wagner , p .
( 2011 ) rapid assessment of the stability of dna duplexes by impedimetric real - time monitoring of chemically induced denaturation . _ lab chip _ , * 11 * , 16561663 .
hooyberghs , j . and
carlon , e .
( 2010 ) hybridisation thermodynamic parameters allow accurate detection of point mutations with dna microarrays . _
_ , * 26 * , 16921695 .
shanahan , h.p .
, memon , f.n . , upton , g.j.g . and
harrison , a.p .
( 2012 ) normalized affymetrix expression data are biased by g - quadruplex formation . _
nucleic acids res .
_ , * 40 * , 33073315 . * supplementary data *
according to the nearest - neighbor model , the total hybridization free energy of a target to a probe can be expressed as a sum of the dinucleotide parameters @xmath96 accounting for hydrogen bonding and stacking interactions .
the index @xmath59 covers all possible dinucleotide parameters .
some examples are : @xmath97 where the underlined nucleotides indicate mismatches . in total
there are @xmath41 perfect match parameters ( taking into account symmetries ) and @xmath42 parameters in the case of a single mismatch .
these dinucleotide parameters are known not to be unique , see e.g. @xcite .
thermodynamics predicts that the intensity measured from a spot @xmath30 is given by : @xmath98 where @xmath22 is the total hybridization free energy between a target and a probe , @xmath23 is a parameter which sets the intensity scale , @xmath24 the target concentration , @xmath25 the gas constant and @xmath26 the temperature .
@xmath29 is the aspecific signal that can be considered as background . in this paper
the stability of duplexes was always compared to that of the perfect match , i.e. @xmath99 which defines the free energy penalty of probe @xmath20 with respect to the perfectly matching probe .
this penalty can be expressed as a sum of nearest - neighbor dinucleotide parameters : @xmath100 where @xmath101 is the frequency matrix , which counts the number of times a given dinucleotide term contributes to @xmath43 . as an example , for an isolated mismatch of type ga we have : @xmath102 for notational convenience we used , by symmetry , the equality of @xmath103 to have the mismatch on the right hand side of the dinucleotide . for any given @xmath20 , the matrix elements @xmath104
are all zero except for the four dinucleotide terms of equation ( [ yi_eq2_supp ] ) which contribute by @xmath105 for the two dinucleotides with mismatches and @xmath106 for the two perfect matching dinucleotides .
equation ( [ lin_prob_suppl ] ) defines a multiple linear regression , from which the @xmath2 dinucleotide parameters can be fitted to match all the observed free energy penalties of mismatches .
note that it defines the dinucleotide parameters not in a unique way , e.g. the following transformation @xmath107 in which the same constant @xmath108 is added and subtracted to different dinucleotide parameters , leaves equation ( [ yi_eq2_supp ] ) invariant .
the triplet parameters , such as defined in the last line of equation ( [ yi_eq2_supp ] ) are however unique as expected , since they are directly physically accessible .
as discussed above , the dinucleotide parameters can be obtained from a linear fit from @xmath9 independent experimental measurements .
such an approach always contains some uncertainties .
these uncertainties can be lowered if one takes @xmath9 large . in our specific case @xmath9
equals the number of spots on the microarrays , and can be increased by combining data from more arrays ( see main paper for experimental setup ) .
further , for a given fixed value of @xmath9 one can use some optimization criterion to select the best @xmath9 measurements which minimize the uncertainties on fitted parameters . in our case
this comes down to the selection of a target sequence with good statistical properties .
the theory of _ optimal design _ establishes some criteria for this purpose and we briefly discuss this theory here . before entering into the details of the optimization followed in the microarray experiment
we discuss a one dimensional example , which illustrates the optimization method .
let us take the example of a simple linear regression with an intersect set to zero ( corresponding to a one - dimensional system ) : @xmath109 where @xmath84 is the unknown of the problem , @xmath110 and @xmath43 are respectively the input and output of the experiment @xmath20 and can take any real value .
the parameter @xmath84 can be obtained by the least square method : @xmath111 where the symbol @xmath112 means the average over the @xmath9 elements .
the error on @xmath84 is given by : @xmath113 where @xmath114 is the cost function of the system .
equation ( [ costfunc ] ) implies that the error can be decreased by enlarging the sampled points ( @xmath9 ) or , for @xmath9 fixed , by increasing the variance of the variable @xmath110 .
the latter criterion can be used in the design of the experiment by performing measurements @xmath43 for a well spread set of points @xmath110 .
indeed , it is intuitively clear that when @xmath110 are very close to each other ( small variance ) one has a large uncertainty on the estimate of the slope @xmath84 . in what follows
we discuss about optimal design criteria in higher dimensions , which roughly correspond to the idea of the maximization of the variance in the previous one - dimensional example .
we define first the so - called _ information matrix _ @xmath115 , where @xmath116 is the frequency matrix defined in equation ( [ lin_prob_suppl ] ) and where @xmath117 denotes its transpose . in terms of matrix elements : @xmath118 which is thus in our case a square symmetric matrix of dimension @xmath119 .
the information about the quality of the experimental design is encoded in @xmath120 and in our case is defined by the sequence of the target oligo in the experiment ( see main paper for experimental setup ) .
the three most used criteria in optimal design are the a- , d- and e - optimality .
a - optimality corresponds to minimizing the trace of @xmath121 , d - optimality corresponds to minimizing the determinant of @xmath121 and e - optimality corresponds to maximizing the lowest eigenvalue of @xmath120 . roughly speaking , these strategies amounts to maximize the information encoded in @xmath120 @xcite .
we note that in the linear problem of equation ( [ lin_prob_suppl ] ) the information matrix has a minimum of @xmath122 null eigenvalues ( see the supplementary material of ref .
@xcite for a detailed explanation ) .
these come from unavoidable degeneracies of the problem , or equivalently from the fact that the dinucleotide parameters are not unique ( see e.g. equations ( [ shift1_suppl ] ) and ( [ shift2_suppl ] ) ) . having some zero eigenvalues , the information matrix @xmath120 is not invertible , therefore we are working with pseudo - inverse which is obtained from the singular value decomposition of @xmath120 .
the three target sequences , @xmath123 and @xmath8 , which were used for the experiments and which are mentioned in table 1 of the main article were selected as follows .
we collected a set of candidate targets by scanning over a piece of the human genome and taking subsequences of length 25 .
the first criterion was to choose sequences with minimum , unavoidable , number of 7 zero eigenvalues in order to get the minimum number of degeneracies when solving the linear system to estimate the nearest - neighbor parameters , as discussed above . for @xmath6
, we considered a subset of sequences with a minimum distance of 3 nucleotides from the border and a minimum distance of 3 nucleotides between 2 mismatches . for @xmath7 and @xmath8 , the minimal distance from the border
is 4 nucleotides and the distance between 2 mismatches is at least 5 nucleotides .
therefore , the constraint on the subset to select @xmath6 was weaker than @xmath7 and @xmath8 .
since the constraint for @xmath7 and @xmath8 is stronger than @xmath6 , in this case , the number of equations in the linear system is lower and it is more difficult to find subsequences of length 25 which display the minimum number of 7 zero eigenvalues . for the same order of calculation
, we managed to find 130 sequences for @xmath6 and only a few sequences for @xmath7 and @xmath8 . for @xmath6 , this set of candidates
was subsequently ranked according to the three optimal design criteria a , d and e. finally , the candidate targets which ended up as top - ranked on all three criteria were retained .
moreover , we checked the energy for the target to fold on itself .
for the 3 targets , it takes a reasonable value .
, @xmath124 , as a function of the intensity @xmath125 .
the used target concentrations are @xmath126 , @xmath127 , @xmath128 of target @xmath7 . ] as a measurement device the microarray technology is faced with a detection limit in the low measurement regime and a saturation in the high end : see sketch in figure [ fig_linear_sketch ] . in our research we want to limit ourselves to measurements in the linear regime .
to assess which data meet this requirement we combine experiments which are identical ( identical target sequence , identical probe sets , identical hybridisation conditions ) except for the concentration @xmath129 of the target . if the data is in the linear regime we expect the intensity of a spot @xmath20 of the experiment with target concentration @xmath129 to be @xmath130 if we now combine two experiments , one with target concentration @xmath129 and one with @xmath124 , and define for each spot @xmath20 the quantity @xmath25 as @xmath131 than we expect @xmath25 to be equal to @xmath132 when both intensities are in the linear regime
. however for low @xmath129 the spot intensity @xmath125 can be close to detection limit and consequently be higher than predicted by the theory of equation ( [ i_i ] ) , or for high @xmath133 the intensity @xmath134 can be close to saturation and consequently lower than theoretically expected . in both cases
@xmath25 will be above one .
the result of this analysis is shown in figure [ fig_linear_regime ] for the combinations @xmath135 and @xmath136 of target @xmath7 . from this picture
it is clear that for a large part of the intensity range @xmath25 equals one and supports the linear regime . for the green dots ,
there is a deviation in the high intensity range due to the proximity of saturation of these spots in the @xmath137 experiment .
for the red dots a deviation is present due to proximity of the detection limit for these spots in the @xmath138 data .
this approach gives a criterion to assess the validity of the linear regime per spot and the possibility to make a correction for the non - linear behaviour close to saturation or detection limit .
in the main article the additivity of free energy penalties of mismatches was shown when mismatches were separated by more than four nucleotides . for two examples , this
was explicitly shown in figure 5 of the main article . in this section
we add some further examples of the additivity with similar plots .
these are shown in figure [ fig_additivity ] . ) . ]
in the main article we present two different approaches that can be used to estimate free energy penalties of single mismatches in a triplet of nucleotides such as in equation ( 3 ) of the main article . the first method , i.e. by linear fitting , produces a robust estimation provided that each of the 58 nn dinucleotide parameters are equally well - represented .
this was achieved by the use of optimal design principle in designing the experiments .
another method is by taking the median of data points from ratios of intensities following equations ( 4)-(6 ) of the main article .
figure [ fig_additivity ] of this document shows six of these unique triplets in which the free energy penalties are indicated by the horizontal line from taking the median of each independent estimates .
it is then imperative to see if these two methods are equivalent in providing the estimates .
figure [ fig_median_vs_fit ] shows that the free energy penalties calculated from the two methods are well - correlated with pearson correlation @xmath82 ( such as mentioned in the main article ) .
this is indicating the equivalence of the two methods .
this is also a proof that our experiments are self - consistent from the different perspective of these two approaches .
gray , d.m .
( 1997 ) derivation of nearest - neighbor properties from data on nucleic acid oligomers .
i. simple sets of independent sequences and the influence of absent nearest neighbors . _
_ , * 42 * , 783793 . | in this article it is shown how optimized and dedicated microarray experiments can be used to study the thermodynamics of dna hybridization for a large number of different conformations in a highly parallel fashion . in particular ,
free energy penalties for mismatches are obtained in two independent ways and are shown to be correlated with values from melting experiments in solution reported in the literature .
the additivity principle , which is at the basis of the nearest - neighbor model , and according to which the penalty for two isolated mismatches is equal to the sum of the independent penalties , is thoroughly tested .
additivity is shown to break down for a mismatch distance below @xmath0 nt .
the behavior of mismatches in the vicinity of the helix edges , and the behavior of tandem mismatches are also investigated . finally , some thermodynamic outlying sequences are observed and highlighted .
these sequences contain combinations of ga mismatches .
the analysis of the microarray data reported in this article provides new insights on the dna hybridization parameters and can help to increase the accuracy of hybridization - based technologies . |
trapped atomic ions have been regarded as a very promising physical system in quantum information processing .
they have been used for many proof - of - principle experiments such as fundamental quantum gate operations@xcite , generation of entangled states@xcite , and quantum simulation of coupled spins@xcite . in these pioneering studies , the quantized motional and internal states of trapped ions
were controlled using laser radiation . in spite of such successful works ,
the use of laser radiation for coherent manipulation has a few issues .
if two levels are coupled by a raman transition , spontaneous emission causes the destruction of quantum coherence@xcite . on the other hand ,
if a ground state and a metastable state are chosen as a two - level system and driven by a quadrupole transition , hertz - level laser stability is typically demanded .
in order to achieve further scalability , it is desirable to develop a system that is not affected by the decoherence due to spontaneous emission and is less demanding of the laser system .
the use of radio frequency ( rf ) or microwave for the manipulation of quantum states has several advantages from these viewpoints@xcite .
spontaneous emission does not affect the gate operation , which would lead to better fidelity in quantum - state manipulation .
in addition , the generation and control of rf or microwave radiation requires a simpler system compared to that of laser radiation .
thus , a laser system is used only for laser cooling and state detection , which reduces the demand on their stability . to perform the quantum simulation of interacting spins or two - ion entangling gates ,
the state coupling of motional and internal quantum states is required . as proposed in @xcite , exposing an ion string to a magnetic field with spatially varying magnitudes enables such coupling .
the coupling strength is proportional to the square of the field gradient .
several studies aimed at generating a magnetic - field gradient at trapped ions have been reported .
individual addressing of trapped ions and coupling of motional and internal states have been performed using permanent magnets , which were set at the outside of end electrodes of a linear paul trap @xcite .
however , the magnitude of the field gradient is limited by the distance between the ion and magnets because it is generally difficult to integrate magnets inside a conventional linear paul trap .
one approach involved a current introduced on wires fabricated in a surface - electrode trap for the generation of magnetic - field gradients@xcite , which can reduce the distance between the ions .
this design is suitable for improving scalability ; however , it requires heat management . another approach involved the use of oscillating magnetic fields in a surface - electrode trap@xcite .
the use of the near - field amplitude gradient produced in a surface - electrode trap was proposed @xcite , and gate operation was demonstrated @xcite . here , we present a different approach to generate a large field gradient at trapped ions .
we integrate permanent magnets underneath a surface - electrode trap .
this structure remarkably reduces the distance between permanent magnets and trapped ions .
a large field gradient can be expected without any heat management .
we discuss design considerations and present the numerical results of the magnetic field at the trapping region .
we show that the trap is implemented by assembling a multi - segmented surface - electrode trap and an alumina plate in which permanent magnets are buried . to evaluate the field gradient , we observe the zeeman splitting of the @xmath2 - @xmath3 transition in a @xmath0ca@xmath1 ion at several positions .
finally , we discuss the generation of an even larger field gradient .
the trap is composed of two parts ; a surface - electrode trap and a magnet layer . the surface - electrode trap is made of a square alumina substrate , the surface of which is gold - plated to form electrodes .
the magnet layer is also made of a square alumina plate , which has holes for placing permanent magnets .
both the trap electrode and the magnet layer are squares of a side 11.5 mm .
the magnet layer is glued underneath the trap with a conductive epoxy to make these two squares overlap . then , the trap is glued on a ceramic pin grid array ( cpga ) mount
. structure of trap electrode with permanent magnets .
( a ) surface - electrode trap and magnet layer .
ions are trapped above the surface along the @xmath4 direction .
though the trap and the magnets are shown separately in this drawing for clarity , the eight permanent magnets are attached underneath the trap .
the labels a and b represent the trapping regions discussed in this work ( see text ) .
( b ) photograph of the trap chip .
( c ) photograph of the magnet layer placed underneath the trap chip ( b).,width=264 ] the trap design is based on a segmented linear surface - electrode trap ( fig . [ figureone ] ) .
an rf voltage is applied to the two rf electrodes for radial confinement .
a direct current ( dc ) electrode called center electrode is placed between the rf electrodes , and it is used for adjusting the height of the static potential minimum .
there are nine dc electrodes on either side of the rf electrodes for axial confinement .
ions can be trapped in any region except for the two outer regions , and they can also be shuttled between these regions by controlling the applied dc voltages . in the present work ,
we focus on two regions labeled a and b. these regions have different magnetic - field gradients when the magnet layer is attached .
the trap substrate is an alumina plate of area 11.5 mm@xmath5 square and thickness 0.2 mm .
electrodes are formed through gold plating using ti and pd as adhesive layers .
the thickness of the gold electrode is approximately 5 @xmath6 m .
the widths of the rf electrodes and center electrode are 300 @xmath6 m and 100 @xmath6 m , respectively .
all the dc electrodes for axial confinement are squares of dimensions 1 mm by 1 mm .
the spacing around the rf electrodes is 50 @xmath6 m , while the other spacing is 25@xmath6 m .
eight rectangular parallelepiped magnets are located underneath the trap electrode .
we choose the quantized axis in the @xmath7 direction and the trap axis in the @xmath4 direction .
our requirement is a large gradient of the @xmath7 component of magnetic field along the @xmath4 direction . to realize this ,
four of the magnets are aligned in the quadrupole configuration at the center of the layer .
ideally , at @xmath8 , no components exist along the @xmath9 and @xmath4 directions .
further , only the magnitudes of @xmath7 components exist and cross zero at @xmath10 .
the large magnetic field causes a large detuning in cooling laser frequency ; therefore , we place four other magnets in order to reduce the magnitude of the magnetic field outside the quadrupole configuration . in addition , these four magnets enable experiments under different field - gradient conditions .
smco magnets with dimensions of 1 mm @xmath11 1 mm @xmath11 2 mm are fit in a pattern made of alumina with a thickness of 1 mm , as shown in fig .
[ figureone ] ( c ) .
the magnet layer is glued underneath the trap chip . for a quantitative estimation , we numerically calculated the magnetic field using radia software package @xcite for the configuration of magnets shown in fig .
[ figuretwo](a ) .
figure [ figuretwo](b ) shows the @xmath7 component of the magnetic field along the @xmath4 direction ( horizontal line in fig.[figuretwo](a ) ) 350 @xmath6 m away from the top surface of the magnet layer .
owing to the symmetry of quadrupole configuration , both @xmath12 and @xmath13 are equal to zero in the ideal case . because of the existence of the outer four magnets ,
the magnetic field is reduced in the outer region .
the maximum field gradient is obtained at @xmath10 , and a smaller gradient is obtained around the other zero - crossing points . because the trap chip and the magnet layer are separated parts originally
, a shift between two squares may exist when they are glued , which results in the generation of an unwanted magnetic field at the trapped ions .
figure [ figuretwo ] ( c ) shows a calculation similar to that in ( b ) but with a shift between the trap chip and the magnetic layer of 100 @xmath6 m , as shown by the dotted horizontal line in fig .
[ figuretwo](a ) . at @xmath10
, @xmath13 becomes large , which changes the direction of the quantized axis .
@xmath12 also arises around @xmath14 mm . in the region @xmath151 mm ,
the unwanted magnetic - field components are close to zero .
results of calculation of magnetic field generated by the magnet layer .
( a ) configuration of eight magnets .
the arrows indicate magnetization vectors inside the magnets .
( b ) magnetic field along the @xmath4 direction ( horizontal line ) at @xmath9=150 @xmath6 m , which is 350 @xmath6 m from the top surface of the magnetic layer . in this ideal case ,
i.e. , when the trap and the magnet layer completely overlap , both @xmath9 and @xmath4 components are zero .
( c ) same calculation as ( b ) , with a shift between the trap and the magnet layer of 100 @xmath6 m assumed in the @xmath7 direction ; that is , ions are located along the dotted horizontal line in ( a ) and 350 @xmath6 m above the magnet - layer surface . in this case ,
@xmath12 and @xmath13 are not zero.,width=264 ] magnitude of the magnetic - field gradient versus the distance between the ion and the top surface of the magnet layer for trapping regions a and b.,width=264 ] the dependence of the calculated magnetic - field gradient on the distance between the trapped ion and the top surface of the magnet layer is shown in fig .
[ figurethree ] for the different trapping regions .
when an ion is trapped in region b at a height of 350 @xmath6 m from the magnet layer , a magnetic - field gradient greater than 100 t / m is estimated .
likewise , according to fig .
[ figurethree ] , when an ion is confined in region b at a height of 350 @xmath6 m , a magnetic - field gradient of approximately 38 t / m is estimated .
the trap on the cpga mount is placed in a vacuum chamber with a vacuum level of 10@xmath16 pa .
calcium ions are loaded by photo - ionization with a 423-nm laser for the @xmath17-@xmath18 transition and a 375-nm laser for ionization .
doppler cooling is conducted by the @xmath2-@xmath19 transition at 397 nm with an 866-nm laser for pumping back the ion from the @xmath20 state to the @xmath19 state .
the zeeman splitting is measured by using the quadrupole transition between the @xmath2 and @xmath3 states at 729 nm with a quenching laser connecting the @xmath3 and @xmath21 states at 854 nm .
all of the lasers for photo - ionization , doppler cooling , and quenching are introduced in the @xmath4 direction .
the 729-nm laser is introduced from the opposite side in the @xmath4 direction .
the polarization of the 729-nm laser is set to be in the @xmath7 direction so that only the @xmath22 transitions occur according to the selection rule @xcite .
coils exist for the compensation of unwanted magnetic field outside the chamber .
the fluorescence of ions is collimated by lenses set above the chamber and then divided into two paths with a beam splitter .
one is detected with a photomultiplier , while the other is detected with an image intensifier .
the trap was typically driven by rf voltages of 112 @xmath23 at 22.2 mhz .
for axial confinement , de voltages ranging from 3 v to 19 v were applied .
in order to evaluate the magnetic - field gradient , we measured the zeeman splitting of the @xmath2 - @xmath3 transition at several points along the @xmath4 direction in region b because the second - largest field gradient is expected without a large offset field of @xmath12 and @xmath13 . after trapping a single ca@xmath1 ion in region b ,
a stray dc field was compensated for to minimize the excess micromotion of ions .
in addition , we compensated for as much of the offset field of @xmath24 and @xmath25 as possible by using external coils attached outside the vacuum chamber .
we then measured the splitting between two zeeman components of the @xmath2 and @xmath3 transition at several points along the z axis .
figure [ figurefour ] shows the spectra of the @xmath26 transition at 729 nm obtained at different positions of the trap . among the four @xmath22 transitions shown in the top panel in fig.[figurefour ] , we focus on the spectra of two , @xmath27 and @xmath28 , which are indicated by vertical arrows , to measure the zeeman splitting of the @xmath2 state .
the ion position is changed by the dc control voltages .
figure [ figurefive ] shows the dependence of the magnetic - field magnitude estimated from the observed zeeman splitting on the ion position .
the error is mainly due to the width of the spectrum . by fitting a linear function , we estimated a magnetic - field gradient of 36 t / m .
spectra of the @xmath26 transition of a single @xmath0ca@xmath1 .
@xmath29 represents the distance from the original ion position.,width=302 ] magnitude of the magnetic field estimated from the zeeman splitting at different positions of a single trapped ion.,width=264 ]
we measured the field gradient in region b , where the second - largest magnetic - field gradient is expected .
we also attempted to perform the experiments in region a , where the largest magnetic - field gradient is estimated ; however , the fluorescence from ca@xmath1 ion could not be observed .
we also attempted moving the ion from the region next to a to region a. we could shuttle the ion , however , it was not possible to monitor the ion image while the ion remained in region a. we consider that this is due to the large offset field originating from a shift between the trap electrode and the magnet layer , which causes a large detuning in the transition at 397 nm .
we measured the size of the shift between the trap and the magnet layer by using a laser microscope and found that the size of the shift was on the order of several tens of micrometers .
the adjustment of the ion position by dc voltages is possible ; however , after the micromotion compensation @xcite , we can not apply an additional dc voltage freely . according to the calculation , a shift as high as 50 @xmath6 m can produce an offset field of approximately 5 mt in the @xmath4 direction at @xmath10 .
it is difficult to cancel such a high offset field with our present external coils attached outside the vacuum chamber . to overcome the offset - field problem due to the shift between the trap electrode and the magnet layer , it is preferable to fabricate a magnet layer integrated with the trap electrode .
other possible cause of excess offset magnetic field is that the magnitude of the magnetic field generated by each permanent magnet is uneven . to overcome this problem
, we need to adjust the positional relation between the rf node and magnetic field .
one possibility is to set the magnet layer on a movable stage inside the vacuum chamber that can be three - dimensionally moved with respect to the position of ions .
alternatively , it is possible to move the rf node by introducing an additional rf field to a trap electrode @xcite .
the obtained field gradient of 36 t / m at a distance of approximately 350 @xmath6 m from the magnet layer is in good agreement with the calculation . from this result
, we can infer that a field gradient of 100 t / m is generated in region a.
we have demonstrated a surface - electrode trap with smco magnets arranged in the quadrupole configuration underneath the trap electrode . by utilizing the great advantage of the permanent magnets , which do not require any heat management and
can be placed as close to ions as possible , a large magnetic - field gradient can be generated .
a field gradient as large as 36 t / m has been estimated by measuring the zeeman splitting of a single trapped @xmath0ca@xmath1 ion at several positions , which is in good agreement with the calculation . from this result
, we can infer that a field gradient of 100 t / m is generated in region a. a large field gradient is necessary to implement a sufficient state - dependent force , which is vital for quantum simulation and/or quantum gate operation using radio - frequency or microwave radiation .
it is important to assemble the trap electrode and the magnet layer without any shift between them .
the fabrication of a magnet layer integrated with the trap electrode would be a solution to overcome this problem .
a field gradient on the order of 100 t / m would be possible with the presented method .
we are grateful to yasuhiro tokura for helpful advice on the numerical calculation of magnetic fields . | we report on a surface - electrode trap with smco magnets arranged in a quadrupole configuration underneath the trap electrode .
because the distance between the magnets and the trapped ions can be as little as several hundred micrometers , a large magnetic field is produced without any heat management .
the magnetic - field gradient was measured using the zeeman splitting of a single trapped @xmath0ca@xmath1 ion at several positions , and a field gradient of 36 t / m was obtained .
such a field gradient is useful for the generation of a state - dependent force , which is important for quantum simulation and/or quantum gate operation using radio - frequency or microwave radiation .
_ keywords _ : ion trap , magnetic - field gradient , quantum simulation |
several studies have found variable stars in fornax or have detected variability ( light et al .
1986 , demers & irwin 1987 , buonanno et al . 1985 ,
stetson et al . 1998 ) .
however they all suffered from some limitations .
for instance @xcite cover a large area with deep photometry but they do nt have enough epochs to determine periods for their candidate variables .
@xcite did have a lot of data but they covered a very small area ; they have light curves for only two variable stars .
@xcite covered the whole galaxy and had @xmath3 epochs but their photometry did nt go deep enough to detect rr lyrae . for these reasons , there are few variable stars known in fornax with published periods and light curves .
we thus thought that it would be worthwhile to survey a fair fraction of this galaxy and look for variable stars , in particular rr lyrae stars .
imaging observations of fornax were made with the 1 m telescope at siding spring observatory and with the 1.3 m telescope at mount stromlo observatory .
roughly half the epochs of observation came from each telescope .
table [ tbl_dates ] lists the dates of observation for each field .
the two sets of observations are described separately below .
the part of the data obtained with the 1 m telescope at siding spring observatory was taken during two runs of a few consecutive nights .
we used a site @xmath4 ccd that gives a field of view of @xmath5 ( 0.602 arcsecond per pixel ) .
four fields have been observed in @xmath6 and @xmath7 filters , covering a square on the sky .
the four fields were centered approximately on @xmath8 , @xmath9 , @xmath10 , @xmath11 ( thus field 1 is north - east , field 2 is north - west , field 3 is south - east and field 4 is south - west ) .
the total area observed is almost half a square degree , covering the central regions of the galaxy .
each field has been observed between one and four times per night , with a typical exposure time of 10 minutes , resulting in about 12 to 15 exposures per field .
our average seeing is 2 , with the best value being 1.5 . during two nights
we also observed standard stars taken from the list of @xcite .
the data have been processed in a standard way using iraf procedures .
we performed aperture photometry for the standard stars and solved the transformation equations between the instrumental and standard systems .
we obtained @xmath12 where @xmath13 is the magnitude of the standard star , @xmath14 is the actually observed magnitude , @xmath15 is the airmass and @xmath16 is the color of the star .
the residuals were 0.010 @xmath17 in @xmath6 and 0.011 @xmath17 in @xmath18 .
the extinction coefficients are close to the average values found for sso @xcite and the color terms are virtually identical to those found by @xcite for the particular camera and filter set that we used .
the photometry for each fornax frame was extracted using dophot @xcite in fixed position mode .
we first created a `` template '' frame for each field .
each template is the average of 4 - 6 good seeing images .
photometry was done on this deep template and the star list of the template was used as input for the photometry of each program frame .
then all images were put on the same zero - point ( still in the instrumental system ) .
we applied an aperture correction to these magnitudes and eqs ( [ eq_mv ] ) and ( [ eq_mi ] ) were inverted to provide calibrated photometry for each frame .
we estimate that our absolute zero point is accurate to @xmath19 .
we obtained 20 images with the wide - field camera mounted on the mount stromlo observatory 50 telescope ( mso ) .
this camera was used by the macho project and is a mosaic of four 2k@xmath202k ccd with a pixel scale of 0.628/pixel , covering approximately @xmath21 .
a feature of this camera is a dichroic mirror that allows to obtain simultaneous measurements in two passbands , @xmath22 and @xmath23 , corresponding ( very roughly ) to @xmath6 and @xmath24 .
we refer the reader to @xcite and @xcite for descriptions of the camera system and filters .
the exposure time was 10 minutes for each image .
the typical seeing was 2.2 .
photometry was also performed in fixed - position mode and the frames were then put on a common zero point .
since the passbands are non - standard , we needed to calibrate @xmath22 and @xmath23 to @xmath6 and @xmath18 .
for this we could use our already calibrated photometry from the 1 m sso telescope .
the passbands @xmath22 and @xmath23 are quite broad but they have been calibrated to @xmath6 and @xmath24 ( bessell & germany 1999 , alcock et al .
1999 ) .
the areas covered with each telescope / instrument combination are virtually identical , thus we have @xmath1 and @xmath25 photometry for most stars .
figure [ fig_vrmvi ] illustrates the procedure we used to calibrate the color @xmath26 to @xmath27 . for each chip , we constructed plots like fig .
[ fig_vrmvi ] and plots of @xmath28 vs @xmath26 .
the `` knee '' at @xmath29 can already be found in @xcite in a @xmath30 vs @xmath31 diagram ( their fig .
this knee also happens when we use @xmath27 since there is an almost linear relation between @xmath27 and @xmath32 .
we then made separate fits for stars bluer and redder than @xmath33 .
for blue stars ( @xmath34 ) a typical calibration would be @xmath35 and for red stars
( @xmath36 ) @xmath37 from chip to chip there were slight differences in the coefficients , amounting to a few percents at most .
our separate chip - by - chip calibration might not have been warranted ( the approach followed by bessell & germany 1999 ) .
however , we certainly do not lose accuracy in the process , it is only a little more cumbersome .
after the calibration of the whole data set an astrometric calibration was done , using the usno - a2.0 catalog @xcite giving residuals between 0.3 and 0.5 arcseconds . with calibrated photometry in hand , the last step is to apply a reddening correction .
given the high galactic latitude of fornax ( @xmath38 ) , the average reddening should be very low .
we thus used the maps of @xcite to determine the reddening to every star in our data set . the average @xmath39 is around 0.025 mag .
we used the method proposed by @xcite .
it makes use of the fact that when a star is brighter ( or fainter ) than average in one passband it is also brighter ( fainter ) than average in another passband .
the variability index @xmath40 has been computed for each star , and fig [ fig_iws ] presents the results .
fig [ fig_cmdvar ] shows the color - magnitude diagram with the variables we have identified .
fornax has five globular clusters @xcite , two of which are in our field ( cluster 2 and cluster 4 ) .
however it was impossible to find variables in these clusters , because of the exceedingly high density of stars .
while it is possible to resolve some stars in cluster 2 ( no star could be resolved in cluster 4 ) , the effect of a varying seeing made the photometry extremely unreliable , although we found one candidate long period variable in cluster 2 .
detection of rr lyrae in fornax s globular clusters can only be done with a much better spatial resolution than we have .
the probability of detecting a variable star depends on a number of things , an important one being the actual time distribution of the observations , since this can affect the detection of some periods more than others .
we carried out numerical simulations ( much in the spirit of saha & hoessel , 1990 ) to estimate the impact that our sampling may have on the discovery of variable stars .
for a star of period @xmath41 and initial phase @xmath42 ( phase at the time of the first observation ) , we can compute the phase at each of our observations .
we then determine which of the following criteria are fulfilled : 1 .
at least 3 observations between phases 0 and 0.2 ( @xmath43 corresponds to maximum light ) 2 .
at least 2 phases between 0.2 and 0.5 ( descending light ) 3 .
at least 3 phases between 0.5 and 0.8 ( around minimum light ) 4 . at least 15 measurements in total for all periods between 0.1 and 50 days , and for all initial phases ,
we simply count the number of observations satisfying each criterion , after having attributed a weight to each observation : we estimate that the _ photometric _ detection efficiency of the sso 40
data is 0.8 , and 0.4 for the mso 50 data .
this means that , for instance , four actual sso 40
observations are nedded to satisfy criterion 1 . for each period
we then count the number of phases that fulfill these criteria .
after appropriate normalization , we obtain a period spectrum which is simply the probability of detecting a star as a function of period ( see fig [ fig_effic ] ) .
we selected all stars in a broad region around the horizontal branch in order to look for rr lyrae variables .
specifically the criteria were @xmath44 , @xmath45 and @xmath46 .
our data are somewhat inhomogeneous in the sense that that sso 40 data are of better quality than the mso 50 data . in the search for rr lyrae
we thus used only sso data when computing the @xmath40 index .
the value @xmath47 has been chosen as a compromise between finding the largest number of genuine variables while minimizing the number of false detections .
it can be seen from fig [ fig_iws ] that the bulk of the stars are below @xmath47 . applying the criteria given above yielded 851 objects .
we used the phase - dispersion minimization algorithm ( stellingwerf , 1978 ) to look for periodicity in @xmath6-band data .
a rapid inspection of some light curves revealed that for a fairly large number of candidates no period could give a reasonable light curve .
this is because the magnitude of a rr lyrae variable is close to the detection limit of our photometry .
we thus decided to apply a conservative approach and to select the best period by examining each light curve .
this led to rejecting a fairly large number of candidates .
our goal was to produce a reasonably clean sample of rr lyrae , not a complete one . among the stars rejected
, a significant number may be genuine rr lyrae .
the main reason to reject a candidate variable was the total number of measurements per star . among the rejected candidates , 48% had less than 25 measurements whereas 75% of the confirmed rr lyrae have more than 25 data points .
there is no obvious correlation between the value of the variability index @xmath40 and the probability of finding a period .
this is because we used @xmath48 epochs to construct the index @xmath40 ( only sso 40 data ) .
some stars may show a large @xmath40 but may have only 15 measurements , making it difficult to find a reliable period .
the second reason for rejecting a variable is aliasing .
for some stars it was impossible to decide between two periods .
in particular , for a given frequency @xmath49 , the light curves for @xmath50 or @xmath51 would be almost as convincing as @xmath49 . in the end
, this selection yielded 515 rr lyrae stars for which one single frequency could be clearly identified , hence 60% of our candidates turned out to be `` good '' rr lyrae variables .
they are listed in table [ tbl_rrl ] where we give the phase - weighted average magnitudes and associated uncertainties .
some light curves are shown in fig .
[ fig_lcrr ] .
when considering the @xmath18-band light curves in this figure , it is quite remarkable that the welch - stetson method still works , even though the variability signal is extremely weak in the @xmath18 band .
this shows that the method is very powerful even at faint magnitudes .
one has to bear in mind that the number of variables is likely to be a lower limit to the total number of rr lyrae stars in fornax . in particular ,
a consequence of the quality of our data ( the average measurement error for a typical rr lyrae variable is @xmath52 ) is that our survey is probably biased against rrc variables , which have a low amplitude
. this will affect the relative number of rrab and rrc variables but the incompleteness should not depend on period . with better photometric data it will be possible to obtain good light curves for many more rr lyrae .
the period histogram is shown on fig .
[ fig_rrhistop ] .
there are two peaks , one at @xmath53 days and one at @xmath54 days .
there seems to be a gap between @xmath55 and @xmath56 .
the probability of detecting variables with periods around @xmath57 is very low but for periods @xmath58 our detection probability is always larger than 66% of the maximum ( see fig .
[ fig_rrhistop ] ) so we should be able to discover variables with periods between @xmath55 and @xmath56 , if they exist .
we thus interpret this gap as the transition period from rrab to rrc .
the shortest period for rrab type is thus @xmath56 . given that our data are not good enough to determine the type of variable on the basis of the light curve , we _ defined _ rrab type variables as the stars having periods larger than 0.47 days , and rrc as having periods shorter than 0.47 days .
the number of stars in each category is @xmath59 and @xmath60 .
we then determined the average periods @xmath61 and @xmath62 .
the longest period for rrab variables is 0.737@xmath63 . in galactic globular clusters ,
the average period of ab type rr lyrae variables @xmath64 is either near @xmath65 or near @xmath66 ( oosterhoff class i and ii respectively ) .
the surprising fact is that dwarf spheroidal galaxies do not fall in either of these categories . from table 7 in @xcite
, one sees that all dwarf spheroidal galaxies are intermediate between oo i and oo ii categories . adding the result of @xcite for leo ii ( @xmath67 ) and the present value of @xmath68 for fornax only reinforces this statement .
the average properties of rr lyrae variables are related to their metallicity .
there are relations between several characteristic periods ( longest , shortest , average ) and the metal content @xcite . for the shortest period one
has @xmath69 } ) - 0.500 \label{eq_logpab}\end{aligned}\ ] ] which yields @xmath2 } = -1.5 $ ] ( for @xmath70 ) . through the relation @xmath71 } ) - 0.389 \label{eq_logp}\end{aligned}\ ] ] the average period yields @xmath2 } = -1.7 $ ] .
the longest period is related to [ fe / h ] via @xmath72 } ) - 0.280 \label{eq_logplong}\end{aligned}\ ] ] which gives @xmath2 } = -1.64 $ ] .
all these values of [ fe / h ] are on the butler - blanco metallicity scale ( used by sandage ) which is more metal - rich than the widely used scale of @xcite by 0.2 dex .
when considering in particular the average period @xmath64 , there is some scatter in this relation , about 0.2 dex ( see siegel & majewski 2000 , their figure 6 , which also displays the other eight dwarf spheroidal galaxies ) .
our estimates of the metal abundance cluster around @xmath73 , we will then take as a final value @xmath2 } = -1.6 \pm 0.2 $ ] .
in contrast with this small dispersion in [ fe / h ] is the fairly large dispersion in average magnitudes for rr lyrae stars ( @xmath74 ) .
this is significantly larger than what is observed in other dwarf spheroidal galaxies [ e.g. @xmath75 in leo ii
@xcite , @xmath76 in sculptor @xcite ] .
it seems likely that photometric scatter is reponsible for the dispersion in @xmath77 .
we found a number of variables slightly or significantly brighter than the horizontal branch , most of which are anomalous cepheids ( acs ) or population ii cepheids .
their magnitudes and periods are given in table [ tab_acs ] .
we found 6 pop ii cepheids ( one unsure ) , 17 anomalous cepheids ( one unsure ) and one blue variable
whose nature is unclear .
the position of a variable in the period - luminosity diagram ( fig .
[ fig_pl ] ) is determined by its metallicity and pulsation mode ( e.g. nemec et al 1994 ) .
fornax has a large range in [ fe / h ] and it has had a complex star formation history @xcite .
this means that cepheids in this galaxy can have a range of ages and/or metallicities .
unfortunately , in the absence of any information on [ fe / h ] it is impossible to assign a pulsation mode to these variables . by looking at fig [ fig_pl ]
it is clear however that there must be a range of metallicity among these stars .
some of the extreme stars in fig .
[ fig_pl ] could be metal - poor and overtone pulsators while others are metal - rich and fundamental pulsators .
there is a surprising correlation between the specific frequency of anomalous cepheids and the luminosity of the dwarf spheroidal galaxies @xcite .
according to this relation , about a dozen anomalous cepheids should have been found .
we found 17 , fairly close to the prediction and not large ( nor small ) enough to destroy the correlation found by @xcite .
the nature of anomalous cepheids is unclear .
they are found in systems where the most massive stars still burning nuclear fuel are @xmath78 , however the masses of anomalous cepheids are around @xmath79 ( e.g. wallerstein & cox 1984 , bono et al .
1997 ) the two explanations usually put forward to explain their large masses are that they are intermediate - age stars or that they are the product of a binary star merging ( the two component having the mass of a turn - off star ) .
fornax has an important intermediate - age population so the first hypothesis is perfectly plausible , however it is not easy to rule out the second possibility .
a remarkable fact is that , among the nine dwarf galaxies satellites of the milky way , fornax is the only one containing pop .
ii cepheids .
these stars are in a short - lived evolutionary stage and fornax is probably the only dwarf galaxy that is massive enough to have a few of these stars .
there has already been a wide - field survey for long period variables ( lpvs ) in fornax @xcite .
they covered the whole galaxy and found 30 variables .
it was based on photographic plates obtained at the uk schmidt telescope and they acknowledge the fact that `` a certain number of bright variables must have been missed ... because of blended images '' .
it thus seemed worthwhile to look specifically for bright red variables showing a long - term trend , even though our sampling is far from ideal for these stars .
@xcite presented a list of very red stars in fornax .
based on two @xmath6 measurements separated by 132 days , they could flag some of these stars as candidate variables .
we also tried to recover objects from that paper .
eighty five stars turn out to be reasonably good candidates .
they are given in table [ tbl_lpv ] . in this table
, a question mark means that the variability is not certain ; this happens particularly for crowded stars .
given that our data do not allow to obtain periods for such objects , the classification as lpvs is only tentative .
however , given their magnitudes and colors , most objects in table [ tbl_lpv ] are good candidates . in this table
we also indicate the numbers given by @xcite in their table 1 for red stars .
we have data for 19 variables listed by @xcite , 3 of which appear non variable in our data simply because we do nt have enough measurements .
the remaining 11 stars in the list of demers & irwin are not in our survey area . in addition
we find 38 good candidate lpvs and 28 possible variables ( indicated with a ? in table [ tbl_lpv ] ) . in fig .
[ fig_lpv ] we show two examples of lpvs .
there are a few dozens of red giants ( @xmath80 ) showing some variability ( i.e. with @xmath81 ) , most with small amplitude ( less than 0.1 @xmath17 ) .
variability along the red giant branch has been detected in a number of cases in galactic field and globular clusters giants ( e.g. smith & dupree 1988 and references therein ) , however it is not clear whether this is related to pulsation , spots , or any other mechanism .
the typical periods of these objects are likely to be in the range of days to weeks , where our ability to find a period is seriously affected by the sampling ( see fig [ fig_effic ] ) .
we have presented a survey of @xmath82 square degree covering the central region of the fornax dwarf galaxy . we found and determined periods for 17 anomalous cepheids , 6 population ii cepheids and more than 500 rr lyrae .
in addition , 85 lpv candidates were identified .
it is almost certain that there are more rr lyrae in this galaxy : the accuracy of our data do not allow us to produce a complete catalog of rr lyrae .
the average metal abundance of rr lyrae is @xmath2}_{rr } \simeq -1.6 \pm 0.2 $ ] .
we can use the known distance to fornax ( @xmath83 with the tip of the red giant branch , bersier 2000 ) to estimate the brightness of rr lyrae . with an average magnitude of @xmath84 , one obtains @xmath85 .
this is intermediate between other estimates of the magnitude of rr lyrae stars .
@xcite found @xmath86 for [ fe / h ] = -1.7 , while from @xcite one would get @xmath87 .
we thank the director of mt stromlo & siding spring observatories , jeremy mould , for the allocation of director s discretionary time on the mso 50 telescope .
we also thank the referee for a swift and constructive report that helped us improve this paper .
db acknowledges partial support from the swiss national science foundation ( grant 8220 - 050332 ) . this work has also been supported by nsf grant ast-9979812 .
alcock , c. et al .
( the macho collaboration ) 1999 , , 111,1539 bersier , d. 2000 , , 543 , l23 bessell , m. s. , germany , l. m. 1999 , , 111 , 1421 demers , s. , irwin , m. j. 1987 , , 226 , 943 demers , s. , kunkel , w.e . 1979 , , 91 , 761 gould , a. , popowski , p. , 1998
, , 508 , 844 hodge , p. w 1961 , , 66 , 83 kaluzny , j. , kubiak , m. , szymaski , m. , udalski , a. , krzemiski , w. , mateo , m 1995 , , 112 , 407 landolt , a. u. , 1992 , , 104 , 340 light , r. m. , armandroff , t. a. , zinn , r. 1986 , , 92 , 43 mateo , m. , fischer , p. , krzemiski , w. 1995 , , 110 , 2166 monet , d. , et al . 1998 ,
usno - sa2.0 , ( washington : us naval obs . ) nemec , j. m. , nemec , a. f. l. , lutz , t. e. 1994 , 108 , 222 saha , a. , hoessel , j , g .
1990 , , 99 , 97 sandage , a. 1993 , , 106 , 687 sandage , a. 1993 , , 106 , 703 saviane , i , held , e. v. , bertelli , g. 2000 , 355 , 56 schechter , p. l. , mateo , m. , saha , a. , 1993 , , 105 , 1342 schlegel , d. j. , finkbeiner , d. p. , davis , m. 1998 , , 500 , 252 siegel , m. h. , majewski , s. r. 2000 , , 120 , 284 smith , g. h. , dupree , a. k. , 1988 , , 95 , 1547 stellingwerf , r. f. 1978 , , 224 , 953 stetson , p. b. , hesser , j. e. , smecker - hane , t. a. 1998 , , 110 , 533 stubbs , c. , et al .
( the macho collaboration ) 1993 , proc .
spie , 1900 , 192 sung , h. , bessell , m. s. , lee , s .- w . 1998 , , 115 , 734 sung , h. , bessell , m. s. 2000 , pasa , 17 , 244 udalski , a. , kubiak , m. , szymanski , m. , kaluzny , j. , mateo , m. , krzeminski , w. , 1994 , aca , v.44 , 317 walker , a.r .
1994 , , 108 , 555 wallerstein , g. , cox , a.n . 1984 , , 96 , 677 welch , d. l. , stetson , p.b .
1993 , , 105 , 1813 zinn , r , west , m. j. 1984 , , 55 , 45 l l l l field & jd@xmath88 & jd@xmath89 & observatory + 1 & 652.26667 & 652.26667 & mso + 1 & 657.26976 & 657.26976 & mso + 1 & 658.27549 & 658.27549 & mso + 1 & 659.28311 & 659.28311 & mso + 1 & 660.25075 & 660.25075 & mso + 1 & 661.23904 & 661.23904 & mso + 1 & 662.23573 & 662.23573 & mso + 1 & 663.23345 & 663.23345 & mso + 1 & 664.23512 & 664.23512 & mso + l r c c r c c r c c name & @xmath90 & @xmath6 & @xmath91 & @xmath92 & @xmath18 & @xmath93 & @xmath40 & period ( days ) & @xmath39 + fbw j023749.8 - 342427 & 24 & 21.389 & 0.027 & 24 & 20.755 & 0.036 & 1.422 & 0.54365 & 0.028 + fbw j023750.8 - 342736 & 32 & 21.346 & 0.021 & 31 & 20.744 & 0.035 & 2.741 & 0.58252 & 0.030 + fbw j023751.9 - 342620 & 32 & 21.419 & 0.023 & 32 & 20.887 & 0.033 & 1.139 & 0.58052 & 0.029 + fbw j023752.8 - 344615 & 32 & 21.469 & 0.029 & 32 & 20.560 & 0.033 & 2.005 & 0.16459 &
0.034 + fbw j023753.4 - 345215 & 28 & 21.344 & 0.026 & 28 & 20.781 & 0.035 & 1.584 & 0.59260 & 0.035 + fbw j023755.0 - 343726 & 30 & 21.401 & 0.026 & 29 & 20.911 & 0.035 & 1.506 & 0.40322 &
0.035 + fbw j023756.0 - 342732 & 32 & 21.468 & 0.020 & 32 & 21.101 & 0.036 & 5.965 & 0.33797 & 0.030 + fbw j023756.1 - 344332 & 31 & 21.544 & 0.022 & 31 & 20.849 & 0.036 & 2.666 & 0.51202 & 0.034 + fbw j023756.6 - 343507 & 32 & 21.392 & 0.022 & 32 & 20.766 & 0.031 & 2.208 & 0.54362 & 0.035 + fbw j023757.0 - 342201 & 32 & 21.455 & 0.021 & 32 & 20.838 & 0.032 & 2.497 & 0.56046 & 0.028 + l
c c c c c r l name & @xmath6 & @xmath27 & @xmath91 & @xmath93 & @xmath39 & period & class + fbw j023803.9 - 343822 & 19.478 & 0.483 & 0.007 & 0.012 & 0.036 & 15.2283 & p2c + fbw j023811.9 - 344023 & 20.651 & 1.087 & 0.015 & 0.021 & 0.036 & 1.21327 & p2c + fbw j023843.0 - 344825 & 19.801 & 0.558 & 0.016 & 0.031 & 0.031 & 1.04503 & ac + fbw j023852.4 - 343012 & 19.905 & 0.602 & 0.007 & 0.011 & 0.027 & 1.24968 & ac +
fbw j023853.4 - 343048 & 19.712 & 0.564 & 0.006 & 0.010 & 0.027 & 2.57976 & p2c + fbw j023907.1 - 343316 & 20.786 & 0.774 & 0.017 & 0.025 & 0.026 & 0.50812 & ac + fbw j023926.9 - 342806 & 20.272 & 0.667 & 0.010 & 0.016 & 0.025 & 2.08323 & p2c + fbw j023926.8 - 343122 & 20.828 & 0.673 & 0.017 & 0.028 & 0.024 & 0.50398 & ac + fbw j023927.0 - 342427 & 21.052 & 0.082 & 0.019 & 0.042 & 0.026 & 0.56965 & blue + fbw j023937.7 - 343621 & 20.645 & 0.725 & 0.015 & 0.024 & 0.025 & 0.546 & ac + fbw j023941.5 - 343125 & 20.616 & 0.475 & 0.012 & 0.022 & 0.024 & 0.57345 & ac + fbw j023945.2 - 344314 & 21.230 & 0.481 & 0.019 & 0.037 & 0.024 & 1.14259 & p2c ? + fbw j023946.2 - 343001 & 20.568 & 0.786 & 0.014 & 0.023 & 0.023 & 0.92161 & ac + fbw j023952.5 - 343340 & 20.252 & 0.904 & 0.014 & 0.023 & 0.024 & 1.31071 & ac + fbw j023953.5 - 342433 & 20.846 & 0.656 & 0.015 & 0.026 & 0.022 & 0.61144 & ac + fbw j023954.8 - 343601 & 21.022 & 0.478 & 0.016 & 0.031 & 0.025 & 0.57442 & ac + fbw j024000.9 - 341846 & 20.828 & 0.476 & 0.015 & 0.025 & 0.023 & 0.41592 & ac + fbw j024002.7 - 341928 & 20.362 & 0.849 & 0.009 & 0.015 & 0.022 & 0.53299 & ac + fbw j024010.2 - 343212 & 20.624 & 0.412 & 0.013 & 0.025 & 0.022 & 1.33531 & p2c + fbw j024016.9 - 343641 & 20.840 & 0.430 & 0.014 & 0.030 & 0.025 & 0.53294 & ac + fbw j024017.4 - 342024 & 19.994 & 0.608 & 0.007 & 0.011 & 0.021 & 1.19847 & ac + fbw j024022.8 - 342802 & 20.480 & 0.526 & 0.011 & 0.019 & 0.020 & 0.83757 & ac + fbw j024050.2 - 344335 & 20.826 & 0.729 & 0.015 & 0.024 & 0.024 & 0.50578 & ac + fbw j024058.3 - 344552 & 20.786 & 0.663 & 0.018 & 0.028 & 0.024 & 0.48106 & ac + l r c c
r c c c l l name & @xmath90 & @xmath94 & @xmath95 & @xmath92 & @xmath96 & @xmath97 & @xmath39 & other i d & note + fbwj023800.0 - 344422 & 32 & 18.562 & 0.021 & 32 & 1.860 & 0.026 & 0.034 & & + fbwj023806.1 - 343119 & 30 & 20.233 & 0.065 & 30 & 3.765 & 0.070 & 0.033 & & + fbwj023812.6 - 345606 & 32 & 18.504 & 0.022 & 32 & 2.291 & 0.026 & 0.033 & dk1 , di30 & + fbwj023817.1 - 342548 & 32 & 18.591 & 0.023 & 32 & 2.314 & 0.030 & 0.029 & s3 , di13 & + fbwj023821.3 - 343618 & 32 & 18.335 & 0.021 & 32 & 1.942 & 0.026 & 0.034 & & ? + fbwj023822.0 - 345213 & 32 & 17.960 & 0.014 & 32 & 1.577 & 0.019 & 0.032 & & + fbwj023822.6 - 343804 & 32 & 18.731 & 0.023 & 31 & 2.296 & 0.029 & 0.035 & dk15 & + fbwj023826.3 - 342533 & 30 & 18.451 & 0.017 & 30 & 1.649 & 0.022 & 0.029 & & + fbwj023830.2 - 344503 & 28 & 18.324 & 0.017 & 28 & 1.875 & 0.021 & 0.033 & & + fbwj023833.4 - 342824 & 32 & 18.403 & 0.018 & 32 & 1.717 & 0.025 & 0.029 & & ? | we present a search for variable stars in the fornax dwarf galaxy covering an area of 1/2 a square degree .
we have @xmath0 epochs of @xmath1 data .
we found and determined periods for more than 500 rr lyrae , 17 anomalous cepheids , 6 population ii cepheids .
in addition we have 85 candidate long period variables , the majority of which were previously unknown .
we estimated that the average metal abundance of rr lyrae stars is @xmath2 } \simeq -1.6 $ ] dex . |
recent advances in studying the world - volume theory of m2-branes @xcite have allowed new perspectives on the ads / cft correspondence . indeed , the three - dimensional theory on the m2-branes is now understood as an ordinary gauge theory of chern - simons ( cs ) type @xcite .
the investigation of ads@xmath3/cft@xmath4 subsequently took flight , ever augmenting our arsenal of dual pairs of theories .
one strand of development has been the extension of the elaborate structure of the ads@xmath5/cft@xmath3 situation wherein the d3-brane world - volume gauge theory and the corresponding calabi - yau cone over sasaki - einstein 5-folds have been probed in great detail over the past decade
. of particular interest had been the cases where the calabi - yau threefold admits toric description . there , a rich tapestry , enabled by the abundant techniques of toric geometry , had been woven over such themes as toric duality @xcite , dimer models and brane tilings @xcite . along this vein , the parallel story for m2-branes probing toric calabi - yau 4-fold singularities met with rapid progress @xcite , addressing such issues as tiling or its 3-dimensional counter - part of crystal models , toric duality , as well as partition functions in the light of the plethystic programme @xcite etc .
thus inspired , especially by the fortitude to attack the general singularity @xcite , it is expedient for us to take a synthetic approach .
the space of toric singularities of calabi - yau 4-folds is largely unchartered ; nevertheless , we can gradually and systematically list , starting from the simplest imaginable , the possible @xmath6-dimensional quiver chern - simons theories .
these theories are , of course , infinite in number ; however , we will see how the number of nodes in the quiver and the number of terms in the superpotential can be used as order parametres to begin a classification . even at low numbers
, we will encounter many highly non - trivial theories .
furthermore , we will establish generating functions which counts the number of inequivalent theories at each level of our enumeration .
remarkably , we can geometrically interpret these functions .
thus , let us we embark on a journey through the complex _ terra incognita _ of new graphs and associated superpotentials , guided by our experience with the techniques of @xmath7-dimensional @xmath8 gauge theory , _ mutatis mutandis _ , and slowly probe the dual pairs of toric geometries and cs theories .
a wealth of interesting properties will be encountered .
our taxonomic study is organised as follows .
we begin , in section [ s : mcs ] , with a brief review of the computation of the moduli space of quiver chern - simons theories in @xmath6-dimensions , culminating in an algorithmic flow - chart which generalises ( and encompasses ) the `` forward - algorithm '' of @xcite to the present case .
next , in section [ s : tax ] , we begin our systematic investigation of what toric chern - simons quiver theories can exist .
we start with two terms in the superpotential satisfying the toric condition and exhaustively present what possibilities could arise , using the number of nodes @xmath9 and the number of fields @xmath10 as order parametres .
we then explicitly show how the moduli spaces of these theories are families of toric calabi - yau 4-folds , indexed by the chern - simons levels .
we find a new manifestation of toric duality . in section [ s : dimer ]
, we show how all these theories admit a dimer - model , or periodic - planar - tiling description , contrary to what one would naively expect . finally ,
stepping back to have a bird s eye view , we find a generating function in section [ s : count ] which counts all the theories encountered . we conclude with prospect in section [ s : conc ] .
in this section , we briefly outline the computation of moduli space of the theories of our interest : @xmath6-dimensional quiver chern - simons theories with @xmath11 supersymmetry .
we point out that henceforth we are interested in the classical mesonic moduli space , which we will denote as @xmath12 .
the theories are with four supersymmetries and the gauge groups have no kinetic terms but instead have cs terms , moreover , they have product gauge groups together with bifundamental and adjoint matter . the computation of @xmath12 is a direct generalisation of the so - called `` * forward algorithm * '' for @xmath7-dimensional @xmath8 gauge theories @xcite .
let the quiver cs theory have gauge group consisting of @xmath9 factors , and a total of @xmath10 fields , we then have the action , written in @xmath11 superspace notation : @xmath13 where @xmath14 indexes the factors in the gauge group , @xmath15 are the superfields accordingly charged , @xmath16 are the gauge multiplets , @xmath17 is the superspace derivative , @xmath18 is the superpotential and @xmath19 are the chern - simons levels which are integers ; an overall trace is implicit since all the fields are matrix - valued .
we take the following two constraints on the cs levels , the reader is referred to @xcite for details : @xmath20 the classical mesonic moduli space @xmath12 is determined by the following equations @xmath21 & = & 4k_a\sigma_a \\ \label{df } \sigma_a x_{ab } - x_{ab } \sigma_b & = & 0 \ , \end{aligned}\ ] ] where @xmath22 is the scalar component of @xmath16 . indeed , this is in analogy to the f - term and d - term equations of @xmath8 gauge theories in @xmath7-dimensions , with the last equation being a new addition .
we are particularly interested in the case when @xmath12 is a toric variety where the forward algorithm conveniently uses the combinatorial power of lattice geometry and toric cones @xcite .
this is the abelian case where the gauge group is simply @xmath23 .
physically , @xmath12 is a toric calabi - yau 4-fold transverse to an m2-brane on whose world - volume lives the @xmath6-dimensional cs theory . for a stack of @xmath24 parallel , coincident m - branes ,
the moduli space is the @xmath24-th symmetrised product of ( or more precisely the @xmath24-th hilbert scheme of points on ) the 4-fold @xcite . in this abelian ,
toric case then , the third equation of ( [ df ] ) sets all @xmath22 to a single field , say @xmath25 , on the coherent component of the moduli space .
the second equation causes the d - terms to have fi - parametres @xmath26 .
the moduli space @xmath12 is a symplectic quotient of the space of solutions to the f - terms prescribed by the first equation modulo the gauge conditions prescribed by the d - terms . because of the condition that all @xmath19 sum to 0 imposed in ( [ k - con ] ) there is an overall @xmath27 ( corresponding to the center of mass motion of the m2-brane ) which can be factored out .
furthermore , there is another @xmath27 which can be factored .
this is because the presence of cs couplings induces fayet - iliopoulos ( fi ) parameters on the space of d - terms . in a generic @xmath7-dimensional theory
all these fi parameters are arbitrary , but here they are all aligned along a line which is parameterized by @xmath25 and has a direction set by the cs integers .
this picks out a very specific baryonic direction out of all possible directions which are present in @xmath7-dimensions .
this direction becomes mesonic in @xmath6-dimensions and fibers over the calabi - yau 3-fold to give a total space as a calabi - yau 4-fold .
thus , in all , there is a net of @xmath28 d - terms .
the space of solutions of the f - terms is itself a toric variety , of dimension @xmath29 , this is the so - called `` * master space * '' @xmath30 , studied in detail in @xcite . indeed , the @xmath31 d - terms are all the directions which remain baryonic in @xmath6-dimensions and give rise to the master space baryonic directions for a given lagrangian with @xmath9 gauge groups .
this does not imply that all possible baryonic directions of the particular cy4 are given by these @xmath31 directions .
it only provides a lower bound .
there are at least @xmath31 such baryonic directions and a different formulation may give more than this number .
such a situation is evident from the study of models presented in @xcite . ] .
since we are interested in the mesonic moduli space , we impose gauge invariance with respect to all of them . in summary , @xmath32 where we have marked the complex dimensions explicitly as subscripts .
the @xmath31 fi - parametres are shown in @xcite to be in the integer kernel of the matrix @xmath33 to proceed we now recall the essential features in the computation of the 3 + 1 dimensional mesonic moduli space of complex dimension 3 .
computationally , the charges of the fields are given by the so - called * incidence matrix * of the quiver ; this is a @xmath34 matrix .
each column of this matrix representation of the quiver consists of two possible choices ( 1 ) a single pair @xmath35 and @xmath36 denoting an arrow beginning and ending at the appropriate node and zero elsewhere , or ( 2 ) an entire column of zeros , denoting an adjoint field charged only under one single node .
the incidence matrix is customarily denoted as @xmath37 and specifies the d - terms , i.e. , the @xmath27 toric actions .
the crucial property of @xmath37 is that each of its columns sums to 0 ( since each column corresponds to one arrow that necessarily has one head and one tail . ) and that each of its rows also sums to 0 ( this ultimately ensures that the moduli space be calabi - yau ) .
we emphasise the ambiguity for an adjoint : we do not know under which precise node it is charged ; from the point of view of the incidence matrix , there is no way to distinguish .
we will see later how using information from the superpotential one may overcome this ambiguity . of course , one row of @xmath37 is redundant because of the summation rule , we delete it and customarily call the result @xmath38 . on the other hand ,
the f - terms can be solved in terms of a matrix @xmath39 , whose dual cone we call @xmath40 . from these
we can extract so - called @xmath41 and @xmath42 matrices .
we refer the reader to section 2 of @xcite for the details .
schematically we can summarise the procedure of the `` forward algorithm '' , taken from _ cit .
as follows : @xmath43^t & & \longrightarrow & & ( q_t)_{(c-3 ) \times c } = \left ( \begin{array}{c } ( vu)_{(g-1)\times c } \\
q_{(c - g-2 ) \times c } \\ \end{array } \right ) & \end{array}\ ] ] we have marked the dimensions of each matrix for clarity .
note that @xmath44 is the number of perfect matching in the dimer model @xcite description of the theory to which we shall later turn .
the key point is that we can combine the matter content ( specified by the incidence matrix ) and the superpotential ( specified by the @xmath39-matrix ) into a single charge matrix @xmath45 ; its kernel , @xmath46 , of dimension @xmath47 , encodes the toric diagram of the calabi - yau threefold moduli space .
refreshed with this recollection let us make one more step before getting back to the main case of interest .
having outlined the computational procedure , let us now proceed with a systematic study of examples .
ideally , one would wish for a classification of all possible theories , their quiver diagrams , interactions and subsequent geometries .
this is a daunting task of organising an infinite number of models .
nevertheless , we can proceed cautiously and modestly : let us start with theories with a single pair of black - white nodes in the dimer picture . these correspond to cases where the superpotential vanishes for a single brane ; this is because the so - called * toric condition * @xcite requires that each field appears exactly twice with opposite signs . in terms of the dimer model @xcite
, this condition is what gives rise to the bi - partite nature of the tiling .
of course , for multiple branes , because the fields become matrix - valued and do not commute necessarily , the superpotential may no longer vanish .
this is familiar to us .
for example , in the 3 + 1 dimensional gauge theory for the conifold 3-fold singularity there is a quartic superpotential which vanishes for a single brane .
we will encounter this situation in detail below .
hence , with a single m2 brane , and with only two terms therein , the superpotential actually vanishes , subsequently , the master space is freely generated by the @xmath10 fields and is simply @xmath57 .
thus , the matter content completely specifies the 4-fold singularity .
the symplectic quotient ( [ symp ] ) and the traditional approach both become relatively simple in this case .
in fact , in ( [ gt ] ) , @xmath58 and the perfect matching matrix @xmath54 is just the identity matrix .
the @xmath53 matrix for the f - terms are not present and we have that : @xmath59 where @xmath60 means @xmath12 has the toric diagram given by @xmath46 of dimensions @xmath61 .
indeed , from ( [ symp ] ) , we see that @xmath62 ; let us hence use @xmath10 as a single order parametre and proceed gradually .
we initiate with the case of two nodes , i.e. , @xmath63 . for each incremental value of @xmath10
, we classify all @xmath37-matrices satisfying the constraints which define the incidence matrix .
indeed , in ( [ symp2 ] ) , @xmath64 has dimensions @xmath65 and its kernel @xmath46 will have columns of length 4 , as is required for a toric diagram of a 4-fold . in summary , our classification scheme for 2 terms in the superpotential proceeds as follows : 1 .
fix @xmath9 , the number of nodes in the quiver .
this also fixes the number of arrows as @xmath62 ; 2 .
find all @xmath66 matrices which are incidence matrices , i.e. , ( a ) each column is one of only zeros ( adjoint ) or consists of a single pair of @xmath35 and @xmath36 and zero otherwise ( bi - fundamental ) , ( b ) each row sums to 0 ( calabi - yau condition ) ; 3 . identify all loops ( gauge invariant operators ) in the quiver drawn from each incidence matrix and construct possible 2-term superpotentials satisfying ( a ) the toric condition , i.e. , each bi - fundamental field occurs exactly twice and with opposite sign and ( b ) for a single brane when all the fields reduce from matrix - valued to complex numbers , the 2 terms conspire to cancel .
how do we attack step 2 , the most computationally intense one ?
luckily , the algorithms for such a matrix - partitioning problem were implemented in @xcite and we can thus happily proceed with presenting the solutions .
we adhere to the standard notation that _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for bi - fundamentals @xmath67 denotes the @xmath68-th arrow from node @xmath14 to @xmath69 and similarly that @xmath70 denotes the @xmath68-th adjoint on node @xmath14 ( when there is only a single arrow the @xmath68-index is dropped ) .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ already , there are many highly non - trivial theories . here
, there are 2 nodes and we find only two non - isomorphic solutions .
these are presented in figure [ f : e=4 ] in the appendix .
there are two possibilities . in model
( 1 ) we see that this is a non - chiral theory with 4 bi - fundamentals .
we recognise the quiver as that of the conifold quiver .
indeed , the full superpotential for arbitrary @xmath24 of that theory also has 2 terms : calling the bi - fundamentals @xmath71 and @xmath72 , ( @xmath73 and @xmath74 matrices ) we have that @xmath75 , which indeed satisfies the toric condition and vanishes for @xmath76 .
the fact that we have cs constraints , of course , modifies the moduli space from the 3-dimensional conifold to a cy 4-fold . in model
( 2 ) , there are 2 adjoint fields and 2 bi - fundamentals . where shall we place the 2 non - bi - fundamental fields in model ( 2 ) ?
we can be guided by promoting to @xmath77 number of branes .
there could be two possible placements : both as adjoints on 1 node or one on each node .
calling the two bi - fundamentals @xmath78 and @xmath79 , for both adjoints on the same node , say node 2 without loss of generality , we have 2 gauge invariant terms : @xmath80 and @xmath81 . on the other hand , for one adjoint on each node we have only one possible invariant : @xmath82 ; this is because we need to be careful in matrix composition when we go about the loops in the quiver to construct the gauge invariants .
therefore , model ( 2 ) has a natural candidate for a 2-term superpotential satisfying the toric condition and having two terms , namely @xmath83x_{21})$ ] . in summary ,
the two full models are drawn in figure [ f : e=4full ] .
= 10 cm @xmath84 ; @xmath85 @xmath86x_{21})$ ] of course , the actual moduli space @xmath12 for both models are easily determined using ( [ symp2 ] ) . here
@xmath87 and there is no symplectic action .
thus the moduli space is simply the master space , viz . , @xmath88 .
the toric diagram is merely 4 corners of a tetrahedron .
we are assured by the fact that these two models have appeared in the literature , model ( 1 ) , in @xcite and model ( 2 ) , in @xcite , in each case from different lines of reasoning .
here we have arrived at these from yet another viewpoint a systematic scan of theories .
it is perhaps important to emphasize that model ( 2 ) is not considered to be a consistent model in 3 + 1 dimensions even though it does admit a two dimensional tiling @xcite ( here consistent is taken to mean that it leads to a 3 + 1 dimensional scft under the rg flow to the ir , with a known ads dual ) .
the essential feature of model ( 2 ) is node 2 which has 1 flavor ( or taking the sqcd conventions of the number of flavors @xmath89 equal to the number of colors ) .
henceforth we shall call such nodes as `` * one - flavored nodes * '' .
indeed it is a standard lore that such theories develop a scale and confine in 3 + 1 dimensions , thus not leading to scft . on the other hand , such theories in 2 + 1 dimensions
do not necessarily develop a scale @xcite and can perfectly lead to a non - trivial scft under the rg flow .
once we allow for theories with one - flavored nodes , a whole space of opportunities reveals itself and a zoo of new scft s in 2 + 1 dimensions becomes available .
we encounter more examples of this type in the following sections . here , there are 3 nodes and we find 5 distinct solutions .
these are shown in figure [ f : e=5 ] in the appendix .
model ( 1 ) has 5 bi - fundamentals .
models ( 2 ) and ( 3 ) have 4 bi - fundamentals and 1 adjoint field .
note that model ( 2 ) has a disconnected node labelled 2 .
model ( 4 ) has 3 bi - fundamentals and 2 adjoint fields and finally model ( 5 ) , also with a disconnected node , has only 2 bi - fundamentals and 3 adjoint fields .
let us ignore models ( 2 ) and ( 4 ) since they inevitably have reducible moduli spaces due to disconnected nodes . for model ( 1 )
, using the standard notation , we clearly have the 2-term superpotential : @xmath90 , which satisfies the toric condition and vanishes for @xmath76 when all fields are simply complex numbers . for model ( 3 ) , clearly there are two invariants @xmath91 and @xmath92 . where shall we place the missing adjoint @xmath93 ?
it is easy to see that placing it on node 2 gives two invariants once we promote to @xmath77 branes , i.e. , to matrix - valued fields : @xmath94 and @xmath95 .
this indeed gives a two - term superpotential satisfying the toric condition of each field appearing exactly twice with opposite signs : @xmath96 ; moreover , @xmath18 indeed vanishes at @xmath76 when all field are merely complex numbers .
finally , for model ( 4 ) , there are 2 ways of placing the 2 adjoints : both on the same node , say node 1 , or one each on 2 of the nodes .
the latter possibility is ruled out since it would be impossible to have 2 terms and satisfying the toric condition .
however , putting both @xmath97 on node 1 , gives us a superpotential satisfying all requisites : @xmath98x_{13}x_{32})$ ] .
we summarise all these good models in figure [ f : e=5full ] .
it is important to note that none of these 3 models appeared previously in the literature .
indeed , our systematic study revealed the existence of these 3 prototypical models .
we will proceed later by analyzing their properties and further introduce them from yet another taxonomic viewpoint .
it is further important to notice that all these models have a one - flavored node and therefore do not correspond to consistent models in 3 + 1 dimensions even though they all admit a brane tiling .
nevertheless they present a rich structure of non - trivial scft s in 2 + 1 dimensions with a highly intricate spectrum of scaling dimensions .
= 14 cm @xmath99x_{13}x_{32 } ) . \ \\
\end{array}$ ] here , there are 4 nodes and there is a total of 18 distinct solutions .
these are shown in figure [ f : e=6 ] in the appendix .
models ( 1 ) to ( 10 ) have 6 bi - fundamentals .
note that models ( 1 ) , ( 2 ) , ( 5 ) and ( 9 ) all have disconnected nodes .
model ( 8) , though apparently acceptable , has a superpotential in terms of its bi - fundamentals which looks like @xmath100 .
this vanishes by cyclicity of the trace and hence it is really just a theory without superpotential at all .
models ( 11 ) and ( 12 ) have 5 bi - fundamentals and 1 adjoint field .
models ( 13 ) to ( 16 ) have 4 bi - fundamentals and 2 adjoint fields .
note that models ( 13 ) and ( 14 ) both have disconnected nodes .
model ( 17 ) has 3 bi - fundamentals and 3 adjoint fields , as well as a disconnected node .
finally , model ( 18 ) has 2 bi - fundamentals , 4 adjoint fields and 2 disconnected nodes .
once again , we select the diagrams without detached nodes , insert the appropriate adjoint fields , and also write down the possible 2-term superpotentials .
we find a total of 6 possible models .
these are presented in figure [ f : e=6full ] .
= 14 cm @xmath101 ) \end{array } $ ] happily , we have again recovered and extended some of the known models in the literature .
model ( 6 ) was proposed in figure 3 of @xcite for the @xmath102 geometry while model ( 10 ) was proposed in figure 7 of @xcite for the so - called @xmath103 theory .
the other models appear for the first time in the literature .
all models have a one - flavored node and so none correspond to consistent models in 3 + 1 dimensions even though they all admit a brane tiling description . for completeness
we also include the two disconnected quivers , models ( 3 ) and ( 15 ) .
these engender multi - trace superpotentials because of the quivers factorise .
we present these in figure [ f : e=6disconnect ] .
we see that they are essentially products of models ( 1 ) and ( 2 ) of the 4-edged case , together with a 2-edged bi - fundamental non - chiral quiver .
also , we include model ( 8) , which has a completely vanishing superpotential if it were to be 2-termed .
we will exclude these models from discussion in the following .
= 14 cm @xmath104x_{41 } ) \ ; \\ w_{(8 ) } = { \mathop{\rm tr}}(x_{31}x_{14}x_{42}x_{24}x_{41}x_{13 } - x_{42}x_{24}x_{41}x_{13}x_{31}x_{14 } ) = 0 \ ; \\ \end{array } $ ] let us take , for concreteness , model ( 1 ) of the @xmath105 quivers , given in figure [ f : e=5full ] .
there are 3 nodes and hence 2 independent cs levels , @xmath106 and @xmath107 . the toric diagram , according to ( [ gt ] ) , is given by @xmath108 in the penultimate step , we have been mindful of the fact we need to find the integer kernel of the charge matrix and not merely the nullspace and hence we wrote gen ( ) therein to denote that we should reduce to a basis over the integers for whichever choice of @xmath106 and @xmath107 .
this means that each row of @xmath46 must have gcd being 1 . guaranteed by the condition @xmath109 , rows 2 and
3 are acceptable but rows 1 and 4 need to divide out the common factor of @xmath107 ; this gives the last step .
indeed , each column of @xmath46 is of length 4 , signifying that the resulting moduli space corresponding to this theory is a toric 4-fold .
furthermore , we see that the vector @xmath110 is perpendicular to every pair - wise linear combination between the 5 column vectors , this means that the columns are actually co - spatial , i.e. , live on a dimension 3 hypersurface in @xmath111 ; this , of course , guarantees that our 4-fold is in fact calabi - yau .
thus re - assured , we can , without loss of generality , delete any row of @xmath46 ( call it @xmath112 ) and represent the toric 4-fold by an integer polytope in 3-dimensions .
thus , we can write , for the toric diagram , @xmath113 where for convenience we have deleted the second row . now , depending on the choice of @xmath106 and @xmath107 obeying the coprimarity condition , we have an infinite family of toric cy4s . as two illustrious examples we have that @xmath114
we see that the two are not related by any @xmath115 transformations and are thus inequivalent toric varieties .
we draw these diagrams explicitly in figure [ f : torice=5 ] .
= 12 cm the other two models of the @xmath105 quiver theories can be similarly treated . for model ( 3 )
, we have that @xmath116 this , as above , gives an infinite family , parametrised by choices of @xmath106 and @xmath107 , of possibilities for the moduli space @xmath12 and hence inequivalent theories .
finally , for model ( 4 ) , we have @xmath117 [ [ an - interesting - family ] ] an interesting family : + + + + + + + + + + + + + + + + + + + + + + + examining model ( 1 ) of figure [ f : e=5full ] and model ( 4 ) of figure [ f : e=6full ] , it is clear that for the general case of @xmath9 nodes and @xmath118 fields , there will always be a cyclically - directed @xmath9-gon graph with 1 edge having an extra pair of arrows in opposite directions .
thus all @xmath10 fields are bi - fundamentals and let us , without loss of generality , place the extra pair of bi - fundamentals between nodes 1 and @xmath9 , and let the direction of the arrows be @xmath119 , @xmath120 , ,
@xmath121 and @xmath122 .
the quiver and adjacency matrices are @xmath123 the 2-term superpotential is also straight - forward to write down : @xmath124 we can also apply ( [ gt ] ) to the incidence matrix to find the toric diagram .
the result has @xmath118 lattice points : @xmath125 [ [ six - field - models ] ] six field models : + + + + + + + + + + + + + + + + + + for completeness , let us present the matrices @xmath46 encoding the toric diagrams for the 6 models at @xmath126 , drawn in figure [ f : e=6full ] .
they are , respectively , @xmath127 in its original guise , toric duality @xcite referred to the phenomenon of 4-dimensional , @xmath8 gauge theories having the same vacuum moduli space as toric varieties , some classes of these were later realised to be seiberg dualities . in our host of examples
above , we have again encountered this duality , now in a more general setting .
take the two models of @xmath128 , they , even though having quite different quivers , share the same infrared moduli space as @xmath88 .
perhaps more dramatic is the following pair : take model ( 4 ) of @xmath105 at @xmath129 and inspect ( [ gt5 - 4 ] ) , then take model ( 16 ) of @xmath126 at @xmath130 and inspect ( [ gt6 ] ) , we see that they are @xmath131 we see that if we removed a repeated column in the latter , which does not influence the toric description , the moduli spaces of the two theories , which have different number of gauge group factors , different matter content and interactions and different chern - simons levels , are identical as toric varieties .
clearly there are infinitely many such cases and it is interesting to study the systematics of this phenomenon .
it is now well - established that the most convenient and elegant way of encoding the @xmath7-dimensional quiver gauge theory of d3-branes probing toric calabi - yau threefold singularities is through the formalism of dimer models , or , equivalently , brane - tilings @xcite .
the afore - mentioned `` toric condition '' @xcite of the superpotential is naturally interpreted as the bi - partite ( 2-colour ) property of the tiling while the calabi - yau condition of the threefold , which compels the toric diagram to be planar , gives the inherent structure of periodic tiling of the 2-dimensional plane .
one question which has emerged is how this would generalise to toric varieties of higher dimension .
indeed , proposals have been made which suggest that a 3-dimensional analogue of the dimer , a so - named `` crystal model '' , should encode the toric calabi - yau 4-fold case @xcite .
is this so for the 4-folds we have encountered in our investigation ?
surprisingly , we find all our above cs theories to afford 2-dimensional tilings , rather than the naively expected crystals which tile 3-dimensions .
let us recall that for the @xmath7-dimensional gauge theories , an important relation exists @xcite : @xmath132 where @xmath133 is the number of terms in the superpotential , while @xmath10 and @xmath9 , as above , are the number of fields and gauge group factors .
the recognition of ( [ euler ] ) as the topological euler equation for the simplex decomposition of a genus 1 riemann surface @xmath134 , with a number @xmath133 of vertices , a number @xmath10 of edges and a number @xmath9 of faces was key to the birth of the dimer model . indeed , the graph dual of this simplex is a periodic version of writing the quiver diagram together with the superpotential , the periodicity further supporting the existence of @xmath134 as a torus .
how does this crucial relation read for our @xmath6-dimensional cs theories ? here , since we are only considering 2-term superpotentials , @xmath135 .
moreover , recall that @xmath62 from ( [ symp ] ) .
therefore @xmath136 and ( [ euler ] ) is still satisfied !
this is not what one would expect from a crystal model which is not a periodic tiling of the plane but which is perhaps at first expected for all toric 4-fold theories . in summary , we have that _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ all quiver chern - simons theories corresponding to a m2-brane probing a toric calabi - yau 4-fold , such that the superpotential has two terms , admit a dimer model ( 2d - tiling ) description .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
we could have continued the process of section [ s : tax ] _ ad infinitum _ , listing more and more graphs and then for each , construct possible superpotentials with 2 terms or fix various values of chern - simons levels to obtain infinite families of toric moduli spaces .
this , though explicit , is perhaps not so illustrative , let alone computationally prohibitive .
it would , however , be most enlightening if we could count , say , the number of possible quivers for a given number of nodes . in this section ,
let us give a generating function to perform this count ; we will find an elegant result very much in the spirit of the plethystic programme @xcite . to this end
, we shall introduce a systematic enumeration and construction of the quivers .
let us do so by the concept of _ base node _ which we now introduce .
examining model ( 2 ) of figure [ f : e=4full ] , models ( 3 ) and ( 4 ) of figure [ f : e=5full ] , as well as models ( 7 ) , ( 11 ) and ( 16 ) of figure [ f : e=6full ] , we see that they , perhaps not immediately obviously , fall into a family .
these are all models which have a single base node , viz . , a single reference node whence loops depart and thence return .
let us consider a chain of closed paths beginning and ending on this same node ( corresponding to a possibly multi - trace gauge invariant operator ) and denote it by a sequence of non - negative integers each entry of which encodes the length of one loop . clearly , this sequence is unordered .
for example , for the simple quiver which has a single node with 3 self - loops attached ( incidentally , this is the quiver of the @xmath137 super - yang - mills theory for d3-branes in flat @xmath138 paradigmatic in the first ads / cft pair ) , we would denote it as 000 , drawn in figure [ f : c3 ] .
= 3 cm next , what would 001 ( and of course any of its permutations ) denote ?
this is model ( 2 ) of figure [ f : e=4full ] , which has 2 loops of length 0 and 1 loop of length 1 .
note that by length here we mean the number of different nodes traversed before returning .
similarly , 011 and 002 correspond to models ( 3 ) and ( 4 ) of figure [ f : e=5full ] respectively .
likewise , we retrieve the 3 aforementioned models for @xmath126 fields corresponding to 111 , 012 and 003 .
the systematics is thus clear . for @xmath10 fields and @xmath139 nodes , we are counting graphs which are in correspondence with unordered partitioning of @xmath140 into 3 parts of non - negative integers .
this is a standard problem , whose solution is simply @xmath141 this function was indeed encountered in @xcite as the generating function for 1 adjoint field with @xmath142 d3 branes .
the above is the generating function such that the coefficient in front of @xmath143 gives the number of quivers of 1 base node with @xmath9 nodes .
these are all quivers with 3 primitive loops .
having addressed one family of quivers transcending across @xmath10 , the family with a single base node , let us move onto 2 base nodes , say 1 and 2 .
now , it is more convenient to count by open paths : we let @xmath144 denote a configuration which has open paths of length @xmath145 , @xmath146 , @xmath147 , @xmath148 respectively starting at node 1 , ending at node 2 , starting at node 2 and ending at node 1 . again , by length we mean distance away from the base pair .
these are all quivers with 4 primitive paths between the 2 base nodes .
specifically , 0000 would correspond to model ( 1 ) of figure [ f : e=4full ] .
next , taking the base pair to be nodes 3 and 1 , 0001 would denote model ( 1 ) of figure [ f : e=5full ] . moving on to 4 nodes and 6 fields
, we see that 0002 , 0011 and 0101 denote respectively models ( 4 ) , ( 6 ) and ( 10 ) of figure [ f : e=6full ] when taking nodes 4 and 1 and the base pair .
this is analogous to the above , but is the unordered partitioning of @xmath31 into 4 non - negative integers with the extra complication that we must respect the di - hedral symmetry .
specifically , the transpositions @xmath149 , @xmath150 , @xmath151 generate a dihedral group of order 8 .
orbits under these 8 elements must be quotiented out .
the generating function for this also admits a standard solution by method of molien series @xcite for the dihedral group , in a 4-dimensional representation acting on our 4-vector : @xmath152 here , the coefficient of @xmath153 is the number of models with @xmath154 nodes . what about triplets or more of base nodes ?
note that our above two have already exhausted all the quivers we have so far explicitly constructed .
could there be more families which one might encounter at higher @xmath10 ?
we now argue that we shall , in fact , not .
first , let us recall that our counting procedure above of course does not include any disconnected graphs and moreover excludes shapes such as model ( 8) of figure [ f : e=6disconnect ] .
this is a 3-base - node example ; however , we have explicitly shown that the superpotential vanished .
indeed , the cases of base nodes being one or two precisely permitted us to write a commutation relation allowing for the two terms in the superpotential : the case of the 1-base - node achieved with the adjoint and the case of the 2-base - node , with the pair of bi - fundamentals between them .
any other case must either be reducible to these two situations or have vanishing superpotential .
indeed , we can use an additional symmetry to restrict the possible models . consider the number of flavours @xmath155 for a given node @xmath68 : this is either an incoming - outgoing pair of arrows or an adjoint .
the vector @xmath155 for @xmath156 is an additional order parameter . from our examples
, we see that it is 1 for most nodes .
this is the one - flavoured node we discussed earlier .
however , it must be at least 1 for every node since we are not counting disconnected quivers and it must never exceed 3 since there are only @xmath118 number of arrows in total . clearly , the sum of @xmath155 over @xmath68 is equal to @xmath118 . therefore , there are only 2 possibilities for the components of the vector : ( a ) @xmath140 of them being 1 and a single 3 or ( b ) @xmath31 of them being 1 and 2 of them being 2 . subsequently , this places a significant restriction on the total number of adjoints . since each node must have at least one bi - fundamental , case ( a )
could have up to 2 adjoints on the last node and case ( b ) could have up to a single adjoint on the last two nodes .
hence , in total there could really only be 0 , 1 or 2 adjoints .
now , we see that any node which has no adjoint and only an incoming / outgoing pair of arrows is a descendent of a simpler configuration : namely replace this node , together with its 2 attached arrows , with a single arrow between the 2 nodes from which the said node emanates .
therefore , we need only consider the parent quivers of ( a ) and ( b ) , viz . , a single node with flavour number 3 or two nodes each with flavour number 2 , respectively .
the former is 000 .
the latter has two possibilities : 0000 and a quiver which looks like @xmath157 ; we can easily show that the latter admits no non - vanishing 2-term superpotential . indeed ,
model ( 8) of figure [ f : e=6disconnect ] discussed above is a descendent thereof and also needs to be eliminated . in summary
, we only need to consider descendents of the 3-vector 000 and the 4-vector 0000 and no quivers with more than 2 base nodes survive .
hence , @xmath158 and @xmath159 together exhaust all counting . furthermore ,
noticing the shift in power of @xmath159 with respect to @xmath158 , we conclude that : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the number of theories of our interest , i.e. , all non - trivial , connected models with 2 terms in a non - vanishing superpotential and @xmath154 nodes , is counted by the coefficient of @xmath160 in the expansion of the generating function _ @xmath161 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the forms of @xmath158 and @xmath159 are perhaps familiar to the astute reader .
the first , is the hilbert series ( cf .
@xcite ) for the weighted projective plane @xmath162}$ ] ( or , alternatively the quotient @xmath163 of the symmetric group on 3 objects ) , and the second , that of the 4-fold quotient singularity @xmath164 , i.e. , the orbifold of @xmath88 by the dihedral group of order 8 ( cf . @xcite for discrete subgroups of @xmath165 ) .
it is quite elegant that these two geometrical spaces should encode the entire space of quivers of our type .
in fact , quivers of the first family , the single - base - node type , obey a simple rule of composition , reflecting the fact that there is an underlying commutative algebra which is freely generated : the zero element is the self - adjoining loop , 1 is the quiver with 2 nodes and a pair of bi - fundamentals in opposite directions between them , and thus generalising to @xmath166 , which is the quiver that is an @xmath166-gon with bi - fundamentals cyclically going around once .
we can create a quiver by selecting any 3 from this list and composition is by pasting the three at a chosen common node : @xmath167 in terms of the 3-vector of integers , we can also see this composition .
for example , 001 + 011 = 012 refers to the following pasting of quivers @xmath168 indeed , we have drawn the result in a suggestive form , so as to indicate that such a composition corresponds to the insertion of nodes .
specifically , adding @xmath169 signifies inserting , after choosing an absolute ordering of the loops , @xmath145 , @xmath148 and @xmath170 nodes into the loops .
in this fashion , we readily see that the quivers @xmath171 , @xmath172 and @xmath173 generate , by this insertion procedure , all quivers with one base node .
thus , we have an algebra freely generated by the three elements : @xmath174 the second family is a little more involved
. we can define the 0 element as a node with an out - going arrow , 1 as a node with an incoming and an out - going arrow , etc .
: @xmath175 composition is by again pasting , though this time we must first paste to create a pair of base nodes , unto which we may then attach the open arrows of the element @xmath166 .
finally , we must eliminate the redundancies of quivers which obey dihedral symmetry . in terms of the 4-vector
, we now have an algebra generated by 5 elements , 0001 , 0011 , 0101 , 0111 and 1111 : @xmath176 obeying a single relation .
we have thus uncovered an algebraic structure on the space of cs quivers with 2-term superpotentials .
this is expected to persist to higher number of terms and indeed to the space of all quivers ; we have thus found an underlying algebraic variety for the set of chern - simons quivers .
we remark also that taking the plethystic logarithm @xcite of the total @xmath177 reveals that it is a non - terminating series suggesting that the algebraic variety is not a complete intersection .
it is in the form of a set union of the two orbifolds , though there is a degree shift on the second .
in this paper we have started a taxonomic study of quiver chern - simons theories in @xmath6-dimensions .
in particular , we study the theories arising from m2-branes probing toric calabi - yau 4-fold singularities .
our purpose is not to merely provide a catalogue of such theories since there are clearly infinite families thereof , but to study the space of these theories from both a synthetic and an analytic approach , and to uncover interesting physical phenomena as well as mathematical structure .
some of the techniques and concepts can indeed be generalised to quiver theories in arbitrary dimension and with any supersymmetry .
we have first presented , in flow - chart ( [ gt ] ) , a succinct way of a `` forward algorithm '' which takes the quiver and superpotential data as input and the moduli space as an output : the former as a pair of matrices , respectively the incidence and the perfect matching matrices , and the latter , as a matrix representing the integer coordinates of the toric diagram of the calabi - yau 4-fold .
the intermediate computations involve nothing more than matrix multiplications and finding kernels .
indeed , this generalises and significantly improves on the forward - algorithm of the case of @xmath7-dimensional gauge theories of @xcite : we only need to insert a @xmath48-matrix of the chern - simons levels and have obviated the need for finding dual cones .
thus armed , we have commenced the classification of all quiver chern - simons theories .
listing the quivers is matter of combinatorics and for simplicity we have taken the superpotential to have two terms .
it is clearly a pressing problem what happens at an arbitrary number @xmath133 of terms , this we shall leave to forth - coming work .
subsequently our order parametre becomes @xmath9 , the number of nodes and @xmath10 , the number of arrows , is equal to @xmath178 since the master space here is @xmath179 .
in general , we should have 3 order parametres @xmath180 .
we exhaustively list the first members at @xmath181 and find many non - trivial theories , including and extending beyond what has so far emerged in the literature .
because of our forward algorithm , we can readily determine the toric calabi - yau 4-fold moduli spaces for all the theories presented .
these exhibit as many toric varieties , each of which is an infinite family , indexed by the integer chern - simons levels .
remarkably we have once more encountered the phenomenon of `` toric duality '' first noted in @xcite for @xmath7-dimensional gauge theories .
now , it has manifested in perhaps some more striking guises : we find not only theories with the same number of nodes but also those with different number of nodes , fields , as well as chern - simons levels , to flow to the same ir moduli space as toric calabi - yau 4-folds .
it is certainly of importance to study what is precisely happening in the field theory of such pairs .
building upon our host of examples , we have stepped back for a panoramic view of all our theories : connected quiver chern - simons with 2-term toric superpotentials . though the graphs seem unwieldy and growth rapidly in number as we increase the number of nodes , we have found a generating function ( [ gen ] ) which counts the number of inequivalent theories given the number of nodes .
remarkably , this generating function splits conveniently into 2 pieces , each of which is the hilbert series of a precise algebraic variety : the non - abelian quotient space @xmath163 by the symmetric group on 3 objects of order 6 and the abelian quotient @xmath164 by the ordinary dihedral group of order 8 ( i.e. , the symmetry group of the square ) .
could we package the case with arbitrary @xmath182 into a convenient tri - variate generating function and would the result have an interpretation as a hilbert series , whereby suggesting that the space of quiver theories is itself some algebraic variety ? in answering this and many questions raised we are presently engaged .
indeed , we have only tread upon the fringe of a vast and fertile land , into her bountiful bosom we must onwardly march .
_ scientiae et technologiae concilio y .- h . h. hoc opusculum dedicat cum gratia ob honorem officiumque socii progressi collatum .
et ricardo fitzjames , episcopo londiniensis , ceterisque omnibus benefactoribus collegii mertonensis oxoniensis quorum beneficiis pie , studiose , iucunde vivere licet , pro amore catharinae sanctae alexandriae et ad maiorem dei gloriam . _
in this appendix , we present the results for a preliminary classification of the case of 2 terms in the superpotential .
this will include cases of disjoint quivers and unmarked adjoint fields , we present them here for completeness .
the cases of 4 , 5 and 6 fields are respectively shown in figures [ f : e=4 ] , [ f : e=5 ] and [ f : e=6 ] . in the text we will sift through these graphs carefully , adding adjoints wherever consistent .
= 10 cm = 14 cm = 16 cm
in this appendix , we extract the matrix @xmath54 of perfect matchings directly from the superpotential @xmath183 of the fields @xmath184 . in the archaic days of the forward algorithm ,
@xmath54 is found to be the product of the matrix @xmath39 ( which encodes how a total of @xmath10 f - terms can be solved in terms of a fewer number ) with its dual cone matrix @xmath40 .
now we can follow the ensuing prescription to directly obtain @xmath54 without the rather computationally intensive procedure of finding dual cones : 1 .
group all the monomials appearing in @xmath18 into those with positive coefficients and those with negative , giving us two sets of equal length ( since each field must appear exactly twice with opposite signs by the toric condition ) , on each of which we choose an order ; 2 .
find the position of field @xmath185 in each of the two sets ; say it is @xmath68-th in the positive set and @xmath186-th in the negative set .
construct a matrix whose @xmath187-th entry is @xmath185 ( should the index ever repeat , simply add , to the entry , the new field ) .
do so for all @xmath10 fields and patch in zeros otherwise .
this is the kasteleyn matrix .
compute its determinant @xmath17 .
3 . @xmath17 is a new polynomial in the @xmath10 fields with @xmath44 ( the number of perfect matchings ) terms .
fix an order for these terms , and construct and @xmath188 matrix . for each of the @xmath10 fields
find the location in the monomial in @xmath17 .
we place a 1 at this location in the @xmath10-th row correspondingly and 0 otherwise .
the resulting matrix is the desired @xmath54 . as an illustrative example , let is consider the famous cone over the zeroth del pezzo surface , i.e. , the abelian orbifold @xmath189 .
its quiver is a directed loop over 3 nodes , each with multiplicity 3 and its superpotential is a six term cubic in 9 fields : @xmath190 we recall that ( cf .
e.g. , eq .
( 2.12 ) - ( 2.17 ) of @xcite ) the explicit solutions to the f - terms are @xmath191 and hence @xmath192 where dual refers to finding the dual cone .
the product @xmath193 is the matrix @xmath54 . using our new algorithm
, we find that there are 3 positive terms and 3 negative terms , giving us a kasteleyn matrix and its determinant @xmath17 as : @xmath194 upon comparison , we find that both procedures give ( up to trivial permutation of columns , note that we have fixed the row by a canonical ordering of the fields ) @xmath195 @xmath196
. we must point out that on a simple timing contrast , our new algorithm for this simple example is about a factor of 10 faster ! this is significant since the algorithm for finding dual cones is exponential running time so for bigger examples , we expect that our new method to be a substantial improvement . for example , for the cone over the third del pezzo surface , with 14 fields and 8 terms in the superpotential , our present algorithm constitutes a time reduction of more than 1000-fold over the old dual - cone method !
it would be interesting to see whether this technique could be generalised to rapidly find the dual cone of arbitrary integer cones ; this would of tremendous use to toric geometry .
99 j. bagger and n. lambert , `` modeling multiple m2 s , '' phys .
d * 75 * , 045020 ( 2007 ) [ arxiv : hep - th/0611108 ] . `` gauge symmetry and supersymmetry of multiple m2-branes , '' phys .
d * 77 * , 065008 ( 2008 ) [ arxiv:0711.0955 [ hep - th ] ] .
`` comments on multiple m2-branes , '' jhep * 0802 * , 105 ( 2008 ) [ arxiv:0712.3738 [ hep - th ] ] . s. benvenuti , b. feng , a. hanany and y. h. he , `` counting bps operators in gauge theories : quivers , syzygies and plethystics , '' jhep * 0711 * , 050 ( 2007 ) [ arxiv : hep - th/0608050 ]
. b. feng , a. hanany and y. h. he , `` counting gauge invariants : the plethystic program , '' jhep * 0703 * , 090 ( 2007 ) [ arxiv : hep - th/0701063 ] . c. beasley , b. r. greene , c. i. lazaroiu and m. r. plesser , `` d3-branes on partial resolutions of abelian quotient singularities of calabi - yau threefolds , '' nucl .
b * 566 * , 599 ( 2000 ) [ arxiv : hep - th/9907186 ] .
j. gray , y. h. he , v. jejjala and b. d. nelson , `` exploring the vacuum geometry of n = 1 gauge theories , '' nucl .
b * 750 * , 1 ( 2006 ) [ arxiv : hep - th/0604208 ] ; y. h. he , `` vacuum geometry and the search for new physics , '' in the proceedings of 9th workshop on non - perturbative qcd , paris , france , 4 - 8 jun 2007 , pp 04 .
d. forcella , a. hanany , y. h. he and a. zaffaroni , `` the master space of n=1 gauge theories , '' jhep * 0808 * , 012 ( 2008 ) [ arxiv:0801.1585 [ hep - th ] ] ; ` mastering the master space,' lett .
phys . *
85 * , 163 ( 2008 ) [ arxiv:0801.3477 [ hep - th ] ] . b. feng , a. hanany , y. h. he and a. m. uranga , `` toric duality as seiberg duality and brane diamonds , '' jhep * 0112 * , 035 ( 2001 ) [ arxiv : hep - th/0109063 ] .
+ c. e. beasley and m. r. plesser , `` toric duality is seiberg duality , '' jhep * 0112 * , 001 ( 2001 ) [ arxiv : hep - th/0109053 ] .
o. aharony , a. hanany , k. a. intriligator , n. seiberg and m. j. strassler , `` aspects of n = 2 supersymmetric gauge theories in three dimensions , '' nucl .
b * 499 * , 67 ( 1997 ) [ arxiv : hep - th/9703110 ] .
a. hanany and y. h. he , `` a monograph on the classification of the discrete subgroups of su(4 ) , '' jhep * 0102 * , 027 ( 2001 ) [ arxiv : hep - th/9905212 ] . | we initiate a systematic investigation of the space of 2 + 1 dimensional quiver gauge theories , emphasising a succinct `` forward algorithm '' .
few `` order parametres '' are introduced such as the number of terms in the superpotential and the number of gauge groups .
starting with two terms in the superpotential , we find a generating function , with interesting geometric interpretation , which counts the number of inequivalent theories for a given number of gauge groups and fields .
we demonstratively list these theories for some low numbers thereof .
furthermore , we show how these theories arise from m2-branes probing toric calabi - yau 4-folds by explicitly obtaining the toric data of the vacuum moduli space . by observing equivalences of the vacua between markedly different theories ,
we see a new emergence of `` toric duality '' .
imperial / tp/08/ah/10 + 0.25 in m2-branes and quiver chern - simons : @xmath0 a taxonomic study + amihay hanany@xmath1 and yang - hui
he@xmath2 + + [ cols= " < , < " , ] |
numerous independent surveys have demonstrated evidence on both cosmological and galactic scales that about 30% of the matter energy density of the universe consists of non - baryonic , non - luminous matter . a leading candidate for this dark matter
is a yet - to - be - discovered weakly interactive massive particle ( wimp ) which could directly interact with detectors based on earth leading to kev - scale nuclear recoils .
direct dark matter detection experiments are now probing well - motivated models of extensions to the standard model such as supersymmetry which naturally predict dark matter candidates @xcite . as the exposures of direct dark matter detection experiments continue to improve , they will soon have enough sensitivity to detect neutrinos from several astrophysical sources such as the sun , the atmosphere , and diffuse supernovae @xcite .
for example , a 1 kev threshold xe based experiment with a 1 ton - year exposure will detect about 100 @xmath1 solar neutrino events via coherent neutrino - nucleus scattering ( cns ) .
in fact for some wimp masses , such neutrino backgrounds can almost perfectly mimic a wimp signal .
it has been shown in ref .
@xcite that the cns background leads to a strong reduction of the discovery potential of upcoming experiments . therefore , though neither coherent neutrino scattering nor the wimp - nucleus interaction have conclusively been observed yet , the search for discrimination methods to disentangle wimps from neutrino events is a necessity .
several methods to improve on the discrimination power between a wimp and a neutrino origin of the observed nuclear recoils have been suggested @xcite , including using the annual modulation signal @xcite or directional detection methods @xcite . in this paper
, we propose a new method to reduce the effect of this neutrino background by looking for a possible complementarity between different target nuclei .
in addition to discussing the target complementarity for different spin - independent ( si ) targets , for the first time we discuss the prospects for complementarity using spin - dependent ( sd ) targets . this paper is organized as follows . in sec .
[ sec : neutfloor ] , we briefly review how to compute the wimp and neutrino event rates and we explain the test statistic that has been used to generate discovery limits .
we then compute some discovery limits for several upcoming direct dark matter detection experiments in light of the neutrino background . in sec .
[ sec : beyond ] , we show how target complementarity can help on improving the wimp discovery potential by considering both si and sd interactions .
we then illustrate the improvement on the discovery limits by combining data from different targets . in sec .
[ sec : optimization ] , we explore how one can optimize the relative exposures of the different target materials in a given experiment in order to maximize its discovery potential using the effect of target complementarity .
[ cols="^,^,^,^,^,^,^,^ " , ] interestingly , from figs .
[ fig : evolution_sys ] and [ fig : evolution_mass ] we can clearly see that the neutrino background will impact the low wimp mass ( below 10 gev / c@xmath0 ) and the high wimp mass ( above 10 gev / c@xmath0 ) regions at very different exposures due to the vastly different rates of neutrino backgrounds seen in fig [ fig : eventrate ] .
indeed , the discovery limit evolution with exposure transitions from 1/mt to 1/@xmath2 as soon as the expected neutrino background nears 1 event , which corresponds to exposures of a few kg - years for very - low - threshold experiments in the low - mass region and a few tens of ton - years for the high - mass region for idealized perfect - efficiency experiments ( actual exposure values depend on the target , threshold , and efficiency of a given experiment ) .
the saturation regime exists for exposures between 100 kg - year and 10 ton - year and between 10@xmath3 ton - year and 10@xmath4 ton - year for the low and high wimp mass regions respectively .
therefore , one can conclude that the next generation of low - threshold experiments focusing on the low wimp mass region could reach well into the @xmath5 regime and even begin to see the effects of saturation , while at high wimp masses the next generation of experiments will at most begin to see the 1/@xmath2 effects from the neutrino background . from fig .
[ fig : evolution_sys ] and [ fig : evolution_mass ] , we have seen that the neutrino background can lead to a saturation regime of the discovery potential over certain mass ranges that can span over 2 orders of magnitude in exposure .
however , with enough exposure or lower systematics , one can always improve the discovery limit . in the following , we will define an arbitrary neutrino induced discovery limit within the saturation regime , derived from a set of two different exposures ( and energy thresholds ) for the low and high wimp mass regimes .
the thresholds are defined such that for the low mass region we expect no @xmath6 neutrino events in the data and that for the high mass region we expect no @xmath1 neutrinos events in the data .
the exposures are set such that the low wimp mass regime detects 200 events from @xmath1 ( about 1660 neutrino events total including pep , hep and @xmath7 ) , and the high wimp mass regime detects 400 neutrino events .
the corresponding exposures and thresholds for each target discussed in the following are summarized in table [ tab : exposures ] .
one can then see , using figure [ fig : evolution_sys ] , that the computed limits are always within the saturation regime of the discovery potential for both the low and the high wimp mass regions .
as previously discussed , this is where adding a new observable such as annual modulation @xcite and/or directionality @xcite , or combining data from several experiments , as suggested in sec .
[ sec : beyond ] , can lead to the most substantial improvement in the discovery potential .
figure [ fig : discovery_limit_si ] , [ fig : discovery_limit_sdp ] and [ fig : discovery_limit_sdn ] show the computed limits for single - target and multi - target - based experiments for both si and sd interactions using the target and neutrino properties shown in tables [ tab : nuclei ] and [ tab : neutrino ] , and the energy thresholds and exposures from table [ tab : exposures ] .
it should be noted that these are limits calculated for idealized experiments with perfect efficiency , and no background except for neutrinos , and thus represent the best attainable discovery limit for the listed exposures . in figure
[ fig : discovery_limit_si ] ( left panel ) we show the discovery limits of elemental targets considering only the si interaction .
one can first notice that , using a binned likelihood function , we have been able to reproduce the results from @xcite . in the si case
the equivalent wimp models corresponding to a given neutrino type are only weakly dependent on the considered target ( see sec .
[ sec : beyond ] for more details ) , thus all single - nucleus based targets share a very similar discovery limit . on figure [ fig : discovery_limit_si ] (
right panel ) we show the discovery limits for the si interaction considering compound targets such as @xmath8 ( used in cresst ) , @xmath9 ( used in coupp ) , and @xmath10 ( used in pico and picasso ) .
since experiments that use @xmath10 and @xmath9 do nt currently measure the recoil energy of the event , we computed limits with and without energy sensitivity , shown as dashed and solid lines respectively . in the case of compound targets ,
we chose the thresholds by considering the lightest nucleus of the mixture as it has its neutrino spectra the most shifted to the high recoil energies . for example , considering @xmath8 , in order to compute the low threshold part of the discovery limit , we chose a threshold such that there is no pp neutrino events in the data due to the o target , _ i.e _ 25 ev . by choosing this threshold we do not consider the same event rate proportions for the three different targets .
this explains the three successive exponential - like falls from a wimp mass of @xmath11 to about @xmath12 for this target .
we can apply the same reasoning for the @xmath9 target where we can see two different exponential - like falls due to i and f targets . for @xmath10
this effect is not visible as f and c have very similar atomic masses .
note that the discovery limit for cf@xmath13 ( used in experiments such as mimac @xcite and dmtpc @xcite ) is almost identical to the one from a c@xmath14f@xmath15 based experiment with energy sensitivity .
therefore , all the following results using @xmath10 are also relevant to cf@xmath13 .
figures [ fig : discovery_limit_sdp ] and [ fig : discovery_limit_sdn ] show the discovery limits considering a sd interaction on the proton and on the neutron , respectively .
these discovery limits have similar shapes to those derived for the si interaction but their relative amplitudes can be very different . for example , considering a sd interaction on the proton , a si - based experiment and an experiment using @xmath10 have their discovery limits around @xmath16 and @xmath17 in the low mass region , respectively .
this difference is due to the nature of the sd interaction .
indeed , the neutrino - nucleus cross section is still described by a coherent effect ( evolution in @xmath18 ) while the sd wimp - nucleus cross section depends on the total angular momentum and mean spin contents of the considered target ( see eq .
[ eq : sigsd ] and table [ tab : nuclei ] ) .
another interesting feature that one can notice from figures [ fig : discovery_limit_sdp ] and [ fig : discovery_limit_sdn ] concerns the multi - target based experiments . indeed , for such experiments
, some nuclei from the target material can not lead to a sd interaction as they have @xmath19 .
this effect leads to different shapes in the discovery limits between the si and sd cases , as one can see for @xmath10 where @xmath20 and @xmath21 .
one may also notice that @xmath8 does not appear on figures [ fig : discovery_limit_sdp ] and [ fig : discovery_limit_sdn ] as it has no sd sensitivity .
for these six figures , the discovery limit indicates the region of parameter space where the neutrino background has its highest impact on the reach of upcoming direct detection experiments . indeed , with the considered energy thresholds and exposures , the discovery potential is in the saturation regime over most of the mass range , where a slight increase in sensitivity would be at the cost of a large increase in exposure .
this clearly highlights the need for developing alternative strategies to be able to probe dark matter models lying underneath this neutrino - induced bound on the wimp discovery reach .
as shown in the previous section , it can be extremely difficult to claim a discovery of dark matter if the true wimp model lies below the neutrino background .
however , there are several possible ways to improve the discovery potential , even in the neutrino - induced saturation regime .
the first one is to improve the theoretical estimates and experimental measurements of neutrino fluxes as it has been shown in figure [ fig : evolution_sys ] .
a second possibility is to add some new observables that could help at disentangling between the wimp and neutrino origin of the observed nuclear recoils .
this could be done by searching for annual modulation @xcite and/or by measuring the nuclear recoil direction @xcite , as suggested by upcoming directional detection experiments @xcite .
indeed , since the main neutrino background has a solar origin , the directional signature of such events is expected to be drastically different from the wimp induced one @xcite .
we are now in position to study the gains in discovery potential that can be achieved by combining data from different target materials in light of the neutrino background .
as discussed in the previous section , the effect of neutrino background is notably important at particular wimp masses and cross sections where the wimp and neutrino spectra are similar , both in shape and magnitude .
the wimp mass and cross section where this occurs is target - material dependent .
one may then wonder whether combining different experiments could alleviate such degeneracies between the wimp and the neutrino hypotheses . in order to investigate this possibility , we determine what wimp models can be mimicked by @xmath22b neutrinos by computing the maximum likelihood distributions under the wimp only hypothesis in the ( @xmath23 , @xmath24 ) plane from fake data containing only nuclear recoils from @xmath22b neutrinos . because of the small differences between the wimp and neutrino energy distributions , the reconstructed wimp masses for each target are slightly different .
this is result shown in figure [ fig : complementarity ] ( left panel ) . for a si interaction , the reconstructed cross sections for different targets are all roughly the same as is shown on figure [ fig : complementarity ] ( right panel ) .
this is because both neutrino- and wimp - nucleus cross sections evolve as @xmath25 ( eq . [ eq : sigsi ] ) .
therefore , we expect the target complementarity in the si case to be fairly weak
. however , for a sd interaction the situation is different . as figure [ fig : complementarity ]
( right panel ) clearly indicates , the reconstructed cross sections for the different targets can be different by many orders of magnitude .
this is due to the nature of the wimp sd cross section which depends on the angular momentum ( @xmath26 ) , spin content ( @xmath27 ) and isotopic fractions ( @xmath28 ) of the considered targets ( eq . [ eq : sigsd ] ) .
thus , as the reconstructed wimp models of a given neutrino for different target nuclei are strongly distinguishable , one could expect to gain in discrimination power between the wimp and the neutrino hypothesis by combining data from several experiments . as a matter of fact ,
figure [ fig : complementarity ] ( right panel ) can be seen as a way to quantify the target complementarity . indeed ,
the larger the difference in reconstructed wimp cross sections between two targets is , the greater the discrimination power and improvement in a combined discovery potential becomes . as mentioned above , target complementarity is not very efficient for the si interaction as one can see on figure [ fig : complementarity_si ] ( left ) where we have computed the discovery limits ( solid curves ) for progressive combinations of xe , ge , si , and @xmath10 targets using thresholds taken from table [ tab : exposures ] for the low wimp mass region .
the exposures for each individual target have been set such that we expect a total of 1,000 @xmath1 neutrino events equally distributed amongst each experiment .
for example , the xe+ge curve has exposures set such that 500 events are detected in xe and 500 in ge .
the slight shift to higher wimp masses when adding si to the xe+ge mixture comes from the differences in the reconstructed wimp masses from a @xmath22b signal shown in fig .
[ fig : complementarity ] ( left panel ) .
@xmath10 is currently used in bubble chamber experiments such as coupp which do not have sensitivity to the recoil energy since they detect events which deposit an energy density above the threshold required to generate bubble nucleation @xcite .
we have thus included curves with and without energy sensitivity for this target . for reference ,
the background - free sensitivity limits for the chosen exposures are also shown as dashed lines .
the ratio between the sensitivity and the discovery limits , shown in fig .
[ fig : complementarity_si ] ( right panel ) , allows us to quantify the impact of the neutrino background on the wimp discovery potential of these idealized experiments . as expected , for si wimp interactions target complementarity has a modest effect and can at most reduce by a factor of 2 the impact of the neutrino background on the discovery potential .
[ fig : complementarity_sdp ] ( left panel ) is similar to fig .
[ fig : complementarity_si ] ( left panel ) but for the sd interaction on the proton .
from fig .
[ fig : complementarity ] ( right ) we see that fluorine targets have a large advantage in sensitivity for sd proton interactions .
one can see this in the large improvement in discovery potential when adding @xmath10 to the compliment of targets .
one might wonder if it is worth combining @xmath10 with other experiments due to its advantage for this interaction . to consider this
, we have added a @xmath10-only discovery and sensitivity limit .
one can see that even though one obtains the best background - free sensitivity with @xmath10 , the best discovery limit is obtained by adding other targets which helps at reducing the effect of the neutrino background by about an order of magnitude ( see fig .
[ fig : complementarity_sdp ] right panel ) even while the background - free sensitivity is reduced .
it is also worth recalling that the curves that consider an experiment using @xmath10 with recoil energy sensitivity have very similar neutrino background implications than cf@xmath13-based experiments such as mimac and dmtpc .
finally fig .
[ fig : complementarity_sdn ] shows the results of complimentarity for the sd neutron case . for this interaction ,
the improvement in discovery potential is intermediate between the si and sd proton case . while we have considered only models in which a wimp has one of these three types of nuclear interactions , a true wimp may have non - zero cross sections in all three of these channels .
if a potential signal is discovered near the neutrino - induced discovery limit , a more detailed scan of the parameter space would be required to disentangle the possible wimp and neutrino signals . in the previous section
, we focused on the effect of target complementarity on the discovery potential of upcoming experiments for a fixed exposure . in the following
, we will describe the effect of target complementarity on the dynamics of the discovery limit when increasing the exposure around the saturation regime .
+ figure [ fig : evolution ] shows the evolution of the discovery limit for xe ( red @xmath29 ) , xe+ge ( blue @xmath30 ) and xe+ge+si ( green @xmath31 ) based experiments considering both si ( left panel ) and sd ( right panel ) interactions for a @xmath11 wimp mass .
we can see that in the case of a si interaction , combining xe , ge and si does not allow one to greatly improve on its discovery potential .
the green points ( xe+ge+si ) still clearly show a saturation regime in the discovery potential .
however , considering a sd interaction on the proton , adding successively ge and si allows one to consequently improve the evolution of the discovery potential with exposure and keep it close to the @xmath5 best - case scenario . by comparing the xe - only and xe+ge+si points
, we can see that target complementarity allows one to reach a discovery limit of about @xmath32 with an exposure which is 50 times lower .
this clearly highlights the interest of combining data from different experiments to bypass the saturation regime induced by the neutrino background .
we considered both the si and the sd interaction on the proton . for the si interaction
there is no optimal si fraction because there is no visible effect from target complementarity .
however , for the sd interaction on the proton , the discovery limit at @xmath11 has a minimum value for a certain si fraction which depends on the considered exposure . ]
the previous results have shown combinations of several single - nucleus based experiments with a fixed relative exposure . in this section
, we will address the optimization of several target materials when designing a future experiment that would maximize its discovery potential .
we will consider the relative exposures of si and ge crystals in this example since at least one experiment ( supercdms ) can use both of these targets in their future payloads . as suggested by our previous study of the target complementarity , in the context of a si interaction
the discovery limit is only slightly affected by changing the si fraction of the total exposure .
we thus concentrate on the sd proton interaction .
figure [ fig : optimization ] shows the discovery limit for a combination of ge and si - based experiments as a function of the considered si fraction in the total exposure at a fixed wimp mass of 6 gev / c@xmath0 .
the considered thresholds for the two experiments target nuclei are taken from table [ tab : exposures ] .
we can see that for exposures which are low enough to be in the systematics - dominated regime ( below 0.2 ton - year ) , adding si in the experiment does not improve the discovery potential because the reconstructed cross section for si is about two orders of magnitude higher than the one for ge ( see fig.[fig : complementarity ] ) .
however , when the exposure is high enough to be in the discovery potential saturation regime , the complementarity of the two targets leads to an improvement of the combined discovery potential .
the optimum si fraction that maximizes the effect of the complementarity of ge and si corresponds to the minimum of these curves which has been marked by a `` @xmath31 '' .
we can therefore see that as the exposure increases , the si fraction that maximizes the discovery potential has to be higher . for a 0.4 ton - year experiment it is around 30% but for a 10 ton - year experiment
it is around 60% .
[ fig : optimization ] suggests that for very high exposures , it is worth tuning the relative exposures of ge and si detectors in order to maximize the discovery potential in light of the neutrino background .
since dark matter experiments will soon begin be sensitive to coherent neutrino scattering from solar neutrinos , the search for discrimination methods to disentangle wimp from neutrino events is a necessity . for both spin - independent and spin - dependent targets ,
we have examined the effect of target complementarity on the discovery potential of single nucleus and multi - target based experiments . in general , we find that combining different targets such as xe and ge will allow for an improvement in the discovery potential . for spin - independent interactions , we show that the similarity between the scaling of the response of wimps and neutrinos with different targets ultimately limits the gain in discovery potential achievable .
however , this situation is different for spin - dependent interactions . in this case
, the differences in the cross sections of wimps and neutrinos with different targets allows one to bypass the saturation of the discovery potential and allow the search for dark matter to be pursued to lower cross sections .
the improvement is most dramatic for spin - dependent wimp - proton interactions . in particular combining c@xmath14f@xmath15 with other targets results in a significant decrease of the impact of the neutrino background on the discovery potential .
we have also shown that the effect of target complementarity can be used to optimize the different element fractions for experiments which have more than one target material .
in particular for the sd interaction on the proton , there is an optimum si fraction for a ge@xmath29si experiment which maximizes the discovery potential .
we have discussed how this fraction depends on the exposure and increases with the exposure . in summary ,
the results that we have presented in this paper provide valuable input for future direct dark matter searches .
we have provided the first suggestion for expanding the reach of wimp dark matter searches in the regime where the neutrino backgrounds are substantial .
the methods we have discussed rely only on measuring energy depositions , and can ultimately be complementary to methods that reduce the neutrino background that rely on different experimental signatures such as annual modulation or the direction of the recoiling nucleus . + * acknowledgments* les was supported by nsf grant phy-1417457 .
fr , jb and eff were supported by nsf grant no .
the authors would like to thank laurent derome and frederic mayet for useful discussions on an earlier version of this work .
g. bertone , d. hooper and j. silk , phys .
rept . * 405 * , 279 ( 2005 ) [ hep - ph/0404175 ] .
l. e. strigari , phys .
rept . * 531 * , 1 ( 2013 ) [ arxiv:1211.7090 [ astro-ph.co ] ] .
b. cabrera , l. m. krauss , and f. wilczek , phys .
lett . * 55 * , 25 ( 1985 ) .
a. gtlein , g. angloher , a. bento , c. bucci , l. canonica , a. erb , f. v. feilitzsch and n. f. iachellini _ et al .
_ , arxiv:1408.2357 [ hep - ph ] .
r. harnik , j . kopp and p. a. n. machado , j. cosmol .
. phys . * 1207 * , 026 ( 2012 ) [ arxiv : 0706.3019 [ astro - ph ] ] . | direct detection dark matter experiments looking for wimp - nucleus elastic scattering will soon be sensitive to an irreducible background from neutrinos which will drastically affect their discovery potential . here
we explore how the neutrino background will affect future ton - scale experiments considering both spin - dependent and spin - independent interactions .
we show that combining data from experiments using different targets can improve the dark matter discovery potential due to target complementarity .
we find that in the context of spin - dependent interactions , combining results from several targets can greatly enhance the subtraction of the neutrino background for wimp masses below 10 gev / c@xmath0 and therefore probe dark matter models to lower cross - sections . in the context of target complementarity , we also explore how one can tune the relative exposures of different target materials to optimize the wimp discovery potential . |
it is well - known that the frictional resistance is independent of the apparent area of the sliding surface and it is proportional to the normal load with a proportional constant @xmath0 , which is known as the friction coefficient . traditionally , the friction coefficient has two distinct values : the static friction coefficient @xmath1 , determined from the minimum force needed to move a slider at rest , and the dynamic friction coefficient @xmath2 , used for the friction force when a steady sliding motion is established .
friction is often described using the well - known amontons - coulomb s laws : ( @xmath3 ) both @xmath1 and @xmath2 are independent of the apparent area of the contacting surfaces and the normal load ; ( @xmath4 ) both @xmath1 and @xmath2 depend on the shear characteristics of the contacting materials @xcite ; ( @xmath5 ) for most cases , @xmath2 is appreciably lower than @xmath1 .
this standard picture is still widely accepted nowadays ; however , in many cases , significant refinements must be taken into account for the explanation of the observed friction .
our common experiences tell us that the sliding of contact bodies subjected to a steady pulling velocity sometimes proceeds in an alternation of periods of rest and slide rather than moving steadily ( this is the reason , for example , for the occurrence of squeaking noises ) .
this unstable motion consists of periods of a stick state followed by a sudden slip is known as stick - slip oscillations .
the stick - slip motion occurs on both large and small scales .
understanding stick - slip phenomenon is not only important for many engineering applications , but also for understanding the mechanism of earthquakes ( which is a stick - slip phenomenon at geological scale @xcite ) .
obviously , stick - slip is caused by the variation of frictional resistance during sliding . in certain range of the velocity @xmath6
, a velocity weakening behavior ( corresponding to a decreasing function @xmath7 with respect to the steady relative sliding velocity @xmath6 of the contact surfaces ) is often observed in many different materials including metals @xcite and rocks @xcite . in this range , which generally lies in low - velocity regime ,
the steady motion is unstable with respect to perturbation ; this gives rise to stick - slip phenomena . on the other hand ,
the static friction coefficient @xmath1 is generally found to be an increasing function with respect to the contact age of contact surfaces @xcite .
this can be explained by the fact that the plastic relaxation of the stressed contact junctions during a stick period leads to an increasing real contact area , hence strengthening the contact .
the mechanism for the velocity - dependence of @xmath8 and the mechanism for the age - dependence of @xmath9 are not un - related ; both can be understood using the concept of a memory length @xcite .
the purpose of the paper is to show , by explicit numerical simulation , that the model based on the phenomenological description of ref .
@xcite can be used to explain qualitatively ( and sometimes even quantitatively ) many interesting features found in experiments .
the model incorporates the existing understanding of the age - dependence of the static friction and the velocity - dependence of the dynamic friction at low velocity , but does not make a distinction between the dynamic and static frictions .
there is no abrupt change from `` static '' friction to `` dynamic '' friction as in the traditional description of friction ; this is important for numerical simulation used to construct the entire phase diagram for the dynamics of dry friction .
the rest of the paper is organized as follows .
we first review the phenomenology of dry friction , focusing on the recent experiments by t. baumberger et al .
we then show that the phenomenological description of ref .
@xcite is equivalent to the well - known rice - ruina friction law on rock friction in the case of small deviation from the constant velocity motion @xcite .
the results from simulation of the model will be presented in the last section together with some concluding remarks .
following the pioneering work of rabinowicz on metals @xcite , many experimental studies have been performed to study the low velocity friction properties of various materials .
it is beyond the scope of the paper to review these experimental work on the dynamics of dry friction .
instead , we focus on a recent extensive experimental study of the dry friction dynamics of a paper - on - paper block - spring system ( shown in fig .
1 ) by t. baumberger et al .
their experiments verified many known properties of dry friction and gave a rather complete picture of dry friction in various regimes .
in particular , they explored systematically the dynamical phase diagram by varying the driving velocity @xmath6 , the slider mass @xmath10 and the spring stiffness @xmath11 .
the features of their experiments are summarized as follows .
* the phase diagram in control parameter space @xmath12 consists of two regions that can be characterized by stick - slip and steady sliding respectively and they are separated by a bifurcation curve .
the character of the bifurcation changes from a direct ( supercritical ) hopf type in the creep - dominated regime to an inverted ( subcritical ) one in the inertial regime as @xmath6 increases .
* in the steady sliding region , the measured dynamic friction coefficient @xmath2 exhibits velocity weakening in low - velocity range ( @xmath13 ) which can be fitted as @xmath14 where @xmath15 is an arbitrary velocity scale ; and velocity strengthening at larger velocities which can be characterized as @xmath16 * the static friction coefficient @xmath1 was found to increase with the contact age @xmath17 as @xmath18 * by introducing a characteristic memory length @xmath19 , a relation between eq . [ mu_dv ] and eq . [ mu_st ] ,
i.e. @xmath20 can be established .
here @xmath19 , which is given experimentally as @xmath21 @xcite , can be interpreted as an average sliding displacement needed to move to new micro - contacts .
these results are rather general ( rock - rock friction , for example , exhibits similar features ) .
the experiments show that the motion of the system at low - velocity ( @xmath13 ) is primarily controlled by a creep process .
based on these results , a phenomenological model of dry friction dynamics at low - velocity regime @xcite of the paper - on - paper system has been proposed . both linear @xcite and nonlinear @xcite stability analysis of the model near the bifurcation give excellent quantitative agreements with experiments . beside this , the transient behavior in the steady sliding region of the system after setting the driving velocity to zero suddenly has been studied under the aid of the model @xcite , where two - stage process has been observed . however , there is no systematic theoretical study of the entire regime covering the crossover from the `` static '' friction regime to the `` dynamic '' friction regime and the crossover to the high - velocity ( inertial ) regime . in this paper
we show numerically that most important experimental features in various regimes can be reproduced in a single model .
we follow the phenomenological approach used in ref . @xcite . consider the experimental setup shown in fig . 1 . the motion of the block under the influence of the external driving force as shown in fig . 1 , is assumed to be a thermally activated creeping motion in a periodic pinning potential biased by the external driving force . in the low bias regime , where the barrier heights of the effective potential are comparable with the thermal activation energy , the velocity of the slider is given by @xmath22 where @xmath23 is
the typical distance between two potential minima , and @xmath24 ( @xmath25 ) is the thermal time for escaping from a given well into its downstream ( upstream ) nearest neighbor .
if we let the corresponding barrier height to be @xmath26 ( @xmath27 ) , then we have @xmath28 where @xmath29 is the thermal activation energy .
@xmath29 can be written as @xmath30 , where @xmath31 is the number moles of the degrees of freedom involved in the creep motion .
@xmath32 is the oscillation frequency about the minimum of the effective potential .
the amplitude @xmath33 of the periodic pinning potential is assumed to increase with the dynamical contact age variable @xmath34 defined in ref .
@xcite , hence the barrier heights are given by @xmath35 where @xmath36 is the external force ( in the experimental setup , it is the spring force induced by the driving velocity @xmath6 ) .
although eqs .
[ dot_x ] and [ tau_inverse ] are valid strictly only for a time- and position - independent external force and a time - independent pinning potential , they can also be used in more general cases under the assumption that the changes in @xmath36 and @xmath37 are so slow that an _ adiabatic _
approximation is valid . combining the above equations we can write the velocity of the slider as @xmath38 in the case that @xmath39 , the above equation can be approximated as @xmath40 we now proceed to derive @xmath41 by considering the case of constant velocity motion ( @xmath42 ) . in this case @xmath43 , and the friction force ( given in eq .
[ mu_dv ] ) is equal to the external force @xmath36 given in eq .
[ f_ext ] .
this gives rise to @xmath44 by using the average contact age @xmath34 , we can rewrite the above equation as @xmath45 where @xmath46 . given @xmath41 , we can write the friction coefficient ( which is equal to the external force divided by the weight @xmath47 under the quasi - stationary approximation ) in term of @xmath34 @xmath48 when the velocity is not a constant , the contact age @xmath34 is assumed to satisfy the following equation @xmath49 by using this equation , the rate - dependence of the dynamic friction and the contact time dependence of the static friction can be taken into account properly .
eqs.[mu_phi ] and [ dot_phi ] form the basis for the phenomenological description of dry friction given in ref .
@xcite .
we now show that this description of friction is in fact equivalent to the state- and rate - dependent friction law proposed by rice and ruina @xcite in the case of small deviation from constant velocity motion . to make this connection
, we choose @xmath50 ( @xmath51 is the state variable in the rice - ruina theory ) , then eqs.[mu_phi ] and [ dot_phi ] become @xmath52 in rice - ruina theory , the dynamical equations for the friction coefficient @xmath0 and state variable @xmath51 in the state- and rate - dependent friction law @xcite can be expressed as @xcite @xmath53 consider the case that there is only a small deviation from the steady sliding state , we write @xmath54 expending eqs . [ dot_theta1 ] and [ dot_theta2 ] to first order of @xmath55 and @xmath56 give the same dynamical equation for @xmath57 , @xmath58 by choosing @xmath59 , @xmath60 and @xmath61 , eqs .
[ mu_theta1 ] and [ mu_theta2 ] are exactly the same .
thus these two descriptions are in fact equivalent in the case where the deviation from the steady sliding motion is small .
this is a strong indication that the phenomenological theory of dry friction discussed in ref .
@xcite is not limited to the explanation of experimental results on the paper - on - paper system , but a rather general theory that can be used as a basis for studying the dynamics involving dry friction on a range of materials .
for example , it will be useful for studying rock friction , which is important in the study of earthquake dynamics . for the case where @xmath29 is comparable with @xmath62 ( this can be thought of as in the _
static friction _
regime ) , we have to use the original equation , i.e. eq
. 8 , which can be rewritten as @xmath63 \}.\ ] ] the friction coefficient is then given as ( with quasi - stationary approximation ) @xmath64 \},\ ] ] where @xmath65 here , @xmath66 is an approximation of @xmath67 we describe in eq .
[ mu_phi ] . in this _
static friction _ regime , the velocity @xmath68 is very small and it can also change its sign , thus the equation for @xmath69 needs to be modified .
it is reasonable to simply use @xmath70 there are a few advantages of using eqs.[mu_phi_new ] , [ bar_mu ] , and [ dot_phi_new ] to describe friction force : ( @xmath3 ) there is no need to make a distinction between _
static _ and _ dynamic _ frictions , thus there is no _ stopping _ condition ( which is used to indicate the crossover from the `` dynamic '' friction regime to the `` static '' friction regime ) to worry about ; ( @xmath4 ) the contact age is a well - defined quantity with eq . [ dot_phi_new ] ; ( @xmath5 ) there is no numerical singularity in the dynamical equations for the sliding block when the velocity goes to zero and changes sign . to include the inertial regime
, we have to take into account the velocity strengthening described by eq . [ mu_d_highv ] ; we include this effect by simply adding a damping term @xmath71 in eq .
[ mu_phi_new ] .
we now present the numerical results from the simulation of the block - spring system . using the friction law described in the previous section , we can write down the dynamical equations of the block - spring system shown in fig . 1 .
@xmath72 where @xmath73 \}+\eta \dot{x},\ ] ] @xmath74 and @xmath75 we compare our result for the friction coefficient @xmath76 with the experimental results when the sliding motion is steady ( @xmath42 and @xmath77 ) .
all the coefficients in the friction law are determined experimentally for the paper - on - paper system except for the value of @xmath78 , which is determined as follows .
we consider the deflection point at @xmath79 from velocity weakening to velocity strengthening ; this is determined using @xmath80 .
this leads to @xmath81}{\sqrt{\frac{1}{4 } \exp [ \frac{2\bar{\mu}}{a}]+1}},\ ] ] where @xmath82 . substituting the experimental values of @xmath83 , @xmath84 , @xmath85 , @xmath86 @xcite and @xmath87 @xcite into eq .
[ eta_eqn ] gives @xmath88 .
using the values given above , the friction coefficient @xmath89 in steady sliding region is shown in fig . 2 , which shows an excellent agreement with experiments @xcite .
the logarithmic velocity weakening behavior of @xmath89 can be found at low velocities ( @xmath13 ) ; in this regime , the motion of the slider is creep - dominated and the effect of the damping term @xmath90 is negligible . in the high - velocity regime , @xmath89 becomes an increasing function of @xmath6 which behaves as @xmath90 asymptotically .
hence , the experimentally observed dynamical friction coefficient @xmath7 @xcite can be described by our simple expression of @xmath89 without imposing any extra condition .
we explore the dynamical phase diagram in control parameter space @xmath12 by systematically varying @xmath6 ( within @xmath91 to @xmath92 ) and @xmath11 ( within @xmath93 to @xmath94 ) with @xmath95 ( this is the value used in the experiment ) . the phase diagram is shown in fig . 3 .
stick - slip occurs below a bifurcation curve @xmath96 , and steady sliding is found above ( or to right of ) it .
the dots refer to the first steady sliding motions observed when increasing @xmath11 or @xmath6 .
the phase diagram consists of a creep - dominated regime which can be characterized by the time scale @xmath97 at low velocities and an inertial regime with a characteristic time @xmath98 at higher velocities .
the dashed line indicates @xmath99 . in the creep - dominated ( low - velocity ) regime ,
the bifurcation occurs at a constant @xmath100 ( @xmath101 ) .
4(a ) shows the bifurcation from stick - slip to steady sliding as @xmath11 is increased in this regime .
the bifurcation in the inertial ( high - velocity ) regime , on the other hand , occurs at a constant @xmath6 ( @xmath102 ) .
the bifurcation from stick - slip to steady sliding in this regime is shown in fig .
4(b ) . except for the finite slope of the experimental bifurcation curve which we are not able to reproduced
, our theoretical phase diagram agrees reasonably well with the experimental phase diagram . to investigate the nature of the bifurcation
, we have measured the amplitude and period of the stick - slip oscillation by using different @xmath11 and @xmath6 .
the results are shown in fig .
the amplitude and period decrease as the stiffness @xmath11 is increased .
the amplitude approaches zero continuously as @xmath11 becomes larger and larger ; this suggests that the bifurcation from stick - slip to steady sliding by increasing @xmath11 is a direct hopf bifurcation .
this agrees with previous theoretical and numerical analysis @xcite . on the other hand ,
the transition from stick - slip with a finite amplitude to steady sliding ( with zero amplitude ) when increasing the pulling velocity @xmath6 is rather sharp ( the transition velocity is around @xmath103 ) .
this supports that the bifurcation encountered by increasing velocity is of the inverted hopf type as was suggested previously @xcite based on experimental data ( but was not understood theoretically ) .
we have shown numerically that the inverted hopf bifurcation can be obtained within our fricition model .
we have shown that the phenomenological description of dry friction proposed by baumberger et al .
is equivalent to the friction law of rice and ruina in the case of small deviation from the constant sliding motion .
we extend the phenomenological description of baumberger et al . and construct a simple model that can be used to simulate dynamics of dry friction in various regimes .
we have shown that , by associating the thermally activated creeping motion with a damping term which is significant only when the velocity is large , the model gives rise to satisfactory agreement with the experimental phase diagram . except for the finite slope of the bifurcation curve observed in experiments , the essential features of the bifurcation in both creep - dominated and inertial regime can be reproduced using our simple model .
we believe that the model will also be very useful for numerical study of earthquake models which are often modeled using block - spring systems .
t. baumberger , f. heslot and b. perrin , nature * 367 * , 544 ( 1994 ) .
f. heslot , t. baumberger , b. perrin , b. caroli , and c. caroli , phys .
e * 49 * , 4973 ( 1994 ) .
a. l. ruina , j. geophys .
b * 88 * , 10359 ( 1983 ) .
+ j. r. rice and a. l. ruina , j. appl .
* 50 * , 343 ( 1983 ) .
f. p. bowden and d. tabor , _ friction and lubrication of solids _
( clarendon press , oxford , 1950 ) .
e. rabinowicz , _ friction and wear of materials _
( wiley , new york , 1965 ) . w. f. brace and j. d. byerlee ,
science * 15 * , 990 ( 1966 ) . c. h. scholz , _ the mechanics of earthquakes and faulting _ ( cambridge university press , cambridge , 1990 ) , chap . 2 and references therein .
j. t. oden and j. a. c. martins _ computer methods in appl .
mech . and engrg .
_ * 52 * , 527 ( 1985 ) .
t. baumberger , c. caroli , b. perrin , and o. ronsin , phys .
e * 51 * , 4005 ( 1995 ) .
t. baumberger and l. gauthier , j. phys .
i ( france ) * 6 * , 1021 ( 1996 ) . j. r. rice , j. geophys .
98 * , b6 , 9885 ( 1993 )
. _ physics of sliding friction _
b. n. j. persson and e. tosatti ( kluwer , dordrecht , 1996 ) . | we perform extended numerical simulation of the dynamics of dry friction , based on a model derived from the phenomenological description proposed by t. baumberger et al . @xcite . in the case of small deviation from the steady sliding motion ,
the model is shown to be equivalent to the state- and rate - dependent friction law which was first introduced by rice and ruina @xcite on the basis of experiments on rocks .
we obtain the dynamical phase diagram that agrees well with the experimental results on the paper - on - paper systems .
in particular , the bifurcation between stick - slip and steady sliding are shown to change from a direct ( supercritical ) hopf type to an inverted ( subcritical ) one as the driving velocity increases , in agreement with the experiments .
pacs numbers : 05.45.+b , 46.30.pa , 62.20.hg , 91.30.px |
er uma stars are a still enigmatic small subgroup of su uma - type dwarf novae [ for a review of dwarf novae , see @xcite ] , which have extremely short supercycle lengths ( @xmath0 the interval between successive superoutbursts ) of 1950 d [ for a review , see @xcite ] and regular occurrence of superoutbursts .
only five definite members have been recognized up to now : er uma ( @xcite , @xcite , @xcite ) ; v1159 ori ( @xcite , @xcite ) ; rz lmi ( @xcite , @xcite ) ; di uma ( @xcite ) ; and ix dra ( @xcite ) .
some helium - transferring cataclysmic variables have become recognized as helium counterparts " of er uma stars [ cr boo : @xcite ; v803 cen : @xcite , @xcite ] . from a theoretical side ,
er uma stars have been understood as a smooth extension of normal su uma - type dwarf nova toward higher mass - transfer rates ( @xmath1 ) @xcite .
the exact origin of such a high mass - transfer rate is still a mystery . even considering a higher mass - transfer rate , the shortest period systems ( rz lmi and di uma ) are difficult to explain without a special mechanism of prematurely quenching a superoutburst @xcite . in recent years
, there have been an alternative attempt to explain the er uma - type phenomenon .
@xcite tried to explain the er uma - type phenomenon by considering a decoupling between the thermal and tidal instabilities [ see @xcite for details of the thermal - tidal instability model ] under extremely small binary mass - ratio ( @xmath2=@xmath3/@xmath4 ) conditions .
@xcite speculated that repeated post - superoutburst rebrightenings in wz sge - type dwarf novae ( hereafter wz sge stars ) or large - amplitude su uma - type dwarf novae ( e.g. @xcite ; see @xcite for a recent observational review of wz sge - type stars ) .
@xcite tried to explain er uma - type phenomenon by ( rather arbitrary ) introducing an inner truncation of the accretion disk and irradiation on the secondary star on a numerical model developed by @xcite .
@xcite further tried to explain the unification idea by @xcite using the same scheme as in @xcite .
although the results partly reproduced the characteristics of er uma stars and wz sge stars , they failed to quantitatively reproduce the light curves of these dwarf novae . from the observational side , the existence of a gap between distributions of er uma stars and usual " su uma - type dwarf novae has been a challenge . the shortest known @xmath0 in usual su uma - type dwarf novae
had been 130 d ( yz cnc , see also table 1 in @xcite ) at the time of the initial proposition of er uma stars . although further works have slightly shortened this minimum @xmath0 [ ss umi : 84.7 d , @xcite ; bf ara : 83.4 d , @xcite ] , there still remains a undisputed gap .
in addition to these usual su uma - type dwarf novae with the shortest @xmath0 s , there exists a seemingly different population of su uma - type dwarf novae with short @xmath0 s , but with infrequent normal outbursts .
v503 cyg [ @xmath0 = 89 d , only a few normal outbursts in a supercycle @xcite ] and ci uma [ @xmath5 140 d , infrequent normal outbursts @xcite ; @xmath0 variable ?
@xcite ] are the best - known examples .
the relation , however , between these objects and er uma stars ( and short @xmath0 usual su uma - type dwarf novae ) are unknown . in most recent years
, some instances of strong @xmath0 variations have been reported in er uma stars ( @xcite , @xcite ) . in this letter
, we report on the dramatic changes in the outburst properties in v503 cyg .
we examined the observations of v503 cyg posted to vsnet collaboration , http://www.kusastro.kyoto - u.ac.jp / vsnet/@xmath6 . ] and found an appreciable change of the outburst properties .
the observations used @xmath7-band comparison stars , and typical errors of individual estimates are smaller than 0.3 mag , which will not affect the following discussion .
the object has been well sampled by many observers around the world except for periods of solar conjunctions .
table [ tab : burst ] lists the observed outbursts since 1997 .
the outburst lengths listed in the table approximately correspond to the durations above @xmath8 15.5 .
although occasional observational gaps introduced an uncertainty of a few days , most of these superoutbursts were well recorded .
many of normal outbursts during the favorably observed seasons were recorded , although some outbursts must have been missed . in order to estimate the numbers of missed outbursts , we performed monte - carlo simulations .
the fractions of missing simulated 1000 normal outbursts ( the maximum magnitude and the rate of decline have been adjusted to those of actual outbursts ) were 10% , 18% and 30% for the three representative epochs shown in figure
[ fig : lc ] .
these fractions of missing normal outbursts will not affect the discussion given in section [ sec : dis ] .
cccccccc jd start & peak mag & length & type & jd start & peak mag & length & type + 2450545 & 14.2 & 3 & normal & 2451519 & 14.4 & 2 & normal + 2450574 & 13.7 & 3 & normal & 2451531 & 13.8 & 10 & super + 2450601 & 13.1 & 14 & super & 2451641 & 14.5 & 1@xmath9 & normal + 2450643 & 14.4 & 2 & normal & 2451666 & 14.5 & 2 & normal + 2450669 & 14.1 & 3 & normal & 2451673 & 14.4 & 3 & normal + 2450697 & 13.1 & 13 & super & 2451691 & 14.2 & 3 & normal + 2450719 & 14.5 & 2 & normal & 2451704 & 13.4 & 12 & super + 2450758 & 14.3 & 1@xmath9 & normal & 2451727 & 13.7 & 2 & normal + 2450776 & 13.8 & 12 & super & 2451736 & 14.0 & 2 & normal + 2450896 & 14.3 & 2 & normal & 2451744 & 14.6 & 3 & normal + 2450952 & 13.3 & 10 & super & 2451753 & 15.6 & 1 & normal + 2450989 & 14.4 & 3 & normal & 2451785 & 13.4 & 11 & super + 2451018 & 14.1 & 3 & normal & 2451813 & 14.8 & 2 & normal + 2451051 & 13.3 & @xmath108 & super & 2451819 & 14.8 & 2 & normal + 2451072 & 14.6 & 2 & normal & 2451826 & 14.5 & 2 & normal + 2451124 & 13.4 & 12 & super & 2451837 & 14.3 & 2 & normal + 2451220 & 14.0 & 1@xmath9 & normal & 2451848 & 13.6 & 1@xmath9 & normal + 2451237 & 14.0 & 1@xmath9 & normal & 2451865 & 13.8 & 2 & normal + 2451323 & 14.5 & 3 & normal & 2451873 & 13.6 & 11 & super + 2451337 & 14.4 & 3 & normal & 2451964 & 14.7 & 1@xmath9 & normal + 2451348 & 14.3 & 1 & normal & 2452014 & 14.1 & 1@xmath9 & normal + 2451359 & 14.4 & 1@xmath9 & normal & 2452046 & 13.5 & 13 & super + 2451372 & 13.6 & 12 & super & 2452114 & 13.3 & 3 & normal + 2451406 & 14.4 & 2 & normal & 2452135 & 13.0 & 14 & super + 2451428 & 14.1 & 3 & normal & 2452176 & 13.9 & 3 & normal + 2451435 & 14.2 & 2 & normal & 2452231 & 13.4 & 10 & super + 2451453 & 13.7 & 13 & super & 2452278 & 13.7 & 2 & normal + 2451479 & 14.4 & 1@xmath9 & normal & 2452321 & 13.8 & @xmath105 & super + 2451485 & 14.4 & 2 & normal & 2452397 & 13.8 & 19@xmath11 & super + 2451495 & 14.4 & 2 & normal & 2452466 & 13.7 & 3 & normal + + + ( 88mm,130mm)fig1.eps
in the standard disk instability model , the recurrence time of normal outbursts ( @xmath12 ) is mainly governed by the diffusion process , while @xmath0 represents the increasing rate of net angular momentum in the accretion disk @xcite .
if the quiescent viscosity parameter has a fixed value between various su uma - type dwarf novae , both @xmath12 and @xmath0 are unique functions of @xmath1 @xcite .
this relation has been observationally confirmed in most of su uma - type stars @xcite .
v503 cyg apparently violates this relation in its low frequency of normal outbursts ( figure [ fig : lc ] , upper panel ) , and several other stars ( v344 lyr , sx lmi ) have been proposed to be analogous to v503 cyg ( @xcite , @xcite ) .
there must be an unknown suppression mechanism of normal outbursts in these systems . in 19992000
, v503 cyg showed a very frequent occurrence of normal outbursts ( minimum @xmath13 79 d , figure [ fig : lc ] , middle panel ) .
this @xmath12 is just what is expected for a @xmath0 = 89 d usual su uma - type dwarf nova @xcite .
this fact indicates that the usually outbursting su uma - type state and unusually outbursting ( in the sense of low frequency of normal outbursts ) v503 cyg - type state are interchangeable . since @xmath0 during this period was not appreciably different from the canonical @xmath0 = 89 d , there should have not been an appreciable change in the @xmath1 .
the suppression mechanism of normal outbursts must have been somehow unlocked " during this period . in 20012002
, v503 cyg showed another different aspect ( figure [ fig : lc ] , lower panel ) . during this period ,
the number of normal outbursts in a supercycle dramatically decreased to @xmath141 .
there is some hint of alternating occurrence of a superoutburst and a normal outburst with a period of 4080 d. such a sequence of outbursts is only known in rarely outbursting su uma - type dwarf novae [ cf .
sw uma , v844 her cf .
@xcite for a discussion ] , and is unprecedented in short @xmath0 systems . during this period ,
some normal outbursts have comparable peak magnitudes to those of superoutbursts .
some of superoutbursts showed rather short durations , which seems to be incompatible with a high @xmath1 necessary to reproduce the short @xmath0 @xcite .
these findings suggest that premature quenching of superoutbursts , as proposed by @xcite and @xcite , indeed occurred during this period , although v503 cyg ( orbital period = 0.0757 d ) is unlikely to have a small @xmath2 required in @xcite and @xcite .
the overall light curve more or less resembles that of ci uma ( @xcite , @xcite ) .
although exact mechanisms have not been yet identified , the present remarkable alternations between the outbursting states in v503 cyg support the presence of mechanisms of suppressing normal outbursts and premature quenching superoutbursts .
the most important finding is that the effects of these mechanisms are temporarily variable even in the same object , and are not a fixed character of a certain system .
this finding suggests that the shortest @xmath0 usual su uma stars and unusual v503 cyg - like stars can represent different aspects of the same system . among er uma stars , di uma can be a similar system with systematic state changes @xcite .
the observed temporal variability of the suppressing / quenching mechanisms in the same object suggests that these mechanisms are not primarily governed by a fixed system parameter [ i.e. mass of the white dwarf @xcite ; @xmath2 @xcite etc . ] but more reflect state changes in the accretion disk .
we are grateful to many amateur observers for supplying their vital visual and ccd estimates via vsnet .
this work is partly supported by a grant - in - aid ( 13640239 , tk ) from the japanese ministry of education , culture , sports , science and technology .
part of this work is supported by a research fellowship of the japan society for the promotion of science for young scientists ( mu ) . | we examined the vsnet light curve of the unusual su uma - type dwarf nova v503 cyg which is known to show a short ( 89 d ) supercycle length and exceptionally small ( a few ) normal outbursts within a supercycle . in 19992000 , v503 cyg displayed frequent normal outbursts with typical recurrence times of 79 d. the behavior during this period is characteristic to an usual su uma - type dwarf nova with a short supercycle length . on the other hand ,
v503 cyg showed very infrequent normal outbursts in 20012002 .
some of the superoutbursts during this period were observed shorter than usual .
the remarkable alternations of the outbursting states in v503 cyg support the presence of mechanisms of suppressing normal outbursts and premature quenching superoutbursts , which have been proposed to explain some unusual su uma - type outbursts .
the observed temporal variability of the suppressing / quenching mechanisms in the same object suggests that these mechanisms are not primarily governed by a fixed system parameter but more reflect state changes in the accretion disk . |
recently it has been argued that astrophysical observations might favor a dark energy ( de ) component with `` supernegative '' equation - of - state ( eos ) parameter @xmath0 @xcite , where @xmath1 is the pressure and @xmath2 the energy density of the i - th component of the cosmic fluid .
sources sharing this property violate the null dominant energy condition ( ndec ) .
otherwise , well - behaved energy sources ( positive energy density ) that violate ndec ( necessarily negative pressure ) , have eos parameter less than minus unity @xcite .
ndec - violating sources are being investigated as possible dark energy ( de ) candidates and have been called as `` phantom '' components @xcite . since ndec prevents instability of the vacuum or propagation of energy outside the light cone , then
, phantom models are intrinsically unstable . nevertheless , if thought of as effective field theories ( valid up to a given momentum cutoff ) , these models could be phenomenologically viable @xcite .
another very strange property of phantom universes is that their entropy is negative@xcite . to the number of unwanted properties of a phantom component with `` supernegative '' eos parameter @xmath3 , we add the fact that its energy density @xmath4 increases in an expanding universe .
this property ultimately leads to a catastrophic ( future ) big rip singularity@xcite that is characterized by divergences in the scale factor @xmath5 , the hubble parameter @xmath6 and its time - derivative @xmath7@xcite .
although other kinds of singularity might occur in scenarios with phantom energy , in the present paper we are interested only in the big rip kind of singularity .
this singularity is at a finite amount of proper time into the future but , before it is reached , the phantom energy will rip apart all bound structures , including molecules , atoms and nuclei . to avoid this catastrophic event (
also called `` cosmic doomsday '' ) , some models and/or mechanisms have been invoked . in ref.@xcite , for instance , it has been shown that this singularity in the future of the cosmic evolution might be avoided or , at least , made milder if quantum effects are taken into consideration . instead
, a suitable perturbation of de sitter equation of state can also lead to classical evolution free of the big rip @xcite .
gravitational back reaction effects@xcite and scalar fields with negative kinetic energy term with self interaction potentials with maxima@xcite , have also been considered .
another way to avoid the unwanted big rip singularity is to allow for a suitable interaction between the phantom energy and the background ( dm ) fluid@xcite . in effect , if there is transfer of energy from the phantom component to the background fluid , it is possible to arrange the free parameters of the model , in such a way that the energy densities of both components decrease with time , thus avoiding the big rip@xcite .
models with interaction between the phantom and the dm components are also appealing since the coincidence problem ( why the energy densities of dark matter and dark energy are of the same order precisely at present ? ) can be solved or , at least , smoothed out@xcite .
aim of the present paper is , precisely , to study models with interaction between the phantom ( de ) and background components of the cosmological fluid but , unlike the phenomenological approach followed in other cases to specify the interaction term ( see references @xcite ) , we start with a general form of the interaction that is inspired in scalar - tensor ( st ) theories of gravity .
we concentrate our investigation on solutions that are free of the coincidence problem . in this paper no specific model for the phantom energy is assumed .
a flat friedmann - robertson - walker ( frw ) universe ( @xmath8 ) is considered that is filled with a mixture of two interacting fluids : the background ( mainly dm ) with a linear equation of state @xmath9 , and the phantom fluid with eos parameter @xmath3 .
additionally we consider the `` ratio function '' : @xmath10 between the energy densities of both fluids .
it is , in fact , no more than a useful parametrization .
we assume that @xmath11 can be written as function of the scale factor . the paper has been organized in the following way : in section 2 we develop a general formalism that is adequate to study the kind of coupling between phantom energy and the background fluid ( inspired in st theories as said ) we want to investigate
. models of references @xcite can be treated as particular cases . the way one can escape one of the most undesirable features of any de model : the coincidence problem ,
is discussed in section 3 for the case of interest , where the de is modelled by a phantom fluid . in this section
it is also briefly discussed under which conditions the big rip singularity might be avoided . with the help of the formalism developed before , in section 4
, particular models with interaction between the components of the mixture are investigated where the coincidence problem is solved .
the cases with constant and dynamical eos parameter are studied separately .
it is found that , in all cases , there is a wide range in parameter space , where the solutions are free of the big rip singularity also .
it is argued that observational data seems to favor phantom models without big rip singularity .
this conclusion is reinforced by a physical argument related to the possibility that the interaction term is always bounded .
conclusions are given in section 5 .
since there is exchange of energy between the phantom and the background fluids , the energy is not conserved separately for each component .
instead , the continuity equation for each fluid shows a source ( interaction ) term : @xmath12 @xmath13 where the dot accounts for derivative with respect to the cosmic time and @xmath14 is the interaction term . note that the total energy density @xmath15 ( @xmath16 ) is indeed conserved : @xmath17 . to specify the general form of the interaction term
, we look at a scalar - tensor theory of gravity where the matter degrees of freedom and the scalar field are coupled in the action through the scalar - tensor metric @xmath18@xcite : @xmath19 where @xmath20 is the coupling function , @xmath21 is the matter lagrangian and @xmath22 is the collective name for the matter degrees of freedom .
it can be shown that , in terms of the coupling function @xmath23 , the interaction term @xmath14 in equations ( [ backconteq ] ) and ( [ phconteq ] ) , can be written in the following form : @xmath24 , \label{interactionterm}\ ] ] where we have introduced the following `` reduced '' notation : @xmath25 and it has been assumed that the coupling can be written as a function of the scale factor @xmath26 .
this is the general form of the interaction we consider in the present paper .
comparing this with other interaction terms in the bibliography , one can obtain the functional form of the coupling function @xmath27 in each case . in ref.@xcite , for instance , @xmath28 , where @xmath29 denotes the transfer strength .
if one compares this expression with ( [ interactionterm ] ) one obtains the following coupling function : @xmath30 where @xmath31 is an arbitrary integration constant . if @xmath32 and @xmath33 , then @xmath34 .
another example is the interaction term in ref .
@xcite : @xmath35 , where @xmath36 is a dimensionless coupling function .
it is related with the coupling function @xmath27 ( eqn .
( [ interactionterm ] ) ) through : @xmath37 and , for @xmath38 ; @xmath39 .
if one substitutes ( [ interactionterm ] ) in ( [ backconteq ] ) , then the last equation can be integrated : @xmath40 where @xmath41 is an arbitrary integration constant . if one considers eqn .
( [ ratio ] ) then , eqn.([phconteq ] ) can be arranged in the form : @xmath42 $ ] or , after integrating : @xmath43\ } , \label{rhoph}\ ] ] where @xmath44 is another integration constant .
using equations ( [ rhom ] ) , ( [ ratio ] ) and ( [ rhoph ] ) , it can be obtained an equation relating the coupling function @xmath27 , the phantom eos parameter @xmath45 and the `` ratio function '' @xmath11 : @xmath46 where , as before , @xmath31 is an integration constant .
therefore , by the knowledge of @xmath47 and @xmath48 , one can describe the dynamics of the model under study .
actually , if @xmath45 and @xmath11 are given as functions of the scale factor , then one can integrate in eqn .
( [ chi ] ) to obtain @xmath26 and , consequently , @xmath49 is given through eqn .
( [ rhom ] ) . the energy density of the phantom field can be computed through either relationship ( [ ratio ] ) or ( [ rhoph ] ) .
besides , the friedmann equation can be rewritten in terms of @xmath11 and one of the energy densities : @xmath50 so , the hubble parameter @xmath51 is also known . due to eqn.([friedmanneq ] ) , the ( dimensionless ) density parameters @xmath52 ( @xmath53 ) are given in terms of only @xmath11 : @xmath54 respectively .
it is useful to rewrite eqn.([chi ] ) , alternatively , in terms of @xmath55 and @xmath45 : @xmath56 another useful parameter ( to judge whether or not the expansion if accelerated ) , is the deceleration parameter @xmath57 , that is given by the following equation : @xmath58 , \label{q}\ ] ] or , in terms of @xmath55 and @xmath45 : @xmath59 . \label{q1}\ ] ] when the eos parameter of the phantom field is a constant @xmath60 , @xmath61 , we need to specify only the behavior of the ratio function @xmath48 or , equivalently , of the phantom density parameter @xmath62 . in this case the equation for determining of the coupling function @xmath27 ( [ chi ] ) or , alternatively , ( [ chialternat ] ) , can be written in either form : @xmath63 or , @xmath64 the deceleration parameter is given , in this case , by @xmath65 $ ] so , to obtain accelerated expansion @xmath66 .
in this section we discuss the way in which one of the most undesirable features of any de model ( including the phantom ) can be avoided or evaded . for completeness
we briefly discuss also the conditions under which the doomsday event can be evaded .
it is interesting to discuss , in the general case , under which conditions the coincidence problem might be avoided in models with interaction among the components in the mixture . in this sense one
expects that a regime with simultaneous non zero values of the density parameters of the interacting components is a critical point of the corresponding dynamical system , so the system lives in this state for a sufficiently long period of time and , hence , the coincidence does not arises . in the present case where we have a two - component fluid : dm plus de , one should look for models where the dimensionless energy density of the background ( dm ) and of the de are simultaneously non vanishing during a ( cosmologically ) long period of time . otherwise , we are interested in ratio functions @xmath67 that approach a constant value @xmath68 at late times
. a precedent of this idea but for a phenomenologically chosen interaction term can be found , for instance , in ref.@xcite , where a scalar field model of phantom energy has been explored ( a self - interacting scalar field with negative kinetic energy term @xmath69 ) . in this case , as already noted , the interaction term is of the form @xmath28 .
for this kind of interaction , stability of models with constant ratio @xmath70 ( and constant eos parameter ) yields to very interesting results . in effect
, it has been shown in @xcite that , if @xmath71 ( phantom - dominated scaling solution ) , the model is stable .
the solution is unstable whenever @xmath72 ( matter - dominated scaling solution ) .
one can trace the evolution of the model that evolves from the unstable ( matter - dominated ) scaling solution to the stable ( phantom - dominated ) one . as a consequence
, this kind of interaction indicates a phenomenological solution to the coincidence problem @xcite .
actually , once the universe reaches the stable phantom - dominated state with constant ratio @xmath73 , it will live in this state for a very long time .
then , it is not a coincidence to live in this long - living state where , since @xmath68 , @xmath74 . the above stability study can be extended to situations where no specific model for the phantom energy is assumed ( like in the case of interest in the present paper ) and where a general interaction term @xmath14 is considered ( in our case @xmath14 is given in eqn.([interactionterm ] ) ) , if one introduces the following dimensionless quantities@xcite : @xmath75 where the `` kinetic '' @xmath76 and `` potential '' @xmath77 terms have been considered .
besides , if one where to consider frw universe with spatial curvature @xmath78 , an additional variable : @xmath79 is needed .
the governing equations can be written as the three - dimensional autonomous system : @xmath80,\nonumber\\ y'&=&-(1-\frac{dp_{ph}}{d\rho_{ph}})[3x+\frac{q}{2h\rho_v}y]\nonumber\\ & & + 3y[2x+\frac{2}{3}z+(\omega_m+1)(1-x - y - z)],\nonumber\\ z'&=&-2z+3z[2x+\frac{2}{3}z+(\omega_m+1)\nonumber\\&&(1-x - y - z ) ] \label{dsystem}\end{aligned}\ ] ] together with the constrain @xmath81 , where the comma denotes derivative with respect to the variable @xmath82 .
the parameter range is restricted to be @xmath83 . in the remaining part of this section , for simplicity ,
we assume that @xmath84 , i. e. , the background fluid is dust . for the flat space model with constant ratio @xmath85 and eos parameter @xmath86 are shown for background dust ( @xmath84 ) . in the upper part of the figure we have considered @xmath87 ( @xmath88 , @xmath89 ) .
the point @xmath90 is an attractor . in the lower part of the figure @xmath91 ( @xmath88 , @xmath92 ) .
all trajectories in phase space diverge from an unstable node ( third critical point @xmath93 ) .
the first critical point is a saddle . in any case
the model has a late time phantom dominated attractor solution ( the second critical point @xmath94).,title="fig:",width=264 ] for the flat space model with constant ratio @xmath85 and eos parameter @xmath86 are shown for background dust ( @xmath84 ) . in the upper part of the figure we have considered @xmath87 ( @xmath88 , @xmath89 ) .
the point @xmath90 is an attractor . in the lower part of the figure @xmath91 ( @xmath88 , @xmath92 ) .
all trajectories in phase space diverge from an unstable node ( third critical point @xmath93 ) .
the first critical point is a saddle . in any case
the model has a late time phantom dominated attractor solution ( the second critical point @xmath94).,title="fig:",width=264 ] let us consider first the spatially flat case @xmath96 .
consider , besides , a constant ratio @xmath97 and a constant eos parameter @xmath98 . according to ( [ chiomegaconst ] ) ,
one obtains a coupling function of the form : @xmath99 . if one substitutes this expression for the coupling function into eqn .
( [ interactionterm ] ) , then : @xmath100 the fixed points of the system ( [ dsystem ] ) can be found once the interaction function @xmath14 given in eqn.([interaction ] ) and the expression @xmath101 are substituted into ( [ dsystem ] ) .
the fixed points of the system @xmath102 are : 1 ) @xmath103 , 2 ) @xmath104 and 3 ) @xmath105,(2+\xi^2)/[2(r_0 + 1)])$ ] . of these ,
the first point is a saddle if @xmath91 and an unstable node if @xmath87 . the second critical point is always a stable node . in this case
@xmath106 so , there is phantom ( dark energy ) dominance .
the third point is an unstable node whenever @xmath91 and a saddle otherwise . in this case @xmath107 and @xmath108 are simultaneously non vanishing so , it corresponds to matter - scaling solution . in fig .
[ phaseportrait ] the convergence of different initial conditions to the attractor ( phantom dominated ) solution is shown in the phase space @xmath109 ( flat space case ) for given values of the parameters @xmath85 and @xmath110 . if @xmath91 all trajectories in phase space diverge from an unstable node ( third critical point ) .
the first critical point is a saddle . in any case
the model has a late - time phantom dominated attractor solution ( the second critical point ) .
the critical points @xmath112 are : 1 ) @xmath113 ( de sitter space ) , 2 ) @xmath114 , 3 ) @xmath115,(2+\xi^2)/[2(r_0 + 1)],0)$ ] and 4 ) @xmath116 ( milne space ) .
the first point is always a saddle , meanwhile , the second critical point is always a stable node ( a sink ) .
the third critical point is a saddle whenever @xmath87 and an unstable node ( a source ) otherwise .
this point corresponds to matter - scaling solution where both @xmath117 and @xmath55 are simultaneously non negligible .
the fourth critical point ( milne space ) corresponds to curved ( @xmath111 ) vacuum space ( @xmath118 ) .
this point is an unstable node if @xmath119 and a saddle whenever @xmath120 .
note that , even if the evolution of the universe begins with initial conditions close to the fourth critical pint ( curved flat space ) , the model drives the subsequent evolution of the universe to the stable node corresponding to a flat universe dominated by the phantom component . in summary ,
the way to evade the coincidence problem when an interaction term of the general form ( [ interactionterm ] ) is considered , is then clear : one should look for models where , whenever expansion proceeds , both @xmath45 and the ratio @xmath121 , approach constant values @xmath122 and @xmath68 , respectively . then , if trajectories in phase space move close to the third critical point ( matter - scaling solution ) , these trajectories will stay close to this point for an extended ( arbitrarily long ) period of time , before subsequently evolving into the phantom dominated solution ( second critical point ) that is a late - time attractor of the model .
if we assume expansion ( the scale factor grows up ) then , to avoid the future big rip singularity , it is necessary that the background fluid energy density @xmath123 and the phantom pressure @xmath124 and energy density @xmath4 , were bounded into the future .
once the `` ratio function '' @xmath11 ( eqn.([ratio ] ) ) and the parametric eos for the phantom fluid ( @xmath125 ) are considered , the above requirement is translated into the requirement that @xmath4 , @xmath126 and @xmath127 were bounded into the future .
let us consider , for illustration , the case with constant eos : @xmath98 , and the following ratio function : @xmath128 @xcite .
integrating in the exponent in eqn.([chiomegaconst ] ) one obtains the following expression for the coupling function : @xmath129 the expression for the dm and phantom energy densities are then given through equations ( see equations ( [ rhom ] ) and ( [ ratio ] ) ) : @xmath130 and @xmath131 where @xmath132 and @xmath133 .
we see that , for @xmath134 , @xmath135 , while @xmath136 . for @xmath137 the dm energy density @xmath123 decreases with the expansion of the universe .
however , the phantom energy density always increases with the expansion and , consequently , the big rip singularity is unavoidable in this example .
the conclusion is that , for a phantom fluid with a constant eos @xmath138 and a power - law ratio of phantom to background ( dust ) fluid energy densities @xmath139 , the big rip singularity is unavoidable . for a constant ratio @xmath97 and a constant eos ( @xmath98 ) @xcite , direct integration in eqn .
( [ chiomegaconst ] ) yields to the following coupling function : @xmath140 so , the background and phantom energy densities in eqn .
( [ rhom ] ) are given by : @xmath141 where @xmath142 $ ] .
we see that , whenever there is expansion of the universe , for @xmath143 , there is no big rip in the future of the cosmic evolution , since @xmath144 are bounded into the future . in order to get accelerated expansion , one should have @xmath119 so , the ratio of background ( dust ) energy density to phantom energy density should be in the range @xmath145 .
( dashed curve ) and @xmath55 ( thin solid curve ) as well as of the ratio function @xmath11 ( thick solid curve ) vs redshift @xmath146 , is shown for the model with constant eos parameter .
for simplicity background dust ( @xmath84 ) was considered .
the values of the free parameters are : @xmath147 , @xmath148 and @xmath149 .
the ratio function @xmath11 approaches a constant value at negative redshift @xmath150 . at present ( @xmath151 )
@xmath152 , which means that we are already in the long living matter - scaling state .
the background energy density dominates the early stages of the cosmic evolution . at @xmath153
both density parameters equate and , since then , the phantom component dominates the dynamics of the expansion .
a very similar behavior is obtained for @xmath117 , @xmath55 and @xmath11 in the case with dynamical eos parameter ( case b).,width=264 ]
in the present section we look for a model with an appropriated coupling function @xmath27 that makes possible to avoid the coincidence problem .
we study separately the cases with a constant eos parameter @xmath154 and with variable eos parameter . in order to have models where both @xmath117 and @xmath55 are simultaneously non - vanishing during a long interval of cosmological time ( no coincidence issue ) and that , at the same time , are consistent with observational evidence on a matter dominated period with decelerated expansion , lasting until recently ( redshift @xmath155 @xcite ) , we choose the following dimensionless density parameter for the phantom component : @xmath156 where @xmath157 , @xmath158 , and @xmath159 are arbitrary constant parameters .
this kind of density parameter function can be arranged to fit the relevant observational data , by properly choosing the free parameters ( @xmath160 ) ( @xmath157 controls the curvature of the curve @xmath161 , meanwhile @xmath159 controls the point at which @xmath162 ) . in this sense ,
the model under study could serve as an adequate , observationally testable , model of the universe . the choice of @xmath55 in equation ( [ omegaph1 ] ) , when substituted in eqn .
( [ chialternatomegaconst ] ) yields to : @xmath163 in consequence , equations ( [ rhom ] ) and ( [ ratio ] ) lead to the following expressions determining the background energy density and the energy density of the phantom component respectively ; @xmath164 where @xmath165 , @xmath166 and @xmath167 where @xmath168 .
it is worth noting that , if @xmath169 , there is no big rip singularity in the future of the cosmic evolution as described by the present model . in effect , for large @xmath170 , the following magnitudes @xmath171 are bounded into the future . in particular , the phantom energy density is a decreasing function of the scale factor . note that the ratio function @xmath172 approaches the constant value @xmath173 so both @xmath117 and @xmath55 are simultaneously non vanishing .
as explained in the former section , this means that the universe is in the third critical point ( matter - scaling solution ) which is a saddle for @xmath87 . in consequence ,
the universe will evolve close to this critical point for a sufficiently long period of time and then , the model approaches a stable dark energy dominated regime ( second critical point ) .
is plotted as function of the redshift @xmath146 for the model a ( constant eos parameter ) , for three different values of @xmath174 and @xmath175 ( thick solid , thin solid and dashed curves respectively ) .
note that the transition from decelerated ( positive @xmath176 ) into accelerated ( negative @xmath176 ) expansion occurs at @xmath177 for the value @xmath149 .
it is apparent that this is , precisely , the observationally favored solution , since , according to a model - independent analysis of snia data , @xmath178@xcite . in consequence , solutions without big rip are preferred by observations.,width=264 ] in fig .
[ densityp1 ] we show the behavior of the density parameters @xmath117 and @xmath55 and of the ratio function @xmath11 as functions of the redshift ( for simplicity we have considered background dust : @xmath84 ) .
the following values of the free parameters : @xmath179 have been chosen , so that it is a big rip free solution .
besides we set the de density parameter at present epoch @xmath180 so , the following relationship between the free constant parameters @xmath158 , @xmath157 and @xmath159 should be valid : @xmath181 $ ] . in consequence , for the values of @xmath157 and @xmath159 given above
: @xmath182 so , @xmath183 , meaning that there is no big rip singularity in the future of the cosmic evolution in this model ( see the condition @xmath184 under eqn .
( [ rhoph1 ] ) ) . in fig .
[ decelerationp1 ] , the plot of the deceleration parameter @xmath176 vs redshift is shown for three different values of the eos parameter @xmath174 and @xmath175 respectively .
it is a nice result that the solution without big rip ( thick solid curve ) is preferred by the observational evidence , since , according to a model - independent analysis of snia data@xcite , the mean value of the present value of the deceleration parameter is @xmath185 .
the way the coincidence problem is avoided in this model is clear from fig.[densityp1 ] also .
the ratio between the energy density parameters of the background and of the de approaches a constant value @xmath150 for negative redshift so , both @xmath117 and @xmath55 are non negligible . as already said , this is a critical point ( third critical point in the stability study of section 3 ) and the universe can live for a long time in this state .
moreover , since at present ( @xmath151 ) ; @xmath186 , we can conclude that the universe is already in this long living state .
( upper part of the figure ) , phantom component @xmath4 ( middle part ) and of the hubble parameter @xmath6 vs redshift , is shown for three values of the constant parameter @xmath122 : @xmath187 ( thick solid curve ) , @xmath188 ( thin solid curve ) and @xmath175 ( dashed curve ) for the model with dynamical eos parameter .
it is apparent that , only in the first case the model is big rip free .
note that , due to the values chosen for the free parameters , in particular @xmath147 , the phantom energy density vanishes as one approaches the initial big bang singularity .
this is true whenever @xmath189 ( see eqn .
( [ rhoph2])).,title="fig:",width=264 ] ( upper part of the figure ) , phantom component @xmath4 ( middle part ) and of the hubble parameter @xmath6 vs redshift , is shown for three values of the constant parameter @xmath122 : @xmath187 ( thick solid curve ) , @xmath188 ( thin solid curve ) and @xmath175 ( dashed curve ) for the model with dynamical eos parameter .
it is apparent that , only in the first case the model is big rip free .
note that , due to the values chosen for the free parameters , in particular @xmath147 , the phantom energy density vanishes as one approaches the initial big bang singularity .
this is true whenever @xmath189 ( see eqn .
( [ rhoph2])).,title="fig:",width=264 ] ( upper part of the figure ) , phantom component @xmath4 ( middle part ) and of the hubble parameter @xmath6 vs redshift , is shown for three values of the constant parameter @xmath122 : @xmath187 ( thick solid curve ) , @xmath188 ( thin solid curve ) and @xmath175 ( dashed curve ) for the model with dynamical eos parameter .
it is apparent that , only in the first case the model is big rip free .
note that , due to the values chosen for the free parameters , in particular @xmath147 , the phantom energy density vanishes as one approaches the initial big bang singularity .
this is true whenever @xmath189 ( see eqn .
( [ rhoph2])).,title="fig:",width=264 ] is plotted vs redshift @xmath146 .
note that for @xmath190 , @xmath14 is negative , meaning that there is transfer of energy from the dm into de component . for higher @xmath146
the model proceeds without interaction .
@xmath14 is bounded into the future only for the big rip free case ( thick solid curve).,width=264 ] we now consider a dynamical eos parameter @xmath47 . in this case
we should give as input the functions @xmath191 and @xmath192 .
in consequence , one could integrate in the exponent in eqn.([chialternat ] ) so that the dynamics of the model is completely specified . in order to assure avoiding of the coincidence problem ,
let us consider the same phantom energy density parameter as in the former subsection ( eqn . ( [ omegaph1 ] ) ) , but rewritten in a simpler form : @xmath193 where @xmath157 , @xmath194 and @xmath195 are non negative ( constant ) free parameters . to choose the function @xmath192 one should take into account the following facts : i ) at high redshift and until recently ( @xmath196 @xcite ) the expansion was decelerated ( positive deceleration parameter @xmath176 ) and , since then
, the expansion accelerates ( negative @xmath176 ) , ii ) for negative redshifts the eos parameter approaches a constant value @xmath122 more negative than @xmath197 ; @xmath198 .
we consider , additionally , the product @xmath199 to be a not much complex function , so that the integral in the exponent in eqn.([chialternat ] ) could be taken analytically . a function that fulfills all of the above mentioned requirements is the following : @xmath200 where @xmath36 is another free parameter .
the parameter @xmath157 controls the curvature of the density parameter function @xmath161 , while @xmath36 controls the point of equality @xmath162 .
the resulting coupling function is @xmath201 therefore , the background and phantom anergy densities are given by the following expressions : @xmath202 \;a^{-3(\omega_m+1)}\nonumber\\ \times\;(a^{2m}+\delta)^{-\frac{3\alpha}{2m}\omega_{ph,0 } } \;e^{\frac{3\alpha\sqrt\delta}{m}\arctan(\frac{a^m}{\sqrt\delta } ) } , \label{rhom2}\end{aligned}\ ] ] @xmath203 \;(a^{2m}+\delta)^{-\frac{3\alpha}{2m}\omega_{ph,0}}\nonumber\\ \times\;e^{\frac{3\alpha\sqrt\delta}{m}\arctan(\frac{a^m}{\sqrt\delta } ) } , \label{rhoph2}\end{aligned}\ ] ] where the constant @xmath204 . at late times
( large @xmath5 ) , the functions @xmath205 .
this means that , whenever @xmath206 , these functions are bounded into the future and , consequently , there is no big rip in the future of the cosmic evolution in the model under study . and
the deceleration parameter @xmath176 are plotted as functions of the redshift @xmath146 ( model b ) , for three values of the constant parameter @xmath174 and @xmath175 ( thick solid , thin solid and dashed curves respectively ) .
it is worth noting that at high redshift @xmath45 approaches a constant value @xmath207 .
it is a minimum at @xmath208 .
the dark energy became a phantom fluid only recently , at @xmath209 .
note that the transition from decelerated ( positive @xmath176 ) into accelerated ( negative @xmath176 ) expansion occurs at @xmath177 for the value @xmath149 .
this last ( big rip free ) solution , seems to be preferred by observations since , according to ref .
@xcite , @xmath210 . in the lower part of the figure the ratio function
is plotted for three values of the parameter @xmath157 : @xmath211 ( thick solid curve ) , @xmath212 ( thin solid curve ) and @xmath213 ( dashed curve ) respectively . in any case , at present , the universe is already in the state with constant ratio @xmath11.,title="fig:",width=264 ] and the deceleration parameter @xmath176 are plotted as functions of the redshift @xmath146 ( model b ) , for three values of the constant parameter @xmath174 and @xmath175 ( thick solid , thin solid and dashed curves respectively ) .
it is worth noting that at high redshift @xmath45 approaches a constant value @xmath207 .
it is a minimum at @xmath208 .
the dark energy became a phantom fluid only recently , at @xmath209 .
note that the transition from decelerated ( positive @xmath176 ) into accelerated ( negative @xmath176 ) expansion occurs at @xmath177 for the value @xmath149 .
this last ( big rip free ) solution , seems to be preferred by observations since , according to ref .
@xcite , @xmath210 . in the lower part of the figure the ratio function
is plotted for three values of the parameter @xmath157 : @xmath211 ( thick solid curve ) , @xmath212 ( thin solid curve ) and @xmath213 ( dashed curve ) respectively . in any case , at present , the universe is already in the state with constant ratio @xmath11.,title="fig:",width=264 ] and the deceleration parameter @xmath176 are plotted as functions of the redshift @xmath146 ( model b ) , for three values of the constant parameter @xmath174 and @xmath175 ( thick solid , thin solid and dashed curves respectively ) .
it is worth noting that at high redshift @xmath45 approaches a constant value @xmath207 .
it is a minimum at @xmath208 .
the dark energy became a phantom fluid only recently , at @xmath209 .
note that the transition from decelerated ( positive @xmath176 ) into accelerated ( negative @xmath176 ) expansion occurs at @xmath177 for the value @xmath149 .
this last ( big rip free ) solution , seems to be preferred by observations since , according to ref .
@xcite , @xmath210 . in the lower part of the figure the ratio function
is plotted for three values of the parameter @xmath157 : @xmath211 ( thick solid curve ) , @xmath212 ( thin solid curve ) and @xmath213 ( dashed curve ) respectively . in any case , at present , the universe is already in the state with constant ratio @xmath11.,title="fig:",width=264 ] in fig.[tresenuno ] the plot of the energy densities of dm @xmath123 ( upper part of the figure ) , phantom component @xmath4 ( middle part ) and of the hubble parameter @xmath6 vs redshift , is shown for three values of the constant parameter @xmath122 : @xmath187 ( thick solid curve ) , @xmath188 ( thin solid curve ) and @xmath175 ( dashed curve ) .
the following values of the free parameters : @xmath147 , @xmath214 , and @xmath215 have been chosen .
we assume @xmath216@xcite so , the following relationship between @xmath194 and @xmath195 should take place : @xmath217 .
it is apparent that , only in the first case ( @xmath218 ) the model is big rip free . actually ,
as already noted , the necessary requirement for absence of big rip is , for the chosen set of parameters , @xmath219 .
it is noticeable that , since we chose @xmath220 , the de energy density diminishes as one goes back into the past ( it is a maximum at present ) . in all cases
the interaction term is negative during a given period of time in the past ( see fig.[q2 ] ) , meaning that the dm transferred energy to the de component . at higher
redshift the evolution proceeds without interaction .
worth noting that only in the singularity free case ( @xmath218 ) the interaction term is bounded into the future . a physical argument against doomsday event
can be based on the following analysis . at late time , the interaction term @xmath221 .
this means that , only in the case when the evolution proceeds without big rip ( @xmath123 and @xmath6 are bounded ) , the interaction between the components of the cosmic fluid does not become unphysically large at finite time into the future . besides , for the big rip free case , as the expansion proceeds into the future , the hubble parameter approaches a negligible value , meaning that the ratio @xmath222 is small . in the model
the transition from decelerated into accelerated expansion takes place at @xmath177 ( see fig .
[ deceeos ] ) . at late times
the deceleration parameter approaches the constant value @xmath223 .
therefore , as in the case with constant eos parameter ( model a ) , the observational evidence ( as argued in ref .
@xcite , @xmath210 ) seems to favor solutions without big rip singularity .
it is interesting to note that , in this model , the dark energy component behaved like dust ( @xmath207 ) at early times ( high @xmath146 ) , i. e. , it behaved like an `` ordinary '' fluid with attractive gravity and , just until very recently ( @xmath224 ) , it has not became a phantom fluid . at present , the universe is already in a state characterized by constant @xmath225 and @xmath226 , meaning that @xmath227 and @xmath228 are simultaneously non vanishing . as shown before , this is a critical point of the corresponding dynamical system so , probably , the universe will stay in this state for a long time and , consequently , the coincidence problem does not arise .
in this paper we have investigated models with additional ( non gravitational ) interaction between the phantom component and the background .
this kind of interaction is justified if the interacting components are of unknown nature , as it is the case for the dm and the de , the dominant components in the cosmic fluid . for the sake of simplicity , baryons , a non vanishing , but small component of the background ,
have not been considered in the present study , however , in a more complete study these should be included also .
a comment should be made in this regard : there are suggestive arguments showing that observational bounds on the `` fifth '' force do not necessarily imply decoupling of baryons from de@xcite . unlike the phenomenological approach followed in other cases to specify the interaction term ( see references @xcite ) , we started with a general form of the interaction that is inspired in st theories of gravity ( see eqn.([interactionterm ] ) ) .
we have considered no specific model for the phantom fluid . within this context ,
cosmological evolutions without coincidence problem have been the target of this investigation .
we have studied two different models : model a , where the de eos parameter is always a constant ( phantom de ) and model b , where we considered a dynamical de eos parameter ( the de became a phantom just recently ) . in these models there
is a wide range in the parameter space where the cosmic evolution is free of the unwanted big rip singularity also .
the interaction between the different components of the cosmic mixture , enables the cosmic evolution to proceed without big rip singularity . in effect , if the interaction term is chosen so that there is transfer of energy from the phantom component to the background , then it is possible to arrange the free parameters of the model to account for decreasing energy density of the phantom fluid ( as well as of the background fluid ) . a physical argument against big rip
can be based on the analysis of the interaction term @xmath14 . in model b , for instance , at late time the interaction term @xmath229 .
this means that , only whenever @xmath123 and @xmath6 are bounded ( no big rip ) , the interaction between the components of the cosmic fluid does not become unphysically large at finite time into the future . besides , for the big rip free case , as the expansion proceeds into the future , the hubble parameter approaches a negligible value , meaning that the ratio @xmath222 is small . from the observational perspective it seems that the evolution without doomsday event is preferred also
for this purpose we have considered model - independent analysis of snia data yielding to a mean value of the present value of the deceleration parameter @xmath230@xcite .
interacting models of dark energy are also useful to account for the coincidence problem . in the present study , it has been shown that , a solution with simultaneously non vanishing values @xmath227 and @xmath228 ( @xmath85 is the constant ratio of the dm to de densities ) , is a critical point of the corresponding dynamical system ( third critical point in section 3 ) .
hence , if trajectories in phase space get close to this point , these trajectories will stay there during a sufficiently long period of time .
this means that , once the universe is driven into the state with simultaneously non vanishing @xmath117 and @xmath55 , it will live in this state for a ( cosmologically ) long period of time until it is finally attracted into the phantom dominated phase being a stable node of the models ( second critical point in section 3 ) .
this is the way the coincidence problem is solved in the present investigation .
it seems that we live already in the matter - scaling regime ( third critical point ) and will live in it for long time until , finally , the phantom dominates the destiny of the cosmic evolution . in any case
the evolution of the universe will proceed without the risk of any catastrophic ( doomsday ) destiny .
an obvious limitation of the present study is that only two specific models ( model a and model b ) have been considered .
however , we think this suffices to illustrate how interacting models can be constructed that avoid the coincidence problem and , at the same time , fit some of the observational evidence .
the outcome that big rip free models are preferred by some of these observations is just a nice result .
we are very grateful to alan coley and sigbjorn hervik for valuable conversations that helped us to improve the original version of the manuscript .
we thank marc kamionkowski and bret mcinnes for calling our attention upon references @xcite and @xcite respectively .
we thank also joerg jaeckel , sergei odintsov , vakif onemli , diego pavon and thanu padmanabhan for useful comments and for pointing to us important references .
the authors thank the mes of cuba by partial financial support of this research .
u. alam , v. sahni , t. d. saini and a. a. starobinsky , mon . not .
* 354 * ( 2004 ) 275 ( astro - ph/0311364 ) ; u. alam , v. sahni and a. a. starobinski , jcap * 0406 * ( 2004 ) 008 ( astro - ph/0403687 ) ; b. feng , x. l. wang and x. zhang , phys . lett . b*607 * ( 2005 ) 35 - 41 ( astro - ph/0404224 ) ; t. r. choudhury and t. padmanabhan , astron
. astrophys .
* 429 * ( 2005 ) 807 ( astro - ph/0311622 ) ; h. k. jassal , j. s. bagla and t. padmanabhan , mon . not .
letters * 356 * ( 2005 ) l11-l16 ( astro - ph/0404378 ) .
s. nojiri and s. d. odintsov , phys .
lett . b*595 * ( 2004 ) 1 - 8 ( hep - th/0405078 ) ; phys .
rev . d*70 * ( 2004 ) 103522 ( hep - th/0408170 ) ; e. elizalde , s. nojiri and s. d. odintsov , phys .
rev . d*70 * ( 2004 ) 043539 ( hep - th/0405034 ) .
w. zimdahl , d. pavon and l. p. chimento , phys .
lett . b*521 * ( 2001 ) 133 - 138 ( astro - ph/0105479 ) ; l. p.
chimento , a. s. jakubi , d. pavon and winfried zimdahl , phys .
d*67 * ( 2003 ) 083513 ( astro - ph/0303145 ) ; l. p. chimento , a. s. jakubi and d. pavon , phys . rev .
d * 62 * ( 2000 ) 063508 ( astro - ph/0005070 ) ; l. p. chimento , a. s. jakubi and d. pavon , phys .
rev . d*67 * ( 2003 ) 087302 ( astro - ph/0303160 ) . | models of the universe with arbitrary ( non gravitational ) interaction between the components of the cosmic fluid : the phantom energy and the background , are investigated .
a general form of the interaction that is inspired in scalar - tensor theories of gravity is considered .
no specific model for the phantom fluid is assumed .
we concentrate our investigation on solutions that are free of the coincidence problem .
we found a wide region in the parameter space where the solutions are free of the big rip singularity also .
physical arguments , together with arguments based on the analysis of the observational evidence , suggest that phantom models without big rip singularity might be preferred by nature . |
polarization - based measurements are increasingly important , both as a fundamental tool for scientific research @xcite , and as a vital tool for applications in the chemical , food , and pharmaceutical industries .
indeed , proposed classical communications systems , fluorescence measurements , non - invasive blood - sugar measurements , lidar , imaging through scattering media , strain , and temperature sensors all have implementations that rely on polarization based measurements @xcite . in surface science , it is well known that polarization based measurements are very sensitive .
for instance , ellipsometry has been used to measure sub - monolayer changes in surface coverage @xcite . as such
, it is a nearly - ideal technique for monitoring modern epitaxial fabrication techniques @xcite and for contamination monitoring of vacuum components ( e.g. , synchrotron optics ) @xcite .
the sensitivity of ellipsometry comes from the ability to set and measure polarization states with a high degree of accuracy . as a result
, it is generally thought that , in order to make polarization - based measurements , one must know the polarization state of the probing light field throughout the optical train , and , especially , the polarization state of the light incident on the surface of interest .
the corollary to this is that the use of optical components that disturb the polarization ( e.g. , optical fibers ) require calibration , and , in the case of optical fibers , where temperature and stress induced birefringence have a large influence on the polarization state , this is not possible in all but the most limited circumstances .
this is because , in contrast to what the name suggests , polarization maintaining ( pm ) fibers do not preserve arbitrary polarization states @xcite .
for instance , if linearly polarized light is injected at an arbitrary polarization orientation with respect to the axes of birefringence of the fiber , then the output is an elliptically polarized state with an unpredictable orientation and degree of ellipticity . as a consequence , fiber - based ellipsometry and polarimetry
have relied on rather complicated experimental apparatus @xcite , use a wavelength that is only supported by one polarization mode in a pm fiber @xcite , or only provide qualitative information @xcite .
the core discovery that we present in this paper is that , given an input polarization state of light and an optical fiber that is subject to environmentally induced birefringence variations , the output state does not map to every point on the poincare sphere , but rather , only a discrete set of polarization states are accessible .
as the temperature , for instance , varies in time , the set of output polarization states appear as a fixed orbit on the surface of the poincare sphere . in our work , we show that the existence of such orbits , as an intrinsic property of the fiber , can be seen as a higher dimensional type of polarization preservation that survives severe environmental perturbances . once this orbit is known , polarization - based measurements can be made by measuring deviations from the orbit .
this is very different from direct calibration , where an input polarization state is mapped to a single output polarization state for a known set of environmental parameters . instead
, a single input polarization state is mapped to a set of output states as the environment varies over some ( generally unknown ) range of temperatures and stresses .
this approach turns a measurement problem environmental noise into an advantage by using the statistical properties of the noise to provide increased sensitivity . in this paper , we present experimental data and modeling results that demonstrate and make use of the observation that the polarization follows a random path , but , importantly , remains bound to well - defined subspace . as an example of how such knowledge can be applied ,
we demonstrate a fiber - based ellipsometer .
we show that this ellipsometer is capable of detecting a well - defined carbon layer that is less than 1 nm thick on top of a multilayer bragg reflecting mirror .
the experimental setup is shown schematically in fig .
[ setup ] . as a light source
, we used a helium - neon laser that emitted light that was nominally linearly polarized .
the polarization of the light field was purified to a linearly polarized state , of which the azimuth angle was set to be parallel to the plane of the optical table , using a wollaston prism .
the wollaston prism ensures that the polarization purity is 100000:1 .
the input polarization state to the fiber was controlled using a soleil - babinet compensator or a quarter - wave plate , which were manually controlled using precision rotation and translation stages .
this allowed the individual relative stokes vector intensities to be set within 0.02% of a desired value . after setting the polarization ,
the laser beam was passed through a non - polarizing beam splitter , before being coupled into a 2 m long pm - fiber ( thorlabs , pm-630-hp ) using a glass aspheric lens .
light was coupled out of the output end of the fiber and collimated with an identical coupling lens .
the output fiber end was mounted on a manual rotation stage to set the angle of incidence with respect to the sample s plane of reflection .
the accuracy of the angle of incidence on the sample was found to be 0.5@xmath2 . in all measurements ,
the middle section of the fiber ( about 1.3 m ) was coiled up on a metal cylinder and placed in a water bath with a heater to simulate environmental changes by varying the temperature between 19 and 30@xmath3c .
only a small fraction of this range ( 2 - 3@xmath3c ) was used during the experiments , though , because that was sufficient to obtain the full range of accessible polarization states ( see below ) .
after reflection from the sample , the polarization state of the light was measured using a rotating compensator , which consisted of a quarter waveplate , rotating at an angular frequency of 0.1 hz , a wollaston prism , and a si pin photodiode ( pm 1 and 2 in fig .
[ setup ] ) .
the photodiode voltage was observed on an oscilloscope and transferred to a computer for analysis .
the photodiode signals were evaluated to obtain the polarization in terms of the stokes vector components @xcite .
the polarization states are calculated as stokes vectors and presented graphically in either 2d or 3d projections of the stokes vector space . in order to restrict our analysis to the polarization properties of the pm fiber and sample , we compensate for the ( small ) modification of the polarization state due to the non - polarizing splitter with a predetermined matrix .
ideally , neither the pm fiber nor the sample are changing , and the frequency components of the photodiode signal should only contain the dc component and two harmonics of the rotating compensator s angular frequency . however , this is true only when the polarization state is constant during the measurement period . in our case
, the fiber s output states are constantly changing and , thus , allows a small non - zero amplitude for additional frequency components . by analyzing the amplitude of these components ,
the direct polarization state measurement noise was estimated to be 0.02@xmath2 ( for the ellipsometric parameter @xmath1 ) , which is comparable to a conventional ellipsometer @xcite .
the measured polarization states of the output of the pm fiber are plotted in fig .
[ polarization - sphere - exp](a ) as zero - dimensional points on the two - dimensional surface of the poincare sphere , where all data are , unless it is specifically stated otherwise , normalized so that the total intensity is unity .
the notable feature is that despite the wide variation in temperature , the polarization states are not observed to fall randomly on the two - dimensional poincare surface . instead , a single , one - dimensional orbit is traced out , which is shown more clearly by taking a cross section through the sphere to obtain a projection of the orbit in a stokes vector plane ( fig . [ polarization - sphere - exp](b ) ) .
the analysis of the orientation of this orbit is key to performing polarization - based measurements without precise knowledge of the polarization of the probing light field . + to show that this polarization orbit can be used as a generalized , higher dimensional polarization state and be used to make polarization - based measurements , we show how the two ellipsometric parameters , @xmath0 and @xmath1 can be calculated from changes to the polarization orbit .
we begin by considering light exiting a fiber in a set of polarization states denoted by @xmath4 where the subscript and superscript denote the polarization coordinate systems of the sample ( @xmath5 ) and the fiber ( @xmath6 ) .
an isotropic reflecting surface transforms the polarization state of incident light according to the following mueller matrix in the sample coordinate system @xcite @xmath7 \label{iso refle surf}\ ] ] where @xmath8 and @xmath9 and @xmath10 are intensity reflectivities for the @xmath11 and @xmath12 polarized light respectively .
the initial states become @xmath13 in other words , the ellipsometric information of the sample is carried in a global transformation from @xmath4 to @xmath14 the matrix @xmath15 is a direct sum of the matrices of its two block diagonal subspaces .
the ( desired ) values of @xmath16 and @xmath17 can , thus , be obtained separately .
the set of polarization states of the light field incident on , and reflected by the sample are given by @xmath18 , & \mathcal{r}_{xy } = \left[\begin{array}{c } r_{1}\\ r_{2}\\ r_{3}\\ r_{4 } \end{array}\right ] \end{array}\ ] ] substituting equations [ eq : stokes ] into equation [ eq : transform ] gives @xmath19 dividing eqs .
[ eq : rforpsi1 ] by [ eq : rforpsi2 ] and rearranging gives : @xmath20 where @xmath21 and @xmath22 .
@xmath1 is obtained by requiring consistency between equations [ eq : rfordelta1 ] and [ eq : rfordelta2 ] .
note that equations [ eq : rfordelta1 ] and [ eq : rfordelta2 ] have the form of a scaling factor and a rotation , which , when applied to an ellipse , reduces its area and changes the orientation of its major axis in stokes space . in practice , since the polarization orbit , when projected in the plane of the 3rd and 4th stokes vectors , is an ellipse , @xmath1 is the angle difference between the orientations of the ellipses given by @xmath4 and @xmath23 .
it is important to note that our analysis assumes that all states are measured in the @xmath6 coordinate system .
slight changes of alignment between the coordinate system of the fiber and the sample introduces an extra rotation to the polarization state that depends on the misalignment angle , @xmath24 . in that case
, it can be shown that the systematic error added to @xmath0 and @xmath1 is of the order of @xmath25 .
provided that the physical alignment of the pm - fiber is held sufficiently constant , the uncertainty in @xmath1 and @xmath0 should be comparable to that of conventional ellipsometry . to demonstrate the applicability of this analysis , we performed fiber - based ellipsometry on three multilayer bragg reflecting samples @xcite , two of which have been coated with amorphous hydrogenated carbon @xcite .
spectroscopic ellipsometry ( wollam m2000 ) revealed that the carbon layers on these samples are 0.3 , and 0.8 nm thick , while the third , supposedly uncoated sample , has @xmath260.1 nm naturally occurring carbon layer with a different composition to that of the first two samples .
both conventional and fiber - based ellipsometry measurements were performed at an angle of incidence of 66@xmath3 .
the raw ellipsometric data from the fiber - based ellipsometer , using the experimental procedure described described above , is shown in fig .
[ carbon samples](a ) .
the measured orbits have been projected onto the plane of the third and fourth stokes vector components ( the inset shows the three dimensional representation of the data ) . as illustrated by the zoomed in section , shown in fig .
[ carbon samples](b ) , the data for each sample are systematically modified by the sub - nanometer layers of carbon to generate three different ellipses .
the data for each sample show very little deviation from the fitted ellipse . + using the analysis described above
, we obtain the @xmath0 and @xmath1 values presented in tab .
[ psi and delta values ] .
these values agree well with those obtained from a conventional ellipsometer .
the small systematic difference between the values from the two instruments is due to a small difference in the angle of incidence between the two measurements ( the accuracy of the angle of incidence is @xmath270.5@xmath3 ) . .@xmath0 and @xmath1 values for fiber - based ellipsometric measurements ( columns 2 and 3 ) and conventional ellipsometer measurements ( columns 4 and 5 ) on carbon - coated mlms [ cols="^,^,^,^,^",options="header " , ] [ psi and delta values ]
to understand why pm fibers that are subject to varying temperature and strain , produce a polarization orbit , a numerical model of a non - ideal pm fiber was developed .
in simple terms , the fiber is a strongly birefringent material with a time and space - dependent beat length , and some degree of polarization mode cross - talk . in this model ,
the orientation of the axes of birefringence and the refractive index difference between the two axes changes as a function of position along the fiber and time . to analyze the effect of such a fiber
, we break the fiber up into ideal segments , @xmath28 .
mathematically , this corresponds to rotating the coordinate system of the fiber for light entering a fiber segment and then reversing the rotation upon exiting .
the phase delay experienced by the light field then depends on its orientation relative to the rotated coordinate system .
this introduces a polarization change that is purely due to the relative phase delay of the components of the electric field along the local axes of birefringence .
the transfer matrix of the entire fiber is then given by a series of matrix multiplications @xcite : @xmath29 @xmath30 and @xmath31 are rotations due to an angular ( @xmath32 ) misalignment of the axes of birefringence of two neighboring ideal segments , while @xmath33 is the matrix transfer function due to the phase delay . the phase delay is expressed as a function of the beat length , @xmath34 , normalized to the segment length .
both @xmath34 and @xmath35 are time dependent in a constrained but random manner . in these simulations ,
the maximum limits of @xmath32 and changes to @xmath36 were given by the yield strength of glass the strain required to introduce these variations should not break the fiber . to allow direct comparison between the model results and experimental results , the average variation in @xmath32 was determined by specifying a maximum polarization cross - talk per unit length .
as the phase change varies as a function of position and time in the fiber , the output polarization states form a simple , circular orbit in stokes space .
the effect of cross - talk is to introduce deviations from a simple orbit in stokes space , instead , a more complicated , but repeating , orbit is obtained . numerically evaluating equation [ fiber cross talk ] under conditions
that the maximum total cross talk is 20 db , which is the maximum specified cross - talk for our fiber , shows ( see fig .
[ model - results ] ) that , even in the presence of a small amount of cross - talk , the polarization orbit can still be approximated by a simple orbit ( the elongation of the red markers in fig . [ model - results ] indicate the deviation from a simple orbit ) .
importantly , these simulations indicate that , as long as we can quantify the polarization orbit of an optical component , then it will be possible to use light exiting that component for polarization - based measurements .
it should be noted there is the ( slim ) possibility that the polarization change induced by the sample is a rotation in the plane of the polarization orbit . in this case , the sample is undetectable .
this problem can be avoided by making measurements at more than one angle of incidence . for a total polarization mode cross talk of 20 db .
time - dependent temperature changes cause changes to the local degree and orientation of the fiber s birefringence .
the resulting output is an orbit on the poincare sphere.,width=226 ] in conclusion , we have demonstrated that sensitive polarization - based measurements are possible even when the polarization state of the probing light field has been passed through an optical element with an unpredictably fluctuating birefringence .
this is possible because , for a fixed input polarization , the environmentally induced fluctuations in the output polarization after propagation through an optical element or material still lie in a single orbit of the poincare sphere , creating what we have termed a generalized , higher dimensional polarization state .
polarization - based measurements can then be made by analyzing changes to the higher dimensional state , rather than the individual polarization states that make up the higher dimensional state . to illustrate this , we demonstrated a fiber - based ellipsometer , capable of detecting carbon layers with a thickness of 0.3 nm .
fl , cjl , and fb acknowledge funding from m2i and `` controlling photon and plasma induced processes at euv optical surfaces ( cp3e ) '' of the `` stichting voor fundamenteel onderzoek der materie ( fom ) '' , which is financially supported by the `` nederlandse organisatie voor wetenschappelijk onderzoek ( nwo ) '' .
the cp3e programme is co - financed by carl zeiss smt and asml . | we show that , under the right conditions , one can make highly accurate polarization - based measurements without knowing the absolute polarization state of the probing light field .
it is shown that light , passed through a randomly varying birefringent material has a well - defined orbit on the poincare sphere , which we term a generalized polarization state , that is preserved .
changes to the generalized polarization state can then be used in place of the absolute polarization states that make up the generalized state , to measure the change in polarization due to a sample under investigation .
we illustrate the usefulness of this analysis approach by demonstrating fiber - based ellipsometry , where the polarization state of the probe light is unknown , and , yet , the ellipsometric angles of the investigated sample ( @xmath0 and @xmath1 ) are obtained with an accuracy comparable to that of conventional ellipsometry instruments by measuring changes to the generalized polarization state . |
our knowledge about the structure of elliptical galaxies has progressed considerably in the last decade .
no longer are we satisfied with fitting a de vaucouleurs ( 1948,1953 ) @xmath0 intensity distribution to the light profiles of elliptical galaxies .
modern imaging techniques are providing ever - more precise measurements , revealing a range of light profile shapes which have been shown to be better described by the sersic ( 1968 ) @xmath5 model ( capaccioli 1989 ; caon , capaccioli & donofrio 1993 ; graham et al .
1996 , and references within ) . through varying the parameter @xmath17 in the sersic model
, one can reproduce an exponential light distribution ( @xmath17=1 ) , the classic de vaucouleurs profile ( @xmath17=4 ) , approximate a power - law for large values of @xmath17 ( approximately @xmath1815 ) , or any of the intermediate forms which galaxies exhibit ( michard 1985 ; schombert 1986 ; binggeli & cameron 1991 ; caon et al .
1993 ; graham & colless 1997a , hereafter paper i , to mention a few ) .
consequently , for those galaxies which are not well described by an @xmath0 profile , the effective radius and mean surface brightness within this radius will be different when obtained using the de vaucouleurs model and the sersic model .
more importantly , these differences have been shown to have a systematic trend : galaxies with @xmath17@xmath194 have their effective half - light radii over - estimated by the @xmath0 model , and galaxies better described with values of @xmath17@xmath184 have their effective half - light radii under - estimated by the @xmath0 model .
similarly , the mean surface brightness is under- and over - estimated by the @xmath0 model when @xmath17@xmath194 and @xmath17@xmath184 respectively ( graham & colless 1997b )
. observed variations in the galaxy structure , as described by the @xmath5 model , imply a variety of dynamical structures ( ciotti 1991 ; paper i ) are required to maintain these predominantly pressure - supported stellar systems .
this conjecture is borne out in the observations of elliptical galaxy dynamics ( davies et al .
1983 ; illingworth 1983 ; capaccioli & longo 1994 ) and has implications for the construction of the fundamental plane ( fp ) .
in particular , the use of a galaxy s central velocity dispersion ( cvd , @xmath3 ) to represent the overall kinetic energy of a galaxy has been shown to be inappropriate , ( jrgensen , franx & kjrgaard 1993 ; bender , saglia & gerhard 1994 ) .
the velocity dispersion derived from the kinetic energy of random motion over the entire galaxy surface has been shown to be proportional to @xmath20 rather than @xmath21 ( busarello et al .
a procedure is described in paper i , based upon the theoretical study of ciotti ( 1991 ) , which enables the velocity dispersion to be measured in a consistent fashion between galaxies . rather than simply using cvd measurements , that sample different fractions of different galaxies velocity dispersion profiles ( in terms of the half - light radii ) , through application of the jeans equation , paper i presented a method to compute the asymptotic value of the velocity dispersion within a projected aperture of infinite radius .
this measure , denoted by @xmath22 , has the added significance of being equal to one - third of the virial velocity for spherical systems ( ciotti 1994 ) .
in paper i the light profile data ( aperture magnitudes from bower , lucey & ellis 1992 ) from a sample of 26 e / s0 virgo galaxies was analysed .
departures from the @xmath0 model were found to be well - described by the sersic model , and the implied non - homology in the velocity dispersion profile was explored . using the sersic model parameters , rather than @xmath0 model parameters , and using the computed , infinite aperture velocity dispersion rather than the cvd , the fp was observed to change from @xmath1@xmath23@xmath24 to @xmath25@xmath23@xmath26 .
broken structural and dynamical homology appeared to be responsible for some of the _ tilt _ of the fp , i.e. its departure from the plane predicted by the virial theorem , @xmath27@xmath23@xmath28 .
ultimately , this broken dynamical and structural homology needs to be dealt with if one is to understand the nature of elliptical galaxies and use them for measuring distances ( dressler et al .
1987 ) . given the substantial improvements in describing elliptical galaxies , it is apt that the fp , and its application to distance estimates , is revisited . using a new data base ( taken from caon et al .
1994 ) obtained by different observers with a different instrument and using a different reduction procedure , the influence of broken structural and broken dynamical homology upon the fp is again investigated .
the data used in this paper is the surface brightness profiles along the major axis of each galaxy , as opposed to the integrated aperture magnitudes used in paper i. the influence of s0 galaxies upon the slope of the fp is also investigated , with comparison of the fps constructed with , and without , their inclusion .
this paper presents the fp for the galaxy clusters fornax and virgo , two key clusters in the extra - galactic distance scale ( jacoby et al .
1992 ; mcmillan , ciardullo & jacoby 1993 ; bureau , mould & staveley - smith 1996 ) .
these clusters are sufficiently distant to possess significant motion due to the hubble expansion and yet they are close enough that their distances may be obtained with a plethora of independent observational techniques . the following section presents the galaxy sample and the observational / model data which is used in this study .
the fundamental plane is computed , and its method of construction , including a new error analysis routine , is described in section 3 . using two different methods of analysis , a distance modulus between virgo and fornax
is calculated in section 4 .
this result is compared with other estimates of the virgo - fornax distance modulus in section 5 , along with a discussion of the other results .
section 6 provides a summary and concluding remarks .
the galaxy sample from caon et al .
( 1994 ) has been used for this investigation .
this data set has been used in a benchmark study ( caon et al .
1993 ) which gave insight into the nature of elliptical galaxy light profiles by quantifying their deviations from the @xmath0 model . knowing that this galaxy sample does not posses structural homology amongst its members makes it an ideal sample to explore the necessarily implied broken dynamical homology and the combined influence of these effects upon the fp . from the sub - sample of galaxies for which major - axis profiles have been modeled ( see caon et al .
1993 for a discussion of some of the problems in doing this ) , those which have no published velocity dispersion data have been excluded here .
this reduced the sample of virgo galaxies from 33 ( caon et al .
1993 ) to 25 , and the sample of fornax galaxies from 16 ( donofrio , capaccioli & caon 1994 ) to 10 .
the 10 fornax galaxies were further reduced to a sample of 9 after the exclusion of ngc 1399 , which , with its shallow light profile ( @xmath17@xmath1815 ) , could not have its velocity dispersion profile accurately modeled due to the large uncertainties associated with the effective radius @xmath29 .
the remaining galaxies are listed in table 1 .
the b band surface brightness profiles from caon et al .
( 1994 ) have been used in this study .
they were constructed by these authors using the ` global mapping ' procedure of capaccioli & caon ( 1989 ) which couples ccd data for the bright inner galaxy profile with photographic data for the outer profile .
this technique allows very deep photometry of galaxies , and alleviates the problem of uncertain sky subtraction errors that is associated with the use of small area ccd s .
the method gives internal errors of less than 0.1 mag at @xmath30=26 mag . in this study
the major - axis surface brightness profiles have been truncated at the same inner and outer limits that were applied by caon et al .
( 1993 ) and donofrio et al .
( 1994 ) for the virgo and fornax galaxies respectively . in general
, this meant fitting the entire light profile after the exclusion of the inner portion affected by seeing , and the outer portion where sky subtraction uncertainties became large ( @xmath180.25 mag ) .
this left a typical range in surface brightness from @xmath30=20 to 27 mag .
the sersic ( 1968 ) @xmath5 light profile model was fitted to the remaining profile , with the intensity @xmath31 given as a function of radius @xmath27 such that @xmath32\right].\ ] ] @xmath33 is the intensity at the radius @xmath34 , and @xmath35 ( @xmath361.9992n-0.3271 , capaccioli 1989 ) is defined such that @xmath34 is the radius that effectively encloses half of the total light of the profile model ( caon et al .
1993 ; graham & colless 1996 ) .
the ` shape parameter ' @xmath17 describes the curvature of each light profile .
application of a non - linear least - squares fit of the @xmath5 model to the galaxy surface brightness data , with equal weight for each data point , produced similar values for the shape parameters to those derived by caon et al .
( 1993 ) and donofrio et al .
( 1994 ) , confirming that the models are being fitted consistently between authors .
table 1 lists the parameters from the best fitting @xmath5 model and also those from the best fitting @xmath0 model , which was fitted in an identical fashion .
@xmath37 is the mean surface brightness within the effective radius @xmath29 , computed using the technique described in appendix a of paper i. the subscripts @xmath17 and 4 on the model parameters are used to denote the model from which the particular parameters have come ; @xmath17 signifying @xmath5 model parameters , and 4 signifying @xmath0 model parameters .
ciotti ( 1991 ) , and also paper i , explored the relationship between the projected and spatial ( deprojected ) quantities . by definition
, the surface brightness terms had a one - to - one mapping , while there was an approximately constant offset ( @[email protected] , ciotti 1991 ) between the effective radii .
therefore either of these quantities could be used for the construction of the fp .
the deprojected velocity dispersion at the effective radius was found not to have such a simple relation with its projected counterpart , the central velocity dispersion measurement .
this deprojected velocity dispersion is preferred over the cvd measurement because it not only allows for the range of velocity dispersion profiles present , but also samples different galaxies in a consistent fashion ( cvd measurements sample typically anywhere between 0.01 - 0.2 @xmath29 ) .
however , due to the physical significance of the projected velocity dispersion within an aperture of infinite radius , this term , which also samples different galaxies in a consistent fashion , is preferred in the analysis which follows .
following caon et al .
( 1993 ) ( their fig .
3 ) , the light profile shape parameter is plotted against galaxy scale - size in figure [ fig1 ] . although our methods of parameterisation differ ( caon et al .
1993 have used model - independent estimates for the half - light radius ) , the same trend of increasing values of @xmath17 with @xmath29 is evident here .
the present study also includes fornax galaxies as well as those from virgo .
also shown in the figure are the 99% confidence intervals for the model parameters .
these were computed by evaluating the chi - square values for a fine grid of points around the optimum solution .
the contours drawn correspond to @xmath38=9.21 ( i.e. 4@xmath39 ) , scaled after normalisation of the reduced @xmath40 of the optimum solution . in exploring the sensitivity of the @xmath5 model parameters to the radial range of the surface brightness profile ,
the sersic model was fitted to that data within the inner 40@xmath41 .
this outer truncation is the same as the outermost data points used in paper i , enabling a further comparison of the present data with that used in paper i. the inner truncation used by caon et al .
( 1993 ) and donofrio et al .
( 1994 ) was again employed , except for ngc 4473 and ngc 4636 for which only the inner 3@xmath41 was excluded . in figure
[ fig2 ] , the resulting model parameters are shown plotted against those obtained using the entire surface brightness profile ( table 1 ) .
it is noted that the light profiles from the smaller radial extension ( @xmath1940@xmath41 ) may not be representative of the galaxy as a whole , and therefore may not represent the _ over - all _ global galaxy shape .
this is more likely to be true for the larger galaxies , especially those having @xmath29@xmath1840@xmath41 .
it is also true that the sersic index @xmath17 is less sensitive at larger values ( graham et al .
thus , figure [ fig2 ] offers an alternative estimate for the uncertainty of the @xmath5 model parameters .
some of the more discrepant points have been labeled in the plot .
it is noted that they are predominantly the s0 galaxies .
if one is to think of these galaxies as possessing an @xmath0 ( @xmath17=4 ) central bulge and an exponential ( @xmath17=1 ) outer disk , or some approximately equivalent sersic form , then this could account for the larger index obtained using the smaller radial range for the s0 galaxies ngc 1380 , 4339 , and 4550 .
in fact , such measurements offer themselves as possible diagnostics for distinguishing s0 galaxies from ellipticals . for the fornax galaxies , the tabulated extinctions range from @xmath42=-0.05 to -0.09
. some of the virgo galaxies also have negative extinctions , with a mean virgo cluster value of @[email protected] ( standard deviation ) for the sample .
the negative values are due to the variable ratio of h@xmath43 gas - to - dust in the galaxy ( burstein & heiles 1978 ) , and have no physical meaning .
such values arise in the method used by burstein & heiles for regions where there is zero or negligible reddening , and they should be set equal to 0.00 .
corrections for galactic extinction ( burstein & heiles 1984 ; burstein et al .
1987 ) were not applied by caon et al .
( 1994 ) . this practice has been maintained here . at a redshift of around 0.003
, the k - correction and redshift dimming will be equally small for both clusters , at 0.015 mag and 0.013 respectively .
consequently these terms have not been incorporated into the analysis .
cvds are typically used as a measure of the kinetic energy within elliptical and s0 galaxies .
there are many problems associated with doing this .
firstly , rotational energies , which are likely to be significant for s0 galaxies , are ignored .
secondly , fixed aperture sizes ( on the sky ) in which the cvd is measured correspond to different physical sizes for galaxies in clusters at different distances .
this effect alone may likely introduce systematic errors into cluster distance determinations which have incorporated fixed aperture sizes to measure the cvd .
thirdly , given that galaxies have a range of sizes and structural profiles , they will also have a range of dynamical profiles which support the various structures that are observed .
therefore , even within clusters , fixed aperture sizes will sample different fractions of individual galaxies , as measured in terms of their effective radii . in an attempt to find a better measure of the kinetic energy within each galaxy , a method of computing the velocity dispersion in a consistent manner for all galaxies
was presented in paper i. deprojecting the galaxy light profile , the spatial luminosity density ( ciotti 1991 ) was computed . assuming a constant @xmath44 ratio within each galaxy , such knowledge of the internal density profile coupled with the application of the jeans hydrodynamical equation enables one to compute the internal velocity dispersion profile .
this can then be projected outward to give the projected radial velocity dispersion profile , which in turn can be integrated over increasing aperture sizes to give the projected aperture velocity dispersion profile .
this is then calibrated using the measured cvd .
the computed spatial ( deprojected ) velocity dispersion at the spatial half - light radius was used in paper i to represent the kinetic energy of the galaxies .
unlike use of cvd , this approach consistently samples the velocity dispersion profile in the same way for all galaxies .
as an alternative , the computed velocity dispersion within an aperture of infinite radius was also used .
this quantity has been shown by ciotti ( 1994 ) to be equal to one - third of the virial velocity dispersion , independent of any orbital anisotropies that may be present .
this measure for the kinetic energy is adopted here , not only for the consistent fashion in which galaxies are sampled but also due to the significance of the infinite aperture velocity dispersion @xmath22 .
these values are listed in table 1 , where the subscripts 4 and @xmath17 refer to use of the @xmath0 and @xmath5 models respectively .
also listed are the cvd measurements , @xmath3 , from mcelroy ( 1995 ) .
[ cols="<,<,>,>,>,^,>,>,^ , > " , ] the two techniques described in feigelson & babu ( 1992 ) for estimating the offset of parallel data sets have both been implemented here .
these methods are subject to the condition that the virgo and fornax elliptical galaxy population define fps which are parallel to each other . given the uncertainties on the derived fp slopes ( table [ tab7 ] ) , the fornax cluster data does indeed define a plane which is consistent with being parallel to the virgo cluster data .
this is true when @xmath0 model parameters and cvd s are used , and also when @xmath5 model parameters and infinite aperture velocity dispersions are used to construct the fp . in the following method of analysis ,
a linear regression is performed on one of the cluster data sets , and the resulting regression line from this _ calibration _ is then applied to the second ( parallel ) cluster .
one technique for measuring the intercept offset between parallel data sets was pioneered by working & hotelling ( 1929 ) and further developed by feigelson & babu ( 1992 ) to permit the @xmath45-values ( i.e. distance - independent values ) to be random variables rather than fixed quantities . with this method ,
the calibrated regression line for virgo was applied individually to the data points in fornax and a weighted mean offset attained ; the method is discussed in section 3.2 of feigelson & babu ( 1992 ) and their appendix a2 .
this procedure encompasses the fact that the confidence intervals about the @xmath46-value one is trying to predict , given its @xmath45-value , will be larger for those data points at the ends of the distribution .
this is because the uncertainty in the slope of the calibration line / plane gives greater freedom of movement to the calibration line at the ends of the distribution than in the center .
this method of analysis gave an intercept offset , along the @xmath12 axis between the fp s defined by the virgo and fornax data sets , of @[email protected] dex and [email protected] dex for the ( @xmath0,@xmath3 ) parameters and the ( @xmath5,@xmath48 ) parameters , respectively .
in a similar fashion , the fornax cluster has been used as the calibration sample , where the fitted regression line for fornax is applied to the virgo cluster .
this was done with the working - hotelling confidence bands . using the ` standard ' fp parameters ( @xmath49,@xmath1,@xmath2 ) , an intercept offset of
+ [email protected] dex was computed . using
the ` improved ' fp parameters ( @xmath6,@xmath25,@xmath50 ) , an intercept offset of + [email protected] dex was computed .
these intercept offsets translate into a distance modulus of + [email protected] mag ( using @xmath0 model parameters and @xmath3 ) and + [email protected] mag ( using @xmath5 model parameters and @xmath48 ) . the former estimate placing fornax some ( 23@xmath158)% further away than the virgo cluster , compared to the latter estimate of ( 12@xmath156)% .
these solutions are fully consistent with those obtained using the virgo cluster as the calibration sample ( see table [ tab8 ] ) .
an alternative approach for computing the offset of parallel data sets was achieved following the methods in section 3.1 of feigelson & babu ( 1992 ) and their appendix a1 .
the fp data set from the virgo cluster of galaxies was used to calibrate the _ regression line _ which was then applied to the fp data set of the fornax cluster as a whole ( and _ vice - versa _ ) .
this was done in order to derive the intercept offset , along the @xmath12 axis , between the fp s defined by the virgo and fornax clusters .
the solutions obtained were similar , within 0.001 dex , to those calculated using the working - hotelling confidence bands , although the associated uncertainties were generally equal or larger .
this served as a useful consistency check on the determination of the cluster - cluster distance modulus . with the working - hotelling confidence bands ,
the uncertainty on the intercept offset , and hence the distance estimate , is reduced by the square - root of the sample size one is applying the calibrated solution to .
consequently , applying the fornax _ regression _ line to virgo , with a sample size of 18 , results in a more accurate solution than achieved by applying the virgo _ regression _ line to fornax , with a sample size of only 7 . however , this effect does have to compete with the fact that the uncertainty on the regression line is larger for a smaller calibration sample . as the generalised working - hotelling confidence bands propagate the errors in the calibrated line , this effect can not be avoided , and the size of these errors influence the associated errors for the intercept offset .
the lack of symmetry for the error estimates when using virgo or fornax as the calibrator is due to following : the error estimates on the slope of the regression line , which were propagated in the above analysis , were derived from the analytical expression for these terms .
however , the errors on the slope and intercept will be under - estimated by the analytical expressions when the number of data points is small .
this is seen in the fp solutions shown in table [ tab7 ] , where the jackknife and bootstrapping methods of analysis provide a better estimate of the true size of these errors .
consequently , the error estimates on the previously given distance moduli should be multiplied by approximately the ratio of the bootstrapping error ( or jackknife error if this is larger ) to the asymptotic error ( feigelson & babu 1992 ) .
in addition , it should be noted that the fp slope uncertainties in table [ tab7 ] were estimated when the mixing value , @xmath51 , was held fixed and assumed to have zero error . it should be remembered that it is the offset between two parallel planes , not two parallel lines , that is being measured . in order to fully estimate the uncertainties on the derived distance modulus , one must allow for errors in the optimal fp slopes for all axes . in so doing , one should return to the bootstrapping errors obtained with the 3-dimensional data set ( table [ tab6 ] ) .
working in 3-dimensions , the errors associated with the fp exponent @xmath52 , are obtained by also letting the exponent @xmath53 ( and hence @xmath51 ) vary , as permitted by the bootstrapping analysis of the data .
therefore , the error obtained in the 3-dimensional analysis represents the true error estimate of the exponent @xmath52 . allowing for this ,
the errors to the solution for the virgo - fornax distance modulus are again multiplied by the ratio of the 3-d bootstrapping errors to the 2-d asymptotic errors .
this procedure does not effect the optimal solution , only the associated uncertainties .
the resulting solutions place fornax ( 23@xmath1544)% ( @xmath0,@xmath3 ) , and ( 12@xmath1536)% ( @xmath5,@xmath6 ) further away than virgo .
in the interest of statistical completeness , galaxies with a cvd of less than 100 km s@xmath54 were not explicitly removed from the analysis . due to the greater scatter in the fp below this value
, past studies have enforced this arbitrary cut - off in their galaxy sample .
however , it seems plausible to attribute the higher level of scatter to an inadequate treatment of the kinetic energy for these galaxies . as this is an issue which is addressed in this study , it made sense to leave these galaxies in the analysis .
contributions from rotational energy to the total kinetic energy of a galaxy can in principle influence the slope of the fp .
djorgovski & santiago ( 1993 ) noted that gains from including a rotational energy measurement would be offset by the introduction of additional measurement errors .
indeed , a marginal effect has been reported many times ( simien & prugniel 1992 ; busarello , longo & feoli 1992 ; prugniel & simien 1994 ) . an asymmetric trend between the galaxy fp residuals and their maximum rotational velocity has also been found by donofrio et al .
more quantitative estimates from prugniel & simien ( 1996a , b ) , using a sample of 400 galaxies , claim that rotation can account for 25@xmath1515% of the departure of the fp from the virial expectation . while this figure may be true for fp s constructed using early - type galaxies ( e and s0 ) , which includes almost all fp s constructed to date , this figure can be reduced by the exclusion of s0 galaxies , known to have rotating disks .
this was hinted at in the study by busarello et al .
( 1997 ) , who used only _ bona fide _ ellipticals to obtain a contribution of 15@xmath55% to the tilt to the fp due to galaxy rotational energies . in this paper ,
the fp was constructed using a culled set of galaxies , removing the s0s and keeping the es .
contributions from possible rotational support to the kinetic energy term are expected to be minimal in this study of the fp .
four of the nine s0 galaxies which were excluded did actually have cvd s less than 100 km s@xmath54 , leaving only one galaxy in the remaining sample of 25 ellipticals with a cvd less than 100 km s@xmath54 .
it may indeed be that the increased scatter observed for the fp at the low velocity dispersion end is due to the presence of s0 galaxies that harbour rotating disks with significant rotational energies that are simply not measured through cvd measurements .
for the virgo cluster , exclusion of these objects resulted in a reduction of some 15 - 35% in the scatter about the fp , as evidenced by the rms residuals listed in table [ tab2 ] and table [ tab3 ] .
for the virgo cluster , exclusion of the s0 s caused the final fp solution ( incorporating the observed range of galaxy profiles and implied dynamics ) to change from @xmath7 to @xmath10 .
this suggests that the inclusion of the s0 population not only increases the scatter about the fp ( see tables [ tab2 ] and [ tab3 ] ) but is also partly responsible for the departure of the fp from the relation @xmath8 , predicted by the virial theorem .
this is again easily understood as due to the neglect of rotational energies in the construction of the observed fp .
it is therefore suggested that in addition to treating broken structural and dynamical homology , the construction of the fp always adopt the procedure of only using _ bona fide _ elliptical galaxies , unless one allows for rotational energies .
it is found that by addressing the range of structural and dynamical profiles evident amongst the elliptical galaxy population , the departure of the fp from the plane predicted by the virial theorem is reduced , confirming the results in paper i. in addition , the scatter of the data points about the fp is reduced by some 20% for the sample of fornax ellipticals and by the same amount for the sample of virgo e and s0 galaxies .
for the sample of virgo ellipticals , the mean rms scatter about the fp is reduced by 40% after treating the range of galaxy structures and dynamics .
this is evident in tables 2 , 3 and 4 , which give the best fitting fp relation when one does and does not treat broken structural homology , and when one does and does not consistently measure the kinetic energy of the galaxies ( i.e. uses @xmath22 or cvd s , respectively ) .
the first row of these 3 tables gives the fp which was constructed assuming that all galaxies are described by an @xmath0 profile and uses cvd s to represent the kinetic energy of each galaxy ; this procedure gives the greatest departures of the fp from the virial plane .
the final row in each table gives the fp that was constructed allowing for a range of structural profiles amongst the galaxies , as described by the @xmath5 model , and measures the kinetic energy of each galaxy in a consistent manner by using the computed velocity dispersion within an aperture of infinite radius ; this procedure gives the closest agreement between the fp and the virial plane .
in fact , the formal solution for the fp of the fornax cluster ( @xmath11 ) is found to be in very close agreement with the prediction of the virial theorem , but does have large uncertainties associated with it because of the small number ( 7 ) of fornax elliptical galaxies in the sample .
overall , it is concluded that addressing the range of structural profiles amongst the elliptical galaxies and better treating the galaxy dynamics reduces both the departure of the fp from the virial plane and the scatter about the fp .
in paper i the sersic light profile model was fitted to a sample of 26 e / s0 virgo galaxies using the aperture magnitude profile data from bower et al .
( 1992 ) . of these ,
18 galaxies are in common with the sample of 25 used here .
comparison between the galaxy model parameters obtained in each study is not straight - forward , however .
the galaxy profile data used here is in the form of surface brightness measurements along the major - axis , where as paper i uses circular aperture magnitudes .
consequently , the models fitted in paper i are more heavily weighted by the inner profile , and by necessity incorporate the central parts of the galaxy light . by contrast
, the treatment of the surface brightness profiles here gives all data points along the profile equal weight , and does not fit the central portion of the profile .
in addition , the use of aperture magnitude data samples light from all position angles , and in so doing represents a mean light distribution for the galaxy as a whole .
the presence of ellipticity gradients will result in the major- and minor - axis light profiles being described by different shape parameters so that @xmath56@xmath57@xmath58 .
this is clearly evident in the range of shape parameters , @xmath17 , for the major- and minor - axis , shown in the study of the virgo ellipticals by caon et al .
consequently , the mean light distribution obtained through the use of circular aperture magnitudes should not be expected to be identical to the light distribution along the major - axis . finally , the data of bower et al .
( 1992 ) was taken in the v - band , and that of caon et al .
( 1993 ) in the b - band .
possible colour gradients , although expected to be small , may also contribute to differing results . despite the above factors ,
there is still fair agreement between the sersic model parameters obtained using the different data sets .
galaxy scale - size and light profile shape for those galaxies in common between this paper and paper i are displayed in figure [ fig3 ] .
for comparison between the data sets , the sersic model was fitted to the same radial extent ( @xmath1940@xmath41 ) for each data set .
this outer radius cut - off comes from the aperture magnitude profile limit of bower et al .
( 1992 ) , and has been applied , for this comparison , to the major - axis profiles of caon et al .
the correlation seen between the shape parameters has a linear correlation coefficient of 0.55 , which increases to 0.82 upon removal of the 3 more discrepant points that are labeled in figure [ fig3 ] .
the correlation coefficient between the effective half - light radii is stronger at 0.93 , suggesting that the range of galaxy sizes , and structures , can be _ largely _ reproduced from both types of observational data .
the galaxies which appear to have the more discrepant shape parameters are the flattened s0 galaxy ngc 4550 , ngc 4486 , and ngc 4564 which has a value of @xmath17=4.64 ( paper i ) lying between @xmath56=1.51 and @xmath58=5.08 ( caon et al .
1993 ) and is therefore not a discrepant point after all .
despite the above differences , it is interesting to compare the resulting fp s obtained using the independent data sets .
firstly , assuming structural homology and using @xmath0 model parameters and cvd measurements for the virgo cluster of galaxies , the fp was found in paper i to be described by @xmath59 .
this is to be compared with @xmath4 found in this study .
the exponents on the velocity dispersion term are consistent within the 1@xmath39 joint errors , and the exponents on the surface brightness term are consistent within the 2@xmath39 joint errors .
allowing for broken structural homology and consistently measuring the velocity dispersion within an aperture of infinite radius , it was found in paper i that the fp for virgo is described by @xmath60 .
this should be compared with @xmath61 .
the relative consistency between the exponents is the same as before .
it is noted that the uncertainties on the fp exponents are based upon the statistics of the linear regression to the data points ; they do not incorporate possible measurement errors in the data points themselves .
it is also noted that the sample of virgo galaxies used in paper i contains 7 galaxies that this study does not include , and this study also contains 7 galaxies not present in the sample of paper i. using only those 18 galaxies in common , the resulting fp s were found to be in acceptable agreement with each other , with the fp described by the data in paper i given by the relation @xmath62 , and that obtained using this studies data given by the relation @xmath63 .
the main point is that both studies show that by treating the range of luminosity and dynamical structures inherent amongst the early type galaxies the departure of the fp from the virial plane is reduced .
donofrio et al . (
1996 ) also constructed a fp for the virgo and fornax clusters using the data of caon et al .
the approach taken in this study differs in several ways to that of donofrio et al .
this work has used model - determined effective radii and the model - determined mean surface brightness within these radii , where as donofrio et al .
( 1996 ) used model - independent values for these quantities ( caon et al .
the model - determined quantities are those derived from the @xmath5 model fitted to the undisturbed portion the galaxy light profile .
this portion is free from any peculiarities of the inner light profiles ( noted by caon et al .
1993 ) , is not heavily distorted by the presence of a disk or dust lanes , and is uncontaminated from possible tidal interactions which may have perturbed the original galaxy profile , particularly in the outer parts .
the parameters used by donofrio et al .
( 1996 ) are better estimates of the half - light radii , in the true sense of the term , in that they are the radii which enclose half of the total light from the galaxy , although the total light was derived by extrapolating the observational data to @xmath30=32 with @xmath0 profiles ( caon et al .
1994 ) , the preferred dynamical term used here differs from that used by donofrio et al .
( 1996 ) . while they used cvd values
, this paper has also used , and in fact prefers , the infinite aperture velocity dispersion derived from the application of the jeans equation ( ciotti 1991 ; paper 1 ) .
donofrio et al .
( 1996 ) gave their virgo fp solution as @xmath64@xmath23@xmath65 .
this is in good agreement with the fp solution derived here , which addressed the range of structural profiles but used cvd measurements . for the sample of virgo e and s0 galaxies
the relation @xmath25@xmath23@xmath66 was obtained here .
donofrio et al .
( 1996 ) computed a virgo - fornax distance modulus of 0.50 mag ( using the fp ) and [email protected] mag ( using the @xmath67-@xmath39 relation ) .
this result is in close agreement with the distance modulus derived in this study based upon galaxy parameters from the @xmath0 profile and using cvd s , where a value of [email protected] mag ( asymptotic error ) was computed . using @xmath5 model parameters and infinite aperture velocity dispersion measurements
, however , this study found the virgo - fornax distance modulus to be [email protected] mag ( asymptotic error ) .
given that donofrio et al .
( 1996 ) allowed for different galaxy structures through using model - independent estimates of the half - light radius and associated mean surface brightness , but used cvd s , the final estimate of 0.25 mag appears to be different because of the treatment of the dynamical term rather than the treatment of the structural quantities . indeed , donofrio et al .
( 1997 ) introduced their own aperture correction to the velocity dispersion term , finding a reduced virgo - fornax distance modulus of [email protected] mag ( formal error ) .
comparison with other methods of distance determination yields good agreement with the distance modulus of + [email protected] mag obtained in section [ sectd ] .
the investigation by mcmillan et al .
( 1993 ) , using the planetary nebula luminosity function , found a virgo - fornax distance modulus of + [email protected] mag .
another accurate estimate has been derived using the method of surface brightness fluctuations , where a value of [email protected] mag has been obtained ( tonry et al .
1997 ) . using a sample of e and s0 galaxies , dressler et al .
( 1987 ) applied the @xmath67-@xmath39 relation to compute a value of + [email protected] mag .
an alternate approach using type ia supernovae ( hamuy et al .
1991 ) found a yet lower value of [email protected] mag , although still consistent within the errors of the value derived here .
however the result presented here is in poorer agreement with studies using the globular cluster luminosity function ( [email protected] mag , bridges , hanes & harris 1991 ) and with studies which have used the tully fisher ( tf ) relation for spiral galaxies .
the h - band tf relation used by aaronson et al . ( 1989 ) placed the fornax cluster closer than virgo , giving a distance modulus of [email protected] mag for fornax relative to virgo .
more recently , bureau et al . ( 1996 ) used an i - band tf relation . demanding equal slopes for the tf relation in each cluster , and allowing for errors on both axes
, they obtained a value of [email protected] mag .
this result is not as inconsistent with the analysis in this paper as one might first think .
it should be noted that the bootstrapping error analysis , performed in section [ sectd ] , revealed that the estimate of the error based upon analytical asymptotic expressions was too small due to the small sample size ( see table [ tab7 ] ) .
correction for this increased the error estimate from 0.12 mag to 0.27 mag .
the treatment in bureau et al.(1996 ) , and indeed most papers of this kind , has similarly under - estimated the true size of their errors for the same reason the analytical expression for the size of the errors under - estimates the true size of the errors when the sample size is small ( approximately @xmath1950 ) .
in addition , it is known that spiral galaxies are less centrally concentrated within a cluster than the elliptical galaxies , so they may not provide as accurate a measurement of the cluster distance as the ellipticals can
. this could be particularly relevant for virgo , a cluster known to have a considerable line - of - sight depth ( tonry , ajhar & luppino 1990 ; jacoby et al .
1992 ) .
the virgo - fornax distance modulus obtained in this paper is also only marginally consistent with the study by bothun , caldwell & schombert ( 1989 ) who used the surface brightness profiles of dwarf ellipticals to compute a distance modulus of [email protected] mag . by the nature of their study , they also sampled a different galaxy population than used here , which again may partly explain the disagreement in distance modulus .
it is also observed that they made no mention of having used any numerical resampling technique to estimate the size of their errors . to extrapolate , this opens up the possibility that many past inconsistencies between cluster distance estimations may be simply due to an inadequate , and under - estimation , of the associated errors .
there are other effects which have been proposed to influence the tilt of the fp .
one of these is orbital anisotropy in the galaxies internal velocity fields .
this has been explored in detail in the studies by ciotti , lanzoni & renzini ( 1996 ) and ciotti & lanzoni ( 1997 ) . using the jeans equation
, they performed theoretical modeling of the dynamics of a range of @xmath5 surface brightness profiles ( @xmath17=1 - 10 ) , and for the hernquist ( 1990 ) and jaffe ( 1983 ) density models .
they explored possible variations in the orbital anisotropy with radius . rejecting those models which are likely to be unstable
, they found that variations to the tilt of the fp due to orbital anisotropy are confined within the known thickness of the fp and can thus have no significant influence on the fp tilt .
the influence of differing stellar populations , as given by line strength indices such as mg@xmath68 or broad band colours such as b - v , was initially determined to have little or no effect on the slope of the fp ( djorgovski & davis 1987 ; lynden - bell et al .
1988 ; jrgensen et al .
however , more recent estimates of the influence from differing stellar metallicities and ages are suggesting a greater effect on the slope of the fp ( gregg 1992 ; guzman & lucey 1993 ) .
prugniel & simien ( 1996a , b ) combined four broad band colours and the mg@xmath68 index in the extraction and computation of a mean measure of the influence of differing stellar populations .
they found that stellar populations may account for half of the observed tilt to the fp .
if confirmed , this would easily explain any remaining tilt to the fp , and may place one in the unexpected position of having the fp tilt the other way .
for a sample of virgo and fornax early - type galaxies , the surface brightness profiles from caon et al .
( 1994 ) , which are known to represent a range of luminosity structures , have been used to explore the tilt and thickness of the fp .
this study also explores the range of luminosity structures present amongst the early - type galaxies using the sersic @xmath5 model rather than forcing structural homology through use of the @xmath0 model ( see also caon et al .
the @xmath5 light profile model has been used to parameterise the size ( as given by the effective radius @xmath25 ) and mean surface brightness ( @xmath50 ) for each galaxy . solving the jeans equation ,
the associated range of dynamical structures has been computed and the projected velocity dispersion within an aperture of infinite radius ( = 1/3 of the virial velocity dispersion ; ciotti 1994 ) has been derived .
this approach measures the velocity dispersion of each galaxy in a consistent fashion .
cvd measurements sample different fractions of different galaxy s dynamical profiles , where as the infinite aperture velocity dispersion consistently measures galaxies in the same way , and therefore provides a better estimate of a galaxy s kinetic energy .
the trend of increasing shape parameter @xmath17 with increasing values of the model - independent half - light radius ( caon et al .
1993 ) , is also shown to be present when using the @xmath5 model - determined effective radius @xmath25 .
a @xmath40 error analysis of the fitted model parameters shows that this is not due to coupling between the model parameters implying it is a real physical relation .
a principal component analysis reveals that greater than 97% of the variance in the fp data set ( @xmath39,@xmath27,@xmath13 ) resides within 2 dimensions , confirming the appropriateness of fitting a plane to this data .
it is concluded that broken structural and dynamical homology are responsible , at least in part , for the departure of the fp from the relation @xmath27@xmath23@xmath28 predicted by the virial theorem .
having used an independent set of data obtained by different observers with a different instrument and reduction procedures , this confirms the results of paper i. this conclusion is supported by the additional analysis of the fornax cluster . for comparison of the the fp with the virial theorem ,
the fp was constructed using the bisector method of linear regression in 3 dimensions . due to the small galaxy sample size
, a bootstrapping procedure was employed to estimate the uncertainty on the fp slope . using a sample of 25 virgo e ( 18 ) and s0 ( 7 ) galaxies ,
the fp constructed using @xmath0 model parameters and cvd measurements was found to be given by the relation @xmath4 . allowing for broken structural and broken dynamical homology , this plane was observed to change to @xmath7 .
the perpendicular rms scatter about these fp s is 0.099 to 0.077 dex respectively .
given that the s0 galaxies likely possess significant rotational energy that is not dealt with through use of velocity dispersion measurements , they were removed from the sample of virgo galaxies and the above analysis was re - performed .
the resulting fp was found to change from @xmath9 to @xmath10 . the perpendicular rms scatter to the plane
was found to decrease by 40% from 0.084 to 0.050 with the fuller treatment of galaxy structure and dynamics .
the fp for a sample of 7 elliptical galaxies in the fornax cluster , was also computed under the same conditions as above .
the fornax fp was found to change from @xmath69 to @xmath11 , with perpendicular rms scatter of 0.064 and 0.050 respectively .
the formal solution of the latter fp is in close agreement with the plane predicted by the virial theorem , and it will be of interest to see how / if this solution changes for a larger sample of galaxies . constructing
the virgo fp by minimising the residuals of the @xmath12 variable and applying this to the fornax data , through use of the improved working - hotelling confidence bands approach developed by feigelson & babu ( 1992 ) , enabled an estimation of the relative distance between the two clusters . forcing structural homology ( i.e. fitting @xmath0 profiles ) and using cvd s , a virgo - fornax distance modulus of [email protected] mag ( analytical error ) was obtained . allowing for the range of galaxy structures observed in the data , and using the infinite aperture velocity dispersion ,
the virgo - fornax distance modulus was computed to be [email protected] mag ( analytical error ) .
a fuller estimate of the associated uncertainties to these distance moduli was achieved through a numerical resampling of the 3-dimensional fp data set .
this analysis made the fornax cluster ( 23@xmath1544)% and ( 12@xmath1536)% more distant than the virgo cluster , for the respective models above .
i thank mauro donofrio for making his galaxy profile data available to me in electronic form .
i also thank eric feigelson and ron kollgaard for providing me with their computer codes to compute the intercept offset between parallel data sets , and for their quick help with my queries .
i am also grateful for the use of eric feigelson s computer code slopes .
i thank matthew colless for his proof reading of this manuscript and his useful suggestions .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
12345 aaronson m. et al . , 1989 , apj , 338 , 654 bender r. , saglia r.p . , gerhard o.e . , 1994 , mnras , 269 , 785 bevington p.r . ,
1969 , data reduction and error analysis for the physical sciences .
new york , mcgraw - hill binggeli b. , cameron l.m .
, 1991 , a&a , 252 , 27 bothun g.d . , caldwell n. , schombert j.m . , 1989 ,
aj , 98 , 1542 bower r.g . , lucey j.r . ,
ellis r.s . , 1992 , mnras , 254 , 589 bridges t.j . , hanes d.a . ,
harris , w.e .
, 1991 , aj , 101 , 469 bureau m. , mould j.r . , staveley - smith , l. , 1996 , apj , 463 , 60 burstein d. , heiles c. , 1978 , apj , 225 , 40 burstein d. , heiles c. , 1984 , apjs , 54 , 33 burstein d. , davies r.l . , dressler a. , faber s.m . ,
stone r.p.s . , lynden - bell d. , terlevich r.j . , wegner g. , 1987 , apjs , 64 , 601 busarello g. , longo g. , feoli a. , 1992 , a&a , 262 , 52 busarello g. , capaccioli m. , capozziello s. , longo g. , puddu e. , 1997 , a&a , 320 , 415 caon n. , capaccioli m. , donofrio m. , 1993 , mnras , 265 , 1013 caon n. , capaccioli m. , donofrio m. , 1994 , a&as , 106 , 199 capaccioli m. , 1989 , in corwin h g. , bottinelli l. , eds , the world of galaxies .
springer - verlag , berlin , p.208 capaccioli m. , caon n. , 1989 , in grosbl p.j . , murtagh f. , warmels r.h .
eso / st - ecf workshop on : data analysis .
eso , garching , p. 107
capaccioli m. , longo g. , 1994 , a&a review , 5 , 293 ciotti l. , 1991 , a&a , 249 , 99 ciotti l. , 1994 , cel . mech . & dynam . astr . , 60 , 401 ciotti l. , lanzoni b. , 1997 , a&a , 321 , 724 ciotti l. , lanzoni b. , renzini a. , 1996 , mnras , 282 , 1 davies r.l . , efstathiou g. , fall s.m . , illingworth g. , schecter p.l . , 1983 , apj , 266 , 41 de vaucouleurs g. , 1948 , ann dap . , 11 , 247 de vaucouleurs g. , 1953 , mnras , 113 , 134 de vaucouleurs g. , de vaucouleurs a. , corwin h.g.jr . , buta r.j . , paturel g. , fouque p. 1991 , in third reference catalogue of bright galaxies .
springer - verlag djorgovski s. , davis m. , 1987 , apj , 313 , 59 djorgovski s. , santiago b.x . , 1993 , in danziger j. et al./ , eds , proc . of the eso / eipc workshop on structure , dynamics , and chemical evolution of early - type galaxies .
eso publication 45 , p.59 donofrio m. , capaccioli m. , caon n. , 1994 , mnras , 271 , 523 donofrio m. , capaccioli m. , zaggia , s.r . , caon n. , 1997 , mnras , in press donofrio m. , longo g. , capaccioli m. , 1996 , in buzzoni a. , renzini a. , serrano g. , eds , asp conf .
86 , fresh views of elliptical galaxies .
, san francisco , p. 143
dressler a. , lynden - bell d. , burstein d. , davies r.l . ,
faber s.m . ,
terlevich r.j . , wegner g. , 1987 , apj , 313 , 42 feigelson e.d .
, babu g.j . , 1992 ,
apj , 397 , 55 graham a. , colless m.m . ,
1997a , mnras , 287 , 221 , paper i graham a. , colless m.m .
, 1997b , in arnaboldi m. , da costa , g.s .
, saha p. , eds , asp conf .
116 , the second stromlo symposium the nature of elliptical galaxies .
, san francisco , p. 175
graham a. , lauer t.r .
, colless m. , postman m. , 1996 , apj , 465 , 534 gregg m.d . , 1992 , apj , 384 , 43 guzman r. , lucey j.r . , 1993 , mnras , 263 l47 hamuy m. , phillips m.m .
, maza j. , wischnjewsky m. , uomoto a. , landolt a.u . , khatwani r. , 1991 , aj , 102 , 208 hernquist l. , 1990 , apj , 356 , 359 illingworth g. , 1983 , in athanassoula e. , ed .
100 , internal kinematics and dynamics of galaxies .
kluwer , dordrecht , p. 257 isobe t. , feigelson e.d .
, akritas m.g . ,
babu g.j . , 1990 ,
apj , 364 , 104 jacoby g.h .
et al . , 1992 , in ?
eds , asp conf .
ser . vol .
pac . , san francisco , p. 599
jaffe w. , 1983 , mnras , 202 , 995 jrgensen i. , franx m. , kjrgaard p. , 1993
, apj , 411 , 34 lynden - bell d. , burstein d. , davies r.l . ,
dressler a. , faber s.m .
, terlevich r. , wegner g. , 1988a , in pritchet c.j .
, vanderbergh s. , eds , asp conf .
4 , the extragalactic distance scale , asp , san francisco , p.307 mcelroy d.b . , 1995 ,
apjs , 100 , 105 mcmillan r. , ciardullo r. , jacoby g.h . , 1993 , apj , 416 , 62 michard r. , 1985 , a&as , 59 , 205 murtagh f. , heck a. , 1987 , multivariate data analysis .
reidel , dordrecht prugniel ph . , simien f. , 1994 , a&a , 281 , l1 prugniel ph . , simien f. , 1996a , a&a , 309 , 749 prugniel ph . , simien f. , 1996b , in buzzoni a. , renzini a. , serrano g. , eds , asp conf .
86 , fresh views of elliptical galaxies .
soc . pac .
, san francisco , p. 143
schombert j.m . , 1986 , apjs , 60 , 603 sersic j .- l . , 1968 , atlas de galaxias australes , observatorio astronomico , cordoba simien f. , prugniel ph . , 1992 , in longo g. , capaccioli m. , busarello g. , eds , morphological and physical classification of galaxies .
kluwer , dordrecht , p. 431
tonry j.l . ,
ajhar e.a .
, luppino g.a . , 1990 ,
aj , 100 , 1416 tonry j.l . , blakeslee j.p . , ajhar e.a .
, dressler a. , 1997 , apj , 475 , 399 working h. , hotelling h. , 1929 , j. am .
suppl . , 24 73 | the influence of broken structural homology , and the implied broken dynamical homology , is examined for the fundamental plane ( fp ) . requiring a symmetrical treatment of the fp variables , a bisector method of linear regression
was applied , in 3-dimensions , to derive the best fp .
a bootstrapping procedure has been used to estimate the uncertainties associated with the slope of the fp . for 25 e and s0 virgo galaxies ,
the ` standard ' fp , constructed using @xmath0 model parameters for the effective radii ( @xmath1 ) and the mean surface brightness within this radius ( @xmath2 ) and using central velocity dispersion ( cvd ) measurements ( @xmath3 ) , gave a relation described by @xmath4 . using sersic @xmath5 light profile model parameters and
the projected , infinite aperture , velocity dispersion ( @xmath6 ) , derived from application of the jeans equation to the observed intensity profiles , gave an ` improved ' fp described by the relation @xmath7 .
this result , based on independent data , supports the previous finding by graham & colless ( 1997a ) that assumptions of structural and dynamical homology are partly responsible for the departure of the observed fp from the plane expected by the virial theorem , which predicts @xmath8 . upon removal of the known s0 galaxies from the sample of virgo galaxies , the above planes
were observed to change to @xmath9 and @xmath10 .
the perpendicular rms residuals about these planes are 0.084 and 0.050 dex , respectively .
the fornax cluster was similarly treated , although removal of the s0 galaxies left a sample of only 7 ellipticals which had published cvd measurements . treating the range of structural and dynamical profiles present in this sample
produced a fp given by the relation @xmath11 , in tantalizing agreement with the plane expected from the virial theorem , but with discouragingly large errors due to the small sample size . similarly to virgo , the perpendicular rms residuals about this plane is 0.050 dex .
the fp was also constructed with the purpose of using it as a distance indicator , achieved by minimising the distance - dependent quantity @xmath12 against the distance - independent quantities @xmath13 and @xmath14 .
a virgo - fornax distance modulus was computed using working - hotelling confidence bands ( feigelson & babu 1992 ) .
the ` standard ' fp parameters gave a value of [email protected] mag , where - as the ` improved ' fp parameters gave a value of [email protected] mag .
however , a full treatment of the uncertainties on the fp slopes , derived through a bootstrapping procedure of the 3-dimensional fp data set , revealed that the analytical expressions for the uncertainties on the estimated distance moduli , given above , should be increased by a factor of @xmath165 .
galaxies : fundamental parameters galaxies : structure galaxies : kinematics and dynamics .
galaxies : elliptical and lenticular , cd distance scale |
since its launch in 1999 , the chandra x - ray observatory has been used to detect a large number of x - ray jets in active galactic nuclei ( agn ) , where prominent radio jets were previously known to exist ( see e.g. , * ? ? ? * and associated website ) .
the recent report @xcite of an extended x - ray jet originating from the z=4.3 quasar gb 1508 + 5714 , where previous observations showed no obvious sign of extended radio emission , presents an interesting case .
the x - ray feature is strong well over 100 counts were detected from it in the @xmath090 ksec chandra exposure .
an archival hst image helps rule out the possibility that it is due to a foreground galaxy or a gravitationally lensed image of the quasar @xcite . based on deep x - ray source counts
, it has a low probability of being a random unassociated x - ray field source .
as discussed by the previous authors , such detections of x - ray jets at large redshifts are actually to be expected as a natural consequence of the inverse compton ( ic ) off the cmb model ( e.g. , * ? ? ?
* ; * ? ? ?
this is because the ( 1+z)@xmath1 dependence of the cmb energy density compensates for cosmological dimming of radiation , so that ic / cmb x - ray jets should remain detectable out to large cosmological distances @xcite .
the model has been successfully applied to account for x - ray jets in many other powerful quasars at more modest redshifts ( e.g. , * ? ? ?
* ) , requiring that the jets are still highly relativistic on kilo - parsec scales , in order that the electrons in the jet frame see an adequately boosted photon source .
however , the lack of a detection of the gb 1508 + 5714 jet at lower frequencies , along with only a rough constraint on the x - ray spectrum , could not rule out a synchrotron origin for the x - rays @xcite .
a previous search in the radio for a proposed x - ray jet in another high redshift ( z=5.99 ) quasar , sdss 1306 + 0356 @xcite , yielded only an upper limit of @xmath20.1 mjy at 1.4 ghz @xcite .
distinguishing between the two possible emission processes is important , as they probe different energetic phenomena . in the case of synchrotron x - ray emission , the x - rays mark sites of particle acceleration with electrons accelerated up to @xmath3 with very short lifetimes ( e.g. m87 ; * ? ? ?
when considered together with minimum energy / equipartition conditions , the ic / cmb model offers important constraints on the beaming , magnetic field , and jet power @xcite .
both the ic and synchrotron interpretations of x - ray jet emission usually require a population of relativistic electrons emitting synchrotron radiation at lower ( radio ) frequencies .
based on the x - ray flux and spectrum , the two models give different predictions of the radio component flux and spectrum . here , we report the detection of such a radio component coincident with the x - ray feature extending from gb 1508 + 5714 from an analysis of archival vla data .
this supports its interpretation as an x - ray jet and we discuss the x - ray emission in the context of the new 1.4 ghz detection , along with the previously set optical , and a new 8.4 ghz limit . following @xcite and @xcite , h@xmath4 km s@xmath5 mpc@xmath5 , @xmath6 and @xmath7 @xcite are assumed throughout , so 1 = 6.871 kpc .
two datasets which utilized the vla in its highest resolution a - configuration were obtained from the nrao archive .
the data were calibrated in the nrao aips package @xcite and brought into difmap @xcite for editing , self - calibration , and imaging .
the flux density scale was set on the vla 1999.2 scale using scans of 3c 286 as outlined in the vla calibrator manual @xcite .
the 1.4 ghz dataset a single five minute snapshot observation yielded a 10@xmath8 detection of a 1.2 mjy feature extended from the quasar ( figure [ fig-1 ] ) .
this feature is only about @xmath9 of the core flux ( @xmath10 mjy ) , so it is no surprise that the original investigators @xcite simply concluded that the quasar was an unresolved point source from these data .
the measured off - source rms of @xmath110.12 mjy in the image is within 50@xmath12 of the thermal noise limit expected from the integration time .
the resultant image has a dynamic range of almost 2,000:1 .
the jet knot was modeled with an elliptical gaussian profile in both the ( u , v ) plane with difmap s ` modelfit ` utility , and in the map plane with the jmfit task in aips .
the fits gave consistent measures of the deconvolved size : @xmath13 , elongated along the jet direction .
we note that this is comparable to the beamsize for this observation so the source is consistent with being unresolved , and the size should be considered an upper limit .
the separation from the core is 2.5 at a position angle of @xmath14 , and is comparable to that measured from the x - ray data .
interestingly however , the peak in the x - ray jet appears somewhat closer to the nucleus than the radio centroid in the image overlay ( figure [ fig-2 ] ) , by about one chandra acis pixel ( 0.492 per pixel ) , or equivalently @xmath03 kpc projected distance .
the images were aligned by the cores to better than half of a pixel .
@xcite found the peak in the x - ray jet to be @xmath02 away ( see also their subpixel re - binned radial profile published in their figure 2 ) , consistent with our qualitative assessment .
@xcite stated the same peak to be @xmath03 distant from the nucleus , although we measured a smaller value off their figure 1 . much more apparent x - ray /
radio offsets are seen in other powerful quasars ( e.g. up to @xmath02 in pks 1127145 ; * ? ? ?
most , if not all of the x - ray counts from the gb 1508 + 5714 jet lie within the outermost radio contour in the vla image ( figure [ fig-2 ] ) . a 10-minute 8.4 ghz observation obtained on nov 3 , 1996 ( program ad388 ) yielded no detection of the radio jet .
the measured off - source rms in the naturally weighted image ( @xmath00.35 beam ) is at about the expected thermal noise limit of 0.045 mjy . in order to judge
if the non - detection was a result of the greater resolution in this image compared to the 1.4 ghz map , the 8.4 ghz dataset were tapered by different amounts .
no outstanding feature appeared above the residual artifacts from the dirty beam .
the measured rms at the expected position of the jet in a tapered image restored with a 0.75 beam is 0.3 mjy ( 3@xmath8 ) .
this , along with the 1.4 ghz detection , is consistent with an @xmath150.8 radio spectrum ( @xmath16 ) , which agrees with the measured x - ray spectral index : [email protected] @xcite and 0.92@xmath18 @xcite .
the spectral energy distribution of the jet knot in gb 1508 + 5714 is shown in figure [ fig-3 ] .
if the x - ray and radio emission are drawn from the same population of relativistic electrons emitting synchrotron radiation , the spectral index will be about what is measured between the 1.4 ghz and x - ray ( using 1.68@xmath1910@xmath20 ph/@xmath21/s / kev at 1 kev , reported by @xcite ) detections : @xmath22 = 0.73 .
this is a typical value for the spectrum seen in radio jets @xcite , and is consistent with the measured x - ray spectrum of @xmath23 @xcite , and our constraint on the radio spectrum of @xmath240.8 ( the modest 8.4 ghz dataset does not preclude that the higher frequency radio emission , @xmath2540 ghz in the source frame , was actually resolved out ) . the optical hst limit ( 3@xmath8 ) from @xcite
is not useful in this case , as it hovers over the radio - to - x - ray spectrum , and only weaker constraints are available at other optical bands @xcite .
we can estimate an equipartition magnetic field of @xmath26 g , by adopting the observed @xmath22 value as the optically thin spectral index , and that the particles and field fill a fraction @xmath27 of a 2.9@xmath1910@xmath28 @xmath29 sphere , corresponding to the size upper limit derived from the radio detection .
the factor @xmath30 is the ratio of magnetic field and electron energy densities , and @xmath31 is the lower energy cutoff of a power law distribution of relativistic particles .
if we further assume @xmath30=1 , @xmath27=1 , and @xmath31=10 , then @xmath32 g , where @xmath33 is the unknown jet doppler beaming factor . in this field , electrons emitting the highest energy x - rays detected ( @xmath05 kev ; * ? ? ?
* ) will have @xmath34 and short lifetimes ( @xmath050 years ) this would require continuous in - situ reacceleration of high energy particles in order that the x - ray jet not be a transient feature . at this high redshift , the energy density of the cmb already exceeds that of the equipartition field , even without bulk motion ( see below ) , so inverse compton losses will already be dominant .
the lifetime of the @xmath34 electrons calculated above is then a strict upper limit . taking this evidence together ,
it is unlikely that the jet x - ray emission is dominated by synchrotron losses .
an ic / cmb origin for this high redshift x - ray jet is preferred @xcite , and give us an additional constraint on the allowed range of magnetic field and jet doppler factor to the equipartition condition .
the following expression is taken from @xcite : where , @xmath36 is a dimensionless function of @xmath37 ( e.g. , * ? ? ?
* ) , @xmath38 is the thomson cross section , and @xmath39 ergs @xmath40 for the cmb ( e.g. , * ? ? ?
* ) , whose spectrum is peaked near @xmath41 ( 1+z ) hz @xcite .
assuming @xmath42 , the observed ratio of the compton ( x - ray ) and synchrotron ( radio ) fluxes ( @xmath43 and @xmath44 , observed at @xmath45 and @xmath46 , respectively ) , allow us to estimate @xmath47 g. the corresponding equipartition calculation gives @xmath48 g , making the assumptions as above about @xmath30 , @xmath27 , and @xmath31 .
both estimates of @xmath49-field are reasonably insensitive to the assumed spectral index : up to @xmath025@xmath12 smaller for @xmath50 , the measured x - ray spectral index . in order to reconcile the differences between the two equations ( opposite dependencies of @xmath49 on @xmath33 ) , a doppler factor greater than one
is required .
these two constraints bracket a plausible range of solutions around a 30@xmath51 g field and @xmath524 .
@xcite arrived at similar conditions using slightly different assumptions .
the unresolved nucleus is a variable x - ray and ( flat spectrum ) radio source with a very high x - ray to optical luminosity ( see , * ? ? ?
* and references therein ) .
these observations have been taken as evidence of beaming on smaller scales , in the nucleus .
the ratio of the monochromatic x - ray to radio luminosities ( i.e. , in @xmath53f@xmath54 ) of the gb 1508 + 5714 jet is 158 .
this is one of the highest amongst the known x - ray jets which can be attributed to ic / cmb emission so far . in figure
[ fig-4 ] , this ratio is plotted vs. redshift along with ic / cmb x - ray jets taken from the literature . the values seem to vary widely over 4 orders of magnitude , but predominantly show x - ray to radio ratios greater than 1 .
the jet in gb 1508 + 5714 sees 12 orders of magnitude times greater energy density from the cmb than other jets at lower redshift , so it is tempting to speculate that its extreme redshift may account , to first order , for its large x - ray to radio luminosity ratio .
this could similarly , account for the large x - ray to radio ratio limit obtained for the proposed x - ray jet in the z=5.99 quasar sdss 1306 + 0356 ( * ? ? ?
* ; * ? ? ?
[ fig-4 ] ) .
the observed monochromatic x - ray to radio luminosity ratios can be compared to what is expected in the ic / cmb model .
we can obtain an expression for the expected ratio by rearranging equation [ eqn-1 ] ( and setting @xmath42 ) : the flux ratio is simply proportional to the ratio of the energy densities in the cmb and magnetic field , with a doppler factor term .
this expression is plotted in figure [ fig-4 ] for several different combinations of @xmath49 and @xmath33 , one of which , best suits the observations of gb 1508 + 5714 .
it appears that varying @xmath49 and @xmath33 can account for the large spread of observed x - ray to radio luminosity ratios within a given redshift range .
larger @xmath49-field , or more likely , smaller @xmath33 may account for the low values observed in q0957 + 561 ( z=1.41 , * ? ? ?
* ) and 3c 9 ( z=2.012 , * ? ? ?
* ) , as they are known to have low radio core dominance , and weak vlbi structures @xcite compared to the other sources studied here . in all but 3c 179 ( z=0.846 , * ? ? ?
* ) where multiple knot regions along the jet can be distinguished , the x - ray to radio flux ratio decreases with increasing distance from the nucleus , and can similarly be accounted for by varying @xmath56 . as discussed specifically in the case of 3c 273 ( z=0.158 , * ? ? ?
* ) , the variations along the jet may indicate deceleration , or increasing @xmath49 along the jet , possibly by compression in strong shocks .
future detections in the redshift range @xmath024 or higher with large x - ray / radio flux ratios can lend further support to the currently preferred ic / cmb model used to explain the x - ray emission in quasar jets .
an anonymous referee is thanked for making the point that compton losses are already dominant in the synchrotron x - ray case , and several other useful comments .
the author is grateful to john wardle for his advice and encouragement throughout the course of this work , and to dan harris , aneta siemiginowska , dave roberts , and fabrizio tavecchio for useful discussions .
radio astronomy at brandeis university is supported by the national science foundation through grants ast 98 - 02708 and ast 00 - 98608 .
additional support by nasa grant go2 - 3195c from the smithsonian astrophysical observatory , and hst - go-09122.08-a from the space telescope science institute , which is operated by aura , inc .
, under nasa contract nas 5 - 26555 is acknowledged . | the recent discovery of an x - ray jet in the z=4.3 quasar gb 1508 + 5714 by yuan et al .
( astro - ph/0309318 ) and siemiginowska et al .
( astro - ph/0310241 ) prompted a search for its radio counterpart .
here , we report the successful discovery of faint radio emission from the jet at 1.4 ghz using archival vla data .
the x - ray emission is best interpreted as inverse compton ( ic ) emission off the cmb as discussed by the previous investigators . in this scenario ,
its high x - ray to radio monochromatic luminosity ratio , compared to previously detected ic / cmb x - ray jets at lower redshift , is a natural consequence of its high redshift . |
the goal of the lisa mission @xcite is to detect and analyze low frequency gravitational signals essentially in the range from 0.1 mhz to 0.1 hz . in this frequency range ,
sources of gravitational waves are mainly the wide population of galactic compact binaries , and interactions between black holes with a range of complexity .
the study of the emission of gw from known binaries could be extremely useful for firstly , direct determination of distances , and secondly , possible small general relativistic effects , if the signal to noise ratio ( snr ) is large enough . for this reason ,
we focus on optimising the sensitivity of lisa for a given astrophysical source with known direction .
the sensitivity of lisa can be improved by solving technological problems , but it can also be improved by employing certain optimal data analysis strategies . in this paper , we show how algebraic methods previously developed in @xcite , henceforth referred to as paper i , can be used to design optimal strategies for combining data with appropriate time - delays .
we show in this paper that it is possible to maintain the optimality during the year by continuously updating the parameters of the combination or simply by switching to optimal data combinations as the source appears to move in the lisa frame , as lisa moves in its complex orbit around the sun .
the problem of optimisation of snr , in this context , has been addressed before in @xcite and @xcite . in @xcite optimisation
has been carried out before averaging over the directions and polarisations , while in @xcite , henceforth referred to as paper ii , the averaging is done first and then the optimisation . in this paper , the averaging over the polarisations is performed first and then the snr is optimised for the average signal for a given direction over the relevant data combinations - those data combinations canceling laser frequency noise .
thus in this optimisation , the direction of source is assumed to be known , but not its polarisation .
this would be the case for several binaries in our galaxy .
we analyse two observables in this context : ( i ) optimal data combination in the module which yields the maximum snr for a given direction and ( ii ) a ` network ' observable which is obtained by squaring and adding the snrs of two independent(orthogonal ) data combinations one of them being the optimal combination mentioned in ( i ) and another orthogonal to it .
the network observable in general yields higher snr .
we analyse how these snrs depend on direction of the source . for an integration time of one year
, we compare our results by plotting sensitivity curves with those obtained from , ( a ) the michelson combination @xmath1 @xcite , which has been used for obtaining the standard sensitivity curve for lisa , and ( b ) the observable @xmath2 obtained by switching the three michelson combinations optimally .
we find that for sources lying in the ecliptic plane , the network sensitivity at low frequencies is about 34% more than the optimally switched michelson combinations which rises to nearly 90% at 20 mhz .
finally we compute the optimal snrs for six known binary systems in our galaxy , whose snrs are significantly high , for an integration time of one year .
we show that if at low frequencies the two snrs of the orthogonal data combinations can be measured , then it is possible to estimate the inclination angle of the binary s orbit .
the paper is organised as follows : in section [ module ] we briefly describe the formalism involving commutative algebra developed in paper i. we also present a convenient set of generators for the data combinations found in paper ii , which consisted of the eigenvectors of the noise - covariance matrix , and in which the computations are simplified . in section [ signal ]
we discuss the gravitational wave ( gw ) signal from non - spinning compact binaries and average the signal over the polarisations . in section [ snr ]
we obtain the optimum snr by the method of lagrange multipliers , where we must solve an eigenvalue problem . in section [ track ]
we describe the transformations from the lisa frame to the barycentric frame and vice - versa . in section [ sens ]
we compute the optimal snrs and the sensitivity curves and compare them with the standard curve for lisa obtained with the michelson data combination @xmath1 and the optimally switched michelson combination @xmath2 . in this section
we also estimate the snrs for a few known binaries in our galaxy and describe a method to estimate the angle of inclination of the binary s orbit , if the two snrs of the orthogonal combinations can be measured .
in this section , we briefly review the results from papers i and ii which are needed for the analysis that follows . in lisa
the six elementary data streams are labeled as @xmath3 and @xmath4 ; if the space - crafts are labeled clockwise , the @xmath3 beams also travel clockwise , while the @xmath5 beams travel counter - clockwise . the beam @xmath6 travels from space - craft 3 to space - craft 1 along the arm of length @xmath7 in the direction @xmath8 , while @xmath9 represents the beam traveling from space - craft 2 to space - craft 1 along the arm of length @xmath10 in the direction of @xmath11 .
the remaining 4 beams are described by cyclically permuting the indices .
these beams contain the laser frequency noise , other noises such as optical path , acceleration etc . and also the gw signal . for performing the analysis
, we choose a frame @xmath12 tied to the lisa constellation .
we choose the centroid of the lisa triangle as the origin , the lisa triangle to lie in the @xmath13 plane and space - craft 2 to lie on the @xmath14-axis .
we will henceforth drop the suffix ` @xmath15 ' from the lisa frame quantities and only introduce it when other frames are being considered in the discussion . a schematic diagram of lisa
is shown in fig.[lisagm ] .
[ lisagm ] in paper i , we showed that all data combinations canceling laser frequency noise form a module over the ring of polynomials in three indeterminates , these being the three - time delay operators @xmath16 of the light travel time along the three arms with lengths @xmath17 respectively .
thus for any data stream @xmath18 . for the lisa specifications @xmath19 sec .
corresponding to an armlength of 5 million km .
a general data combination is a linear combination of the elementary data streams @xmath20 : @xmath21 where @xmath22 and @xmath23 are polynomials in the time - delay operators @xmath16 .
thus any data combination can be expressed as a six - tuple polynomial ` vector ' @xmath24 .
for cancellation of laser frequency noise only the polynomial vectors satisfying this constraint are allowed and they form the module of syzygies mentioned above .
an important advantage of this formalism is that , given the generators of the module , any other data combination canceling the laser frequency noise is simply a linear combination of the generators with the coefficients being polynomials in the ring .
several sets of generators have been listed in paper i. based on the physical application as also on convenience , we may choose one set over another .
for the monochromatic sources , as argued in paper ii , only three generators are sufficient to generate the relevant data combinations .
in general 4 generators are necessary to generate the module but if the source is monochromatic , except at certain frequencies which are solutions of @xmath25 , the fourth generator can be effectively eliminated for the purposes of maximising the snr . since
this maximisation is possible arbitrarily close to the singular frequencies the singularities do not seem to be important .
we found in paper ii that a convenient set of generators for our purpose is the one that diagonalises the noise covariance matrix .
these generators are the eigenvectors of the noise covariance matrix .
the noise covariance matrix @xmath26 is defined as the outer product @xmath27 where @xmath28 is a 12 dimensional complex noise vector of the generator @xmath29 ( the @xmath30 could be any set of generators not necessarily the eigenvectors of the noise covariance matrix @xmath31 ) .
the noise vector is given by : @xmath32 where the @xmath33 polynomials are defined through the equations @xmath34 plus cyclic permutations for @xmath35 and @xmath36 .
the @xmath37 and @xmath38 are the one - sided power spectral densities ( psd ) of the proof - mass noise and the optical - path noise respectively @xcite . for the purposes of this paper when computing the snr for any given data combination , the differences in armlengths are extremely small compared to the wavelength of the gw we are interested in detecting .
thus we may take the armlengths to be equal , i.e. , @xmath39 .
for the purposes of this analysis it is sufficient to consider lisa as an equilateral triangle ; we ignore the deviations arising from this assumption .
this simplifies the expressions for the noise and further also for the signal .
the eigenvectors now contain overall common factors which are polynomials in @xmath40 .
these common factors make no difference to the computation of snr as they cancel out from the numerator and denominator which comprise of the signal and the noise respectively .
the unnormalised eigenvectors ( with common factors canceled ) of the noise covariance matrix is the set @xmath41 which we list below : @xmath42 in paper ii , the @xmath43 have been listed with the uncanceled common factors .
we adopt the following terminology : we refer to a single element of the module as a data _ combination _ ; while a function of the elements of the module , such as taking the maximum over several data combinations in the module or squaring and adding data combinations belonging to the module , is called as an _
observable_. the important point to note is that the laser frequency noise is also suppressed for the observable although it may not be an element of the module .
since binaries will be important sources for lisa the analysis of such sources is relevant .
one such class is of massive or super massive binaries whose individual masses could range from @xmath44 to @xmath45 and which could be around a gpc away .
another class of interest are known binaries within our own galaxy whose individual masses are of the order of a solar mass but are just at a distance of a few kpc or less . here
we will assume the direction of the source to be known , which is justified for known binaries in our galaxy ; but even for the former case of distant binaries , it amounts to ` looking ' in a specific direction .
the spacetime metric perturbation for a gravitational wave propagating in a @xmath46 direction can be written as , @xmath47 where the direction to the source is described by direction angles @xmath48 , the unit vector pointing to the source is @xmath49 and @xmath50 and @xmath51 are unit vectors in the direction of the coordinates @xmath52 and @xmath53 , such that @xmath54 form a right - handed triad . for a binary that is not chirping , and hence monochromatic with frequency @xmath55 ,
the two polarisation amplitudes @xmath56 and @xmath57 @xcite are given by , @xmath58\ , , \nonumber \\
h_{\times}(t ) & = & h\left[-\frac{1+\cos^{2}\epsilon}{2}\sin2\psi\,\cos\omega t+\cos\epsilon\,\cos2\psi\,\sin\omega t\right]\,,\end{aligned}\ ] ] where the angles @xmath59 describe the orientation of the binary orbit ( @xmath60 could be the direction angles of the orbital angular momentum vector ) and @xmath61 the overall amplitude which depends on the masses , the distance and the frequency @xmath62 as given below : @xmath63^{\frac{5}{3 } } \cdot \left[\frac{r}{1\,\mathrm{gpc}}\right]^{-1}\cdot\left[\frac{f}{1\,\mathrm{mhz}}\right]^{2/3}\,,\ ] ] where @xmath64 is the distance to the source , @xmath65 is the gw frequency of the source and @xmath66 the chirp - mass , where @xmath67 and @xmath68 are respectively the reduced and the total mass of the binary . for the typical parameters taken above the frequency
evolves very slowly , so much so that the time for the system to coalesce is more than 20 years . for an observational time of the order of an year , the frequency of the binary changes very little so that the source can be practically taken to be monochromatic .
since we deal with essentially monochromatic sources , the fourier domain is appropriate for further analysis . in the fourier domain
we have , @xmath69\ , , \nonumber \\
h_{\times}(\omega ) & = & h \left[-\frac{1+\cos^{2}\epsilon}{2}\sin2\psi - i\,\cos\epsilon\,\cos2\psi\right]\ , \label{sig}.\end{aligned}\ ] ] the gw response for a generator @xmath70 is ( see papers i and ii ) , @xmath71 where , @xmath72 where we have expressed the response of the two polarisations in terms of the responses of the elementary data streams for each of the polarisations .
below we state the responses for the @xmath73 and @xmath74 polarisations for the beams @xmath6 and @xmath75 only ; the responses for the remaining four beams @xmath76 are obtained from cyclic permutations : @xmath77 where , @xmath78 we define the signal covariance matrix @xmath79 in analogous fashion as the noise covariance matrix
. it is given as follows : @xmath80 in general we may not have any knowledge of the polarisation of the gw binary source .
we therefore average over the polarizations and assume that the direction of the orbital angular momentum of the binary is uniformly distributed over the sphere .
the orientation of the binary ( its orbital angular momentum vector ) has been described in terms of the angles @xmath81 and @xmath82 in eq.([sig ] ) .
thus we carry out the averaging of @xmath79 over @xmath83 which results in an overall factor of @xmath84 .
the averaged matrix we denote by @xmath85 . in the fourier domain it is given by , @xmath86 the signal matrix so averaged has the following properties : * @xmath87 is the sum of outer products of two vectors : @xmath88 with its complex conjugate and @xmath89 with its complex conjugate .
thus the natural basis for expressing @xmath87 consists of the two vectors @xmath88 and @xmath89 . in the analysis that follows we will use this fact . *
because we average over the polarisations , @xmath87 is constructed out of two vectors .
its rank is two , everywhere except on the @xmath90 plane where it is one when @xmath89 goes identically to zero . in paper
ii we had obtained a signal matrix of rank three because there we had averaged over the directions as well . while in @xcite , since optimisation is performed first before averaging , the signal matrix is constituted from a single vector and thus has rank one .
for a generic data combination @xmath91 where @xmath92 are polynomials in @xmath40 , the snr is given by : @xmath93 if the psd is given in units of hz@xmath94 then the above equation yields the square of the snr integrated over one second . because one second is short compared to the various time - scales envisaged in the problem we call this snr the _ instantaneous _ snr .
since we are dealing with monochromatic sources , @xmath95 , the coefficients @xmath92 reduce to just complex numbers .
also in the generating set @xmath96 the noise covariance matrix is diagonal with diagonal elements @xmath97 .
however , we find @xmath98 in our case , so that the first two eigenvectors correspond to the same eigenvalue .
thus , the denominator of eq.([gensnr ] ) simplifies to a sum of squares @xmath99 which we can set equal to unity because the snr does not depend on the normalisation of the data combination .
moreover , it is convenient to define coefficients @xmath100 which are scaled by the noise , @xmath101 so that the @xmath102 satisfy , @xmath103 the expression for snr simplifies to , @xmath104 where we define a snr matrix @xmath105 by , @xmath106 in the above equations the @xmath107 are construed of as @xmath108 column matrices and @xmath87 and @xmath105 as @xmath109 square matrices .
the extremisation of snr is now carried out with the method of lagrange multipliers because of the normalisation constraint on @xmath100 .
this procedure yields an eigenvalue equation with the lagrange multiplier appearing as an eigenvalue : @xmath110 since @xmath87 has atmost rank 2 , one eigenvalue is necessarily zero .
we now proceed to compute the other two eigenvalues .
the analysis is simplified if we go to the basis consisting of the two vectors @xmath111 and @xmath112 : @xmath113 where @xmath114 is the amplitude of the gw averaged over the polarisation states at frequency @xmath55 .
we can then write the matrix @xmath105 as a tensor product in terms of these two vectors : @xmath115 in general the vectors @xmath116 and @xmath117 are not orthogonal .
the action of the matrix @xmath105 on any vector @xmath118 is given by , @xmath119 for the eigenvalue problem , we have the eigenvalue equation , @xmath120 expressing the eigenvector @xmath121 as a linear combination of @xmath122 and @xmath123 , @xmath124 , and from the linear independence of @xmath116 and @xmath117 ( they are linearly independent in general ) we obtain the system of equations for @xmath125 and @xmath126 as : @xmath127 setting the determinant of this system to zero , we obtain eigenvalue equation : @xmath128 which can be solved to yield the eigenvalues : @xmath129 where , @xmath130 in the low frequency limit as we shall find @xmath131 so that the positive square root of @xmath132 is just @xmath133 and the eigenvalues are just @xmath134 and @xmath135 ( infact the eigenvalue equation factorises into linear factors ) with @xmath116 and @xmath117 as eigenvectors respectively .
this can also be directly inferred from the structure of @xmath105 .
so it may be appropriate to call @xmath136 as @xmath137 even in the general case . in the general case
the eigenvectors can be easily determined from eq.([eqn : eigncnd ] ) : @xmath138 where @xmath139 are eigenvectors belonging to the eigenvalues @xmath140 respectively . since @xmath105 is hermitian
the eigenvectors are orthogonal , @xmath141 as can be verified also directly from the expressions of the eigenvectors .
these eigenvectors are not normalised .
the eigenvalues @xmath140 are the squares of the instantaneous snrs for the two data combinations described by the two corresponding eigenvectors .
the data combinations are @xmath142 and @xmath143 , which we will call eigen - combinations or alternatively eigen - observables , and which give the instantaneous snrs : snr@xmath144 respectively . for a given direction @xmath145 in the lisa frame
, snr@xmath146 is the maximum instantaneous snr among all data combinations . while snr@xmath147 is the minimum instantaneous snr among data combinations that are linear combinations of @xmath148 and @xmath149 ; those which lie in the ` plane ' of @xmath148 and @xmath149 .
however , snr@xmath147 is not zero in general and therefore not the absolute minimum ; the absolute minimum snr is zero corresponding to the third eigenvalue which is zero .
moreover , the eigenvectors are orthogonal , which means they yield statistically independent observables .
thus these observables can be combined in quadratures to form a network observable with instantaneous snr@xmath150 given by : @xmath151 the third eigenvector is @xmath152 ; it is orthogonal to @xmath116 and @xmath117 with eigenvalue zero .
this means that the data combination corresponding to this vector gives zero response in that particular direction , which may be important if one wishes to ` switch off ' the gw coming from that direction .
in general the amplitude of the gw source will be small and it would be necessary to track or follow the source for a considerable period of time in order to accumulate an adequate snr .
this period of time could range from few days to a year or even years - the life of the lisa experiment .
the lisa configuration performs a complex motion due to which the source will appear to move in the lisa frame even if it is stationary in the barycentric frame . for the purposes of this analysis , we will take the observation time to be a year and integrate the snr for this period of time .
it is possible that some gw sources may be sufficiently powerful that snr integration for an year is not required . in that case
it is easily possible to limit the integration to the required period of observation and obtain useful results .
we will also assume that the source is stationary in the barycentric frame , that is , its direction remains constant during the observation period . in this section
we will present the transformations connecting the barycentric frame and the lisa frame and hence obtain the apparent motion in the lisa frame for a gw source fixed in the barycentric frame .
the lisa constellation trails the earth in its orbit by @xmath153 around the sun .
the plane of lisa makes an angle of @xmath154 with the plane of the ecliptic and the lisa triangle rotates in its own plane completing one rotation in a year .
we describe the barycentric frame by the cartesian coordinates @xmath155 .
the @xmath156 plane coincides with the orbital plane of lisa .
the @xmath157-axis is orthogonal to this plane forming a right - handed coordinate system .
the transformation from the barycentric frame to lisa frame is given as follows @xcite : a vector @xmath158 is transformed to @xmath159 by the matrix @xmath160 as , @xmath161 where @xmath160 is a product of the three matrices containing euler angles : @xmath162 where the matrices @xmath163 , @xmath164 and @xmath165 are given by , @xmath166 here @xmath167 , @xmath168 and @xmath169 .
@xmath170 is the orbital period of lisa which we take to be of one year duration .
the @xmath171 and @xmath172 are constants fixing initial conditions when the observation begins .
the matrix @xmath160 is time - dependent and is a product of two time - dependent matrices .
the unit vector @xmath173 of the source direction described by the angles @xmath174 in the barycentric frame is given by @xmath175 .
the corresponding vector @xmath176 in the lisa frame is then , @xmath177 . from this vector
we can explicitly work out the angles @xmath178 : @xmath179 where , @xmath180 for a fixed source direction @xmath174 in the barycentric frame , the apparent source direction depends on time in the lisa frame ; @xmath181 are functions of time ; the source appears to move in the lisa frame .
thus a given data combination even if it is optimal initially , will not continue to remain optimal subsequently .
our strategy is then to _ switch _ the data combinations continuously so that the snr remains optimal at all times . in this way we can optimally track the source and accumulate maximum snr .
the following figures show the apparent trajectory of the source for one year in the lisa frame when ( i ) the source lies at the pole of the barycentric frame - it is just a circle in the lisa frame : @xmath182 ( see fig.[trk1 ] ) and ( ii ) the source lies in plane of the lisa orbit - a figure of 8 is described by the source ( see fig.[apptrk ] ) . .
the track of the source for a period of one year is shown on the unit sphere in the lisa frame .
, title="fig : " ] [ trk1 ] .
the track of the source for a period of one year is shown on the unit sphere in the lisa frame .
, title="fig : " ] [ apptrk ]
for optimal tracking of the source , we switch the data combinations so that the snrs yielded are optimal at all times .
three snrs are of interest ; @xmath183 and snr@xmath150 , that is , the two eigenvalues and their sum .
the eigenvalues are functions of @xmath181 which are in turn functions of @xmath184 as the gw source describes its apparent motion in the lisa frame .
the three integrated snrs are given by : @xmath185 where we have taken the total observation time to be @xmath186 and the initial time of observation to be zero .
if we integrate for a complete year , we can set the initial observation time to be zero without loss of generality and at the same time set the constants @xmath187 to be zero .
however , if we make observations for times that are not integral number of years , the constants @xmath187 must be chosen appropriately in the transformation matrix @xmath160 .
an important point is that , since lisa moves in an heliocentric orbit , a doppler phase depending on the position vector of the lisa projected on to the direction to the source @xmath188 and the gw frequency will be added to the phase of the gw signal .
this doppler phase will be a function of time which will be added to the monochromatic part of the phase @xmath189 of the signal .
we assume in this analysis that this phase has been accounted for when integrating the signal .
one way is to ` stretch ' the time - coordinate so that the signal appears monochromatic ( a technique well - known to radio astronomers)@xcite .
an important case arises when we consider gw of low frequencies , say below 3 mhz . for this case
it is possible to obtain analytical expressions for the optimal snrs . a large fraction of the sources for lisa fall into this category , for example , massive / supermassive blackhole binaries , several galactic and extragalactic binaries which contribute to the ` confusion noise ' .
we need to compute the @xmath116 and @xmath117 in order to compute the snr matrix @xmath105 and then obtain its eigenvalues . integrating the eigenvalues for the period of observation will yield the relevant snrs . to this end , we expand @xmath190 and @xmath191 to the lowest order in the dimensionless frequency @xmath192 : @xmath193 where , @xmath194 where , @xmath195 is the unit vector in the direction of @xmath196-th spacecraft and @xmath197 and the angles @xmath198 refer to the lisa frame ( as before we have dropped the subscript ` @xmath15 ' ) .
the transfer functions for the four other elementary data streams are obtained by cyclic permutations . in order to get the @xmath199 we must operate on the @xmath20 with the polynomial operators @xmath200 given in eq.([basis ] ) and
then scale them by the averaged signal divided by the noise .
thus , we obtain after some algebra , @xmath201 where , @xmath202 the vectors @xmath116 and @xmath117 are expressed in the @xmath43 basis and in this basis they have real components .
we observe the following properties of the vectors : * the vectors @xmath116 and @xmath117 lie in the @xmath203 plane and the @xmath204 component is zero for both vectors .
this can be understood if we recall that @xmath204 is proportional to symmetric sagnac combination @xcite which is insensitive to gw at low frequency .
* the apparent motion of a gw source in the lisa frame can be optimally and continuously tracked by @xmath205 and @xmath206 by rotating the pair @xmath207 by @xmath208 at each instant of time as follows : @xmath209 here @xmath205 and @xmath206 are unit vectors in the directions of @xmath116 and @xmath117 respectively .
thus , optimally tracking a source amounts to orienting the data combinations along @xmath116 and @xmath117 .
* moreover , the vectors are orthogonal : @xmath210 in this low frequency limit .
the orthogonality implies that these eigen - observables @xmath116 and @xmath117 have zero response to @xmath74 and @xmath73 polarizations of gw respectively .
we use this fact to estimate the polarisation angles ; namely @xmath211 at the end of this section .
the eigenvalues of the eigen - vectors @xmath116 and @xmath117 are just @xmath134 and @xmath135 and are explicitly given by , @xmath212 the eigenvalues are the squares of the instantaneous snrs .
we notice that @xmath213 is the maximum instantaneous snr obtained when the source lies at the poles @xmath214 or @xmath215 in the lisa frame .
if the source is observed over a period of an year , the eigenvalues must be integrated over this length of time .
we notice that the eigenvalues do not depend on @xmath53 .
thus our next task of integrating the snr becomes somewhat simplified . for a given source direction @xmath174 the corresponding track @xmath216 of the source in lisa frame
is given by eq.([trns ] ) .
we may substitute these values into the integrals for the snrs and the integrals yield simple analytical expressions for the snrs if the integration is over an integral number of years .
the instantaneous snrs however change with time . in the fig.[snrtime ] we show how the snrs change with time as lisa orbits the sun during the course of a year .
the various snrs are shown for a source lying in the ecliptic plane , @xmath217 for the gw frequency of 1 mhz .
we have chosen this direction because the snrs show considerable variations during the course of a year . and
snr@xmath150 as functions of time for the source direction @xmath218 at the gw frequency of @xmath219 mhz in units of @xmath213.,title="fig:",scaledwidth=50.0% ] [ snrtime ] integration over a period of @xmath220 leads to the following results : @xmath221 where , @xmath222 and @xmath223.\end{aligned}\ ] ] we have purposely not ` simplified ' the formulae in powers of @xmath224 because in this form it is easy to see the limits @xmath225 corresponding to @xmath226 respectively .
infact since only @xmath224 occurs in the expressions of @xmath227 there is symmetry about the ecliptic plane .
the network snr is just the root mean square of the two snrs : @xmath228 where , @xmath229 . in fig.[intsnr ] we plot the functions @xmath227 and @xmath230 as functions of @xmath231 between @xmath232 .
the factors @xmath233 are of the order of unity and the snr@xmath234 gives essentially the integrated snr of a gw source .
recall that this is an snr averaged over the polarisations . and
@xmath230 as functions of @xmath231,title="fig:",scaledwidth=50.0% ] [ intsnr ] we observe from fig.[intsnr ] that the maximum integrated snr is obtained for sources lying in the ecliptic plane @xmath235 .
this can be readily explained from fig.[apptrk ] where the trajectory of such a source is plotted .
one observes that a large fraction of the orbit in the lisa frame is _ away _ from the lisa plane @xmath236 . as seen from eq.([eigsnr ] ) the sensitivity of lisa increases as we go away from the plane of lisa , that is , towards the poles .
we also observe that the variations in the three curves are small ; among these the network snr shows the highest variation of @xmath237 .
this shows that lisa has a more or less uniform average response over the year as it moves in its orbit .
the above formulae are also more generally valid if the integration time @xmath186 is taken to be in integral multiples of @xmath170 ; then the @xmath170 in eq.([snr0 ] ) should be replaced by @xmath186 .
we compute the snrs of six binaries in our galaxy which give high average snr ( averaged over polarisations ) if we integrate the snr optimally over the period of a year .
the following table lists six binary systems which give the network snrs ranging from about 3 to over 100 .
the snrs have been computed assuming circular orbits for the binaries .
the information about the binaries has been obtained from @xcite .
we observe that the binary masses are small ; @xmath238 and companion mass @xmath239 .
the reason for such small masses is that the orbital periods of the binaries must be short , in this case ranging from about .01 to .03 of a day .
the gw quadrupole frequency @xmath240 is related to the orbital period by the equation : @xmath241 thus the sources radiate gw at frequencies @xmath242 1 or 2 mhz , the band in which lisa has maximum sensitivity .
this yields the high snrs .
they are also close by ; @xmath243 or 200 pc which tends to increase the raw gravitational wave amplitude @xmath61 .
note that the snrs listed in the table are average and actual observations can yield different values depending on the orientation of the binary orbit .
also the observables used are optimal in the average sense .
if the orientation of the binary is known then in general a better observable can be found .
.in this table we list six binaries in our galaxy whose parameters are known . for these
we compute the optimum snrs where the optimisation has been performed after averaging over the polarisations . [ cols="^,^,^,^,^,^,^,^,^,^",options="header " , ] [ bin ] we showed in previous sections that if we do not have any prior information of the inclination angle @xmath81 of the binary orbit and the the angle @xmath82 , then tracking the source with @xmath205 and @xmath206 are optimal in the average sense . typically , the orbital inclination is difficult to estimate from other astrophysical observational means .
however , for binaries with known masses and distances ( e.g. binaries in table-[bin ] ) , we can estimate @xmath211 from the output of eigen - observables @xmath116 and @xmath117 .
the integrated snrs over the period of one year of @xmath116 and @xmath117 , _ without averaging over polarisations _ are given by @xmath244 where @xmath245 and @xmath246 ( see eq .
( [ sig ] ) ) .
we note that here the factor @xmath247 appears , since there is no averaging over the polarisations .
the @xmath248 can be estimated , if the snrs appearing on the lhs of eq .
( [ est ] ) can be measured .
further , straightforward algebra shows that @xmath249 from the above equation , we can estimate @xmath81 . substituting back in @xmath250
, one can also estimate @xmath82 if needed .
this exercise can be carried out for the binaries listed in the table [ bin ] . in this section
we relax the condition of dealing only with low frequencies and consider the entire band - width of lisa .
we compare the sensitivities for lisa obtained by using the optimum snr with the michelson data combination usually denoted by @xmath1 . as a polynomial vector @xmath24 it is given by : @xmath251 when we set all the @xmath252 to be equal , the factor @xmath253 factors out and one is left with a simple polynomial vector @xmath254 .
this combination has been used to plot the standard lisa sensitivity curve .
two other michelson observables are obtained by cyclic permutations called @xmath255 and @xmath256 in the literature @xcite . in this section
we will compare the sensitivities obtained by using the eigen - combination @xmath257 and the network observable with the michelson combination @xmath1 and the observable @xmath2 obtained by ` switching ' the michelson s @xmath258 optimally , that is , @xmath259 . extending the definition of sensitivity of a data combination from earlier literature @xcite to an observable
, we define the sensitivity of an observable @xmath260 as , @xmath261 where @xmath262 is the integrated snr over a observation period @xmath186 .
the number @xmath263 has been chosen following earlier literature .
( [ eq : intsen ] ) for a fixed data combination @xmath264 reduces to the standard one in literature : @xmath265 where @xmath266 . as before we take @xmath220 for plotting the sensitivity curves in the figures below .
the results of our findings are displayed in the plots .
fig.[fig : sens1 ] displays the sensitivity curves for the observables ( a ) michelson - @xmath1 ( dotted curve ) , ( b ) switched michelson @xmath2 ( dash - dotted curve ) , ( c ) eigen - combination @xmath257 ( dashed curve ) and ( d ) network observable ( solid curve ) . and network for the source direction @xmath267.,title="fig:",scaledwidth=50.0% ] [ fig : sens1 ]
we observe that the sensitivity over the band - width of lisa increases as we go from ( a ) to ( d ) .
we observe that the @xmath2 does not do much better than @xmath1 .
this is because for the source direction chosen @xmath1 is reasonably well oriented and there is not much switching to @xmath255 and @xmath256 combinations .
however , the network and @xmath257 observables show significant improvement in sensitivity over both @xmath1 and @xmath2 .
the sensitivity curves ( except @xmath1 ) do not show much variations for other source directions and the plots are similar .
the quantitative comparison of sensitivities is shown in fig.[fig : snrratio ] .
with @xmath2 for the source direction @xmath267.,title="fig:",scaledwidth=50.0% ] [ fig : snrratio ] here we have compared the network , and the eigen - combinations @xmath139 with @xmath2 . defining : @xmath268 where the subscript @xmath269 stands for network or @xmath270 and @xmath271 the snr of the observable @xmath2 , we plot these ratios of sensitivities over the lisa band - width .
we notice from the @xmath272 curve that the improvement in sensitivity for the network observable is about 34@xmath0 at low frequencies and rises to nearly 90 @xmath0 at about 20 mhz , while at the same time the @xmath257 combination shows improvement of 12 @xmath0 at low frequencies rising to over 50 @xmath0 at 20 mhz .
finally , fig.[fig : sens - net ] exhibits the sensitivities of the network observable over various source directions .
since the sensitivity of this observable is independent of @xmath273 , we plot the curves for several values of @xmath274 degrees . also since the network observable possesses reflection symmetry about the ecliptic plane @xmath275
, we do not need to plot the curves @xmath231 between @xmath276 and @xmath277 .
the important observation from this figure is that not much variation in sensitivity is seen as all source directions are scanned .
thus the network observable integrated over the year has essentially uniform sensitivity to all source directions over the frequency range @xmath278 hz . ,
title="fig:",scaledwidth=50.0% ] [ fig : sens - net ]
we have shown how the snr can be optimised for a gw source with known direction but with unknown polarisation . while obtaining the snr the signal
is averaged over the polarisations and then optimised .
because of this procedure we lose out to some extent on the snr but on the otherhand it leads to robust results .
the optimisation methods are algebraic in that one must solve an eigenvalue equation to determine the optimum snrs .
we separately deal with the low frequency case as it is of considerable astrophysical importance - a large fraction of gw sources are expected to be of this category .
secondly , it lends itself to simple analytical approximations which throw light on the results obtained .
lastly , we deal with the general case covering the full band - width of lisa .
we have compared the sensitivities obtained with our strategy to those obtained in the standard way .
we find that the improvement in sensitivity of the network observable over the @xmath1 or @xmath2 ranges from about 34% to nearly 90% over the bandwidth of lisa for a source lying the ecliptic plane .
finally we present a list of few binaries in our galaxy for which the optimal snrs and the network snrs have been computed .
we also describe a method of extracting information about the inclination angle of the orbit of the binary if @xmath183 can be measured .
the authors would like to thank the ifcpar under which this work has been carried out .
a. pai would like to thank the cnrs , observatoire de la cte dazur for post - roug position under which this work was carried out . 1 p. bender _
et al . _ "
lisa : a cornerstone mission for the observation of gravitational waves " , system and technology study report esa - sci(2000 ) 11 , 2000 .
s. v. dhurandhar , k. rajesh nayak , j - y .
vinet , _ phys .
* d 65 * , 102002(2002 ) . t. prince , m. tinto and s. larson , phys .
d 66 * , 122002 ( 2002 ) .
k. rajesh nayak , a. pai , s. v. dhurandhar and j - y .
vinet , _ class .
quantum grav _ , * 20 , * 1217(2003 ) .
the michelson combination was first described in schilling , _ class .
quantum grav _ , * 14 * , 1513 ( 1997 )
. j. w. armstrong , f. b. eastabrook and m. tinto , _ ap .
j _ , * 527 * , 814(1999 ) .
f. b. estabrook , m. tinto and j.w .
armstrong , _ phys .
rev _ , * d 62 * , 042002(2000 ) .
meliani , j.c.n . de araujo and o.d aguiar , astron .
astrophys . , 358 , 417(2000 ) ; h. ritter and u. kolb , astron .
astrophys , 129 , 83(1998 ) .
p. r. brady , t. creighton , c. cutler and b. f. schutz , _
phys.rev_ , * d57 * , 2101(1998 ) . b. f. schutz , `` data processing analysis and storage for interferometric antennas '' , in d. g. blair , ed .
, _ the detection of gravitational waves _ , cambridge university press , cambridge england , 406(1991 ) . | it was shown in a previous work that the data combinations canceling laser frequency noise constitute a module - the module of syzygies .
the cancellation of laser frequency noise is crucial for obtaining the requisite sensitivity for lisa . in this work
we show how the sensitivity of lisa can be optimised for a monochromatic source - a compact binary - whose direction is known , by using appropriate data combinations in the module .
a stationary source in the barycentric frame appears to move in the lisa frame and our strategy consists of _ coherently tracking _
the source by appropriately _ switching _ the data combinations so that they remain optimal at all times . assuming that the polarisation of the source is not known
, we average the signal over the polarisations .
we find that the best statistic is the ` network ' statistic , in which case lisa can be construed of as two independent detectors .
we compare our results with the michelson combination , which has been used for obtaining the standard sensitivity curve for lisa , and with the observable obtained by optimally switching the three michelson combinations .
we find that for sources lying in the ecliptic plane the improvement in snr increases from 34@xmath0 at low frequencies to nearly 90@xmath0 at around 20 mhz .
finally we present the signal - to - noise ratios for some known binaries in our galaxy .
we also show that , if at low frequencies snrs of both polarisations can be measured , the inclination angle of the plane of the orbit of the binary can be estimated . |
the fundamental problem of aerodynamics is to find the action force allocated along with the surface of a body in the flow ( such as the lift and drag force on airfoils , etc ) , thus we shall first study how the physical quantities distribute in the entire flow field .
the relevant theory development peaked at the time shortly after joukowski and prandtl dedicated their effort .
however navier - stoke equation and reynolds equation had become to their fellow researchers formidable mathematical obstacles ahead of them . in order to attain accurate data records in a flow field
, we have to be aware of the complete figure of the field with great details , such as the regulatory development routine for the boundary layer , the laminar - turbulent transition , the development of the separation region , and the mechanism of turbulence energy dissipation in all sections . only through unveiling the mysteries of these principles
, can we learn more about the pressure distribution as well as the appropriate circumstances for applying euler equations .
when we study turbulence into depth , we shall refer to a few people of importance : g.i taylor , l. prandtl , von karman , kolmogorov , and r. kraichnann .
it s worthy to note that , while potential flow theory and viscous flow theory can neither provide complete sustainable explanations for the physical principles behind the development of a flow field , the quantum flow field we discuss in this paper shall be recognized as a hypothesis .
the studies we are doing today , apart from what our elder generation of researchers had aimed for , are to actualize the quantization of flow fields by incorporating a generalized de broglie relation .
the diagram of the quantized flow field is as shown in _
figure 1_. the topic regarding traditional flow field problems would focus on the physical quantity distributions of a uniform straight flow along with the surface of a rigid body in an arbitrary shape , and on the computation of the resulting force distributions acting on the particular object . similarly to the classical potential theory where we use a dipole instead of a sphere , or use the distribution function of a source / sink instead of an ordinary airfoil model , in the quantization theory of a flow field , we replace the uniform straight flow with a plane wave and the body in the field with a generalized potential function .
but we ought to mention that here the plane wave , is in fact a generalized de broglie wave , i.e. a material wave ; is not an oscillation from any classical physics phenomena .
once the geometrical figuration of an object in the flow field is steady to the observers in classical potential theory , there exists an exact potential function , which is the potential of velocity , and this potential function will be the solution of the entire flow field . in resemblance of the potential theory , the quantization method of a flow field
thus seeks a particular wave function for some generalized schrodinger equation .
mathematically speaking , the generalization of de broglie relation does nt encounter any difficulties in the effort doing logical advancement .
the following equations hold up the fundamental structures in the quantization of a flow field : 1.generalized de broglie relation : @xmath0 @xmath1 2.generalized schrodinger equation : @xmath2\varphi = e_{\infty}\varphi\ ] ] 3.generalized velocity(in probability ) distributions : @xmath3\ ] ] the above equations are the theoretical framework of a flow over a body under certain circumstances completely made up of material waves , as already shown in _ figure 1_. as indicated , @xmath4 and @xmath5 are respectively the quantity of the momentum and of the kinetic energy for the uniform straight flow in unit volume . and @xmath6 , where @xmath7 is the generalized planck constant , its according dimension is the fluid angular moment in the unit volume .
@xmath8 is a generalized de broglie wavelength ; @xmath9 , @xmath10 , @xmath11 is the generalized potential function ; its dimension corresponds to the energy in unit volume ; similarly , the dimension for @xmath12 stands for the velocity .
if we plug in the generalized de broglie relation ( 1 ) into the explicit expression for reynolds number : @xmath13 we can get the most significant dimensionless quantity in our wave features , i.e. the generalized re number : @xmath14 these are the _ similarity criteria _ for dynamics in a quantized environment , and in the above formula , both @xmath7 and @xmath15 are attached to various numerical values with respect to different fluid mediums ; in addition , they will have the same dimension .
the existing and only admitted dimensionless parameter for all types of waves diffractions exhibits as @xmath16 .
the ratio for @xmath17 settles the exact correspondence in coordinate parametrization between @xmath18 and @xmath19 .
as shown in figure 2 , when we set the origin for polar coordinates to be the center of a sphere , the generalized potential function @xmath11 would have a simplified expression written in polar coordinates : @xmath20 and @xmath21 where @xmath22 is the radius of the spherical body . here
@xmath23 is as required for the boundary condition .
the formulas ( 5 ) and ( 6 ) are the _ no slipping condition _ and _ impenetrability condition _ on the surface of the body in a quantized flow field . noticing that the problem of the flow over a sphere , as we have discussed from above , has turned into a spherical symmetric generalized potential function acting on the plane wave presenting the uniform straight flow as a scattering process , and
such a problem can have a relatively simple solution with partial wave method . at this moment
the generalized schrodinger equation may have an asymptotic solution : @xmath24 where @xmath25 which is the so called scattering amplitude , and @xmath26 is the differential cross section .
@xmath27 correspond to the partial waves @xmath28 etc .
@xmath29 is the phase shift for the @xmath30 partial wave .
and @xmath31 is the @xmath30 term for legendre polynomial .
substituting ( 7 ) into ( 4 ) , we can obtain the asymptotic solution for the velocity distribution in probability , i.e the solution to the far - field wake behind the spherical body .
because of our limitation on the computational skills and the enormous amount of work required for further analysis , here we only discuss the velocity distribution in probability parallel to the z - axis for some cross - section of a wake .
its asymptotic solution has the form : @xmath32\ ] ] where @xmath33 , @xmath34 and @xmath35 are the real and imaginary parts for @xmath36 .
this is an approximated one omitting some higher order terms when @xmath37 , _ figure 3 _ displays it in its typical prominence with 2 examples of @xmath38 ( as @xmath39 grows to @xmath40 ) and @xmath41 ( as @xmath39 grows to @xmath42 . ) in a graph carrying the main properties of the curve @xmath43 , it demonstrates which we shall consider as the basic form of what @xmath43 really is , the velocity distribution in terms of its probability natures , and thus we shall average @xmath43 in some way to get a result assimilating the curves being confirmed from wake attending experiments .
but it s worth noting that even if we do nt take some form of average to the @xmath43 curve , there still always exists a _ deficit area _ on the graph for the velocity near z - axis ( when @xmath44 ) . when @xmath45 becomes big enough
, we can see a special character called _ modulation _ in the `` tail '' ( i.e. perpendicular to the z - axis ) of the curve @xmath43 describing the velocity in a quantized wake .
the _ `` modulation '' _ implies that when @xmath45 ( @xmath46 ) increases , wake width decreases , and it may also suggest there exists a multitude of _ wave packets _ in the flow field but outside of the wake .
we call the indented parts between these wave packets _ wasp - waist shape _ ( when @xmath45 is small , we ca nt see a second wave packet on the @xmath43 curve so we may consider the _ transient point _ to be the _ wasp - waist shape _ as the first wave packet narrowing down until where its two horizontal `` boundary lines '' become relatively parallel . ) these mysterious figures of the velocity distribution in probability are first found in data analysis , which shall draw our special attention since they may be encrypted with unpredictable physical importance . we can at least propose that one among many of the quantum flow field properties is the complex interference between the bouquets of de broglie waves .
in classical flow field theory , the transition mechanism turning from turbulence flow to laminar flow and the displacement of such a transition lack a creditable theoretical explanation , and therefore this becomes one of the key reasons why we are still having troubles determining the numerical values of the dynamical parameters ( especially for the lift and drag force parameters ) on bodies in a flow field .
however , if we use the quantization method of the flow over a body , including formulas of computations in the entire flow field and even for the computations of the mechanism of emerging turbulence flow , we may lead ourselves to an overall new prospect for analysis of the flow field .
in fact , @xmath47 i.e. the momentum distribution in probability can be extracted from the fourier transform of wave functions to the generalized schrodinger equation , which implies we may get the spectral function directly .
this is already defined by de broglie relation @xmath48 . and
it also constitutes the basics of the statistical description of turbulence , i.e. the concepts behind the energy spectral density function must come from the wave function in the momentum space , which is in resemblance with the turbulence spectral theory @xmath49 forwarded by g.i.taylor .
in contrast to what g.i.taylor believed about the mechanism of energy - spectral analysis , we think that it s more reasonable and perceivable to consider the family of waves in momentum space to be the generalized de broglie waves , which is a physics cause in principle of the energy spectral density function . by the observation of some phenomenon in data analysis
, we accommodate the u function into some form of its average and thus lead to the formula of the computation for drag under quantized environment : @xmath50 where @xmath51 is an undetermined constant , @xmath52 is the velocity distribution in probability and its average ( in quantum flow field ) , which is very close to the classical integral computational method for drag .
@xmath53 stands for some necessary complementing term added to the drag formula upon the fact that the occurring pattern in the graph of velocity distribution in probability inside a wake is in correspondence to the energy of turbulence fluctuation . but
mind that it differs from the energy dissipation term of the turbulence theory .
given that we already have the simplified expression of a solution to the velocity in probability of flow over a sphere ( as shown in equation ( 9 ) ) , taking account of the existence of the _ wasp - waist shape _ between wave packets , we shall expect that @xmath53 having a mathematical form of : @xmath54 where @xmath55 are parameters involving contents from the _ wasp - waist shapes , _ which shall be chosen based on individual experiments using simulations . in our data analysis by simulations ,
we emphasize some of the features which should be carefully studied in the future : \1 ) when @xmath45 approaches 0 , the velocity distribution in probability would be back to the classical velocity profile as for @xmath56 , i.e. the width of the wake grows to infinity ( or similarly @xmath57 . )
this shows for @xmath58 or for @xmath56 the quantized flow field and the classical flow field would lead to the same solution for the velocity distribution .
_ figure 4 _ shows the @xmath43 curves for @xmath59 with each curve rotated @xmath60 counter - clockwise . counter - clockwise for various @xmath45 values _
_ ] \2 ) if we pick out the first _ wasp - waist shapes _ from each @xmath43 graph and place them all together into one curve as _ figure 5 _ displays ( for 17 typical @xmath45 values : 0.000001 , 0.01 , 0.1 , 0.9 , 1.6 , 2.3 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 ) , we first see a drastic drop off between @xmath61 since we ca nt even see clear oscillating features especially when @xmath45 becomes smaller than 0.01 .
so we may assert that the first _ wasp - waist shape _ appears very close to @xmath60 , in other words , close to _ infinity _ if we use cartesian coordinates
. moreover the rate of change of the positions where the first wasp - waist shape figures appearing on @xmath43 curves getting closer to @xmath62 relatively accelerates for another time around the region @xmath63 .
we hold that this is a primary reason for the existence of the drag crisis .
if we leave the `` red curve '' on _ figure 5 _ only and adjust _ ka values _ to correct scales , i.e. @xmath64 , we will get a curve as depicted in the graph of _ figure 6_. but at this point our knowledge about the @xmath65 as well as where it would appear is minimal .
so we should note that our reading to the locations of the first @xmath66 now is highly depending on visual premises , although our assumptions as follows would maintain some theoretical validity , this evaluation should only be considered as a hypothetical reference .
notice the first wave packet does nt emerge distinctively until @xmath45 reaches @xmath67 , and the second wave packet appears when @xmath68 , so for @xmath69 we claim that the _ wasp - waist shape
_ sits on the region separating the first wave packet from the `` tail '' , i.e. the part where the `` amplitude '' of the wave increases and the part where the amplitude stays relatively constant ; and for @xmath70 ( where the growth rate of the wave amplitude is nt quite distinguishable ) , we pick our point based on the same assumption as before but this reading may include some inconsistency needed to be clarified in future studies because of the limited information we know about the nature behind these wave packets and time required for computations in depth supporting the legitimate evidence . _
( note that in figure 5 and 6 we only compute the u values from @xmath62 to @xmath71 for @xmath45 is greater than @xmath72 whereas @xmath73 to @xmath74 for @xmath75 and @xmath73 to @xmath76 for @xmath77 due to the exponentially increasing number of points we need to select for computations as @xmath45 grows bigger . ) _ in quantum flow field , @xmath78 would attain a wide range of efficiency .
this is related to the fact that the impact of s wave ( the component wave corresponding to the first term in the series embedded in the simplified solution of ( 8) ) on the cross - section of the scattering reduces to a much less significant extent , and causing by generalized ramsauer effect , reduces to 0 when @xmath45 reaches @xmath79 .
the generalized ramsauer effect depicts that every component wave would encounter a _ transmission _ when @xmath45 equals certain values .
then the cross - section would become smaller as a result of a transmission .
thus this overall affects the value of the drag coefficient @xmath80 .
notice that the _ wasp - waist shape _ on each @xmath43 curve is not unique , there exist multiple ones especially when @xmath45 becomes greater than 10 ( in our original calculation when @xmath45 grows up to 40 we saw as many as 4 of these shapes on each @xmath43 curve . )
this is a result emerging specifically in wave mechanics .
the quantization method transforms the classical flow field into a capricious interfering wave family where the existence for modulation is a primary characteristic . from the figure
we shall notice the undeniable importance of the first _ wasp waist shape _ to the calculation of drag coefficient ; thus it would be reasonable to predict the rest multiple wasp waist shapes on the @xmath43 curves when @xmath45 gets higher also devote into the formation of drag coefficients , i.e. in last integral formula presented above the @xmath81 term is an essential term directly related to the number or the properties of all the wasp waist shapes .
but at the moment we could not provide precise information about them or about the numerical relations between the other multiple wasp waist shapes and the drag coefficient .
in the future potential researches following this paper , we encourage people to specifically work on these features .
boundary layer theory by prandtl is the most widely accepted theory for illustrating the actual process of flow over a body to our current researchers in the field of aerodynamics .
yet we still lack some valid theoretical explanations for experimental data , i.e. when @xmath19 number grows greater than @xmath78 , why are both drag coefficient decreasing and its increasing gradual changes over the entire phase instead of sudden shifts ?
prandtl s well known experiment was to place a thin metal ring to the windward side of a sphere and to make the separation point on the surface of that sphere move to the lower course ; it does nt provide coordinate - dependent demonstration especially on the time - space evolution for this separation process . on the contrary ,
suppose the transition from laminar flow to turbulence flow is a morphological abruptness , then there will be a great scale of topological changes , and @xmath78 is a relatively small region , further , @xmath80 will have a relatively minor change .
however this does nt agree with experimental facts .
not only @xmath80 changes over a wide range of efficiency , but its numerical value grows back slowly as @xmath19 becomes greater than @xmath82 . and the separation point on the surface of the sphere moves to the upper course instead .
boundary layer theory has so far failed to provide valuable insights on this disagreement .
the authors are sincerely grateful to the people for making this paper possible including family and friends .
the late professors zhuang fenggan , and xu huafang , founders of china s studies in aerodynamics who had encouraged the authors for the development of this theory ; professor li chunxuan at beihang university who offered treasured consultation for advancing this paper ; professors zhang fuqing , liu ziqiang and yuan youxin for their detailed advice during the discussions about this paper ; mrs.tan liuning for her generous help financially and her first approach in some preliminary computations which turned out to be significant results
. pratyush sarkar , ethan yale jaffe , freid tong , michael yu , who all have been great friends and been utterly patient providing the author professional opinions regarding the paper and other mathematical topics ; yuqing tang , tom wang , and guanchu liu for their help initiating this paper ; li , andrew , tomas , zhifei , eric , yuan , david , christopher , adam , samer , jonathan , and all the other friends without whom this paper can not be possible . formulas ( 4 ) to ( 8) @xcite have the almost identical presentations to the similar setting ups for the quantum scattering theory .
the referenced books are the ones we used when consulting background knowledge from quantum mechanics @xcite @xcite , but we believe most of the earlier books studying quantum should all have these formulas and it s hard to track the true origin of them
. the theoretical as well as the computational results presented in this paper do nt have relative fellow researchers work to be compared with , and the theory been developed behind this paper originated from the author yufei liu s book _ an introduction to quantum methods for flows around a body _ published in 2013 ( currently only formally available in mandarin . ) as almost all of the named theories or problems in the text aside from the new method introduced the first time are very well known and should not cause any confusion , we decided not to list any specific articles or books for references aside from the above ones . | in this paper we briefly introduce the quantum methods for computations of the drag coefficients for flows around a body , using the flows around a rigid sphere as an example , and we aim for comparing the wake under quantized environment and the classical computational strategies for finding the drag force .
this paper however does nt provide discussion on the pressure distribution over the surfaces of a spherical body and neither the analytical solution for the flows around an airfoil due to the greatly limited computational capacities and resources . |
the identity of the non - baryonic dark matter , believed to account for about @xmath3 of the total density of the universe @xcite , is one of the major unanswered questions in cosmology today .
there is much speculation that the dark matter consists of massive non - standard model elementary particles .
the so - called `` wimps , '' or weakly interacting massive particles , are the best known such candidate .
these are often taken as the lightest supersymmetric particle , with a mass of no more than about @xmath4 @xcite .
any dark matter particle which is a thermal relic can not have a mass of more than about @xmath5 @xcite .
it has been shown that wimps could be captured by the sun @xcite , and that their subsequent annihilations could produce an observable flux of high - energy neutrinos @xcite .
recently , another candidate for particle dark matter has been proposed , the `` wimpzilla '' @xcite .
these are gravitationally produced towards the end of inflation by the interaction of the wimpzilla field with the inflating space - time , and in sufficient abundance to be the dark matter .
the wimpzilla is usually not assumed to be coupled to any other fields , although the case of coupling between the wimpzilla and the inflaton has been studied and wimpzilla production still found to be robust @xcite ; inflaton decays have also been shown to be a possible source of supermassive particles @xcite .
the original wimpzilla calculations were done in the context of chaotic inflation ( @xmath6 ) , but they have been shown to be abundantly produced for other inflaton potentials too @xcite .
wimpzillas are most efficiently produced at extremely high masses , on the order of @xmath7 . a crucial difference between wimps and wimpzillas is that the latter are never in thermal equilibrium , and as such , their masses are not thermodynamically constrained .
it also follows from this that their present - day abundance does not depend on whether they have strong , weak , electromagnetic , or only gravitational interactions ( although there are other considerations which generally rule out charged dark matter @xcite ) .
we assume in this paper that the non - baryonic dark matter consists of wimpzillas with strong interactions , referred to from now on as `` simpzillas . ''
simpzillas , like thermal wimps , can be captured by the sun , and their annihilations in the solar core can produce high - energy neutrinos .
we present the results of a detailed monte carlo calculation simulating the propagation of high - energy neutrinos from simpzilla annihilations through the sun , and also taking oscillations in matter and vacuum into account .
our general conclusion is that the flux of high - energy neutrinos from simpzilla annihilations should be well above background for a broad range of parameter space , and observable by next - generation neutrino detectors such as icecube .
albuquerque , hui and kolb @xcite , henceforth ahk , derived the capture rate of simpzillas by the sun .
the capture of thermal wimps by the sun had previously been studied in @xcite .
ahk assumed that the dark matter consists entirely of simpzillas in a maxwell - boltzmann velocity distribution , and that the interaction cross section @xmath8 is of the order of the strong force , implying that the sun is many interaction lengths thick .
formulas for the capture rate are given in equations ( 2.5 ) and ( 2.7 ) of their paper .
we plot the capture rate @xmath9 in figure [ gamma_c_fig ] .
the capture rate has two different forms , determined by the efficiency of the simpzilla energy loss in the sun .
once captured by the sun , the simpzillas rapidly fall to the core and annihilate with each other .
ahk showed that equilibrium between annihilation and capture is reached very early in the lifetime of the sun , and would obtain today . in equilibrium , by definition , @xmath10 ( each annihilation destroys two simpzillas ) . note that this has the effect of making the equilibrium value of @xmath11 dependent only on the interaction cross section @xmath8 and not the annihilation cross section @xmath12 .
high - energy neutrinos are produced by the simpzilla annihilations , which produce a quark or gluon pair that then fragment into hadronic jets containing a large number of particles .
ahk used the fragmentation function formalism of @xcite to calculate the numbers of hadrons produced per annihilation . in the dense solar core ,
hadrons composed of light and charmed quarks lose most of their energy before decaying . hadrons with bottom and top quarks , however , have much shorter lifetimes and decay before substantial energy losses .
the neutrinos from these decays are at high energies . in this paper , as ahk did , we consider the high - energy neutrinos from top hadrons produced in simpzilla annihilations .
although high - energy neutrinos can also come from bottom hadrons , it is generally quite difficult to calculate their flux due to the large number of @xmath13 meson decay modes , the products of which may interact in the sun before themselves decaying . since we find that the neutrino flux from top hadron decays alone
is well above background , including the neutrinos from @xmath13 decays will only add to an already detectable signal , although most of these will be below the @xmath14 cutoff energy we choose .
approximately @xmath15 top hadrons are produced per simpzilla annihilation , where @xmath16 .
top quarks almost always decay in the channel @xmath17 .
the @xmath18 , in turn , decays with equal branching ratios ( each about @xmath19 ) into @xmath20 , @xmath21 , and @xmath22 .
the @xmath18 , like the @xmath23 , decays before virtually any energy loss .
when the @xmath22 pair is produced by the @xmath18 decay , the @xmath24 also decays before losing much energy , producing a second @xmath25 .
about @xmath26 of the time , it also produces a second @xmath27 and another @xmath26 of the time a second @xmath28 .
ahk showed that the top hadrons have an energy distribution proportional to @xmath29 .
they also calculated the distribution of the subsequent neutrinos .
the total number of neutrinos produced above @xmath14 from this @xmath30 decay chain is @xmath31 where @xmath32 denotes the neutrino flavor ( @xmath33 , @xmath34 , or @xmath24 ) .
@xmath35 for @xmath36 and @xmath37 for the other two flavors , representing the fact that roughly twice as many @xmath25 are produced per annihilation .
the initial neutrino energy distribution calculated by ahk is @xmath38
\left [ \left ( e + m_t \right)^2 - m_t^2 \right ] \left [ \left ( e + m_w \right)^2 - m_w^2 \right ] } } \nonumber \\ & \times & \theta \left ( e - e_{min } \right ) \,.\end{aligned}\ ] ] here , @xmath39 .
this is near the lower limit of @xmath28 energies detectable by icecube @xcite , and also roughly the lowest neutrino energy possible in the two - body @xmath18 decay . in figure
[ initial_flux_fig ] , we show the initial simpzilla neutrino flux at the core of the sun .
= 3.25 in = 3.25 in
neutrinos with energies above about @xmath40 have significant interactions with matter as they propagate through the sun . the majority are charged - current ( cc ) interactions with nucleons .
the rest are neutral - current ( nc ) scatterings ; the cross sections for the latter are about one - third of the charged - current scattering cross sections at the same energy .
neutrino - electron scatterings in this energy range are negligible compared to neutrino - nucleon scatterings in the vicinity of the glashow resonance at @xmath41
. however , the fluxes we consider are far below this . ]
@xcite , and accordingly we do not consider them in our code .
we have calculated the charged and neutral - current neutrino - nucleon cross sections between @xmath42 and @xmath43 .
the most recent previous calculations in this energy range @xcite have assumed an isoscalar nucleon ; that is , one in which the quark distribution functions are the average of those for the proton and neutron . this is a good approximation for a medium such as rock ( the calculations were in the context of high - energy neutrino beams propagating through the earth ) .
however , the proton number density in the sun is between two and six times the neutron number density , and the isoscalar approximation does not hold .
we show the ratio of proton and neutron number densities in figure [ np_ratio_fig ] , calculated using the standard solar model of @xcite .
we have used the recently published cteq6-l ( leading order ) parton distribution functions @xcite , together with the deep inelastic scattering formalism of @xcite , to calculate the cross sections . given the energies involved
, we may neglect the electron and muon masses , and so the @xmath27 and @xmath28 cc cross sections are identical .
the @xmath24 mass , however , significantly suppresses the @xmath25 cc cross sections up to about @xmath14 .
although the effect of this on our calculation of the emergent simpzilla neutrino fluxes is minor ( we only consider energies above @xmath14 , and neutrino interactions in the sun are significant only above about @xmath40 ) , we have taken it into account .
the nc cross sections are identical for all three flavors .
we show our results in figures [ cs_cc.e_mu_fig]-[cs_nc_fig ] .
note that for @xmath44 ( and in the case of @xmath25 , above energies where the @xmath24 mass suppression is significant ) , the cross sections are approximately proportional to energy . at higher energies , the rate at which the cross section increases is suppressed by the gauge boson propagator @xcite . however , almost all of the simpzilla neutrino flux is below @xmath45 .
we have also calculated the differential cross sections @xmath46 , where the inelasticity parameter @xmath47 is @xmath48 and @xmath49 is the final charged lepton ( neutrino ) energy in a cc ( nc ) scattering .
we show these for @xmath50 and @xmath25 cc scattering in figures [ dsigmady.nu_e.cc_fig ] and [ dsigmady.nu_tau.cc_fig ] ; the others are similar .
our code uses these differential cross sections to obtain probability distributions for @xmath49 .
we find numerically that almost all neutrino interactions take place in the deepest part of the sun , where the density is greatest , out to about @xmath51 .
this is significant when oscillations are taken into account , because at high energies the oscillations and interactions effectively decouple in radius ( see section v ) .
= 3.25 in = 3.25 in = 3.25 in = 3.25 in = 3.25 in = 3.25 in
the charged leptons produced in charged - current interactions are very important in determining the emergent neutrino flux .
electrons and muons tend to be stopped by their electromagnetic energy losses .
however , taus have very short lifetimes .
they tend to decay before losing any significant fraction of their energy , and so the resulting @xmath25 are also high - energy .
cc interactions , in effect , absorb the @xmath27 and @xmath28 but regenerate the @xmath25 .
this phenomenon was first remarked upon by ritz and seckel @xcite in the context of a calculation of neutrino spectra from thermal wimp annihilations in the sun ( see also @xcite ) ; halzen and saltzberg @xcite have shown it to obtain for high - energy neutrino beams propagating through the earth ( see also @xcite ) . in the absence of oscillations
, this means that high - energy @xmath25 are much more likely than high - energy @xmath27 and @xmath28 to survive .
the @xmath25 are moderated down to lower energies , to the point where the probability of further interactions is small .
however , the total number of @xmath25 emerging is essentially equal to the initial number , whereas the @xmath27 and @xmath28 are attenuated .
we will discuss the properties of the @xmath25 versus @xmath50 emergent fluxes in greater detail in the penultimate section , including the effects of oscillations .
in this section , we consider the energy losses and decays of the charged leptons in the solar medium .
as the charged leptons move through the sun , they have electromagnetic interactions with the medium and lose energy .
( weak interactions , apart from decays , are insignificant for charged leptons at energies below @xmath52 @xcite ) . the general equation for energy loss as a function of column depth @xmath53 is @xmath54 where @xmath55 represents interactions with electrons and @xmath56 represent interactions with nuclei , with the summation being over the different processes : bremsstrahlung , pair production , and photonuclear scattering . below @xmath45 , where most of the simpzilla neutrinos ( and charged leptons from their cc scatterings ) are produced ,
electronic losses dominate . in most terrestrial situations ,
these are ionization losses which can be calculated with the bethe - bloch formula @xcite , but inside the sun the electrons are unbound . therefore , @xmath55 is the energy loss rate of a charged particle in a plasma @xcite , @xmath57 \,.\end{aligned}\ ] ] the units of @xmath55 in eq .
( [ ritz_seckel_eq ] ) are @xmath58 .
the quantities @xmath59 and @xmath60 are the atomic number and weight of the nuclei in the medium for the sun , which is ( approximately ) a mixture of @xmath61 and @xmath62 , we calculate the energy losses separately for the two isotopes and then weight them by their mass fractions .
the @xmath63 and @xmath64 are the lorentz parameters of the charged lepton and @xmath65 is the electron number density . the nuclear energy losses become dominant at energies above about @xmath45 , although as mentioned they are not very important for the simpzilla neutrino flux which tends to be at energies below this .
the formulas for @xmath56 are given , for example , in @xcite .
our code simulates both electronic and nuclear energy losses for @xmath34 and @xmath24 .
we do not follow @xmath33 since they never decay . in figure [ eloss_fig ]
, we show the muon and tau ranges and decay lengths in the core of the sun , which is where the majority of neutrino interactions take place .
the range is defined as the distance , in the absence of decays , over which the lepton would lose all its energy ; and similarly the decay length is the ( mean ) distance , in the absence of energy losses , the lepton would travel before decaying .
both muons and taus decay back into neutrinos , but the muons have lost almost all their energy by this point and essentially decay at rest .
the resulting @xmath28 are below @xmath66 and do not concern us .
the taus , on the other hand , have negligible energy losses except at very high energies ( @xmath67 ) , which again does not concern us since very little of the initial simpzilla neutrino flux is in this range .
the @xmath25 from the decay carries on average about @xmath68 the energy of the @xmath24 , or about @xmath69 the energy of the incident @xmath25 ( the mean energy of the @xmath24 relative to the incident @xmath25 can be obtained from the inelasticity parameter distributions ; in the energy range we consider , the mean value of @xmath47 is about @xmath70 , higher than for the energies at which neutrinos interact significantly in the earth ) .
one phenomenon our code takes into account is the production of secondary @xmath27 and @xmath28 in @xmath24 decays , as discussed by @xcite .
approximately @xmath26 of @xmath24 decays produce a @xmath28 ( for @xmath71 , a @xmath72 ( @xmath28 ) ) , and another @xmath26 produce a @xmath27 .
these secondary neutrinos are created with about half the energy of the accompanying @xmath25 , or about @xmath73 the energy of the incident @xmath25 .
when the @xmath25 experiences its last few charged - current scatterings and is downscattered to energies at which it escapes the sun , these secondary neutrinos emerge too .
@xcite considered high - energy neutrino beams from extragalactic point sources propagating through the earth , and showed that the additional flux of secondaries could substantially modify the detected signal .
the major decay channels of @xmath24 are @xmath74 ( 18% ) ; @xmath75 ( 18% ) ; @xmath76 ( 12% ) ; @xmath77 ( 26% ) ; @xmath78 ( 13% ) ; and @xmath79 ( 13% ) , where @xmath80 indicates other massive hadrons .
the combined @xmath25 decay distribution is given in @xcite .
we show it in figure [ dndz_fig ] , along with that of the secondary @xmath27 and @xmath28 , which is straightforward to obtain ( see e.g. @xcite ) .
our code randomly samples over the appropriate distributions to obtain the energies of the neutrinos from @xmath24 decays .
= 3.25 in = 3.25 in
both matter and vacuum oscillations are significant for simpzilla neutrinos , and have a large effect on the emergent flux . our code takes the oscillations of all three flavors into account .
we use values for the mass - squared differences and mixing angles which are consistent with recent observations . from the superkamiokande results @xcite , we choose @xmath81 , @xmath82 ( which is the upper limit on @xmath83 ) , and @xmath84 , where the angles in the three - flavor mixing matrix are as in eq .
( 3.1 ) of @xcite .
the other parameters are constrained by solar neutrino observations , such as those recently recorded by the sudbury neutrino observatory ( sno ) @xcite .
analysis of the sno data favors the `` large mixing angle '' ( lma ) solution @xcite , and so we choose values characteristic of this : @xmath85 and @xmath86 .
we assume the normal mass hierarchy with @xmath87 , and set the cp - violating phase in the mixing matrix to @xmath88 . to simulate the oscillations , we evolve the flavor amplitudes using the analytical solution for the time - evolution operator in @xcite .
our monte carlo step sizes are sufficiently small that the oscillation probabilities do not change significantly over the step , and the density at each step may be taken as constant . if the neutrino is determined to have had a charged - current interaction during the step , we sample over the current probabilities to find the flavor of the resulting charged lepton ; if the lepton subsequently decays at high energy , the resulting neutrino begins purely in that flavor . in figures [ osc.nu.100_fig ] and [ osc.nubar.100_fig ]
, we show the oscillation probabilities of @xmath40 neutrinos and antineutrinos traveling through the sun without interacting ( so their energies do not change ) .
this is roughly the lower limit of where neutrino interactions in the sun are significant .
we have noted above that most neutrino interactions take place between the center of the sun and @xmath51 . at higher energies ,
the oscillation wavelengths increase , to the point where the oscillation probabilities do not vary substantially over this distance .
interactions at energies much greater than @xmath40 thus effectively `` decouple '' in radius from oscillations ( though if our calculation were extended to include sterile neutrinos , this might not hold ; see @xcite ) .
we also would not expect different reasonable choices of the oscillation parameters to make a difference of more than a factor of two or so in the detection rates .
real - world detectors most easily observe the muons created in @xmath28 and @xmath72 charged - current interactions ( the @xmath27 and @xmath25 signals are much less directional ) , so the relevant issue is how much different parameter choices would change the emergent @xmath89 flux .
as can be seen from figures [ osc.nu.100_fig ] and [ osc.nubar.100_fig ] , the worst - case scenario would completely deplete the @xmath28 flux by maximizing the @xmath27 conversion probabilities ; however , the @xmath72 flux would be unaffected since the @xmath90 do not encounter a resonance and barely oscillate .
( a difference in the sign of @xmath91 , which is currently unknown and which we assume to be positive , would cause the @xmath90 rather than the @xmath27 to have a resonance , so the graphs would essentially be interchanged . )
the @xmath92 oscillations in the sun are vacuum oscillations with the atmospheric mixing parameters .
the @xmath27 resonance is governed by the mixing angle @xmath83 ( corresponding to @xmath93 ) and @xmath91 .
this is in contrast to normal solar neutrinos , whose msw resonance is governed by @xmath94 and @xmath95 . for the simpzilla neutrinos ,
that resonance is in the outermost layers of the sun at very low densities , and is not numerically significant compared to the other resonance .
we note that vacuum oscillations outside the sun are completely averaged out by the detector energy resolution ( see section vi ) .
we also note that although the detector is considered to be looking at upcoming simpzilla neutrinos which have come through the earth , the density of the earth is too high for msw oscillations inside it to be important .
= 3.25 in = 3.25 in
using the initial neutrino energy distribution given in ( [ initial_neutrino_flux_eq ] ) and taking all the physical effects discussed above into account , we have calculated the resulting neutrino flux at earth .
we have also , with icecube in mind , calculated the event rates it would produce in an idealized @xmath2 ice detector .
we have written a monte carlo code to simulate the propagation of high energy neutrinos through matter .
each neutrino is followed as it passes through the sun .
the step sizes are chosen to be much smaller than both the interaction length and the neutrino oscillation length . at each step
, we use a pseudo - random number generator to determine whether an interaction occurred over that step , whether it was a charged- or neutral - current interaction , and whether it occurred on a proton or neutron . in the event of a neutral - current interaction ,
the final neutrino energy is determined by randomly sampling over the appropriate inelasticity parameter distribution ( see section iii ) .
charged - current interactions are effectively measurements of the neutrino flavor . as such , when a charged - current interaction occurs , we randomly sample over the current oscillation probabilities to determine the flavor of the neutrino and outgoing lepton . when and
if the lepton decays back into a neutrino , it starts in a pure flavor state .
we calculate the initial energy of the lepton using the inelasticity parameter distribution .
the @xmath34 and @xmath24 produced in charged - current interactions are followed until they decay or their energies fall below @xmath14 .
the @xmath33 produced in @xmath27 charged - current interactions are not followed since they never decay .
( in practice , the @xmath34 almost never decay at high energies either , but we follow them for consistency . ) we calculate their energy losses over each step according to ( [ energy_loss_eq ] ) .
we also use their boosted decay lengths to randomly determine whether a decay occurs on each step .
if so , we sample over the appropriate decay distribution to determine the energy of the outgoing neutrino .
we also calculate the energies of the secondary @xmath27 and @xmath28 in @xmath24 decays using the energy distributions for those modes .
neutrino oscillations are simulated both in the sun and in the vacuum between the sun and the earth .
we average the oscillation probabilities over @xmath96 of the sun - earth distance , which is mathematically equivalent to a @xmath96 energy resolution in the detector ( characteristic of real - world detectors ) .
the detector is considered to be able to detect every neutrino charged - current interaction above @xmath14 taking place in a cubic kilometer of ice . in practice , real detectors are considerably more complex than this and a sophisticated treatment would require a detector monte carlo , which is beyond the scope of this work .
the detector is also assumed to have an angular resolution of @xmath97 , which is approximately the size of the sun on the sky .
having calculated the neutrino flux at the earth , we calculate the detection rate by integrating the product of the flux , the area of the detector ( @xmath98 ) , and the probability of the neutrino having a charged - current interaction , the integration being over the energy @xmath99 . the probability of a charged - current interaction in the detector is @xmath100 with the interaction length @xmath101 and @xmath102 is the size of the detector
. the number densities @xmath103 and @xmath104 are for ice , and the charged - current cross sections @xmath105 were discussed in section iii . the background flux from atmospheric neutrinos above @xmath14 is approximately @xcite @xmath106 \,.\end{aligned}\ ] ] integrating this over energy , the detection probability , the detector area , and the angular size of the sun gives about 2 atmospheric neutrino events per year above @xmath14 which come from the direction of the sun .
these will be almost exclusively muon neutrinos , if we assume the detector is looking at upcoming neutrinos .
we take this as our background .
we also note that the baksan neutrino telescope has placed a @xmath107 confidence level limit of muon fluxes from nonatmospheric neutrinos coming from the direction of the sun of about @xmath108 @xcite .
we take this as a rough upper limit .
the future analysis of the data from amanda @xcite may lower this limit . in figures [ flux.no_sec.no_osc_fig ] and [ flux.sec.osc_fig ] , we show our numerical results for the simpzilla neutrino flux at the earth . in figures [ event_rate.no_sec.no_osc_fig ] and [ event_rate.sec.osc_fig ]
, we show the event rates as a function of simpzilla mass @xmath109 for three different choices of the interaction cross section @xmath8 .
we show two cases : when neither oscillations nor secondaries are included , and when both are included .
we have found that the secondaries alone do not make a dramatic difference to the flux , largely because it falls over several decades in energy , and much of it is in a region where neutrino interactions are not significant and thus secondaries are not produced .
we accordingly do not show the other two permutations .
previously , only the case without oscillations or secondaries has been studied @xcite .
we discuss this case briefly .
as can be seen from figure [ flux.no_sec.no_osc_fig ] , the emergent @xmath25 fluxes exceed those of the other two flavors by more than the initial factor of @xmath110 ( see section ii ) .
this is due to the fact that the @xmath25 which experience charged - current interactions are not , as are the @xmath27 and @xmath28 , absorbed , but rather regenerated at lower energies , as remarked upon in section iv .
the unscattered components of the @xmath27 , @xmath28 , and @xmath25 beams relate to their initial fluxes as @xcite @xmath111 where @xmath112 is the `` transparency energy , '' defined as the energy such that the mean number of interactions experienced by the neutrino is 1 . equation ( [ unscattered_flux_eq ] ) obtains because the cross sections are approximately linear functions of energy in the range we are considering .
numerically , we find that @xmath113 for @xmath27 and @xmath28 , @xmath114 for @xmath25 , @xmath115 for @xmath90 and @xmath72 , and @xmath116 for @xmath117 .
we use ( [ unscattered_flux_eq ] ) as the emergent flux for the electron and muon neutrinos and @xmath28 ( as well as @xmath25 ) experience repeated neutral - current scatterings without any charged - current scatterings .
we neglect this neutral - current - scattered component here , although it is taken into account by our code . ] .
to model the scattered component of the @xmath25 , we use the observation of ourselves and several other groups that the scattered @xmath25 emerge in a roughly log - normal distribution .
we find numerically that about @xmath118 of the emergent @xmath25 and @xmath117 are scattered .
the log - normal distribution is defined as @xmath119 } \,,\ ] ] and when transformed to a distribution in @xmath120 , is a gaussian with mean @xmath121 and standard deviation @xmath8 ( whose units are decades ) .
note that @xmath122 , although they turn out to be similar .
we show how the numerical results compare to a log - normal distribution in figure [ nu_tau.log_normal.fixed_fig ] .
the correspondence is not exact , and the log - normal fit overestimates the high - energy neutrinos while underestimating the low - energy ones .
however , it is evidently a decent approximation .
the @xmath25 log - normal fit has @xmath123 and @xmath124 ; the @xmath117 has @xmath125 and @xmath126 . in figure [ flux.an.no_sec.no_osc_fig ] , we plot these analytical approximations .
the @xmath27 and @xmath28 fluxes are in qualitative agreement with the numerical results .
the @xmath25 fluxes are somewhat lower at low energies and higher at high energies than are the numerical results , again due to the log - normal underestimating the first while overestimating the second .
we note that eq .
( [ log_normal_dist_eq ] ) differs from eq .
( 4.6 ) of @xcite , leading to a much smaller difference between the @xmath25 and @xmath50 event rates ; this is due to our correctly taking the @xmath127 term in the jacobian factor in the log - normal distribution into account .
the oscillations clearly have a substantial effect on the flux .
the @xmath25 are depleted , whereas the @xmath28 and @xmath27 are enhanced . in a real world detector , this is advantageous , because it is generally easiest to observe muons from @xmath28 charged - current interactions .
the detectability of the simpzilla neutrino flux is therefore improved by oscillations . in figure
[ background_threshold_fig ] , we show the excludable region of simpzilla parameter space ; that is , the range of @xmath109 and @xmath8 which would be ruled out by the failure to observe a high - energy neutrino signal coming from the direction of the sun .
even better , of course , would be the observation of such a signal .
however , a more detailed analysis , which we leave to others , would be necessary to rule out other possible sources such as thermal wimps , particularly massive ones that could produce neutrinos in roughly the same energy range .
= 3.25 in = 3.25 in = 3.25 in = 3.25 in = 3.25 in = 3.25 in = 3.25 in
we have shown that the annihilations of simpzillas in the sun should produce a conspicuous high - energy neutrino signal in a @xmath2 ice detector for a fairly large range of simpzilla mass and interaction cross section .
much of this range of parameter space has not yet been ruled out by current observations .
our flux calculation takes all important physical effects into account , including neutrino oscillations , which approximately equalize the fluxes of the different flavors . in calculating the event rates ,
we have made some simplifying assumptions about the detector , and a more realistic calculation ( say , for icecube ) would take the actual detector response characteristics into account .
however , we would not expect this to alter our general conclusion that the simpzilla neutrino signal would be observable in such a detector
. the next generation of neutrino detectors should be able to either rule out a large range of simpzilla parameter space , or provide observational evidence for them .
i thank e. w. kolb for his guidance and support .
i also thank j. f. beacom for assisting me in a large number of ways , and pointing out the phenomenon of neutrino interaction - oscillation decoupling in radius .
i thank a. b. balantekin , a. de gouva , d. e. groom , d. mller , t. ohlsson , d. rainwater , j. truran , m. turner , and t. witten for useful discussions .
m. s. turner , arxiv : astro - ph/0106035 . j. edsjo and p. gondolo , phys .
d * 56 * , 1879 ( 1997 ) [ arxiv : hep - ph/9704361 ] .
k. griest and m. kamionkowski , phys .
lett . * 64 * , 615 ( 1990 ) .
w. h. press and d. n. spergel , astrophys .
j. * 296 * , 679 ( 1985 ) .
a. gould , astrophys .
j. * 321 * , 571 ( 1987 ) .
s. ritz and d. seckel , nucl .
b * 304 * , 877 ( 1988 ) .
d. j. chung , e. w. kolb and a. riotto , phys .
d * 59 * , 023501 ( 1999 ) [ arxiv : hep - ph/9802238 ] .
d. j. chung , e. w. kolb and a. riotto , phys .
lett . *
81 * , 4048 ( 1998 ) [ arxiv : hep - ph/9805473 ] .
d. j. chung , e. w. kolb and a. riotto , phys .
d * 60 * , 063504 ( 1999 ) [ arxiv : hep - ph/9809453 ] .
e. w. kolb , d. j. chung and a. riotto , arxiv : hep - ph/9810361 .
v. kuzmin and i. tkachev , phys .
d * 59 * , 123006 ( 1999 ) [ arxiv : hep - ph/9809547 ] .
d. j. chung , arxiv : hep - ph/9809489 .
r. allahverdi , k. enqvist and a. mazumdar , phys .
d * 65 * , 103519 ( 2002 ) [ arxiv : hep - ph/0111299 ] .
r. allahverdi and m. drees , arxiv : hep - ph/0203118 .
d. j. chung , p. crotty , e. w. kolb and a. riotto , phys .
d * 64 * , 043503 ( 2001 ) [ arxiv : hep - ph/0104100 ] .
a. gould , b. t. draine , r. w. romani and s. nussinov , phys .
b * 238 * , 337 ( 1990 ) .
e. nardi and e. roulet , phys .
b * 245 * , 105 ( 1990 ) .
i. f. albuquerque , l. hui and e. w. kolb , phys .
d * 64 * , 083504 ( 2001 ) [ arxiv : hep - ph/0009017 ] .
c. t. hill , nucl .
b * 224 * , 469 ( 1983 ) .
j. ahrens _ et al .
_ , _ icecube preliminary design document _ ( icecube collaboration , 2001 ) .
r. gandhi , c. quigg , m. h. reno and i. sarcevic , astropart .
phys . * 5 * , 81 ( 1996 ) [ arxiv : hep - ph/9512364 ] .
r. gandhi , c. quigg , m. h. reno and i. sarcevic , phys .
d * 58 * , 093009 ( 1998 ) [ arxiv : hep - ph/9807264 ] .
j. n. bahcall , _ neutrino astrophysics _
( cambridge university press , cambridge , u.k . ,
1989 ) .
j. pumplin , d. r. stump , j. huston , h. l. lai , p. nadolsky and w. k. tung , arxiv : hep - ph/0201195 .
e. a. paschos and j. y. yu , phys . rev .
d * 65 * , 033002 ( 2002 ) [ arxiv : hep - ph/0107261 ] .
m. kowalski , phys .
b * 511 * , 119 ( 2001 ) [ arxiv : hep - ph/0009183 ] .
f. halzen and d. saltzberg , phys .
lett . * 81 * , 4305 ( 1998 ) [ arxiv : hep - ph/9804354 ] .
s. i. dutta , m. h. reno and i. sarcevic , phys .
d * 62 * , 123001 ( 2000 ) [ arxiv : hep - ph/0005310 ] .
f. becattini and s. bottai , astropart .
phys . * 15 * , 323 ( 2001 ) [ arxiv : astro - ph/0003179 ] .
j. f. beacom , p. crotty and e. w. kolb , phys .
d * 66 * , 021302 ( 2002 ) [ arxiv : astro - ph/0111482 ] .
s. i. dutta , m. h. reno , i. sarcevic and d. seckel , phys .
d * 63 * , 094020 ( 2001 ) [ arxiv : hep - ph/0012350 ] .
d. e. groom _ et al . _
[ particle data group collaboration ] , eur .
j. c * 15 * , 1 ( 2000 ) .
d. e. groom , n. v. mokhov and s. i. striganov , atom .
data nucl .
data tabl . *
78 * , 183 ( 2001 ) .
p. renton , _ electroweak interactions _ ( cambridge university press , cambridge , u.k . ,
1990 ) .
y. fukuda _ et al .
_ [ super - kamiokande collaboration ] , phys . rev .
lett . * 81 * , 1562 ( 1998 ) [ arxiv : hep - ex/9807003 ] .
s. fukuda _ et al .
_ [ super - kamiokande collaboration ] , phys . rev .
* 85 * , 3999 ( 2000 ) [ arxiv : hep - ex/0009001 ] .
a. de gouvea , phys .
d * 63 * , 093003 ( 2001 ) [ arxiv : hep - ph/0006157 ] .
j. r. klein [ sno collaboration ] , arxiv : hep - ex/0111040 .
q. r. ahmad _ et al .
_ [ sno collaboration ] , arxiv : nucl - ex/0204008
. q. r. ahmad _ et al .
_ [ sno collaboration ] , arxiv : nucl - ex/0204009 .
p. i. krastev and a. y. smirnov , phys .
d * 65 * , 073022 ( 2002 ) [ arxiv : hep - ph/0108177 ] .
t. ohlsson and h. snellman , j. math .
phys . * 41 * , 2768 ( 2000 ) [ erratum - ibid . *
42 * , 2345 ( 2000 ) ] [ arxiv : hep - ph/9910546 ] .
t. ohlsson and h. snellman , eur .
j. c * 20 * , 507 ( 2001 ) [ arxiv : hep - ph/0103252 ] .
v. a. naumov , phys .
b * 529 * , 199 ( 2002 ) [ arxiv : hep - ph/0112249 ] .
i. f. albuquerque , j. lamoureux and g. f. smoot , arxiv : hep - ph/0109177 .
m. m. boliev , e. v. bugaev , a. v. butkevich , a. e. chudakov , s. p. mikheev , o. v. suvorova and v. n. zakidyshev , nucl .
suppl . *
48 * , 83 ( 1996 ) .
r. wischnewski _ et al .
_ , nucl .
* 75a * , 412 ( 1999 ) . i. f. albuquerque , j. lamoureux and g. f. smoot , arxiv : hep - ph/0204301 . | we calculate the fluxes and energy spectra of high - energy ( @xmath0 ) neutrinos from the annihilations of supermassive ( @xmath1 ) , strongly interacting dark matter particles in the core of the sun .
we take all significant aspects of neutrino propagation through matter into account , as well as oscillations in matter and vacuum .
we also calculate the resulting event rates in an idealized @xmath2 ice detector .
we find that the signal should be well above background and easily observed by next - generation neutrino detectors such as icecube . |
a micro - channel plate ( mcp ) photo - multiplier tube ( pmt ) provides a good time response .
our r&d work had obtained a transit time spread ( tts ) of @xmath4 ps for single photons with a channel - hole size of @xmath5 10 @xmath6m@xmath7 without a magnetic field ( b ) or with [email protected] t strength @xcite .
such an mcp - pmt made it possible to realize @xmath9 ps of a tof counter for 4 gev / c pion beams @xcite .
furthermore , an mcp - pmt with a multi - alkali photocathode maintains its timing performance up to an integrated irradiation of @xmath10 photons/@xmath11 or an integrated output charge amount of @xmath12 c/@xmath11 under an irradiation rate of @xmath13 photons/@xmath11/s @xcite .
the mcp - pmt having these superb properties lets us develop a new type of cherenkov ring imaging counter , a time - of - propagation ( top ) counter @xcite , for particle identification at a planned super - kekb factory with an expected luminosity of @xmath14 /@xmath15sec , which is 50-times as intense as the achieved luminosity of the current kekb factory , and is expected to survive for more than 14 years under this situation .
we here report on r&d results of a newly developed mcp - pmt , which is equipped with 4 anodes rather than a single anode , and shaped to be square , not round , from the required performance on the top counter .
when setting a multi - anode structure , an additional r&d issue arises : it shows a cross - talk effect .
while cross - talk basically occurs among the anodes , its influence is in most cases is harmless for measurements of the incident - particle energies or a number of tracks in terms of the pulse - height of output signals .
however , for a precise time measurement , as discussed here , the cross - talk deteriorates the timing performance , and sometimes makes it ineffectual , in much the same way that occurs for a vacuum pmt , as reported @xcite . based on our tests for various modified versions of the mcp - pmt
, we take a specific measure to reduce the cross - talk while keeping its time resolution , as reported below .
( we hereafter abbreviate mcp - pmt as simply pmt . )
we manufactured two versions of 4-channel linear - array multi - anode mcp - pmt s , and named them sl10 , as can be seen in fig .
the difference in the versions only concerns the electrode of the second mcp - layer , facing to the anodes : sl10-s has a uniform electrode over the whole mcp surface , while sl10-m has an electrode subdivided into 4 segments facing the individual 4 anodes .
a schematic of the internal structure and high - voltage ( hv ) supply network of sl10-s is drawn in fig .
both versions of sl10 s have the same characteristics , as listed in table [ t-1 ] .
they are two - stage mcp s with a multi - alkali photocathode ; the channel - hole size and the thickness of an mcp - layer are 10 @xmath6 m and 400 @xmath6 m , respectively . the anode plane has a size of 22@xmath1622 mm@xmath17 , between which there is a gap of 0.3 mm .
[ t-1 ] .sl10 characteristics . [ cols="<,^ " , ]
the performance of sl10 , square - shaped and a multi - anode pmt , is measured for single photons , and compared with those of a similar type of mcp - pmt s , but with a round - shaped and a single anode ( hamamatsu photonics k.k .
( hpk ) , r3809u-50 - 11x and -25x , referred to as hpk6 ( the channel - hole diameter of 6 @xmath18 m ) and hpk10 ( 10 @xmath18 m ) in @xcite , respectively ) .
a pico - sec light pulser ( hpk , plp-02-sldh-041 ) is used to generate light of wavelength @xmath19 nm with a pulse width of @xmath20 ps at a repetition rate of 1 khz .
its intensity is attenuated down to a single - photon level by diffusers and filters , and is then led to the sl10 surface by an optical fiber .
this system is set inside a magnet in the case of a measurement of the sl10 response based on the b field strength , where the sl10 axis is placed so as to be parallel along the b field .
the waveform of the output signals of the anode is recorded by a digital oscilloscope ( hewlett packard , infinium ) . to measure the charge and timing of the output pulse from individual anodes
, it is fed into an attenuator ( agilent , 8495b : attenuation = 0 - 70 db , frequency @xmath518 ghz ) and then to an amplifier ( hpk , c5594 : gain= 36 db , frequency = 50 khz - 1.5 ghz ) .
one of the amplifier outputs is fed into a camac adc ( 0.25 pc / count ) module , and the other output is sent into a camac tdc ( 25 ps / count ) module through a discriminator ( phillips scientific , model-708 ) with a threshold voltage of 20 mv .
it is @xmath211/7 of the pulse height . a drawing of a similar setup is shown in fig . 1 in @xcite
. data presented in this subsection were obtained for the sl10-s , but there is no essential difference from that of the sl10-m .
figures [ f-3 ] show the output signals for single photons under various b strengths .
the pulse height is @xmath2230 mv and the rise time is @xmath23500 ps at hv = 3.5 kv .
these properties are consistent with those observed by the round - shaped pmt with a single anode , particularly hpk6 ( see @xcite ) .
1.5 t , with hv = 3.5 kv.,width=340 ] figure [ f-4 ] shows the charge distributions of the output pulses under various b fields .
a single photon peak is clearly seen around the 45-th , 70-th , 60-th and 40-th channel at b=0 , 0.5 , 1.0 and 1.5 t , respectively . from these output charges for single photons ,
the typical multiplication gain @xmath24 was evaluated , as indicated in fig .
[ f-4.5 ] , with several different hv s from 3.1 kv to 3.5 kv with a 0.1 kv step : the maximum @xmath24 of @xmath25 was attained at b = 0.5 t. the behavior of @xmath24 vs b is quite similar to that of hpk10 , but its absolute magnitude is 5 - 10 times larger than that of hpk10 .
this difference is inferred to be due to a variation of individual mcp manufacturing batches .
vs b relations for single photons with different hv.,width=302 ] a huge rate of medium / high energy photons could be produced regarding the beams of the super - kekb accelerator ; their @xmath26 pair creation of the top quartz radiator then yielded many cherenkov photons as backgrounds . at our top - counter configuration
@xcite , the photon background rate at the super - kekb was estimated to be @xmath27/@xmath11/s on the pmt surface @xcite , by linearly extrapolating from the single rate of the present belle tof counter .
the irradiation of a pmt by such a high - rate of photons could cause a @xmath24 drop due to a finite recovery time of the effective hv field inside the channel - hole , apart from a shortening of the pmt lifetime .
the photon rate on the sl10 was varied by changing the light - pulser frequency from 1 khz up to 1 mhz , and the light intensity from 1 up to 600 photons / pulse .
the resulting @xmath24 variation is shown in fig .
[ f-5 ] as a function of the photon rate .
the sl10 stably functions up to a photon rate of @xmath28/@xmath11/s without significantly deteriorating its @xmath24 .
this rate corresponds to about 100-times higher than the expected rate at the super - kekb factory .
vs photon rate .
expected photon rate on the sl10 surface at super - kekb is indicated by the dashed vertical line.,width=302 ] the time - walk corrected time response for single photons is plotted in fig .
[ f-6](a ) .
the dominant peak ( @xmath29 tdc @xmath30 ) yields the time resolution of @xmath31 ps . the tail component is supposed to be due to the delayed photoelectrons bouncing from the surface of the first mcp layer .
[ f-6](b ) shows the relation of @xmath32 vs. b ; @xmath0 ps can be attained in the case of @xmath33 .
this performance is similar to those of both hpk6 and hpk10 ( see fig . 7 in @xcite ) . as a function of b. , title="fig:",height=219 ] as a function of b. , title="fig:",height=226 ]
the cross - talk among the anodes is a general issue for multi - anode detectors ; moreover , it is substantial for a precise time measurement of the single photons , as reported in @xcite . in the case of the top - counter
, @xmath2120 cherenkov photons could be detected by 15 pieces of the sl10 s , each of which is equipped with 4 anodes , on a side edge of the quartz - bar , when a charged track passes through the bar .
there are then appreciable probabilities that plural sl10 s detect @xmath34 photons in an event .
a primary single - photon would induce cross - talk on other anodes in the same sl10 , so that the pulse shape of the secondary arriving single - photon on the anode would be affected , and thus deteriorate the timing information .
we studied and then reduced the influence of the cross - talk by modifying the electrode structure as in the case of the sl10-s to sl10-m .
no b - field is applied in the following studies .
[ f-7 ] shows the anode signals from the sl10-s in the case that an anode , channel 2 , is irradiated by a sufficient number of photons , @xmath2120 , through a spot of 1 mm@xmath7 , to observe the cross - talk signals at any other channel .
the cross - talk pulses show a differential shape of the signal pulse , and its relative pulse height to the signal is @xmath211/5 @xmath35 1/3 , and does not depend on the distance among anodes that measure the signals and cross - talks .
as is can be seen in fig .
[ f-3 ] , the signal pulse height of single photons is @xmath36 mv , so that the induced cross - talk height for single photons is below the threshold voltage of 20 mv .
however , as demonstrated later in fig .
[ f-10 ] , the base - line fluctuation caused by the cross - talk would make the time measurement ineffectual for a consecutive photon on a channel other than ch.2 . .
, width=302 ] second signals can be seen at @xmath37 ns after the primary signals in fig .
it is caused by feedback @xmath38 ions , yielded in the multiplication process of the micro - channel plate ; the second pulses also induce cross - talk at other anodes .
we extensively studied the cross - talk on a multi - anode 16-channel linear array vacuum pmt @xcite , and here just also found the same phenomena for the sl10 .
the cross - talk in this case might be induced through , especially , the last electrode , facing the anodes , of the second mcp - layer .
an internal structure of sl10-s , and its readout and hv circuits are illustrated in fig .
the anode , @xmath39 mm@xmath17 , and the last electrode of the mcp separated by a 1 mm gap , would form a capacitance of @xmath401 pf ; a short hv line from the hv bleeder network to the last electrode would yield an inductance of @xmath4150 nh ; also , individual anodes are terminated with a resistance of @xmath42 @xmath43 .
we thought out that one way to reduce the cross - talk by disconnecting this rlc loop circuit is to make a segmentation of the mcp electrode facing to 4 anodes , and to prepare 4 separate hv networks , as illustrated in fig .
[ f-9 ] . since the mcp is a high - resistance material , 20 m@xmath43 per plate , the impedances between the segmented electrodes are quite large , electrically well disconnected .
anode signals of the sl10-m can be seen in fig .
[ f-10 ] , where the setup condition is the same in the case of fig .
[ f-7 ] , except that the number of photons irradiated is @xmath2110 , rather than @xmath2120 .
the pulse - height ratio between the cross - talk and the signal is now @xmath2110 .
it has been improved by a factor of @xmath212 , compared to that of sl10-s .
we prepared the setup shown in fig .
[ f-11 ] to measure the time resolution in a case that two photons hit different anodes of the sl10-s and sl10-m within an adjacent narrow time period , supposing cherenkov ring imaging photons in the top - counter .
the light beams from the pulser are divided into two by a half mirror .
the primary light is detected by ch.3 through a mask , and the divided and then delayed light is by ch.1 as shown in fig .
the delay time , @xmath44 , is adjusted to be within a range of up to @xmath453 ns , by changing the distance of the second mirror to the sl10 .
a light pulse is attenuated down to single - photon level by beam expanders and filters , upstream of the half mirror . the observed time resolution , @xmath32 , for the delayed photons is plotted in fig . [ f-12 ] as a function of @xmath44 by open circles for sl10-s and open squares for sl10-m , using a discriminator of the voltage threshold type ( phillips scientific , model-708 , threshold = 20 mv ) . for sl10-s , which provides @xmath46 ps in the case of single photons with no cross - talk ( see , fig.[f-6 ] ) , there is no data point at @xmath47 ns because the discriminator hits the cross - talk pulse before the delayed signal arrives . and ( b ) difference between the observed time and the delayed time for the delayed photons at ch.1 .
data with open circles and open squares are those obtained by sl10-s and sl10-m , respectively , using a voltage threshold type of discriminator ; the closed squares indicate data by the sl10-m , but with constant fraction discrimination .
, width=340 ] for sl10-m , the cross - talk pulse is sufficiently suppressed so that only the delayed signal is detected by the discriminator at any @xmath44 range .
however , @xmath32 strongly depends on @xmath44 . for @xmath48
ns , around which the cross - talk effect from the primary signal is the largest , @xmath32 becomes @xmath49 ps , while in the other @xmath44 region , where the fluctuation of the base line is rather small , @xmath50 ps .
even so , the latter resolution is worse than the @xmath51 ps attained at the incidence of sole single photons .
although sl10-m successfully reduces the cross - talk , it does not recover the original time resolution of @xmath51 ps for single photons .
the cross - talk effect also appears as a shift of the observed absolute time of the delayed photons , as plotted in fig .
[ f-12](b ) .
the difference , @xmath52 , between the observed time and @xmath44 behaves as expected from the cross - talk pulse ( see , fig .
[ f-10 ] ) . when the base line shifts to the positive side in voltage , @xmath52 moves to a positive delay , and vise - versa .
this fact indicates that the signal pulse is not broken away by the cross - talk , rather it keeps its intrinsic signal shape upon a tiny cross - talk disturbance .
we replaced a leading - edge type of discriminator with a constant fractional discriminator ( cfd , kaizu - works kn381 ) , since the latter provides an output pulse at zero - crossing timing of a differentiated input pulse so that it would not be strongly affected by the cross - talk , as the former would be . the resulting @xmath32 and @xmath52 with the cross - talk are plotted by closed squares with error bars in fig .
the time resolution largely improved to @xmath53 ps with no large variation in @xmath44 .
the timing shift has also been reduced by a factor of 2 : @xmath54 ps . however , because its maximum shift is about the same size as the time resolution it is not a trivial matter , unless it is corrected by referring to the timing of the primary signal .
we have developed a square - shaped 4-anode mcp - pmt , sl10 , to measure the precise timing of single photons .
it provides a performance that satisfies the requirement on our top - counter : @xmath55 , @xmath56 ps for single photons under a 1.5 t magnetic field strength , and durable up to @xmath28 /@xmath11/s of a high photon rate . in order to suppress the cross - talk effect on the timing measurement in the case of plural photon injection , we have taken two measures .
one way is to subdivide the electrode of the last mcp - layer into 4 pieces , facing the anodes : it successfully reduces the cross - talk size to one half .
the second way is to use a constant fractional discriminator , instead of the leading - edge type one .
the time resolution is largely recovered up to @xmath57 ps , while not yet restoring the original resolution of @xmath51 ps .
+ * acknowledgments * + this work is supported by a grant - in - aid for science research on priority area ( mass origin and supersymmetry physics , 18071003 ) from the ministry of education , culture , sports , science and technology of japan , and by a grant - in - aid for creative scientific research ( 18gs0206 ) from the japan society for the promotion of science .
n. kishimoto et al .
instr . and meth . a 564 ( 2006 ) 204 .
the mcp - pmt ( hpk - w : hpk6 equipped with a feedback protection layer between the photocathode and the first mcp ) held a stable quantum efficiency ( qe ) up to an integrated irradiation of @xmath58 photons/@xmath11 .
after the publication , we extended the measurement up to @xmath59 photons/@xmath11 , and no appreciable degradation of qe was still observed , as shown below .
m. akatsu et al . , nucl .
instr . and meth . a 440 ( 2000 ) 124 ; t. ohshima , icfa instr .
bull . 20 ( 2000 ) 2 ; t. ohshima , nucl .
instr . and meth . a 453 ( 2000 ) 331 ; m. hirose et al .
. instr . and meth . a 460 ( 2001 ) 326 ;
s. matsui et al . , nucl .
instr . and meth . a 463 ( 2001 ) 220 ; t. hokuue et al .
instr . and meth . a 494 ( 2002 ) 436 ; y. enari et al . , nucl .
instr . and meth . a 494 ( 2002 ) 430 ; m. akatsu et al .
instr . and meth . a 502 ( 2003 ) 133 .
a segment of top counters will comprize two different lengths of the quartz radiators : 2,000/1,000 mm@xmath60 400 mm@xmath61 20 mm@xmath62 , each of which equips 15 pieces of sl10 s at an end .
such 18 segments will form a barrel with r=1,200 mm from the beam - colliding point . | we have developed a 4-channel multi - anode mcp - pmt , sl10 , which exhibits a performance of @xmath0 ps for single photons with @xmath1 and @xmath2 under a magnetic field of b @xmath3 .
the cross - talk among anodes has been extensively studied .
we have taken two measures to suppress it : one is to configure the sl10 to an effectively independent 4 small pieces of mcp - pmt s by segmenting an electrode of the second mcp - layer ; the other is to use a constant fractional discriminator .
remarkable improvement has been achieved . , , , , , , , , , , , , , , , mcp - pmt , cross - talk , top counter |
geometrical frustration of antiferromagnetic interactions is the object of intense research as a route to generate novel exotic ground states .
first proposed by anderson for the triangular lattice @xcite , spin - liquid - like behavior is now commonly observed in many compounds with corner - shared triangular lattices .
archetypes are the kagome or the pyrochlore lattices @xcite .
the surge of interest for such weakly connected lattices in the 1990s came from ( i ) the lack of good experimental realizations of heisenberg triangular antiferromagnets and ( ii ) the idea that a weaker connectivity offers an additional ingredient to destabilize nel states and generate novel exotic physics . only very recently , the discovery of new experimental systems revived the interest in the pure triangular lattice . from the theoretical point of view , there was an early consensus that the heisenberg triangular antiferromagnet with nearest - neighbor coupling orders at @xmath6 in the so - called 120@xmath7 configuration of spins @xcite .
however , to date , there has been little theoretical exploration in the dynamics associated with the triangular topology . kawamura and miyashita proposed in the 1980s that , even though no magnetic transition is possible at a finite temperature , the system goes through a cross - over regime due to the proliferation of topological defects z@xmath1
observations of exotic fluctuating behaviors are at the heart of the recent experimental explorations of the triangular lattice since they might have some connection with the aforementioned topological excitations . in niga@xmath1s@xmath8 , a fluctuating behavior is observed down to low temperatures as compared to the exchange interaction @xcite and recently a transition to an ordered state was finally observed at @xmath9 k @xcite .
the case of cs@xmath1cucl@xmath8 @xcite is more controversial . initially thought to be an anisotropic triangular lattice
, the fluctuating character might be due to a very pronounced one - dimensional ( 1d ) character .
in the two previous examples , the interaction paths are not straightforward .
nacro@xmath1 , re - discovered in the context of the surge of interest for na cobaltates , offered a third , much simpler route to tackle the issue of the nature of the ground state .
as shown in @xcite , it features well - decoupled perfect heisenberg triangular planes .
a broad fluctuating regime was observed well below the temperature at which a peak was observed in the magnetic susceptibility and was tentatively associated with topological excitations such as z@xmath1 vortices . in this paper , we present results on another stable member of the acro@xmath1 family ( a = li , na , k ) , namely , licro@xmath1 . our main motivation is to probe the changes in the exotic dynamics observed in nacro@xmath1 through the modification of the structural parameters and , hence , a modification of the magnetic couplings .
as reviewed by collins and petrenko , most of the studied triangular compounds had been far from ideal since they presented either a large anisotropy of the interactions or a sizeable single - ion anisotropy and/or next - neighbor coupling @xcite . to cite some of them , compounds from the family @xmath13 , where @xmath14 is an alkali metal
, @xmath15 is a transition metal , and @xmath16 is a halogen atom , were very well studied in the past .
they have quasi-1d character , with coupling between planes much larger than the intra - plane one .
closer to the ideal heisenberg case are the systems that belong to the family v@xmath17 , where @xmath16=cl , br , and i. however , in these compounds single - ion anisotropy was observed while the triangular planes are weakly coupled . in the acro@xmath1 family ,
the magnetic structure is built by stacking well - separated , triangular planes of cr@xmath2 with spin @xmath18 in an @xmath19 sequence .
a schematic structure is given in fig .
[ structure ] .
the r@xmath20 m symmetry of the structure ensures the isotropy of the in - plane - exchange coupling .
also , at least in the first two members of the series , this coupling is restricted to the nearest - neighbors @xcite .
electron spin resonance ( esr ) measurements indicate a very small single - ion anisotropy , thus revealing a very pronounced heisenberg character which is quite common for cr - oxides @xcite . in this family ,
the @xmath21-lattice parameter increases sizably from 14.4 for li to 17.9 for k , which tunes the separation between cr planes . here
the @xmath21 parameter is given for the rhombohedral unit cell .
an early macroscopic susceptibility study on a wide temperature range presented in ref .
@xcite showed that all three compounds present antiferromagnetic correlations .
for licro@xmath1 , @xmath22 k and @xmath23 k , while for nacro@xmath1 @xmath24 k and @xmath25 k. structure is built of triangular planes of li , o and cr in an @xmath19 sequence . ]
neutron - scattering measurements on a single crystal of licro@xmath1 @xcite show the presence of three - dimensional ( 3d ) antiferromagnetic order that develops below @xmath26 k in a complicated 120@xmath7 structure based on 18 sublattices .
accordingly , macroscopic susceptibility and specific - heat - capacity data show maxima around @xmath27 k @xcite .
recent first - principles calculations on licro@xmath1 indicated the existence of weak coupling between the planes , which is possibly antiferromagnetic in nature @xcite .
a @xmath28li nmr study indicated the presence , above 60 k , of a line shift , which demonstrates the existence of hyperfine coupling between li and cr , probably mediated by the oxygen @xcite .
below this temperature , a wipe out of the signal occurs , probably because of the fast relaxation of nuclei in the critical regime around the phase transition . in a recent nmr and @xmath0sr study on nacro@xmath1 ,
an unusual regime of excitations was revealed , likely related to the triangular nature of the magnetic structure @xcite .
note that no shift was found by @xmath29na measurements on nacro@xmath1 in the high-@xmath30 regime , which was attributed to the absence of hyperfine coupling between na and cr @xcite .
as we will detail further in this paper , the presence of shift on @xmath28li nmr spectra reveals a more substantial 3d coupling between layers in licro@xmath1 . in this paper
we present a similar @xmath0sr study of licro@xmath1 . due to the 10 ns-15@xmath0s window that allows detection of fast - relaxing sites
, @xmath0sr enables a more complete study of licro@xmath1 over the whole temperature range , than in nmr .
we compare the temperature dependence of the relaxation rate for both compounds and discuss the relation with the strength of the coupling between planes .
our polycrystalline samples were synthesized by the stoichiometric mixture of li@xmath1co@xmath31 and cr@xmath1o@xmath31 , heated at 800@xmath7c for 48 h , with one intermediate grinding @xcite .
x - ray measurements showed them to be single phased with lattice parameters @xmath32 and @xmath33 , in good agreement with previous results @xcite .
@xmath0sr is a sensitive technique for the study of magnetic properties at a microscopic level .
it relies on the implantation of positive muons , that have a spin @xmath34 and a very large gyromagnetic ratio , @xmath35 mhz / tesla .
the muon spin is initially 100 % polarized opposite to the direction of the beam .
it is commonly assumed that in oxides @xmath36 stop at @xmath37 from the o@xmath38 sites @xcite and the muon spin feels the dipolar magnetic field created by the local environment of the sample .
the measured quantity is the asymmetry of the muon as a function of time , that provides information about the static and dynamical magnetic behavior of the system under study .
complementary measurements presented in this paper were performed either at psi ( switzerland ) or at isis ( united kingdom ) facilities .
the continuous character of the beam at psi enables one to get a high time resolution as needed at low @xmath30 whereas the pulsed beam of isis yields the possibility to follow in detail slow relaxing signals as needed for the dynamical study . .
the spontaneous oscillation is a signature of an internal field indicating ordered freezing at this temperature .
the line is a fit to the data .
inset : temperature dependance of the magnetic internal field , as extracted from the frequency of the oscillations .
lines are guides for the eye . ]
the typical @xmath0sr asymmetry in zero external field ( zf ) , well below @xmath5 , such as detected at 5 k with high statistics , is presented in fig .
[ zf_5kzoom ] .
the spontaneous oscillations visible up to @xmath39s indicate the presence of well - defined internal magnetic fields in the sample , showing that at this temperature the system is frozen and ordered .
this finding perfectly agrees with neutron evidence @xcite for a magnetic long - range order which can be described with 18 sublattices of spins at 120@xmath7 . in line with this
quite involved structure , one can expect many magnetically inequivalent muon sites .
for each site , the muon will experience a well - defined internal field and will contribute to the asymmetry with a given frequency .
after inspection of the signal of fig .
[ zf_5kzoom ] , we were led to a minimal fitting model corresponding to a double - peaked histogram of the internal fields .
this leads to a two - frequencies fit with a damping for each frequency corresponding to the expected but unresolved multiplicity of magnetically inequivalent @xmath40 sites .
the asymmetry was therefore fitted according to the formula @xcite : @xmath41 here , @xmath42 represent the weight of the two frequency components , found to be around 0.5 by the fit at 5 k and then fixed at this value at all temperatures .
@xmath43 stands for the initial asymmetry at @xmath44 when muons are fully polarized ( typically @xmath45 at psi and @xmath46 at isis ) .
the fit obtained at 5 k is represented as a continuous line in fig .
[ zf_5kzoom ] .
we obtained 43 and 26 mhz for the two frequencies of oscillations , which correspond to 317 and 192 mt for the internal fields at the two muon sites .
their evolution with temperature is presented in the inset of fig .
[ zf_5kzoom ] .
the field values decrease from a saturation value at low temperature down to zero around 60 k , where the oscillation pattern becomes difficult to follow .
the fast damping of the oscillations , taken into account by the gaussian exponential in the fitting function , indicates a distribution of sites of width @xmath47 mt , in agreement with a large number of inequivalent magnetic sites .
we now turn to the study of dynamical effects , i.e. , of electronic fluctuations , that can be tracked by the study of the asymmetry curves at long times .
we used a combination of data from psi and isis facilities .
the advantage of the former is the absence of background due to muons not stopping into the sample , whereas the latter can be taken at long times with a good statistics which proved to be useful around the transition . yet the uncertainty of the background values of the order of 0.005 is responsible for error bars in the relaxation rate .
spectra obtained up to @xmath48s at psi in zero external field are presented in fig .
in the case of a complete freezing , @xmath49 , and one obtains the so - called `` one - third '' tail at long times ( @xmath50 ) .
when dynamical effects are present , fluctuations induce a relaxation of this tail . in the fitting formula , a stretched exponential with @xmath51
was used empirically to account for the distribution of magnetically inequivalent muon sites as mentioned above , which yield a distribution of relaxation rates . at 5 k ,
the relaxation due to electronic fluctuations is small , @xmath52s@xmath53 , which shows that the behavior is mainly static . as the temperature increases
, the relaxation becomes more important .
the one - third tail relaxation increases progressively up to @xmath54 k. at 56 k , the tail has the fastest decrease toward zero and , therefore , corresponds to the maximum of relaxation . in the opposite limit of the fast fluctuating regime , above 65 k
, the early time depolarization of the muon spin is mainly due to static nuclear dipoles , which create a small random field of @xmath550.4 mt , independent of temperature .
the induced asymmetry has the well - established kubo - toyabe ( kt ) shape , which is gaussian - like on the time scale of fig .
in order to keep track of the slight relaxation of electronic origin , the kt function was multiplied by an exponential @xmath56 with a weak value of @xmath57s@xmath53 . in the intermediate regime between 56 and 65 k
, the data could not be fitted using the previous approach and we were led to conclude that two types of clusters , either static or dynamical , coexist . in order to simplify the fitting procedure , we performed experiments under a weak magnetic field of 3 mt , applied along the initial direction of the muon spin .
thus , the muon is no longer depolarized by the static field of the nuclear dipoles , which is much smaller than the applied field . on the other hand , this value is small enough so that it does not impact the electronic fluctuations .
asymmetry obtained at the isis facility is represented in fig .
we fitted the asymmetry for @xmath58s according to the ansatz @xmath59 here the term @xmath15 stands for the background due to muons that do not stop into the sample ( @xmath60 ) .
the term @xmath61 was introduced empirically in order to track the paramagnetic islands that survive below @xmath5 , which adds to the background @xmath15 .
for simplicity , the relaxation corresponding to these sites was crudely neglected . the relaxation rate estimated for different values of
@xmath61 varying from 0 to 0.035 were used in order to calculate the error bars .
the obtained fits are represented as continuous lines in fig [ lf ] .
the temperature dependence of the relaxation rate is presented in fig .
[ relaxationrate ] , and values obtained below 56 k are in good agreement with those obtained using the zero field spectra .
the temperature dependence of the relaxation rate is represented in fig .
[ relaxationrate ] and compared to the na analog .
the temperature scale is normalized by @xmath5 ( 62 k for licro@xmath1 , and 41 k for nacro@xmath1 ) . according to this figure , one can distinguish two features : \(i ) in licro@xmath62the relaxation rate peak appears at @xmath63 , while in nacro@xmath1 it appears at @xmath64 .
\(ii ) in licro@xmath1 , the relaxation rate peak is sharper than for nacro@xmath1 . for nacro@xmath1 and licro@xmath1 . lines are guides for the eye . ]
note that in compounds that present a regular transition , the relaxation - rate peaks where the specific heat does , and the width of the relaxation - rate peak is much smaller than in the case of nacro@xmath1 , typically 20% of @xmath5 @xcite . in nacro@xmath1 ,
the relaxation rate cross - over is very broad and appears well below @xmath5 , which indicates the presence of an _ extended fluctuating regime_. accordingly , neutron - scattering results @xcite show no sign of 3d ordering above @xmath64 but only the presence of 2d correlations . below this temperature ,
an incommensurate 3d magnetic order develops but remains short ranged down to lowest temperature . in licro@xmath1
, this unusual fluctuating regime appears to be very reduced in comparison with the na case .
the neutron - scattering study revealed the growing of 3d long - range magnetic order very close to the transition temperature @xcite .
replacing na by li reduces the separation between successive cr planes , which is likely to create a more important coupling in licro@xmath1 .
note that in both na and li compounds the dominant mechanism for in - plane coupling is made by _
direct _ exchange , as shown by band - structure calculations @xcite , as well as by calculations that take into account the value of the cr - o - cr angle @xcite .
this angle varies only little , from 96.4@xmath7 in nacro@xmath1 to 98.8@xmath7 in licro@xmath1 .
in fact only minor changes occur in the cro@xmath1 planes which can not yield the observed differences between the two compounds .
the scenario of a 3d coupling in licro@xmath1 is further supported by @xmath28li nmr measurements @xcite , which revealed the presence of a line shift , implying the existence of hyperfine coupling between li and cr , probably mediated by the oxygen .
the coupling between adjacent cr@xmath2 planes can therefore proceed indirectly by the paths cr - o - li - o - cr or alternatively cr - o - o - cr .
the coupling through direct overlapping of cr@xmath2 orbitals is probably negligible , in view of the large separation between the planes , of about 4.8 .
in addition to the exchange interaction , the cr@xmath2 planes are also coupled by a small dipolar interaction of less than 1 k. on the contrary , in the na compound , the @xmath29na nmr study revealed the absence of line shift , and by comparison with results on the isostructural na@xmath65coo@xmath1 , it was inferred that there is no overlapping between the na and o orbitals @xcite .
the main mechanisms for coupling , via the paths cr - o - na - o - cr or cr - o - o - cr , are therefore not effective .
the inter - layer coupling appears thus to be mainly dipolar in nature , so extremely small .
the comparative study of li and na chromates allows , thus , to track the influence of the 3d coupling on the magnetic properties and isolate those intrinsic to the triangular lattice . in nacro@xmath1
, it was suggested that the extended fluctuating regime is a signature of topological defects , z@xmath1 vortices , whose existence was proposed long ago by kawamura and miyashita @xcite .
the dissociation of vortex pairs would induce a transition at a temperature lower than that of the specific - heat - capacity peak .
the present comparison underlines the sensitivity of such a mechanism to 3d couplings , and the sensitivity to similar deviations from the ideal 2d heisenberg case might explain why the experimental signature of these defects has not been observed in real systems for a long time .
altogether with the recent discovery of niga@xmath1s@xmath8 , this opens a new route to the exploration of the physics of edge - sharing triangular heisenberg lattices .
indeed , in niga@xmath1s@xmath8 , where ni@xmath66 ions with @xmath67 form a triangular lattice , persisting dynamics was observed down to very low temperature , well below the specific - heat peak @xcite and the magnetic order revealed by neutron scattering measurements remains short ranged down to 0.35 k. along these ideas , it would be very interesting to study kcro@xmath1 , if this compound can be stabilized , for which one expects the most pronounced 2d character in the acro@xmath1 family , since it has the largest separation between cr planes .
an early neutron - scattering work interestingly revealed the absence of any 3d magnetic ordering down to 2 k @xcite . +
our @xmath0sr study shows that the triangular heisenberg compound licro@xmath1 develops a more conventional ordering than nacro@xmath1 , although the fluctuating regime remains sizeable .
we propose an interpretation in terms of a more pronounced 3d coupling between cro@xmath1 triangular planes related to the smaller distance between them .
this point is in line with neutron experiments and recent nmr results which overall reveals the importance of the 2d character in the study of topological excitations typical of triangular systems .
this work is supported by anr under grant no .
nt05 - 4 - 41913 `` oxyfonda '' .
we gratefully acknowledge support from the european commission under framework programme 6 through the key action : strengthening the european research area , research infrastructures under contract no .
rii3-ct-2003 - 505925 . | we report a @xmath0sr study of licro@xmath1 , which has a magnetic lattice made up of a stacking of triangular heisenberg antiferromagnetic ( cr@xmath2 , @xmath3 ) layers .
a static magnetically ordered state is observed below the transition temperature @xmath4 k , while the expected peak of the relaxation rate is slightly shifted downward by a few kelvins below @xmath5 .
we draw a comparison with the isostructural compound nacro@xmath1 , where an exotic broad fluctuating regime has been observed [ a. olariu , p. mendels , f. bert , b. g. ueland , p. schiffer , r. f. berger , and r. j. cava , phys .
rev .
lett . * 97 * , 167203 ( 2006 ) ] and was suggested to originate from topological excitations of the triangular lattice . replacing na by li strongly narrows the exotic fluctuating regime formerly observed in nacro@xmath1 , which we attribute to a more pronounced inter - plane coupling in licro@xmath1 . |
the discovery of electron spin is one of the most interesting stories in the history of quantum mechanics ; told e.g. in van der waerden s contribution to the pauli memorial volume ( @xcite , pp.199 - 244 ) , in tomonaga s book @xcite , and also in various first - hand reports @xcite@xcite @xcite .
this story also bears fascinating relations to the history of understanding special relativity .
one such relation is given by thomas discovery of what we now call `` thomas precession '' @xcite@xcite , which explained for the first time the correct magnitude of spin - orbit coupling and hence the correct magnitude of the fine - structure split of spectral lines , and whose mathematical origin can be traced to precisely that point which marks the central difference between the galilei and the lorentz group ( this is e.g. explained in detail in sects.4.3 - 4.6 of @xcite ) . in the present paper
i will dwell a little on another such connection to special relativity . as is widely appreciated , wolfgang pauli is a central figure , perhaps _ the _ most central figure , in the story of spin .
being the inventor of the idea of an inner ( quantum mechanical ) degree of freedom of the electron , he was at the same time the strongest opponent to attempts to relate it to any kind of interpretation in terms of kinematical concepts that derive from the picture of an extended material object in a state of rotation . to my knowledge ,
pauli s hypothesis of this new intrinsic feature of the electron , which he cautiously called `` a classical non - describable two valuedness '' , was the first instance where a quantum - mechanical degree of freedom was claimed to exist without a corresponding classical one .
this seems to be an early attempt to walk without the crutches of some ` correspondence principle ' .
even though the ensuing developments seem to have re - installed mentally at least the more classical notion of a spinning electron through the ideas of ralph kronig ( compare section 4 of vanderwaerden s contribution to @xcite , pp.209 - 216 ) and , independently , samuel goudsmit and george uhlenbeck @xcite@xcite , pauli was never convinced , despite the fact that he lost the battle against thomas and declared `` total surrender '' in a letter to bohr written on march 12 .
1926 ( @xcite , vol.i , doc.127 , pp.310 ) . for pauli
the spin of the electron remained an abstract property which receives its ultimate and irreducible explanation in terms of group theory , as applied to the subgroup of spatial rotations ( or its double cover ) within the full symmetry group of space - time , may it be the galilei or the lorentz group ( or their double cover ) . or as faithful true representations ( i.e. ` single - valued ' ) of its double - cover group @xmath0 , which are subgroups of the galilei and lorentz groups or their double - cover groups respectively .
] in this respect , pauli s 1946 nobel lecture contains the following instructive passage ( here and throughout this paper i enclose my annotations to quotes within square brackets ) : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ although at first i strongly doubted the correctness of this idea [ of the electron spin in the sense of kronig , goudsmit and uhlenbeck ] because of its classical - mechanical character , i was finally converted to it by thomas calculations on the magnitude of doublet splitting . on the other hand , my earlier doubts as well as the cautions expression @xmath1classically non - describable two - valuedness@xmath2 experienced a certain verification during later developments , since bohr was able to show on the basis of wave mechanics that the electron spin can not be measured by classically describable experiments ( as , for instance , deflection of molecular beams in external electromagnetic fields ) and must therefore be considered as an essential quantum - mechanical property of the electron .
( @xcite , p.30 ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ r0.6 this should clearly not be misunderstood as saying that under the impression of thomas calculations pauli accepted spin in its ` classical - mechanical ' interpretation .
in fact , he kept on arguing fiercely against what in a letter to sommerfeld from december 1924 he called `` model prejudices '' ( @xcite , vol.i , doc.72 , p.182 ) and did not refrain from ridiculing the upcoming idea of spin from the very first moment ( cf .
what pauli accepted was the idea of the electron possessing an intrinsic magnetic moment and angular momentum , the latter being interpreted _ exclusively _ in a formal fashion through its connection with the generators of the subgroup of rotations within the lorentz group , much like we nowadays view it in modern relativistic field theory .
to some extent it seems fair to say that , in this case , pauli was a pioneer of the modern view according to which abstract concepts based on symmetry - principles are seen as primary , whereas their concrete interpretation in terms of localised material structures , to which e.g. the kinematical concept of ` rotation ' in the proper sense applies , is secondary and sometimes even dispensable .
but one should not forget that this process of emancipation was already going on in connection with the notion of classical fields , as einstein used to emphasise , e.g. , in his 1920 leiden address `` ether and the theory of relativity '' ( @xcite , vol.7 , doc.38 , pp.308 - 320 ) .
we will come back to this point below .
besides being sceptical in general , pauli once also made a _
specific _ remark as to the inadequateness of classical electron models ; that was three years after thomas note , in a footnote in the addendum to his survey article `` general foundations of the quantum theory of atomic structure '' , that appeared 1929 as chapter 29 in ` mller - pouillets lehrbuch der physik ' .
there he said : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ emphasising the kinematical aspects one also speaks of the ` rotating electron ' ( english ` spin - electron ' ) .
however , we do not regard the conception of a rotating material structure to be essential , and it does not even recommend itself for reasons of superluminal velocities one then has to accept .
( @xcite , vol.1 , pp.721 - 722 , footnote2 ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ interestingly , this is precisely the objection that , according to goudsmit s recollections @xcite , lorentz put forward when presented with goudsmit s and uhlenbeck s idea by uhlenbeck , and which impressed uhlenbeck so much that he asked ehrenfest for help in withdrawing the already submitted paper @xcite .
he did not succeed , but the printed version contains at least a footnote pointing out that difficulty : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the electron must now assume the property ( a ) [ a @xmath3-factor of 2 ] , which land attributed to the atom s core , and which is hitherto not understood .
the quantitative details may well depend on the choice of model for the electron .
[ ... ] note that upon quantisation of that rotational motion [ of the spherical hollow electron ] , the equatorial velocity will greatly exceed the velocity of light .
( @xcite , p.954 ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ this clearly says that a classical electron model can not reproduce the observable quantities , mass , charge , angular momentum , and magnetic moment , without running into severe contradictions with special relativity . and the magnetic moment to be one magneton ( @xmath4 ) '' .
] the electron model they had in mind was that developed by abraham in his 1903 classic paper on the `` principles of electron dynamics '' @xcite ( cited in footnote2 on p.954 of @xcite ) . since then it has become standard textbook wisdom that classical electron models necessarily suffer from such defects ( compare , e.g. , @xcite , p.155 ) and that , even in quantum mechanics , `` the idea of the rotating electron is not be taken literally '' , as max born once put it ( @xcite , p.188 ) .
modern references iterate this almost verbatim : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the term ` electron spin ' is not to be taken literally in the classical sense as a description of the origin of the magnetic moment described above . to be sure , a spinning sphere of charge can produce a magnetic moment , but the magnitude of the magnetic moment obtained above can not be reasonably modelled by considering the electron as a spinning sphere .
+ ( taken from @xmath5 http://hyperphysics.phy-astr.gsu.edu/hbase/spin.html@xmath6 ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in this contribution i wish to scrutinise the last statement .
this is not done in an attempt to regain respect for classical electron models for modern physics , but rather to illuminate in some detail a specific and interesting case of the ( well know ) general fact that progress is often driven by a strange mixture of good and bad arguments , which hardly anybody cares to separate once progress is seen to advance in the ` right direction ' .
also , the issues connected with an inner rotational motion of the electron are hardly mentioned in the otherwise very detailed discussion of classical electron theories in the history - of - physics literature ( compare @xcite@xcite ) .
last but not least , the present investigation once more emphasises the importance of special - relativistic effects due to stresses , which are not necessarily connected with large velocities , at least in a phenomenological description of matter .
but before giving a self - contained account , i wish to recall pauli s classic paper of december 1924 , where he introduced his famous `` classically non - describable two - valuedness '' .
we begin by recalling the notion of _ gyromagnetic ratio_. consider a ( not necessarily continuous ) distribution of mass and charge in the context of pre - special - relativistic physics , like , e.g. , a charged fluid or a finite number of point particles .
let @xmath7 denote the corresponding velocity field with respect to an inertial frame and @xmath8 and @xmath9 the density distributions of electric charge and mass corresponding to the total charge @xmath10 and mass @xmath11 respectively .
the total angular momentum is given by ( @xmath12 denotes the antisymmetric vector product ) @xmath13 the electric current distribution , @xmath14 , is the source of a magnetic field which at large distances can be approximated by a sum of multipole components of increasingly rapid fall - off for large distances from the source .
the lowest possible such component is the dipole .
it has the slowest fall - off ( namely @xmath15 ) and is therefore the dominant one at large distances .
( a monopole contribution is absent due to the lack of magnetic charges . )
the dipole field is given by and @xmath16 will appear explicitly . note that @xmath17 and that @xmath18 _ exactly _ , where c stands for ` coulomb ' , the unit of charge . ]
@xmath19 where @xmath20 , @xmath21 and where @xmath22 denotes the magnetic dipole moment of the current distribution , which is often ( we shall follow this ) just called _ the _ magnetic moment : @xmath23 note the similarity in structure to ( [ eq : introangmom ] ) , except for the additional factor of 1/2 in front of ( [ eq : defmagnmoment ] ) .
the _ gyromagnetic ratio _ of a stationary mass and charge current - distribution , @xmath24 , is defined to be the ratio of the moduli of @xmath22 and @xmath25 : @xmath26 we further define a dimensionless quantity @xmath3 , called the _ gyromagnetic factor _ , by @xmath27 these notions continue to make sense in non - stationary situations if @xmath22 and @xmath25 are slowly changing ( compared to other timescales set by the given problem ) , or in ( quasi ) periodic situations if @xmath22 and @xmath25 are replaced by their time averages , or in mixtures of those cases where , e.g. , @xmath25 is slowly changing and @xmath22 rapidly precesses around @xmath25 ( as in the case discussed below ) .
an important special case is given if charge and mass distributions are strictly proportional to each other , i.e. , @xmath28 , where @xmath29 is independent of @xmath30 .
then we have @xmath31 in particular , this would be the case if charge and mass carriers were point particles of the same charge - to - mass ratio , like @xmath32 particles of one sort , where @xmath33 after these preliminaries we now turn to pauli s paper . on december 2nd 1924
, pauli submitted a paper entitled `` on the influence of the velocity dependence of the electron mass upon the zeeman effect '' ( @xcite , vol.2 , pp.201 - 213 ) to the zeitschrift fr physik . in that paper
he starts with the general observation that for a point particle of rest - mass @xmath11 and charge @xmath10 , moving in a bound state within a spherically symmetric potential , the velocity dependence of mass , @xmath34 affects the gyromagnetic ratio .
here @xmath35 , where @xmath36 .
the application he aims for is the anomalous zeeman effect for weak magnetic fields , a topic on which he had already written an earlier paper , entitled on the rules of the anomalous zeeman effect ( @xcite , vol.2 , pp.151 - 160 ) , in which he pointed out certain connections between the weak - field case and the theoretically simpler case of a strong magnetic field .
note that `` weak '' and `` strong '' here refers to the standard set by the inner magnetic field caused by the electrons orbital motion , so that `` weak '' here means that the zeeman split is small compared to the fine - structure .
since the charge is performing a quasi periodic motion , its magnetic moment due to its orbital motion is given by the time average ( i will denote the time average of a quantity @xmath37 by @xmath38 ) @xmath39 on the other hand , its angular momentum is given by @xmath40 it is constant if no external field is applied and slowly precessing around the magnetic field direction if such a field is sufficiently weak , thereby keeping a constant modulus .
hence we can write @xmath41 where the averaging period is taken to be long compared to the orbital period of the charge , but short compared to the precession period of @xmath25 if an external magnetic field is applied .
this gives @xmath42 where with the lorentz factor @xmath43 , which nowadays is usually abbreviated by @xmath44 , though not in the present paper . ]
@xmath45 more specifically , pauli applies this to the case on an electron in the coulomb field of a nucleus .
hence @xmath11 from now on denotes the electron mass .
its charge is @xmath46 , and the charge of the nucleus is @xmath47 . using the virial theorem
, he then gives a very simple derivation of @xmath48 where @xmath49 denotes the electron s total energy ( kinetic plus potential ) .
for the quantised one - electron problem , an explicit expression for @xmath49 in terns of the azimuthal quantum number @xmath50 ( @xmath51 in modern notation , where @xmath52 is the quantum number of orbital angular - momentum ) and the principal quantum number @xmath53 ( @xmath54 , where @xmath55 is the radial quantum number ) was known since sommerfeld s 1916 explanation of fine structure ( see , e.g. , @xcite , p.53 , formula ( 17 ) ) .
hence pauli could further write : @xmath56 where the approximation refers to small values of @xmath57 and where @xmath58 is the fine - structure constant .
for higher @xmath59 one obtains significant deviations from the classical value @xmath60 . for example
, @xmath61 gives @xmath62 .
the relativistic correction factor @xmath44 affects the angular frequency .
this is also related to our usage of @xmath63 rather than @xmath64 ( planck s constant ) . ]
with which the magnetic moment created by the electron s orbital motion will precess in a magnetic field of strength @xmath65 .
this angular frequency is now given by @xmath66 , where @xmath67 is the larmor ( angular ) frequency : @xmath68 here we explicitly wrote down the gyromagnetic ratio , @xmath69 , for of the electron s orbital motion even though @xmath70 , just to keep track of its appearance .
the energy for the interaction of the electron with the magnetic field now likewise receives a factor of @xmath44 .
pauli now applies all this to the `` core model '' for atoms with a single valence electron . according to the simplest version of this model , the total angular momentum ,
@xmath25 , and the total magnetic moment , @xmath22 , are the vector sums of the angular and magnetic momenta of the core ( indicated here by a subscript @xmath71 ) and the valence electron ( indicated here by a subscript @xmath72 ) : [ eq : totalmomenta ] @xmath73 the relations between the core s and electron s magnetic momenta on one side , and their angular momenta on the other , are of the form [ eq : relationsmomenta ] @xmath74 the point is now that @xmath22 is not a multiple of @xmath25 if @xmath75 . assuming a constant @xmath25 for the time being , this means that @xmath22 will precess around @xmath25 .
hence @xmath22 is the sum of a time independent part , @xmath76 , parallel to @xmath25 and a rotating part , @xmath77 , perpendicular to @xmath25 .
the time average of @xmath77 vanishes so that the effective magnetic moment is just given by @xmath76 . using ( [ eq : totalmomenta ] ) and ( [ eq : relationsmomenta ] ) , and resolving scalar products into sums and differences of squares , .
] we get @xmath78 setting again @xmath70 , the expression in curly brackets gives the gyromagnetic factor of the total system with respect to the effective magnetic moment .
its quantum analog is obtained by replacing @xmath79 and correspondingly for @xmath80 and @xmath81 , which is then called the _ land factor _ @xmath82 . hence @xmath83
all this is still right to a good approximation if @xmath25 is not constant , but if its frequency of precession around the direction of the ( homogeneous ) external field is much smaller than the precession frequency of @xmath22 around @xmath25 , which is the case for sufficiently small external field strength .
basically through the work of land it was known that @xmath84 fitted the observed multiplets of alkalies and also earth alkalies quite well .
this value clearly had to be considered anomalous , since the magnetic moment and angular momentum of the core were due to the orbital motions of the electrons inside the core , which inevitably would lead to @xmath85 , as explained in section[sec : preliminaries ] .
this was a great difficulty for the core model at the time , which was generally referred to as the `` magneto - mechanical anomaly '' .
pauli pointed out that one could either say that the physical value of the core s gyromagnetic factor is twice the normal value , or , alternatively , that it is obtained by adding 1 to the normal value .
these two ways of looking at the anomaly suggested two different ways to account for the relativistic correction , which should only affect that part of the magnetic moment that is due to the orbital motion of the inner electrons , that is , the ` normal ' part of @xmath86 .
hence pauli considered the following two possibilities for a relativistic correction of @xmath86 , corresponding to the two views just outlined : @xmath87 then comes his final observation , that neither of these corrections are compatible with experimental results on high-@xmath59 elements by runge , paschen and back , which , like the low-@xmath59 experiments , resulted in compatibility with ( [ eq : landefactor ] ) only if @xmath84 . in a footnote
pauli thanked land and back for reassuring him that the accuracy of these measurements where about one percent .
pauli summarises his findings as follows _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ if one wishes to keep the hypothesis that the magneto - mechanical anomaly is also based in closed electron groups and , in particular , the @xmath88 shell , then it is not sufficient to assume a doubling of the ratio of the group s magnetic moment to its angular momentum relative to its classical value .
in addition , one also needs to assume a compensation of the relativistic correction .
( @xcite , vol.2 , p.211 ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ after some further discussion , in which he stresses once more the strangeness that lies in @xmath84 , he launches the following hypothesis , which forms the main result of his paper : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the closed electron configurations shall not contribute to the magnetic moment and angular momentum of the atom . in particular , for the alkalies , the angular momenta of , and energy changes suffered by , the atom in an external magnetic field shall be viewed exclusively as an effect of the light - electron , which is also regarded as the location [ `` der sitz '' ] of the magneto - mechanical anomaly .
the doublet structure of the alkali spectra , as well as the violation of the larmor theorem , is , according to this viewpoint , a result of a classically indescribable two - valuedness of the quantum - theoretic properties of the light - electron .
( @xcite , vol.2 , p.212 ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ note that this hypothesis replaces the atom s core as carrier of angular momentum by the valence electron .
this means that ( [ eq : totalmomenta ] ) , ( [ eq : relationsmomenta ] ) , and ( [ eq : landefactor ] ) are still valid , except that the subscript @xmath71 ( for `` core '' ) is now replaced by the subscript @xmath89 ( for `` spin '' , anticipating its later interpretation ) , so that we now have a coupling of the electron s orbital angular momentum ( subscript @xmath72 ) to its intrinsic angular momentum ( subscript @xmath89 ) . in ( [ eq : landefactor ] ) , with @xmath86 replaced by @xmath90 , one needs to set @xmath91 in order to fit the data .
but now , as long as no attempt is made to relate the intrinsic angular momentum and magnetic moment of the electron to a common origin , there is no immediate urge left to regard this value as anomalous .
also , the problem in connection with the relativistic corrections ( [ eq : correctionsforg_c ] ) now simply disappeared , since it was based on the assumption that @xmath92 and @xmath93 were due to orbital motions of inner ( and hence fast ) electrons , whereas in the new interpretation only @xmath94 and @xmath95 are due to orbital motion of the outer ( and hence slow ) valence electron .
it is understandable that this hypothesis was nevertheless felt by some to lack precisely that kind of ` explanation ' that pauli deliberately stayed away from : a common dynamical origin of the electron s inner angular momentum and magnetic moment . from here
the ` story of spin ' takes its course , leading to the hypothesis of the rotating electron , first conceived by kronig and a little later , and apparently independently , by goudsmit and uhlenbeck , and finally to its implementation into quantum mechanics by pauli @xcite ( `` pauli equation '' for the non - relativistic case ) and dirac @xcite ( fully lorentz invariant `` dirac equation '' ) . since then many myths surrounding spin built up , like that the concept of spin , and in particular the value @xmath4 , was irreconcilable with classical ( i.e. non - quantum ) physics and that only the dirac equation naturally predicted @xmath4 . as for the latter statement
, it is well known that the principle of minimal coupling applied to the pauli equation leads just as natural to @xmath4 as in case of the dirac equation ( cf .
@xcite and @xcite , p.37 ) . also , the very concept of spin has as natural a home in classical physics as in quantum physics if one starts from equally general and corresponding group - theoretic considerations.the spaces of states in quantum and classical mechanics are hilbert spaces and symplectic manifolds respectively .
an _ elementary system _ is characterised in quantum mechanics by the requirement that the group of space - time symmetries act unitarily and irreducibly on its space of states .
the corresponding requirement in classical mechanics is that the group action be symplectic and transitive @xcite .
the classification of homogeneous ( with respect to the space - time symmetry group , be it the galilei or lorentz group ) symplectic manifolds @xcite@xcite leads then as natural to a classical concept of spin as the classification of unitary irreducible ( ray- ) representations leads to the quantum - mechanical spin concept .
the mentioned classical structures are related to the quantum structures by various concepts of ` quantisation ' like ` geometric quantisation ' .
compare @xcite , in particular chap.6 on elementary systems . ] for the rest of this contribution i wish to concentrate on the particular side aspect already outlined in the introduction .
let me repeat the question : in what sense do the actual values of the electron parameters , mass , charge , intrinsic angular - momentum , and gyromagnetic factor , resist classical modelling in the framework of special relativity ?
in this section we will give a self - contained summary of the basic features of simple electron models .
the first model corresponds to that developed by abraham @xcite , which was mentioned by goudsmit and uhlenbeck as already explained.since we are mainly concerned with the spin aspects , we will ignore the differences between abraham s and , say , lorentz model ( rigid versus deformable ) , which become important as soon as translational motions are considered .
we mention abraham not for any preference for his ` rigid ' model , but for the reason that he considered rotational motion explicitly .
its interaction with the translational motion was further worked out in detail by schwarzschild in @xcite , but this is not important here . ]
we will see that this model can only account for @xmath3 factors in the interval between @xmath96 and @xmath97 if superluminal speeds along the equator are to be avoided .
we also critically discuss the assumption made by goudsmit and uhlenbeck that this ( i.e. abraham s ) model predicts @xmath4 .
since this model neglects the stresses that are necessary to prevent the charge distribution from exploding , we also discuss a second model in which such stresses ( corresponding to a negative pressure in the electron s interior ) are taken into account , at least in some slow - rotation approximation .
this model , too , has been discussed in the literature before @xcite .
here it is interesting to see that due to those stresses significantly higher values of @xmath3 are possible , though not for small charges as we will also show .
finally we discuss the restriction imposed by the condition of energy dominance , which basically says that the speed of sound of the stress - supporting material should not exceed the speed of light .
this sets an upper bound on @xmath3 given by @xmath98 .
note that all these statements are made only in the realm where the slow - rotation approximation is valid .
i do not know of any fully special - relativistic treatment on which generalisations of these statements could be based . in that sense ,
the general answer to our main question posed above is still lacking .
consider a homogeneous charge distribution , @xmath99 , of total charge @xmath100 on a sphere of radius @xmath101 centred at the origin ( again we write @xmath20 and @xmath21 ) : @xmath102 for the moment we shall neglect the rest mass of the matter that sits at @xmath103 and also the stresses it must support in order to keep the charge distribution in place .
the charge is the source of the scalar potential @xmath104 with corresponding electric field @xmath105 let now the charge distribution rotate rigidly with constant angular velocity @xmath106 .
this gives rise to a current density @xmath107 which , in turn , is the source of a vector potential according to @xmath108 hence , in the rotating case , there is an additional magnetic field in addition to the electric field ( [ eq : electricfield ] ) : @xmath109 where @xmath110 for @xmath111 this is a constant field in @xmath106 direction .
for @xmath112 it is a pure dipole field ( i.e. all higher multipole components vanish ) with moment ( [ eq : magndipolemoment ] ) .
the general expression for the energy of the electromagnetic field is etc . ]
@xmath113 for the case at hand , the electric and magnetic contributions to the energy are respectively given by [ eq : electric - magneticfieldenergy ] @xmath114 the total magnetic contribution can be written as @xmath115 where @xmath116 may be called the _ electromagnetic moment of inertia _
it has no mechanical interpretation in terms of a rigid rotation of the electrostatic energy distribution ( see below )
! the total electromagnetic energy can now be written as @xmath117 where we used @xmath17 and set @xmath118 . the ratio of magnetic ( ` kinetic ' ) to total energy is then given by @xmath119 which is a strictly monotonic function of @xmath120 bounded above by 1 ( as it should be ) . however ,
if we require @xmath121 , the upper bound is @xmath122 .
the momentum density of the electromagnetic field vanishes for @xmath111 and is given by @xmath123 for @xmath112 ( @xmath124 times ` poynting vector ' ) .
the angular - momentum density also vanishes for @xmath111 .
for @xmath112 it is given by @xmath125 hence the total linear momentum vanishes , whereas the total angular momentum is given by @xmath126 with the _ same _ @xmath127 ( moment of inertia ) as in ( [ eq : emmomentinertia ] ) .
the gyromagnetic ratio now follows from expressions ( [ eq : magndipolemoment ] ) for @xmath22 and ( [ eq : totalangularmomentummod1 ] ) for @xmath25 : @xmath128 where @xmath129 denotes the total mass , which is here given by @xmath130 hence @xmath3 can be solved for : @xmath131 so that @xmath132 even with that simple model we do get quite close to @xmath4 .
it is sometimes stated that abraham s model somehow ` predicts ' @xmath4 ( e.g. , @xcite p.39 or @xcite p.206 ) , though this is not at all obvious from @xcite .
my interpretation for how such a ` prediction ' could come about can be given in terms of the present special - relativistic model .. ] it rests on an ( inconsistent ) combination of the following two observations . first , if we lorentz transform the purely electric field ( [ eq : electricfield ] ) into constant translational motion with velocity @xmath133 , we obtain a new electric and also a non - vanishing magnetic field .
the integrated poynting vector then gives the total electromagnetic momentum of the charged shell at speed @xmath133 : @xmath134 where @xmath135 the infamous factor 4/3 results from the contribution of the ( unbalanced ) electromagnetic stresses . generally speaking , the factor 4/3 marks the discrepancy between two definitions of ` electromagnetic mass ' , one through the electromagnetic momentum , the other , called @xmath136 above , through the electrostatic energy .
this discrepancy is nothing to get terribly excited about and simply a consequence of the non - conservation of the electromagnetic energy - momentum tensor , i.e. , @xmath137 , a result of which is that the ( unbalanced ) electromagnetic stresses contribute to the electromagnetic momentum another third of the expression @xmath138 that one naively obtains from just formally transforming total energy and momentum as time and space components respectively of a four vector .
much discussion in the literature was provoked by getting confused whether this state of affairs had anything to do with lorentz non - covariance .
see , e.g. , @xcite for a good account and references . ] in this way one is led to assign to the electron a dynamically measurable rest - mass of @xmath139 _ if one neglects the rotational energy_. second , we may ask how fast the electron is to spin for ( [ eq : total mass ] ) to just give @xmath139 ( rest energy of the spinning electron ) .
the immediate answer is , that this is just the case if and only if @xmath140 , which in view of ( [ eq : em - gfactor ] ) is equivalent to @xmath4 .
it is now obvious how this argument rests on the conflation of two different notions of mass .
the factor @xmath141 will consistently be dealt with by taking into account the stresses that balance electrostatic repulsion , not by trying to account for it in letting the electron spin fast enough . in the introduction
we stressed that the emancipation of the notion of angular momentum from the usual kinematical notion of rotation in space had already begun in classical field theory .
more precisely this applies to maxwell s theory , in which the notion of a field differs from that of , say , hydrodynamics in that it is _ not _ thought of as being attached to a material carrier .
this has consequences if we wish to apply kinematical states of motion to the field itself . at first sight ,
faraday s picture of lines of force in space suggests to view them as material entities , capable of assuming different kinematical states of motion .
if so , the time - dependence of the electromagnetic field might then be interpreted as , and possibly explained by , the motions of such lines ( given by some yet unknown equations of motion , of which the maxwell equations might turn out to be some coarse grained version ) . that this is not possible
has been stressed by einstein in his 1920 leiden address `` ether and the theory of relativity '' , where he writes _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ if one wishes to represent these lines of force as something material in the usual sense , one is tempted to interpret dynamical processes [ of the em .
field ] as motions of these lines of force , so that each such line can be followed in time .
it is , however , well known that such an interpretation leads to contradictions . + in general we have to say that it is possible to envisage extended physical objects to which the notion of motion [ in space ] does not apply .
( @xcite , vol.7 , doc.38 , p.315 ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the reason why we mention this is that the notion of an `` electromagnetic moment of inertia '' , introduced in ( [ eq : emmomentinertia ] ) , nicely illustrates this point . assume that the electrostatic energy density @xmath142 of the coulomb field of charge @xmath100 corresponded to a mass density according to a local version of @xmath143 , i.e. , @xmath144 if the electrostatic energy is now thought of as being attached to the somehow individuated lines of force , a moment of inertia for the shell @xmath145 would result , given by @xmath146 but this diverges as @xmath147 , in contrast to ( [ eq : emmomentinertia ] ) , showing that we may not think of the energy distribution of the electromagnetic field as rigidly rotating in the ordinary sense . in this section we will modify the previous model for the electron in the following three aspects * the infinitesimally thin spherical shell is given a small rest - mass of constant surface density @xmath148 .
* stresses in the shell are taken into account which keep the electron from exploding .
they are called `` poincar stresses '' since poincar was the first in 1906 to discuss the dynamical need of balancing stresses @xcite@xcite . *
the rotational velocity is small , so that @xmath149 terms are neglected for @xmath150 .
the second modification needs further explanation .
if we view the surface @xmath103 as a kind of elastic membrane , there will be tangential stresses in the surface of that membrane that keep the charged membrane from exploding . in the present approximation , which keeps only linear terms in @xmath151 , these stresses need only balance the electrostatic repulsion , which is constant over the surface @xmath103 . in quadratic order
the stresses would , in addition , need to balance the latitude dependent centrifugal forces , which we neglect here . to calculate the surface stress that is needed to balance electrostatic repulsion we recall the expression ( [ eq : electric - magneticfieldenergy - a ] ) for the electrostatic energy as function of radius @xmath101 :
@xmath152 varying @xmath101 gives us the differential of work that we need to supply in order to change the volume through a variation of @xmath101 .
equating this to @xmath153 gives the pressure inside the electron : @xmath154 now , imagine the sphere @xmath103 being cut into two hemispheres along a great circle .
the pressure tries to separate these hemispheres by acting on each with a total force of strength @xmath155 in diametrically opposite directions .
, over a hemisphere .
] this force is distributed uniformly along the cut ( the great circle ) , whose length is @xmath156 . hence the force per length is just @xmath157 . the surface stress , @xmath158 , ( force per length ) that is needed to prevent the electron from exploding is just the negative of that . using ( [ eq : electronpressure ] ) , we therefore get @xmath159 the energy - momentum tensor now receives a contribution that accounts for the presence of the surface stress ( [ eq : surfacestress ] ) that acts tangential to the surface @xmath103 _ in the local rest frame corresponding to each surface element of the rotating sphere_. the four - velocity of each surface element is given by where the latter three are standard spherical polar coordinates .
i also employ the notation @xmath160 for the chart - induced vector fields , so that , e.g. , @xmath161 . ]
@xmath162 which is normalised ( @xmath163 ) up to terms @xmath164 ( which we neglect ) .
recall that the space - time metric of minkowski space in spatial polar coordinates is ( we use the `` mostly plus '' convention for the signature ) @xmath165 the energy - momentum tensor has now three contributions , corresponding to the matter of the shell ( subscript @xmath129 ) , the poincar stresses within the shell ( subscript @xmath158 ) , and the electromagnetic field ( subscript @xmath166 ) : [ eq : fullem - tensor ] @xmath167 the first two comprise the shell s contribution and are given by @xmath168 here @xmath169 is the orthogonal projector onto the 2-dimensional subspace orthogonal to @xmath170 and @xmath171 , which is the subspace tangential to the sphere in each of its local rest frames .
it can be written explicitly in terms of local orthonormal 2-legs , @xmath172 and @xmath173 , spanning these local 2-planes .
for example , we may take @xmath174 and write ( since @xmath173 must be orthogonal to @xmath171 and @xmath175 ) @xmath176 , where the coefficients @xmath177 follow from @xmath178 and normality .
this gives [ eq : projector ] @xmath179 where @xmath180 note that @xmath181 and @xmath182 .
equation ( [ eq : projector - a ] ) may therefore be be written in the form ( again neglecting @xmath164 terms ) @xmath183 for us the crucial term will be the last one , which is off - diagonal , since it will contribute to the total angular momentum .
more precisely , we will need to invoke the integral of @xmath184 ( the dot ( @xmath185 ) refers to the inner product with respect to the minkowski metric ) over the sphere @xmath103 : @xmath186 where we used @xmath187 and @xmath188 from ( [ eq : minkmetric ] ) .
the addition of the stress part has the effect that the total energy - momentum tensor is now conserved ( here in the slow - rotation approximation ) : @xmath189 as one may explicitly check .
note that since we use curvilinear coordinates here we need to invoke the covariant derivative .
, where @xmath190 , with @xmath191 taken from ( [ eq : minkmetric ] ) .
the @xmath192 s are most easily computed directly from the geodesic equation .
] indeed , writing the shell s energy momentum tensor as @xmath193 , it is not difficult to show that @xmath194 is zero for @xmath195 , and for @xmath196 is given by @xmath197 with @xmath198 as in ( [ eq : electronpressure ] ) . but
this clearly equals @xmath199 since , according to maxwell s equations , this quantity equals minus the electromagnetic force density on the charge distribution , which is obviously @xmath200 .
in fact , this is precisely the interpretation that we used to determine @xmath198 in the first place .
the conservation equation ( [ eq : em - conservation ] ) generally ensures that total energy and total momentum form , respectively , the time- and space component of a four vector .
let us now show explicitly that @xmath201 removes the factor @xmath141 in the calculation of the linear momentum when the system is boosted in , say , the @xmath202 direction . to do this we need to calculate the integral of @xmath203 over all of space and
show that it precisely cancels the corresponding integral of the electromagnetic part , i.e. the integral over @xmath204 . noting that @xmath205
, we have @xmath206 whereas the tracelessness of @xmath207 together with isotropy immediately imply @xmath208 that the sum of ( [ eq : verlauetheorem-1 ] ) and ( [ eq : verlauetheorem-2 ] ) vanishes is a consequence of laue s theorem , which basically states that the integral over all of space of the space - space components of a time - independent conserved energy - momentum tensor vanish . here
this was achieved by including stresses , which subtracted one third of the electromagnetic linear momentum . to be such that the total energy and momentum derived from @xmath209 should transform as a four vector clearly still leaves much freedom in the choice of @xmath210 .
the choice made here is such that the total rest energy equals the electrostatic self energy .
but other values for the rest energy ( like , e.g. , 4/3 of the electrostatic contribution ) would also have been possible .
in particular , the ` covariantisation through stresses ' does not as such prefer any of the ` electromagnetic masses ' mentioned above ( footnote[foot : fourthirds ] ) , as has also been demonstrated in an elegant and manifestly covariant fashion in @xcite . ] similarly , the stresses will also subtract from the electromagnetic angular momentum , this time even the larger portion of three quarters of it .
moreover , since the magnetic moment is the same as before , the stresses will have the tendency to increase the gyromagnetic ratio .
this we will see next in more detail the total angular momentum represented by ( [ eq : fullem - tensor ] ) is calculated by the general formula [ eq : angmom2mod ] @xmath211 the matter part , @xmath212 , corresponding to ( [ eq : fullem - tensor - b ] ) , yields the standard expression for a mass - shell of uniform density : @xmath213 the electromagnetic part is the same as that already calculated , since the electromagnetic field is the same .
therefore we just read off ( [ eq : totalangularmomentummod1 ] ) and ( [ eq : emmomentinertia ] ) that @xmath214 finally , using ( [ eq : intp ] ) , the contribution of the stresses can also be written down : @xmath215 adding the last two contributions shows that the inclusion of stresses amounts to reducing the electromagnetic contribution from the value given by ( [ eq : angmom2mod - b ] ) to a quarter of that value : @xmath216 in total we have @xmath217 to linear order in @xmath151 the kinetic energy does not contribute to the overall mass , @xmath129 , which is therefore simply given by the sum of the bare and the electrostatic mass @xmath218 using this to eliminate @xmath136 in ( [ eq : totangmom2mod-1 ] ) gives @xmath219 since the electromagnetic field is exactly as in the previous model , the magnetic moment in the present case is that given by ( [ eq : magndipolemoment ] ) .
the gyromagnetic factor is defined through @xmath220 which leads to @xmath221 this allows for a range of @xmath3 given by @xmath222 where @xmath223 corresponds to @xmath224 , i.e. , no electromagnetic contribution and @xmath225 corresponds to @xmath226 , i.e. , all mass is of electromagnetic origin .
the interval ( [ eq : rangeg - fac2mod ] ) looks striking , given the modern experimental values for the electron and the proton : @xmath227
however , we have not yet discussed the restrictions imposed by our slow - rotation assumption .
this we shall do next .
our model depends on the four independent parameters , @xmath228 . on the other hand , there are four independent physical observables , @xmath229 ( @xmath230 is dependent through ( [ eq : defgyrfac2mod ] ) ) .
our model provides us with a functional dependence expressing the observables as functions of the parameters : @xmath231 .
since @xmath100 is already an observable , it remains to display @xmath232 in terms of the parameters : [ eq : functobspara ] @xmath233 these relations can be inverted so as to allow the calculation of the values of the parameters from the values of the observables .
if we choose to display @xmath118 rather than @xmath151 , this gives [ eq : functparaobs ] @xmath234^{-1 } \frac{9(g-1)}{5}\,,\end{aligned}\ ] ] where the last equation ( [ eq : functparaobs - d ] ) follows from ( [ eq : functobspara - c ] ) using ( [ eq : functparaobs - a]-[eq : functparaobs - c ] ) .
it is of particular interest to us since it allows to easily express the slow - rotation assumption @xmath235 .
for this it will be convenient to measure @xmath100 in units of the elementary charge @xmath72 and @xmath236 in units of @xmath237 .
hence we write @xmath238 then , using that the fine - structure constant in si units reads @xmath239 , we get @xmath240 this nicely shows that the slow - rotation approximation constrains the given combination of angular momentum , charge , and gyromagnetic factor .
in particular , any gyromagnetic factor up to @xmath225 can be so obtained , given that the charge is sufficiently large . if we set @xmath4 and @xmath241 ( corresponding to the electron s values ) , we get @xmath242 this means that indeed we can not cover the electrons values with the present model while keeping the slow - rotation approximation , though this model seems to be able to accommodate values of @xmath3 up to six if the charge is sufficiently high .
however , we did not check whether the assumption that the matter of the shell provided the stabilising stresses is in any way violating general conditions to be imposed on any energy - momentum tensor .
this we shall do next .
energy dominance essentially requires the velocity of sound in the stress - supporting material to be superluminal .
it is conceivable that for certain values of the physical quantities @xmath243 the stresses would become unphysically high . to check that , at least for the condition of energy dominance , we first note from ( [ eq : fullem - tensor - c ] ) and ( [ eq : elmass ] ) that the stress part of the energy - momentum tensor can be written in the form @xmath244 hence the ratio between the stress within the shell ( in any direction given by the unit spacelike vector @xmath53 tangent to the shell , so that @xmath245 ) and its energy density , as measured by a locally co - rotating observer , is given by @xmath246 where we used ( [ eq : functparaobs - a ] ) and ( [ eq : functparaobs - b ] ) in the last step .
the condition of energy dominance now requires this quantity to be bounded above by 1 , so that @xmath247 interestingly this depends on @xmath3 only .
hence we get , after all , an upper bound for @xmath3 , though from the condition of energy dominance , i.e. a subluminal speed of sound in the shell material , and not from the condition of a subluminal rotational speed .
what is the size of the electron ? according to ( [ eq : functparaobs - c ] ) , its radius comes out to be @xmath248 where we set @xmath249 and @xmath4 . on the other hand , in quantum mechanics ,
the compton wavelength of the electron is @xmath250 so that their quotient is just @xmath251 this might first look as if the classical electron is really small , at least compared to its compton wavelength .
however , in absolute terms we have ( @xmath252 stands for the length scale `` fermi '' ) @xmath253 which is very large compared to the scale of @xmath254 at which modern high - energy experiments have probed the electron s structure , so far without any indication for substructures . at that scale
the model discussed here is certainly not capable of producing any reasonable values for the electron parameters , since the electrostatic mass ( and hence the total mass , if we assume the weak energy - condition , @xmath255 , for the shell matter ) comes out much too large and the angular momentum much too small ( assuming @xmath121 ) .
one might ask whether the inclusion of gravity will substantially change the situation . for example
, one would expect the gravitational binding to reduce the electrostatic self - energy .
an obvious and answerable questions is whether the electron could be a black hole ?
what is particularly intriguing about spinning and charged black holes in general relativity is that their gyromagnetic factor is @xmath4 , always and exactly !
is already a preferred value in special - relativistic electrodynamics @xcite , a fact on which modern precision measurements of @xmath256 rest .
see @xcite and @xcite for instructive discussions as to what makes @xmath4 also a special value in general relativity . ] for a mass @xmath230 of about @xmath257 to be a black hole it must be confined to a region smaller than the schwarzschild radius @xmath258 , which is almost 40 orders of magnitude below the scale to which the electron structure has been probed and found featureless . hence , leaving alone quantum theory , it is certainly a vast speculation to presumes the electron to be a black hole . but
would it also be inconsistent from the point of view of general relativity ?
the kerr - newman family of solutions for the einstein - maxwell equations allow any parameter values for mass ( except that it must be positive ) , charge , and angular momentum . as already stated , @xmath4 automatically .
hence there is also a solution whose parameter values are those of the electron .
however , only for certain restricted ranges of parameter values do these solutions represent black holes , that is , possess event horizons that cover the interior singularity ; otherwise they contain naked singularities .
more precisely , one measures the mass @xmath230 , angular momentum per unit mass @xmath259 , and charge @xmath100 of a kerr - newman solution in geometric units , so that each of these quantities acquires the dimension of length .
if we denote these quantities in geometric units by the corresponding lower case letters , @xmath129 , @xmath260 , and @xmath10 respectively , we have [ eq : geometricunits ] @xmath261 the necessary and sufficient condition for an event horizon to exist is now given by @xmath262 the relevant quantities to look at are therefore the dimensionless ratios$ ] to denote the number that gives the physical quantity @xmath263 in units of @xmath37 . ] [ eq : bh - ratios ] @xmath264}% { m[\mathrm{kg}]}\cdot 5.5\cdot 10^{18}\,,\\ \label{eq : bh - ratios - b } & \frac{q}{m}&&\quad=\quad & & \frac{q}{m}&&\cdot\sqrt{\frac{\mu_0}{4\pi}\,\frac{c^2}{g } } & & \quad \approx\quad \frac{q[\mathrm{c}]}{m[\mathrm{kg}]}\cdot 10^{10}\,.\end{aligned}\ ] ] now , if we insert the parameter values for the electron with @xmath265 ( modulus of electron spin ) and use the approximate values @xmath266\approx 10^{-34}$ ] , @xmath267=10^{-30}$ ] , and @xmath268=1.6\cdot 10^{-19}$ ] . ] ( we take for @xmath100 the modulus @xmath72 of the electron charge ) we arrive at the preposterous values [ eq : electronbh - ratios ] @xmath269 so that we are indeed very far from a black hole .
classically one would reject the solution for the reason of having a naked singularity .
but note that this does not exclude the possibility that this exterior solution is valid up to some finite radius , and is then continued by another solution that takes into account matter sources other than just the electromagnetic field . means a very small angular momentum indeed .
recall that in newtonian approximation the angular momentum of a homogeneous massive ball of radius @xmath101 is @xmath270 , so that @xmath271 translates to the following inequality for the spin period @xmath272 : @xmath273}{m[\mathrm{kg}]}\cdot 10^{19}\ , \mathrm{sec}\,,\ ] ] which for a ball of radius one meter and mass @xmath274 kilogrammes sets an upper bound for @xmath275 of @xmath276 ! in fact , ( [ eq : macrbodiesam - ratio ] ) is violated by all planets in our solar system . ]
understanding the generation of new ideas and the mechanisms that led to their acceptance is a common central concern of historians of science , philosophers of science , and the working scientists themselves .
the latter might even foster the hope that important lessons can be learnt for the future .
in any case , it seems to me that from all perspectives it is equally natural to ask whether a specific argument is actually true or just put forward for persuasive reasons . within the history of quantum mechanics the history of spin
is , in my opinion , of particular interest , since it marks the first instance where a genuine quantum degree of freedom without a classically corresponding one were postulated to exist . if this were the general situation , our understanding of a quantum theory as the quantisation of a classical theory can not be fundamentally correct
. on the other hand , modern theories of quantisation can explain the quantum theory of a spinning particle as the result of a quantisation applied to some classical theory , in which the notion of spin is already present .
hence , from a modern perspective , it is simply not true that spin has no classical counterpart .
that verdict ( that is has no classical counterpart ) , which is still often heard and/or read , is based on a narrow concept of ` classical system ' , which has been overcome in modern formulations , as was already mentioned in footnote[foot : classicalspin ] to which i refer at this point . from that point of view , spin is no less natural in classical physics than in quantum theory , which has now become the standard attitude in good textbooks on analytical mechanics , e.g. @xcite@xcite as well as in attempts to formulate theories of quantisation @xcite@xcite . in the present contribution i concentrated on another aspect , namely whether it is actually true that classical models for the electron ( as they were already , or could have been , established around 1925 ) are not capable to account for the actual values of the four electron parameters : mass , charge , angular momentum , and the gyromagnetic factor .
this criticism was put forward from the very beginning ( lorentz ) and was often repeated thereafter .
it turns out that this argument is not as clear cut as usually implied .
in particular , @xmath4 is by no means incompatible with classical physics .
unfortunately , explicit calculations seem to have been carried out only in a simplifying slow - rotation approximation , in which the poincar stresses may be taken uniform over the charged shell . in the regime of validity of this approximation
@xmath4 is attainable , but not for small charges .
i do not think it is known whether and , if so , how an exact treatment improves on the situation . in that sense , the answer to the question posed above is not known .
an exact treatment would have to account for the centrifugal forces that act on the rotating shell in a latitude dependent way . as a result ,
the poincar stresses can not retain the simple ( constant ) form as in ( [ eq : fullem - tensor - c ] ) but must now also be latitude dependent . in particular , they must be equal in sign but larger in magnitude than given in ( [ eq : surfacestress ] ) since now they need in addition to balance the outward pushing centrifugal forces .
on one hand , this suggests that their effect is a still further reduction of angular momentum for fixed magnetic moment , resulting in still larger values for @xmath3 . on the other hand ,
fast rotational velocities result in an increase of the inertial mass according to ( [ eq : massveldep ] ) and hence an increase of angular momentum , though by the same token also an increase in the centrifugal force and hence an increase in stress .
how the account of these different effects finally turns out to be is unclear ( to me ) without a detailed calculation .
it would be of interest to return to this issue in the future .
domenico giulini .
algebraic and geometric structures in special relativity . in claus lmmerzahl and jrgen ehlers , editors ,
_ special relativity : will it survive the next 101 years ? _ , volume 702 of _ lecture notes in physics _ , pages 45111 .
springer verlag , berlin , 2006 .
arxiv math - ph/0602018 .
samuel a. goudsmit .
the discovery of the electron spin .
online available at @xmath5www.lorentz.leidenuniv.nl / history / spin / goudsmit.html@xmath6 .
concluding lecture given at the occasion of the golden jubilee meeting of the dutch physical society in april 1971 .
henri poincar . .
, 21:129176 , 1906 .
marcus schulte . .
online available at @xmath5http://deposit.ddb.de / cgi - bin / dokserv?idn=960517863@xmath6 .
dissertation at fb 14 of the university of dortmund , december 1999 . | in december 1924 wolfgang pauli proposed the idea of an inner degree of freedom of the electron , which he insisted should be thought of as genuinely quantum mechanical in nature .
shortly thereafter ralph kronig and , independently , samuel goudsmit and george uhlenbeck took up a less radical stance by suggesting that this degree of freedom somehow corresponded to an inner rotational motion , though it was unclear from the very beginning how literal one was actually supposed to take this picture , since it was immediately recognised ( already by goudsmit and uhlenbeck ) that it would very likely lead to serious problems with special relativity if the model were to reproduce the electron s values for mass , charge , angular momentum , and magnetic moment .
however , probably due to the then overwhelming impression that classical concepts were generally insufficient for the proper description of microscopic phenomena , a more detailed reasoning was never given . in this contribution
i shall investigate in some detail what the restrictions on the physical quantities just mentioned are , if they are to be reproduced by rather simple classical models of the electron within the framework of special relativity .
it turns out that surface stresses play a decisive role and that the question of whether a classical model for the electron does indeed contradict special relativity can only be answered on the basis of an _ exact _ solution , which has hitherto not been given . |
in this paper , we consider the construction of absorbing boundary conditions ( abcs ) for the nonlinear schrdinger equation ( nls ) , @xmath0\psi,\label{nls}\ ] ] where @xmath1 is the atomic mass , @xmath2 represents the planck constant , and @xmath3 is the potential function .
this equation appears in many different applications @xcite , such as the gravity waves on deep water in fluid dynamics , the pulse propagations in optics fibers , and the bose - einstein condensation . for the cubic nonlinear term @xmath4 , eq .
( [ nls ] ) reduced to gross - pitaevskii equation , which be numerically studied by @xcite and references therein . for a quintic nonlinearity
, one can see @xcite .
a current challenge for the numerical solutions of this kind of problems is due to the unboundedness of the physical domain .
the common practice to overcome this difficulty is to limit the computational domain and solve a reduced problem . to make the truncated problems complete , boundary conditions
should be added to the reduced problem .
abcs applied at prescribed artificial boundaries have been widely studied in recent decades , such as @xcite for hyperbolic wave equations , @xcite for elliptic equations , @xcite for parabolic equations ; see also the reviews by givoli@xcite , tsynkov@xcite , hagstrom@xcite and han@xcite .
the purpose of designing abcs is to annihilate all the incident waves so that minor reflected waves may propagate into the computational domain .
abcs can be distinguished into two categories : the global abcs and local abcs .
the global abcs usually lead to the well - approximated and well - posed truncated problems , but the implementation cost for the global abcs is expensive .
fast evaluation methods @xcite should be applied to treat the global abcs . on the other hand ,
local abcs are computationally efficient , but the accuracy and stability are the main concerns .
there is another way to design abcs based on the media of the material , called the perfectly matched layer ( pml ) methods @xcite , which have been applied to many complicated wave propagation problems . for the nonlinear equations ,
however , it is difficult to find a suitable boundary condition on the artificial boundary in general , except for some special cases such as the problem can be linearized @xcite , or a burgers type questions @xcite which can be corresponded with the linear parabolic equations by the cole - hopf transformation . for the nls equation on the unbounded domain
, zheng @xcite obtained the transparent boundary condition using the inverse scattering transform approach for the cubic nonlinear schrdinger equation in one dimension .
antonie et al .
@xcite also studied the one - dimensional cubic nonlinear schrdinger equation and constructed several nonlinear integro - differential artificial boundary conditions . in @xcite , szeftel designed absorbing boundary conditions for one - dimensional nonlinear wave equation by the potential and the paralinear strategies .
soffer and stucchio@xcite presented a phase space filter method to obtain absorbing boundary conditions .
the pml @xcite was also applied to handling the nonlinear schrdinger equations .
the recent papers @xcite , split local absorbing boundary ( slab ) conditions have been developed through a time - splitting procedure for one - dimensional nonlinear schrdinger equations .
the local absorbing boundary conditions were imposed on the split linear subproblem and yielded a full scheme by coupling the discretizations for the interior equation and boundary subproblems .
in addition , an adaptive method was developed to determine the parameter @xmath5 involved in the local boundary conditions . in this paper
, we present an efficient implementation of the nonlinear absorbing boundary conditions for the nls equation by making use of the splitting idea @xcite .
we distinguish incoming and outgoing waves along the boundaries for the linear kinetic subproblem , and then unite the potential energy subproblem to yield nonlinear boundary conditions which are of local form and examine the coupling between the equation governing the wave in the computational domain and the boundary conditions on the boundary yields a well - posed problem .
furthermore , the obtained boundary conditions are easy " to discretize and cheap " to compute in terms of the computational time .
we will then perform numerical tests to illustrate the effectiveness and tractability of this approach .
let us first consider the local abcs of the linear schrdinger equation , @xmath6 which will be the basis of constructing abcs of nonlinear problems , because it is the kinetic part of the nls equation ( [ nls ] ) . in the frequency domain of the fourier transform , the one way equations at the two boundaries , which annihilate all the outgoing waves ,
can be represented by @xmath7 where the subscript represents the partial derivative , @xmath8 is the fourier transform of @xmath9 , @xmath10 is the frequency , the plus sign in @xmath11 " corresponds to the right boundary condition , and the minus sign corresponds to the left . the transparent boundary conditions in exact manner of the problem
is then derived by an inverse fourier transform : @xmath12 which is a nonlocal boundary condition . in order to get local boundary conditions
, a usual way is to approximate the square root @xmath13 by using polynomials or rational polynomials .
for instance , applying the approximations @xmath14 we obtain the first three absorbing boundary conditions in the physical domain : @xmath15 here , @xmath16 is a positive constant , and @xmath17 which is called the wavenumber parameter . in this section ,
we derive the absorbing boundary conditions for the one - dimensional nls equation ( [ nls ] ) on the region @xmath18 $ ] by assuming that there do not exist outgoing waves at the two ends @xmath19 and @xmath20 .
we remark that this assumption is not correct for a general wave due to the nonlinear or space - dependent potential @xcite , but it is reasonable for many situations of physical applications ; in fact , this is a fundamental assumption so that we can design absorbing boundary conditions . to understand the philosophy of obtaining nonlinear absorbing boundary conditions , let us rewrite eq .
( [ nls ] ) into the following operator form : @xmath21\psi(x , t ) , \label{operform}\ ] ] where @xmath22 represents the linear differential operator which accounts for the kinetic energy part , and @xmath23 is the nonlinear operator that governs the effects of the potential energy and nonlinearity .
these are given by @xmath24 in a time interval from @xmath25 to @xmath26 for small @xmath27 , the exact solution of eq .
( [ operform ] ) can be approximated by @xmath28\tau/\hbar}\psi(x , t ) \approx e^{-i\hat{t}\tau/\hbar}e^{-i\hat{v}\tau/\hbar}\psi(x , t ) .
\label{splitting}\ ] ] this means in a small time step @xmath27 that the approximation carries out the wave propagation in a kinetic energy step and a potential energy step separately .
this is the basic idea of the well - known time - splitting method ( or the split - step method ) which is effective to numerically solve the nonlinear schrdinger - type equations .
it is natural to examine the well - posedness that we concern most of the induced initial boundary value problem confined in the finite computational interval for the nonlinear schrdiger equation by using the nonlocal absorbing boundary condition .
the classical energy method seems to be difficult to obtain the stability result of the problem under consideration here . in this section
, we will investigate the stability based on kreiss s normal mode analysis method@xcite and references therein , which was also used for the wave equations and the linear schrdinger equation .
the idea of the general algebraic normal mode analysis is based on the fact that the well - posed problem does not admit any complex eigenvalues with positive real parts @xmath41 , or generalized eigenvalues with @xmath42 and the negative ( positive ) group velocity of normal mode on the right - hand ( left - hand ) boundary .
the eigenvalues and generalized eigenvalues satisfy the dispersion relations of both the interior equation and the equation on the boundaries .
if there exist such eigenvalues , the problem will admit an unboundedly grown normal mode @xmath43 and is hence unstable .
if there exit such generalized eigenvalues , the boundary conditions will admit a incoming wave which will propagate energy into the interior domain to disrupt the solution in the computational interval and generate a spurious wave solution .
we need to check if there exist eigenvalues and generalized eigenvalues by replacing the solution of the plane wave form @xmath44 into eq .
( [ ibvp - e ] ) and boundary conditions ( [ ibvp - bl])([ibvp - br ] ) .
there are left - hand and right - hand boundary , they can be considered by the same argument . here we only consider the wellposedness with the right - hand boundary conditions .
for n=2 , substituting the plane wave into the dispersion relation and the boundary condition yields @xmath45 , \label{i}\\ & & i\hbar s =- i\hbar\sqrt{\frac{2\hbar}{m}}k_0k- \hbar k^2_0+[v(x)+f(|\psi|^2)].\label{ii}\end{aligned}\ ] ] in very small time step @xmath27 the basic idea of the well - known time - splitting method is to carry out the wave propagation in a kinetic energy step and a potential energy step separately . when time step is taken limit to zero , that is the reason why we can write the nonlinear shrdinger equation in above forms , the nonlinear term is considered as a potential function .
hence , let ( [ i ] ) minus ( [ ii ] ) and substitute the result into ( [ i ] ) to arrive at @xmath46 where @xmath47 .
for the @xmath48 order approximation on the right boundary , by the same argument , we have @xmath45,\label{aa}\\ & & i\hbar s=-\frac{\hbar^2k_0 ^ 2}{2m}\frac{3ik+k_0}{ik+3k_0 } + [ v(x)+f(|\psi|^2)].\label{bb}\end{aligned}\ ] ] combine ( [ aa ] ) and ( [ bb ] ) , we have @xmath49 solve ( [ k ] ) and substitute the results @xmath50 into ( [ aa ] ) to obtain @xmath51}{\hbar}\\ & & s =- i\frac{\hbar}{2m}k_0 ^ 2-i\frac{v_1+f_1}{\hbar } + \frac{v_2+f_2}{\hbar}.\label{order3}\end{aligned}\ ] ] it is easy to have @xmath52 for @xmath53 from ( [ order2 ] ) and ( [ order3 ] ) whatever @xmath54 or @xmath55 . in particular , when @xmath56 and @xmath57 are both real functions , i.e. @xmath58 , we thus have @xmath59 hence @xmath60 is wholly imaginary .
since the boundary condition is designed to annihilate all the outgoing wave , this implies that the group velocity is positive on the right - hand boundary .
hence there is no generalized eigenvalue which will propagate energy into the interior interval to influence on the true wave solution .
if any instabilities which might be admitted by the generalized eigenvalues of the abcs exit , they would not propagate wave to affect the interior solution . keep this concept in mind , by linearizing the nonlinear terms in a sufficiently small time step , we can say that the boundary conditions of the problem ( [ ibvp - e])-([ibvp - br ] ) are well - posed if functions @xmath61 and @xmath62 satisfy @xmath63 .
hence the boundary condition for n=2 or 3 is well - posed and numerical tests below show that the boundary condition is stable .
in this section we will test the efficiency of the nonlinear absorbing boundary conditions by using the finite - difference scheme to the induced initial boundary value problem ( [ ibvp - e])-([ibvp - br ] ) .
+ in the computational interval @xmath39,$ ] let @xmath64 is a positive integer , denote @xmath65 and @xmath66 by grid sizes in space and time , respectively .
we use the following crank - nicolson - type difference scheme @xcite @xmath67\psi_j^{n+\frac{1}{2}},\nonumber\\ \label{appe}\end{aligned}\ ] ] to discretize the nonlinear schrdinger equation inside the computational domain , which is unconditionally stable and of second order in time . here
@xmath68 and @xmath69 represents the approximation of wave function @xmath70 at the grid point @xmath71 with @xmath72 @xmath73 and @xmath74 .
obviously , the above systems of equations can not be solved uniquely since the number of equations are less than that of unknowns . at the two unknown ghost points @xmath75 and @xmath76
no equation can be defined from ( [ appe ] ) . thus the absorbing boundary conditions are required to provide two extra equations on the grid points @xmath77 for @xmath78 in the vicinity of boundary .
the @xmath48 order abcs are discretized as follows : @xmath79\frac{\psi_{s+1}^{n+\frac{1}{2}}-\psi_{s-1}^{n+\frac{1}{2}}}{2\delta
x } -\hbar\frac{\psi_{s+1}^{n+1}-\psi_{s-1}^{n+1}}{\delta x \delta t } + \hbar\frac{\psi_{s+1}^{n}-\psi_{s-1}^{n}}{\delta x \delta t } \nonumber\\ & \pm6k_0i\frac{\psi_{s}^{n+1}-\psi_{s}^{n}}{\delta t}\pm\left(\frac{\hbar^2k_0 ^ 3}{m}-6k_0v_s - 6k_0f(|\psi_s^{n+\frac{1}{2}}|^2)\right)\psi_s^{n+\frac{1}{2 } } = 0 , \label{appeb}\end{aligned}\ ] ] where the positive sign in @xmath38 corresponds to the right boundary @xmath80 and the negative one corresponds to the left boundary @xmath81 the above scheme is implicit , we need use an iterative strategy to solve the nonlinear equations ( [ appe])-([appeb ] ) by replacing @xmath82 with @xmath83 , where the superscript @xmath84 denotes the @xmath84-th iteration at each time step and the initial iteration is given as @xmath85 .
+ without loss of generality , we rewrite the original nls equation ( [ nls ] ) as @xmath86 where @xmath87 represents different physical phenomena when chosen different nonlinear forms . in example 1 , we take @xmath88 , @xmath89 and @xmath90 , which corresponds to a focusing @xmath91 effect of the cubic nonlinearity .
many applications in science and technology have connections with this kind of nls equations @xcite .
furthermore , in this case equation ( [ equation ] ) can be solved analytically by the inverse scattering theory , its exact soliton solution has the following form @xmath92 where @xmath93 and @xmath94 are real parameters:@xmath93 determines the amplitude of the wavefield and @xmath94 is related to velocity of the soliton . in example 2 , we take @xmath95 , @xmath89 and @xmath90 , the nls equation becomes of a quintic nonlinearity .
the computational interval is set to be @xmath39 = [ -5,5].$ ] for this kind of nls equation with power - like nonlinearity , if the initial energy satisfies : @xmath96 the solution @xmath70 will blow up in finite time ( @xcite and references therein ) , i.e , there exists @xmath97 such that @xmath98 in example 3 , we take @xmath99 and the temporal evolution of an initial bright solition @xmath100 with @xmath101 .
the exact solution is the form of ( [ exact_soltion ] ) with @xmath102 and @xmath103 .
this case is used to test the performance of the nonlinear abcs when the wave impinges on the right - hand - side boundary of the computational interval @xmath104 $ ] .
1 shows the evolution of absolute error @xmath105 on the artificial boundary point by picking wavenumber @xmath106 , spatial size @xmath107 and time size @xmath108 and its refinement mesh .
when refining the mesh , one can see that the error will converge quickly . in order to explore the convergence rates in space and time ,
we take time size as @xmath109 .
denote the @xmath110norm by @xmath111 fig .
2 shows the evolution of the @xmath110norm with different mesh sizes . from this figure
, we observe a second - order accuracy of the errors . to character convergence rates exactly , we take the different values of the @xmath112-norm at different times and mesh sizes .
table i shows the second - order convergence rate .
.the values and orders of @xmath112-norm at different times and mesh sizes . [ cols="^,^,^,^,^,^,^",options="header " , ] * example 2 . *
we consider the quintic nls equation with the initial value @xmath113 where the wavenumber @xmath114 .
a direct calculation obtains the initial energy @xmath115 which implies that the numerical experiment will not blow - up . by using the crank - nicolson scheme proposed by durn et al @xcite , fig .
4 shows the reflection ratios for different wavenumbers under spatial size @xmath116 and its refinement mesh size @xmath117 , respectively . comparing the two curves
, we can see that the reflection ratios decrease fast when wavenumber taken in the vicinity of the initial wavenumber @xmath114 . and
the reflection ratios in larger mesh is larger than these in smaller mesh when @xmath118 .
this indicates that the parameter @xmath5 in nonlinear abcs should be given in suitable way , i.e , we had better endow the wavenumber with being equal to the initial wavenumber if known .
we can also see if the @xmath5 is equal to half of the group velocity of the wave impinged on the boundary point for third - order case , the nonlinear abcs will not disrupt the the interior solution when mesh is refined and refined .
5 shows the wave amplitude of utilizing the crank - nicolson method proposed by durn under mesh size @xmath119 , @xmath120 and @xmath121 .
the reflecting wave can not be observed and the reflection ration is 7.637e-5 . +
* example 3 .
* we consider a nonlinear wave with repulsive interaction with @xmath122 in ( [ equation ] ) .
the initial data and potential function are taken to be gaussian pulses @xmath123 with @xmath101 .
it has been an example in @xcite , and is used to model expansion of a bose - einstain condensate composed of waves with different group velocities . in the calculation , we choose @xmath124 .
( a)-(b ) depict the evolution of the wave s amplitude with the third - order nonlinear abcs for different wavenumbers @xmath5 .
we emphasize the very small spurious reflections by highlighting the figures , which can make us visualize them by showing the associated shadow zones . and
the reflective waves are too small to be seen with wavenumbers chosen in the appropriate range [ 1.25,2 ] . in fig .
6 ( a ) and 6(b ) , no observable reflection can be defected at all .
if we take the boundary conditions as dirichlet or neumann boundary condition , the reflective wave is very large ( ref . to @xcite ) at t=6 with the same meshes .
we remark that picking the suitable wave - number @xmath5 in the nonlinear abcs plays a very important role if we want to obtain an efficient nonlinear absorbing boundary condition with more and more refinement mesh .
the ideal wavenumber @xmath5 is equal to a half of group velocity when a single solitary wave impinges on the boundary . for an arbitrary wave packet
, a suitable wavenumber can be picked adaptively by using the gabor transform @xcite to capture local frequency information in the vicinity of artificial boundary , for which we will not discuss in detail in this paper .
through approximating the kinetic energy term by a one - way equation and uniting it with the potential energy equation , we proposed an efficient method for designing absorbing boundary conditions of nonlinear schrdinger equations .
the obtained boundary conditions , which are of nonlinear form , can perfectly absorb the outgoing waves .
we examined the coupling between the equation governing the wave in the computational domain and the boundary conditions on the boundary yielded a well - posed problem in a sufficiently small time interval .
furthermore , some numerical examples are given to demonstrate the effectiveness and tractability of the artificial boundary conditions .
the authors would like to thank professor houde han for fruitful discussion on designing the nonlinear bcs .
this research is supported in part by rgc of hong kong and frg of hong kong baptist university .
a. du@xmath126n , j.m .
sanz - serna , 20(2)(2000)235 - 261 . b. enguist , a. majda , 31(1977)629 - 651 .
d. givoli , elsevier , amsterdam , 1992 .
m. m. cerimele et al . , phys .
e 62 , 1382(2000 ) .
s. k. adhikari , phys .
e 62 , 2937(2000 ) .
weizhu bao , dieter jaksch , peter a. markowich , 187(2003 ) 318 - 342 . c. farrell and u. leonhardt .
, 7:14 , 2005 .
thomas fevens , hong jiang , vol .
21 , no . 1(1999)255 - 282
. t. hagstrom , acta numer .
( 1999)47 - 106 .
l. halpern , j. rauch , numer .
71(1995)185 - 224 . y. b. gaididei et al .
e 60 , 4877(1999 ) .
t. hagstrom , h.b .
keller , 48(1987)449 - 470 .
h. han , in : t. li , p. zhang(eds . ) , frontiers and prospents of contemporary applied mathematics , higher education press and world scientific , 2006 , pp .
h. han , z. huang , 44(2002)655 - 666 .
han , x.n .
wu , 3(1985)179 - 192 .
han , x.n .
xu , , 24(2006)295 - 304 .
higdon , 47(1986)437 - 459 .
s.d jiang , l. greengard , 47(2004)955 - 966 .
kuska , , vol 46 , no .
8(1999)5000 - 5003 .
kreiss , j. lorenz , vol .
136(1989 ) . l. p. pitaevskii and s. stringari .
oxford university press , oxford , 2003 .
a. soffer , c. stucchio , 225(2)(2007 ) 1218 - 1232 .
j. szeftel . , 195:37603775 , 2006 .
c. sulem and p.l .
springer , 1999 .
tsynkov , appl .
numer . math .
27(1998 ) 465 - 532 . c. zheng . ,
215:552565 , 2006 . j. szeftel .
, 104:103127 , 2006 .
zhang , , to appear .
z. xu and h. han .
, 74:037704 , 2006 . | an efficient method is proposed for numerical solutions of nonlinear schrdinger equations in an unbounded domain . through approximating the kinetic energy term by a one - way equation and uniting it with the potential energy equation , absorbing
boundary conditions are designed to truncate the unbounded domain , which are in nonlinear form and can perfectly absorb the waves outgoing from the truncated domain .
we examine the stability of the induced initial boundary value problems defined on the computational domain with the boundary conditions by a normal mode analysis .
numerical examples are given to illustrate the stable and tractable advantages of the method . |
it is not unusual in science to find that the initial early detection of a phenomenon is followed by a rapidly increasing discovery rate as interest intensifies and new technology is developed .
such has been the case with the search for substellar companions to stars in the solar neighborhood . while there have been claims of detections of companions to stars with masses less than the lower limit for the mass of a star
, most have not stood the test of time . however , that has changed following the announcement of the detection of a companion to the star hd 114762 ( latham et al .
1989 ) . during the interval of time from 1995 to early 1997 there have been reported detections of 20 substellar mass companions to nearby stars .
the recent avalanche of detections began with the paper announcing the discovery of a companion to the star 51 pegasi ( mayor and queloz 1995 ) .
the lower limit to the mass of the companion is 0.45 jupiter masses .
more remarkable than the mass of the companion was its orbital period , 4.23 days .
the semi - major axis of this companion is only 0.05 au ! the detection of the radial velocity variation for 51 peg was confirmed and there began a rapid sequence of detections of what have been called by their discoverers `` extrasolar planets '' .
but , are these companions really planets ? on what basis was that interpretation of the data made , and how firm is it ? could these objects be something else , like the bottom end of the star formation process , viz . , brown dwarfs , or an as yet unidentified class of astronomical object ? indeed , there now is a serious challenge ( gray 1997 ) to the interpretation of the 51 pegasi signal as being due to a companion of any kind
. it may be that all of the short period signals that have been attributed to planetary companions may be intrinsic to the star , but that will become clear with additional observations of the type conducted by gray .
the criteria suggested here do not depend upon the reality of those companion systems .
the fundamental distinction between brown dwarfs and planets is the manner in which they are formed ( e.g. , kaftos et al . 1986 ) .
brown dwarfs are formed in the same manner as a star , which means that if they are found as companions to stars then they formed by that process involved in the formation of a binary star system .
the underlying mechanism in that case is not fully understood , but is thought to involve large - scale gravitational instabilities .
planets , at least the nine that we can study in detail , in contrast appear to be formed by an accretion process beginning from small dust grains building up to planetesimals to lunar - sized objects to terrestrial planet sized objects and then , depending on the availability of gas , giant gas - rich planets such as jupiter .
thus , the distinction between brown dwarfs and planets is fundamental .
it is not a `` matter of semantics '' .
an incorrect identification of the nature of these companions would lead in turn to erroneous notions about the basic processes involved in their formation and evolution .
importantly in this regard , as we do not at this time know either the lower limit to the mass of a brown dwarf , or the upper limit to the mass of a planet , one can not at this time use mass alone as the basis for identifying a given companion to a star as a planet , unless the masses are well below likely masses for brown dwarfs ( i.e. , earth mass companions ) .
the sub - stellar mass companions discovered since 1989 are listed in table 1 .
there are two exceptions .
the companion to hd 98230 was discovered , or at least the observations made , in 1931 ( bergman 1931 ) .
also , the table lists only those companions discovered by radial velocity observations .
the possible companion(s ) to lelande 21185 ( gatewood 1996 ) are not included as there is no reliable estimate at present of their eccentricity .
crrr 51 peg & 0.45 & 4.23 & 0.00 v and & 0.65 & 4.61 & 0.11 55 cnc & 0.84 & 14.65 & 0.05 rho crb & 1.1 & 39.64 & 0.03 16 cyg & 1.6 & 804 & 0.65 47 uma & 2.3 & 1090 & 0.08 @xmath0 boo & 3.9 & 3.31 & 0.00 70 vir & 7.4 & 116.7 & 0.37 hd 114762 & 9.0 & 84.02 & 0.33 hd 110833 & 17 & 270 & 0.69 bd -04:782 & 21 & 240.92 & 0.28 hd 112758 & 35 & 103.22 & 0.16 hd 98230 & 37 & 3.98 & 0.00 hd 18445 & 39 & 554.67 & 0.54 hd 29587 & 40 & 1471.70 & 0.37 hd 140913 & 46 & 147.94 & 0.61 bd + 26:730 & 50 & 1.79 & 0.02 hd 89707 & 54 & 298.25 & 0.95 hd 217580 & 60 & 454.66 & 0.52
the observational criterion suggested here for distinguishing between brown dwarf and planetary companions to a star concerns the relationship between two properties of the companion s orbit , namely its eccentricity as a function of its period .
figure 1 shows orbital eccentricity versus the logarithm of orbital period , expressed in days , for two populations .
one population is pre - main - sequence ( pms ) binaries ( mathieu 1994 ) .
another is objects thought to have formed via accretion in a disk , the giant planets in the solar system .
note that low eccentricity is a characteristic for other objects such as the terrestrial planets , the regular satellites of the giant planets in the solar system , and the companions to the pulsar psr 1257 + 12 ( wolszczan and frail 1992 ) , all of which are thought to have formed by accretion from a disk .
data for pms binaries were used as they presumably reflect any signature of the binary formation process with minimal alteration . also , only systems with periods less than 10@xmath1 days were used because of the apparent difference in the mass function for this class of companions as compared with field stars and companions with periods longer than 10@xmath1 days ( abt and levy 1976 , mazeh et al .
this difference has been taken as evidence by these authors and by mathieu ( 1994 ) that the formation process for short - period binaries differs from that for field or long - period companions ( see also trimble and cheung 1976 , trimble 1990 ) .
main - sequence binaries with periods in the range considered here display a similar ( e , log p ) distribution as do the pms binaries .
orbital evolution is thus not likely to be a major consideration for periods much beyond a few weeks .
furthermore , it appears that the ( e , log p ) distribution is established early in the history of these systems ( mathieu ) , consistent with a signature of their formation .
the distribution of ( e , log p ) for stellar companions is markedly different from that for planetary companions .
this difference is a reflection of a corresponding difference in the manner by which objects in these two classes were formed .
a least - squares fit to the binary star distribution gives e = 0.2 log p - 0.03 .
this yields a pearson correlation coefficient of r = 0.76 .
this statistical test suggest that the causal relationship between eccentricity and period is a modest but significant one .
while a trend is clear , there is a range of eccentricities at a given period .
if the proposed ( e , log p ) criterion is valid one expects that the ( e , log p ) distribution for binary brown dwarfs would mimic , if not be indistinguishable from , that shown in figure 1 .
ten of the companions in table 1 have 17 m@xmath2sin_i _ 80 and as such are likely to be brown dwarfs .
the ( e , log p ) set for those ten companions have a pearson coefficient r = 0.67 .
this is similar to that for pms binary companions .
the least - squares fit for the data gives e = 0.22 log p - 0.05 , indistinguishable from that of pms binaries given the uncertainties in the coefficients .
a better measure of the reality of these correlations is the spearman coefficient as it is derived from a simple ranking of the data and does not depend on knowing the distribution function that might underlie the data ( e.g. , normal bivariate ) .
the spearman coefficient for the brown dwarf companions is @xmath3 = 0.58 .
this suggests that the correlation between e and log p for this group of ten systems is significant at the 95 percent level .
it is clear from the above that the orbital signature of formation is the same for pms binaries and the brown dwarf companions , demonstrating the validity of the criterion .
we now apply the criterion to the group of substellar companions listed in table 1 with m@xmath2sin_i _ 10 .
the least - squares fit gives e = 0.14 log p - 0.04 .
this is similar to , but differs from the best fits to the pms and brown dwarf populations .
the pearson coefficient for this group is r = 0.63 , and the spearman coefficient is @xmath3 = 0.73 .
it should be noted that the results for this group are skewed heavily by two systems , 47 u ma and rho crb . if these two systems are ignored , and there is no a priori justification for so doing , the remaining systems yield a best fit of e = 0.23 log p - 0.10 .
this gives values of e at a given value of log p that are in good agreement with the values calculated from the best - fit lines for pms and brown dwarfs .
the pearson coefficient for this fit is r = 0.95 , and the spearman coefficient is @xmath3 = 0.96 .
this trend is both statistically significant and consistent with that of the pms binaries and brown dwarfs .
shown in figure 2 are all three populations , pms , brown dwarfs , and companions with m@xmath2sin_i _ 10 .
also shown for reference is the best fit line for the combined population .
the spearman coefficient for the three populations combined is 0.75 . with a sample consisting of 44 members
, this value of the spearman coefficient indicates that the observed correlation is significant at greater than the 99.999 percent confidence level .
it would appear that with two exceptions , a third if we include the astrometric system lelande 21185 , all of the newly discovered , sub - stellar mass companions , have an ( e , log p ) distribution that could be drawn from the same population and that this distribution is indistinguishable from that of pms binary companions .
given that this distribution is a signature of their formation process , it would appear that these companions formed by the same process as did the binary stellar companions .
that is , those companions display the orbital signature of brown dwarfs and not that of planets .
some authors ( e.g. , butler et al . ) have suggested that companions with low eccentricity , such as the putative companion to 51 pegasi , are planets by virtue of their low eccentricity .
however , as is clear from figure 1 and knowledge of tidal effects ( e.g. , duquennoy and mayor 1992 ) , any companion that is that close to a star will be in a low eccentricity orbit .
the substellar companions with short periods are fully consistent with them being members of the stellar population . for any companion that has a sufficiently short period that its orbit will be tidally circularized , it s eccentricity alone can not be used as a criterion for identifying it as a planet . there is an additional orbital signature that could be useful as a criterion on whether a companion is a planet ( black 1980 ) .
this arises again from the formation mode , namely , more than one companion is formed and the resultant orbital architecture is characteristic of that process , and that architecture differs from that of a multiple star system .
the regular satellite systems of jupiter , saturn , and uranus , as well as the companions to the pulsar psr 1257 + 12 , and the planets in the solar system are all multiple systems with geometric spacing of the orbits .
the five systems that we have reason to believe have companions formed by accretion all have the same basic architecture .
if the companions that have been discovered are planets , one can expect that other companions are present .
so far , no such companions have been found .
their continued absence would weaken the interpretation of the companions in table 1 as planets , whereas detection of such companions strengthens that interpretation .
nature knows how to make even stellar mass companions with short orbital periods , so if these objects are brown dwarfs , their presence close to the star is not a mystery , but is to be expected . also , the high eccentricities that are seen for the putative planets with orbital periods longer than a few weeks is perfectly natural if they are brown dwarfs .
for example , given the orbital period of the companion to 16 cyg b , the best fit line _ based just on the pms binary star distribution _ would predict an eccentricity of 0.55 for the companion .
the observed eccentricity is 0.65 , good agreement given the uncertainty in both the fit and the data .
similarly , the predicted eccentricity for the companion to 70 vir is 0.38 , virtually identical with the observed value of 0.37 .
the fit is even better if one uses the regression defined by all the data points shown in figure 2 .
there is no need to postulate a form of dynamical instability arising from the presence of a distant binary to account for the observed eccentricity ( e.g. , holman et al .
1997 , innanen et al 1997 ) .
indeed it is problematic whether the proposed mechanism would work in the presence of a system where there are other sources of gravity such as other companions , as would be the case if the companion is a planet , or a disk , in the vicinity of the companion to 16 cyg .
this caveat is pointed out by both holman et al . and innanen et al .
there is no binary in the case of 70 vir to cause the eccentricity of this companion so if it is a `` planet '' , one must invoke yet another mechanism .
the binary formation mechanism would appear to provide a single , unifying context for the observed orbital eccentricities and their systematic variation with orbital period .
the interpretation of the nature of the newly discovered substellar companions that emerges from the ( e , log p ) criterion also removes the need for mechanisms ( lin et al .
1996 , trilling et al .
1996 , weidenschilling and marzari 1996 ) that postulate significant ( i.e. , nearly unity fractional change in semi - major axis ) orbital evolution to account for short - period companions .
it should be noted that contrary to statements elsewhere ( e.g. , marcy and butler ) the location of the apparent companion to 47 u ma has not been shown to be inconsistent with where a giant planet might form based upon our current paradigm for gas giant planet formation .
the companion to rho crb , if it is a planet , would likely have experienced orbital migration , but an alternative interpretation is that it is a brown dwarf , formed near its current location , with a slightly lower eccentricity than is typical for such companions with that orbital period .
this does not mean that such mechanisms can not , or do not , occur in planetary systems , but there is no independent evidence that such large - scale migration occurs in those disks that end up making planetary systems . on the contrary , the five bona fide systems where it is believed that accretion - formed companions exist all show no signs ( e.g. , eccentric and non - geometric spaced orbits ) of such motion .
it will be important to understand the _
testable _ , that is observable , consequences of models that suggest that systems such as 51 peg are the result of large - scale motion through a disk .
clearly one such observable test for those models that rely upon gravitational scattering of one giant planet by another would be the presence of a second companion in a very high eccentricity , longer period orbit .
failure to find such companions would rule out such models .
models that rely solely on interactions between a single large planet and its parent disk are more difficult to verify by examining only their end state .
their verification will likely await instruments capable of very high spatial resolution far - infrared to sub - millimeter studies of young systems where the possible detection of density waves in such disks could signal that the process in question is taking place .
another observational criterion for distinguishing between brown dwarfs and planets was suggested originally by lunine ( 1986 ) and refined recently ( saumon et al . 1996 ) .
the basic notion is that as planets form via accretion with solid material playing a key role , one expects that the abundance of metals relative to hydrogen will be greater than in the central star .
that is the case for jupiter relative to the sun .
in contrast , as a brown dwarf is formed by the same process as a star , one expects that the composition of a brown dwarf companion will be similar to , if not identical with , that of its stellar companion .
a spectroscopist confronted with two jupiter - mass objects , one a planet and the other a brown dwarf , should be able in principle to distinguish which is which on the basis the metal to hydrogen ratio normalized to the same ratio in the central star .
spectroscopic studies of companions to other stars are beyond the capability of existing instruments except for situations where the companion is both relatively luminous and well - separated from its stellar companion .
thus , while this observational test has potential , its full application must also await future instrumentation .
the perspective offered here raises the important point that these new detections are providing the first observational guide to the lower limit to the mass of a brown dwarf .
assuming that there is nothing pathological in the viewing geometry for these systems ( i.e. , sin_i _
@xmath4 0 ) , then it would appear that the lower limit is comparable to the mass of jupiter .
it will be interesting to see whether even less - massive companions are discovered with short orbital periods .
systems with the accuracy of those now in use should be able to detect companions with masses as low as a few tens of earth masses in orbits with periods of a few days .
* acknowledgement : * i wish to thank the referee for useful comments on the manuscript .
this research was conducted at the lunar and planetary institute which is operated by the universities space research association under contract nasw-4574 with the national aeronautics and space administration .
abt , h.a . and
levy s. 1976 , ap .
j. supp . , 30 , 273 - 306 bergman , l. 1931 , lick obs .
, 15 , 109 black , d.c .
1986 , in astrophysics of brown dwarfs , kaftos , m.c . , harrington , r. s. & maran , s. p. , cambridge univ . press , cambridge , 139 - 147 butler , r. p. , and marcy , g. w. 1996 , ap.j.lett . , 464 , l153 - 156 butler , r. p. , marcy , g. w. , williams , e. , hauser , h. , and shirts , p. 1997
, in press cochran , w. d. , and hatzes , a. p. 1996
, bull.a.a.s .
, 28 , 1111 duquennoy , a. and mayor , m. in binaries as tracers of stellar formation , duqennoy , a. & mayor , m. , cambridge univ .
press , 52 gatewood , g.d .
1996 , bull .
, 28 , 885 gray , d. f. 1997 , nature , 385 , 449 holman , m. , touma , j. and tremaine s. 1997 , nature , 386 , 254 innanen , k. a. , zheng , j. q. , mikkola , s. , and voltonen , m. j. 1997 , to appear , astron . j. kaftos , m. c. , harrington , r s. , and maran , s. p. 1986
, astrophysics of brown dwarfs , cambridge univ . press
, cambridge latham , d. w. , mazeh , t. , stefanik , r. p. , mayor , m. , and burki , g. , 1989 , nature , 339 , 38 - 40 lin , d. n. c. and papaloizou , j. c. b. 1985 , in protostars and planets ii , black , d. c. & matthews , m. s. , univ . of arizona press , tucson , 981 - 1072 lin , d. n.
c. , bodenheimer , p. , and richardson , d. c. 1996 , nature , 380 , 606 - 607 lunine , j. i. 1986 in astrophysics of brown dwarfs , kaftos , m. c. , harrington , r. s. & maran , s. p. , cambridge univ .
press , cambridge , 170 - 176 marcy , g. w. , and butler , r. p. 1996
, ap.j.lett . , 464 , l151 - 152 mathieu , r. d. 1994 , ann .
astron . & astrophys . , 32 ,
465 - 530 mayor , m. and queloz , d. 1995 , nature , 378 , 355 - 359 mayor , m. , queloz , d. , udry , s. , and halbwachs , j - l 1996 , in bioastronomy 96 , iau colloquium # 161 , capri , italy mazeh , t. , goldberg , d. , duquennoy , a. , and mayor , m. 1992 , ap .
j. , 401 , 265 - 68 noyes , r. w. , jha , s. , korzennik , s. g. , krockenberger , m. , nisenson , p. , brown , t. , kennelly , e. , horner , s. 1997 , ap . j. lett .
, 483 , l111 - 14 saumon , d. , hubbard , w. b. , burrows , a. , guillot , t. , lunine , j. i. , and chabrier , g. 1996 , ap.j . , 460 , 993 - 1018 trilling , d. e. , benz , w. , guillot , t. , and lunine , j. i. 1996 , bull.a.a.s .
, 28 , 1113 trimble , v. 1990 , mnras , 242 , 79 - 87 trimble , v. and cheung , c. 1976 , in iau symposium 73 , eggleton , p. , mitton , s. , and whelan , j. , reidel , dordrecht , 369 weidenschilling , s. j. & marzari , f. 1996 , bull.a.a.s . , 28 , 1113 wolszczan , a. & frail , d. a. 1992 , nature , 355 , 145 | the difference in formation process between binary stars and planetary systems is reflected in their composition as well as their orbital architecture , particularly orbital eccentricity as a function of orbital period .
it is suggested here that this difference can be used as an observational criterion to distinguish between brown dwarfs and planets .
application of the orbital criterion suggests that with three possible exceptions , all of the recently - discovered substellar companions discovered to date may be brown dwarfs and not planets .
these criterion may be used as a guide for interpretation of the nature of sub - stellar mass companions to stars in the future . |
interactions between turbulent convection and magnetic field at the photospheric layers play central roles in structuring and driving varied forms of dynamical phenomena in the atmospheric layers above , and hence in the energetics ( see , _ e.g. _ @xcite and references therein ) .
these interactions in the near - surface layers are also basic to local dynamo action @xcite , which , if present , could explain the large amount of quiet - sun magnetic flux inferred from high - resolution observations @xcite .
an important aspect of these interactions is the role of helical or swirly fluid motions which can similarly twist or inject helicity to the magnetic field , and vice versa .
helicity of a vector field , in general , is defined as the volume integral of the scalar product of the field vector and its curl ( or rotation ) and it quantifies the amount of twistedness in the vector field @xcite . for fluid flow ,
the kinetic helicity , @xmath0 , is such a quantity derived from velocity @xmath1 and its curl or vorticity , @xmath2 : @xmath3 , where @xmath4 is the volume element . for magnetic fields
, helicity can be calculated in two different ways @xcite : the magnetic helicity , in general , is obtained by applying the above definition on the vector potential @xmath5 and its curl ( _ i.e. _ , the magnetic field @xmath6 ) , @xmath7 ; and use of magnetic field and its curl in the above definition gives the so called current helicity , @xmath8 .
both @xmath9 and @xmath10 are measures of twistedness in magnetic field , and they normally preserve signs over a volume of physical interest @xcite .
interactions between kinetic and magnetic helicities play fundamental roles in magnetic field generation ( or dynamo action ) as well as in the magneto - hydrodynamical evolution of the fluid and magnetic field @xcite
. helicities ( magnetic , and kinetic ) of solar active regions have been extensively studied using observations of magnetic and velocity fields in and around them : a well known property of the active region magnetic fields is the hemispheric sign rule , originally discovered by @xcite ( see also , @xcite ) : active regions in the northern hemisphere show a preferential negative magnetic helicity while those in the southern hemisphere show positive helicity .
the origin or exact cause of such pattern in large - scale magnetic helicity is still not fully understood .
dynamo mechanisms that impart helicity while the field is being generated as well as transfer of kinetic helicity from fluid motions to the magnetic field as it rises through the convection zone @xcite or during and after its emergence at the surface ( photosphere ) are thought to play roles in the observed pattern @xcite . on the small scale , away from the active and emerging flux regions , the magnetic and kinetic helicities and their interactions are even more poorly understood as measuring them reliably poses significant difficulties .
although significant advances have been made in mapping horizontal velocities through correlation tracking of surface features such as granulation or magnetic structures ( _ e.g. _ , see @xcite and references therein ) , significant uncertainties remain in measuring horizontal components of vector magnetic field at the small - scale @xcite and hence in estimating reliably the current or magnetic helicities there [ y. liu and a. norton private communication ; _ cf .
_ further discussion below . ] .
observational studies by @xcite have explored vertical vorticities associated with supergranular scale flows and such results have guided some theoretical studies of the relations between kinetic and magnetic helicities in the context of turbulent dynamo mechanisms @xcite .
apart from the above described aspects of interactions between fluid motions and magnetic field , recently , vortex motions around small - scale magnetic flux tubes and transfer of helicity from fluid motions to magnetic fields have been identified as key players in the upward energy transport and thus in the heating of solar corona @xcite . while @xcite find vortex flows of life - time of about an hour , numerical simulations of @xcite show no long - lived vortex flows in the solar photosphere .
detection of vortex flows at the granular scale in the photosphere date back to the studies by @xcite and @xcite , who infered that such motions could be common features in the granular and supergranular inflow regions .
a slightly excess correlation between negative divergence of the horizontal flows ( or inflows ) and vertical vorticity was found by @xcite .
@xcite detected a lot of small vortices in the inflow regions and found a clear association between them and magnetic bright points .
@xcite have detected vorticites in the inflow regions by calculating horizonatal flows using balltracking technique .
@xcite detected strong magnetic flux at the centers of the vortex flows .
vortical flow maps in the quiet - sun were calculated by @xcite and circular flow component of the inflows around active regions were calculated by @xcite using helioseismic ring - diagram technique .
a recent work has compared spatially resolved veritical vorticites calculated from two independent techniques , local correlation tracking and time - distance helioseismology @xcite . despite a good number of studies on vortex
flows themselves , there have not been detailed analyses of relationships between such fluid motions and magnetic field in the small - scale .
for example , there have not been reliable inferences on the connections between helicities of fluid motion and magnetic field , and on the back - reaction of magnetic field on the fluid .
much of the difficulties lie in reliably measuring the @xmath9 or @xmath10 of the small - scale magnetic fields as vector field measurements are often subject to large uncertainties outside of sunspots or active regions @xcite . for these reasons , there are no reliable measurements to ascertain if the helicity of small - scale magnetic fields follow the hemispheric sign pattern obeyed by active regions .
there are conflicting findings regarding the dominant signs of current helicity in the weak or small - scale fields over the hemispheres @xcite .
helioseismology results on supergranular - scale flows , however , show the effect of coriolis force introducing a hemispheric sign pattern in the vorticity ( and hence kinetic helicity ) of such flows @xcite .
in this work , we focus on examining how the magnetic field modifies the relationships among the fluid dynamical quantities , divergence , vorticity and kinetic helicity on the one hand , and how these quantities themselves scale against the strength of the magnetic field on the other .
such an analysis is facilitated by the continuous full - disk coverage of the sun in velocity ( doppler ) , granulation ( continuum intensity ) , and magnetic field provided by the helioseismic and magnetic imager ( hmi ) onboard the solar dynamics observatory ( sdo ; @xcite ) .
though the spatial resolution of about 1 arc - sec ( about 720 km ) provided by hmi enables us to track the granulation features in both the continuum and velocity images and to derive the horizontal flow field , it is not sufficient to resolve the sub - granular scale vortex flows that possibly surround the tiny magnetic flux concentrations .
hence , the vertical vorticity that we measure from hmi data through local correlation tracking ( lct ) of granular motions would have contributions mainly from vortical flows of the size of several granules .
since such flows are likely of preponderance at the supergranular boundaries and junctions , we , in our analyses here , are able to study also the effects of coriolis force @xcite and their influence on the relations between vortex motions and magnetic fields .
we discuss the data and the analysis methods in section 2 , results in section 3 , and discussions , conclusions and future studies in section 4 .
we have used the three major observables from the helioseismic and magnetic imager ( hmi ) onboard the solar dynamics observatory ( sdo ) : doppler velocities ( @xmath11 ) , continuum intensities ( @xmath12 ) , and line - of - sight magnetic fields @xmath13 ( hereafter , we denote @xmath13 simply as @xmath14 ) .
the basic datasets are cubes of above observables over about nineteen large regions , typically of size 30.7 @xmath15 30.7@xmath16 ( in heliographic degrees , or @xmath17 with 0.03@xmath16/pixel ) covering both northern and southern hemispheres in the latitude range @xmath1830and about @xmath1815 in longitude about the central meridian , tracked for 14 hours and remapped ( postel projected ) to a uniform pixel size of 0.5@xmath19 per pixel .
the total area covered by the nineteen regions on the solar surface is @xmath20 = @xmath21 , which is about 0.87 solar hemispheres .
the nineteen regions chosen are from eleven dates distributed over a period of years 2010 - 2012 .
on each of the eight dates , july 11 , august 3 , october 8 , and november 3 of 2010 , and february 8 , february 19 , july 3 , and october 2 of 2011 , we have two regions , one centered at 15n and the other centered at 15s , and with central longitudes within about 15 of the central meridian . on the dates may 8 , 2011 , june 21 , and july 2 , 2012 , three regions centered at 0 latitude and central meridian were included .
thus the included regions cover equal amount of northern and southern hemispherical areas . of the nineteen regions , fourteen are quiet - network regions chosen by examining the magnetograms for the presence of mixed - polarity network field well away from active regions .
the remaining five regions on the five dates , august 3 , 2010 , february 19 , july 3 , october 2 , 2011 , and july 2 , 2012 , however , have a sunspot in each ; data of the first four regions were available with us already and had been used in a different analysis published by the second author @xcite .
we included quiet areas of these regions by carefully excluding the sunspots and surrounding plages ( one example is shown figure 1 ) covering about 8 of the total area , and hence the total quiet - network area included in the analysis is about 0.8 solar hemispheres .
since this discarded area of about 0.07 solar hemispheres is all from the north , we have about 15 excess southern hemispherical area over that of north in the analyses here .
figure [ contourplot ] shows two examples from among the analysed regions : derived flow divergence and vorticity maps overlaid with magnetic field contours ( refer to the following section ) over a mixed - polarity quiet - network area ( left panels ) and over a region covering a sunspot and plages ( right panels ) .
the white - line boundaries in the right panel of figure [ contourplot ] separate the quiet - sun areas included in the analysis for this region , and these were chosen by visually examining the magnetograms to avoid sunspot and surrounding plages and to include only the quiet - sun network .
these straight - line boundaries are just for convenient and easy inclusion of the chosen areas in the analyses .
we apply the local correlation tracking ( lct ) technique @xcite on @xmath11 and @xmath12 to derive horizontal motions of convective granules .
we use the code flct @xcite that implements lct through measurement of correlation shifts in the fourier space .
flct is applied on two images separated by a time @xmath22 .
each image is multiplied by a gaussian of width @xmath23 centered at the pixel where velocity has to be derived .
cross - correlation is done within this gaussian window to calculate the shifts in x- and y - directions that maximise the correlation .
the shifts in x- and y - directions are divided by @xmath22 to get velocities in x- and y - directions .
we remove the @xmath24 ( surface gravity ) and @xmath25 mode oscillation signals in @xmath11 and @xmath12 before applying the lct to derive fluid motions .
this is done using a gaussian tapered fourier frequency filter that removes frequency components above @xmath26 .
the fwhm of gaussian window for lct is @xmath27 pixels and @xmath22 is about 2 minutes .
we apply flct at every time - step , _
i.e. _ every 45 seconds , to derive the horizontal velocity components @xmath28 and @xmath29 with the original resolution as the data . from these horizontal components of velocity ,
we calculate the z - component of the vorticity and the horizontal divergence as , @xmath30 @xmath31 calculation of kinetic helicity @xmath0 requires the vertical component of @xmath32 and its gradient , which we do not have .
we follow @xcite in deriving a proxy for kinetic helicity from the calculated vertical component of vorticity and horizontal divergence , @xmath33 this proxy for kinetic helicity is similar to the relative kinetic helicity , @xmath34 , used by @xcite @xmath35 in situations dominated by two - dimensional flows .
spatial maps of horizontal divergence , @xmath36 , the vertical vorticity @xmath37 , and the kinetic helicity @xmath0 ( hereafter we denote @xmath38 defined above simply as @xmath0 ) , derived at each time step as described in the previous section , form our basic fluid dynamical quantities . to improve the signal - to - noise of these measurements
, we use a running temporal average over about 4.5 minutes , _
i.e. _ average of six individual measurements ; this is found suitable as typical life - time of granules is about 5 - 7 minutes .
this running average is taken on the flows derived but not on the cross - correlations of lct to avoid missing any granular signals that have life - time smaller or close to the averaging interval .
an example map of full 14 hr temporal average of @xmath39 and @xmath40 with overlaid contours of similarly averaged @xmath13 is shown in figure [ contourplot ] .
the velocity maps @xmath41 and @xmath42 calculated from doppler observations show a systematic variation in x- and y - directions , and the magnitude of change across the spatial extent covered is about @xmath43 .
this is attributed to the shrinking sun effect @xcite that shows an apparent disk - centered ( or radially directed ) inflows .
origin of this constant flow signal ( i.e. time independent ) is not fully known , although it has been attributed to selection bias of lct method and to insufficient resolution of the instrument to resolve fully the granules on the solar surface @xcite .
whatever the origin , this constant disk - centered flow signal is easily determined by taking temporal averages ( of both @xmath41 and @xmath42 ) , spatillay smoothing and obtaining a low degree 2-d fit of it.so determined background artefact is then removed by subtracting it out from maps @xmath41 and @xmath42 at every time step . in this study , we examine ( 1 ) the hemispherical dependence of the signs of @xmath40 or @xmath0 arising from the coriolis force , ( 2 ) how the magnetic field modifies the relationship between @xmath44 and @xmath40 or @xmath0 , and ( 3 ) how these quantities themselves scale against the strength of the magnetic field .
since these quantities are highly dynamic with typical time scales of the order of granular life - time , we derive the relationship between these quantities at each time step of measurement .
this we do by determining , at each time - step , the dependence of @xmath40 ( or @xmath0 ) on @xmath44 and @xmath14 by calculating their average values over chosen small intervals ( bin sizes ) in @xmath14 and @xmath44 : 10 g leads to , when much of the magnetic field is vertically oriented on the surface , a systematic bias towards lower field strengths that are about @xmath45 times the true value for an angular distance of @xmath46 from the disk center . since the maximum off - center location does not exceed @xmath47 , we have at the most a 14 lower values for @xmath14 . however , average deviations of derived dependences on @xmath48 would be off from true values of @xmath14 by a much smaller amount than 14 . ]
bins in @xmath14 ( a magnetic bin ) and 20@xmath49 bins in @xmath44 .
this is implemented by sub - dividing every magnetic bin , _
i.e. _ pixels having magnetic field spread of 10 g , in terms of @xmath44 with a bin size 20@xmath49 .
we do this for both signed and absolute values of @xmath40 or @xmath0 , denoting them respectively as @xmath50 or @xmath51 and @xmath52 or @xmath53 ; in the former case of averages of signed values we perform the calculations for northern and southern hemispheric regions separately so as to check the hemispheric trends such as those introduced by coriolis force , and in the latter case of average of absolute values we combine both hemispheric regions together .
the fill - factor for magnetic field is small ( _ i.e. _ , much of the area is occupied by @xmath54 0 g pixels , refer to fig .
[ contourplot ] ) and , in general , area occupied by magnetic pixels goes down sharply as @xmath55 increases ( refer to the bottom panel of figure 3 , where a histogram of area ( in logarithmic scale ) against @xmath14 is plotted ) . in such situations a better statistic is provided by median values rather than averages .
we tested this by taking median values over the chosen bins of @xmath14 and @xmath44 and found , however , that the derived relationships are nearly the same for averages and medians .
the above analysis process is repeated for each region , and average @xmath50 , @xmath52 , @xmath51 and @xmath53 are determined from those for all the regions , making sure that the included areas contain only the quiet - sun magnetic network .
the resulting average relationships @xmath50 and @xmath51 determined for northern and southern hemispheric regions separately are shown in figures [ 2dplotic ] and [ 2dplotv ] : results based on lct velocities derived using hmi continuum intensities @xmath12 are in figure [ 2dplotic ] and that from doppler velocities are in figure [ 2dplotv ] , and they agree well .
it is seen that the doppler velocities yield a little less noisy results for @xmath50 and @xmath51 and hence we use these for further analyses in the following sections .
it is to be noted that the signed averages @xmath50 ( @xmath51 ) have cancellations and hence measure only the excess of one sign of rotation [ either clock - wise ( negative ) or counter - clockwise ( positive ) rotation ] over the other .
hence , vanishing of these averages do not necessarily mean absence of rotations , and this is easily checked by the averages of absolute values @xmath52 ( @xmath53 ) shown in figures [ cavsdivav ] and [ kavsdivav ] . as can clearly be seen , the excesses @xmath50 ( @xmath51 ) of one sign over the other are about one - tenth of @xmath52 ( @xmath53 ) .
further , we see that such cancellations are the largest for non - magnetic ( quiet ) flows .
the dominance of one sign of @xmath50 for flows around magnetised regions is due to the phenomenon of flux expulsion @xcite that leads to magnetic flux occupying preferentially the inflow ( negative @xmath44 ) locations .
we discuss this aspect further later in the next section .
results in figures [ 2dplotic ] and [ 2dplotv ] show that , to a large extent , the sign of @xmath40 ( @xmath0 ) is positive ( negative ) or counter - clockwise rotation in the northern hemisphere and negative ( positive ) or clockwise rotation in the southern hemisphere .
it is to be noted that , in the absence of coriolis and any other large scale force , the flows are expected to have a roughly equal distribution of clock - wise ( negative ) and anti - clockwise rotations , and hence a signed average of @xmath40 or @xmath0 over magnetic and divergence bins is expected to yield near - zero values due to cancellation among positive and negative vorticities .
however , as earlier studies have shown , supergranular scale flows are indeed subject to coriolis force , and as results in figures [ 2dplotic ] and [ 2dplotv ] show even the smaller - sized inflows ( negative @xmath44 ) show predominant hemispheric sign pattern consistent with the coriolis effect
. a closer look at this striking pattern requires understanding first the @xmath56 relationship for the non - magnetic flows .
quiet - sun areas devoid of significant magnetic field is captured in the central vertical area close to @xmath57 g in figures [ 2dplotic ] and [ 2dplotv ] .
we take the average of @xmath40 within @xmath58 g , _ i.e. _ a horizontal average over three magnetic bins of -10 , 0 , and 10 g in figure [ 2dplotv ] , to determine @xmath59 = @xmath60 for the non - magnetic flows .
it is to be noted that the observational errors in hmi los magnetograms are about 10 g @xcite and hence the pixels within @xmath58 g can be considered as non - magnetic .
the resulting variation of @xmath59 is shown as blue lines in figure [ cavsdivsv ] ( other colored lines in this figure show averages over different ranges of @xmath55 , and we discuss them in the next section ) . signed values of @xmath40 against @xmath44
are shown in the left panel of figure [ cavsdivsv ] , and a comparison of absolute amplitudes of @xmath40 at inflows and outflows is shown in the right panel by plotting them against absolute @xmath44 .
results plotted similary for @xmath0 are in figure [ kavsdivsv ] . in these figures , we show only the results obtained from lct velocities derived using hmi doppler velocities ( results obtained from hmi @xmath12 are very similar ) .
the variation of @xmath59 clearly brings out the effect of coriolis force on fluid flows on the supergranular scale , _ viz . _
a radial outflow ( positive @xmath44 ) rotates clock - wise ( negative @xmath40 ) and a inflow ( negative @xmath44 ) rotates counter - clock - wise ( positive @xmath40 ) in the northern hemisphere and vice versa in the southern hemisphere .
these results are in agreement with earlier known rotation properties of supergranular flows @xcite . further the magnitudes of @xmath61 derived as above increase more or less linearly against magnitudes of @xmath44 .
although the above result indicates that the predominant variation of @xmath59 in quiet - sun flows at the scales that we are measuring is due to coriolis effect , it is expected that not all size vortex flows have the signs following the coriolis effect .
especially , any smaller scale vortex flows at granular inflows or junctions within the supergranular cells can potentially be of larger magnitude and not influenced by the coriolis force .
this indeed turns out to be true as the averages of absolute values , @xmath52 ( @xmath53 ) , shown in figures [ cavsdivav ] and [ kavsdivav ] portray .
the top panels in these figures show the 2-d binned maps , while the bottom panels show the non - magnetic @xmath62 ( blue lines ) and averages over a few selected magnetic field ranges ( same as done in figures [ cavsdivsv ] and [ kavsdivsv ] ) against abs(@xmath44 ) for a comparison of outflows and inflows . in the analysis and plots here
, the error bars represent standard deviations within the ranges of magnetic bins used .
firstly , as noted earlier , magnitudes of vorticities ( determined through averages of absolute values ) are about ten times those of the excesses determined to be resulting from the hemispherical preference of one sign over the other due to the coriolis force ( comapre the values in figures [ cavsdivsv ] and [ kavsdivsv ] with those in figures [ cavsdivav ] and [ kavsdivav ] )
. magnitudes of vorticities estimated in our work here compare very well with earlier local helioseismic results from @xcite and with local helioseismic as well as lct analyses of @xcite .
secondly , in agreement with the earlier findings @xcite , the inflow regions have slightly excess vorticity ( and helicity ) as compared to outflow regions for non - magnetic flows ( blue curves in figures [ cavsdivav ] and [ kavsdivav ] ) .
we discuss the magnetic modifications in the following section . to study how the above described non - magnetic relationship @xmath59 is influenced by the magnetic field ,
we average @xmath50 and @xmath52 ( as well as @xmath51 and @xmath53 ) over three different ranges of magnetic field : 20 - 80 g ( denoted as 50 g ) , 150 - 250 g ( denoted as 200 g ) , and 250 - 350 g ( denoted as 300 g ) centered around 50 , 200 and 300 g. the results are shown in the figures [ cavsdivsv ] , [ kavsdivsv ] , [ cavsdivav ] and [ kavsdivav ] .
firstly , we see from figures [ 2dplotic ] and [ 2dplotv ] that the sign of @xmath50 over magnetized regions does not follow the dependence on @xmath44 expected from the action of coriolis force : while the non - magnetic ( @xmath6310 g ) central vertical regions of figures [ 2dplotic ] and [ 2dplotv ] show the sign change ( the blue curves in the left panel of figure [ cavsdivsv ] ) through @xmath44 = 0 . , the magnetised regions show predominantly one sign , _
i.e. _ positive in the north and negative in the south , which should hold only for inflows or converging flows ( negative @xmath44 ) if the coriolis effect is the cause .
however , magnetic fields over outflows or diverging flows ( positive @xmath44 ) too show the same sign as that seen over converging ( negative @xmath44 ) regions , although it is noted that the majority of the magnetic fluxes lie over the coverging or inflow ( negative @xmath44 ) regions .
it is clear that the dominantly inflow located magnetic fields are connected to those located in the interiors of supergranules ( with positive @xmath44 ) , and that such connectivity transfers vorticity from the inflow ( negative @xmath44 ) regions to fluid surrounding such connected fields within the supergranules ( positive @xmath44 ) .
this magnetic connectivity is expected in the predominantly mixed - polarity flux , which can have a large horizontal component or loops arching above the photosphere .
this transfer can happen in either direction , but it is seen in figures [ 2dplotic ] and [ 2dplotv ] that tranfer of vorticity ( and kinetic helicity ) is predominantly from inflow regions ( negative @xmath44 ) to outflows or diverging flows ( positive @xmath44 ) ; this is consistent with the fact that stronger magnetic fields with larger fluxes are found mainly over the inflow regions .
the cross - sections over three different magnetic field ranges plotted in figures [ cavsdivsv ] , [ kavsdivsv ] , [ cavsdivav ] and [ kavsdivav ] show the above described signatures of transfer of vorticity from inflow to outflow regions more clearly , and also show how the strengths of the magnetic field influence this phenomenon .
we add a caveat here to the above inferences on possible transfer of vorticity from supergranular inflow ( negative @xmath44 ) to outflow ( positive @xmath44 ) regions : the rearrangement of magnetic flux during the process of evolution of supergranules , viz .
dying " inflow that is replaced by the outflow of a new supergranule , can lead to our analyses finding vorticity in outflow regions that matches the vorticity in inflow regions ( as described above ) .
this can arise because the newly formed outflow ( in the place of old inflow ) has not had time to sweep the flux to the new inlfow region and the coriolis force too has not had time to act and hence reverse the vorticity .
however , we note that the averaging over 14 hours , which is more than about half the life time of a supergranule and hence is long enough for both the above processes to establish , should tend to smooth out such signatures of dying and newly forming supergranular flows .
nevertheless , a residual signature of this process is certainly possible in our analyses .
in addition , we find that these differences between inflow and outflow vorticities ( and helicities ) are much more pronounced for flows around magnetic fields refer to the black ( 50 g ) , red ( 200 g ) and pink ( 300 g ) curves in figures [ cavsdivav ] and [ kavsdivav ] .
it is to be noted however that these enhanced excesses of inflow vorticities , both in signed averages ( figures [ cavsdivsv ] and [ kavsdivsv ] ) as well as in averages of absolute values ( figures [ cavsdivav ] and [ kavsdivav ] ) , also depend on the fact that much of the magnetic fields , especially stronger concentrations , are located in inflow regions .
this is clearly seen in the 2d maps in figures [ 2dplotic ] and [ 2dplotv ] and in the top panels of figures [ cavsdivav ] and [ kavsdivav ] , which show the large asymmetry in the areas occupied by non - zero vorticities in the @xmath64 region : relatively weaker fields are found in the supergranular outflows ( positive @xmath44 ) , and majority of the magnetic flux and stronger field concentrations are in the inflow ( negative @xmath44 ) regions .
a basic feature in all the above magnetic effects is that the linear relationship characterising @xmath59 has given way to more complicated variations .
we present and describe the dependences on magnetic field strength further in the following section . to examine the influence of magnetic field strengths on the sign and amplitudes of @xmath40 and @xmath0 ( figure [ 2dplotv ] ) , and on the absolute magnitudes of @xmath65 , and @xmath66 ( top panels of figures [ cavsdivav ] and [ kavsdivav ] ) , we take averages of these quantities over the divergence bins .
these results are shown in figures [ ca1vsbv ] and [ cavsbv ] .
as noted earlier , at @xmath55=0 g , the opposite contributions to @xmath40 and @xmath0 from the positive ( outflow ) and negative ( inflow ) @xmath44 regions cancel out yielding near - zero values for these quantities .
however , averages of absolute values @xmath65 and @xmath66 show that the largest magnitude vorticities are to be found over non - magnetic flows , and they decline sharply as @xmath55 increases .
the excesses @xmath67 and @xmath68 , however , show a rapid increase at low magnetic field strengths , reaching a maximum between about 150 and 200 g. the sign pattern of magnetic field correlated @xmath40 and @xmath0 preserves that expected from coriolis force action on inflows ( negative @xmath44 regions ) . taken together ,
these results show that a major contribution to larger magnitudes of vorticities and kinetic helicities in inflow regions , as compared to those over diverging flows in the interiors of supergranules , leading to the hemispheric excesses seen in figure [ ca1vsbv ] are due to the predominance of magnetic fields at inflow locations .
it is also possible that the supergranular inflows have more vigorous swirls around them due to larger thermal perturbations that strong magnetic flux concentrations cause in the solar photospheric layers : cooler magnetic field trapped photospheric gas acts as channels for further heat ( radiative ) loss vertically outwards accelerating the flows surrounding hotter plasma towards them , thus enabling stronger inflows and vorticities . as magnetic field increases further beyond equipartition strength , which is typically about 300 - 400 g for the solar photosphere , the back - reaction of the magnetic field via lorentz force , involving both the pressure and tension forces , acts to inhibit the flows .
this effect eventually becomes dominant and acts to compensate the contributions from the coriolis force assisted combinations of inflows and magnetic fields leading to the decreases seeen in figure [ ca1vsbv ] .
as to the averages of absolute magnitudes @xmath65 and @xmath66 shown in figure [ cavsbv ] , it is clear that , when we ignore the outflow - inflow asymmetry in the magnetic field locations and the resulting influence of coriolis force , the net effect of magnetic forces is to suppress vortical motions .
there might also be contributions due to the fact that field strengths @xmath14 here ( from hmi ) are flux densities within the resolution element of the instrument , and hence increase of @xmath55 here may mainly reflect the increase of flux while the actual field strengths being already in strong saturated super - equipartition values . that the photospheric small - scale network field concentrations outlining supergranular inflows are predominantly of super - equipartition kg field strengths is a well established fact .
hence , the results in figure [ cavsbv ] for vorticities and kinetic helicities may correspond to the so - called highly @xmath69-quenched state , if photospheric flows are considered for a ( helical)-turbulent local dynamo ( refer to , _ e.g. _ , @xcite and references therein ) . in non - linear mean - field dynamo theory ,
both the hydrodynamic ( @xmath70 ) and magnetic ( @xmath71 ) parts of the the @xmath69-effect are coupled via the magnetic field , and explicit algebraic expressions for the dependences @xmath72 and @xmath73 are available @xcite .
the @xmath70 is proportional to @xmath66 shown in figure [ cavsbv ] .
however , we note that theoretical quenching functions that determine @xmath72 are in terms of mean magnetic field @xmath14 that develops out of dynamo action at the small - scales , whereas in our observations and analysis here @xmath14 refers to the field at the small - scale
. a more detailed look at @xmath74 or @xmath72 , their spectra , both spatial and temporal , and their relations to various other local and global properties of solar magnetic fields in the context of dynamo mechanisms is beyond the scope of work reported here .
moreover , higher spatial resolution to resolve sub - granular scale flows and magnetic fields than provided by hmi is important to address such details .
we have derived and analysed horizontal fluid motions on the solar surface over large areas covering the quiet - sun magnetic network using local correlation tracking of convective granules imaged in continuum intensity and doppler velocity by the hmi onboard sdo .
we have studied the relationships between fluid divergence and vorticity , and that between vorticity ( kinetic helicity ) and magnetic field .
these relationships are derived through both signed and absolute averages of @xmath40 and @xmath0 over time and space .
these latter dependences are studied through fluid divergence @xmath75 and magnetic field @xmath76 , yielding @xmath50 , @xmath51 , @xmath52 , and @xmath53 ( figures [ 2dplotic ] , [ 2dplotv ] , [ cavsdivsv ] , and [ kavsdivsv ] ) .
the main results obtained can be summarised as below : + ( 1 ) the correlations between vorticity and divergence of non - magnetic flows at the supergranular scale have the dominant hemispheric sign pattern brought about by the action of the coriolis force .
the rotations of outflows and inflows are roughly of equal magnitude and of opposite sign , although there is a slight excess of inflow vorticities which increase as magnitude of divergence increases .
further , the non - magnetic vorticities scale linearly with the divergences . + ( 2 ) for magnetised flows , the sign pattern corresponding to the inflows dominate over the whole divergence field , _ i.e. _ even the outflow regions ( positive divergence , interiors of supergranules ) exhibit rotations expected of the action of coriolis force acting on inflows . in other words , for magnetised flows the dominant sign pattern is negative helicity @xmath0 ( positive @xmath40 ) in the north and positive helicity @xmath0 ( negative @xmath40 ) in the south . we have identified and attributed this to the mechanism of transfer of vorticity from network inflow regions to interiors of supergranules by the magnetic connectivity or by horizontal component of magnetic field , with a caveat that supergranular evolution ( dying " inflow that is replaced by the outflow of a new supergranule ) too can contribute to the observed signatures .
+ ( 3 ) the excess of inflow vorticities over those of outflows increases dramatically for magnetised flows .
this has been identified as due to the preferential inflow locations of magnetic fields resulting from the convective flux expulsion mechanism .
contributions of instrinsically larger rotations of inflows around stronger fields there due to thermal causes , however , can also play a role in the above asymmetries between outflow and inflow rotations . + ( 4 ) in terms of absolute magnitudes of vorticities , it has been found that the non - magnetic flows have the largest values . as a function of magnetic field strengths , as observed by hmi , magnitudes of absolute vorticities decrease almost exponentially bringing out the magnetic suppression of flows due to lorentz forces .
in particular , we find that the magnetic fields largely suppress the amplitudes of vortical motions when magnetic flux densities exceed about 300 g ( hmi )
. this magnetic suppression of vorticities or helicities is identified as that arising from the @xmath69-quenching action of magnetic field .
this identification is found reasonable as much of the magnetic flux in the solar photosphere is known to be of kg strengths , which are well above the equipartition values .
the above results have wider implications and relations to several other related phenomena pertaining to magneto - convective processes in the solar photosphere .
the results ( 1 ) and ( 2 ) above related to the hemispheric pattern of fluid kinetic helicity and its relationship to magnetic field are of importance to understanding transfer of helicities between the fluid and magnetic field , especially in situations where reliable measurements of magnetic helicities in the small - scale are hard to come by .
since we have not measured magnetic helicities , we can not establish the connections or transfer of helicities between fluid motions and the magnetic field .
however , if transfer of helicities happen between fluid motions and magnetic field , irrespective of which direction it happens , we expect both to have the same sign . in our analysis
, we find kinetic helicity is negative ( positive ) in northern ( southern ) hemisphere and this is consistent with magnetic helicity followed by active regions . on the other hand ,
if the small - scale fields have opposite sign of current helicity as compared to active regions as reported by @xcite , then it could not have gone from fluid motions as this is not consistent with our observation .
thus , if the magnetic helicity of small - scale magnetic field all over the sun has contributions from kinetic helicity , then we expect it to have the same sign as active regions .
hence , our results indicate that small - scale magnetic helicity would indeed be of same sign as that of active regions that follow the usual hemispheric trend .
transfer and redistribution of vorticities by magnetic fields implied by results ( 2 ) and ( 3 ) exemplify the basic magnetohydrodynamic effect known from early laboratory experiments involving conducting liquid metals performed and expounded by @xcite .
these same laboratory experiments also demonstrate suppression of swirly motions by magnetic field , when the field and fluid velocity directions are not aligned .
our result ( 4 ) here is an example for this phenomenon on the sun .
as regards the relevance of our results for understanding the nature of fluid turbulence and associated dynamo - sustaining state of the near - surface convection @xcite , we note that the dominance of negative kinetic helicity in the north , and hence the expected same sign for magnetic helicity , indicate a positive @xmath69-effect in the north in contrast to that required for explaining the observed butterfly diagram of large - scale active regions . as referred to earlier , more detailed analyses of distribution of kinetic helicity , its interactions with magnetic field , its spatio - temporal spectra and their relations to various other local and global properties of solar magnetic fields are necessary to make progress towards devising observational diagnostics of possible dynamo actions happening in the near - surface layers .
we note that continuous wide field - of - view observations of higher spatial resolution to resolve sub - granular scale flows and magnetic fields than provided by sdo / hmi is important to address such details .
we believe the kind of analyses undertaken here , _
e.g. _ such as those derived from the results in figure [ 2dplotv ] , as applied to magnetic regions in different dynamical state such as emerging flux regions , decaying active regions , and those before , during and after major atmospheric activity such as eruptions or flares , would aid in understanding the interactions between fluid and magnetic helicities , their evolution , exchange and transport upwards in the atmosphere .
+ data - intensive numerical computations required in this work were carried out using the high performance computing facility of the indian institute of astrophysics , bangalore .
we thank ravindra , b. for help with running the flct code .
this work has utilised extensively the hmi / sdo data pipeline at the joint science operations center ( jsoc ) , stanford university .
we thank the jsoc team at stanford solar observatories group .
our thanks are also due to an anonymous referee for a number of constructive comments and suggestions that improved the discussion of results and the presentation in this paper . | we derive horizontal fluid motions on the solar surface over large areas covering the quiet - sun magnetic network from local correlation tracking of convective granules imaged in continuum intensity and doppler velocity by the helioseismic and magnetic imager ( hmi ) onboard the solar dynamics observatory ( sdo ) . from these
we calculate horizontal divergence , vertical component of vorticity , and kinetic helicity of fluid motions .
we study the correlations between fluid divergence and vorticity , and that between vorticity ( kinetic helicity ) and magnetic field .
we find that the vorticity ( kinetic helicity ) around small - scale fields exhibits a hemispherical pattern ( in sign ) similar to that followed by the magnetic helicity of large - scale active regions ( containing sunspots ) .
we identify this pattern to be a result of the coriolis force acting on supergranular - scale flows ( both the outflows and inflows ) , and is consistent with earlier studies using local helioseismology .
further , we show that the magnetic fields cause transfer of vorticity from supergranular inflow regions to outflow regions , and that they tend to suppress the vortical motions around them when magnetic flux densities exceed about 300 g ( hmi ) .
we also show that such action of magnetic fields leads to marked changes in the correlations between fluid divergence and vorticity .
these results are speculated to be of importance to local dynamo action if present , and to the dynamical evolution of magnetic helicity at the small - scale . |
over the last decades strangeness has been a matter of extensive study in connection to exotic atoms @xcite as well as heavy - ion collisions at sis / gsi energies @xcite .
phenomenology of antikaonic atoms shows that the @xmath2 feels an attractive potential at low densities .
this attraction results from the modified @xmath3-wave @xmath4 resonance in the medium due to pauli blocking effects @xcite together with the self - consistent consideration of the @xmath2 self - energy @xcite and the inclusion of self - energies of the mesons and baryons in the intermediate states @xcite
. attraction of the order of -50 mev at normal nuclear matter density , @xmath5 , is obtained by different approaches , such as unitarizated theories in coupled channels based on meson - exchange models @xcite or chiral dynamics @xcite .
higher - partial waves beyond the s - wave contribution have been also studied @xcite as they become relevant for heavy - ion collisions at beam energies below 2agev @xcite .
also the charm degree of freedom is a recent topic of analysis in heavy - ion experiments .
the cbm experiment of the future fair project at gsi will investigate highly compressed dense matter in nuclear collisions with a beam energy range between 10 and 40 gev / u .
an important part of the hadron physics project is devoted to extend the sis / gsi program for the in - medium modification of hadrons to the heavy quark sector providing a first insight into charm - nucleus interaction .
thus , the possible modifications of the properties of open and hidden charm mesons in a hot and dense environment are matter of recent studies .
the in - medium modification of the open charm mesons ( @xmath6 and @xmath7 may help to explain the @xmath8 suppression in a hadronic environment as well as the possible formation of @xmath6-mesic nuclei .
moreover , changes in the properties of open charm mesons will affect the renormalization of charm and hidden charm scalar meson resonances in nuclear matter , providing information about their nature , whether they are @xmath9 states , molecules , mixtures of @xmath9 with meson - meson components , or dynamically generated resonances resulting from the interaction of two pseudoscalars . in the present article , we present a study of the properties of strange and charm mesons in hot and dense matter within a self - consistent approach in coupled channels for the conditions expected at cbm / fair .
we analyze the behaviour of dynamically generated baryonic resonances as well as the strange and charm meson spectral functions in this hot and dense medium .
we then test our results for strange mesons using energy - weighted sum rules and analyze the effect of the self - energy of @xmath6 mesons on dynamically - generated charm and hidden charm scalar resonances .
the self - energy and , hence , the spectral function at finite temperature for strange ( @xmath2 and @xmath10 ) and charm ( @xmath6 and @xmath11 ) mesons are obtained following a self - consistent coupled - channel procedure .
we start by solving the bethe - salpeter equation in coupled channels or @xmath12-matrix ( @xmath12 ) taking , as bare interaction , a transition potential coming from effective lagrangians .
details about this bare interaction for strange and charm mesons are given in the next sections .
the self - energy is then obtained summing the transition amplitude @xmath12 for the different isospins over the nucleon fermi distribution at a given temperature , @xmath13 , as @xmath14\ , \label{eq : selfd}\end{aligned}\ ] ] where @xmath15 and @xmath16 are the total energy and momentum of the meson - nucleon pair in the nuclear matter rest frame , and ( @xmath17,@xmath18 ) and ( @xmath19,@xmath20 ) stand for the energy and momentum of the meson and nucleon , respectively , also in this frame .
the self - energy must be determined self - consistently since it is obtained from the in - medium amplitude @xmath12 which contains the meson - baryon loop function , and this last quantity itself is a function of the self - energy .
the meson spectral function then reads @xmath21
the kaon self - energies in symmetric nuclear matter at finite temperature are obtained from the in - medium kaon - nucleon interaction within a chiral unitary approach .
the model incorporates the @xmath3- and @xmath22-waves of the kaon - nucleon interaction @xcite . at tree level
, the @xmath3-wave amplitude arises from the weinberg - tomozawa ( wt ) term of the chiral lagrangian .
unitarization in coupled channels is imposed by solving the bethe - salpeter equation with on - shell amplitudes and a cutoff regularization .
the unitarized @xmath23 amplitude generates dynamically the @xmath4 resonance in the @xmath24 channel and provides a satisfactory description of low - energy scattering observables .
the in - medium solution of the @xmath3-wave amplitude accounts for pauli - blocking effects , mean - field binding on the nucleons and hyperons via a @xmath25 model , and the dressing of the pion and kaon propagators via their corresponding self - energies in a self - consistent manner .
the model includes , in addition , a @xmath22-wave contribution to the self - energy from hyperon - hole ( @xmath26 ) excitations , where @xmath27 stands for @xmath28 , @xmath29 and @xmath30 components .
self - energy and spectral function for @xmath31 , @xmath32 mev and two momenta showing the partial wave decomposition .
right : evolution of the @xmath2 spectral function with density and temperature for two momenta.,title="fig:",width=6,height=7 ] self - energy and spectral function for @xmath31 , @xmath32 mev and two momenta showing the partial wave decomposition . right : evolution of the @xmath2 spectral function with density and temperature for two momenta.,title="fig:",width=6,height=7 ] on the l.h.s . of fig .
[ fig1 ] we show the real and imaginary parts of the @xmath2 self - energy together with the @xmath2 spectral function as a function of the @xmath2 energy at @xmath33 for two different momenta , @xmath34 mev / c ( left column ) and @xmath35 mev / c ( right column ) .
the different curves correspond to @xmath36 and @xmath37 mev including the @xmath3-wave and the @xmath38-wave contributions .
the self - energy is dominated by the @xmath3-wave dynamics and the @xmath22-wave contributions from @xmath39 , @xmath40 and @xmath41 excitations become evident at a finite momentum of @xmath35 mev / c .
the effect of these subthreshold excitations is repulsive at the @xmath23 threshold .
this repulsion together with the strength below threshold can be easily seen in the spectral function at finite momentum ( third row ) . at finite momentum
the quasi - particle peak moves to higher energies while the spectral function falls off slowly on the left - hand side .
temperature results in a softening of the real and imaginary part of the self - energy as the fermi surface is smeared out .
the peak of the spectral function moves closer to the free position while the spectral function extends over a wider range of energies .
the evolution with density and temperature of the @xmath42 spectral function is depicted on the r.h.s . of fig .
the spectral function shows a strong mixing between the quasi - particle peak and the @xmath43 and @xmath44 excitations . as we have seen before , the effect of the @xmath22-wave @xmath44 subthreshold excitations is repulsive for the @xmath2 potential , compensating in part the attraction from the @xmath3-wave @xmath45 interaction .
temperature softens the @xmath22-wave contributions to the spectral function at the quasi - particle energy .
moreover , together with the @xmath3-wave mechanisms , the @xmath22-wave self - energy provides a low - energy tail which spreads the spectral function considerably .
increasing the density dilutes the spectral function even further .
the hadron propagator or single particle green s function has well defined analytical properties that impose some constraints on the many - body formalism as well as the interaction model .
an excellent tool to test the quality of our model for hadrons in medium is provided by the energy - weighted sum rules ( ewsrs ) of the single - particle spectral functions .
the ewsrs are obtained from matching the dyson form of the meson propagator with its spectral lehmann representation at low and high energies @xcite .
the first ewsrs in the high - energy limit expansion , @xmath46 , together with the zero energy one , @xmath47 , are given by @xmath48 = \frac{1}{\omega_{\bar k}^2(\vec{q}\ , ) + \pi_{\bar k}(0,\vec{q}\,;\rho , t ) } \ \ \ \ \\ m_{0}^{(\mp ) } & : & \int_0^{\infty } \textrm{d}\omega \ , [ s_{\bar k}(\omega,\vec{q}\;\rho , t ) - s_{k}(\omega,\vec{q}\,;\rho , t ) ] = 0 \nonumber \\ & & \int_0^{\infty } \textrm{d}\omega \ , \omega \ , [ s_{\bar k}(\omega,\vec{q}\,;\rho , t ) + s_{k}(\omega,\vec{q}\,;\rho , t ) ] = 1 \ .
% \nonumber \\ % \\ % m_{1}^{(\mp ) } & : & % \int_0^{\infty } \textrm{d}\omega \ , % \omega^2 \ , [ s_{\bar k}(\omega,\vec{q}\,;\rho , t ) - s_{k}(\omega,\vec{q}\,;\rho , t ) ] = 0 % \nonumber \\ % & & \ \ \ % \int_0^{\infty } \textrm{d}\omega \ , % \omega^3 \ , [ s_{\bar k}(\omega,\vec{q}\,;\rho , t ) + s_{k}(\omega,\vec{q}\,;\rho , t ) ] % = % \omega_{k}^2(\vec{q}\ , ) + \pi_{\bar k}^{\infty}(\vec{q}\,;\rho , t ) \ .
% \nonumber \\\end{aligned}\ ] ] , @xmath49 and @xmath50 sum rules for the @xmath10 and @xmath2 spectral functions at @xmath51 mev / c , @xmath31 and @xmath36 mev .
the @xmath2 and @xmath10 spectral functions are also displayed for reference in arbitrary units.,width=6,height=7 ] the sum rules for the antikaon propagator are shown in fig .
[ fig2 ] as a function of the upper integral limit in the case of @xmath31 , @xmath52 and @xmath53mev / c .
the contributions from @xmath2 and @xmath10 to the l.h.s . of the sum rule are depicted separately .
the @xmath2 and @xmath10 spectral functions are also shown for reference in arbitrary units .
note that saturation is progressively shifted to higher energies as we examine sum rules involving higher order weights in energy . the l.h.s
. of the @xmath47 sum rule ( upper panel ) converges properly and saturates a few hundred mev beyond the quasiparticle peak , following the behaviour of the @xmath2 and @xmath10 spectral functions .
we have also plotted in fig .
[ fig2 ] the r.h.s . of the @xmath47 sum rule both for the antikaon and kaon ,
namely their off - shell propagators evaluated at zero energy ( modulo a minus sign ) .
the difference between both values reflects the violation of crossing symmetry present in the chiral model employed for the kaon and antikaon self - energies as we neglect the explicit exchange of a meson - baryon pair in a @xmath54-channel configuration .
however , we may still expect the saturated value of the l.h.s . of the @xmath47 sum - rule to provide a constraint for the value of the zero - mode propagator appearing on the r.h.s
, because the most of the strength sets in at energies of the order of the meson mass , where the neglected terms of the @xmath55 amplitudes are irrelevant .
the @xmath56 sum rule shows that the areas subtended by the @xmath10 and @xmath57 spectral functions should coincide .
this is indeed the case for the calculation considered here , as can be seen in the middle panel of fig .
the fullfilment of this sum rule is , however , far from trivial .
we recall that whereas one expects the @xmath2 and @xmath10 spectral functions to be related by the retardation property , @xmath58 , the actual calculation of the meson self - energies is done exclusively for positive meson energies .
the @xmath59 sum rule saturates to one independently of the meson momentum , nuclear density or temperature , thus posing a strong constraint on the accuracy of the calculations .
the lower panel in fig .
[ fig2 ] shows that the @xmath10 and @xmath2 spectral functions fulfill this sum rule to a high precision .
we have also tested those sum rules for higher momenta and temperature . as the meson momentum is increased ,
the saturation of the integral part of the sum rules is progressively shifted to higher energies , following the strength of the spectral distribution . at finite temperature
the @xmath2 spectral function spreads considerably @xcite , and in particular acquires a sizable low energy tail from smearing of the fermi surface , which contributes substantially to the l.h.s . of the sum rule below the quasi - particle peak .
the @xmath10 contribution also softens at finite temperature and increasing momenta , as the @xmath10 in - medium decay width is driven by the @xmath60 thermal phase space .
the properties of charm mesons is a topic of recent analysis and lot of effort is being invested in constructing effective models for the meson - baryon interaction in the charm sector . in this section
we present two different approaches to this effective meson - baryon interaction .
we then discuss possible experimental scenarios where the charm meson properties can be tested , such as scalar resonances in nuclear matter .
the @xmath6 and @xmath11 meson spectral functions are obtained from the multichannel bethe - salpeter equation taking , as bare interaction , a type of broken @xmath61 @xmath3-wave wt interaction supplemented by an attractive isoscalar - scalar term and using a cutoff regularization scheme .
this cutoff is fixed by the position and the width of the @xmath24 @xmath62 resonance . as a result , a new resonance in @xmath63 channel @xmath64
is generated @xcite .
the in - medium solution at finite temperature incorporates , as well , pauli blocking effects , baryon mean - field bindings and @xmath65 and @xmath6 meson self - energies @xcite .
the @xmath24 @xmath66 and @xmath63 @xmath67 resonances in hot dense matter are shown in the l.h.s .
[ fig3 ] for two different self - consistent calculations : i ) including only the self - consistent dressing of the @xmath6 meson , ii ) including mean - field binding on baryons and the pion self - energy .
medium effects at @xmath36 lower the position of the @xmath66 and @xmath67 with respect to their free values .
their width values , which increase due to @xmath68 processes , differ according to the phase space available .
the pion dressing induces a small effect in the resonances because of charm - exchange channels being suppressed .
finite temperature results in the reduction of the pauli blocking due to the smearing of the fermi surface .
both resonances move up in energy closer to their free position while they are smoothed out , as in @xcite . and @xmath67 resonances .
right : the @xmath6 meson spectral function , title="fig:",width=6,height=7 ] and @xmath67 resonances .
right : the @xmath6 meson spectral function , title="fig:",width=6,height=7 ] in the r.h.s of fig .
[ fig3 ] we display the evolution with density and temperature of the @xmath6 meson spectral function for ( ii ) . at @xmath36
the spectral function presents two peaks : @xmath69 excitation at a lower energy whereas the second one at higher energy is the quasi(d)-particle peak mixed with the @xmath70 state .
those structures dilute with increasing temperature while the quasiparticle peak gets closer to its free value becoming narrower , as the self - energy receives contributions from higher momentum @xmath71 pairs where the interaction is weaker .
finite density results in a broadening of the spectral function because of the increased decay and collisional phase space .
similar effects were observed previously for the @xmath2 in hot dense nuclear matter @xcite . and
@xmath72 spectral functions in nuclear matter at @xmath34 mev / c.,title="fig:",width=6,height=7 ] and @xmath72 spectral functions in nuclear matter at @xmath34 mev / c.,title="fig:",width=6,height=6 ] heavy - quark symmetry ( hqs ) is a proper qcd spin - flavor symmetry that appears when the quark masses , such as the charm mass , become larger than the typical confinement scale . as a consequence of this symmetry ,
the spin interactions vanish for infinitely massive quarks .
thus , heavy hadrons come in doublets ( if the spin of the light degrees of freedom is not zero ) , which are degenerated in the infinite quark - mass limit . and
this is the case for the @xmath6 meson and its vector partner , the @xmath72 meson .
therefore , we calculate the self - energy and , hence , the spectral function of the @xmath6 and @xmath72 mesons in nuclear matter simultaneously from a self - consistent calculation in coupled channels . to incorporate hqs to the meson - baryon interaction
we extend the wt meson - baryon lagrangian to the @xmath73 spin - flavor symmetry group as we include pseudoscalars and vector mesons together with @xmath74 and @xmath75 baryons @xcite , following the steps for @xmath76 of ref.@xcite .
however , the @xmath73 spin - flavor is strongly broken in nature .
on one hand , we take into account mass breaking effects by adopting the physical hadron masses in the tree level interactions and in the evaluation of the kinematical thresholds of different channels , as done in @xmath61 models . on the other hand , we consider the difference between the weak non - charmed and charmed pseudoscalar and vector meson decay constants . we also improve on the regularization scheme in nuclear matter going beyond the usual cutoff scheme @xcite .
the @xmath73 model generates a wider spectrum of resonances with charm @xmath77 and strangeness @xmath78 content compared to the previous @xmath61 models , as seen in the l.h.s of fig .
while the parameters of both @xmath61 and @xmath73 models are fixed by the ( @xmath24,@xmath79 ) @xmath80 resonance , the incorporation of vectors mesons in the @xmath73 scheme generates naturally @xmath81 resonances , such as @xmath82 , @xmath83 , @xmath84 and @xmath85 , which might be identified experimentally @xcite .
new resonances are also produced for @xmath79 , as @xmath86 and @xmath87 , while others are not observed in @xmath61 models due to the different symmetry breaking pattern used in both models .
the modifications of the mass and width of these resonances in the nuclear medium will strongly depend on the coupling to channels with @xmath6 , @xmath72 and nucleon content . moreover
, the resonances close to the @xmath71 or @xmath88 thresholds change their properties more evidently as compared to those far offshell .
the improvement in the regularization / renormalization procedure of the intermediate propagators in the nuclear medium beyond the usual cutoff method has also an important effect on the in - medium changes of the dynamically - generated resonances , in particular , for those lying far offshell from their dominant channel , as the case of the @xmath80 .
( left ) and @xmath1 ( right ) resonances.,title="fig:",width=6,height=6 ] ( left ) and @xmath1 ( right ) resonances.,title="fig:",width=6,height=6 ] in the r.h.s of fig .
[ fig4 ] we also display the @xmath6 and @xmath72 spectral functions , which show then a rich spectrum of resonant - hole states .
the @xmath6 meson quasiparticle peak mixes strongly with @xmath89 and @xmath90 states while the @xmath91 is clearly visible in the low - energy tail .
the @xmath72 spectral function incorporates the @xmath81 resonances , and the quasiparticle peak fully mixes with @xmath92 and @xmath93 . as density increases , these @xmath94 modes tend to smear out and the spectral functions broaden as the collisional and absorption processes increase .
the nature of a resonance , whether it has the usual @xmath95/@xmath96 structure or is better described as being dynamically generated , is an active matter of research , in particular , for scalar resonances .
the excitation mechanisms in the nucleus together with the properties of those particles can be extracted studying their renormalized properties in nuclear matter .
we study the charmed resonance @xmath0 @xcite together with a hidden charm scalar meson , @xmath1 , predicted in @xcite , which might have been observed by the belle collaboration @xcite via the reanalysis of @xcite .
those resonances are generated dynamically solving the coupled - channel bethe - salpeter equation for two pseudoscalars @xcite .
the kernel is derived from a @xmath61 extension of the @xmath97 chiral lagrangian used to generate scalar resonances in the light sector .
the @xmath61 symmetry is , however , strongly broken , mostly due to the explicit consideration of the masses of the vector mesons exchanged between pseudoscalars @xcite .
the analysis of the transition amplitude close to each resonance for the different coupled channels gives us information about the coupling of the resonance to a particular channel .
the @xmath0 mainly couples to the @xmath98 system , while the hidden charm state @xmath1 couples most strongly to @xmath99 .
therefore , any change in the @xmath6 meson properties in nuclear matter will have an important effect on these resonances .
those modifications are given by the @xmath6 meson self - energy , as discussed in sec .
[ su4 ] , but supplemented by the @xmath22-wave self - energy through the corresponding @xmath94 excitations @xcite . in fig .
[ fig5 ] , the resonances @xmath0 and @xmath1 are shown by displaying the squared transition amplitude for the corresponding dominant channel at different densities .
the @xmath0 and @xmath1 resonances , which have a zero and small width , respectively , develop widths of the order of 100 and 200 mev at normal nuclear matter density , correspondingly .
the origin can be traced back to the opening of new many - body decay channels , as the @xmath6 meson gets absorbed in the nuclear medium via @xmath71 and @xmath100 inelastic reactions . in our model
, we do not extract any clear conclusion for the mass shift .
we suggest to look at transparency ratios to investigate those in - medium widths .
this magnitude , which gives the survival probability in production reactions in nuclei , is very sensitive to the absorption rate of any resonance inside nuclei , i.e. , to its in - medium width .
in this paper we have studied the properties of strange and charm mesons in hot and dense matter within a self - consistent coupled - channel approach .
the in - medium solution at finite temperature accounts for pauli blocking effects , mean - field binding on all the baryons involved , and meson self - energies .
we have analyzed the behaviour in this hot and dense environment of dynamically - generated baryonic resonances together with the evolution with density and temperature of the strange and open - charm meson spectral functions .
the spectral function for @xmath2 and @xmath6 mesons dilutes with increasing temperature and density while the quasiparticle peak moves closer to the free position . the spectral function for strange mesons
is also tested using energy - weighted sum rules and we found that the sum rules for the lower energy weights are fulfilled satisfactorily .
we have finally discussed the implications of the properties of charm mesons on the @xmath0 and the predicted @xmath1 .
we suggest to look at transparency ratios to investigate the changes in width of those resonances in nuclear matter .
l.t . acknowledges support from the rff - open and hidden charm at panda project of the university of groningen .
this work is partly supported by the eu contract no .
mrtn - ct-2006 - 035482 ( flavianet ) , by the contracts fis2006 - 03438 , fis2008 - 01661 , fis2008 - 01143 and fpa2009 - 00592 from micinn ( spain ) , by the spanish consolider - ingenio 2010 programme cpan ( csd2007 - 00042 ) , by the generalitat de catalunya contract 2009sgr-1289 , by the ucm - bsch contract gr58/08 910309 contracts and by junta de andaluca under contract fqm225 .
we acknowledge the support of the european community - research infrastructure integrating activity `` study of strongly interacting matter '' ( hadronphysics2 , grant agreement n. 227431 ) under the 7th framework programme of eu . e. friedman and a. gal , _ phys .
rept . _ * 452 * , 89 ( 2007 ) . c. fuchs , _ prog .
nucl . phys .
_ * 56 * , 1 ( 2006 )
. v. koch , _ phys .
lett . _ * b337 * , 7 ( 1994 ) .
m. lutz , _ phys .
lett . _ * b426 * , 12 ( 1998 ) . a. ramos and e. oset ,
_ * a671 * , 481 ( 2000 ) .
l. tolos , a. ramos , a. polls and t. t. s. kuo , _ nucl .
_ * a690 * , 547 ( 2001 ) ; l. tolos , a. ramos and a. polls , _ phys .
* c65 * , 054907 ( 2002 ) .
l. tolos , a. ramos and e. oset , _ phys .
rev . _ * c74 * , 015203 ( 2006 ) .
m. f. m. lutz , c. l. korpa and m. moller , _ nucl .
phys . _ * a808 * , 124 ( 2008 ) .
l. tolos , d. cabrera and a. ramos , _ phys .
rev . _ * c78 * , 045205 ( 2008 ) .
d. cabrera , a. polls , a. ramos and l. tolos , _ phys .
_ * c80 * , 045201 ( 2009 ) .
m. f. m. lutz and c. l. korpa , _ phys .
lett . _ * b633 * , 43 ( 2006 ) . ; t. mizutani and a. ramos , _ phys .
rev . _ * c74 * , 065201 ( 2006 ) .
l. tolos , a. ramos and t. mizutani , _ phys .
rev . _ * c77 * , 015207 ( 2008 ) . c. garcia - recio , v. k. magas , t. mizutani , j. nieves , a. ramos , l. l. salcedo and l. tolos , _ phys .
rev . _ * d79 * , 054004 ( 2009 ) . c. garcia - recio ,
j. nieves and l. l. salcedo , _ phys .
_ * d74 * , 034025 ( 2006 ) .
l. tolos , c. garcia - recio and j. nieves , arxiv:0905.4859 [ nucl - th ] c. amsler _ et al . _
[ particle data group ] , _ phys .
lett . _ * b667 * , 1 ( 2008 ) .
e. e. kolomeitsev and m. f. m. lutz , _ phys .
_ * b582 * , 39 ( 2004 ) ; f. k. guo , p. n. shen , h. c. chiang and r. g. ping , _ phys .
lett . _ * b641 * , 278 ( 2006 ) .
d. gamermann , e. oset , d. strottman and m. j. vicente vacas , _ phys .
* d76 * , 074016 ( 2007 ) .
k. abe _ et al . _
[ belle collaboration ] , _ phys .
lett . _ * 100 * , 202001 ( 2008 ) .
d. gamermann and e. oset , _ eur .
phys . j. _
* a36 * , 189 ( 2008 ) .
r. molina , d. gamermann , e. oset and l. tolos , _ eur .
j _ * a42 * , 31 ( 2009 ) . | we study the properties of strange and charm mesons in hot and dense matter within a self - consistent coupled - channel approach for the experimental conditions of density and temperature expected for the cbm experiment at fair / gsi .
the in - medium solution at finite temperature accounts for pauli blocking effects , mean - field binding of all the baryons involved , and meson self - energies .
we analyze the behaviour in this hot and dense environment of dynamically - generated baryonic resonances together with the evolution with density and temperature of the strange and open - charm meson spectral functions .
we test the spectral functions for strange mesons using energy - weighted sum rules and finally discuss the implications of the properties of charm mesons on the @xmath0 and the predicted @xmath1 scalar resonances . |
solar neutrinos have now been detected with the geochemical method @xcite via the + @xmath3cl @xmath4ar + @xmath5 reaction at homestake , and via the + @xmath6ga @xmath7ge + @xmath5 reaction at the gallex , sage and gno experiments .
direct detection with cherenkov techniques using @xmath8 scattering on water at super - k @xcite , and heavy water at sno @xcite have given a robust confirmation that the number of solar neutrinos detected in underground experiments is less than expected from theories of energy generation in the sun @xcite .
especially relevant is the sensitivity of the sno experiment to the neutral current ( nc ) .
altogether these experiments provide a solid evidence for solar neutrino conversions and , as a result , imply that an extension of the standard model of particle physics in the lepton sector is needed .
although not yet unique , the most popular explanation of solar neutrino experiments is provided by the neutrino oscillations hypothesis .
present data indicate that the mixing angle is large @xcite , the best description being given by the lma msw - type @xcite solution , already hinted previously from the flat super - k recoil electron spectra @xcite . the absence of a clear hint for day - night or seasonal variation in the solar neutrino signal places important restrictions on neutrino parameters . in ref .
@xcite the solar neutrino data , including the latest 1496day data and the sno nc result have been analysed in the general framework of mixed active - sterile neutrino oscillations , where the electron neutrino produced in the sun converts to a combination of an active non - electron neutrino @xmath9 ( a combination of @xmath10 and @xmath11 ) and a sterile neutrino @xmath12 : @xmath13 .
the setting for such scenarios are four - neutrino mass schemes @xcite which try to accommodate the solar and atmospheric mass - splittings with the hint for short baseline oscillations from lsnd @xcite indicating a large @xmath14 .
the parameter @xmath15 with @xmath16 describes the fraction of sterile neutrinos taking part in the solar oscillations . in fig .
[ fig : sol - osc - par ] we display the regions of solar neutrino oscillation parameters for 3 d.o.f .
with respect to the global minimum , for the standard case of active oscillations , @xmath17 , as well as for @xmath18 and @xmath19 . the first thing to notice
is the impact of the sno nc , spectral , and day / night data in improving the determination of the oscillation parameters : the shaded regions after their inclusion are much smaller than the hollow regions delimited by the corresponding sno@xmath20 confidence contours . especially important
is the full sno@xmath21 information in closing the lma region from above : values of @xmath22 appear only at @xmath23 .
previously solar data on their own could not close the lma region , only the inclusion of reactor data @xcite probed the upper part of the lma region @xcite .
furthermore , the complete sno@xmath21 information is important for excluding ` maximal ` solar mixing in the lma region . at @xmath23
we find the upper bound ( 1 d.o.f . ) : @xmath24 in order to compare the allowed regions in fig .
[ fig : sol - osc - par ] with others @xcite , one must note that our c.l .
regions correspond to the 3 d.o.f .
: @xmath25 , @xmath26 and @xmath15 . therefore at a given c.l .
our regions are larger than the usual regions for 2 d.o.f . , because we also constrain the parameter @xmath15 .
allowed @xmath25 and @xmath26 regions for @xmath27 ( active oscillations ) , @xmath18 and @xmath19 .
the lines and shaded regions correspond to the sno@xmath20 and sno@xmath21 analyses , respectively , as defined in ref .
@xcite . the 90% , 95% , 99% c.l .
and 3@xmath28 contours are for 3 d.o.f ..
,scaledwidth=47.0% ] our global best fit point occurs for active oscillations with @xmath29 a concise way to illustrate the above results is displayed in fig .
[ fig : chi - sol ] .
we give the profiles of @xmath30 as a function of @xmath26 ( left ) as well as @xmath25 ( right ) , by minimizing with respect to the undisplayed oscillation parameters , for the fixed values of @xmath31 , @xmath32 , @xmath33 .
@xmath30 as a function of @xmath26 and @xmath25 , for pure active ( @xmath17 ) , pure sterile ( @xmath34 ) and mixed neutrino oscillations ( @xmath19 ) .
upper and lower panels correspond to the sno@xmath20 and sno@xmath21 samples defined in ref .
@xcite.,scaledwidth=47.0% ] by comparing top and bottom panels one can clearly see the impact of the full sno@xmath21 sample in leading to the relative worsening of all non - lma solutions with respect to the preferred active lma solution .
one sees also how the preferred lma status survives in the presence of a small sterile admixture characterized by @xmath15 ( also seen in figs .
[ fig : sol - osc - par ] and [ fig : sol02.etas ] ) .
increasing @xmath15 leads to a deterioration of all oscillation solutions .
it is also instructive to display the profile of @xmath30 as a function of @xmath35 , irrespective of the detailed values of the solar neutrino oscillation parameters @xmath26 and @xmath36 , as shown in fig .
[ fig : sol02.etas ] .
one can see that there is a crossing between the lma and quasi vacuum
oscillations ( qvo ) solutions .
this implies that the best pure sterile description lies in the qvo regime .
however , in the global analysis pure sterile oscillations with @xmath37 are highly disfavored .
we find a @xmath38-difference between pure active and sterile of @xmath39 if we restrict to the lma solution , or @xmath40 if we allow also for qvo . for 3 d.o.f .
the @xmath40 implies that pure sterile oscillations are ruled out at 99.996% c.l .
compared to the active case . from the figure we obtain the bound @xmath41 at 99% c.l . for 1 d.o.f
.. a complete table of best fit values of @xmath26 and @xmath36 with the corresponding @xmath42 and gof values for pure active , pure sterile , and mixed neutrino oscillations is given in @xcite , both for the sno@xmath20 ( @xmath43 d.o.f . ) and the sno@xmath21 analysis ( @xmath44 d.o.f . ) .
here i summarize the analysis of atmospheric data given in ref .
@xcite , in a generalized oscillation scheme in which a light sterile neutrino takes part in the oscillations , under the approximation @xmath45 .
in order to comply with the strong constraints from reactor experiments @xcite we completely decouple the electron neutrino from atmospheric oscillations .
in contrast with the case of solar oscillations , the constraints on the @xmath10content in atmospheric oscillations are not so stringent .
thus the description of atmospheric neutrino oscillations in this general framework requires two parameters besides the standard 2-neutrino oscillation parameters @xmath46 and @xmath47 .
we will use the parameters @xmath48 and @xmath49 already introduced in ref .
@xcite , and defined in such a way that @xmath50 ( @xmath51 ) corresponds to the fraction of @xmath10 ( @xmath52 ) participating in oscillations with @xmath47 .
hence , pure active atmospheric oscillations with @xmath47 are recovered when @xmath53 and @xmath54 . in four - neutrino models
there is a mass scheme - dependent relationship between @xmath49 and the solar parameter @xmath15 .
for details see ref .
@xcite . to get a feeling on the physical meaning of these two parameters ,
note that for @xmath53 we obtain that the @xmath10 oscillates with @xmath47 to a linear combination of @xmath11 and @xmath52 given as @xmath55 our global best fit point occurs at @xmath56 and has @xmath57 .
we see that atmospheric data prefers a small sterile neutrino admixture .
however , this effect is not statistically significant , since the pure active case ( @xmath58 ) also gives an excellent fit : the difference in @xmath38 with respect to the best fit point is only @xmath59 . for the pure active best fit point
we obtain , @xmath60 with the 3@xmath28 ranges ( 1 d.o.f . ) @xmath61 the determination of the parameters @xmath46 and @xmath47 is summarized in figs .
[ fig : atm - osc - par ] and [ fig : chi - atm ] . allowed regions of @xmath62 and @xmath47 at 90% , 95% , 99% and 3@xmath28 for 4 d.o.f . and different assumptions on the parameters @xmath49 and @xmath48 , from @xcite .
the lines ( shaded regions ) correspond to 1289 ( 1489 ) days of super - k data .
, scaledwidth=46.0%,height=188 ] note that fig .
[ fig : chi - atm ] considers several cases : arbitrary @xmath49 and @xmath48 , best
fit @xmath49 and @xmath48 , and pure active and mixed active sterile neutrino oscillations , as indicated . at a given c.l .
we cut the @xmath63 at a @xmath64 determined by 4 d.o.f . to obtain 4-dimensional volumes in the parameter space of ( @xmath65 ) .
in the upper panels we show sections of these volumes at values of @xmath54 and @xmath53 corresponding to the pure active case ( left ) and the best fit point ( right ) .
again we observe that moving from pure active to the best fit does not change the fit significantly . in the lower right panel
we project away both @xmath48 and @xmath49 , whereas in the lower left panel we fix @xmath66 and eliminate only @xmath48 . comparing the regions resulting from 1489 days super - k data ( shaded regions ) with the one from the 1289 days super - k sample ( hollow regions )
we note that the new data leads to a slightly better determination of @xmath46 and @xmath47 .
however , more importantly , from the lower left panel we see , that the new data shows a much stronger rejection against a sterile admixture : for @xmath66 no allowed region appears at 3@xmath28 for 4 d.o.f .. @xmath67 as a function of @xmath62 ( left ) and @xmath47 ( right ) , using 1289 ( upper ) and 1489 ( lower ) days of super - k data @xcite.,scaledwidth=46.0%,height=151 ]
the basic structure of the neutrino sector required to account for present solar and atmospheric data has been developed in the early eighties , motivated mainly by theory @xcite . in a gauge theory of the weak interaction
the ` simplest ` lepton mixing matrix is characterized by 3 angles and 3 cp violating phases @xcite : * the solar angle @xmath1 * the atmospheric angle @xmath2 * the reactor angle @xmath0 * one kobayashi - maskawa - like cp phase * 2 extra ( majorana - type ) cp phases the structure of leptonic weak interactions is more complex in theories containing 1 singlet leptons @xcite , especially if , for some symmetry reason , they happen to be light . in such case the charged current mixing matrix is rectangular and the neutral current is non - trivial , with yet new angles and phases present ( see @xcite for a detailed discussion and parametrization ) .
as seen above @xcite , current solar and atmospheric data fit very well with oscillations among the three active neutrinos , provided that @xmath1 is large as seen in eq .
( [ t12d12 ] ) , but non - maximal , given in eq .
( [ t12bound ] ) , while @xmath2 must be nearly maximal , from eq .
( [ t23d23 ] ) .
as mentioned , @xmath0 must be rather small @xcite .
note from eqs .
( [ t12d12 ] ) and ( [ t23d23 ] ) that @xmath68 . depending on the sign of @xmath69
there are two possible neutrino mass schemes : normal and inverse - hierarchical neutrino masses .
if solar and atmospheric data are combined with short baseline data including the lsnd hint , then one needs , in the framework of the oscillation hypothesis , the existence of a light sterile neutrino @xcite taking part in the oscillations @xcite . with the latest data one finds that ,
even though 4-neutrino models can not be ruled out as such , the resulting global fit of all current oscillation data is extremely poor @xcite .
the large solar mixing indicated by present data will lead to significant deformation of the energy spectra of supernova neutrinos , affecting the resulting signal @xcite .
however , a global analysis shows that the lma
msw may remain as the best solution even after combining sn1987 with solar neutrino data @xcite .
finally we note that solar neutrino oscillations with large mixing may allow us in the future to obtain otherwise inaccessible features of sn neutrino spectra .
[ minak ] from @xcite shows how one can determine temperatures and luminosities of non - electron flavor neutrinos by observing @xmath70 from a galactic supernova in massive water cherenkov detectors using the cc reactions on protons , especially at a hyper - k type detector .
last , but not least , note that neutrino oscillations are sensitive only to mass splittings , not to the absolute scale of neutrino mass , nor to whether neutrinos are dirac or majorana particles . the main process relevant to decide this fundamental issue is decay @xcite . the ` black - box theorem ` states that , in a `` natural '' gauge theory , irrespective of how is engendered , it implies a majorana neutrino mass , and vice - versa , as illustrated by fig .
[ fig : bbox ] .
one may quantify the implications of the black - box argument once a particular model is specified .
the strength of neutrino - exchange - induced is characterized by an `` effective '' neutrino mass parameter @xmath71 which takes into account possible cancellations among individual neutrino amplitudes @xcite . as seen in fig .
[ fig : bbmass ] this is directly correlated @xcite with the neutrino mass scales probed in tritium beta decays @xcite and cosmology @xcite .
it is therefore important to probe in a more sensitive experiment , such as genius @xcite . moreover @xmath72 processes , such as decay , are sensitive to the `` majorana - type '' phases @xcite which drop out from ordinary ( @xmath73 ) neutrino oscillations .
however it is unlikely that these phases can ever be reliably extracted from alone .
the cp violation induced by the `` dirac '' phase is very hard to probe in oscillations , since it disappears as two neutrinos become degenerate @xcite and as @xmath74 @xcite .
fortunately the lma solution helps , so with good luck , neutrino factories may probe leptonic cp violating effects , through the measurement of cp asymmetries @xcite .
basic uncertainties hinder the prediction of neutrino masses from first principles @xcite : no knowledge of the underlying scale ( the scale of gravity / strings ?
the gut scale ?
an intermediate left - right scale ?
the weak interaction scale itself ? ) , no knowledge of the underlying mechanism ( tree level ?
hybrid ? ) and , last but not least , lack of a theory of flavor , which makes it especially difficult to make an honest prediction of mixing angles .
nevertheless there has been an explosion of models in the last few years , most of which based on the so - called seesaw scheme @xcite .
an interesting of feature of such seesaw models is that the amount of lepton asymmetry produced in the early universe by the out of thermal equilibrium decay of the right - handed neutrino may be enough to explain the current baryon - to - photon ratio of the universe @xcite .
this asymmetry arises from leptonic cp violation associated to the majorana nature of neutrinos @xcite .
various yukawa textures and gauge groups have been proposed in order to `` predict '' neutrino mixing angles in the seesaw approach , as described by king @xcite . here
i simply mention a few models from our own crop .
they fall into two classes : ` top - down ` and ` bottom - up ` , and cover both ` hierarchical ` and ` quasi - degenerate ` neutrino mass spectra .
the simplest way to give neutrino masses makes use of the basic dimension-5 operator in fig .
[ d-5.ps ] , first noted by weinberg @xcite whose coefficient is unknown .
since gravity is believed to break global symmetries , such as lepton number , it may induce this operator .
alternatively it may arise from physics at some grand - unified or intermediate scale , ` a la seesaw ` .
a radical idea @xcite is that , due to some symmetry , valid at some high - energy scale @xmath75 , neutrino masses `` unify '' at that scale , as indicated in fig .
[ neunif2.ps ] .
gev , scaledwidth=40.0%,height=132 ] such a simple ansatz naturally leads to quasi - degenerate neutrinos at low scales , which could lie in the electron - volt range , while the tiny solar and atmospheric neutrino mass splittings are induced by renormalization effects .
an interesting choice for the underlying symmetry is the discrete non - abelian symmetry @xmath76 @xcite , which can be realized without spoiling the hierarchy of charged - lepton masses .
solar and atmospheric neutrino mass splittings , maximal atmospheric and large solar mixings are then induced radiatively in softly broken supersymmetry .
the quark mixing matrix is also calculable in a similar way .
the mixing parameter @xmath77 is predicted to be small and imaginary , leading to maximal cp violation in neutrino oscillations .
the is a majorana neutrino while the muon and tau neutrinos form a pseudo - dirac pair @xcite .
we found that and @xmath78 decay rates fall in the experimentally accessible range . another minimalistic way to generate neutrino
mass and mixings has been suggested in @xcite .
it consists in producing the atmospheric neutrino scale ` a la seesaw ` with just one 21 singlet lepton @xcite . in this approximation two of the neutrinos
are massless , their degeneracy being lifted by the gravitationally induced dimension-5 operator discussed above .
the required neutrino parameters can be easily accommodated . since the solar scale comes from planck - mass effects ,
the solar neutrino problem is explained by vacuum oscillations , which may be tested through the search for anomalous seasonal effects at borexino .
i now turn to the possibility that neutrino masses may have an intrinsically supersymmetric origin , through the breaking of r
parity @xcite .
i focus on the bilinear violation of r
parity @xcite , described by @xmath79 where @xmath80 is the mssm superpotential , @xmath81 ( @xmath82 ) denote the strength of bilinear terms involving the lepton ( @xmath83 ) and up - type higgs ( @xmath84 ) superfields .
the bilinear terms lead to ` one ` tree level neutrino mass , chosen to lie in the range required by the atmospheric neutrino data , while calculable radiative contributions lift the degeneracy of the other neutrinos giving rise to the solar neutrino scale @xcite .
this provides the simplest , most predictive and systematic effective r parity violation model at low energies .
its theoretical basis can be found either in the context of models where r
parity breaking is introduced explicitly ` ab initio ` @xcite , or in models where the violation of r parity occurs ` spontaneously ` , through a non - zero 1 singlet sneutrino vacuum expectation value @xcite .
a recent example of the former kind was given in @xcite using an anomalous u(1 ) horizontal symmetry which forbids all trilinear r - parity violating superpotential terms , selecting only the bilinear ones , and relating their strength @xmath81 to powers of the u(1 ) breaking parameter @xmath85 .
this gives a common origin for the @xmath86term related to electroweak breaking and for the l - violating terms generating neutrino masses .
the latter are suitable for explaining neutrino anomalies , though radiative contributions prefer the presently disfavored qvo solutions to the solar neutrino problem .
this can be tested through the search for anomalous seasonal effects .
the neutrino mixing angles are not suppressed by powers of @xmath87 and can naturally be large .
spontaneous r
parity breaking models require the addition of 1 singlet superfields , e. g. right - handed neutrinos , and give a dynamical origin for the bilinear strength @xmath81 identified as @xmath88 where @xmath89 is the dirac yukawa matrix .
a characteristic feature of these models is the existence of an novel variant ( l=1 ) of the singlet seesaw majoron @xcite .
majoron emission induces `` invisible '' neutrino @xcite and higgs boson decays @xcite .
the former are relevant in astrophysical environments @xcite , while the latter lead to sizable higgs to missing transverse
momentum signals @xcite .
in addition to explaining neutrino anomalies , supersymmetry with bilinear breaking of r parity leads to a variety of phenomenological implications @xcite .
some examples are : unification predictions for gauge and yukawa couplings , @xmath90 , @xmath91 and @xmath92 @xcite , @xmath93 @xcite , @xcite , charged and neutral higgs boson decays @xcite , top quark decays @xcite , chargino @xcite and neutralino decays @xcite , and gluino cascade decays @xcite . barring fine - tuning or other assumptions ,
the smallness of the neutrino masses suppresses many of these effects in the most generic bilinear model of neutrino anomalies
. however , there is at least one which survives and which may lead to a dramatic confirmation of the neutrino anomalies at high energy accelerator experiments such as the lhc or nlc : the decays of the lightest supersymmetric particle ( lsp ) .
this is a very striking feature of the model which holds irrespective of what is the nature of the lsp .
we have considered the cases of lightest neutralino as lsp @xcite , lightest stop as lsp @xcite and lightest stau as lsp @xcite . as an example , the left panel in fig .
[ anomalytest ] shows the lsp decay length in cm versus mass in gev , when it is the lightest neutralino . on the other hand
the right panel gives the ratio of predicted lsp decay semileptonic decay branching ratios to muons over taus ( ordinate ) versus the atmospheric neutrino mixing angle ( abscissa ) , illustrating a perfect correlation : if @xmath94 one expects equal numbers of muons and taus in semileptonic neutralino decays .
similarly , the solar and reactor neutrino angles can be probed in various lsp decay scenarios , see @xcite for details .
non - standard neutrinos interactions ( nsi ) are expected in most neutrino mass models @xcite and can be of two types : flavour - changing ( fc ) and non - universal ( nu ) .
they may arise from a nontrivial structure of cc and nc weak interactions characterized a non - unitary lepton mixing matrix and a correspondingly non - trivial nc matrix @xcite . such `
gauge - induced ` nsi may lead to flavor and cp violation , even with degenerate massless neutrinos @xcite . in radiative models of neutrino mass @xcite and supersymmetric models with broken r parity @xcite fc - nsi can also be ` yukawa - induced ` , from the exchange of spinless bosons . in supersymmetric unified models , they may be calculable as renormalization effects @xcite . at the moment one can not yet pin down the exact profile of the survival probability and , as a result ,
the underlying mechanism of solar neutrino conversion remains unknown .
alternatives to oscillations have been suggested since the eighties , including non - standard neutrino matter interactions ( nsi ) @xcite and spin - flavor precession ( sfp ) @xcite .
the former may be represented as effective dimension-6 terms of the type @xmath95 , as illustrated in fig .
[ fig : nunsi ] , where @xmath96 specifies their sub - weak strength . analyses of solar neutrino data in terms of nsi @xcite and sfp @xcite have been given recently .
sfp solutions exist both in the resonant ( rsfp ) @xcite and non - resonant regimes ( nrsfp ) @xcite .
the nsi solar neutrino energy spectrum is undistorted , in agreement with data . on the other hand fig .
[ probs.eps ] shows the predicted modification of the solar neutrino spectra for the `` best '' lma solution , and for the `` best '' sfp solutions @xcite , from latest solar data .
clearly the spectra in the high energy region are nearly undistorted in all 3 cases , in agreement with observations .
although lma oscillations are clearly favored over other ` oscillation ` solutions @xcite , present solar neutrino data ` can be equally well
described by sfp and nsi solutions ` .
[ chi2sfp ] shows that this is indeed the case , the resulting parameter regions is given in fig .
[ sfpregions ] .
note that in the presence of a neutrino transition magnetic moment of @xmath98 bohr magneton , a magnetic field of 80 kgauss eliminates all oscillation solutions other than lma .
ways to separate these 3 solutions at borexino have been considered in @xcite .
similarly the regions for the nsi mechanism are shown in fig .
[ fig : hybparam ] .
the required nsi values indicated by the solar data analysis are fully acceptable also for the atmospheric data . can nsi play a role in the atmospheric neutrino signal ?
fc - nsi interactions in the -channel without neutrino mass nor mixing have been shown to account for the zenith angle
dependent deficit of atmospheric neutrinos observed in ` contained ` super - k events @xcite .
however such nsi explanation fails to reconcile these with super - k and macro ` up - going muons ` , due to the lack of energy dependence intrinsic of nsi conversions . as a result , a pure nsi conversion in the atmospheric channel
is ruled out at 99% c.l .
thus , unlike the case of solar neutrinos , the oscillation interpretation of atmospheric data is robust , nsi being allowed only at a sub - leading level .
such robustness of the atmospheric @xmath99 oscillation hypothesis can be used to provide the most stringent present limits on fc and nu neutrino interactions , as illustrated in fig . [
fig : atmnsibds ] .
future neutrino factories aim at probing the lepton mixing angle @xmath0 with much better sensitivity than possible at present @xcite .
they may also probe nsi in the -channel @xcite , with substantially improved sensitivity in the case of fc - nsi , especially at energies higher than approximately 50 gev .
for example , a 100 gev ` nufact ` can probe fc - nsi interactions at the level of @xmath100 at 99 % c.l .
note also that in such hybrid solution to the neutrino anomalies , with fc - nsi explaining the solar data , and oscillations accounting for the atmospheric data , the two sectors are connected not only by the neutrino mixing angle @xmath0 , but also by the -fc - nsi parameters . as a result nsi and oscillations may be confused , as shown in @xcite .
this implies that information on @xmath0 can only be obtained if bounds on nsi are available .
taking into account the existing bounds on fc interactions , one finds a drastic loss in ` nufact ` sensitivities on @xmath0 , of at least two orders of magnitude .
a near detector offers the possibility to obtain stringent bounds on some nsi parameters and therefore constitutes a crucial necessary step towards the determination of @xmath0 and subsequent study of leptonic cp violation .
last , but not least , note that the kamland experiment @xcite will provide vital information very soon .
even if the lma solution is finally confirmed by kamland , such alternative mechanisms will still be interesting for study since their sub - leading admixture may be testable , as discussed in @xcite .
they may also affect the propagation of neutrinos in a variety of astrophysical environments @xcite .
l. wolfenstein , phys .
d * 17 * , 2369 ( 1978 ) ; s. p. mikheev and a. y. smirnov , sov . j. nucl .
* 42 * ( 1985 ) 913 m. c. gonzalez - garcia , p. c. de holanda , c. pena - garay and j. w. valle , nucl .
b * 573 * , 3 ( 2000 ) j. t. peltoniemi , d. tommasini and j. w. valle , phys .
b * 298 * , 383 ( 1993 ) .
j. t. peltoniemi and j. w. valle , nucl .
b * 406 * , 409 ( 1993 ) ; d. o. caldwell and r. n. mohapatra , phys .
d * 48 * ( 1993 ) 325 ; additional references in giunti and lavader s excellent web
page http://www.to.infn.it/ giunti / neutrino/ m. apollonio et al , chooz coll . , phys .
b * 466 * ( 1999 ) 415 ; f. boehm et al , palo verde coll .
d * 64 * ( 2001 ) 112001 m. c. gonzalez - garcia , m. maltoni , c. pena - garay and j. w. valle , phys .
d * 63 * , 033005 ( 2001 ) m. maltoni , t. schwetz and j. w. valle , phys .
d * 65 * , 093004 ( 2002 ) ; phys .
b * 518 * ( 2001 ) 252 s. weinberg , proceedings of ` neutrinos 78 ` , p , 1 - 21 ; r. barbieri , j. r. ellis and m. k. gaillard , phys .
b * 90 * , 249 ( 1980 ) .
m. gell - mann , p. ramond and r. slansky in _ supergravity _ ( north holland , amsterdam 1979 ) ; t. yanagida in _ proc . of the workshop on unified theory and baryon number of the universe _ , kek , japan , 1979
. r. n. mohapatra and g. senjanovic , phys . rev
. lett .
* 44 * ( 1980 ) 912 ; j. schechter and j. w. valle , phys .
d * 22 * , 2227 ( 1980 ) ; c. wetterich , nucl .
b * 187 * ( 1981 ) 343 .
y. chikashige , r. n. mohapatra and r. d. peccei , phys .
b * 98 * ( 1981 ) 265 ; j. schechter and j. w. valle , phys . rev .
d * 25 * ( 1982 ) 774 ; g. b. gelmini and j. w. valle , phys .
b * 142 * , 181 ( 1984 ) .
j. schechter and j. w. valle , phys .
d * 22 * , 2227 ( 1980 ) ; phys .
d * 23 * , 1666 ( 1981 ) . for a phenomenological review see s. m. bilenky , c. giunti and w. grimus , prog .
* 43 * ( 1999 ) 1 m. maltoni , t. schwetz , m. a. tortola and j. w. valle , arxiv : hep - ph/0207157 .
a. y. smirnov et al , phys .
d * 49 * ( 1994 ) 1389 ; b. jegerlehner , f. neubig and g. raffelt , phys .
d * 54 * , 1194 ( 1996 ) ; m. kachelriess et al , jhep * 0101 * , 030 ( 2001 ) m. kachelriess , a. strumia , r. tomas and j. w. valle , phys .
d * 65 * , 073016 ( 2002 ) h. minakata , h. nunokawa , r. tomas and j. w. valle , phys .
b * 542 * , 239 ( 2002 ) a. morales , nucl .
suppl . * 77 * , 335 ( 1999 ) ; a. faessler and f. simkovic , prog .
* 46 * ( 2001 ) 233 .
j. schechter and j. w. valle , phys .
d * 25 * , 2951 ( 1982 ) . l. wolfenstein , nucl .
b * 186 * ( 1981 ) 147 .
j. w. valle and m. singer , phys .
d * 28 * , 540 ( 1983 ) .
j. w. valle , phys .
d * 27 * , 1672 ( 1983 ) . v. barger et al , phys .
b * 532 * ( 2002 ) 15 .
f. feruglio , a. strumia and f. vissani , arxiv : hep - ph/0201291 ; h. minakata and h. sugiyama , phys .
b * 532 * , 275 ( 2002 ) ch .
weinheimer , these proceedings .
o. elgaroy et al , arxiv : astro - ph/0204152 .
h. v. klapdor - kleingrothaus et al , [ genius collaboration ] , arxiv : hep - ph/9910205 . m. doi , t. kotani and e. takasugi , prog . theor .
phys . suppl .
* 83 * , 1 ( 1985 ) .
j. schechter and j. w. valle , phys .
d * 21 * , 309 ( 1980 )
. talks by f. dydak , s. geer and m. lindner .
h. chankowski et al , phys . rev .
* 86 * ( 2001 ) 3488 k. s. babu , e. ma and j. w. valle , arxiv : hep - ph/0206292 .
a. de gouvea and j. w. valle , phys .
b * 501 * , 115 ( 2001 ) c. s. aulakh and r. n. mohapatra , phys .
b * 119 * ( 1982 ) 13 ; g. g. ross and j. w. f. valle , phys .
b * 151 * ( 1985 ) 375 ; j. r. ellis et al , phys .
b * 150 * ( 1985 ) 142 ; a. santamaria and j. w. f. valle , phys .
b * 195 * , 423 ( 1987 ) ; phys .
lett . *
60 * , 397 ( 1988 ) ; phys .
d * 39 * , 1780 ( 1989 ) .
l. j. hall and m. suzuki , nucl .
b * 231 * ( 1984 ) 419 , see also @xcite and papers therein .
j. w. valle , proceedings of 6th int .
symposium on particles , strings and cosmology ( pascos 98 ) , p. 502 - 512 ; arxiv : hep - ph/9808292 and references therein .
m. a. diaz , j. c. romao and j. w. valle , nucl . phys .
b * 524 * , 23 ( 1998 ) m. hirsch et al , phys .
d * 62 * , 113008 ( 2000 ) [ err .- ibid .
d * 65 * , 119901 ( 2002 ) ] ; j. c. romao et al , phys .
d * 61 * , 071703 ( 2000 ) j. m.
mira et al , phys .
b * 492 * , 81 ( 2000 ) j. c. romao , f. de campos and j. w. valle , phys .
b * 292 * , 329 ( 1992 ) a. s. joshipura and j. w. valle , nucl . phys .
b * 397 * ( 1993 ) 105 m. kachelriess , r. tomas and j. w. valle , phys .
d * 62 * ( 2000 ) 023004 m. carena et al , `` higgs physics '' , report of the workshop on physics at lep2 , ed .
g. altarelli et al , arxiv : hep - ph/9602250 .
f. de campos et al , phys .
d * 55 * ( 1997 ) 1316 ; m. a. diaz et al , nucl .
b * 527 * , 44 ( 1998 ) ; phys .
d * 58 * , 057702 ( 1998 ) b. allanach et al , arxiv : hep - ph/9906224
. m. a. diaz , j. ferrandis and j. w. valle , nucl .
b * 573 * , 75 ( 2000 ) m. a. diaz et al , nucl .
b * 590 * , 3 ( 2000 ) m. a. diaz et al , phys .
b * 453 * , 263 ( 1999 ) m. a.
diaz et al , nucl .
b * 551 * , 78 ( 1999 ) m. hirsch et al , phys .
b * 486 * , 255 ( 2000 ) ; m. hirsch and j. w. valle , nucl .
b * 557 * , 60 ( 1999 ) a. g. akeroyd et al , nucl .
b * 529 * , 3 ( 1998 ) ; f. de campos et al , nucl .
b * 451 * , 3 ( 1995 ) l. navarro , w. porod and j. w. valle , phys .
b * 459 * , 615 ( 1999 ) f. de campos et al , nucl . phys .
b * 546 * , 33 ( 1999 ) a. bartl et al , nucl .
b * 600 * , 39 ( 2001 ) a. bartl et al , nucl .
b * 502 * , 19 ( 1997 ) w. porod et al , phys .
d * 63 * , 115004 ( 2001 ) d. restrepo et al , phys .
d * 64 * , 055011 ( 2001 ) ; m. a. diaz et al , nucl .
b * 583 * , 182 ( 2000 ) m. hirsch et al , arxiv : hep - ph/0207334 .
r. n. mohapatra and j. w. f. valle , phys .
d * 34 * ( 1986 ) 1642 .
d. wyler and l. wolfenstein , nucl .
b * 218 * ( 1983 ) 205 .
a. zee , phys .
b * 93 * ( 1980 ) 389 [ err - ibid .
b * 95 * ( 1980 ) 461 ] .
k. s. babu , phys .
b * 203 * ( 1988 ) 132 .
l. j. hall , v. a. kostelecky and s. raby , nucl .
b * 267 * ( 1986 ) 415 r. barbieri , l. j. hall and a. strumia , nucl .
b * 445 * , 219 ( 1995 ) f. gabbiani et al , nucl .
b * 477 * , 321 ( 1996 ) j. w. valle , phys .
b * 199 * , 432 ( 1987 ) .
e. k. akhmedov , phys .
b * 213 * ( 1988 ) 64 . c. s. lim and w. j. marciano , phys .
d * 37 * ( 1988 ) 1368 .
j. schechter and j. w. valle , phys .
d * 24 * , 1883 ( 1981 ) [ err .- ibid .
d * 25 * , 283 ( 1982 ) ] .
m. guzzo et al , nucl .
b * 629 * , 479 ( 2002 ) j. barranco et al , arxiv : hep - ph/0207326 .
o. g. miranda et al , phys .
b * 521 * , 299 ( 2001 ) o. g. miranda et al , nucl .
b * 595 * , 360 ( 2001 ) e. k. akhmedov and j. pulido , phys .
b * 529 * , 193 ( 2002 ) n. fornengo et al , jhep * 0007 * , 006 ( 2000 ) m. c. gonzalez - garcia et al , phys .
* 82 * , 3202 ( 1999 ) n. fornengo et al , phys . rev .
d * 65 * , 013010 ( 2002 ) p. huber and j. w. valle , phys .
b * 523 * , 151 ( 2001 ) p. huber , t. schwetz and j. w. valle , phys .
lett . * 88 * , 101804 ( 2002 ) and phys .
d * 66 * , 013006 ( 2002 ) j. shirai , these proceedings .
w. grimus et al , arxiv : hep - ph/0208132 . h. nunokawa et al , phys .
d * 54 * , 4356 ( 1996 ) ; h. nunokawa et al , nucl .
b * 482 * , 481 ( 1996 ) ; d. grasso et al , phys .
lett . * 81 * , 2412 ( 1998 ) ; g. l. fogli et al , phys .
d * 66 * , 013009 ( 2002 ) | i review the interpretation of solar and atmospheric neutrino data in terms neutrino oscillations and describe some ways to account for the required neutrino masses and mixing angles from first principles , both within top - down and bottom - up approaches .
i also discuss non - oscillation phenomena such as which may probe the absolute scale of neutrino mass , and also reveal its majorana nature .
i note that leptonic cp violation induced by `` majorana '' phases drop from oscillations but play a role in the leptogenesis scenario for the baryon asymmetry of the universe .
direct tests of leptonic cp violation in oscillation experiments , such as neutrino factories , will be a tough challenge , due to the hierarchical neutrino mass splittings and the smallness of @xmath0 indicated by reactors .
the large solar mixing angle @xmath1 offers a way to probe otherwise inaccessible features of supernova physics .
finally , i note that in low - scale models of neutrino mass , one may probe ` all ` mixing angles , including the atmospheric @xmath2 , at high energy accelerator experiments such as the lhc or nlc .
a neat example is supersymmetry with bilinear breaking of r parity , where the lsp decay branching ratios are directly correlated to the neutrino mixing angles .
i also discuss non - oscillation solutions to the solar neutrino problem in terms of spin - flavor precession and non - standard neutrino interactions , which will be crucially tested at kamland . |
the one - way quantum computational model , proposed by raussendorf and briegel @xcite , is remarkable in many aspects .
it represents an approach to quantum computation very different to more conventional `` circuit - based '' approaches which were derived in close analogy to classical logic circuits . in the one - way model ,
computation proceeds by the generation of a particular entangled multi - qubit state - a cluster state - followed by the adaptive measurement of individual qubits .
the choice of basis for the measurements , and their adaptive dependency encodes the computation .
the dependancy of the bases upon the outcome of previous measurements is a necessary part of the model .
it compensates for the inherent randomness of the outcome of individual measurements allowing deterministic computation .
measurements which are not directly or indirectly dependent upon each other can be performed simultaneously .
thus the one - way model offers a radically different approach to the parallelisation of computations .
broadbent and kashefi @xcite have pointed out a depth separation between the quantum circuit model and the one - way model . indeed , there is a constant depth one - way quantum computation for implementing the parity gate , whereas there is no poly - size constant - depth circuit for this gate .
moreover they have proved that for any problem , the depth - separation between quantum circuits and one - way model is at most logarithmic
. however , the exact power of the one - way quantum computation remained unknown . in this paper
, we mainly prove that the computational power of the one - way model is equivalent , up to classical side - processing , to the quantum circuit model augmented with unbounded fanout gates .
this model , first explored by hyer and spalek @xcite is a computational model which allows any two commuting gates to be performed simultaneously .
the unbounded fanout model is surprisingly powerful , for example , the quantum component of shor s algorithm reduces to constant depth in this model .
our results imply that the one - way model shares this power , provided the depth of the classical parity computations which make up the dependency calculations is neglected . in section [ sec : meascal ]
, we review the measurement calculus @xcite , a formal framework in which the one - way model can be succinctly represented .
section [ sec : mp - qc ] is dedicated to the comparison of the computational power of the one - way model and the quantum circuit model . in section [ sec : fan - out ]
we present the unbounded fanout circuit model and prove the main result of this paper , its equivalence with the one - way model . in section [ sec : gen ] , we show that reasonable classically controlled models can be efficiently implemented in the one - way model , answering an open question stated in @xcite . in section [ sec : app ] , we discuss some applications of these results and in section [ sec : assumption ] we discuss the assumptions underlying each of these models .
one - way quantum computations can be rigourously described in the measurement calculus formalism @xcite .
a term of measurement calculus , called _ measurement pattern _ and playing the role of a circuit in the quantum circuit model , is a quadruplet @xmath0 .
@xmath1 is a finite register of qubits , @xmath2 are subresgisters denoting respectively the inputs and output qubits . the non input qubits ( @xmath3 ) are initialised in state @xmath4 .
@xmath5 is a sequence of quantum operations , called commands .
they are three kinds of commands : * @xmath6 is an entangling operation , which is nothing but the controlled - z @xmath7 unitary gate on qubits @xmath8 where @xmath7 is definied as follows : @xmath9 * @xmath10 is a measurement of the qubit @xmath11 in the basis @xmath12:@xmath13 this measurement produces a classical outcomes @xmath14 . *
@xmath15 and @xmath16 are pauli corrections or _
dependant corrections _ , where @xmath17 is a finite sum modulo @xmath18 of classical outcomes @xmath19 s .
@xmath20 is applied on qubit @xmath11 and depends on the sum of @xmath21 as follows : @xmath22 and @xmath23 .
@xmath24 measurements are supposed to be destructive ( a measured qubit can not be reused anymore ) , as a consequence a measurement pattern is well - formed if no command is applied on already measured qubits .
@xmath25 note that the commands are read from right to left .
this pattern is implementing an @xmath26-rotation @xmath27 from qubit @xmath28 to qubit @xmath29 ( see section [ sec : mp - qc ] for the definition of @xmath30 ) .
another kind of command , the _ dependant measurements _ exist in the measurement calculus .
they can be defined as a combination of measurements and dependant corrections : @xmath31^\sigma : = m_i^\alpha x^\sigma_iz_i^\tau = m_i^{(-1)^s\alpha + \tau\pi}\ ] ] corrections of the form @xmath32 clearly illustrates the hybrid nature of measurement - based quantum computing : a classical control is collecting classical outcomes of previous measurements and computes the sum for deciding what the next quantum operation is .
let @xmath33 and @xmath34 be two measurement patterns .
the sequential composition @xmath35 is defined as @xmath36 where we assume , up to a relabelling of the qubits in @xmath37 that @xmath38 . the parallel ( or tensor ) composition @xmath39 is defined as @xmath40 where we assume , up to a relabelling of the qubits in @xmath41 that @xmath42 .
the size of a command is the number of qubits affected by it .
notice that for any finite sums @xmath43 and @xmath44 of classical outcomes , @xmath45 , @xmath46 and @xmath47^\sigma$ ] are of size @xmath28 .
the _ size _ of a measurement pattern is the total size of all its commands .
the classical dependency between one - qubit operations is an important ingredient of the depth of a measurement pattern .
indeed , a correction of the form @xmath48 has to be applied after the measurement of the qubit @xmath49 .
the depth of a measurement pattern is longest path of dependant commands : for a given patterm @xmath50 , its quantum depth @xmath51 is defined as the longest sub - sequence @xmath52 of @xmath5 s.t . for any @xmath53 , @xmath54 , where @xmath55 , @xmath56 , @xmath57 and @xmath58^\sigma):=\{i\}\cup\{j\text { s.t .
$ s_j$ appears in $ \sigma$ or in $ \tau$}\}$ ] the measurement pattern @xmath59 @xmath60 is of size @xmath61 and depth @xmath62 ( @xmath63 is a dependant sub - sequence of size @xmath62 ) .
notice that the quantum depth does not take into account the depth of the classical side - processing coming from computation of the classical sums . as a consequence , the quantum depth of a measurement pattern
is based on the assumption that the classical computation is free , which can be motivated by the fact that the physical implementation of the quantum part of the computation is much more challenging than the classical part which can be considered at first approximation as free .
this assumption is discussed in details in section [ sec : assumption ] . for any measurement patterns @xmath64 and @xmath65 , @xmath66 .
moreover , @xmath67 and @xmath68 . since @xmath47^\sigma= m_i^\alpha x^\sigma_iz_i^\tau$ ] , any measurement pattern @xmath69 can be rewritten into a measurement pattern @xmath70 without dependant measurements such that @xmath71 and @xmath72 . as a consequence , w.l.o.g , we consider in the rest of the paper only measurement patterns without dependant measurements . in the following , we define some classes of complexity for measurement patterns .
we consider only _ uniform families _ , whose description can be generated by a log - space turing machine .
moreover we consider a fixed basis of measurement angles i.e. @xmath73 and an irrational multiple of @xmath74 .
@xmath75 contains decision problems computed exactly by uniform families of measurement patterns of input size @xmath76 , depth @xmath77 , polynomial size , and over a fixed basis .
let @xmath78 and @xmath79 contain decision problems computed with two - sided , polynomially small error .
the comparison of the computational power of quantum circuits and measurement patterns has been extensively studied @xcite .
indeed , since the introduction of the one - way model , the advantage of the one - way model in terms of depth complexity has been pointed out on some examples . in this section ,
we review the main results and state them in terms of complexity classes .
a quantum circuit is a sequence of quantum gates .
these gates are acting on three kinds of qubits : _ input _ , _ output _ and _ ancilla _ qubits . input and
output qubits may overlap .
a quantum circuit is a quadruplet @xmath80 , where @xmath1 is the set of all the qubits ( input , output and ancilla qubits ) , @xmath81 are sets of input and output qubits .
@xmath82 is a sequence of gates .
the size of a gate is the number of affected qubits , the size of a circuit is the total size of all its gates .
the depth is longest path of dependant gates .
a gate is a unitary operation applied on a bounded number of qubits .
we consider the following one- and two - qubit gates : @xmath83 , @xmath30 and @xmath7 ( see section 2 for a definition of @xmath7 ) : @xmath84 @xmath85 with @xmath86 an irrational multiple of @xmath74 is a universal family that can approximate any quantum gate with good precision @xcite . running a quantum circuit
consists in initialising the non input qubits in the @xmath87 state , then applying sequence of gates , and finally measuring the non output qubits in the standard basis .
we consider the classes of complexity @xmath88 and @xmath89 for quantum circuits @xcite : @xmath90 contains decision problems computed exactly by uniform families of quantum circuits of input size @xmath76 , depth @xmath77 , polynomial size , and over a fixed basis .
let @xmath91 , and @xmath92 contain decision problems computed with two - sided , polynomially small error .
measurement patterns can be translated into quantum circuits and vice versa .
[ lem : qc - to - mp ] any quantum circuit @xmath93 can be simulated by a measurement pattern @xmath69 of size @xmath94 and depth @xmath95 .
[ thm : mp - to - qc ] any measurement pattern @xmath69 can be simulated by a quantum circuit @xmath93 of size @xmath96 and depth @xmath97 .
these lemmas lead to the following inclusions of complexity classes : for any @xmath98 , @xmath99 @xmath100 lemma [ lem : qc - to - mp ] implies @xmath101 .
moreover , for any measurement pattern @xmath69 of input size @xmath76 and polynomial size , @xmath102 .
so , according to lemma [ lem : qc - to - mp ] , @xmath103 .
@xmath104 there is potentially a logarithmic depth separation between quantum circuits and measurement patterns , and indeed such a separation has been pointed out for the computation of the parity : @xmath105 @xmath106 is a so called clifford operation , a class of operations that can be computed in constant depth using measurements patterns @xcite . on the other hand , the depth of a quantum circuit for computing
@xmath107 i of depth @xmath108 @xcite .
it implies that @xmath109 .
such a separation is an open question for @xmath110 . in the next section ,
we characterise the computational power of the measurement patterns using a reduction to quantum circuits with unbounded fan - out .
measurement patterns and quantum circuits does not have the same computational power . in this section
we compare the computational power of the measurement patterns with a stronger version of the quantum circuits , the quantum circuits with unbounded fan - out .
this model of quantum circuits was introduced by hyer and spalek @xcite .
the original motivation for introducing such a model of quantum circuits with unbounded fan - out circuits is that in addition to parallelising operations acting on distinct qubits , commuting operations can also , in certain circumstances , be applied simultaneously .
see @xcite for details on quantum fan - out circuits .
a quantum circuit with unbound fan - out is a circuit with the usual gates and also fan - out gates which have unbounded input / output size : fan - out gate maps @xmath111 the depth and size are defined like for quantum circuits .
notice that a fan - out gate acting on @xmath76 qubits has size @xmath76 .
we consider the classes of complexity @xmath112 and @xmath113 for quantum circuits @xcite : @xmath114 contains decision problems computed exactly by uniform families of quantum circuits with unbounded fan - out of depth @xmath77 , polynomial size , and over a fixed basis .
let @xmath115 , and @xmath116 contain decision problems computed with two - sided , polynomially small error .
[ lemma : parityfan - out ] there exists a fan - out circuit of depth @xmath29 which implements the parity gate @xmath117 @xmath118 @xmath119 quantum fan - out circuit can be used to parallelising commuting operations if a basis change making these operators diagonal can be implemented efficiently : [ lem : commuting ] ( * ? ? ?
* theorem 1.3.19 ) for every set of pairwise commuting unitary gates , there exists an orthogonal basis in which all the gates are diagonal .
[ th : parallel ] @xcite let @xmath120 be pairwise commuting gates on @xmath121 qubits . gate @xmath122 is controlled by qubit @xmath123 .
let @xmath124 be a gate changing the basis according to lemma [ lem : commuting ] .
there exists a quantum circuit with fan - out computing @xmath125 having depth @xmath126 , size @xmath127 , and using @xmath128 ancillas . in the following we prove that the quantum depth of the measurement patterns and the depth of the quantum circuits with unbounded fan - out coincide .
first , we prove that measurement patterns are at least as powerful as quantum circuits with unbounded fan - out : [ lem : mp - qfoc ] any quantum circuit with unbounded fan - out @xmath93 can be simulated by a measurement pattern of depth @xmath95 and size @xmath129
. each gate of @xmath93 can be simulated by a measurement pattern .
indeed @xmath83 , @xmath130 , @xmath7 can be simulated by constant depth , constant size measurement patterns : @xmath131 since @xmath132 is equal to @xmath133 and since it exists a measurement pattern for @xmath107 of depth @xmath134 and size @xmath135 ( see section [ sec : mp - qc ] ) , @xmath132 can be simulated by a measurement pattern of constant depth and @xmath135 size . by sequential and parallel compositions
, @xmath93 is simulated by a measurement pattern @xmath69 of size @xmath129 and depth @xmath95.@xmath104 moreover , quantum circuits with unbounded circuits are at least as powerful as measurement patterns : [ lem : qfoc - mp ] any measurement pattern @xmath69 can be implemented by a quantum circuit with unbounded fan - out of depth @xmath136 and size @xmath137 . for a given measurement pattern @xmath50
, the sequence of commands @xmath5 can be rewritten into @xmath138 layers @xmath139 of depth @xmath28 such that @xmath140 . in the following
we show that each of these layers can be translated into a constant depth piece of quantum fan - out circuit . given a layer @xmath139 , since @xmath139 is of depth @xmath28
, it implies that each operation is acting on distinct qubits .
thus , up to some commutations , we assume w.l.o.g . that the commands of @xmath139 are performed in the following order : the entangling operations first , followed by measurements , @xmath26-corrections and finally @xmath141-corrections .
* the sub - sequence of @xmath139 composed of entangling operations is translated into a quantum circuit composed of @xmath7 .
the depth of this circuit is one since the entangling operations are acting on distinct qubits . * the sub - sequence of @xmath139 composed of measurements
is translated into a quantum circuit where each measurement @xmath142 is replaced by a @xmath143 .
the depth of this circuit is @xmath18 . * the sub - sequence of @xmath139 composed of @xmath26-corrections
is translated to a quantum circuit where each @xmath144 is replaced by @xmath145 .
this piece of circuit is not of constant depth , however , all the unitary transformations are diagonal , thus according to the theorem [ th : parallel ] , this piece of circuit can be simulated by quantum circuit with unbounded fan out of constant depth and size @xmath146 . * similarly , the sub - sequence of @xmath139 composed of @xmath141-corrections is translated into a quantum circuit with unbounded fan out of constant depth and size @xmath146 . in this case , since @xmath141 is not diagonal , @xmath83 is used as basis change in theorem [ th : parallel ] .
thus each layer @xmath139 is translated to a piece of circuit of constant depth and size @xmath147 , where @xmath17 is the size of @xmath139 . as a consequence
the whole pattern is translated into a quantum circuit with unbounded fan - out of detph @xmath136 and size @xmath137 .
@xmath104 the combination of lemmas [ lem : mp - qfoc ] and [ lem : qfoc - mp ] leads us to the main result of this paper .
it implies that measurement patterns and quantum circuits with unbounded fan - out have the same computational power : [ thm : same - power ] for any @xmath98 , @xmath148 @xmath149
in this section a general scheme of classically controlled quantum computation is considered .
one - way quantum computation is a special instance of this scheme as well as the teleportation based model @xcite , the state transfer model @xcite and the ancilla - driven quantum computation with twisted graph states @xcite .
a classically controlled scheme , generalization of the measurement calculus , is characterized by a quadruplet @xmath150 where @xmath151 is a set of quantum states ( for initialising ancillary qubits ) ; @xmath152 is a set of unitary transformations ; @xmath153 is a set of measurements with classical outcomes in @xmath154 ; @xmath155 is a set of corrections that are classically controlled by a sum modulo two of measurement outcomes .
for instance , the one - way model is a @xmath156 @xmath157-scheme ; the state transfer model
@xcite is a @xmath158-scheme ; the quantum circuit model is a @xmath159 @xmath160-scheme ; and the ancilla - driven model @xcite is a @xmath161 @xmath162-scheme .
size and depth of a classically controlled pattern are defined as for the measurement patterns , where every primitive operation has a constant depth and its size is the number of qubits affected by it .
the classes @xmath113 can be extended to unitary transformations , quantum states and quantum measurements as follows : a unitary @xmath163 is in @xmath164 if @xmath163 can be implemented by a quantum circuit with unbounded fan - out of depth @xmath121 ; a quantum state @xmath165 is in @xmath166 if there exists @xmath167 such that @xmath168 ; finally a linear map @xmath169 is in @xmath170 if there exists @xmath167 such that @xmath171 is diagonal . if @xmath172 is an observable ( self adjoint ) then @xmath169 is describing a measurement that can be implemented in depth @xmath173 by a circuit with unbounded fan - out : the measurement according to @xmath169 is transformed into a measurement in the standard basis thanks to the unitary @xmath174 such that @xmath171 is diagonal .
we show that among all the classically controlled schemes which commands can be implemented in constant depth , the measurement calculus is optimal in term of depth : [ thm : gen ] given a classically controlled scheme @xmath150 , if @xmath175 , @xmath176 , @xmath177 , @xmath178 , then any pattern @xmath179 of that scheme can be implemented by a measurement pattern @xmath69 of depth @xmath180 . we prove that any classically controlled pattern @xmath179 can be implemented by a quantum circuit of unbounded fan - out of depth @xmath180 , which can then be implemented by a measurement pattern @xmath69 of depth @xmath180 according to lemma [ lem : qfoc - mp ] .
the proof generalizes the proof of lemma [ lem : qfoc - mp ] .
@xmath179 is composed of @xmath181 layers @xmath182 .
notice that for each layer , the operations are acting on distinct qubits and then can be reorganised into subsequences of each type ( intialisation , unitary , measurement and correction ) . each of these subsequences can be implemented in constant depth : * the subsequences of @xmath183 composed of initialisations , unitaries and measurmeents can be implemented in constant depth since @xmath175 , @xmath176 , and @xmath177 ; * the subsequence of @xmath183 composed of corrections can also be implemented in constant depth .
indeed , since every correction of the subsequence are acting on distinct qubit , all these corrections are commuting .
thus we can apply theorem [ th : parallel ] , leading to a quantum circuit with unbounded fanout of constant depth since each @xmath184 is of constant depth and the gate @xmath124 changing the basis is of constant depth as well since @xmath185 . thus each layer is implemented by a constant depth piece of circuit , so @xmath179 is translated into a unbounded fan - out circuit of depth @xmath181.@xmath119 notice that the one - way , the teleportation - based , the state - transfer - based , and the ancilla - driven models are all classically controlled models for which theorem [ thm : gen ] applies . as a corollary , it exists a depth - preserving translation from the ancilla - driven model to the one - way model , answering an open question stated in @xcite .
the complexity classes of the quantum circuits with unbounded circuits have been studied and compared to other classes .
thanks to theorem [ thm : same - power ] , all known complexity results about quantum circuits with unbounded fan - out can be applied to the measurement patterns . among them , it is known that the quantum fourier transform ( qft ) is in @xmath186 @xcite , so qft is in @xmath187 .
whereas the one - way model have mainly been introduced as a promising model of physical implementation , it turns out that this model is very powerful : qft can be approximated in constant depth in the one - way model .
this result confirms that the semi - classical quantum fourier transform can be done efficiently @xcite .
moreover , factorisation is in @xmath188={\mathrm{rp}}[{\mathrm{bqmnc}}^0]$ ] @xcite , thus the factorisation can be approximated efficiently on a probabilistic machine which has access to a constant depth one - way quantum computer .
parallelisation in the quantum circuit model is based on the following assumption : ( _ i _ ) gates acting on distinct qubits can be performed simultaneously .
measurement patterns and quantum circuits with unbounded fan - out and have an extra power compared to quantum circuits because they are based on stronger assumptions .
the additional assumption for quantum circuits with unbounded fan - out is : ( _ ii _ ) commuting operations can be done simultaneously .
implies assumption @xmath189 whereas the additional assumption for measurement patterns is : ( _ iii _ ) classical part of any measurement pattern can be done in constant depth . in this section ,
we investigate assumption @xmath190 , and we compare assumptions @xmath191 and @xmath190 which lead to the same extra power .
we show that these two assumptions can be related , and that assumption @xmath190 is weaker than @xmath191 .
the classical part of a measurement pattern is reduced to the computation of sum modulo @xmath18 ( i.e. , parity ) @xcite in the dependant correction commands of the form @xmath192 .
thus , assumption @xmath190 can be rephrased as : _
boolean unbounded fan - in parity gates can be done in constant depth_. if boolean unbounded fan - in parity gates can be done in constant depth , any measurement pattern @xmath69 can be done in @xmath51 quantum layers of constant - depth interspersed by constant depth classical layers .
notice that without assumption @xmath190 , the classical parity gate on @xmath76 bits can be computed in depth @xmath193 .
thus , any measurement pattern @xmath69 can be done in @xmath51 constant - depth quantum layers interspersed by @xmath194-depth classical layers .
now we show that assumption @xmath190 is weaker than assumption @xmath191 .
indeed , assumption @xmath191 is that commuting operations can be done simultaneously , which implies that classical is classical if @xmath163 maps basis states to basis states in the computational basis . ]
commuting operations can be done simultaneously .
so , the classical parity can be implemented in constant depth using classical control - not which are commuting since target and controlled bits are not overlapping . in other words , whereas the assumption associated to the quantum circuits with unbounded fan - out is that commuting operations can be done in parallel , the classical version of this assumption is enough for the measurement patterns .
in this paper , we have shown that the measurement patterns has an equivalent computational power to the quantum circuits enhanced with unbounded fanout .
this characterises the power of the one - way model for parallelisation of algorithms and demonstrates that a number of quantum algorithms can be achieved in this model with a modest quantum depth .
the equivalence of these two models is at first sight surprising .
the parallelisation structure of the one - way model would appear to be radically different to the quantum circuit model .
nevertheless , our results indicate a close underlying connection .
there remain a number of open questions .
can the separation @xmath109 ( proved by broadbent and kashefi with the parity problem ) be proved for any @xmath121 : @xmath195 ? or at least for @xmath196 .
moreover , given the examples above , to what extent can quantum depth be minimised for generic quantum algorithms ?
recently , it has been shown that non - trivial quantum algorithms can be implemented with a single round of simultaneously applied commuting gates @xcite . moreover , in @xcite richard josza made the following conjecture : our results may have an implication for fault tolerant quantum computation .
thresholds for fault tolerant computation are typically derived under the assumption of a polynomial computational depth .
how could such fault tolerant models be relaxed if only constant or logarithmic quantum depth would suffice ?
we have focussed on classically controlled quantum computation where the classical control consists sums modulo two in our study .
these results applies to one - way quantum computation and various measurement - based models , like the teleportation - based and the ancilla - driven models .
but , other variants of measurement - based quantum computation with a different dependency structure @xcite could have a quite different characterisation .
we hope that this work motivates further study of this rich area . | we prove that one - way quantum computations have the same computational power as quantum circuits with unbounded fan - out .
it demonstrates that the one - way model is not only one of the most promising models of physical realisation , but also a very powerful model of quantum computation .
it confirms and completes previous results which have pointed out , for some specific problems , a depth separation between the one - way model and the quantum circuit model .
since one - way model has the same computational power as unbounded quantum fan - out circuits , the quantum fourier transform can be approximated in constant depth in the one - way model , and thus the factorisation can be done by a polytime probabilistic classical algorithm which has access to a constant - depth one - way quantum computer .
the extra power of the one - way model , comparing with the quantum circuit model , comes from its classical - quantum hybrid nature .
we show that this extra power is reduced to the capability to perform unbounded classical parity gates in constant depth . |
it has long been noted that most old halo stars show little magnetic activity relative to young disk stars indeed , joy ( 1947 ) used `` a weakening of emission lines '' as one of the criteria for identifying m subdwarfs . for disk stars ,
magnetic activity has been extensively studied from the onset of the convective envelope amongst f stars down to the hydrogen burning limit ( as reviewed by @xcite ) .
comparatively little is known about the activity of population ii stars recent studies of population ii stars have addressed chromospheric activity of single stars ( @xcite ) and x - ray emission in binary systems ( @xcite , hereafter ofp ) , but these studies targeted near - solar mass stars rather than very - low - mass stars .
nevertheless , the population ii m subdwarfs are worthy of study because stars with @xmath5 are thought to be fully convective for all metallicities ( @xcite ) , and therefore may have a different dynamo mechanism than solar mass stars which have radiative / convective zone boundaries ( @xcite ; @xcite )
. moderate resolution spectra of very - low - mass stars allow the identification of high activity ( `` dme '' ) stars by the presence of @xmath0 emission lines while at the same time allowing the classification ( @xcite , hereafter g97 ) of metal - poor stars as sdm ( m subdwarfs ) and esdm ( extreme m subdwarfs ) . in our recent palomar / michigan state university ( pmsu )
nearby - star spectroscopic survey ( @xcite ; @xcite ) , two of 2063 stars showed both @xmath0 emission and anomalously weak tio bands , indicating that these stars are sdme . the pmsu sample consisted of the known main - sequence m stars within 25 parsecs , and
therefore was largely made up of disk near - solar metallicity stars .
subsequently , g97 surveyed a sample of cool high - velocity stars , identifying an additional 27 sdm and 17 esdm .
including our recent spectra of previously unobserved faint high - proper - motion stars ( @xcite ; @xcite ) , we have observed a total of 68 sdm and esdm stars . only the original two subdwarfs show emission .
g97 classified gl 781 ( lhs 482 ) as sdm1.5e and gl 455 ( lhs 2497 ) as sdm3.5e . by fitting the synthetic spectra of allard and hauschildt ( 1995 ) , g97 showed that the sdm stars have @xmath3 \approx -1.2 \pm 0.3 $ ] .
we present additional observations of these two sdme systems aimed at understanding their nature and investigating the characteristics of their activity .
both stars prove to be close binary systems .
we present the data in section [ data ] , discuss both the unusual gl 781 system and the magnetic activity levels in section [ discussion ] , and summarize our conclusions in section [ conclusions ] .
spectra were obtained using the echelle spectrograph ( @xcite ) of the palomar 60-in . telescope .
the red cross - dispersing prisms were used , yielding wavelength coverage from @xmath6 to @xmath7 with gaps between orders beyond @xmath8 .
the 2 pixel resolution was @xmath9 .
gl 455 was observed on 31 december 1995 , 4 january 1996 , and twice on 6 january 1996 .
gl 781 was observed 23 times on ut dates 08 - 09 august 1996 ; four other observations were taken in july 1994 .
the echelle data were reduced on a sparc 5 running unix using the echelle figaro routines ( @xcite ) .
radial velocities were found by cross - correlation with the m dwarf standards from marcy & benitz ( 1989 ) .
comparison of other m dwarfs observed on the same observing runs to high precision radial velocities show that the accuracy is @xmath10 km s@xmath2 ; these uncertainties will be discussed in more detail in gizis _
( in prep . ) .
the radial velocities are given in table [ table - radvel ] .
no lines from a companion were seen in the spectrum of gl 781 ( figure [ figure - gl781 ] ) , which shows large velocity variations ( @xcite ) ; however , our monitoring yields a reliable period estimate . t. mazeh and d. goldberg kindly calculated orbital solutions using both june 1995 and august 1996 data . due to the large separation in time between the observations
, there are a number of different possible periods .
orbital solutions for each period are given in table [ orbital ] .
note that all periods lie within the range @xmath11 days .
we will refer to the observed , brighter sdm star as gl 781 a and the unseen , fainter companion as gl 781 b. the mass of gl 781 a is likely to be @xmath12 according to the models of dantona and mazzitelli ( 1996 ) and baraffe _ et al . _
( 1995 ) . with the observed mass function , @xmath13 for gl 781 b. the absorption lines in gl 781 a
are broadened compared to other m dwarfs observed with the same setup . to measure this broadening , we artificially broadened spectra of non - rotating stars obtained on the same night by convolving with the rotation profile ( gray 1992 ) .
we used the m1.5 v star gl 880 as the standard .
marcy and chen ( 1992 ) determined the upper limit on this star s rotational velocity to be @xmath14 km s@xmath2 .
the least squares fit of the atomic lines yields @xmath15 km s@xmath2 . to rule out the possibility of metallicity differences causing the broadening
, we have repeated this procedure using the inactive sdm1.5 star lhs 64 .
lhs 64 does not show measurable broadening relative to gl 880 , and fitting the gl 781 spectra to artificially broadened lhs 64 spectra again yields @xmath16 km s@xmath2 .
we note , however , that stauffer and hartmann ( 1986 ) measured @xmath17 km s@xmath2 for gl 781 a. stauffer and hartmann also measured @xmath17 km s@xmath2 for gl 82 and gl 268 .
we also have spectra of these stars using the same spectrograph setup gl 781 a s absorption lines are clearly broader .
we therefore prefer our own measurement .
the @xmath0 and @xmath18 emission lines show significant variability .
the @xmath0 line varies between @xmath19 and @xmath20 in the august 1996 spectra , with no dependence on orbital phase . in the july 1994 spectra , @xmath0
is seen between @xmath21 and @xmath22 .
previous low resolution observations ( @xcite ) observed @xmath0 in a stronger state ( @xmath23 ) .
the @xmath18 region is observed at low signal to noise , but @xmath18 is definitely also variable with equivalent width of @xmath24 when @xmath25 .
based on low - resolution spectrophotometry , obtained with the palomar 60-in .
telescope and spectrograph in july 1996 , we estimate the continuum flux level as @xmath26 ergs @xmath27 s@xmath2 @xmath28 at @xmath0 and @xmath29 ergs @xmath27 s@xmath2 @xmath28 at @xmath18 .
these indicate a line flux between 1.1 and @xmath30 ergs @xmath27 s@xmath2 for @xmath0 , and a balmer decrement of 1.8 for the @xmath25 observation .
the he d3 line was detected with observed equivalent width of @xmath31 .
the he triplet at @xmath32 was also marginally detected with an equivalent with of @xmath33 ( this line is difficult to measure as it lies near a tio bandhead ) .
our four echelle spectra of gl 455 clearly indicate it is a double - lined spectroscopic binary , but do not provide enough data to determine the orbital parameters . in the first two spectra ( 31 dec . and 4 jan . ) , the absorption lines from both components are resolved , with the cores separated but the wings overlapping .
the other two spectra ( 6 jan . ) do not show line doubling .
the small shift ( @xmath34 km s@xmath2 ) seen for the observations separated by two hours suggests the period is much longer than gl 781 s period .
we note that the velocity data are consistent with a @xmath35 or @xmath36 day period .
it should also be noted that since the lines are never completely separated we can not rule out the possibility that there is a ( fainter ) third star in this system ; g97 showed that the two components of gl 455 appear to lie above the sdm sequence in the hr diagram even after correcting for binarity .
we do not detect rotational broadening in this system , indicating that @xmath37 km s@xmath2 . since our resolution is insufficient to detect rotational broadening in most dme ( @xcite ) , this limit on the rotation
is not surprising .
if the period is 8 days , then we estimate that the equatorial rotational velocities are @xmath38 km s@xmath2 .
the @xmath0 emission of gl 455 ab is not resolved , so the reported equivalent widths are for the combined spectrum . in the december 1995 spectrum
, the @xmath18 line is resolved sufficiently to show that both components are in emission .
significant variability is present : the four equivalent widths of @xmath0 are @xmath39 and @xmath18 are @xmath40 .
our earlier moderate - resolution spectrum ( @xcite ) had an @xmath0 equivalent width of less than @xmath41 .
we therefore adopt the @xmath0 equivalent width of @xmath42 as a `` typical '' value for the discussion of this system s activity level .
our july 1996 spectrophotometry indicates that the continuum flux near @xmath0 for the two unresolved components is approximately @xmath43 ergs @xmath27 s@xmath2 @xmath28 .
we therefore estimate the `` typical '' @xmath0 flux to be @xmath44 ergs @xmath27 s@xmath2 for each star under the assumption of equal - luminosity components . we have searched the rosat all - sky survey ( @xcite ) to determine if the sdme are x - ray sources .
we match gl 781 with 1rxs j200503.8 + 542609 and gl 455 with 1rxs j120219.1 + 283507 . the hardness ratio ( hr1 ) given in the catalog is nearly identical with the hardness ratio ( hr ) used by schmitt _ et al .
_ 1995 to study nearby dm and dme stars .
we therefore determine the count rate - to - energy conversion factor ( cf ) using their equation @xmath45 the derived x - ray fluxes and other stellar parameters used in this paper are given in table [ fluxes ] .
the observed mass function of the system and the theoretically estimated mass of gl 781 a imply that gl 781 b is more probably more massive ( @xmath46 ) .
together with the fact that no lines are seen , this mass implies that gl 781 b is a faint , cool white dwarf .
the line strengths , colors , and absolute magnitude of gl 781 a are similar to other sdm , so we estimate that the white dwarf is @xmath47 magnitudes fainter , i.e. @xmath48 . ordinary white dwarfs ( @xmath49 )
have ages @xmath50 years at this luminosity ( @xcite ; we have used the liebert _
_ 1988 bolometric corrections ) .
as shown below , the most likely possibility is that gl 781 b is a @xmath51 helium white dwarf the corresponding lower limit on the age is @xmath52 years ( @xcite ) .
the detection of rotational broadening serves as a constraint on the possible system parameters . assuming that the gl 781 a s rotational period is the same as that of the system , @xmath53 the radius of the star is not known but can be estimated , either from stellar interior models , or by using the bolometric corrections and effective temperatures from model atmospheres ( @xcite ) as fit by g97 . using the latter method
, we estimate @xmath54 and @xmath55 , so @xmath56 .
this estimate is in good agreement with stellar interior models ( @xcite ) .
we then find that we expect @xmath57 to be @xmath58 km s@xmath2 , compared to the observed @xmath59 of @xmath60 km s@xmath2 , suggesting the system is viewed nearly edge on , with @xmath61 .
if @xmath62 is indeed near unity , then the white dwarf companion has an unusually small mass .
gravitational redshift measurements show that the distribution of white dwarf masses in wide binaries is strongly peaked at @xmath63 ( @xcite ) .
this peak mass can not be reconciled with the radius of gl 781 a deduced from the observed luminosity and effective temperature .
thus , the observations require that gl 781 b is about half the mass of the typical relatively isolated white dwarf .
the small present day separation of the two stars , @xmath64 , implies that gl 781 a was engulfed when gl 781 b became a red giant .
( recall that the present day brighter component , the sdm , was then the secondary by both mass and luminosity ) .
there was therefore a phase of common envelope evolution . since the mass of gl 781 a is @xmath12 and the original mass of gl 781 b was @xmath65 ( since it has already evolved and cooled ) , it is clear that the mass ratio was greater than @xmath66 when the gl 781 b became a red giant and first engulfed the companion .
iben & livio ( 1993 ) have reviewed this situation and shown that a common envelope will form and that the secondary ( gl 781 a ) will not accrete a significant amount of mass . after being engulfed , gl 781 a caused mass loss from gl 781 b , possibly ending b s evolution as a he white dwarf with relatively low mass ( @xmath67 ) instead of allowing b to become a co white dwarf ( @xmath68 ) .
this would be consistent with the estimate of @xmath62 from the detected rotational broadening .
the calculations of iben _ et al . _
( 1997 ) indicate that for systems with 12 hour periods , helium white dwarfs with very - low - mass ( @xmath69 ) main sequence companions should be about seven times more common than co white dwarfs with very - low - mass main sequence companions . for m dwarfs , membership in a very close ( @xmath70 days )
binary system is thought to be a sufficient condition for high chromospheric activity rapid rotation at relatively old age is maintained by tidal interactions between the two stars ( @xcite ; @xcite ) .
both gl 455 and gl 781 prove to be short period binaries , with gl 781 a having detectable rotation , and we therefore attribute their activity to their binary nature .
gl 781 a s companion is not likely to affect the activity ( except by being the cause of gl 781 a s rapid rotation ) . although red dwarfs near hot white dwarfs show @xmath71 emission due to reprocessing of the white dwarf s euv flux in the chromosphere of the red dwarf ( @xcite ) , gl 781 b s low luminosity ( @xmath72 ) implies that it is not a significant source of euv flux . furthermore ,
gl 781 a s emission is not stronger from the hemisphere facing the white dwarf , since there is no correlation between the strength of @xmath0 with orbital phase in the august 1996 data .
the white dwarf is therefore not causing the emission by incident radiation or other line - of - sight effects .
we also attribute the x - ray flux to a normal corona generated by gl 781 a s rapid rotation , rather than accretion onto the white dwarf , since at present gl 781 a is only about @xmath73 of the size of its roche lobe as estimated by eggleton s ( 1983 ) formula .
we note also that the x - ray flux is not due to some sort of `` basal '' emission unique to subdwarfs since none of the known sdm and esdm without @xmath0 emission are detected in x rays by rosat .
metal - poor stars are usually associated with an extremely old ( @xmath74 yr ) population . preston _
( 1996 ) , however , have identified young , a - type metal - poor main - sequence stars with large velocities which they suggest are the residue of a merger with a satellite galaxy ( or galaxies ) .
presumably these stars have lower mass counterparts , and if the m subdwarfs have emission lifetimes comparable to field m dwarfs ( @xcite ) , then these young m subdwarfs will show detectable @xmath0 emission
. we can not directly rule out either gl 455 or gl 781 as members of this young , metal - poor population .
the cooling time for the white dwarf in gl 781 is at least 2 gyr , which is too long to maintain the observed rapid rotation if m subdwarfs lose angular momentum like field m dwarfs .
overall , the binary nature of the two systems appears to be sufficient to explain their activity .
the properties of sdme relative to ( near - solar metallicity ) dme are important because they provide a unique perspective on the generation of strong activity .
gl 781 a and gl 455 ab are the only known highly active , metal - poor , cool main sequence stars .
gl 781 a is of particular interest because its high rotational velocity is similar to the rotational velocities observed in the pleiades cluster members of the same mass ( @xcite ) . in comparing the properties of the sdme , it should be be remembered that considerable dispersion exists in the chromospheric and coronal properties of even coeval , homogeneous samples such as the pleiades and hyades clusters , and that even larger dispersion exists in field samples .
the physical causes of this dispersion are unknown .
thus any relations based upon only two systems must be viewed with caution .
we first note that the data suggest that the chromospheres and coronae of the sdme are quite similar to disk dme stars , despite the great metallicity difference between the two classes .
the sdme x - ray hardness ratios , which are sensitive to the coronae temperatures , are comparable to the field dme with similar x - ray luminosities
. the ratios of @xmath0 to @xmath18 fluxes are similar to that seen in field dme stars ( @xcite ; @xcite ) .
the strength of the he d3 line is normal compared to the @xmath0 line , as is the strength of the he 6678 line compared to the he d3 line .
the correspondence between this sdme and the dme suggest that the physical properties ( e.g. , temperature ) of the lower chromosphere ( traced by the balmer lines ) , upper chromosphere ( helium lines ) , and corona ( x - ray ) do not depend strongly on metallicity .
we measure the overall activity of each star using the ratios of the x - ray luminosity ( @xmath75 ) and the @xmath0 luminosity ( @xmath76 ) to the star s bolometric luminosity .
we focus on gl 781 a since we know the orbital period and @xmath77 , allowing a comparison to disk dme stars with similar rotation rates .
gl 455 a and b are more slowly rotating than gl 781 a , but the period is unknown .
jones _ et al . _
( 1996 ) have measured @xmath77 for m dwarfs in the pleiades using the keck i telescope .
the nine stars with @xmath78 , similar to gl 781 a s mass of @xmath12 , have measured velocities in the range 8 km s@xmath79 km s@xmath2 .
these results suggest that we can directly compare the metal - poor star gl 781 a to a population of stars with similar masses and rotation rates .
although this means two important parameters are the same , there are three notable differences between gl 781 a and the pleiads : first , gl 781 a s age is most likely 10 - 15 gyr , and certainly is 2 gyrs or more which is @xmath80 times the age of the pleiades cluster ; second , gl 781 a is metal - poor ( @xmath3 \approx -1.2 $ ] ) whereas the pleiades have near - solar metallicity ; third , gl 781 a s companion might induce internal and surface differential rotation different from that of single stars ( @xcite ) . direct comparison to the pleiad sample observed by jones _
is hampered by the small sample size and lack of x - ray data for most of their stars .
we instead compare to the much larger survey of the pleiades cluster reported by hodgkin _
( 1995 , hereafter hjs ) .
the hjs survival analysis of both detections and upper limits indicates that stars with masses near @xmath81 ( @xmath82 ) have _ mean _ @xmath83 and @xmath84 of @xmath85 and @xmath86 . the range in activity is large , as some detected stars have @xmath87 while many stars have upper limits indicating @xmath88 .
gl 781 a s ratios are quite to close to the mean values , and on this basis the activity of a rapidly - rotating metal - poor star appears to be virtually identical to its more metal - rich counterparts although factors of @xmath89 in the log of the activity level can not be ruled out . on the other hand , gl 781 a
is an order of magnitude weaker in x - ray activity than the strongest m dwarfs in the pleiades . as noted by hjs , the reason why these sources are stronger is unknown .
indeed , it interesting to note that many of the slowest rotators in the jones _ et al .
_ sample have the strongest @xmath0 emission an expanded data sample would be of considerable interest to establish the nature of a rotation - activity relation for this stellar mass .
we finally note that the hyades cluster offers another point of comparison to the sdme sample .
again , the _ mean _ ratios of @xmath4 and @xmath90 for the hyades m dwarfs ( @xcite ) are similar to the ratio seen in gl 781 a. we may also compare the activity of the sdm binaries to that of nearby field dm binaries . like the m subdwarfs , these m dwarfs have activity induced by the rapid rotation caused by very close companions .
we have detected a further 16 sb2 m dwarf binaries using the same echelle configuration as used for the current observations ( @xcite ) .
none of these systems have @xmath77 as great as gl 781 a ; all are likely to have orbital periods less than 5 - 10 days . matching these objects with the rosat
all sky survey , we find that four have @xmath91 whereas twelve have @xmath92 ( with four systems at the maximum of @xmath93 ) . if rotation were the most important parameter , one might expect gl 781 a to be as active in x - rays as the most active nearby binaries . in this case ,
gl 781 a is deficient by about 0.4 in the log ( a factor of 2.5 ) .
it should also be recalled that the range in observed @xmath0 equivalent widths is 1.0 to 1.9 , which suggests that some of the difference in activity level could be due to gl 781 a happening to be in a relatively weak state at the time of the rosat observations .
( 1995 ) have found that over a decade m dwarfs , unlike the sun , do not show variability in x - rays in excess of a factor of 2 .
one of the most striking aspects of the sdme data is that both systems show similar x - ray activity levels despite their different rotation rates .
ab certainly has a much slower rotation rate than gl 781 a , indeed , @xmath94 is much weaker in gl 455 ab .
the simplest explanation is that the gl 455 ab x - ray observation is affected by a flare or other transient event .
there are a few m dwarfs in the hyades ( @xcite ) which have @xmath95 for non - simultaneous measurements , so gl 455 ab s relative activity levels are not unprecedented . comparing the sdme to the near - solar mass hyades binaries detected by stern
_ et al . _
( 1995 ) , gl 781 a has an x - ray activity level similar to that of the gk - dwarf ( @xmath96 ) binaries with periods of a few days but is less active than the half - day period binary v471tauri .
given the large mass differences , we conclude merely that this example suggests that it is not unreasonable to expect that gl 781 a s half - day period should put it amongst the most active binaries and therefore that gl 781 a may be mildly x - ray subluminous .
there is , however , considerable evidence that the rotation - activity relationship is complex in low - mass stars ( indeed , hawley _
_ 1996 questioned the existence of a correlation between rotation rate and activity level in m dwarfs ) .
( 1997 ) have found that for a sample of hyades m dwarfs ( @xmath97 ; above the fully - convective limit and more massive than the sdme ) that there is no single activity - rotation relation .
they propose that their data can be explained as a `` bifurcation '' with two different rotation - activity relations . on one relation , relatively slowly rotating stars ( @xmath98 km s@xmath2 ) can be as active ( @xmath99 ) as stars ( on the other relation ) with @xmath100 km s@xmath2 .
they suggest that the difference is due to differing rotational evolution histories which results in differing radial and latitudinal differential rotation .
they also find that stars with companions as distant as 1 a.u .
have enhanced activity relative to single stars with the same rotation rate .
in addition , there is evidence that the most rapid ( @xmath101 hr . )
gk rotators may weaken in activity level ( @xcite ) .
clearly , there are many parameters relevant in determining the activity level these or others may be important for gl 781 a. we conclude that gl 781 a does not show evidence for a significant ( i.e. , factor of 10 ) weakening in its activity levels relative to near - solar metallicity stars .
it may , however , be subluminous in x - rays for its bolometric luminosity by up to a factor of @xmath102
. this may be the result of its metal deficiency .
definite conclusions are prevented by the large , and as yet poorly - understood , dispersion in the activity levels of m dwarfs .
the relatively small effect on the x - ray emission due to a factor of @xmath103 deficiency in metals suggests that composition is not a major contributor to the dispersion in activity levels observed for m dwarfs , which are all relatively close to solar metallicity ( @xmath3 \gtrsim -0.6 $ ] , @xcite ) .
the ofp survey of spectral type fgk metal - poor binaries has shown that a sample of population ii short - period binaries analogous to rs cvn binaries are at least one order of magnitude less x - ray luminous than the disk rs cvn binaries , although some population ii binaries are as luminous as the rs cvn binaries . they have suggested that the x - ray line emission is reduced due to the relative lack of metals in the corona of population ii systems with @xmath104 < -0.4 $ ] .
our sdme lie in the low - metallicity end of that range .
based upon our discussion above , we conclude that the sdme are not as x - ray subluminous as expected by analogy with the ofp results .
we have obtained spectra of two halo very - low - mass metal - poor systems that show @xmath0 and x - ray emission . both prove to be close binary systems , which accounts for their high activity level .
we deduce that gl 781 a s unseen companion is a cool white dwarf .
the present - day orbital period of 12 hours implies that the system passed through a phase of common envelope evolution .
gl 455 ab consists of two near - equal luminosity sdme .
the strength of the x - ray coronae of the sdme relative to the @xmath0 chromosphere is well within the range observed in dme stars in the field and in young clusters .
gl 781 a s level of activity , as measured by the ratio @xmath4 , is comparable to the mean activity level in the pleiades .
the comparison to both the pleiades and field dme binaries suggest that gl 781 a is no more than a factor of @xmath102 subluminous in x - rays due to its metallicity .
this is contrast to the factor of ten observed in most ( but not all ) stars in the ofp survey of more massive metal - poor binaries . with only two sdme systems ,
these results are necessarily uncertain .
unfortunately , it will be difficult to significantly increase the sample of sdme and esdme since the two systems discussed here are the only such binaries detected amongst a total of 68 lhs catalog sdm and esdm stars observed from palomar ( @xcite ; @xcite ; @xcite ) . a more optimistic view , however , is that activity due to rapid rotation could allow the identification of , or place limits upon the existence of , short - period binaries and/or relatively young ( few gigayear ) halo populations given a large spectroscopic sample .
i would like to thank the staff of palomar observatory and in particular skip staples for assistance with the observations .
jim mccarthy gave valuable tutorials on his echelle data reduction software .
t. mazeh and d. goldberg kindly computed the orbital solutions .
neill reid and suzanne hawley gave helpful comments .
mark giampapa s referee review substantially improved the paper .
i gratefully acknowledge support by greenstein and kingsley fellowships as well as nasa grants go-06344.01 - 95a and go-05913.01 - 94a . this research has made use of the simbad database , operated at cds , strasbourg , france .
rrrr 2449559.787284 & 70.59 & 2450304.002925 & -134.41 2449560.793864 & 58.07 & 2450304.022841 & -150.44 2449561.789240 & 57.89 & 2450304.661833 & -47.03 2449561.856991 & -37.40 & 2450304.707263 & 20.89 2450303.661066 & -61.09 & 2450304.757906 & 75.85 2450303.685817 & -23.23 & 2450304.801232 & 87.79 2450303.710209 & 16.40 & 2450304.806716 & 86.17 2450303.715577 & 22.64 & 2450304.811984 & 85.73 2450303.742898 & 58.07 & 2450304.816111 & 82.87 2450303.773024 & 77.66 & 2450304.872800 & 34.81 2450303.821616 & 84.31 & 2450304.925684 & -39.83 2450303.881348 & 32.56 & 2450304.975199 & -111.42 2450303.932413 & -42.01 & 2450305.021252 & -152.47 2450303.973081 & -102.64 crrrrrrrcrrrrrlr [ orbital ] a & 0.49603807 & -33.97 & 123.31 & 0.0136 & 212 . &
2450193.974 & 0.0965 @xmath106 & 0.00000057 & 0.40 & 0.63 & 0.0043 & 17 .
& 0.024 & 0.0015 b & 0.49636949 & -34.06 & 123.31 & 0.0123 & 212 .
& 2450194.397 & 0.0966 @xmath107 & 0.00000050 & 0.35 & 0.56 & 0.0038 & 17 . &
0.024 & 0.0013 c & 0.49670134 & -34.15 & 123.29 & 0.0109 & 212 . &
2450194.323 & 0.0966 @xmath108 & 0.00000049 & 0.34 & 0.54 & 0.0037 & 19 .
& 0.026 & 0.0013 d & 0.49703363 & -34.23 & 123.27 & 0.0095 & 211 .
& 2450194.249 & 0.0966 @xmath109 & 0.00000052 & 0.36 & 0.57 & 0.0039 & 23 . &
0.031 & 0.0013 e & 0.49736636 & -34.31 & 123.24 & 0.0081 & 210 . &
2450194.174 & 0.0966 @xmath110 & 0.00000059 & 0.42 & 0.66 & 0.0045 & 31 .
& 0.043 & 0.0015 f & 0.49769953 & -34.38 & 123.20 & 0.0066 & 208 .
& 2450194.097 & 0.0966 @xmath111 & 0.00000070 & 0.49 & 0.77 & 0.0052 & 45 . &
0.062 & 0.0018 g & 0.49803314 & -34.45 & 123.16 & 0.0051 & 204 .
& 2450194.018 & 0.0966 @xmath112 & 0.00000082 & 0.58 & 0.91 & 0.0061 & 70 . &
0.097 & 0.0021 cccc [ fluxes ] v & & 11.98 & 12.85 v - i@xmath113 & & 1.99 & 2.47 d & pc & 16.6 & 20.3 @xmath114 & & 10.88 & 12.06 @xmath115 & & 9.58 & 10.50 luminosity & @xmath116 ergs s@xmath2 & 4.6 & 2.0 @xmath117 & k & 3600 & 3500 radius & r@xmath118 & 0.28 & 0.20 mass & @xmath119 & 0.25 & 0.15hr1 & & -0.41 & -0.74 @xmath120 & @xmath121 ergs @xmath27 s@xmath2 & 6.60 & 3.91 @xmath75 & @xmath122 ergs s@xmath2 & 22 & 9.7 @xmath123 & @xmath124 & 1.3 & 0.6 & & -0.6 & -1.1 & & -3.9 & -4.4 & & -3.3 & -3.3 @xmath77 & km s@xmath2 & 30 & @xmath125 @xmath126 & days & 0.5 & 8 ? | we present spectroscopic observations of two halo m subdwarfs which have @xmath0 emission lines .
we show that in both cases close companions are the likely cause of the chromospheric activity in these old , metal - poor stars .
we argue that gl 781 a s unseen companion is most likely a cool helium white dwarf .
gl 455 is a near - equal - mass m subdwarf ( sdm ) system .
gl 781 a is rapidly rotating with @xmath1 km s@xmath2 . the properties of the chromospheres and x - ray coronae of these systems are compared to m dwarfs with emission ( dme ) .
the x - ray hardness ratios and optical chromospheric line emission ratios are consistent with those seen in dme stars .
comparison to active near - solar metallicity stars indicates that despite their low metallicity ( @xmath3 \approx -1.2 $ ] ) , the sdme stars are roughly as active in both x - rays and chromospheric emission .
measured by @xmath4 , the activity level of gl 781 a is no more than a factor of 2.5 subluminous with respect to near - solar metallicity stars . |
latest developments in cosmology are provided by deep redshift surveys for cosmological structures .
these structures develop from evolution of primordial perturbations generated during inflation .
originally , quantum fluctuations of the inflation field set the seeds of curvature perturbations at primordial epochs .
the nature of these fluctuations are well studied in the cosmic microwave background ( cmb ) .
the fingerprints of these fluctuations on large scale structures ( lss ) are also important since the current and future galaxy surveys will take high precision measurements of galaxy clustering .
this provides enough statistical precision to solve the issues in the standard model of cosmology .
the importance of lss is highlighted by the fact that lss surveys are three dimensional , whereas the cmb is two dimensional .
we study the fingerprints of long modes that come from inflation using the relativistic approach to large scale observables .
+ to interpret galaxy and lss survey observations correctly , one should note that as the light propagates through the cosmos , its path is modified by inhomogeneities .
therefore , one needs to use general relativistic calculations to relate the observables to the inhomogeneities . since the exact relativistic calculation for cosmological structure observables is complicated @xcite , usually the first or the second order perturbations of the einstein equations
are used to calculate these observables .
some of these effects on the observed fluctuations of galaxies are the dark matter density fluctuations , the redshift - space distortions ( the peculiar velocity effect in the redshift space of light ) and the magnification bias which are studied in @xcite . the standard method to derive the full relativistic expression for galaxy clustering
is done by tracing back the photon path given the observed redshift and the angular position of the source galaxies .
a fully general relativistic expression for the observed galaxy density contrast includes volume distortions due to the light deflections , evolving number densities , galaxy bias , as well as the magnification bias generalized to the evolving luminosity function .
these relativistic effects are usually named projection effects .
+ generally , there are different wavelengths for the gravitational potential that come from the early universe fluctuations .
the long wavelength potential can produce a superhorizon perturbation that might have observable effects on the cmb or on other large scale structure observables .
the power asymmetry in the cmb as observed by planck satellite @xcite ( for earlier reports of hemispherical asymmetry in wmap data see @xcite ) can be the generic property of some early universe models .
this issue has been recently revisited in works such as @xcite and @xcite which use a local - variance estimator .
the observations show that the power spectrum in the northern hemisphere is different from the power spectrum in the southern hemisphere .
erickcek et al .
@xcite have proposed that a superhorizon perturbation would introduce a preferred direction that generates the power asymmetry . along this way , the predictions of inflationary models with a long mode modulation of large scale structures are presented in @xcite . as a matter of fact
, it was shown that a long constant @xcite and a long gradient mode @xcite can be gauge artifacts of the perturbation theory which does not leave any effect on the cmb @xcite .
note that the vanishing contribution of a gradient to the cmb is non trivial .
now , our aim is to study the non trivial effects of long modes which could be possibly observed in the future galaxy surveys .
our goal in this paper is to extend the investigation of the long gradient modes effects to other large scale observables like the galaxy number counts considering the subtleties that arise from the bias .
+ given the high precision of lss surveys and some hints of the cmb asymmetry from long mode perturbations , we study the imprints of these long modes on other large scale observables as well .
the cancellation of the pure gradient modes in the galaxy number counts is nontrivial as various new effects appear compared to the cmb case .
specifically , we show that the contribution of a pure gradient mode to large scale observables vanishes . using this frame work
, we show that the coupling of the long gradient modes to short modes does not modulate the smoothed matter variance at a fixed physical scale if some form of consistency condition is satisfied .
therefore , we confirm the consistency of the definition ( [ halo ] ) .
+ the outline of the paper is as follows .
we begin with a review of the long mode effects in the perturbation theory in section ii .
we also discuss the importance of extending the previous studies to other large scale observables .
the study of large scale observables is presented in two sections .
we calculate the three large - scale observables in the presence of long gradient mode in section iii , after reviewing the relativistic derivation of large scale observables .
the main section is section iv .
it is devoted to the study of the galaxy clustering quantities such as the galaxy number counts and the halo bias in presence of a long gradient mode .
finally , our conclusion and discussion are presented in section v. the latin indices indicate the space components and greek indices indicate the space time components and we have set @xmath0 .
in cosmological perturbation theory , gauge invariant quantities are defined with the assumption that perturbations fall off at infinity .
this allows to decompose perturbations as scalars , vectors , and tensors ( svt ) .
examples of perturbations that do not vanish at infinity are zero momentum modes and pure gradient modes . + * zero momentum modes : * in the perturbation theory zero momentum modes are important , because they modulate the power spectrum in the case of single field slow - roll inflation as derived in @xcite .
this modulation is known as the consistency condition .
these zero momentum modes will induce a local type non - gaussianity .
it is argued that the consistency relations are true in models of inflation in which the only dynamical field is the inflaton field @xcite .
these consistency relations are also derived in the newtonian limit in @xcite .
although this type of non - gassainity is small and proportional to @xmath1 , it is crucial to know if our large scale observations are really contaminated by such effects . along this line
, it is shown that even after fixing the gauge , zero momentum transformations are still allowed @xcite .
this remaining gauge freedom is shown to be the source of ir divergences @xcite .
that is because one needs boundary conditions to uniquely solve for the lapse function after imposing the conventional comoving gauge condition .
it is shown that zero momentum modes in the metric can be removed by a coordinate transformation , so they do not have any observable effect as they induce a relative shift in the expansion history .
the vanishing effect of a constant mode in the power spectrum is emphasized in @xcite . + * gradient modes : * let us suppose that we have a long wavelength perturbation expanded as @xmath2 in the fourier space , where @xmath3 , @xmath4 and @xmath5 are the scale factor , the hubble parameter and the fourier mode for the long mode respectively .
this perturbation appears in the expansion of primordial perturbations of the form @xmath6 .
the average of such perturbations should vanish in the whole universe in order to keep homogeneity and isotropy .
we suppose that the effect of such a perturbation does not vanish in our hubble patch .
it is also shown that pure gradient modes are removed by a coordinate transformation in the metric @xcite . in the newtonian gauge
this transformation is given by @xmath7 where @xmath8 and @xmath9 are defined as @xmath10 that is because the effect of a pure gradient mode is an acceleration which is not observable locally .
+ in this manner , we extend the gradient mode studies to other large scale observables including cosmic rulers , galaxy number counts and the halo bias .
the vanishing effect of gradient modes on observables is not trivial given that various relativistic terms appear ( equation ( [ galaxy - p ] ) ) .
a consistent general relativistic formulation of galaxy clustering should take into account all effects such as displacements , velocities and other new relativistic effects naturally .
the especial case of the halo bias needs to be treated in a different way , because we need to consider coupling to short modes .
we will show that in the general relativistic formulation , the effects of a gradient mode vanish if some forms of consistency condition hold .
+ to study gradient mode effects , one needs to define the observables .
this is done in the relativistic frame work for observables .
the observables are gauge invariant as a prior and made of information on the past light cone .
they should to be devoid of gauge artifacts as well .
a long gradient mode is gauge artifact .
the observables are described by the standard clocks , the standard rulers and the galaxy number counts @xcite .
one example of such a clock is the cmb on large scales .
next section is devoted to study of effects of long gradient modes on observables .
+ the other cosmological observable is the galaxy number count .
we also study the gradient mode effects on galaxy number counts .
galaxy number counts importance is highlighted by the fact that future surveys will probe the cosmos on large volumes with unprecedented precision .
because of their significant statistical precision , they will be able to discriminate between inflationary models by the level of non - gaussianity .
it is crucial to consistently formulate all the effects in order to correctly interpret the data .
other relativistic formulations have recently emerged @xcite .
the authors in @xcite use the evolution equation for jacobi map to derive the galaxy number counts .
we found that our result of the galaxy number counts agrees with the results given in @xcite . in next sections ,
we study the imprints of gradient modes using relativistic definition for large scale observables .
one of the interesting observable quantities in the lss is the galaxy number density fluctuation .
this quantity is measured in the redshift space .
it is well known that the observed power spectrum for galaxy number density is modified by the peculiar velocities as described by kaiser formula @xcite . however ,
for large sky surveys this formula is hindered by the fact that this formula does not include all the relativistic terms consistently .
basically , the problem of deriving the relativistic correction terms starts by tracing photons paths in an inhomogeneous universe .
since photons are traveling through inhomogeneities , their path will deviate from the straight path in the frw universe and these inhomogeneities change the observed size of rulers , like the baryon acoustic oscillations ( bao ) or the power spectrum peak and survey volume .
in addition , on large scales , gauge effects become important and should be taken into account . hence , a gauge invariant formulation is important . for example , galaxies are thought to trace the underlying dark matter overdensity and are biased by a scale independent factor @xmath11 with respect to the dark matter overdensity .
this definition is obscured because changing the gauge will introduce scale dependencies .
in addition , observables should have physical interpretation
. we will first review how the relativistic effects are derived .
the details of derivations can be found in @xcite .
we only give the main points of derivation here to introduce the notation .
we will then show that the gradient mode contribution to all cosmic ruler perturbations vanishes identically .
+ an observer screen space is a two dimensional space orthogonal to his velocity @xmath12 and the photon four - vector @xmath13 .
the observer sees the cosmos images by projecting on to his own instantaneous screen space spanned by orthonormal basis @xmath14 and @xmath15 .
these are also orthogonal to the observer s velocity @xcite .
+ when a photon is observed at the redshift @xmath16 ( observed redshift ) , its apparent position is attributed as @xmath17 . in this frame work ,
the apparent position is always computed assuming a homogeneous universe .. refers to apparent quantities . ]
note that @xmath17 also depends on direction at the observer .
the photon trajectory at the observed redshift @xmath16 is deviated from a similar trajectory in a homogeneous universe .
the actual position of the source is obtained by integrating the geodesic equations for the photons in an inhomogeneous universe as shown in fig.([fig:1 ] ) .
these deviations are gauge dependent and unobservable .
because null geodesics are conformally invariant , it is simpler to use conformally transformed metric and affine parameter for the photons . in this way the photon geodesics have simple expressions .
but it is important to use the correct metric in the expressions ( [ proj1 ] , [ redshift ] ) .
the photon geodesics is solved for the source position in @xcite .
we present the final result as @xmath18 where @xmath19 is the conformal affine parameter for null geodesics in the background and @xmath17 is apparent position evaluated at the observed redshift @xmath20 please note that ( @xmath21 ) is used for apparent values of any quantity and apparent values are evaluated using the background equations .
the first term is given by the geodesic equation ( [ geo - eq1 ] , [ geo - eq2 ] ) and the second term is given by the redshift matching equation ( [ redshift - eq ] ) .
the perturbed conformal photon four - vector is also @xmath22 .
the metric in a general gauge is written as @xmath23 typically , the observer fixes the scale factor at his observation time by @xmath24 , while the global scale factor is simply @xmath25 where @xmath26 is the observer s proper time and the subscript @xmath27 refers to the observer .
note that to first order , proper time is not affected by the peculiar velocity of the observer .
we use the conformal newtonian gauge in which the metric is given by @xmath28 in the matter dominated einstein - de sitter ( eds ) universe where there is no anisotropic stress , we have @xmath29 . in this frame work ,
the scale factor difference between the global scale factor @xmath30 and the @xmath31 is @xmath32 because this expressions are correct up to first order , @xmath3 refers to the global scale factor in our notation . before discussing cosmic rulers , it is helpful to mention the main points of deriving the equation ( [ res ] ) .
as the observations are in observer s rest frame , the initial conditions by requiring that in observer s frame the frequency and the direction of the photons are simply @xmath33 where @xmath34
are the orthonormal tetrads given by @xmath35 and @xmath36 where @xmath37 is the tetrad index and @xmath38 refers to space time indexes . in tetrads definition , @xmath39 refers to the conformal velocity of the observer . note that the factor @xmath40 appears as we need the physical photon four - vector @xmath41 in order to project on the observer s frame . in this way
, the perturbation at observer s position is fixed .
we have defined @xmath42 and @xmath43 in which @xmath44 is the projection operator .
] given the initial conditions , integration of the geodesic equations yield temporal and spatial displacements @xcite @xmath45 where @xmath46 and dot denotes derivative with respect to the conformal time @xmath47 .
note that equation ( [ a - def ] ) is also used .
we stress that that @xmath48 is calculated at the observed redshift .
this shifts the real value of @xmath48 by @xmath49 .
the value of @xmath49 is derived by redshift matching .
the redshift for the photons is defined by @xmath50 where @xmath51 is the real emission time .
the scale factor perturbation with respect to the apparent scale factor @xmath52 is defined as @xmath53 .
this is given by @xmath54 the equation ( [ redshift ] ) yields @xcite @xmath55 .
now that we have reviewed the notation and main points , we move on to the observables in the next section .
the cosmic rulers are the cosmological observables that their spatial scale @xmath56 is known like the cmb on large scales . because we know the scale when photons decoupled , in principles , this scale can be measured in a statistical manner .
other cosmic rulers include the known scales in the matter power spectrum or the baryon acoustic oscillations ( bao ) @xcite .
moreover , the angular diameter distance perturbations for cosmic rulers are related to the luminosity distance perturbations by @xmath57 @xcite ( equation ( [ m ] ) ) .
these are used for studying the nature of dark energy @xcite .
the fully relativistic approach to cosmic rulers can be found at @xcite .
here we review the main points .
basically , the size of cosmic rulers which is measured is their apparent size .
the cosmic rulers size change from @xmath56 to the observed apparent size @xmath58 .
because length is a frame dependent quantity , one needs to define the observer who actually measures the length of the rulers .
one natural way is to define their length in the instantaneous rest frame of the comoving observers .
the velocity of comoving observers is given by @xmath59 and their rest frame metric is simply @xmath60 .
provided that the size of the ruler is small , its apparent length yields @xmath61 whereas in the observer rest frame , its physical length is @xmath62 where @xmath63 is the apparent distance between to points of the ruler and @xmath64 is given by the equation ( [ res ] ) . finally , the relative ruler perturbation is defined as @xmath65 where @xmath66 @xcite . to decompose the apparent separation @xmath67 to parallel and perpendicular to line of sight components , the projection operator @xmath68 is introduced . that is the perpendicular components of separation @xmath69
are defined as @xmath70 .
the parallel component is simply given by @xmath71 @xcite and note that the definitions of @xmath72 , @xmath73 and @xmath74 will be given below .
we move on to study the effect of a pure gradient mode on these observables and show that a pure gradient gives vanishing contribution to these observables .
this confirms that consistency of this relativistic approach .
the scalar @xmath74 includes the line of sight perturbations of the cosmic rulers and induces the perturbations in the redshift space .
it also includes the redshift space distortion term which is the dominant term on small scales .
the first term @xmath74 is given by @xmath75 @xcite . + to address the problem of consistency of derivation , we calculate the effect of a pure gradient mode on @xmath74 . in the eds universe
a pure gradient mode is @xmath76 . in the newtonian gauge , the scalar @xmath74 yields @xmath77 where @xmath78 and @xmath79s .
we use the fact that @xmath80 and @xmath81 in the first line .
we also use @xmath82 .
please note that in the last line we need @xmath83 to zero order .
that is we have @xmath84 for the normalization of scale factor .
this means that @xmath85 . as a result
, there is consistency in derivation of the line of sight perturbations .
the second term @xmath86 is basically vector .
this term is determined by @xmath87 where @xmath88 @xcite .
this term produces perturbations both in the line of sight and perpendicular to the line of sight .
this terms can be written in the spin basis as @xmath89 where @xmath90 and @xmath91 and @xmath92 are the basis on the sphere . + to study the effect of the gradient mode , a simple calculation yields @xmath93 .
again the effect of the pure gradient to @xmath94 vanishes identically .
similar to the former case , the @xmath73 term is not modified by the pure gradient mode . the last term @xmath95 in the equation ( [ rulers ] ) is the most crucial term , since this term includes we observe in shear surveys . this term is given in a general gauge as @xcite @xmath96 the magnification is the trace part of @xmath95 . in conformal newtonian gauge , magnification is @xmath97 where @xmath98 and @xmath44 .
note that lensing convergence ( @xmath99 ) is not gauge invariant .
hence , it is not observable .
it has been shown that the magnification is not perturbed by gradient modes @xcite .
the trace free part of the @xmath95 term is called shear , @xmath100 .
the shear term can be written in the spin basis as @xmath101 . in the newtonian gauge in the spin basis
the shear term yields @xmath102 since the shear involves the second derivatives of the metric , it is not modulated by the gradient mode .
consequently , surveys like weak lensing surveys which measure the shear are not contaminated .
the magnification which produces area perturbations is measured in the lensing surveys and surveys which probe the luminosity like supernova surveys .
it is related to angular and luminosity distance by @xmath103 @xcite .
therefore , the pure gradient will also induce no asymmetry in standard candle observations since its contribution to the magnification vanishes .
one of the important observable quantities in the large scale structure surveys is the galaxy number count .
similarly , we stress that the effect a gradient mode should vanishes naturally in a consistent derivation . along this line
, we also study the bias .
since one observes tracers and not the underlying matter field , a general relativistic definition for bias is essential .
one problem is the correct gauge to define bias . it is argued that in the synchronous gauge the conventional definition for bias is correct @xcite
this is in contrast to the advocated gauge in @xcite that chooses constant redshift gauge for the bias .
the synchronous gauge is also proposed for a second order calculation of the galaxy clustering @xcite . + suppose we have a comoving source with four velocity @xmath104 . in the instantaneous source
rest frame , a volume element is defined by @xmath105 .
the number of galaxies expressed in the observed coordinates is @xmath106 where @xmath107 is the proper number density of the galaxies .
this approach to relativistic galaxy number counts is used by @xcite . we know that the number of galaxies in real space is equal to the number of galaxies in redshift space .
this yields @xmath108d^{3}\tilde x=\int \tilde{a}^{3}\bar{n}_{g}^{p}(z)(1+\delta_{g}(z,\tilde{x}))d^{3}\tilde{x},\end{aligned}\ ] ] where @xmath109 is the number density perturbation and @xmath110 .
note that it is assumed that @xmath111 .
thus , the observed number of the galaxies in the observed coordinates is given by @xmath112 where @xmath98 and @xmath113 .
this formula is the relativistic generalization of kaiser formula which includes the new relativistic corrections .
these corrections are dominant on large scales where general relativity and other models deviate from each other .
basically , these corrections should be considered in the future surveys to interpret the observations . in this frame work , we show that pure gradient mode modifications vanish non trivially . on the other hand , on large scales ,
observations are constrained by the cosmic variance .
detectability of these new relativistic corrections is discussed in @xcite .
the method proposed to overcome cosmic variance is given by multi tracer method which uses the fact that different biased tracers trace the same underlying density field @xcite .
consider the case of a matter dominated universe with sources and observers comoving with the cosmic fluid . in the presence of the long mode
, we have @xmath114 and @xmath115 . in the newtonian gauge each term in equation ( [ galaxy - p ] )
is given by @xmath116 the equation ( [ galaxy - p ] ) yields @xmath117 consequently , the long mode does not modify the number count observations .
this confirms the consistency of the relativistic derivation .
+ in addition , as we observe galaxies and not the underlying dark matter density field , we need to know the halo bias to relate @xmath118 in the synchronous gauge to the dark matter perturbation @xmath119 .
the other complexity is that we observe on constant redshift surfaces . given the difference between constant time surfaces and constant redshift surfaces , number density perturbation yields @xmath120 .
the @xmath121 is the cosmic clock perturbation @xcite .
it is important to note that the cosmic clock perturbation by a gradient mode vanishes as @xmath122 where we have @xmath123 and it is assumed that observers are comoving with the cosmic fluid @xmath78 .
the aim in this section is to show that a gradient mode does not contribute to the halo bias ( equation ( [ halo ] ) ) if some form of consistency condition ( equation ( [ ps ] ) ) is satisfied . in other words , gradient modes in single field inflation models
do not induce non - gaussian corrections to the bias .
it is known that the bias in presence of local non - guassianities is modulated @xcite .
this modulation increases as the scale increases .
this is contrary , as equivalence principle requires that effects of very long modes on local scales vanish a correct bias modulation naturally includes equivalence principle .
in addition , general relativity can also induce local non - gaussianities because it couples long modes to shorts modes @xcite .
it is then very crucial to discriminate between primordial non - gaussianities and gravity induced non - gaussianities .
our goal is to use the gradient mode to argue that bias definition given by equation ( [ halo ] ) .
what we do is to circumvent second order calculations by using local transformations . to this end
, we first derive the consistency condition for a gradient mode .
+ in this section we use the @xmath124 gauge .
constant time hypersurfaces have the same proper time in this gauge . the spatial part of the metric in this gauge is given by @xmath125 a gradient mode @xmath126 can be produced by a coordinate transformation of the form @xmath127 note that this transformation does not change the gauge .
the metric takes the new form @xmath128 where @xmath129 . in this metric
, finding long - short effects needs a second order treatment .
note that we neglect terms of second order in @xmath5 because we only study long modes to short modes couplings .
in other words , if we have a metric of the form @xmath130 the transformation @xmath131 will change the metric to a flat frw only locally in patches where @xmath132 . a different approach used for constant modes
is to use a local transformation , named conformal fermi coordinates , adapted to long modes @xcite .
this locally changes the metric to a frw metric .
the extent of validity of our transformation and conformal fermi transformation are the same .
+ in the spherical collapse model , halos are formed when the over density of the collapsing regions exceeds a threshold value @xmath133 .
the perturbation in the number density of halos , @xmath134 , is biased by the presence of the long modes perturbations because it changes the threshold to @xmath135 . in press - schechter method
the number density of halos depends on the height parameter @xmath136 where @xmath137 is the smoothed matter variance @xcite ( some more discussion is in this paper draft @xcite ) .
if there is no coupling of long modes and short modes , the halo number density is simply modeled by @xmath138 . however , gravity couples long modes to short modes . in this case , the halo number density perturbation is also a function of the long mode @xmath139 .
one model which includes the effects of long modes to short modes couplings is @xmath140 where @xmath137 is the smoothed matter variance at the scale @xmath141 @xcite .
using the equation ( [ halo ] ) , we confirm that the bias is not modulated by gradient modes .
first , we need to know how the gravitational coupling of long modes to short modes modulate the power spectrum to understand how the bias changes . + here
, we derive the power spectrum with short modes and long modes coupling corrections .
we first write the power spectrum in local coordinates @xmath142 and then transform this back to @xmath143 coordinates .
the correlation function for the matter perturbation in the @xmath144 coordinate where the pure gradient mode is removed is written as @xmath145 where @xmath146 means the correlation function in the presence of the gradient mode . after substituting the equation ( [ cor ] ) and integration by parts , we find that the power spectrum in fourier space is modulated as @xmath147 this is the consistency condition for gradient modes which is not derived in @xcite as we have taken into account terms of order @xmath148 in the consistency condition .
note that this result is essentially second order .
+ the power spectrum modulation ( equation [ ps ] ) is because of the left freedom to define new coordinates locally after gauge fixing . here , we show that these effects should vanish in a different way compared to @xcite . in this way
, we measure the variance in a the same scale in local coordinates .
that is our local rulers are also perturbed .
this is justified as halo formation is a spherical collapse which is a local process .
we will show that @xmath149 does not change in a fixed physical scale .
this approach and @xcite are similar in the sense that the coordinate transformations change the metric to a flat frw metric locally .
+ the smoothed matter variance is defined as @xmath150 where @xmath151 is the window function and @xmath152 can be substituted from the equation ( [ ps ] ) .
the length scale @xmath141 is the proper length in @xmath143 coordinates .
this variance is measured in @xmath143 coordinates where the gradient mode is present .
fixing the physical scale , using local metric to measure the radius , is equivalent to a transformation in fourier space .
the transformation from to @xmath153 to @xmath154 is given by @xmath155 this transformation keeps the extent of the window function the same . doing this transformation yields @xmath156 where @xmath157 is the jacobin of the transformation .
the new radius @xmath158 which is now measured in @xmath142 coordinates , is basically where spherical collapse calculations are done .
the @xmath158 is calculated keeping @xmath159 the same after using the equation ( [ f - eq ] ) .
note that in the integral one can replace @xmath143 with @xmath142 to first order in @xmath5 .
finally , the matter variance is @xmath160 this will precisely cancel the change in the power spectrum which we previously calculated in the equation ( [ ps ] ) . note that the halo number density depends on the matter variance . since we showed that @xmath161 is not modified , the equation ( [ halo ] ) is nominated by general relativity considerations .
+ as we discussed , the fact that long constant modes can be produced by coordinate transformations is due to consistency condition .
however , they will produce a squeezed bi - spectrum by modulation of power spectrum through gravitational coupling in the coordinates where long mode is present . because we derived a consistency condition for the gradient modes in terms of modulation of the power spectrum
, we expect their effect vanish in local coordinates . here
, we use a different transformation than @xcite and required the same physical scale in local coordinates . as halo formation
is a local phenomena we argue that our transformation is sufficient for capturing the local effects .
+ on large scales , there are other complexities since gauge effects become important .
changing the gauge will produce scale dependencies in the bias .
thus , one needs to find a relativistic definition for the bias .
the synchronous gauge is shown to be the right gauge @xcite . since
halo formation is a local phenomena , it only depends on the dark matter perturbations in this gauge .
this gauge has been implemented for writing the relativistic galaxy clustering .
a relativistic bias has also been defined using fermi normal coordinates and it is shown that it reduces to the usual bias on small scales @xcite . + finally , we compute the effect of the pure gradient mode on the observed power spectrum
. we can compute the intrinsic power spectrum in a chosen gauge and transform it back to the redshift space using the equations in @xcite .
this equation is given by @xmath162 where @xmath163 since @xmath164 and @xmath165 do not change by the pure gradients , the observed large scale power spectrum is not modified .
our aim in this paper is two fold .
first , we emphasize that the effect of a gradient mode on observables should vanish , providing that the derivation is consistent . in this manner
, we extend the investigation of the long mode modulations to other large scale observables taking into account projection effects that are inherent in a consistent relativistic formulation of the large scale observables . in the relativistic formulation ,
defining observables is crucial .
these observables are gauge invariant quantities made of information on the past light cone .
the measured value of these observables is different because of the inhomogeneity effects . as an example , the cosmic rulers are one observable with a known physical scale like the bao . their observed scale can be compared to their physical scale .
the observables in the cosmic rulers can be presented in terms of a scalar , a vector and a tensor quantity .
the relativistic formulation of observables yields that the effects of gradient modes vanishes on all cosmic ruler observables .
the other important quantity in the relativistic formulation is the galaxy number counts .
additionally , we show that different projection effects cancel each other as well , using the relativistic galaxy number counts .
+ second , we emphasize that the effects of a gradient mode on the bias vanishes in a consistent bias formulation .
it is important to note this is possible even though there is a coupling of gradient modes to short modes .
that is non - gaussian corrections induced by gravitational coupling vanish . since the halo number density depends on the matter variance , we first derived a consistency condition for how the matter power spectrum is modulated in the presence of a long gradient mode .
as halo formation is a local phenomena , we used local coordinates to define the matter variance .
this yields that the matter variance does not change in local coordinates .
this confirms that the definition ( [ halo ] ) is consistent with equivalence principle .
+ the vanishing effect of gradient modes is favored by the equivalence principle .
it is shown that the local physical effects of long mode perturbations start at @xmath166 order if one uses conformal fermi coordinates as implied by the equivalence principle @xcite . as a result
, the pure gradient mode does not contribute to the tidal term at linear order .
however , to calculate the observables at late times , we have to transform from the conformal fermi coordinates to the observed coordinates .
the gradient mode should vanish in a consistent general relativistic formulation .
+ * acknowledgments : * + we would like to thank hassan firouzjahi , ali akbar abolhasani and reza mansouri for useful discussions and comments .
+ g. f. r. ellis , class .
* 28 * , 164001 ( 2011 ) ; t. buchert and s. rsnen , ann . rev .
nucl . part .
* 62 * , 57 ( 2012 ) ; m. p. mood , javad t. firouzjaee and r. mansouri , phys .
d * 88 * , 083011 ( 2013 ) ; rahim moradi , javad t. firouzjaee , reza mansouri , class .
grav . *
32 * , no .
21 , 215001 ( 2015 ) ; reza javadinezhad , javad t. firouzjaee and reza mansouri , arxiv:1510.04429 [ gr - qc ] ; javad t. firouzjaee and t. feghhi , arxiv:1608.05491 [ gr - qc ] .
a. kehagias , a. m. dizgah , j. norea , h. perrier and a. riotto , jcap * 1508 * , no .
08 , 018 ( 2015 ) [ arxiv:1503.04467 [ astro-ph.co ] ] .
a. r. ade _ et al .
_ [ planck collaboration ] , astron .
astrophys . * 571 * , a23 ( 2014 ) [ arxiv:1303.5083 [ astro-ph.co ] ] .
y. akrami , y. fantaye , a. shafieloo , h. k. eriksen , f. k. hansen , a. j. banday and k. m. grski , astrophys .
j. * 784 * , l42 ( 2014 ) doi:10.1088/2041 - 8205/784/2/l42 [ arxiv:1402.0870 [ astro-ph.co ] ] .
p. a. r. ade _ et al .
_ [ planck collaboration ] , arxiv:1506.07135 [ astro-ph.co ] .
astrophys .
j. * 784 * , l42 ( 2014 ) [ arxiv:1402.0870 [ astro-ph.co ] ] .
astrophys .
j. * 699 * , 985 ( 2009 ) [ arxiv:0903.1229 [ astro-ph.co ] ] .
a. kehagias and a. riotto , nucl .
b * 873 * , 514 ( 2013 ) [ arxiv:1302.0130 [ astro-ph.co ] ] .
m. mirbabayi and m. zaldarriaga , jcap * 1503 * ( 2015 ) 03 , 056 [ arxiv:1409.4777 [ astro-ph.co ] ] .
p. creminelli and m. zaldarriaga , jcap * 0410 * , 006 ( 2004 ) [ astro - ph/0407059 ] .
n. kaiser , mon . not .
227 * , 1 ( 1987 ) .
j. m. maldacena , jhep * 0305 * ( 2003 ) 013 [ astro - ph/0210603 ] .
e. barausse , s. matarrese and a. riotto , phys .
d * 71 * , 063537 ( 2005 ) [ astro - ph/0501152 ] .
j. yoo , phys .
d * 90 * , no .
12 , 123507 ( 2014 ) [ arxiv:1408.5137 [ astro-ph.co ] ] .
e. pajer , f. schmidt and m. zaldarriaga , phys .
d * 88 * , no .
8 , 083502 ( 2013 ) [ arxiv:1305.0824 [ astro-ph.co ] ] .
l. dai , d. jeong and m. kamionkowski , phys .
d * 88 * , no . 4 , 043507 ( 2013 ) [ arxiv:1306.3985 [ astro-ph.co ] ] . y. urakawa and t. tanaka , phys .
d * 82 * , 121301 ( 2010 ) [ arxiv:1007.0468 [ hep - th ] ] .
f. schmidt and d. jeong , phys .
d * 86 * , 083527 .
m. bruni , j. c. hidalgo and d. wands , astrophys . j. * 794 * , no .
1 , l11 ( 2014 ) doi:10.1088/2041 - 8205/794/1/l11 [ arxiv:1405.7006 [ astro-ph.co ] ] .
a. allahyari and j. t. firouzjaee , arxiv:1510.05531 [ astro-ph.co ] .
d. bertacca , n. bartolo , m. bruni , k. koyama , r. maartens , s. matarrese , m. sasaki and d. wands , class .
grav . * 32 * , no .
17 , 175019 ( 2015 ) doi:10.1088/0264 - 9381/32/17/175019 [ arxiv:1501.03163 [ astro-ph.co ] ] . | we extend the study of long mode perturbations to other large scale observables such as cosmic rulers , galaxy number counts and halo bias .
the long mode is a pure gradient mode that is still outside observer s horizon .
we insist that gradient mode effects on observables vanish .
it is also crucial that the expressions for observables are relativistic .
this allows us to show that the effects of a gradient mode on the large scale observables vanishes identically in a relativistic frame work . to study the potential modulation effect of the gradient mode on halo bias , we derive a consistency condition to the first order in gradient expansion .
we find that the matter variance at a fixed physical scale is not modulated by the long gradient mode perturbations when the consistency condition holds .
this shows that the contribution of long gradient modes to bias vanishes in this frame work . |
there are many important results in the unitary representation theory of lie groups related to analytic continuation . here
a key ingredient is the special case where @xmath17 is a strongly continuous unitary one - parameter group and @xmath18 its self - adjoint infinitesimal generator , i.e. , @xmath19 in the sense of measurable functional calculus .
then the unitary one - parameter group @xmath20 extends to a holomorphic one - parameter semigroup @xmath21 if and only if @xmath18 is bounded below .
this extension then restricts to a locally bounded non - degenerate hermitian one - parameter semigroup @xmath22 .
conversely , any such hermitian one - parameter group has a self - adjoint infinitesimal generator @xmath23 and @xmath24 then yields an extension to @xmath25 , where the boundary values form a unitary one - parameter group .
the key point of this picture is that self - adjoint operators @xmath18 whose spectrum is bounded below can be viewed as the infinitesimal generators of two objects : a unitary one - parameter group ( stone s theorem ) and a hermitian one - parameter semigroup of bounded operators ( hille yosida theorem ) .
this is the one - parameter context of what we are dealing with in the present paper for banach lie groups .
we call a pair @xmath0 consisting of a banach lie group @xmath4 and an involutive automorphism @xmath1 of @xmath4 a _ symmetric banach lie group_. we decompose its lie algebra @xmath26 into @xmath27-eigenspaces of @xmath1 and write @xmath28 for the identity component of the group of @xmath1-fixed points in @xmath4 .
let @xmath29 be an open convex @xmath9-invariant cone for which the polar map @xmath30 is a diffeomorphism onto an open subsemigroup @xmath31 of @xmath4 .
then @xmath32 is an involutive semigroup with respect to the involution @xmath33 which is called an _
olshanski semigroup_. in appendix [ a ] we explain how to obtain such semigroups @xmath34 , where @xmath35 is a connected lie group locally isomorphic to @xmath36 for which @xmath35 and @xmath34 need not be contained in a lie group ( cf .
ii.13 ) ) .
our first main result ( theorem [ thm:3.6 ] , proved in section [ sec:4 ] ) is the following banach version of the lscher
mack theorem ( @xcite ) : let @xmath37 be a non - degenerate strongly continuous @xmath12-representation of @xmath32 which is _ smooth _ in the sense that the space @xmath38 of smooth vectors is dense . if @xmath15 is the simply connected lie group with lie algebra @xmath39 , then there exists a smooth unitary representation @xmath40 of @xmath15 on @xmath41 which is uniquely determined by the requirement that the unitary one - parameter groups corresponding to elements @xmath42 are those obtained by `` extending '' @xmath43 to @xmath36 and the generators of the one - parameter groups corresponding to elements of @xmath44 are of the form @xmath45 , where @xmath46 is the infinitesimal generator of the hermitian one - parameter semigroup @xmath47 . for finite dimensional groups
this result is due to m. lscher and g. mack ( @xcite ; see also @xcite ) .
as their methods make heavy use of coordinates obtained from products of one - parameter groups , we have to develop a completely new approach in the banach context .
actually our approach is more direct and uses only quite general methods , such as the criteria for the integrability of infinitesimal unitary representations of banach lie algebras from @xcite .
the one - parameter case described above corresponds to @xmath48 , @xmath49 , @xmath50 and @xmath51 .
an interesting , much less involved , special case arises if @xmath52 is the whole group .
then our assumption is that @xmath4 has a diffeomorphic polar decomposition @xmath53 and our theorem establishes a correspondence between @xmath12-representations @xmath54 by bounded operators and norm - continuous unitary representations of @xmath15 .
if @xmath4 is a finite dimensional semisimple lie group and @xmath1 a cartan involution , then this is weyl s well - known unitary trick relating finite dimensional representations of @xmath4 to unitary representations of the compact group @xmath15 . in the context of the lscher mack theorem , it is a natural question which unitary representations of @xmath15 are obtained from representations of a semigroup @xmath55 . to answer this and related questions , we consider for a smooth unitary representation @xmath40 of a lie group @xmath4 the convex function @xmath56 we call @xmath20 _ semibounded _
if @xmath57 is bounded in the neighborhood of some point @xmath58 , and if @xmath59 is a convex cone , then we say that @xmath20 is _ @xmath8-semibounded _ if @xmath57 is locally bounded on @xmath8 . now
the converse to the lscher
mack theorem ( corollary [ c : conv ] ) asserts that a representation of @xmath15 is obtained from a smooth representation of @xmath32 if and only if it is @xmath60-semibounded . here
the main difficulty is to show that , for an @xmath60-semibounded unitary representation of @xmath15 , the prescription @xmath61 define a representation and to see that it is actually smooth .
here we use methods developed previously in @xcite and @xcite . with the same tools we obtain the following holomorphic extension theorem ( cf .
@xcite and @xcite for the finite dimensional case ) : for every open invariant cone @xmath59 and a corresponding complex olshanski semigroup @xmath62 , each @xmath8-semibounded unitary representation of @xmath4 extends to a holomorphic representation of @xmath62
. in the finite dimensional context lawson s theorem guarantees the existence of olshanski semigroups @xmath55 under quite simple requirements on the spectra of the operators @xmath63 , @xmath64 ( cf .
@xcite , @xcite ) .
since lawson s arguments involve local compactness in a crucial way , they do not generalize to the banach context .
however , natural examples of olshanski semigroups arise as compression semigroups of bounded symmetric complex domains in banach spaces , symmetric hilbert domains , and real forms of such domains ( cf .
@xcite ) .
as we show in @xcite , the real forms @xmath65 of symmetric hilbert domains @xmath66 , i.e. , the fixed points for an antiholomorphic involution @xmath67 , are of particular interest in representation theory ( see @xcite for a classification ) . here
@xmath15 is a central extension of the identity component of @xmath68 , which leads to the class of _ hermitian lie groups _ whose semibounded representations are classified in @xcite . choosing a base point in @xmath65
leads to an involution @xmath1 on @xmath15 with @xmath69 , and conjugation with @xmath67 induces another involution on @xmath15 .
the complex domain @xmath66 has a natural compression semigroup of form @xmath70 , where @xmath71 is an open invariant cone ( @xcite ) .
the subgroup @xmath72 is invariant under @xmath1 and @xmath73 is a real symmetric space .
the corresponding semigroup is @xmath55 for @xmath74 . here
our lscher
mack theorem provides a bridge between @xmath12-representations of @xmath55 and unitary representations of the group @xmath15 .
a remarkable feature of the infinite dimensional context is that a large class of the irreducible separable continuous unitary representations of @xmath36 extend to contraction representations of @xmath55 with the same commutant and hence further to representations of the larger group @xmath15 .
in particular we obtain an automatic extension of a large class of irreducible representations of @xmath36 to irreducible semibounded representations of @xmath15 . for irreducible domains , the irreducible semibounded representations of @xmath15 are classified in @xcite , and this in turn leads to a classification of the corresponding representations of @xmath36 .
one of the central motivations to study results like the lscher
mack theorem is that there are natural sources of involutive representations of real olshanski semigroups @xmath55 . here
the constructions based on `` reflection positivity '' are of particular interest because of their connections with euclidean , resp . , relativistic quantum field theories ( cf .
@xcite , @xcite , @xcite , @xcite ) . in this context
the semigroup @xmath55 constitutes a bridge between unitary representations of the groups @xmath4 and @xmath15 .
however , these semigroups do not exist in all situations , where the passage from @xmath4 to @xmath15 is of interest , a typical example is the euclidean motion group @xmath75 and the poincar group @xmath76 .
this was the motivation for frhlich , osterwalder and seiler to introduce the concept of a virtual representations of a symmetric lie group ( @xcite ) .
it would be very interesting to see if a suitable variant of this concept can be developed for banach lie groups .
if @xmath41 is a hilbert space , we write @xmath77 for the algebra of bounded linear operators on @xmath41 , @xmath78 for its group of units , and @xmath79 for the unitary group .
if @xmath4 is a topological group and @xmath41 a complex hilbert space , then a _ unitary representation of @xmath4 on @xmath41 _ , denoted @xmath40 , is a homomorphism @xmath80 which is continuous with respect to the strong operator topology , i.e. , all orbit maps @xmath81 are continuous .
if @xmath4 is a banach lie group , then we write @xmath3 for its lie algebra .
it is a banach lie algebra , i.e. , a banach space with a continuous lie bracket . if @xmath32 is a semigroup , we write @xmath82 and @xmath83 for left and right multiplications on @xmath32 . if a neutral element in @xmath32 exists , it is denoted by @xmath84 .
let @xmath85 be a set .
a kernel function @xmath86 is called _ positive definite _ if for @xmath87 , @xmath88 , the matrix @xmath89 is positive definite .
for any such kernel there exists a unique hilbert subspace @xmath90 of functions on @xmath85 with continuous evaluation maps given by @xmath91 , @xmath92 .
in particular , the subspace @xmath93 , spanned by the functions @xmath94 , @xmath95 , is dense in @xmath96 .
if @xmath97 is a function with values in a hilbert space whose range is total in the sense that it spans a dense subspace and @xmath98 for @xmath99 , then we have a unitary map @xmath100 ( cf .
i.1.6 ) ) .
let @xmath101 be a banach manifold , @xmath102 be a smooth positive definite kernel on @xmath101 and @xmath90 be the corresponding reproducing kernel hilbert space . according to ( * ? ? ?
7.1 ) , the map @xmath103 is smooth , so that @xmath104 for @xmath105 implies that @xmath106 .
let @xmath107 be a vector field on @xmath101 , @xmath108 the associated derivation of @xmath109 , and @xmath110 be its local flow , defined at @xmath111 for @xmath112 . the lie derivative on functions
is given by @xmath113 we also consider @xmath108 as an unbounded operator on @xmath96 defined on the domain @xmath114 we say that @xmath107 is _ symmetric with respect to @xmath115 _ ( or _ @xmath115-symmetric _ ) , if @xmath116 the key point in the preceding definition is that it can be expressed in terms of the kernel and the vector field , but the following proposition shows that we can draw interesting conclusions for the corresponding unbounded operator on @xmath96 .
[ prop:2.2 ] let @xmath107 be a @xmath115-symmetric vector field .
then , for every @xmath111 , @xmath117 in particular @xmath118 and the operator @xmath119 is symmetric .
if @xmath120 then @xmath121 since @xmath122 , @xmath123 is smooth , the derivative @xmath124 exists in @xmath96 and , for every @xmath125 , @xmath126 for every @xmath120 we therefore have @xmath127 note that implies in particular @xmath128 below we recall frhlich s theorem on unbounded symmetric semigroups as it is stated in ( * ? ? ?
actually frhlich assumes that the hilbert space @xmath41 is separable , but this is not necessary for the conclusion . replacing the assumption of weak measurability by weak continuity , all arguments in @xcite work for non - separable spaces as well .
thm:2.4 let @xmath36 be a symmetric operator defined on the domain @xmath129 dense in the hilbert space @xmath41 .
suppose that for every @xmath130 there exists @xmath131 such that the equation @xmath132 has a solution satisfying @xmath133 and @xmath134 for @xmath135 .
then the operator @xmath36 is essentially self - adjoint and @xmath136 .
we apply frhlich s theorem to our geometric operator @xmath108 from proposition [ prop:2.2 ] .
the new feature is that we can even describe the closure of the essentially self - adjoint operator we obtain .
t : froelichgeo let @xmath101 be a banach manifold and @xmath115 be a smooth positive definite kernel .
then @xmath96 consists of smooth functions , and if @xmath107 is a @xmath115-symmetric vector field on @xmath101 , then the lie derivative @xmath108 defines an essentially self - adjoint operator @xmath137 whose closure @xmath138 coincides with @xmath139 .
moreover , if the local flow @xmath140 of @xmath141 at time @xmath142 is defined , then @xmath143 we apply frhlich s theorem with @xmath144 and @xmath145 . for @xmath146 , @xmath147 , @xmath148 , we define @xmath149 then implies @xmath150 , so the assumptions of frhlich s theorem are satisfied , and @xmath151 is essentially self - adjoint .
let us denote by @xmath152 its closure .
if @xmath153 is an element of its domain @xmath154 , then leads for every @xmath155 to @xmath156 we conclude that @xmath157 and @xmath158 .
the other inclusion follows from which implies that @xmath159 . indeed , since @xmath151 is essentially self - adjoint we have @xmath160 .
theorem [ thm:2.4 ] above extends directly to the case where @xmath101 is a manifold modeled on a locally convex space , provided the vector field @xmath107 is assumed to have a local flow . indeed
7.1 ) is stated in this generality .
let @xmath0 be a symmetric banach lie group .
we also write @xmath1 for the corresponding automorphism of its lie algebra @xmath3 , which leads to the _ symmetric banach
lie algebra _
we write @xmath162 for the eigenspace decomposition of @xmath163 under @xmath1 .
let @xmath164 be an open convex cone invariant under @xmath165 and @xmath166 be a corresponding olshanski semigroup in the sense of definition [ def : olshsem ] .
in particular , @xmath36 is a connected lie group with lie algebra @xmath167 , but we do not assume that @xmath36 is contained in @xmath4 .
we write @xmath168 for the corresponding exponential map on @xmath8 . in polar
coordinates the involution on @xmath32 is given by @xmath169 ( remark [ rem : a.5 ] ) .
\(a ) let @xmath41 be a real hilbert space and @xmath170 .
we write @xmath171 for the adjoint of an element @xmath172 .
then @xmath173 defines an involution on @xmath4 with @xmath174 ( the orthogonal group of @xmath41 ) and @xmath175 is the space of symmetric operators . in this case
@xmath176 actually is an olshanski ( semi-)group for @xmath177 .
similarly statements hold for complex and quaternionic hilbert spaces .
\(b ) if @xmath36 is a connected lie group for which @xmath9 leaves a compatible norm on @xmath167 invariant , then @xmath167 is called an _ elliptic banach
lie algebra_. then @xmath36 has a universal complexification @xmath178 with a polar decomposition @xmath179 ( ( * ? ?
* ex . 6.9 ) ) .
a finite dimensional lie algebra @xmath167 is elliptic if and only if it is compact , but the class of elliptic lie algebras is quite large . in particular , it contains the algebra @xmath180 of skew - hermitian elements of a @xmath181-algebra @xmath182 and in particular the lie algebra @xmath183 of the full unitary group @xmath184 of a complex hilbert space @xmath41 .
\(c ) if @xmath107 is a banach space and @xmath185 an open convex cone , then @xmath186 is a complex olshanski semigroup with respect to the involution @xmath187 .
\(d ) if @xmath182 is a unital @xmath181-algebra , @xmath188 ( the identity component of its group of units ) and @xmath189 , then @xmath32 is a complex olshanski semigroup @xmath190 , where @xmath191 an important example is the semigroup of invertible strict contractions of a complex hilbert space @xmath41 .
\(e ) let @xmath182 be a unital @xmath181-algebra and @xmath192 . for @xmath193 , we write @xmath194 if there exists an invertible element @xmath195 with @xmath196 .
then @xmath197 is an open subsemigroup of @xmath182 with respect to multiplication . to see that it is non - empty ,
we observe that we may write @xmath198 for a projection @xmath199 . for @xmath200 and @xmath201
we then have @xmath202 if and only if @xmath203 .
the semigroup @xmath32 is of the form @xmath55 for @xmath204 \(f ) as already mentioned in the introduction , the compression semigroups of ( real forms of ) symmetric ( hilbert ) domains are also olshanski semigroups ( cf .
@xcite ) .
let @xmath205 be an _ involutive banach semigroup _
, i.e. , an involutive semigroup carrying a banach manifold structure such that multiplication and inversion are smooth maps .
\(a ) a homomorphism @xmath206 is called a _ @xmath12-representation _ if @xmath207 for every @xmath208 .
such a representation is said to be _ non - degenerate _ if @xmath209 spans a dense subspace of @xmath41 , which is equivalent to the condition that @xmath210 implies @xmath211 .
\(b ) for a representation @xmath212 of @xmath32 , a vector @xmath213 is called _ smooth _ if its orbit map @xmath214 is smooth .
we write @xmath38 for the subspace of _ smooth vectors _ and say that @xmath212 is _ smooth _ if @xmath38 is dense in @xmath41 .
\(c ) a @xmath12-representation @xmath212 of @xmath32 is called _ locally bounded _ if every @xmath215 has a neighborhood on which @xmath216 is bounded .
\(d ) for a unitary representation @xmath217 of a lie group @xmath4 , we also write @xmath218 for the subspace of smooth vectors . this subspace carries the _ derived representation _ @xmath219 of the lie algebra @xmath3 of @xmath4 .
if @xmath20 is smooth , then these operators are essentially skew - adjoint and , for @xmath220 , the closure @xmath221 is the infinitesimal generator of the unitary one - parameter group @xmath222 .
rem : locbo ( a ) since the arguments in ( * ? ? ?
5.1 , lemma 5.2 ) apply also to semigroup actions , the first countability of @xmath32 implies that any strongly continuous representation @xmath206 defines a continuous action @xmath223 .
the continuity in the points @xmath224 implies that @xmath43 is locally bounded .
\(b ) if , conversely , @xmath43 is locally bounded and the orbit maps @xmath225 are continuous for every @xmath226 in a dense subspace , then @xmath43 is strongly continuous ( ( * ? ? ? * lemma iv.1.3 ) ) . in the following we are interested in non - degenerate strongly continuous @xmath12-representations @xmath227 on a hilbert space @xmath41 .
first we observe that , although @xmath36 need not be contained in @xmath32 , any non - degenerate @xmath12-representation of @xmath32 defines in a natural fashion a unitary representation of @xmath36 .
p : reph for every non - degenerate @xmath12-representation @xmath212 of @xmath32 there exists a unique , not necessarily continuous , unitary representation @xmath228 satisfying @xmath229 its space of continuous vectors contains @xmath209 and its space of smooth vectors contains @xmath230 . in particular , @xmath231 is strongly continuous , resp . , smooth if @xmath212 is .
the existence of a unique homomorphism @xmath228 satisfying follows from the fact that @xmath36 acts on @xmath32 by unitary multipliers ( remark [ rem : a.5 ] , ( * ? ? ?
iii.1.5 ) ) .
the smoothness of the action of @xmath36 on @xmath32 now implies that @xmath232 has a continuous ( smooth ) orbit map under @xmath36 if @xmath226 has a continuous ( smooth ) orbit map under @xmath32 .
this completes the proof . to apply the hille
yosida theorem to the symmetric one - parameter semigroups @xmath233 for @xmath64 , we need to know that @xmath234 as the following lemma shows , this follows from the non - degeneracy of the representation and strong continuity .
lem:3.4 if @xmath43 is a non - degenerate @xmath12-representation of @xmath166 and @xmath64 , then @xmath235 is dense in @xmath41 .
furthermore , holds if @xmath236 is strongly continuous on @xmath237 .
let @xmath238 . in view of @xmath239
this is equivalent to @xmath240 . now
ii.4.15 ) implies that @xmath241 and hence that @xmath242 for every @xmath243 .
any @xmath208 can be written @xmath244 with @xmath245 and some @xmath246 because the left invariant vector field @xmath247 generates a local flow on @xmath32 ( remark [ rem : a.9 ] ) .
it follows that @xmath248 .
since @xmath43 is a non - degenerate representation of @xmath32 , it follows that @xmath211 , and hence that @xmath235 is dense in @xmath41 .
we see in particular , that the representation @xmath236 of @xmath249 is non - degenerate .
assume that @xmath236 is strongly continuous , hence locally bounded ( remark [ rem : locbo ] ) .
then lemma ( * ? ? ?
* lemma vi.2.2 ) implies that @xmath236 extends uniquely to a strongly continuous representation on @xmath250 , which implies .
let @xmath212 be a non - degenerate strongly continuous @xmath12-representation of @xmath32 .
in view of lemma [ lem:3.4 ] , we obtain for every @xmath64 a self - adjoint operator @xmath251 the generator of the symmetric one - parameter semigroup @xmath236 .
it is defined on the subspace @xmath252 where the derivative exists ( hille yosida theorem ) . for @xmath42 ,
we likewise write @xmath253 for the generator of the corresponding strongly continuous unitary one - parameter group @xmath254 ( proposition [ p : reph ] ; stone s theorem ) .
the following theorem is our main result .
thm:3.6 let @xmath15 be a simply connected lie group with lie algebra @xmath255 and @xmath227 be a non - degenerate strongly continuous smooth @xmath12-representation .
then there exists a unique smooth unitary representation @xmath40 of @xmath15 on @xmath41 whose space of smooth vectors is contained in @xmath252 for every @xmath256 , and whose derived representation satisfies @xmath257 we will see ( remark [ r : dom ] ) that the space @xmath258 of smooth vectors for the representation @xmath20 of @xmath15 coincides with @xmath259 accordingly , our strategy is to define a representation of the lie algebra @xmath260 on this space so that is satisfies and then verify that we can use the results in @xcite to show that this representation of @xmath260 integrates to a representation of @xmath15 .
\(a ) let @xmath261 be the simply connected covering group of @xmath36 .
then we have a unique morphism @xmath262 integrating the inclusion map @xmath263 .
the relation @xmath264 now implies that @xmath265 .
as @xmath266 acts trivially on @xmath3 , it also acts trivially on @xmath260 , i.e. , @xmath267 .
in particular , it is a normal subgroup and @xmath20 actually factors through a representation of the quotient @xmath268 .
if @xmath269 is discrete , this quotient is a lie group with the same lie algebra . here is an example showing that the group @xmath269 need not be discrete .
we consider the lie algebra @xmath270 with the involution @xmath271 , which leads to @xmath272 , and @xmath273 is a corresponding simply connected group .
for the lie algebra @xmath274 , the corresponding simply connected group is @xmath275 , and the kernel for the adjoint action of @xmath276 on @xmath3 is isomorphic to @xmath277 with @xmath278 .
now @xmath279 is a discrete subgroup of @xmath276 acting trivially on @xmath3 , so that @xmath280 satisfies our condition imposed for the construction of olshanski semigroups and @xmath281 .
now @xmath282 is not discrete in @xmath15 .
\(b ) let @xmath283 be the universal covering of the olshanski semigroup @xmath166 and @xmath284 be a smooth locally bounded @xmath12-representation of @xmath32 , so that theorem [ thm:3.6 ] leads to a unitary representation @xmath217 of @xmath15 .
clearly , the representation @xmath285 of @xmath286 leads to the same representation of @xmath15 .
now @xmath287 implies that @xmath288 , so that the representation @xmath289 of @xmath286 actually factors through a representation of the semigroup @xmath290 , where @xmath291 . from the point of view of the representation theory of the group @xmath15 , all representations obtained by theorem [ thm:3.6 ] can also be obtained from the semigroup @xmath292 , which is locally isomorphic to @xmath32 .
the representation @xmath217 of @xmath15 obtained from the lscher
mack theorem has the remarkable property that the spectrum of the operator @xmath293 is bounded from below for every @xmath294 in the cone @xmath295 . in section [ s : holext ] we will prove a converse to theorem [ thm:3.6 ] : a @xmath296-semibounded representation @xmath20 of @xmath15 comes from a strongly continuous smooth @xmath12-representation of @xmath55 , where @xmath297 is the identity component of the group of fixed points for the involution @xmath298 on @xmath15 acting on the lie algebra by @xmath299 for @xmath300 .
it is instructive to take a closer look at the special case @xmath301 with @xmath302 , i.e. , where @xmath303 is complex conjugation with respect to the real form @xmath167 .
then @xmath55 is a complex olshanski semigroup ( cf
. definition [ def : olshsem](b ) ) .
then the complexification of @xmath3 can be realized by the embedding @xmath304 which leads to @xmath305 , a direct sum of two complex lie algebras . in this picture
the lie algebra @xmath306 is given by @xmath307 we conclude that @xmath308 , where @xmath276 is the simply connected covering group of @xmath36 .
if @xmath309 is a complex linear representation of @xmath3 on the complex linear space @xmath65 , then the complex linear extension to @xmath305 is given by @xmath310 because @xmath311 therefore the corresponding unitary representation @xmath312 of @xmath15 is given by @xmath313 .
sec:4 the following lemma permits us to reduce the proof of theorem [ thm:3.6 ] to the case where @xmath212 is cyclic and generated by a smooth vector .
lem:3.7 a non - degenerate strongly continuous smooth @xmath12-representation is a direct sum of cyclic representations with smooth cyclic vectors .
the set of all families @xmath314 of mutually orthogonal closed @xmath32-invariant subspaces which contain a smooth cyclic vector is well ordered by inclusion and by zorn s lemma it has a maximal element @xmath314 .
let @xmath315 .
then @xmath316 and @xmath317 are @xmath32-invariant , and we claim that if @xmath317 is non - zero , it contains a smooth vector . indeed ,
if @xmath318 , there exists a smooth vector @xmath319 which is not in @xmath316 .
let us write @xmath320 for the orthogonal projection on @xmath317 and @xmath321 .
then the relation @xmath322 , @xmath208 , shows that @xmath226 is a smooth vector in @xmath317 .
since @xmath43 is non - degenerate , we have @xmath323 ( cf .
* lemma ii.2.4 ) ) , hence @xmath324 is a closed @xmath32-invariant subspace with a smooth cyclic vector orthogonal to each @xmath325 , @xmath326 . but this contradicts the maximality of @xmath314 .
therefore @xmath327 , and as representations of @xmath32 , we have @xmath328 .
from now on we assume that @xmath329 is a cyclic vector .
then we obtain a smooth positive definite kernel @xmath330 on @xmath32 which leads to a unitary map @xmath331 onto the reproducing kernel space @xmath332 .
indeed , @xmath333 intertwines @xmath43 with the representation of @xmath32 on @xmath334 given by @xmath335 .
therefore it suffices to prove theorem [ thm:3.6 ] only for the representation @xmath336 of @xmath166 .
the function @xmath337 satisfies @xmath338 , so that @xmath339 for @xmath340 . to each @xmath341
we associate the corresponding left invariant vector field @xmath247 on @xmath32 ( cf .
definition [ d : tangiso ] ) and write the corresponding integral curves as @xmath342 ( cf .
remark [ rem : a.9 ] ) . as in
, we set @xmath343 and extend this definition @xmath344-linearly to any @xmath345 .
then we have on @xmath346 the relation @xmath347=\cl_{[x , y]}.\ ] ] we recall that a _ core _ of a self - adjoint operator @xmath348 is a dense subspace @xmath349 for which @xmath350 .
prop : lieder let @xmath351 .
then @xmath352 , @xmath353 is a core for @xmath253 , and @xmath354 for @xmath355 , the first assertion follows from and proposition [ p : reph ] . for @xmath356 and @xmath208 , the map @xmath357 extends to a smooth map on some @xmath358-neighborhood , and for @xmath359 it coincides with @xmath360 .
therefore @xmath361 is contained in the domain @xmath362 of @xmath363 . in the following we also write @xmath364 for @xmath365 . for @xmath256 , the space @xmath353 is invariant under @xmath366 , @xmath367
, hence is a core for @xmath253 ( ( * ? ? ?
1.7 ) ) , i.e. , @xmath368 .
finally , let @xmath351 and @xmath369 .
then @xmath370 shows that @xmath371 extends @xmath253 .
further , for @xmath372 , @xmath373 shows that @xmath374 extends @xmath371 .
this proves .
the preceding proposition shows that , for any @xmath256 , @xmath371 leaves the subspace @xmath375 invariant , and by linearity this is also true for @xmath376 .
therefore we set @xmath377 and observe that @xmath378 l : lieder if @xmath379 , @xmath380 , then @xmath381 and for @xmath215 , @xmath382 the map @xmath383 defined for small enough @xmath384 , is a smooth @xmath96-valued map ( cf .
remark [ rem : a.9 ] ) .
hence @xmath385 and @xmath386 p : repgc the domain @xmath129 contains @xmath353 and is dense in @xmath96 .
the map @xmath387 is a strongly continuous representation of @xmath388 by skew - symmetric operators in the sense that all the maps @xmath389 are continuous .
more generally , for each @xmath390 the map @xmath391 is continuous .
the first assertion follows from and lemma [ l : lieder ] .
the map @xmath392 is a lie algebra homomorphism because @xmath393 is so by . for the strong continuity it suffices to show that , for @xmath394 , the graph of the map @xmath395 is closed ( cf .
* lemma 4.2 ) ) .
this follows if , for each @xmath215 the map @xmath396 is continuous . that this is the case on the real lie algebra @xmath3 follows from the fact that @xmath153 is a smooth function on @xmath32 because @xmath397 is a topological isomorphism ( cf
. definition [ d : tangiso ] ) .
clearly , the continuity is inherited by the complex linear extension to @xmath398 .
this completes the proof . to show that the representation of @xmath399 integrates to a continuous unitary representation of @xmath15 we will use the following theorem t :
intcrit let @xmath15 be a simply connected banach
lie group with lie algebra @xmath260 .
assume that @xmath400 where @xmath401 and @xmath402 are closed subspaces .
let @xmath392 be a strongly continuous representation of @xmath260 on a dense subspace @xmath129 of a hilbert space @xmath41 such that for every @xmath403 , @xmath404 is essentially skew - adjoint , @xmath405 and @xmath406 then @xmath392 integrates to a continuous unitary representation @xmath40 of @xmath15 for which @xmath407 is a @xmath15-invariant subspace and @xmath408 for @xmath409 .
in particular , for @xmath409 , the infinitesimal generator @xmath221 of the one - parameter group @xmath222 coincides with the closure @xmath410 .
the next two propositions ensure that the assumptions of the theorem are satisfied by the representation @xmath392 for @xmath411 and @xmath412 .
the first one uses the geometric frhlich theorem [ t : froelichgeo ] .
p : sa for @xmath413 , the operator @xmath404 is essentially skew - adjoint with closure @xmath414 . for @xmath415 , we know that @xmath353 is a core for the skew - adjoint operator @xmath416 ( proposition [ prop : lieder ] ) . since @xmath417 ,
the larger subspace @xmath129 is also a core for @xmath414 .
this proves the assertion for @xmath418 .
now let @xmath419 be a general element .
then the vector field @xmath247 is @xmath115-symmetric , since for every @xmath420 , we have @xmath421 we can therefore apply theorem [ t : froelichgeo ] and proposition [ prop : lieder ] to see that @xmath414 is a self - adjoint operator , and that @xmath422 is essentially self - adjoint .
writing @xmath423 with @xmath424 , we see that we also have @xmath425 , so that @xmath426 and thus @xmath427 is essentially skew - adjoint with closure @xmath428 .
l : comrel let @xmath256 , @xmath429 , and @xmath430 such that @xmath431 .
then @xmath432 and @xmath433 for every @xmath208 , @xmath434 leads to @xmath435 l : ext let @xmath436 be a symmetric operator and @xmath437 a dense subspace . if @xmath438 is essentially self - adjoint , then @xmath18 is also essentially self - adjoint , with @xmath439 . from @xmath440
it follows that @xmath441 ( cf .
viii.1 ) ) .
therefore @xmath442 is self - adjoint .
prop:4.9 for @xmath413 , @xmath405 .
moreover for any @xmath443 we have @xmath444 in the following we also write @xmath364 for @xmath365 . + * step 1 : * let @xmath256 and @xmath445
. then lemma [ l : comrel ] implies by induction that for every @xmath380 and @xmath446 , @xmath447 and hence @xmath448 .
it follows that @xmath449 and that we have for @xmath445 and ( by linearity ) for @xmath443 , @xmath450 in particular holds for @xmath451 .
now let @xmath452 .
then can be written @xmath453 and this last equation is all we need for the following . * step 2 : * assume that @xmath356 .
then the spectrum of @xmath253 is bounded from above and hence @xmath454 extends to a strongly continuous holomorphic semigroup @xmath455 which is holomorphic on @xmath456 ( cf .
9.9 ) or ( * ? ? ?
vi.3.2 ) ) . for @xmath457 , the map @xmath458 , @xmath459 is @xmath344-linear and continuous ( proposition [ p : repgc ] ) , hence the function @xmath460
, @xmath461 is analytic .
since the map @xmath462 is @xmath344-bilinear and continuous , it follows that the map @xmath463 is analytic . by the analytic continuation principle
, the equality implies @xmath464 we then have by continuity @xmath465 this shows that , for @xmath466 , @xmath467 .
we thus arrive at @xmath468 by induction , we now obtain @xmath469 * step 3 : * for @xmath380 , @xmath470 , @xmath443 and @xmath471 we now obtain @xmath472 from @xmath473 now let @xmath419 and write it as @xmath474 with @xmath470 . since @xmath475 is essentially self - adjoint on @xmath129 ( proposition [ p : sa ] ) , it is essentially self - adjoint as an operator on its domain @xmath476 , and its closure is @xmath477 ( lemma [ l : ext ] ) .
we therefore have trotter s product formula ( * ? ? ?
viii.31 ) : @xmath478 replacing @xmath479 by @xmath480 , @xmath481 , in and taking the limit we obtain @xmath482 as above we now obtain by induction that @xmath483 and that holds for @xmath484 . in view of proposition [ prop:4.9 ] , we obtain from theorem [ t : intcrit ] a unitary representation of @xmath15 whose space of smooth vectors contains @xmath129 and such that @xmath485 and @xmath486 for @xmath487 .
we conclude with proposition [ prop : lieder ] and [ p : sa ] that @xmath488 for @xmath451 and @xmath489 for @xmath356 .
[ r : dom ] since @xmath490 , it immediatly follows from @xmath491 ( ( * ? ? ?
* lemma 3.4 , remark 8.3 ) ) , that @xmath492 .
in this section we obtain a result which is a converse to the lscher mack theorem . it is new even in the finite dimensional setting . at the same time
, we obtain the existence of holomorphic extensions for semibounded unitary representations .
d:5.3 let @xmath4 be a banach lie group with lie algebra @xmath163 .
for a smooth unitary representation @xmath217 of @xmath4 we consider the map @xmath493
\(a ) a smooth unitary representation @xmath217 of @xmath4 is called _ semibounded _ if @xmath57 is bounded on a non - empty open subset of @xmath163 .
then the cone @xmath494 , consisting of all those @xmath220 for which @xmath57 is bounded on a neighborhood of @xmath294 , is an open @xmath495-invariant convex cone in @xmath3 .
moreover , @xmath496 is a continuous convex function ( @xcite ) . for a convex cone @xmath59
, we say that @xmath20 is _ @xmath8-semibounded _ if @xmath497 and @xmath498 is locally bounded .
\(b ) a convex cone @xmath59 is said to be _ relatively open _
if the linear subspace @xmath499 of @xmath3 is closed and @xmath8 is open in @xmath499 .
for the definition of a @xmath500-map we use in the next lemma see definition [ d : smooth ] .
l : diffexpcone let @xmath8 be a relatively open convex cone in @xmath163 , @xmath501 , and @xmath217 be a smooth @xmath8-semibounded unitary representation of @xmath4 . then
, for every @xmath502 , the map @xmath503 is @xmath500 and @xmath504 the map @xmath505 is also @xmath500 . for @xmath356 , @xmath443 and @xmath506 ,
the function @xmath507 is continuous on the closed upper half - plane @xmath508 and holomorphic on its interior ( ( * ? ? ?
vi.3.2 ) ; see also step @xmath509 in the proof of proposition [ prop:4.9 ] ) .
since @xmath510 for @xmath511 by proposition [ prop:4.9 ] , the schwarz reflection principle ( * ? ? ?
. 11.14 ) implies that @xmath512 vanishes on @xmath25 .
it follows that , for @xmath356 and @xmath513 we have @xmath514 for every @xmath515 with @xmath516 since @xmath517 is again a smooth vector , we can iterate this argument to obtain inductively
@xmath518 we know by ( * ? ? ?
* lemma 3.4 , remark 8.3 ) that @xmath519 .
hence we have @xmath520 for @xmath64 and @xmath521 .
now let @xmath513 .
in view of , proposition [ p : derpath ] implies that @xmath522 is @xmath500 with @xmath523 and that the map @xmath524 is continuous . finally , we observe that the map @xmath525 is continuous because the @xmath4-action on @xmath41 defined by @xmath20 is continuous .
we have just seen that @xmath512 is partially differentiable in @xmath294 with continuous partial derivative @xmath526 we have also seen that @xmath527 , so that the partial derivatives in @xmath528 also exist and are given by @xmath529 as @xmath4 acts continously on @xmath41 , @xmath530 is continuous , and the adjoint action of @xmath4 on @xmath3 is continuous , this function is continuous .
this implies that @xmath512 is @xmath500 ( cf .
@xcite ) .
let @xmath8 be a relatively open convex cone in @xmath163 , @xmath531 , and @xmath167 be a closed subalgebra of @xmath163 .
the cone @xmath8 is called _
@xmath167-compatible _ if @xmath532\subseteq \fh \quad \mbox { and } \quad e^{\ad\fh}w\subseteq w.\ ] ] then @xmath533 is a closed subalgebra which is turned in a symmetric banach
lie algebra by the involution @xmath534 for @xmath535 .
an @xmath167-compatible cone @xmath59 is said to be _
integrable _ if there exists a symmetric banach lie group @xmath536 with symmetric lie algebra @xmath537 such that , for @xmath538 , the polar map @xmath539 is an analytic diffeomorphism onto an open subsemigroup @xmath540 then @xmath32 is invariant under the involution @xmath541 , turning it into an involutive semigroup @xmath205 . from the discussion of banach olshanski semigroups in appendix
[ a ] , it then follows that for each connected banach
lie group @xmath542 locally isomorphic to @xmath36 to which the adjoint action of @xmath167 on @xmath260 integrates , there exists an involutive banach
olshanski semigroup @xmath543 with a polar decomposition which is a quotient of the universal covering semigroup of @xmath544 .
t : analcont let @xmath4 be a banach lie group with lie algebra @xmath163 , @xmath167 be a closed complemented lie subalgebra of @xmath163 , and @xmath545 be the corresponding integral subgroup in @xmath4 .
let @xmath40 be a smooth @xmath8-semibounded unitary representation of @xmath4 for the integrable @xmath167-compatible cone @xmath8 .
then the formula @xmath546 defines a strongly continuous smooth @xmath12-representation @xmath43 of @xmath544 on @xmath41 .
we shall need the following chain rule ( ( * ? ? ?
* lemma 4 ) ) : [ l : pr ] let @xmath547 be an open interval , @xmath548 and @xmath512 be two banach spaces and @xmath549 denotes the space of continuous linear operators from @xmath548 to @xmath512 endowed with the strong operator topology .
let @xmath550 be a continuous path such that @xmath551 is differentiable for every @xmath226 in a subspace @xmath129 of @xmath548 and let @xmath552 be a differentiable path in @xmath129 .
we write @xmath553 for the linear operator obtained by @xmath554 for @xmath555 .
then @xmath556 is differentiable with @xmath557 * step 1 : * let us first prove that , for @xmath470 , @xmath558 for this purpose , let us write for @xmath246 , @xmath559 now let @xmath560 and consider @xmath561
. by lemma [ l : diffexpcone ] and the chain rule ( lemma [ l : pr ] ) , applied with @xmath562 and @xmath563 , the path @xmath564 is differentiable for @xmath243 , and , denoting by @xmath565 the right logarithmic derivative ( see definition [ def : logder ] ) , we obtain with proposition [ p : exp ] @xmath566 since @xmath567 ( lemma [ l : diffexpcone ] , lemma [ lem : approx ] ) , we obtain with @xcite that @xmath568 and follows for @xmath569 .
* step 2 : * for @xmath570 and @xmath356 we have @xmath571 so that @xmath572 from we obtain the relation @xmath573 and we further derive @xmath574 * step 3 : * now we can prove that for @xmath575 and @xmath470 , @xmath576 with and we obtain @xmath577 * step 4 : * it remains to prove that for @xmath502 the map @xmath578 , @xmath579 is smooth . in view of lemma
[ l : diffexpcone ] , this map is @xmath500
. for @xmath451 we have by remark [ rem : a.9 ] and @xmath580 similarly we have for @xmath356 @xmath581 since @xmath582 linear , it follows that @xmath583 now an easy induction shows that the higher partial derivatives of @xmath584 only involve the continuous @xmath585-linear maps @xmath586 and hence that @xmath522 is smooth .
now recall the context of the lscher mack theorem .
we have a symmetric banach
lie algebra @xmath587 and an integrable @xmath588-invariant open convex cone @xmath29 .
we therefore have a banach
olshanski semigroup @xmath589 for each connected lie group with lie algebra @xmath167 .
applying the preceding theorem to @xmath590 and @xmath591 , we obtain the following converse to the lscher mack theorem : [ c : conv ] let @xmath587 be a symmetric banach lie algebra and @xmath8 be an integrable @xmath588-invariant open convex cone in @xmath10 .
let @xmath15 be a banach lie group with lie algebra @xmath592 and let @xmath593 be its integral subgroup with lie algebra @xmath167 .
let @xmath20 be an @xmath296-semibounded unitary representation of @xmath15 .
then @xmath594 defines a strongly continuous smooth @xmath12-representation @xmath43 of @xmath595 .
let @xmath4 be a banach - lie group with lie algebra @xmath163 , @xmath40 be a semibounded unitary representation of @xmath4 , and @xmath596 be an open integrable @xmath495-invariant convex cone .
then @xmath597 defines a holomorphic @xmath12-representation @xmath43 of the complex involutive semigroup @xmath598 in particular the vectors in @xmath599 are analytic for @xmath20 .
note that @xmath62 is a complex olshanski semigroup ( cf
. definition [ def : olshsem ] ) .
first we observe that the cone @xmath8 is @xmath163-compatible .
theorem [ t : analcont ] now applies to the semigroup @xmath62 .
it remains to prove that @xmath600 is holomorphic .
but shows that @xmath584 is complex linear , hence that @xmath522 is holomorphic .
the holomorphy of @xmath43 now follows from * ( * ? ? ?
* lemma iv.2.2)*. now let @xmath601 and @xmath602 .
since @xmath603 , the analyticity of @xmath604 follows from the analyticity of the map @xmath605 , @xmath606 .
the preceding theorem generalizes olshanski s holomorphic extension theorem for highest weight representations ( @xcite , @xcite ) to the banach lie setting . in the finite dimensional case
the proof heavily relies on the existence of a dense space of analytic vectors , which can be derived by convolution with heat kernels ( @xcite ) , but for unitary representations of banach lie groups , not even the space of @xmath500-vectors need to be dense ( cf .
@xcite ) . in the finite dimensional context one proves first that @xmath43 is holomorphic , and then the multiplicativity of @xmath43 is obtained by analytic continuation . in the proof we give here the multiplicativity of @xmath43 follows from the ( assumed ) existence of smooth vectors and then the holomorphy of @xmath43 follows as a bonus from its multiplicativity .
let @xmath0 be a symmetric banach lie group and @xmath161 the corresponding symmetric lie algebra .
we write @xmath607 for the eigenspace decomposition of @xmath163 under @xmath1 and let @xmath608 denote the identity component of @xmath609 .
we set @xmath610 and consider an open convex @xmath611-invariant convex cone @xmath29 for which the map @xmath612 is an analytic diffeomorphism onto an open subsemigroup @xmath7 of @xmath4 . in these
coordinates the involution on @xmath613 is given by @xmath614 in particular @xmath32 is @xmath12-invariant . in the following we write @xmath615 for this involutive semigroup .
[ def:1.9 ] let @xmath32 be an involutive semigroup .
a _ multiplier _ of @xmath32 is a pair @xmath616 of maps @xmath617 satisfying @xmath618 we write @xmath619 for the set of all multipliers of @xmath32 and turn it into an involutive semigroup by @xmath620 where @xmath621 and @xmath622 .
we write @xmath623 for the _ unitary group _ of @xmath619 .
note that @xmath624 is a morphism of involutive semigroups and that @xmath619 acts on @xmath32 from the left by @xmath625 and from the right by @xmath626 .
the group @xmath627 is in general not contained in @xmath628 , but it acts on it by the unitary multipliers @xmath629 , @xmath630 .
let @xmath631 be the universal covering of the banach manifold @xmath632 .
then @xmath633 carries the structure of an analytic banach @xmath12-semigroup such that the covering map @xmath634 is a homomorphism of banach @xmath12-semigroups .
moreover , the simply connected covering group @xmath635 of @xmath627 acts on @xmath286 by unitary multipliers and we thus obtain an analytic diffeomorphism @xmath636 where @xmath637 is a continuous lift of @xmath638 such that @xmath639 and @xmath640 since the polar map @xmath641 is an analytic diffeomorphism , there exists an analytic diffeomorphism @xmath642 with @xmath643 .
we then define @xmath644 .
pick @xmath645 and let @xmath646 be the unique continuous lift of the multiplication map @xmath647 with @xmath648 then the uniqueness of lifts implies that the restriction of @xmath649 to @xmath650 satisfies @xmath651 and we obtain in particular @xmath652 therefore the uniqueness of lifts implies that @xmath649 is associative , hence defines on @xmath286 an analytic semigroup structure .
we also lift the involution on @xmath32 to the unique involutive diffeomorphism @xmath12 on @xmath286 with @xmath653 , and since @xmath654 now holds for @xmath655 , the uniqueness of lifts implies that @xmath656 is an involutive semigroup . the multiplication on @xmath32 can be expressed by analytic maps @xmath657 and @xmath658 as @xmath659 from the continuity of the multiplication @xmath660 in @xmath4 , it follows that both maps @xmath661 and @xmath662 extends continuously to the set @xmath663 if @xmath664 is the unique lift of @xmath665 satisfying @xmath666 , then we obtain the formula @xmath667 for each @xmath64 and @xmath668 we further have @xmath669 because @xmath670 and the subset @xmath671 is connected .
this implies .
the left and right multiplier actions of @xmath627 on @xmath32 lift to unique left and right actions of @xmath635 on @xmath286 , satisfying @xmath672 from we further derive @xmath673 this implies in particular that the left action of @xmath635 on @xmath286 commutes with the right multiplications .
we also obtain from the uniqueness of lifts that @xmath674 so that left multiplications in @xmath286 commute with the right action of @xmath635 . to see that @xmath635 acts on @xmath286 by unitary multipliers , it remains to observe that @xmath675 for @xmath676 , which also follows from .
let @xmath677 be the action of @xmath635 on @xmath10 , obtained from the action @xmath678 of @xmath627 on @xmath10 and the covering map @xmath679 . for a discrete central subgroup @xmath680 acting trivially on @xmath10 , the cosets in @xmath286 satisfy @xmath681 so that the quotient semigroup @xmath682 inherits the structure of an analytic involutive banach semigroup for which the quotient map @xmath683 is a morphism of involutive semigroups .
moreover , the polar map @xmath642 factors through a diffeomorphism @xmath684 and the group @xmath685 acts faithfully on @xmath682 by unitary multipliers . since the left action of @xmath686 on @xmath286 coincides with the right action ( see ) , the relations easily follow .
the remaining assertions are now obvious .
def : olshsem ( a ) the semigroups obtained by the preceding proposition will be called banach olshanski semigroups .
we write @xmath687 and @xmath688 for the exponential function @xmath689 .
\(b ) for the special case where @xmath690 and @xmath691 , the group @xmath4 is complex , so that @xmath692 is a complex manifold on which the multiplication is holomorphic and the involution is antiholomorphic .
these properties are inherited by all other olshanski semigroups @xmath55 , where @xmath693 is a connected lie group with lie algebra @xmath167 .
therefore we call them _ complex olshanski semigroups_. rem : a.5 the basic properties of olshanski semigroups @xmath55 are : the polar map @xmath694 is an analytic diffeomorphism .
@xmath36 acts on @xmath55 smoothly by unitary multipliers . for @xmath64
, we have @xmath695 for @xmath365 and @xmath696 , we have @xmath697 . from now on @xmath36 always denotes a connected lie group with lie algebra @xmath167 and @xmath166 is a corresponding olshanski semigroup .
we do not assume that @xmath36 is contained in @xmath4 .
lem : approx for @xmath698 and @xmath64 we have @xmath699 it suffices to verify this relation in the simply connected covering semigroup @xmath286 , where we have for @xmath700 : @xmath701 since the functions @xmath702 and @xmath661 extend continuously to the domain @xmath703 , formula yields @xmath704 .
the other relation is obtained by applying the involution @xmath12 .
let @xmath705 denote the _
right maurer
cartan form _ , defined by @xmath706 for @xmath707 , where @xmath708 denotes the canonical right action of @xmath4 on @xmath709 .
similarly we define the _ left maurer
cartan form _ by @xmath710 . for the subsemigroup @xmath711
, the restriction @xmath712 defines a trivialization of the tangent bundle of @xmath32 by @xmath713 if @xmath714 is the universal covering , the form @xmath715 likewise trivializes @xmath716 . for every discrete central subgroup @xmath680 acting trivially on @xmath10 and @xmath632 , the form @xmath717 on @xmath286 is @xmath686-invariant ,
hence is the pullback of a form @xmath718 trivializing @xmath719 .
[ d : tangiso ] the preceding discussion shows that on every olshanski semigroup @xmath55 , we have a natural form @xmath720 trivializing the tangent bundle and we similarly obtain a left invariant form @xmath721 .
accordingly , we have natural _ left invariant vector fields _
@xmath247 , @xmath220 , on @xmath32 , defined by @xmath722 . and
_ right invariant vector fields _
@xmath723 , @xmath220 , defined by @xmath724 .
rem : a.9 for @xmath725 and @xmath215 we have @xmath726 both relations are obvious for the semigroup @xmath628 , and they are inherited by the simply connected covering and hence also by its quotients . for @xmath220 and @xmath215 , we write @xmath727 for the integral curve of @xmath247 through @xmath728 and likewise @xmath729 for the local integral curve of @xmath723 .
this is redundant for @xmath711 , and for a general @xmath166 , the preceding observation shows that it is also consistent for @xmath730 with the action of the corresponding one - parameter ( semi)groups .
rem : rightinv ( a ) from the right invariance of @xmath731 on @xmath4 , we obtain @xmath732 by verifying that this property is preserved by the passage to covering semigroups and to quotients by discrete central subgroups .
we likewise get @xmath733 \(b ) as in ( a ) , it follows that , for @xmath365 , the right multiplication @xmath734 also leaves @xmath735 invariant . for the left multiplication @xmath736 , the relation @xmath737 for the maurer
cartan form of a lie group @xmath4 implies that @xmath738 this is also verified by the passage through the universal covering semigroup .
def : logder for a smooth map @xmath739 , where @xmath85 is a smooth manifold , we define the ( right ) logarithmic derivative as the @xmath3-valued @xmath740-form @xmath741 if @xmath547 is an interval and @xmath742 a differentiable path , then the identification of @xmath740-forms on @xmath743 with @xmath3-valued functions leads to @xmath744 we likewise define logarithmic derivatives for maps with values in lie groups . for the exponential map @xmath745 we then have @xmath746 ( ( * ? ? ?
ii.5.7 ) ) and therefore : [ p : exp ] any @xmath500-path @xmath747 satisfies @xmath748 from the differential of the multiplications , we obtain left and right actions @xmath749 and likewise @xmath750 as well as @xmath751 we then also have @xmath752 because this is true for the subsemigroup @xmath692 of @xmath4 .
[ p : pr ] let @xmath753 and @xmath754 be two @xmath500-paths .
for the path @xmath755 in @xmath32 we then have @xmath756 first we note that @xmath757 to evaluate @xmath758 , we first write @xmath759 in @xmath760 . for @xmath761 ,
the relation @xmath762 for @xmath763 leads to @xmath764 so that implies that @xmath765 .
finally , the relation leads to @xmath766
in this section we let @xmath767)$ ] be a locally convex lie algebra .
this means that @xmath163 is a locally convex space and the lie bracket @xmath768 $ ] is continuous .
let us first recall the following basic notions of the differential calculus over locally convex spaces .
\(a ) let @xmath769 be two locally convex spaces , @xmath770 open and @xmath771 be a continuous map on the open set @xmath770 of @xmath548
. then @xmath772 is called @xmath500 if the directional derivatives @xmath773 exist for every @xmath774 and @xmath775 and the map @xmath776 is continuous .
+ ( b ) a continuous map @xmath777 is called @xmath778 , @xmath779 if it is @xmath500 and @xmath780 is @xmath781 .
it is called @xmath782 , or _ smooth _ , if it is @xmath778 for every @xmath783 .
+ ( c ) a locally convex space @xmath548 is called _ mackey complete _ if for each smooth curve @xmath784\ra e$ ] the weak integral @xmath785 exists , i.e. , there exists a ( unique ) element @xmath786 satisfying @xmath787 this implies in particular that the curve @xmath788 is smooth and satisfies @xmath789 .
[ d : smooth ] a locally convex lie algebra @xmath163 is called @xmath790-_integrable _ if for every @xmath341 the ( linear ) vector field defined by @xmath63 is complete , that is , if there exists a smooth map @xmath791 with @xmath792 .
we will then use the notation @xmath793 .
we consider a linear homomorphism @xmath794 in the space of endomorphism of a dense domain @xmath65 of a banach space @xmath548 .
we assume that @xmath392 is strongly continuous in the sense that for every @xmath795 the map @xmath796 , @xmath797 is continuous . now let @xmath8 be a convex cone in @xmath3 which is relatively open in its span @xmath501 ( cf . definition [ d:5.3 ] ) , and assume that , for every @xmath356 , the closure @xmath798 of @xmath404 generates a strongly continuous semigroup @xmath799 and that the map @xmath800 is locally bounded on @xmath8 .
if @xmath801 is a hilbert and for every @xmath356 the semigroup @xmath802 consists of normal operators , then the fact it generates a commutative @xmath181-algebra isomorphic to some @xmath803 implies that @xmath804 and hence that @xmath805 .
[ l : diff ] consider two operators @xmath18 and @xmath806 defined on a common dense domain @xmath129 of the banach space @xmath548 and whose closures generates strongly continuous semigroups @xmath807 and @xmath808 respectively .
assume further that @xmath809 for all @xmath810 .
if , for some @xmath555 the map @xmath811 is continuous on @xmath812 , then @xmath813 [ p : derpath]let @xmath163 be an @xmath790-integrable mackey complete locally convex lie algebra .
if , for every @xmath356 and every @xmath429 , @xmath814 then the map @xmath815 is continuous , and for every @xmath795 , @xmath522 is @xmath500 with @xmath816 let @xmath555 and @xmath817 . from the relation @xmath818 the continuity of the map @xmath819 and the strong continuity of @xmath820 , we derive that @xmath821 is continuous on @xmath822 .
we can therefore apply lemma [ l : diff ] to obtain @xmath823 let @xmath824 let @xmath825 be a @xmath358-neighbourhood in @xmath163 such that for every @xmath826 , @xmath827 .
consider now the continuous map @xmath828\times\fg\ra e,\ ( u , z)\mt\ha^v\big(e^{(u-1)\ad x}z\big).\ ] ] since @xmath829\times \{0\ } ) = \{0\}$ ] , the compactness of @xmath830 $ ] implies the existence of a @xmath358-neighborhood @xmath831 such that @xmath829 \times \v')$ ] is contained in the open ball @xmath832 of radius @xmath833 around @xmath358 in @xmath548 .
thus , for every @xmath834 , @xmath835 , and this proves that @xmath522 is continuous .
the local boundedness of @xmath836 further implies that @xmath522 is continuous for every @xmath775 , and hence the map @xmath837 is continuous .
let @xmath838 and let @xmath839 such that @xmath840 for @xmath841 .
we derive from the formula @xmath842 let us fix @xmath843 .
then the continuity of the map @xmath844 on @xmath845\times[-\tau,\tau]$ ] implies that we can pass to the limit under the integral sign to derive @xmath846 the same type of argument as the one used for the continuity of @xmath522 shows that the integrand of @xmath847 is bounded unformly with respect to @xmath848 , and hence the integral is uniformly small ( with respect to @xmath849 ) when @xmath850 is sufficently close to @xmath358
. therefore @xmath851 where the last equality follows from the uniqueness of the weak integral
. similarly we obtain @xmath852 and now the continuity of @xmath530 implies that @xmath853 is continuous , i.e , that @xmath522 is a @xmath500-map .
jorgensen , p. e. t. , and g. lafsson , _ unitary representations and osterwalder - schrader duality _ , in `` the mathematical legacy of harish - chandra , ''
r. s. doran and v. s. varadarajan , eds .
, proc . symp . in pure math .
* 68 * , amer .
soc . , 2000 | let @xmath0 be a banach lie group with involutive automorphism @xmath1 , @xmath2 be the @xmath1-eigenspaces in the lie algebra @xmath3 of @xmath4 , and @xmath5 be the identity component of its group of fixed points .
an olshanski semigroup is a semigroup @xmath6 of the form @xmath7 , where @xmath8 is an open @xmath9-invariant convex cone in @xmath10 and the polar map @xmath11 is a diffeomorphism .
any such semigroup carries an involution @xmath12 satisfying @xmath13 . our central result , generalizing the lscher mack theorem for finite dimensional groups , asserts that any locally bounded @xmath12-representation @xmath14 with a dense set of smooth vectors defines by `` analytic continuation '' a unitary representation of the simply connected lie group @xmath15 with lie algebra @xmath16 .
we also characterize those unitary representations of @xmath15 obtained by this construction . with similar methods , we further show that semibounded unitary representations extend to holomorphic representations of complex olshanski semigroups . |
the galaxy luminosity function ( hereafter lf ) provides a @xmath14 order description of the gross properties of galaxy populations .
although it may be regarded as somewhat simplistic , in reducing galaxy properties to the two numbers ( @xmath15 and @xmath16 ) that describe the lf in the @xcite form , reproducing the observed lf is a fundamental test for any viable theory of galaxy formation , and one that has proven surprisingly complex to achieve until relatively recently ( e.g. , @xcite ) .
clusters of galaxies are , in several ways , ideal laboratories to study the lf and its evolution .
cluster galaxies may be considered as a volume limited sample of objects observed at the same cosmic epoch and occupy a relatively ` constant ' environment corresponding to the highest density peaks at the redshift of observation .
operationally , cluster members have a much higher surface density on the sky than the surrounding field and often exhibit distinctive morphologies and colors ( see fig . 1 below ) .
it is therefore possible to establish membership in a cluster ( at least in a statistical sense ) without resorting to observationally expensive redshift surveys . on the other hand ,
clusters are obviously special places , containing only a small percentage ( @xmath17@xmath18 ) of all galaxies in the universe .
their galaxy populations clearly implicate the cluster environment in a variety of processes that affect the morphological evolution of galaxies ( e.g. , @xcite ) and suppress or modify their star formation history @xcite . in order to use clusters as probes of galaxy evolution
it is therefore important to understand the effects of environment and carry out careful comparisons with field samples . for local samples , this is now possible thanks to large redshift surveys such as 2df @xcite and sdss @xcite .
although bright galaxies appear to have formed the majority of their stellar populations and assembled their mass at least by @xmath19 ( e.g. , @xcite and references therein ) , there is considerable evidence that fainter galaxies undergo significant evolution since @xmath20 .
the red sequence in clusters in the ediscs sample appears to be truncated at faint magnitudes @xcite , while @xcite show that the luminosity function of red cluster galaxies flattens at higher redshifts , as do @xcite for clusters in a deep _ spitzer _ field . on the other hand , there are clusters with well - formed red sequences at redshifts approaching @xmath21 as well @xcite .
it may not be surprising that the faint end of the lf shows considerable variations from cluster to cluster , as one expects that dwarfs are more sensitive to environmental effects .
although the bright end of the lf ( @xmath22 ) is broadly universal , the faint end varies between clusters and may even change radially within clusters @xcite .
the evolution of the faint end slope of the total lf provides an interesting test of galaxy formation models .
even if the red sequence is truncated at faint luminosities at earlier epochs , one expects a steepening faint end as established by the initial power spectrum of fluctuations , which is very steep @xcite . in higher redshift clusters
, there should be a steeper lf and a larger fraction of bluer dwarfs .
the purpose of this paper is to explore the evolution of the lf to very faint magnitudes in a sample of @xmath11 clusters observed with the hubble space telescope ( hst ) , reaching well below the @xmath10 point and into the domain of dwarf spheroidal galaxies , using a multicolor imaging sample .
this will allow to test scenarios for the evolution of dwarf galaxy populations in @xmath23cdm hierarchical models .
the structure of this paper is as follows : we describe the data and photometry in the next section .
we derive the composite lfs and then discuss the results in the context of galaxy formation models .
we adopt the ` consensus ' cosmology with @xmath24 , @xmath25 and use h@xmath26=100 km s@xmath27 mpc@xmath27 .
all data are photometrically calibrated to the ab system , using the latest acs zeropoints published on the hst instrument web page .
observations for this paper consist of deep hst imaging of six clusters at @xmath6 taken with the advanced camera for surveys ( acs ) in the @xmath0 ( f435w ) , @xmath1 ( f475w ) , @xmath2 ( f555w ) , @xmath3 ( f625w ) , @xmath4 ( f775w ) and @xmath5 ( f850lp ) bands with exposure times of 5,00010,000 seconds in each band .
each cluster was observed in a single acs exposure , covering about @xmath28 on the sky .
the data were originally taken for studies of gravitational lenses in these clusters .
table 1 shows the clusters observed , their redshifts , exposure times in each band , and hst proposal ids .
table 2 summarizes some essential physical characteristics for the observed clusters .
most data are from the compilation of @xcite , from which @xmath29 can be estimated via the formula of @xcite , except for a1413 where @xmath29 was derived by @xcite from the x - ray profile . we were unable to find data in the literature for a1703 .
the data were retrieved from the hst archive as fully reduced and flatfielded files ( * .flt ) and then processed through the multidrizzle algorithm @xcite to produce fully registered and co - added images for each band .
1 shows color composites for our data .
full resolution jpegs will be made available on the @xmath30 web site .
detection of objects and photometry were carried out using sextractor @xcite .
we experimented with sextractor parameters to maximize our detections and minimize the number of spurious objects .
all detections were visually examined to eliminate noise spikes , bleed trails from bright stars and other contaminants , especially the numerous arclets present in the images .
the sextractor parameters eventually employed are shown in table 3 .
we used the same parameters for all images . for each object
we compute a total ( kron - like ) magnitude and an aperture magnitude of area equivalent to the minimum number of connected pixels needed for detection .
this provides an estimate of the central surface brightness for each object .
the motivation behind this procedure is as follows .
detections of objects in deep images depends on their total magnitude but also on their central surface brightness .
an object could be brighter than the magnitude threshold but be lost in the night sky because of low surface brightness ( see discussion in @xcite ) .
for this reason we need to determine both a limiting total magnitude for completeness and a surface brightness threshold for detection .
we do this by plotting central surface brightness vs. total magnitude for all bands in fig . 2 . to save space we only show the data for abell 1703 , but equivalent figures for all other clusters are available on the @xmath30 web site . in these figures
the sequence at high surface brightness that provides an upper limit to the scatter plot is caused by stars .
we can therefore discriminate against stellar contamination in this way ( cf .
the surface brightness threshold is chosen empirically .
the limit is selected to include as many objects as possible , but avoiding regions of the parameter space where the sample is obviously very incomplete .
selection lines in surface brightness , apparent magnitude and the star - galaxy separation line are shown in fig .
2 . in order to determine a limiting magnitude , we plot the raw number counts for galaxies in the abell 1703 field in fig .
3 . as for fig . 2
, figures for all the other clusters are made available electronically .
the completeness magnitude is chosen to lie about 0.5 mag .
brighter than the luminosity at which counts start to decrease , in order to select a highly complete sample of objects . the only way to establish cluster membership for faint galaxies is by statistical background subtraction . in order to do so , we need to analyze ` blank ' ( cluster - less ) comparison fields of similar or greater photometric depth .
we choose to use the two goods fields @xcite as these are the deepest and widest fields where hst multicolor photometry is available .
we used the same sextractor parameters as we used for the cluster fields and impose the same selection limits on the goods fields as we did for the target fields .
4 shows the detection and selection plots for the goods fields ( similar to fig . 2 above ) .
the fields are much deeper than the cluster fields we use .
it must be noted that we plot only @xmath18 of the detections in the goods fields in fig .
4 to avoid saturating the figure .
fig . 5 plots
galaxy number counts for the goods fields and comparison galaxy number counts from the literature .
the hst data were corrected to the sdss and johnson - cousins systems using the transformations tabulated in @xcite .
goods number counts were fitted with a quadratic of the form @xmath31 .
the fit was carried out over galaxies brighter than the completeness limit shown in fig . 4 and fainter than m@xmath32 mag .
as goods counts for brighter galaxies are affected by small scale clustering .
table 4 shows the values of the coefficients for the quadratic fits .
we now subtract the scaled contribution from fore / background counts from galaxy counts . errors in galaxy counts for clusters and the goods fields
are assumed to be poissonian .
we used our smoothed quadratic for the background counts , extrapolating for galaxies brighter than @xmath33 mag .
field counts at these bright limits and over the small ( @xmath34 arcmin@xmath35 ) covered by acs are very small in any case . clustering errors over the cluster and goods fields are estimated following the counts - in - cells approach and the fitting formula described by @xcite
. it may be possible to further refine the selection of cluster members using photometric redshifts or color cuts .
contamination of the lf by background galaxies has been suggested as a possible cause of the faint end upturn ( see below ) by @xcite .
however , this would lead to loss of generality .
the argument by @xcite may be correct for nearby clusters ( although @xcite point out that the simulations used by @xcite are unrealistic and that the background subtraction approach used here works as long as proper care is taken of statistical uncertainties , see their 3.3.6 ) , but it is not appropriate for these distant objects where the redshift is much larger than the typical scale of structure in the universe ( so that the counts should be smooth ) .
there are some mismatches between the goods data ( used for background subtraction ) and the cluster data . for the @xmath1 ( f435w ) and @xmath3 ( f625w ) bands
, we can use the goods f475w and f606w counts ( respectively ) , which are close matches to the bands used , however , for f555w ( @xmath2 ) we used the average of the f475w and f606w counts .
the lf in this bands is more uncertain , because we do not have an appropriate background field .
we @xmath36 and @xmath37 correct the lfs to @xmath38 for ease of comparison with local data . in order to do
so we use a @xcite model for a solar metallicity elliptical formed at @xmath39 with an e - folding time of 1 gyr .
this is appropriate to the red ellipticals that dominate the bright end of cluster lfs . in effect
this approach allows us to consider the differential evolution , if any , between the dwarfs and the giants , who appear to consist mainly of old galaxies with little or no recent star formation .
the lf of each individual cluster is relatively uncertain .
we therefore build composite lfs in each band following the method by @xcite in order to average out errors .
the method assumes that the lf is broadly universal , but this appears to be borne out by observations at low redshift , at least for relatively bright galaxies @xcite .
composite lfs are built as follows : the number of cluster galaxies in the @xmath40 magnitude bin of the composite lf is given by : @xmath41 where @xmath42 is the ( background corrected ) number of galaxies in the @xmath40 magnitude bin of the lf of the @xmath43 cluster lf , @xmath44 is the normalization of the cluster lf ( to take care of the different richnesses of the clusters ) and is taken to be the background corrected number of cluster galaxies brighter than @xmath45 ( or its equivalent in the other bands , assuming the color of an old elliptical ) , @xmath46 is the number of @xmath4 clusters contributing to the @xmath40 magnitude bin of the composite lf and @xmath47 is the sum of all the @xmath4 normalizations .
@xmath48 essentially , this carried out a weighted average of the individual cluster lfs , scaled by the number of galaxies ( in each cluster ) brighter than approximately the @xmath10 point .
the formal errors on @xmath49 are computed as : @xmath50^{1/2}\ ] ] where @xmath51 are the errors on the number counts ( including poissonian and clustering contributions ) for the @xmath40 magnitude bin of the @xmath43 cluster lf . for each cluster
we build the composite lf by summing up the individual lfs in absolute magnitude bins , after correcting for the distance modulus , extinction and @xmath52 correction to the common redshift of @xmath38 .
the resulting lfs are presented in fig .
6 , together with the 1 , 2 and 3@xmath53 error countours .
we fit the data to a single schechter function ( see discussion below , regarding the existence of a faint end upturn ) . in some cases
we exclude the fainter and/or brighter magnitude bin from the fit , because these points are dominated by either the poorly sampled very massive galaxies or the fainter objects that suffer from complex incompleteness due to magnitude and surface brightness selection .
the fit parameters are tabulated in table 5 , together with the @xmath54 errors for each parameter , marginalized over the remaining parameters .
there are only a few deep cluster lfs having comparable depth to our own . in general , we are in good agreement with previous work , based on smaller and shallower samples @xcite .
the @xmath15 point we measure gets steadily brighter with increasing wavelength , tracking the color of old , passively evolving stellar populations that are the main component of bright elliptical galaxies @xcite .
the value we find is also consistent with the local sdss value by @xcite : this is not surprising , as it is well known that massive galaxies evolve passively since high redshift .
the main interest of this investigation is in the evolution of the lf slope .
the lf is well fitted by a single schechter function , with a slope @xmath55 which is essentially the same in all bands .
this suggests that the cluster lf is dominated by galaxies on the red sequence to a luminosity of @xmath56 in all bands , and that therefore the red sequence continues at least to 6 magnitudes below the @xmath10 point even at @xmath11 .
this can be clearly seen in fig .
7 where we show the @xmath57 vs. @xmath5 color - magnitude relation for abell 1703 .
the relation can be clearly followed at least to @xmath58 with little scatter .
a more detailed analysis of the color - magnitude relation is deferred to a future paper ( harsono & de propris 2008 , in preparation ) .
the red sequence appears to be well established and contain the majority of cluster populations even at a lookback time of @xmath59 gyrs ( cf .
@xcite for a similarly deep relation in local clusters ) .
the slope is very similar to the local value for @xmath22 derived by @xcite and @xcite .
this argues that even galaxies to @xmath56 have formed their stellar populations and assembled the majority of their mass at least at @xmath11 .
we can therefore place a significant lower limit to the assembly epoch of galaxies down to 1/600@xmath60 of the mass of the milky way . given the evidence for truncation of the red sequence at @xmath61 @xcite , our data point to the @xmath62 interval as a crucial epoch to investigate the formation of the fainter galaxy populations .
the other issue we wish to address is the faint end upturn .
this has been a controversial subject , ever since its original discovery by @xcite and @xcite .
there have been a series of claims and counterclaims regarding the faint end slope and the existence of an upward inflection , sometimes even in the same cluster .
for instance , @xcite and @xcite observe a steep lf at the faint end for the coma cluster , while @xcite and @xcite quote much flatter slopes .
in virgo , @xcite claim a slope as steep as @xmath63 , while @xcite derive @xmath64 .
@xcite review the existence of the upturn and the shape of the faint end of the lf and conclude that there is a steep mass function .
the observation of the upturn in two composite lfs , derived by two different groups @xcite , suggests that the upward inflection of the lf is real ( however , cf . , @xcite ) .
unlike @xcite and @xcite we do not find a steepening of the lf at @xmath13 .
however , we must consider that the lf upturn is more pronounced at large clustercentric radii , while our fields generally cover only the central region of each cluster .
the size of the acs fields used in this study covers between 350 and 750 @xmath65 kpc on the side . in terms of @xmath66 ( the radius
at which @xcite and @xcite normalize their lfs ) , the areas imaged by acs cover between 20% and 40% of the area of the cluster out to @xmath29 .
we therefore only derive lfs for the cluster cores .
this should not affect our comparisons with the bright end of the lf ( @xmath22 ) , as this does not seem to vary significantly with radius .
however , the steep upturn claimed by @xcite and @xcite is much more pronounced at large clustercentric radii , i.e. , in the cluster outskirts . on the other hand it is possible to see the upturn even in the more central regions of the sample studied by @xcite , inside @xmath67 ( see fig . 10 of @xcite ) , so our data should exhibit an upturn at faint magnitudes , although of course more clusters would be welcome to bolster the argument .
as we do not detect a steepening in the lf , we suggest that , if the steepening is real , the faint galaxies contributing to the upturn consist of a population of recently infalling objects ( e.g. , @xcite ) whose star formation is curtailed by the cluster environment
. this would be consistent with the observation that about 1/2 of the fainter cluster dwarfs in virgo and elsewhere have been forming stars until relatively recently @xcite , until their star formation stopped and their colors moved to the red sequence
. the newly infalling population may be the one now contributing to the steep upturn . in a ` downsizing ' model
, it may be expected that objects undergoing star formation suppression and cluster infall at the present epoch would indeed tend to be among the fainter and less massive galaxies and to have the steep lf characteristic of ` pristine ' cdm power spectra .
our observations imply that the passive evolution of bright galaxies can be extended to faint dwarfs , at least to @xmath11 and suggest that the majority of galaxy evolution may have taken place at surprisingly early epochs even for the least massive objects .
all of the data presented in this paper were obtained from the multimission archive at the space telescope science institute ( mast ) .
stsci is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
support for mast for non - hst data is provided by the nasa office of space science via grant nag5 - 7584 and by other grants and contracts .
adami , c. , nichol , r. c. , mazure , a. , durret , f. , holden , b. & lobo , c. 1998 , , 334 , 765 adami , c. , picat , j. p. , durret , f. , mazure , a. , pell , r. & west , m. 2007 , , 472 , 749 andreon , s. , punzi , g. , grado , a. 2005 , , 360 , 727 andreon , s. 2006a , , 448 , 447 andreon , s. 2006b , , 369 , 969 andreon , s. , cuillandre , j .- c .
, puddu , e. & mellier , y. 2006 , , 372 , 60 baldry , i. k. , glazebrook , k. & driver , s. p. 2008
, , 388 , 945 barkhouse , w. a. , yee , h. k. c. & lopez - cruz o. 2007 , , 671 , 1471 beijersbergen , m. , hoekstra , h. , van dokkum , p. g. & van der hulst , t. 2002 , , 329 , 385 bernstein , g. m. , nichol , r. c. , tyson , j. a. , ulmer , m. p. & wittman , d. 1995 , , 110 , 1507 bertin e. , arnouts s. 1996 , , 117 , 393 boyce , p. j. , phillipps , s. , jones , j. b. , driver , s. p. , smith , r. m. , & couch , w. j. 2001 , , 328 , 277 bruzual , g. & charlot , s. 2003 , , 344 , 1000 capak , p. et al .
2007 , , 172 , 99 carlberg , r. g. , yee , h. k. c. & ellingson , e. 1997 , , 478 , 462 crawford , s. m. , bershady , j. a. & hoessel , j. g. 2008 , astro.ph 0809.1661 colless , m. m. 1989 , , 237 , 799 colless , m. m. et al .
2001 , , 328 , 1039 conselice , c. j. , gallagher , j. s. & wyse , r. f. g. 2001 , , 559 . 791 conselice , c. j. , gallagher , j. s. & wyse , r. f. g. 2003 , , 125 , 66 cool , r. j. , eisenstein , d. j. , scranton , r. , brinkmann , j. , schneider , d. p. & zehavi , i. 2006 , , 131 , 736 cross , n. j. g. & driver , s. p. 2002
, , 329 , 579 croton , d. et al .
2006 , , 365 , 11 de lucia , g. et al .
2007 , , 374 , 809 de propris , r. , pritchet , c. j. , hartwick , f. d. a. & mcclure , r. d. 1995 , , 450 , 534 de propris , r. , eisenhardt , p. r. , stanford , s. a. & dickinson , m. 1998 , , 503 , l45 de propris , r. , stanford , s. a. , eisenhardt , p. r. , holden , b. p. & rosati , p. 2007
, , 133 , 2207 dressler , a. 1980 , , 236 , 351 driver , s. p. , phillipps , s. , davies , j. i. , morgan , i. , disney , m. j. 1994 , , 268 , 393 eisenhardt , p. r. , de propris , r. , gonzalez , a. h. , stanford , s. a. , wang , m. & dickinson , m. 2007 , , 169 , 225 garilli , b. , maccagni , d. & andreon , s. 1999 , , 353 , 479 giavalisco , m. et al .
2004 , , 600 , l93 gomez , p. l. et al .
2003 , , 584 , 210 harsono d. & de propris r. 2007 , , 380 , 1036 hilker , m. , mieske , s. & infante , l. 2003 , , 397 , l9 holberg , j. b. & bergeron p. 2006
, , 132 , 1221 huang j .- s . ,
cowie l. l. , gardner j. p. , hu e. m. , songaila a. , wainscoat r. j. 1997 , , 476 , 12 jenkins , l. p. , hornschemeier , a. e. , mobasher , b. , alexander , d. m. & bauer , f. e. 2007 , , 666 , 846 koekemoer , a. , fruchter , a. s. , hook , r. n. & hack , w. 2002 , in _
the 2002 hst calibration workshop : hubble after the installation of acs and the nicmos cooling system _ ed . by s. arribas , a. koekemoer , b. whitmore ( baltimore : space telescope science institute ) , p. 337
khochfar , s. , silk , j. , windhorst , r. a. & ryan , r. e. 2007 , , 668 , l115 krick , j. e. , surace , j. a. , thompson , d. , ashby , m. l. n. , hora , j. l. , gorjian , v. & yan , l. 2008 , astro - ph 0807.1565 lewis , i. j. et al .
2002 , , 334 , 673 macdonald , e. c. et al .
2004 , , 352 , 1255 mercurio , a. , massarotti , m. , merluzzi , p. , girardi , m. , la barbera , f. & busarello , g. 2003 , , 408 , 57 milne , m. l. , pritchet , c. j. , poole , g. b. , gwyn , s. d. j. , kavelaars , j. j. , harris , w. e. & hanes , d. a. 2007 , , 133 , 177 muzzin , a. , wilson , g. , lacy , m. , yee , h. k. c. & stanford , s. a. 2008 , astro - ph 0807.0227 penny , s. j. & conselice , c. j. 2008 , , 383 , 247 pointecouteau , e. , arnaud , m. & pratt , g. w. 2005 , , 435 , 1 popesso , p. , biviano , a. , bhringer , h. & romaniello , m. 2006 , , 445 , 29 pracy , m. , de propris , r. , driver , s. p. , couch , w. j. & nulsen , p. e. j. 2004 , , 352 , 1135 rines , k. & geller , m. j. 2008 , , 135 , 1837 sabatini , s. , davies , j. i. , scaramella , r. , smith , r. , baes , m. , linder , s. m. , roberts , s. & testa , m. 2005 , , 357 , 819 schechter , p. 1976
, , 203 , 297 smith , r. j. et al .
2008 , , 386 , l96 stott , j. p. , smail , i. , edge , a. c. , ebeling , h. , smith , g. p. , kneib , j .-
& pimbblet , k. a. 2007 , , 661 , 95 trentham , n. 1998a , , 295 , 360 trentham , n. 1998b , , 293 , 71 valotto , c. , moore , b. & lambas , d. g. 2001 , , 546 , 157 wright , e. l. 2006 , , 118 , 1711 wu , x .-
, xue , y .- j . &
fang , l .- z . 1999 , , 524 , 22 yasuda , n. et al . 2001 , , 122 , 1104 york , d. g. et al .
2000 , , 120 , 1579 lcccccccl abell 1413 & 0.143 & & & & & 2440 & 2560 & 9292 + abell 2218 & 0.171 & 7048 & 5640 & 7048 & 8386 & 10732 & 5640 & 9452 , 9717 , 10325 + abell 1689 & 0.183 & & 9500 & & 9500 & 11800 & 16600 & 9289 + abell 1703 & 0.258 & 7050 & 5564 & 5564 & 9834 & 11128 & 17800 & 10325 + ms
1358.4 + 6245 & 0.328 & 7928 & 5470 & 5482 & 9196 & 10952 & 15017 & 9292 , 9717 , 10325 + cl 0024.0 + 1652 & 0.395 & 6435 & 5072 & 5072 & 8971 & 10144 & 16328 & 10325
+ lcccccc abell 1413 & 0.143 & & @xmath68 & @xmath69 & 1.71 & 21 + abell 2218 & 0.171 & @xmath70 & @xmath71 & @xmath72 & 2.94 & 14 + abell 1689 & 0.183 & @xmath73 & @xmath74 & @xmath75 & 2.24 & 19 + abell 1703 & 0.258 & & & & & + ms 1358.4 + 6245 & 0.328 & @xmath76 & @xmath77 & @xmath78 & 2.40 & 28 + cl 0024.0 + 1652 & 0.395 & @xmath79 & & @xmath80 & 1.95 & 39 + cccc @xmath0 ( f435w ) & @xmath89 & @xmath90 & @xmath91 + @xmath1 ( f475w ) & @xmath92 & @xmath93 & @xmath94 + @xmath2 ( f555w ) & @xmath95 & @xmath96 & @xmath97 + @xmath3 ( f625w ) & @xmath98 & @xmath99 & @xmath100 + @xmath4 ( f775w ) & @xmath101 & @xmath102 & @xmath103 + @xmath5 ( f850lp ) & @xmath104 & @xmath105 & @xmath106 + | we present deep composite luminosity functions in @xmath0 , @xmath1 , @xmath2 , @xmath3 , @xmath4 and @xmath5 for six clusters at @xmath6 observed with the hubble space telescope advanced camera for surveys .
the luminosity functions reach to absolute magnitude of @xmath7 mag . and
are well fitted by a single schechter function with @xmath8 mag . and @xmath9 ( in all bands ) .
the observations suggest that the galaxy luminosity function is dominated by objects on the red sequence to at least 6 mags . below the @xmath10 point .
comparison with local data shows that the red sequence is well established at least at @xmath11 down to @xmath12 of the luminosity of the milky way and that galaxies down to the regime of dwarf spheroidals have been completely assembled in clusters at this redshift .
we do not detect a steepening of the luminosity function at @xmath13 as is observed locally .
if the faint end upturn is real , the steepening of the luminosity function must be due to a newly infalling population of faint dwarf galaxies . |
in three previous papers we have developed a method of measuring galaxy power spectrum with discrete wavelet transform ( dwt ) decomposition , which is an alternative of the fourier power spectrum detection ( fang & feng , 2000 ( paper i ) , yang et al 2001a ( paper ii ) , 2001b ( paper iii ) ) .
the dwt power spectrum estimator is information - loseless , and optimized in the sense that the spatial resolution is adaptive automatically to the scale to be studied .
a test with observed sample of the lcrs galaxies showed that the dwt estimator can give a robust measurement of the power spectrum . in this paper
, we continue our effort in this direction .
the topic this time is to develop theory and algorithm of redshift distortion in the dwt representation . in terms of power spectrum measurement ,
the central problem of redshift distortion is to find the mapping between the power spectra in redshift and physical spaces . in our paper ii
, we have already studied this mapping . however , that mapping is directly obtained by a wavelet transform of the mapping of the power spectrum in the fourier representation ( peacock & dodds 1994 ) .
although a fourier mode can be transformed into dwt modes and _ vice versa _ , we should be careful in doing the transform of a second order statistical quantity from the fourier representation into the dwt one .
as we have showed in paper i , the covariance of density contrast in the dwt representation @xmath1 is not equivalent to the fourier counterpart @xmath2 , because the fourier mode is subjected to the central limit theorem , while mode localized in both scale and physical space may not be so .
this is because the dwt mode is characterized by not only the scale , but also the configuration ( or shape ) of the mode .
simply speaking , the variance @xmath1 contains information of both the scale and phase ( shape ) , while @xmath3 does not contain information of phases ( fang & thews 1998 ) .
we have found in paper ii that , the fourier mapping of redshift distortion can be employed for diagonal dwt power spectrum , but not off - diagonal dwt power spectrum .
this is due to the shape(phase)-dependence of dwt modes .
a 3-d fourier mode with wave - vector @xmath4 can be transformed into mode @xmath5 by a coordinate rotation as long as @xmath6 .
therefore , for an isotropic random field , the fourier modes with the same @xmath7 are statistically equivalent . however , the dwt modes do not share the same property
. the length scale of a 3-d wavelet mode with scale index @xmath8 is @xmath9^{1/2}$ ] , where @xmath10 is the length scale of the sample .
generally , one can not transform a mode @xmath11 to @xmath12 by a rotation , even if they have the same scale . as a consequence ,
the redshift distortion in the dwt presentation will be shape - dependent .
for instance , in the fourier representation , the linear redshift distortion factor @xmath13 ( kaiser 1987 ) , where @xmath0 is the so - called redshift distortion parameter , depends only on @xmath14 , i.e. the cosine of the angle between the wavevector @xmath15 and the line - of - sight . while in the dwt representation , the linear redshift distortion factor will depend on not only the scale , but also the shape of dwt modes .
an important application of redshift distortion is to determine the redshift distortion parameter @xmath0 ( e.g. hamilton 1998 ) , which contains valuable information of the cosmological mass density parameter and the bias of galaxies .
the shape - dependence of redshift distortion is very useful for the parameter determination .
we will develop an algorithm of @xmath0 estimation with diagonal and off - diagonal dwt power spectra .
moreover , the dwt representation provides an easy way to study the effect of selection function on the redshift distortion . in the first three papers
, we assumed that the selection function @xmath16 is known .
actually , the known " is not necessary . according to the definition
, selection function is an observed galaxy distribution if galaxy clustering is absent .
therefore , for a consistent algorithm of power spectrum , the selection function should be and can be obtained by the galaxy distribution itself .
we will show that in the dwt representation the selection function can be determined by the distribution of galaxies without other assumption .
this algorithm is convenient to estimate the contribution of selection function to redshift distortion .
the paper will be organized as follows .
2 is a summary of the algorithm of the dwt power spectrum . in 3 , we develop a theory of the redshift distortion with a multiresolution analysis .
the emphases are the real - redshift mapping of diagonal and off - diagonal power spectrum .
we proposed @xmath0-estimators , which are tested by n - body simulation samples of popular dark matter models ( 4 and 5 ) .
finally , the conclusions and discussions will be presented in 6 .
the mathematical stuffs with the relevant dwt quantities are given in appendix .
we summarize the algorithm of the dwt power spectrum .
the details can be found in fang & feng ( 2000 ) and yang et al ( 2001 ) .
if the position measurement is perfectly precise , the number density distribution of galaxies can be written as @xmath17\ ] ] where @xmath18 is the total number of galaxies , @xmath19 is position of @xmath20th galaxy , @xmath21 is its weight , and @xmath22 the 3-d dirac @xmath23 function .
@xmath16 is selection function given by the mean number density of galaxies when galaxy clustering is absent , and @xmath24 is the density contrast fluctuation in the underlying matter distribution . without loss of generality , we enclose the sample in a cubic box with side lengths @xmath25 . in the dwt representation , @xmath26 is decomposed into @xmath27 where , @xmath28 is given by @xmath29 the index @xmath8 stand for 3-d scales @xmath30 , and index @xmath31 for the position of cell @xmath32 , @xmath33 , @xmath34 . in eq.(3 ) , @xmath35 is called scaling function coefficient ( sfc ) of @xmath26 .
they are given by @xmath36 where we have used equation ( 1 ) in the last step .
the scaling functions @xmath37 play the role of window functions for the cell with volume @xmath38 , and located at position @xmath39 .
therefore , the sfc @xmath35 is the mean of field @xmath26 in the volume @xmath38 at position @xmath39 .
thus , the term @xmath28 of eqs.(2 ) and ( 3 ) actually is the smoothed @xmath28 by window on scale @xmath40 .
the 3-d scaling functions @xmath37 can be constructed by a direct product of 1-d scaling functions , i.e. @xmath41 in eq.(2 ) , @xmath42 are wavelets .
an 1-d wavelet @xmath43 is the modes used to extract the fluctuations of a 1-d field @xmath44 on scale @xmath45 at position @xmath39 .
the 3-d wavelets @xmath42 are given by mixed direct products of 1-d scaling functions @xmath46 and wavelets @xmath47 as follows , @xmath48 accordingly , @xmath49 with @xmath50 describe 2-d projected fluctuations on a slice , and @xmath51 1-d fluctuations along a line .
@xmath52 describes the 3-d fluctuations inside the cell @xmath53 on a scale @xmath54 .
hereafter , we consider the projection of density fluctuations onto the space spanned by @xmath52 only . for simplicity , we ignore the upper index @xmath55 in the notations of @xmath49 and @xmath56 without confusion .
thus , the wavelet function coefficient ( wfc ) @xmath57 of @xmath26 is given by @xmath58 therefore , the wfc @xmath57 is the amplitude of the fluctuations of the field @xmath26 on scale @xmath38 at position @xmath39 . the functions @xmath37 and @xmath59 form a complete and orthonormal basis in space of functions defined in 3-dimensional coordinate space .
the decomposition of eq.(2 ) is information loseless .
the dwt power spectrum of @xmath24 can be estimated by @xmath60 ^ 2}{n^2({\bf j , l})}\right \rangle -\left \langle \frac{1}{n({\bf j , l})}\right \rangle,\ ] ] where @xmath61 is the average over ensemble . the first term on the r.h.s .
is the normalized power from @xmath26 , and the second term corrects for the poisson noise .
the normalization factor @xmath62 is the mean selection function in the mode @xmath63 , i.e. , the number density of galaxies in the mode @xmath63 when galaxy clustering is absent .
as has been shown in paper i ( fang & feng 2000 ) , the factor @xmath64 can be absorbed into the weight factor by @xmath65 thus , equation(8 ) yields @xmath66 ^ 2 - \frac{1}{2^{j_1+j_2+j_3}}\sum_{l_1=0}^{2^{j_1}-1 } \sum_{l_2=0}^{2^{j_2}-1}\sum_{l_3=0}^{2^{j_3}-1 } \int\frac{\psi_{\bf j , l}^2({\bf x})}{\bar{n}({\bf x } ) } d{\bf x},\ ] ] where the wfc @xmath67 is defined as @xmath68 by definition eq.(1 ) , selection function @xmath16 is an observed galaxy distribution if galaxy clustering @xmath69 is absent .
that is , equation ( 1 ) requires to decompose an observed distribution @xmath26 into two parts : the `` background '' @xmath16 , and the fluctuations @xmath24 upon the background .
this decomposition is not easy if the selection function @xmath16 is position - dependent , i.e. it will mix with the fluctuations to be detected . with the dwt analysis ,
one is capable of performing this decomposition by a scale - by - scale analysis . when we study the fluctuations on a scale @xmath40 , all @xmath70-dependencies of @xmath26 on scales larger than this scale play the role as a background .
thus , in terms of the scale @xmath40 , the background is given by a smooth of @xmath26 on the scale @xmath40 , i.e. it does not contain information of fluctuations on scales equal to or less than @xmath40 .
this background has already be found from equations ( 2 ) and ( 3 ) . because of the orthogonal relation @xmath71 the function @xmath28 does not contain any information of fluctuations of modes @xmath72 with @xmath73 .
thus , to detect the fluctuation power on the scale @xmath40 , the function @xmath28 is recognized as a background field .
one can identify the selection function as @xmath28 , i.e. @xmath74 where the superscript @xmath40 means that this `` selection function '' is only for the scale @xmath40 . in the plane
parallel approximation , selection function depends on @xmath75 only , i.e. the coordinate in redshift direction . from equation ( 13 ) , the selection function is given by @xmath76 from equation ( 6 ) , the sfc @xmath77 is actually given by an average of @xmath26 over the plane @xmath78 , and decomposition along @xmath75 direction .
using equation ( 14 ) , the wfc @xmath67 is now calculated by @xmath79 which presents a simple algorithm for deriving the selection function from observed galaxy samples .
to test this algorithm , we produce mock galaxy samples using n - body simulation , of which the details will be given in 4 . for simplicity
, the selection effect is applied along one axis ( e.g. , @xmath75 direction ) of 3-dimensional cartesian coordinates under the plane parallel approximation . for the simulation sample in the cubic box with a side length of 256 h@xmath80 mpc , we replicate the sample along @xmath75 direction , and choose the mock galaxies located between 100 - 356 h@xmath80 mpc .
a selection function is given by @xmath81 where @xmath10 is the size of the sample . to be comparable with , for example
, the lcrs selection function , the parameters are adopted to be @xmath82 h@xmath80 mpc , @xmath83 and @xmath84 .
1 displays the dwt diagonal power spectrum @xmath85 of particle distributions for three typical models with the selection function eq.(14 ) , in which we take @xmath86 .
it shows that the power spectrum estimator eqs.(10 ) and ( 15 ) can perfectly recover the dwt power spectrum regardless the selection functions ( 16 ) .
this result will keep unchanged if @xmath87 .
it means that @xmath88 gives a proper estimation of selection function if @xmath89 is high enough .
practically , one can find a properly recovered power spectrum by checking whether it is insensitive to @xmath89 .
it should be pointed out that with the developed algorithm the power spectrum eq.(8 ) is normalized scale - by - scale .
the fluctuation amplitudes @xmath90 ( wfcs ) on scale @xmath40 is normalized by @xmath62 or @xmath91 [ eq.(15 ) ] , which contains fluctuations on all scales larger than @xmath40 .
therefore , the normalization factor generally is scale - dependent .
this is different from conventional normalization , which is scale - independent .
if the field is gaussian , there is no correlation between @xmath90 and @xmath62 or @xmath91 .
eq.(8 ) yields @xmath92 ^ 2\right \rangle - \left \langle \frac{1}{n({\bf j , l})}\right \rangle.\ ] ] it has been shown with the so - called partition of unity " of wavelets that @xmath93 is approximately independent of @xmath40 . in this case , the power spectrum eq.(8 ) is the same as that given by conventional normalization ( jamkhedkar , bi & fang 2001 ) .
however , for non - gaussian field , especially , if the perturbations on different scales are correlated , the small scale fluctuations given by the wfc @xmath90 generally are correlated with large scale fluctuation contained in @xmath62 or @xmath91 . in this case , the power spectrum eq.(8 ) is very different from usual power spectrum which does nt not sensitive to the correlation between perturbations on different scales .
the power spectrum defined by eq.(8 ) will benefit to calculate the effect of selection function upon the redshift distortion ( 3.3 ) .
the redshift distortion is due to peculiar motion of galaxies . for a given mass field @xmath24
, the galaxy velocity @xmath94 is a random field with mean @xmath95 where @xmath96 denotes ensemble average for velocity fields .
@xmath97 is the bulk velocity at @xmath98 . in linear regime ,
the bulk velocity is related to the density contrast by @xmath99 where @xmath100 is the redshift distortion parameter at present , i.e. redshift @xmath101 .
the _ rms _ deviation of velocity @xmath102 from the bulk velocity @xmath97 is @xmath103 ^ 2\rangle_v = \sigma^2_{pv}({\bf x}).\ ] ] in the dwt representation , one can calculate @xmath104 by the variance of the wfcs of velocity field , i.e. @xmath105 , which is actually the dwt pairwise peculiar velocity ( paper iii ) . in the paper iii , we analyzed the pvd of the velocity fields given by the n - body simulation for the cdm family of models . on large scales ,
the velocity field is basically gaussian .
it is completely described by its mean @xmath97 and variance @xmath106 . on small scales ,
the velocity field is significantly non - gaussian .
the one point function of @xmath107 is lognormal .
the two - point pairwise velocity correlation functions are non - zero on scales @xmath108 1 h@xmath80 mpc . moreover , on these scales , the pairwise velocities are also correlated significantly with density fluctuations .
the position of galaxy @xmath20 in redshift space is given by @xmath109 , where @xmath110 is the radial component of @xmath102 .
the number density distribution in redshift space is then @xmath111= \bar{n}^s({\bf s})[1+\delta^s({\bf s})],\ ] ] where @xmath112 is the selection function in redshift space .
similar to equation ( 8) , the power spectrum in redshift space is @xmath113 ^ 2 - \frac{1}{2^{j_1+j_2+j_3}}\sum_{l_1=0}^{2^{j_1}-1 } \sum_{l_2=0}^{2^{j_2}-1}\sum_{l_3=0}^{2^{j_3}-1 } \int\frac{\psi_{\bf j , l}^2({\bf x})}{\bar{n}^s({\bf x } ) } d{\bf x},\ ] ] where @xmath114 the selection function @xmath112 can be determined from the observed distribution @xmath115 by equation ( 13 ) or ( 14 ) .
it can then be absorbed into the weight @xmath21 as equation ( 11 ) or ( 15 ) .
the effect of the difference between @xmath115 and @xmath116 will be studied in next section . in this section , we will ignored this effect .
thus , using an auxiliary vector @xmath117 , equations ( 21 ) and ( 23 ) yield @xmath118 where @xmath119 is gradient operator on @xmath120 . subjecting equation ( 24 ) to an average over ensemble of velocity ,
if the velocity field is _ gaussian _ , we have @xmath121 for simplicity , we will use @xmath122 for @xmath123 below without causing confusion .
if we consider only linear effect of the bulk velocity , equation ( 24 ) gives @xmath124 e^{-(1/2)\sigma^2_{pv}({\bf x}_i)({\bf j}\cdot\hat{\bf r})^2 } \psi_{\bf j , l}({\bf s } ) \right |_{{\bf j}=0 } \\
\nonumber & = & \sum_{i=1}^{n_g}w_i \int d{\bf s } \psi_{\bf j , l}({\bf s } ) e^{-(1/2)\sigma^2_{pv}({\bf s})(\hat{\bf r}\cdot \nabla ) ^2 } \delta_d ( { \bf s } -{\bf x}_i ) \\ \nonumber & - & \frac{1}{h_0}\sum_{i=1}^{n_g}w_i \int d{\bf s } \left .
\hat{\bf r } \cdot [ \nabla_s v_r({\bf s}+i\nabla_j ) \delta_d \left ( { \bf s } -{\bf x}_i+ i \nabla_j \right ) ] e^{-(1/2)\sigma^2_{pv}({\bf x}_i)({\bf j}\cdot\hat{\bf r})^2 } \psi_{\bf j , l}({\bf s } ) \right |_{{\bf j}=0 } \\
\nonumber\end{aligned}\ ] ] neglecting the terms of the order of @xmath125 , and using the linear relation eq.(19 ) , equation ( 26 ) gives @xmath126 because all operators in the integrand of equation ( 27 ) are nearly diagonal in the dwt representation ( farge et al 1996 ) , equation ( 27 ) can be rewritten as @xmath127 where @xmath128 and @xmath129 the method of calculating @xmath130 and @xmath131 is presented in appendix a. substituting equation ( 28 ) into equation ( 22 ) , we have the redshift distorted power spectrum as @xmath132 where @xmath133 ^ 2 $ ] . above equation
formulates the redshift distortion effect in dwt expression .
this is the basic formula for our redshift distortion analysis .
usually , the factor @xmath134 is called linear redshift distortion , and @xmath135 called non - linear redshift distortion due to the pairwise velocity dispersion . however in our derivation , the two parts are not treated separately .
this derivation can be generalized to any velocity fields , which are not simply described by equations ( 18 ) and ( 20 ) .
for instance , if the pairwise velocities are correlated , i.e. @xmath103[v_i({\bf x'})-v_i({\bf x'})]\rangle_v = \sigma^2_{pv}({\bf x - x'}),\ ] ] the pairwise velocity dispersion factor becomes @xmath136 } \psi_{\bf j , l}({\bf x } ) \psi_{\bf j , l}({\bf x'}),\end{aligned}\ ] ] where @xmath137 is gradient operator on @xmath138 .
the theory of redshift distortion presented in last section did not consider the effects of selection functions , @xmath139 and @xmath140 .
since selection function relies on the radial distance , it could be also a source of the anisotropy of power spectrum with respect to redshift direction .
it should be taken into account when analyze observed samples . in the linear approximation of @xmath110 , equation ( 21 )
gives @xmath141.\ ] ] in the second steps , @xmath110 is replaced by @xmath142 , as only the bulk velocity is considered .
using equations ( 1 ) and ( 21 ) , equation ( 34 ) yields @xmath143 \right \},\ ] ] where the second order term @xmath144 is ignored .
the term -1 on the r.h.s . of equation ( 35 ) does not contribute to power spectrum , because of @xmath145 .
in the linear approximation , @xmath146 and therefore , the factor @xmath147 in the third term of the r.h.s . of equation ( 34 ) can be approximated as 1 . the second term in the r.h.s .
of equation ( 35 ) contains a linear term of @xmath142 , i.e. the same order as the third term .
however , with the selection function equation ( 13 ) , we have @xmath148 for plane parallel approximation , @xmath149 , the above equation is always hold if we study fluctuation powers on scales @xmath150 or @xmath151 .
actually , in this case , @xmath152 or @xmath153 depend only on @xmath75 , and thus their projections onto bases @xmath154 or @xmath155 ( @xmath156 ) are always null .
similarly the selection function in redshift space is also only dependent on @xmath75 .
accordingly , in the plane parallel approximation , we have @xmath157 which implies that the second term in the r.h.s . of equation ( 34 )
also has not contribution to the projection on bases @xmath158 or @xmath159 .
in fact , the condition of plane parallel approximation is not necessary . for radial redshift ,
the selection function is still a 1-d function .
its projection onto the bases @xmath160 on the celestial spherical surface is still zero .
thus , the linear redshift distortion mapping is given by @xmath161.\ ] ] hereafter we will use @xmath98 for the variable @xmath162 .
it will not cause confusion as the superscript @xmath163 stands for redshift space . in the plane
parallel approximation , equation ( 38 ) becomes @xmath164 using equation ( 19 ) , @xmath165 can be represented by @xmath69 , we have then @xmath166 \delta({\bf x}).\ ] ] the differential operator of the second term in the bracket of equation ( 40 ) is nearly diagonal in the dwt representation ( farge et al 1996 . )
thus , we have @xmath167 where @xmath168 means the value of @xmath169 in the mode @xmath170 .
the coefficient @xmath171 are defined by @xmath172 the calculations of @xmath171 are given in appendix a. since @xmath173 ( appendix a ) , the first and second terms on the r.h.s .
of equation ( 40 ) are not correlated , we have @xmath174 ^ 2 \langle |\tilde{\epsilon}_{\bf j , l}|^2 \rangle + \left [ \left .
\beta\frac{d\ln \bar{n}(x_3)}{dx_3}\right|_{\bf j , l}\right ] ^2 \sum_{l_3-l'_3 } q^2_{{\bf j } , l_3-l'_3 } \langle |\tilde{\epsilon}_{{\bf j},l_1,l_2,l'_3}|^2 \rangle\ ] ] for uniform fields , @xmath175 and @xmath176 are @xmath39-independent . hence , equation ( 43 )
gives the relation between the dwt power spectra in redshift @xmath177 and real spaces @xmath178 as @xmath179 ^ 2 + \left [ \left .
\beta\frac{d\ln \bar{n}(x_3)}{dx_3}\right|_{\bf j , l}\right ] ^2 \sum_{l_3-l'_3 } q^2_{{\bf j } , l_3-l'_3 } \right \}p_{\bf j}.\ ] ] which quantifies both redshift distortion and selection effect on the dwt power spectrum . using inequality equation ( a15 )
, we can show that if @xmath180 we have @xmath181 ^ 2 \sum_{l_3-l'_3 } q^2_{{\bf j } , l_3-l'_3 } < \beta^2 s^2_{\bf j}.\ ] ] which states that the selection function term in equation ( 44 ) is even less than the second order terms @xmath182 if the selection function varies with @xmath75 slowly . in this case
, the selection function does not significantly disturb the linear redshift distortion described by equation ( 31 ) when we perform a dwt power spectrum analysis using equations ( 10 ) and ( 15 ) . the poisson noise term in equation ( 10 ) is not affected by the redshift distortion , i.e. it is the same as the poisson noise term in equation ( 22 ) . because both poisson noise terms of equations ( 10 ) and ( 22 ) linearly depend on @xmath183 or @xmath184 , for an ensemble average , we have @xmath185 .
to demonstrate the redshift distortion in the dwt representation , we use the n - body simulation samples like that in paper ii .
the model parameters used are listed in table .
1 . we use modified ap@xmath186 m code ( couchman , 1991 ) to evolve @xmath187 cold dark matter particles in a periodic cube of side length @xmath10 .
the linear power spectrum is using the fitting formula given in bardeen et al .
( 1986 ) .
table 1 [ cols="^,^,^,^,^,^,<,^",options="header " , ] in our simulation , we use the so - called glass " configuration to generate the unperturbed uniform distribution of particles , and the zeldovich approximation to set up the initial perturbation . the triangular - shaped cloud ( tsc ) method is used for the mass assignment on the grid and the calculation of the force on a given particle from interpolation of the grid values .
we take 600 total integration steps from @xmath188 for the scdm model , and @xmath189 for @xmath190cdm and @xmath191cdm down to @xmath192 .
the force softening parameter @xmath193 in the comoving system decreases with time as @xmath194 .
its initial value is taken to be 15% , and the minimum value to be 5% of the grid size , respectively .
we try to estimate the redshift distortion parameter @xmath0 using equation ( 30 ) .
obviously , the precision of the @xmath0 determination is dependent on the linear relation of equation ( 18 ) , which is valid only on the scales where the bulk velocity can be described by linear or quasi - linear density perturbations .
it has been already realized that the non - linearity of the relation between @xmath24 and @xmath97 will be significant on small scales ( e.g. kudlicki et al .
therefore , it is necessary to have an estimation of uncertainties due to the non - linear @xmath195 relation .
the non - linear effect can be estimated by a blueshift " ( change the sign of velocity from @xmath102 to @xmath196 ) distorted power spectrum . from the derivation of equation ( 31 ) in last section , it is easy to show that , the blueshifted power spectrum is @xmath197 which has the same non - linear redshift distortion effect @xmath135 as the redshift power spectrum .
the difference between the redshift [ equation ( 31 ) ] and blueshift [ equation ( 47 ) ] distorted power spectra is a sign of the linear term @xmath0 . combining equations ( 31 ) and ( 47 )
, one can determine @xmath0 as @xmath198 we test equation ( 48 ) by simulation samples in the @xmath190cdm model .
first , we calculate the diagonal power spectra of particles , @xmath199 and @xmath200 , in the plane parallel approximation .
we then find @xmath0 by equation ( 48 ) .
the result is presented in fig .
the density parameter used for the simulation is @xmath201 , or @xmath202 .
2 shows that the values of @xmath0 given by equation ( 48 ) is generally overestimated .
but the overestimations are no more than 10% on scales @xmath203 h mpc@xmath80 , and about 20% for @xmath204 h mpc@xmath80 .
it should be pointed out that the goodness of the resulted @xmath0 on small scales @xmath204 h mpc@xmath80 does nt mean that the linear relation eq.(19 ) can be used to describe the redshift distortion on these scales . the non - linear effect on these scales
can not be ignored , and the redshift distortion is dominated by the term @xmath205 . the goodness for @xmath204 h mpc@xmath80 shown in fig .
2 is due to that the non - linear effect is significantly repressed by the test eq.(48 ) based on an assumed blueshifts . at the first glance
, it seems impossible to estimate @xmath0 using the observed redshift distorted power spectrum @xmath177 only , as all the quantities such as @xmath0 , @xmath178 , and @xmath135 ( or @xmath206 ) included on the r.h.s .
of equation ( 30 ) are unknown . usually , equation ( 30 )
is used to fit the observed power spectrum with model - predicted @xmath178 , and then determine the parameter @xmath0 from the most - likely - fitting .
but we try to search for @xmath0-estimators which depend on model as less as possible . to achieve this
, we take the advantage of the wavelet analysis : the dwt modes of @xmath207 are not rotational invariant , but cyclic permutational invariant . as a consequence , all the dwt quantities in the redshift distortion equation ( 30 ) , @xmath178 , @xmath177 , @xmath130 , and @xmath135 , are dependent on the three indexes @xmath208 , rather than the length scale of mode @xmath40 only . the quantities
@xmath178 , @xmath177 , @xmath130 and @xmath135 satisfy the following symmetry .
\1 . if cosmic density and velocity fields are statistically isotropic , the dwt power spectrum in real space is invariant with respect to the cyclic permutations of index @xmath209 , i.e. @xmath210 \2 . in the plane - parallel approximation , e.g. ,
coordinate @xmath75 is in the redshift direction , we have @xmath211 @xmath212 @xmath213 \3 . if @xmath206 is constant , i.e. scale - independent , following equation ( a10 ) , we have @xmath214 i.e. @xmath135 is independent of @xmath215 and @xmath216 .
more generally , equation ( 53 ) also hold even when the radial correlation of pairwise peculiar velocity is considered [ equation ( 33 ) ] .
equations ( 49 ) - ( 53 ) provide the base of designing the @xmath0-estimators with the dwt power spectrum . assuming that the pairwise velocity dispersion is scale - independent , equations ( 31 ) and ( 49 ) give @xmath217 ^ 2.\ ] ] for a given pair @xmath218 , @xmath219 is a constant , and thus the r.h.s . of equation ( 54 ) depends only on two parameters @xmath0 and @xmath219 .
accordingly , these parameters can be found by the best fitting of the r.h.s . of equation ( 54 ) with observed ratios @xmath220 , @xmath221 .
a numerical example of this fitting is demonstrated in fig .
3 , in which we take @xmath222 , and @xmath223 .
we calculate @xmath220 for the @xmath190cdm simulation samples , and the best fitting yields the values of @xmath224 and @xmath225 .
the precision of the estimator eq.(53 ) probably is not better than 20% .
this shows that the assumption of a constant @xmath206 is not too bad , but its effect on the @xmath0-determination can not be neglected .
if @xmath206 is scale - dependent , equation ( 54 ) is no longer correct in general . in this case @xmath226
depends on the three index @xmath208 , rather than @xmath89 only .
so the relation eq.(30 ) is obviously not enough to extract @xmath0 from measuring redshift distorted power spectrum @xmath177 only . to search for an appropriate algorithm for @xmath0 estimation
, we shall first consider the property of @xmath135 . at first , the pairwise velocity dispersion factor @xmath135 is determined by an isotropic function @xmath106 , which is scale - dependent .
let us consider the modes of @xmath227 , where @xmath228 is defined by @xmath229 . in this case ,
the scales of these modes are dominant by @xmath89 , and therefore , @xmath106 is also dominant by @xmath89 .
that is , for a given @xmath89 , @xmath135 will keep constant if @xmath230 . to test this expectation , we calculate @xmath135 in the @xmath190cdm model .
the result is shown in fig .
4 . one can see from fig .
4 that @xmath135 is generally dependent on indexes @xmath228 as well as @xmath89 .
however , for a given @xmath89 , @xmath135 almost keeps constant in the range of @xmath231 . since @xmath232 is the two largest scales of the samples , @xmath233 and @xmath234 keep constant very well for all @xmath235 . using this result
, we can design a @xmath0-estimator as follows @xmath236 or @xmath237 \frac{1}{(s_{j23}-s_{j32}+s_{j'32}-s_{j'23})}\ ] ] obviously , the first factor on the r.h.s . of equation ( 56 )
is to measure the difference between spectra with the same scales , but different shapes of the modes .
the estimator eq.(56 ) is model - free , as equation ( 56 ) is based only on property that two dwt modes of @xmath238 and @xmath239 have the same scale , but different shapes if @xmath231 and @xmath240 . to apply the estimator ( 56 )
, we take @xmath241 and @xmath242 , because modes with @xmath241 and @xmath242 have largest difference in the shape , but very small difference in the scale .
this estimator yields the values of @xmath243 for @xmath190cdm sample with simulation box @xmath244 480 h@xmath80 mpc ; @xmath245 for scdm sample with @xmath244 256 h@xmath80 mpc ; and @xmath246 for @xmath191cdm sample with @xmath244 256 h@xmath80 mpc . considering
the uncertainty of non - linearity is about 20% ( 5.1 ) , the estimator eq .
( 56 ) works very well .
after the @xmath0 estimation , we can also estimate @xmath206 on scale @xmath89 using off - diagonal power spectrum in redshift space . from
5.4 , we know that @xmath135 is almost constant in the range of @xmath231 .
so we have @xmath247 and @xmath248 .
we can calculate the ratio of pairwise velocity dispersion factor @xmath249 by @xmath250 on the largest scale @xmath251 , the nonlinear redshift distortion due to the pairwise velocity dispersion is negligible , we have @xmath252 .
thus , we have all the diagonal members of the pairwise velocity dispersion factor @xmath253 . the parameter @xmath206 on the scale @xmath254
can then be found by @xmath255 ^ 2 $ ] with equation ( a10 ) , i.e. @xmath256.\ ] ] for the @xmath190cdm simulation sample , @xmath253 is shown in fig.5 .
the @xmath206 on scales @xmath257 is shown in fig .
although the values of @xmath206 shown in fig .
6 are correct in average , but it does not match with the direct measurement of @xmath206 given in paper iii .
especially , on small scales @xmath258 6 and 7 , the values of @xmath206 are significantly lower than the direct measurement .
this is not unexpected .
the factor @xmath131 equation ( 58 ) is obtained under the assumption of gaussian velocity field ( 3.1 ) .
however , we have shown in paper iii that the velocity field is actually intermittent on small scales .
the pdf of @xmath259 is not gaussian , but lognormal . nevertheless , the factor @xmath135 is still good for the redshift - real mapping and @xmath0-estimation , because in these calculations , we used only the symmetric properties and scale - dependence of @xmath135 , but not the details of @xmath206 .
this point can also be seen from reconstruction of power spectrum in real space from that in redshift space . using the @xmath0 estimated by equation ( 56 ) , and @xmath253 by equation ( 57 )
, one can reconstruct the diagonal dwt power spectrum in real space @xmath85 from @xmath199 through equation ( 31 ) .
fig.(7 ) compares the recovery of @xmath85 with the original diagonal dwt power spectrum in the @xmath190cdm model .
the result shows that the algorithm of reconstruction is reliable .
we established the mapping between the dwt power spectra in real and redshift spaces . from equations ( 31 ) and ( 44 ) , the mapping in the plane parallel approximation is @xmath179 ^ 2 + \left [ \left .
\beta\frac{d\ln \bar{n}(x_3)}{dx_3}\right|_{\bf j , l}\right ] ^2 \sum_{l_3-l'_3 } q^2_{{\bf j } , l_3-l'_3 } \right \}s^{pv}_{\bf j } p_{\bf j}.\ ] ] which includes the effects of ( 1 ) bulk velocity ( the term of @xmath130 ) , and ( 2 ) selection function ( the term of @xmath260 ) as well as ( 3 ) pairwise peculiar velocity ( the term of @xmath135 ) . in the fourier representation , the redshift distortion mapping is axially symmetric with respect to the redshift direction .
this is because the fourier mode is rotational invariant , and the redshift distortion produces an anisotropy in the spatial distribution of galaxies between the line of sight and directions other than it .
an isotropic statistics plus an axially symmetric violation will result in an axial symmetry . on the other hand , all redshift distortion factors in equation ( 59 ) are no longer axial symmetric , which is due to the modes @xmath40 being not invariant under rotational transformation , but for cyclical permutation . therefore , the symmetry of the dwt redshift distortion results from a cyclical permutation plus an axially symmetric violation . by virtue of this feature
, we develop @xmath0 estimators , which are mainly based on the shape - dependence of the redshift distorted dwt power spectrum , but not the scale - dependence .
the estimators eqs.(54 ) and ( 56 ) look similar to the quadrupole - to - monopole ratio method based on the fourier power spectrum in redshift space , which gives a model - free estimation of @xmath0 ( e.g. cole et al , 1994 , fisher et al , 1994 , hatton & cole , 1998 , szalay et al , 1998 ) .
however , the estimators eqs.(54 ) and ( 56 ) is different from the quadrupole - to - monopole ratio method .
the latter considered only the linear redshift distortion effect , but not the non - linear redshift distortion effect .
if considering the non - linear redshift distortion effect , the quadrupole - to - monopole ratio method needs a fitting of observed redshift distorted power spectrum with model - predicted power spectrum . on the other hand , the estimators eqs.(54 ) and ( 56 )
considered both linear ( @xmath261 ) and non - linear ( @xmath135 ) redshift distortion effects .
these @xmath0-estimators are model - free even when the non - linear redshift distortion effect is not negligible .
we test the @xmath0-estimators using n - body simulation samples .
the result shows that regardless the pairwise velocity dispersion is scale - dependent or not , the @xmath0-estimators can yield the correct number of @xmath0 with about 20% uncertainty .
we also develop an algorithm for reconstruction of the power spectrum in real space from the redshift distorted power spectrum .
the numerical tests also show that the real power spectrum can be reasonably recovered from the redshift distorted power spectrum .
llf and yqc acknowledges support from the national science foundation of china ( nsfc ) and national key basic research science foundation .
let us consider plane - parallel approximation , i.e. coordinate @xmath75 is in the redshift ( @xmath263)-direction .
the linear redshift distortion @xmath130 equation ( 28 ) gives @xmath264 because 1-d wavelet @xmath43 is given by dilating and translating the basic wavelet @xmath265 as @xmath266 if @xmath281 is independent of @xmath98 , @xmath282 [ equation ( 28 ) ] can be calculated by @xmath283,\ ] ] where vector @xmath284 . in plane - parallel approximation
, we have @xmath285.\ ] ] the summation of equations .
( a9 ) and ( a10 ) also runs only over few number of @xmath275 around @xmath276 . \\
\nonumber & & \simeq \frac{l_3}{2\pi n_p2^{j_3 } } \frac{1}{2^{j_1+j_2+j_3 } } \\
\nonumber & & \sum_{n_1,n_2,n_3 = \infty}^{\infty } \frac{(n_3/l_3)^2\sin[2\pi n_3(l_3-l_3')/2^{j_3 } ] } { [ ( n_1/l_1)^2 + ( n_2/l_2)^2 + ( n_3/l_3)^2 ] } | in this paper , we develop a theory of redshift distortion of the galaxy power spectrum in the discrete wavelet transform ( dwt ) representation . because the dwt power spectrum is dependent of both the scale and shape ( configuration ) of the decomposition modes , it is sensitive to distortion of shape of the field . on the other hand ,
the redshift distortion causes a shape distortion of distributions in real space with respect to redshift space .
therefore , the shape - dependent dwt power spectrum is useful to detect the effect of redshift distortion .
we first established the mapping between the dwt power spectra in redshift and real space .
the mapping depends on the redshift distortion effects of ( 1 ) bulk velocity , ( 2 ) selection function and ( 3 ) pairwise peculiar velocity .
we then proposed @xmath0-estimators using the dwt off - diagonal power spectra .
these @xmath0-estimators are model - free even when the non - linear redshift distortion effect is not negligible . moreover
, these estimators do not rely on the assumption of whether the pairwise velocity dispersion being scale - dependent .
the tests with n - body simulation samples show that the proposed @xmath0-estimators can yield reliable measurements of @xmath0 with about 20% uncertainty for all popular dark matter models .
we also develop an algorithm for reconstruction of the power spectrum in real space from the redshift distorted power spectrum .
the numerical test also shows that the real power spectrum can be well recovered from the redshift distorted power spectrum . |
the local group ( lg ) is composed of two massive spiral galaxies , the milky way ( mw ) and m31 , and a smaller one , m33 .
in addition to these normal galaxies , more and more dwarf galaxies are discovered ( see * ? ? ? * for an updated review of the lg dwarf galaxies ) .
the new discoveries came as a result of the sdss @xcite , the pandas survey @xcite , the pan - starrs1 @xmath1 survey @xcite and most recently the dark energy survey @xcite .
most of these objects are dwarf spheroidal ( dsph ) galaxies : spheroidal in shape , devoid of gas and current star formation and typically located close to the mw or m31 ( @xmath2 kpc ) . except for a few ellipticals , the remaining lg dwarf galaxies are dwarf irregulars ( dirr ) : gas - rich , with high star formation rates and located rather far from the mw and m31 ( @xmath3 kpc ) , with a notable exception of the magellanic clouds . in the context of the @xmath4cdm paradigm ,
dwarf galaxies are understood as remnants of primordial galaxies , full of dark matter , which have not yet merged with other objects to build massive galaxies @xcite .
indeed , the lg dirrs have gas fractions larger than 50% @xcite similarly to the high-@xmath5 galaxies @xcite . in the case of dsphs , their properties ( the lack of gas and the spheroidal shape )
are explained by a long - term interaction with their host galaxies due to tidal forces , dynamical friction and ram pressure stripping @xcite . in this study , we investigate specifically the origin of the andromeda ii ( and ii ) dsph galaxy .
this dsph discovered in 1972 @xcite lies in the vicinity of m31 at a distance of 184 kpc @xcite .
it seems to be a typical , relatively luminous dsph devoid of gas , without star formation for the last several gyr ( * ? ? ?
* ; * ? ? ?
* hidalgo et al . in preparation ) , with a low central surface brightness of 24.8 mag arcsec@xmath6 and a quasi - spheroidal shape with ellipticity close to 0.2 ( * ? ? ?
* del pino et al . in preparation ) .
however , when studied in more detail , and ii reveals a complicated star formation history ( sfh ) with two star formation episodes ( hidalgo et al . in preparation ) which generated two stellar populations : one more extended containing primordial stars ( @xmath7 11 gyr old ) and one more spatially concentrated composed of intermediate - age stars ( 5 - 9 gyr old ) ( * ? ? ?
* del pino et al . in preparation ) .
in addition , @xcite measured a clear rotation signal in and ii , which is unusual for a dsph in the lg .
moreover , this rotation is around the major axis of the dwarf and not the minor one as commonly observed in spiral galaxies .
@xcite reanalyzed the same kinematic data and claimed detection of substructure that could form a stellar stream .
they speculated that such substructure could result from a merger between and ii and a smaller object . until recently
, mergers between dwarf galaxies in environments such as lg were believed to be rare events due to large relative velocities of the possible progenitors .
however , investigations of subhalo statistics in simulated lg - like environments revealed @xcite that a non - negligible part ( 10 - 30% ) of present - day dwarf galaxies could have undergone a major merger mainly at high redshift when they were not yet bound to their hosts and when the galaxy density in the universe was higher . these mergers turned out to take place between galaxies accreted by the host in pairs , which explains their low relative velocities .
@xcite resimulated some of such pair - like configurations , populating the subhaloes with stellar disks and demonstrated that such mergers can lead to the formation of dsph - like objects .
some dwarf galaxies of the lg could therefore be remnants of major mergers which have occurred in the field a few gyr ago .
in a recent study @xcite investigated major mergers between gas - rich dwarf galaxies in a constrained simulation of the lg .
they showed that products of such mergers would indeed have two stellar populations with different spatial distributions similar to the ones seen in and ii .
@xcite proposed the first detailed model for the origin of and ii as the result of a major merger of two disky dwarf galaxies of equal mass .
this study used collisionless @xmath0-body simulations with the progenitor galaxies composed of a stellar disk and a dark matter halo .
the different spatial distribution of the two stellar populations was explained by the different size of the stellar disk in the two progenitors .
the unusual prolate rotation was explained as the result of a head - on collision with a particular orientation of the progenitor disks .
@xcite generalized this result by considering other initial configurations in terms of the orbit and inclination of the disks .
they demonstrated that prolate rotation can result from a variety of initial conditions and does not require fine tuning thus adding credibility to the model .
they have also shown that it is practically impossible to produce such type of rotation via tidal stirring of initially disky dwarfs orbiting a mw - like host ( e.g. * ? ? ?
although tidal stirring can be very efficient in replacing rotation by random motions of the stars , it can not induce rotation around the major axis and whatever rotation remains in the dwarf is always around the minor axis .
these studies did not however include gas dynamics and star formation so could not address the origin of stellar populations of different age . if the major merger occurred at high redshift in the field , it is highly probable that the two dwarf galaxies were gas - rich . in this work
we therefore extend the model proposed by @xcite by adding gas and star formation to the picture .
we present a complete model for and ii since the beginning of the gas - rich major merger ( 10 gyr ago ) till the present day . in order to be complete
, this study also takes into account the interaction between m31 and the remnant of the gas - rich major merger in order to strip the gas .
in sections 2 and 3 , we describe the gas - rich major merger : the initial conditions , the numerical configuration and the final remnant .
then in sections 4 and 5 , we investigate the interaction with m31 taking into account the possible orbit of and ii around m31 and the effects of the tidal force and the ram pressure stripping .
finally , sections 6 and 7 present the discussion of our results and conclusions from this work .
in our scenario , and ii is the result of a major merger which occurred at high redshift ( @xmath8 ) in the field between two dwarf galaxies . due to the lack of dwarf galaxy observations at high redshifts , it is uncertain what the progenitor galaxies of and ii could look like .
we therefore have to make some assumptions to define the main properties of and ii progenitors using what is thought reliable concerning high redshift dwarf galaxies .
first , we assume that the two progenitors are identical .
this choice leads to a 1:1 merger and decreases the number of parameters by a factor of two ( see table [ tab : prop_prog ] ) .
it also seems essential to reproduce the and ii kinematics following @xcite and @xcite .
consequently , we did not try to change the mass ratio .
however , one may argue that a major merger is mandatory in order to produce a dsph galaxy from two disky ones .
a minor merger ( with a mass ratio below @xmath9 ) would only perturb the biggest galaxy without significantly affecting its morphology and kinematics ( see e.g. * ? ? ?
second , as the merger occurs at high redshift in the field , we expect the two dwarf galaxies to be dirrs , thus they are gas - rich and disky . in this section ,
we present and explain the initial conditions of the gas - rich major merger .
the final set of parameters used in our simulation was selected in order to provide best match to the observations and was determined as a result of many numerical experiments we performed with different combinations of their values .
we assumed the dwarf galaxy progenitors have a stellar disk with an exponential density profile defined by three most important parameters : the mass , the radial extension and the thickness . and
ii has a @xmath10-band absolute magnitude of @xmath11 mag @xcite . by assuming a mass - to - light ratio of 1 m@xmath12/l@xmath12 as in @xcite , the total stellar mass is estimated at @xmath13 m@xmath12
however , as and ii s stellar population is mainly composed of primordial and intermediate - age stars ( * ? ? ?
* hidalgo et al . in preparation ) , the present mass - to - light ratio could be larger than 1 .
this implies that the total stellar mass of the two progenitors should be larger than @xmath13 m@xmath12 . in our simulation , we have chosen the stellar mass per progenitor of @xmath14 m@xmath12 , so the total stellar mass of @xmath15 m@xmath12 .
different test simulations made clear that this large initial mass ( about twice the final one ) is needed because some of the stars are ejected from the remnant or are too faint to be visible .
the second important parameter is the radial size of the disk . as the progenitor disks have an exponential shape , it is defined by the scale radius , @xmath16 , which should be between 0.4 and 1 kpc .
values smaller than 0.4 would lead to the formation of a remnant which is too dense , larger ones would make it too extended .
we have tested different scale radii between 0.4 and 1.0 kpc .
finally we have chosen a scale radius of 0.8 kpc .
this parameter , as expected , turned out to be the main parameter that needed to be adjusted in order to accurately reproduce the slope of the stellar density profile .
the adopted thickness of the disk , defined by @xmath17 , is a typical value for dwarf galaxies .
we did not vary its value in the test simulations , as it seems to have little effect on the final outcome of the merger . ) with an assumed mass - to - light ratio of 1 m@xmath12/l@xmath12 . for the gas , colours code the column density of the gas ( in @xmath18 atoms cm@xmath6 ) .
top panels show the two components viewed edge - on while in the bottom panels they are viewed face - on . ]
the dirrs in the lg @xcite or in surveys like little things @xcite have high gas fractions which can reach more than 80% of the baryonic mass . at high redshift
, this value must have been statistically larger because less gas was consumed by star formation .
observations of and ii do not provide any hint to derive the gas fraction in the past because it seems to have no more gas at present @xcite . the gas fraction was chosen to be 70% of the baryonic mass . with the stellar mass of @xmath14 m@xmath12
, this leads to the gas mass of @xmath20 m@xmath12 and the total baryonic mass of @xmath21 m@xmath12 . in our test simulations , we have also tried other gas fractions from 60% to 80% . as expected , an increase ( decrease ) of the gas fraction
increases ( decreases ) the star formation .
nevertheless , the star formation also depends on feedback , therefore we have decided to keep the gas fraction constant at 70% and to fine tune the feedback .
this obviously shows that the problem is degenerated as two different sets of parameters can provide the same result .
this is not surprising due to the large number of parameters and the limited and uncertain observations .
we also always assumed that the gas disk has the same extension as the stellar disk ( see figure [ fig : ic_dwarf ] ) .
again , the gas extension has a direct effect on the star formation ; increasing it will decrease the star formation because the gas density will decrease . on the other hand ,
the thickness of the gaseous disk is not a free parameter because it depends on the balance between the pressure force and the gravitational force in this component .
consequently , the gas disk we obtain is about a factor of 1.5 thicker than the stellar disk ( see figure [ fig : ic_dwarf ] ) .
the stellar and gas disks are embedded in a dark matter halo .
different works ( * ? ? ?
* and reference therein ) tend to show that the dark mass - baryonic mass ratio can be very large , even larger than a hundred for faint dwarf galaxies like sextans . we have chosen the ratio of the dark to baryonic mass to be equal to 33 .
this choice is mainly motivated by the need to reproduce the level of velocity and velocity dispersion in the and ii kinematics .
as the initial baryonic mass in the simulation is constrained by the actual and ii s stellar mass , a larger ( smaller ) mass - to - light ratio would imply a larger ( smaller ) dark matter mass and thus larger ( smaller ) kinematics . in our test simulations we tried mass - to - light ratios between 20 and 50 .
the density profile of dark matter is assumed to be similar to the navarro - frenk - white ( nfw ) profile @xcite with a cut - off at the virial radius and the total mass equal to @xmath22 m@xmath12@xmath23 but with a shallower inner slope of @xmath24 instead of @xmath25 .
this choice was made to better match the inner kinematic profile . with a pure nfw profile ,
dark matter mass is too concentrated and the inner slope of the rotational velocity is too steep .
we also note that this shallower slope is consistent with recent findings concerning the effect of baryons on dark matter distribution @xcite . in our case
this flattening of the dark matter slope would have been caused by the first generation of stars prior to the epoch we attempt to model .
llr + components & properties & values + dark matter & total mass & @xmath26 m@xmath12 + & inner slope & @xmath24 + & outer slope & @xmath27 + baryons & baryonic mass & @xmath21 m@xmath12 + & baryon fraction & 0.03 + stars & stellar mass & @xmath14 m@xmath12 + & disk scale - length @xmath16 & 0.8 kpc + & disk thickness & 0.2 kpc + gas & gas mass & @xmath20 m@xmath12 + & gas fraction & 70% + & disk scale - length @xmath16 & 0.8
kpc + [ tab : prop_prog ] to create the initial conditions for the progenitors , we used the method described in @xcite and @xcite .
it allows to create stable numerical realizations of galaxies in near - equilibrium , composed of a halo and a stellar disk .
this method produces objects closer to equilibrium and therefore gives us better control over the structural parameters compared to the one described in @xcite because it is based on the distribution functions , not just their moments .
however , this method does not allow to define the initial conditions for the gas . in order to include the gas component we therefore split the baryonic disk particles randomly into two populations : gas particles and stellar particles matching the assumed gas fraction ( 70% ) .
we end up with a dwarf galaxy with properties given in table [ tab : prop_prog ] but unstable due to the gas .
k. ] in order to avoid gas instabilities due to the gas pressure , we let the dwarf galaxy evolve during 10 gyr by setting the initial gas temperature to @xmath28 k. for this simulation we used the publicly available version of the gadget2 @xmath0-body / sph code @xcite . during this simulation , there is no star formation because our purpose here is just to stabilize the gas component rather than to investigate the evolution of the galaxy in isolation . after 10 gyr ,
the dwarf galaxy is stable again . during this evolution
the stellar disk becomes thicker ( by less than @xmath29 ) but has the same exponential shape and the same kinematics .
the gas is almost 1.5 times thicker than the stars , its radial extension remains similar to the stars , and the gas loss is small ( @xmath30 10% ) .
stars are all conserved so the new gas fraction is 67% , close to the previous one , 70% .
the evolution also results in a slight overdensity of gas close to the centre ( see figure [ fig : denspro_gas ] ) .
however , there is no need to get a perfect exponential shape for the gas because when the star formation with feedback is turned on , it affects the gas density profile , increases the gas bump at the centre and keeps the maximal temperature at @xmath31 k due to the cooling ( see figure [ fig : temp_dens_gas ] ) .
a much lower initial temperature , for example 20 k instead of 8000 k , allows us to reduce the pressure force leading to less gas escaping in the external parts .
however , it creates a much larger gas overdensity in the centre .
the rotating gas kinematics is similar to that of the stars although the rotation velocity is smaller due to the initial instability . however , gas rotation is not important in our study because the gas is stripped in the end .
the small fraction that turns into stars does not significantly affect the overall kinematics of the final remnant .
the velocity dispersion of the gas is lower than for the stars , almost constant with radius with a slight drop at the centre .
this property is due to the gas pressure which does not need velocity dispersion to keep the gas stable ( see figure [ fig : kine_gas ] ) . and a relative one about 30% ) .
the stellar velocity dispersion has a maximum at the centre ( @xmath32 km s@xmath33 ) to support the stellar component . on the contrary , the gas velocity dispersion is smaller with a drop at the centre because it is the pressure force and not the velocity dispersion that stabilizes the gas component.,title="fig : " ] + lrrr + properties & dark matter & stars & gas + number of particles & 200 000 & 60 000 & 140 000 + particle mass ( m@xmath12 ) & 5114.6 & 149 & 149 + softening length ( pc ) & 60 & 20 & 20 + [ tab : param_simu ] a major merger between galaxies can result from a variety of orbital initial conditions .
@xcite studied the statistics of orbits of merging massive haloes in a cosmological simulation and found that an eccentricity around one , corresponding to a parabolic orbit , is the most likely .
unfortunately , no such statistics is available for smaller mass haloes or dwarf galaxies in lg - like environments .
however , @xcite demonstrated that for a major merger with a hyperbolic orbit and a small pericentre after the first passage the orbital angular momentum is lost and the trajectory evolves from a hyperbola to a highly eccentric ellipse for the second passage , which is similar to an almost radial orbit .
a similar result is expected with a parabola which is a limiting case , @xmath34 , of a hyperbola .
in our model there are also two passages
. the first one could be on a parabolic trajectory and a close pericentre which would lead to a nearly head - on second passage . in order to simplify the initial conditions ,
we start with a head - on collision .
therefore at @xmath35 , the two dwarf galaxies are separated by 50 kpc and they are approaching with a velocity of 8 km s@xmath33 along the @xmath36-axis of the simulation box , which corresponds to the initial relative velocity of 16 km s@xmath33 ( see figure [ fig : ic_orbit ] ) .
plane of the simulation box .
the dwarf images include both stars and gas .
the red arrows indicate the direction of the disk angular momenta which lie in the @xmath37 plane .
the velocities and the centres of the dwarf galaxies are aligned with the @xmath36-axis . ]
the disks of the progenitors are rotated so that the disk angular momenta have the same component along the @xmath36-axis .
this implies that @xmath38 , @xmath39 and @xmath40 .
this leaves us only the freedom to choose the inclination of the angular momentum between 0 and 90 deg for the first galaxy ( the left one in figure [ fig : ic_orbit ] ) . with an angle of 0 deg , the total angular momentum is along the positive @xmath36 . with 90 deg ,
the angular momentum is perpendicular to the @xmath36-axis .
a value of 0 deg for the first progenitor implies 0 deg for the second too .
the dependence of the remnant properties on this parameter has been also studied by @xcite . following this study
, we can expect that small inclination angles would provide a large angular momentum for the remnant galaxy but not a spheroidal shape . on the contrary
, a choice of a high inclination angle ( close to 90 degrees ) would produce a galaxy without significant prolate rotation .
we have therefore chosen an angle of 45 deg , as in @xcite , for the first galaxy , and @xmath41 deg = ( @xmath42 deg ) for the second one to match observations ( see figure [ fig : ic_orbit ] ) .
for the simulations described below we used a modified version of the gadget2 @xmath0-body / sph code @xcite with star formation , feedback and cooling processes added as described in @xcite and @xcite following recipes defined in @xcite .
different parameters are available to tune the star formation .
we choose to mainly follow @xcite , i.e. we adopt the star formation efficiency of 0.03 , the feedback index of 2 and the density threshold for star formation of 0.0171 m@xmath12pc@xmath43 .
our assumptions differ only in one feedback parameter , @xmath44 , determining the time - scale of feedback thermalization . when @xmath44 is large the energy due to the star formation is quickly released and vice versa .
this parameter is tuned in order to get the star formation rate which fits the observed sfh in and ii .
the value in the presented simulation is 5.5 myr which equals to 0.7 times the normal feedback timescale defined by @xcite .
the numbers of particles , the softening lengths and the masses of the particles are provided in table [ tab : param_simu ] .
plane of the simulation box as in figure [ fig : ic_orbit ] . the rows of panels from top to bottom show , respectively ,
the initial conditions 12 gyr ago , the first passage 10 gyr ago , the apocentre 9.5 gyr ago and the end of the star formation 5 gyr ago . times and scales are indicated for each phase .
the scales are the same for stars and gas but change with the phase .
the colours code the stellar and gas mass density in logarithmic scale .
the colour coding is different for each phase in order to clearly show the density gradient . ]
the simulation starts with two separated dwarf galaxies .
the stars inside these galaxies were formed during the first gyrs of the evolution of the universe .
we did not try to reproduce this initial star formation period due to its complexity which must have involved gas accretion , formation of primordial stars , cosmological evolution , etc .
we only know that stars older than 11 gyr represent about 70% of all stars as seen in observations .
we assume that the simulated sfh matches the observed sfh during this early period .
our simulation becomes consistent with the data around 11 gyr ago .
between 11 and 10 gyr ago , there is no star formation in the dwarf galaxies , which are then approaching each other , because the gas is not dense enough ( see figure [ fig : temp_dens_gas ] ) .
ten gyr ago , the two progenitors collide for the first time ( the first passage ) . due to the high relative velocity of the two dwarf galaxies ( @xmath45 60 km s@xmath33 ) ,
the collision is quick and the stellar disks are not completely destroyed .
they are warped and tidal tails form for each dwarf galaxy ( see figure [ fig : ic_merger ] ) .
the situation is different for the gas .
in addition to the gravitational force , the gas dynamics is also influenced by pressure forces .
consequently , the gas does not behave collisionlessly .
when the dwarf galaxies collide , the gas is compressed and its density suddenly increases .
this results in the first peak of star formation at this time .
however , this star formation episode lasts only for a very short time , less than 0.1 gyr , because the dwarf galaxies move quickly away from each other reducing the gas density .
thus , this peak of star formation rate produces very few new stars and is barely visible in the sfh ( see figure [ fig : res_majmer_sfh ] ) . between 10 gyr and 8.7 gyr ago , the dwarf galaxies are separated again reaching their apocentre 9.5 gyr ago ( see figure [ fig : ic_merger ] ) . afterwards , they fall towards each other for the second passage 9 gyr ago .
it is quickly followed by a complete fusion of the galaxies which happens 0.3 gyr later , 8.7 gyr ago .
after that , the centres of the dwarf galaxies are indistinguishable ( see figure [ fig : ic_merger ] ) .
the fusion has a clear effect on the sfh with a strong and constant enhancement of star formation ( see figure [ fig : res_majmer_sfh ] ) .
the sfh slope becomes steep because the star formation rate is now steady ( @xmath46 m@xmath12yr@xmath33 ) .
we assume that the star formation stops @xmath45 5 gyr ago as required by the observations ( see figure [ fig : res_majmer_sfh ] ) .
the reason for this stop is assumed to be the interaction with m31 discussed in sections 4 and 5 of this paper .
nearly 5 gyr ago , when the star formation seems to end , we stop the simulation and make an image of the stellar component which we then use to measure the luminosity and ellipticity of the remnant dwarf galaxy . for that , we have chosen the line of sight to be the @xmath5-axis of the simulation box so the positions of the stellar particles are projected onto the @xmath47 plane .
in fact , all lines of sight perpendicular to the @xmath36-axis are nearly equivalent because the dwarf remnant is symmetrical with respect to the @xmath36-axis due to the initial conditions ( see figure [ fig : ic_merger ] ) .
the limiting surface brightness is set to 28 mag arcsec@xmath6 .
the visible mass is converted into the @xmath10-band luminosity and surface brightness , by assuming a constant mass - to - light ratio ( see figure [ fig : res_majmer_morpho ] ) .
the value of the mass - to - light ratio is chosen so that the simulated @xmath10-band luminosity equals to the observed one : @xmath48 l@xmath12 .
we have measured the ellipticity of the simulated dwarf galaxy using the method described in @xcite by fitting iso - density contours by ellipses ( see figure [ fig : res_majmer_morpho ] ) .
the measured stellar mass for the simulated and ii is @xmath49 m@xmath12@xmath23 with a mass - to - light ratio of 1.25 m@xmath12/@xmath50 . concerning the ellipticity , we have found @xmath51 at the half - light radius of 0.8 kpc , in the @xmath47 plane ( see figure [ fig : res_majmer_morpho ] ) , which is within @xmath52 of the observed value @xmath53 @xcite . plane 5 gyr ago obtained with a cut at 28 mag arcsec@xmath6 .
the black lines show the isophotes .
the colour bar codes the surface brightness scale in mag arcsec@xmath6 .
the position angles of the ellipses vary a little with radius due to perturbations resulting from star formation . ]
we have computed the stellar density profiles of the two stellar populations present in and ii .
we used the sfh of the galaxy ( hidalgo et al .
, in preparation ) to assign stellar ages to the objects in the photometric catalog obtained by @xcite
. the procedure is explained in detail in @xcite and del pino et al .
( in preparation ) .
a short summary is given here . a synthetic color - magnitude diagram ( cmd )
is computed based on the sfh of and ii . using an adaptive grid , we sample both cmds , the synthetic and the observed one , assigning ages and metallicities to the observed stars on the basis of their synthetic counterpart .
after the sampling , two distinct stellar populations were recovered : stars older than 11 gyr and stars with ages between 9 and 5 gyr .
we expect this method to provide an improved age resolution compared to the one used in @xcite .
nevertheless , even using the sfh we are not able to completely break the age - metallicity degeneracy present in the cmd of and ii .
it is therefore plausible that a non - negligible mixing between both stellar populations is still present .
this seems to be the case when comparing both radial profiles at large radii ( see figure [ fig : res_majmer_denspro ] ) .
after the age determination , the radial density profiles of the two stellar populations were obtained by fitting 38 ellipses to the isopleths of their surface density maps .
we count the stars lying within each of these regions , adopting as a galactocentric radius the average of the semi - major axes of the two boundary ellipses delimiting a given region .
errors in the galactocentric distance come from the ellipse fitting to the isopleths , while we assume poissonian error for counting stars .
although the derived total density profile is similar to @xcite , the old and intermediate stellar density profiles are different .
@xcite found a density profile flat in the center with a sersic index of 0.3 for the old stellar population .
we have found an exponential density profile for the old stars . concerning the intermediate - age stars
, @xcite found an exponential profile which dominates the old stars in the centre of the dwarf ( @xmath54 pc ) .
consequently , the intermediate - age stars represent more than 70% of the luminosity in the centre , therefore around 60 - 70% of the mass if we take into account that their @xmath55 ratio is larger than the one of the old stars .
this is in contradiction with the fact that with the new sfh determination it is now known that only 30% of the stars have an intermediate age in the center of and ii ( @xmath54 pc ) .
we have arbitrarily rescaled the observed density profile for which the unit is star counts and not surface brightness due to the fact that the mass - to - light ratios are not well constrained .
for the primordial stars , observations show a part of the curve ( @xmath56 kpc ) which can be fitted by an exponential density profile , a straight line , with a slope of 1.52 mag arcsec@xmath6 kpc@xmath33 ( dashed lines in figure [ fig : res_majmer_denspro ] ) . beyond 1.8 kpc ,
the density drops very quickly .
this decrease can be due to the cut in the stellar profile ( a physical explanation ) or the lack of data ( an instrumental explanation ) .
for the simulation , the density profile of old stars also follows a straight line ( from 0 to 1.8 kpc ) with a similar slope of 1.68 mag arcsec@xmath6 kpc@xmath33 .
the difference with respect to the observations is about 10% . for the intermediate - age stars the errors are larger but
it seems that a straight line can be a good approximation of the density profile especially at radii between 0.5 and 1.8 kpc .
in fact , this portion of the intermediate - age stars can be well fitted by the same line as for the old stars with a smaller scale , @xmath57 mag arcsec@xmath6 and shows the same density drop after 1.8 kpc .
this confirms the expectations from del pino et al .
( in preparation ) : at radii larger than @xmath45 0.5 kpc there is no more intermediate - age stars , this stellar population is contaminated by the old stars . on the contrary ,
within @xmath45 0.5 kpc the density profile is different from the one for old stars with a bump in density which is well reproduced by the simulation ( see figure [ fig : res_majmer_denspro ] ) .
+ in the initial conditions , the two progenitors have angular momenta of the same magnitude ( @xmath58 m@xmath12kpc km s@xmath33 ) but with different directions .
the total angular momentum along the @xmath36-axis is equal to @xmath59 m@xmath12kpc km s@xmath33 . after the major merger , 5 gyr ago , the angular momentum of the stars points towards the @xmath36-axis as expected from the conservation of this quantity along the symmetry axis : its direction is almost exactly along the @xmath36-axis and its magnitude is @xmath60 in the same units , very close to the expected value ( the difference being only 3% ) .
figure [ fig : res_majmer_kine ] presents the mean velocity and velocity dispersion of the stars in the remnant galaxy 5 gyr ago within a projected radius of 4 kpc using bins of 0.4 kpc in the @xmath47 plane .
this map shows clear rotation signal of stars around the @xmath36-axis as expected with this value of the angular momentum .
8 kpc with a resolution of 0.4 kpc in the @xmath47 plane .
the colour coding of the values is indicated on the right hand side of each picture .
the top panel shows the map of the mean velocity of the stars while the bottom panel presents the map of the stellar velocity dispersion . ]
we have also calculated the mean velocity and velocity dispersion within 2 kpc along the minor and major axis in order to mimic the observations presented in @xcite .
figure [ fig : res_majmer_kine_2 ] shows that the simulated mean velocity along the minor and major axis reproduces within 1 - 1.5 @xmath61 error bars the observed values except for one point .
concerning the velocity dispersion , the simulation agrees with the observations within @xmath62 except for three data points among 22 for the minor and major axis and agrees within @xmath52 for 11 points out of 22 . at the beginning of the simulation ,
the gas comprises @xmath63% of the baryonic mass ( baryonic mass = @xmath64 m@xmath12 , gas mass = @xmath65 m@xmath12 , stellar mass = @xmath15 m@xmath12 ) .
after the fusion , the gas starts to be turned into stars . until 5 gyr ago ,
m@xmath12 of the gas has been transformed into stars .
in addition some of the gas and stars were ejected due to the major merger .
the visible stellar mass ( above the surface brightness of 28 mag arcsec@xmath6 ) is roughly @xmath49 m@xmath12@xmath23 and the one of the gas is @xmath21 m@xmath12 corresponding to column density of @xmath18 to @xmath67 atoms cm@xmath6 .
the gas fraction represents 71% of the baryonic mass , i.e. it is approximately the same as before the merger .
the gas column densities and the gas total mass are large enough to be detected by radio telescopes as was the case for other dwarf galaxies in the lg with hi masses below @xmath68 m@xmath12@xmath23 @xcite .
and ii seems to contain no gas like the other mw dsph galaxies @xcite .
consequently , our scenario so far missed the physical processes which would allow us to remove all the gas . in the next section ,
we investigate the interaction between m31 and the merger remnant as a possible mechanism to remove the gas and stop star formation .
and ii is likely a satellite of m31 . in the previous sections
, we did not take into account the possible interaction between and ii and m31 and its consequences . as we assume
the gas - rich major merger occurs in the field far from m31 ( 9 - 10 gyr ago ) , the interaction with m31 has to occur later . we assume it occurs 5 - 7 gyr ago to be consistent with the star formation history which ends 5 gyr ago . in this section ,
we use the snapshot which corresponds to the and ii state 5 gyr ago when the star formation ends .
= 7.63 kpc , @xmath69 m@xmath12 , @xmath70 , @xmath71 kpc ) and the red line from @xcite ( nfw profile with : @xmath72 kpc , @xmath73 m@xmath12pc@xmath43 , @xmath74 m@xmath12 ) .
the vertical dashed lines indicate the radial range between 30 and 300 kpc where and ii is assumed to orbit .
middle panel : an approximation for the gravitational force from the m31 components : @xmath75 .
bottom panel : an approximation for the tidal force generated by the same components : @xmath76 .
the dashed curve in the middle and bottom panel corresponds to the total mass ( baryons + dark matter ) . ]
m31 is assumed to be composed of a bulge , a disk and a dark matter halo .
the dark matter halo is a stable spherical collisionless system for which velocity dispersion balances gravitational attraction .
its mass is set at @xmath77 m@xmath12@xmath23 in agreement with the results from @xcite .
its shape is assumed to be given by a hernquist profile with a scale radius of @xmath78 kpc .
these parameters are chosen to be consistent with other studies which have investigated the density profile of m31 between 30 - 300 kpc where and ii is assumed to orbit ( see figure [ fig : m31_mass_dens ] ) .
the disk is modelled by an exponential disk which includes stars and gas .
its mass is taken to be @xmath79 m@xmath12 and its scale radius is 3 kpc .
the bulge is modelled by a hernquist profile with a mass of @xmath80 m@xmath12 and a scale length of 1 kpc @xcite .
if all the components are assumed to be spherical , even the disk , their cumulative masses contribute to their gravitational forces as @xmath81 .
the tidal forces they would exert on and ii are @xmath82 . figure [ fig : m31_mass_dens ] shows that between 30 and 300 kpc , the dark matter component provides a dominant contribution to the cumulative mass , thus the gravitational force and the tidal force .
consequently , in the simulation presented in the next section , where and ii is always at a larger distance from m31 than 30 kpc , m31 is only modelled analytically by its dark matter halo . in order to compute the past orbit of and
ii around m31 , we would need to know its present 3d distance to m31 , its 3d velocity with respect to m31 and the mass and density profile of m31 during the last few gyr .
what we actually know is the 3d distance of and ii and m31 @xcite , the radial velocities of m31 and and ii in the mw rest frame and approximately the mass and the density profile of m31 ( see figure [ fig : m31_mass_dens ] ) .
we therefore have to make some assumptions to derive the orbit of and ii in the past .
we assumed ( 1 ) that the m31 density profile has not changed during the last 6 gyr , ( 2 ) that tangential velocities of m31 and and ii are smaller than 200 km s@xmath33 @xcite , ( 3 ) that the and ii trajectory takes it close to m31 @xmath45 5 gyr ago , to a distance between 30 and 40 kpc , to match the end of star formation 5 gyr ago .
this range is chosen in order not to destroy the dwarf and to agree with our model of m31 ( the pericentric distance @xmath83 kpc ) but at the same time to affect and ii enough to remove its gas ( @xmath84 kpc ) .
plane is defined by the present position of and ii , m31 and the mw .
the plot is centred on m31 .
the @xmath36-axis corresponds to the direction mw - m31 and the @xmath85-axis is the projected direction m31-and ii .
the trajectory of and ii does not lie in this plane .
the most important moments of the orbit are labelled . ] by choosing randomly tangential velocities for and ii and m31 within the limit of 200 km s@xmath33 , we find that there is only one family of trajectories that fit our assumptions .
these trajectories turn out to have only one pericentre close to m31 .
these pericentres can occur between 3.2 and 5.8 gyr ago .
we have selected pericentres to be between 30 and 40 kpc and then the apocentres lie between 330 and 360 kpc .
the properties of the trajectories are mainly constrained by the constraint on the pericentre .
figure [ fig : m31_orbit ] shows one of the trajectories we have selected . for this trajectory , 6
gyr ago and ii was at a distance of 212 kpc from m31 with a relative velocity of 118 km s@xmath33 .
the pericentre passage happened 4.8 gyr ago at a distance of 36 kpc with a velocity of 346 km s@xmath33 and the apocentre is at 294 kpc .
in addition to the trajectory choice , we can choose the orientation of the remnant galaxy .
a purely retrograde interaction , when the direction of the dwarf s angular momentum is opposite to the orbital angular momentum , reduces the internal stirring and mass loss @xcite .
this choice allows us to maintain the internal kinematics of the dwarf remnant which fits the observations but limits the gas ejection .
the inverse , an exactly prograde interaction , has opposite effects .
we launch two simulations , one prograde and one retrograde , which represent two extreme cases , in order to check the range of influence of tidal stripping on and ii . in each simulation
, star formation is turned off to make them computationally faster and study only one effect at a time .
we use the remnant galaxy from the previous merger simulation corresponding to and ii 5 gyr ago at the expected position of and ii 6 gyr ago ( see the previous section ) and we follow the simulation for 6 gyr , i.e. until the present time .
the snapshot choice , 5 gyr ago instead of 6 gyr ago , was done in order to have the right amount of new stars for and ii . there is no star formation during the tidal stripping simulation .
moreover , 5 or 6 gyr ago , and ii had similar properties which do not change the main result of the simulation . after 6 gyr , there remains 93% ( retrograde case ) and 88% ( prograde case ) of the initial gas in the dwarf .
the gas stripping due to the close encounter with m31 turns out to be insufficient .
it is mainly due to the high density of the dark matter halo and the baryonic matter in the dwarf galaxy centre .
the difference due to the orbital orientation takes the form of a nearly twice longer tail and a larger fraction of stripped mass for the prograde with respect to the retrograde trajectory ( 12% instead of 7% ) .
we conclude that the tidal stripping can not be the only mechanism of gas stripping . moreover ,
if the gas could be fully stripped by tidal forces , it would imply that the tidal force was strong enough to also strip the stars and the dwarf galaxy would be completely or partially destroyed .
the only way to remove the gas without destroying the whole dwarf galaxy is to employ a process which affects just the gas and not the stars : the ram pressure stripping . finally , if the tidal force does not affect the gas in the central parts it does not affect the stars neither .
this means that the stellar properties in terms of morphology and kinematics are conserved .
in addition to the disk , bulge and dark matter halo , galaxies like the mw or m31 seem to have a hot gas halo @xcite .
these haloes are believed to be responsible for a well known observational fact in the local universe : dwarf galaxies in the vicinity of massive galaxies or clusters have no gas , while dwarf galaxies in the field have large gas fractions @xcite .
this fact is explained by the mechanism of ram pressure stripping which reduces the gas content in dwarf galaxies as a result of interaction with the hot gas of their host @xcite .
the process depends mainly on two parameters : the density of the hot gas ( @xmath86 ) and the velocity of the dwarf galaxy moving within this medium ( @xmath87 ) .
the ram pressure is proportional to @xmath88 . in our study , in order to have efficient ram pressure stripping , the dwarf remnant of the major merger must be on an eccentric orbit and pass close to the centre of m31 where the gas density and the dwarf velocity are the highest .
these assumptions are consistent with the type of orbit described in the previous section . the hot gas halo of m31 is assumed to be spherical without streaming velocity nor velocity dispersion , i.e. we set @xmath89 for all gas particles in the initial conditions .
the pressure supports gravity only by the internal temperature which is very high @xmath45 10@xmath90 k. there are no strong observational constrains concerning the mass and density distribution of the hot gas halo of m31 .
however , previous works on hot gas haloes seem to indicate that these structures have mass roughly 100 times smaller than the dark matter mass of the host halo and have a similar spatial distribution @xcite . consequently , we approximate the hot gas halo of m31 as a hernquist profile with a total mass of @xmath91 m@xmath12@xmath23 and a scale length of 50 kpc , as for the dark matter halo . at 36 kpc , the assumed pericentre of the orbit , the density reaches @xmath92 m@xmath12kpc@xmath43 corresponding to @xmath46 atoms cm@xmath43 . in order to simulate the complete interaction between m31 and the remnant of the dwarf galaxy major merger with a constant gas resolution
, we would need about 80 million particles for the hot gas halo , which is too expensive computationally . in order to reduce the simulation time and at the same time reliably model the main physical process , we use a periodic box .
we fill the box with hot gas of constant density of @xmath46 atoms cm@xmath43 and temperature of @xmath94 k. the box has a size of 30 kpc in order to be significantly larger than the dwarf galaxy itself ( @xmath95 times the half - light diameter ) and at the same time not to contain too many particles ( close to 1 million ) .
the remnant dwarf galaxy is placed at the centre of the box without changing the orientation of its angular momentum .
we set a rather high bulk velocity of the dwarf @xmath96 km s@xmath33 .
the chosen density and velocity are consistent with the conditions at the pericentre discussed in the previous section .
the velocity vector of the galaxy is parallel to the direction of the angular momentum which helps to remove the gas .
we have decided to not include star formation in this ram - pressure stripping simulation .
however , we have ran a test simulation with star formation and found that , as expected , ram pressure compresses the gas and thus increases the star formation .
this star formation process does not form a stellar population that would be considered young at present because the ram - pressure stripping due to m31 is assumed to occur before 5 gyr ago and to finish nearly 5 gyr ago . due to ram pressure
the star formation can become stronger by a factor of 2 - 3 . at the same time
, ram pressure reduces the amount of gas and thus decreases the global star formation . in our ram - pressure model , we try to reproduce only the period when the ram - pressure stripping is very efficient thus when the dwarf is very close to m31 ( 30 - 60 kpc ) so that the dwarf velocity and the hot gas halo have their highest values .
this period is short , less than 1 gyr .
consequently , the star formation due to the ram pressure would just increase the star formation during a small period before the complete stripping of the gas .
the amount of stars due to the ram pressure would represent less than 15% of the total mass of stars of the second stellar population .
this is therefore an effect of second order on the star formation .
moreover , we do not take into account ram - pressure stripping when the simulated dwarf is further from m31 ( 60 - 150 kpc ) . in this region ,
ram - pressure stripping is less efficient but lasts longer .
it is probably not sufficiently strong to increase the central density and star formation but it can remove the gas in the outer parts and reduce the remaining ` gas fuel ' .
we think that only a complete simulation , including all the processes simultaneously , can properly model the star formation process when and ii is on an orbit around m31 .
-axis because the dwarf s velocity , thus the ram pressure stripping , is along this axis .
the stars and gas separate between 1.8 ( upper panel ) and 2.2 ( lower panel ) gyr of simulation in the periodic box . at 2.2 gyr
the gas is fully stripped and forms a clump 4 kpc behind the stars . ] after 0.5 gyr of evolution in the periodic box with a constant velocity and a high hot gas halo density , the gas remaining around the dwarf galaxy centre represents less than 60% of the initial gas . after about 2 gyr ,
the gas and stars start to separate . after 2.2 gyr
, they are completely disconnected and the dwarf becomes gas free ( see figure [ fig : res_rps ] ) .
this proves that the gas can be fully stripped from and ii via ram pressure stripping .
however , this simulation requires too much time ( more than 2 gyr ) with extreme conditions ( @xmath96 km s@xmath33 and @xmath97 cm@xmath43 ) . for a realistic orbit , and ii would remain close to the pericentre ( between 70 and 30 kpc ) during only @xmath45 0.5 gyr .
how can we reduce the gas stripping time ?
it does not seem reasonable to increase the velocity even more because and ii would then be unbound from m31 . on the other hand , the density of the hot gas could be larger because the mass of the hot gas halo is not strongly constrained .
if the gas density is increased by a factor of two or three , the stripping time decreases to values as low as 0.7 gyr and 0.3 gyr , respectively , for the same simulation .
our simulation is obviously simplified , it does not include star formation which decreases the central density and thus facilitates the ram pressure stripping , nor does it take into account the tidal force which would elongate the gas distribution making it easier to strip .
however , the central density of the gas with or without star formation differs only by a factor of about 2 and the star formation occurs only at the centre of the dwarf galaxy .
concerning the tidal force , it is mostly efficient in the outer parts .
another way to obtain faster gas stripping would be to modify the dark matter density profile .
indeed , the force which resists the ram pressure stripping is the gravitational force mainly due to dark matter . for comparison ,
the mass that generates the gravitational force within 0.3 kpc is @xmath98 m@xmath12 for the dark matter , @xmath99 m@xmath12 for the gas and @xmath100 m@xmath12 for the stars . with a lower central density
, the gas could be more easily stripped . among the simulations we ran ,
those with smaller dark matter content indeed could strip the gas within 1 gyr but the properties of the remnant dwarf did not match the observations so well .
this is consistent with the recent study by @xcite which shows that changing a cusp into a core in the dark matter profile , i.e. decreasing the central dark matter density , helps to remove the gas .
in addition to gas stripping , ram pressure has a small effect on the stars .
the ellipticity slightly increased from 0.22 to 0.29 at the new half - light radius of 1.0 kpc in the @xmath47 plane , which differs by @xmath101 from the observations .
this result can be due to the assumed alignment between the dwarf velocity and angular momentum . the gas is thus ejected along the @xmath36-axis , the major axis of the dwarf , which can increase the extent of the dwarf in this direction and the ellipticity .
figure [ fig : res_majmer_kine_2 ] illustrates the fact that the observed velocity dispersion has decreased .
this is due to the decrease of the mass of the dwarf and more specifically in the center after the stripping of the gas . in parallel ,
the stellar central density has also slightly decreased .
this leads to a better fit to the observations for the density profiles of the old and intermediate - age stars ( see figure [ fig : res_majmer_denspro ] ) .
with the new studies concerning the sfh @xcite and kinematics of dwarf galaxies @xcite , dsphs and dirrs seem more complicated than they have ever been .
dsphs were thought to be the descendants of primordial disky galaxies accreted by a bigger host galaxy and transformed morphologically and kinematically into pressure - supported systems over a time - scale of a few gyr .
this scenario may still be true for galaxies with only one old stellar population as is the case for some ultra - faint dwarf galaxies .
however , it does not hold for dsphs like and ii , fornax , leo i and carina which seem to have complicated and extended sfhs @xcite . for example , leo i was still forming stars 1 - 2 gyr ago . in this context , and ii
appears to be an example of a dwarf with a complicated history .
first , it probably is a result of a major merger between two gas - rich dwarf galaxies .
second , it could have been quickly stripped of its gas 5 gyr ago after one close passage around m31 . in the lg , massive dwarf ellipticals or dsphs like ngc205 , ngc147 , m32 or fornax ,
show either rotation or complicated kinematical structure @xcite and some even retain some gas , e.g. ngc205 which is much more massive than and ii @xcite .
each dwarf galaxy seems to be a specific and complicated puzzle to solve .
the first and most important test of our model would be provided by the velocity of and ii in the m31 rest frame .
when the tangential velocities of m31 and and ii are measured with high enough accuracy , they should indicate a close passage around m31 ( at a distance of 30 - 50 kpc ) in the past ( 4 - 6 gyr ago ) . if it turns out instead that and ii s does not follow a highly eccentric orbit , it will mean that the gas stripping is not due to the interaction with m31 . nowadays , even the tangential velocity of m31 is very uncertain .
two recent measurements give very different , inconsistent results : a value of 17 km s@xmath33 from @xcite and 164 km s@xmath33 by @xcite . for and ii
the tangential velocity is completely unknown
. another test could be the presence of high - velocity clouds ( hvc ) as the result of the quick stripping of gas from and ii 5 gyr ago .
our simulations ( see figure [ fig : res_rps ] ) and those of @xcite show that when a galaxy is stripped of gas by ram pressure some of the stripped gas can form gas clouds .
a new hi survey @xcite of the outskirts of m31 has recently provided some hints that an elongated hvc observed there could be formed from the gas stripped from andromeda xix .
the hvcs between and ii s position and the centre of m31 could also be the traces of the stripping .
however , even if the hypothesis is true , it would be difficult to prove because in our scenario the gas has been decoupled from the dwarf for the last @xmath102 gyr and the two components must have followed different trajectories in the hot gas halo .
a full simulation of the interaction of and ii with the m31 hot gas halo including ram pressure stripping , the formation hvc and tracing its trajectory could provide more insight .
how is it possible to strip and ii s gas in one passage close to m31 ?
one mechanism , that of ram pressure stripping was discussed in the previous section .
another possibility could be the star formation if we assume that during the major merger all the gas is turned into stars .
this would mean that the gas mass during the fusion is the same as the mass of intermediate - age stars : roughly @xmath66 m@xmath12 .
this implies an initial gas fraction close to 20% or a bit larger ( @xmath45 30% ) if we take into account the gas ejected during the major merger , which seems to be too low given the observational constraints .
moreover , from the theoretical point of view it is difficult to consume all the gas in a dwarf galaxy only by the star formation process .
this would require to have a star formation activity even when the gas density is low , i.e. below the commonly accepted density threshold .
the other possibility is to remove the gas by supernovae .
an interesting example in this context is a galaxy like leo p which is full of gas with star formation going on only at the centre where the density is high enough @xcite .
however , this level of star formation ( @xmath103 m@xmath12yr@xmath33 ) is not high enough to blow away the gas and if the gas density is decreased further due to the star formation , the dwarf will not be able to form new stars .
it would remain a dwarf galaxy with a few stars and a large reservoir of gas , not a dsph .
however , since leo p is more than 10 times less massive than and ii , the comparison may be not completely relevant .
this conclusion is supported by @xcite who found that gas stripping by supernova feedback is not very efficient .
the first limitation of our model of and ii is its degeneracy . in order to reduce the number of parameters we have assumed that the two dwarf galaxies are exactly the same .
it is a useful simplification but it indicates that the problem is degenerate .
moreover , even with the same progenitors there are still degeneracies present between the parameters .
for example , the gas fraction and feedback both have an effect on the sfh .
it is possible to choose a different pair of values for the feedback and gas fraction parameters and get a very similar dwarf remnant .
however , even if the parameters are allowed to change in our model , they can not change too much .
if we want a gas - rich major merger to reproduce the observables the second progenitor can not be too small or contain a low gas fraction .
second , another limitation concerns the inclination of and ii .
we assume implicitly in this study that and ii is seen along the line of sight which is perpendicular to its angular momentum vector .
contrary to disky galaxies , where the projected shape allows us to infer inclination , there is no simple way to find the inclination of and ii .
it is quite possible that and ii is seen at a slightly different angle and its rotation is therefore even higher .
third , our model depends on the exact numerical implementation of gas physics .
this physics is much more complicated to simulate than the stellar physics .
for example , the effects of feedback and ram pressure stripping are known to depend on the versions of code used @xcite . with a more complete description of the baryonic processes
, the gas could be stripped faster and the star formation could eject more gas from the dwarf .
fourth , the simulation of the ram - pressure stripping is idealized . the hot gas density and the dwarf velocity are constant which is surely not true if and ii was orbiting m31 on an eccentric orbit . moreover , we do not include star formation in this period .
these drawbacks are directly linked to the fact that our model is constructed from two simulations and not one complete simulation due to numerical limitations .
in this study , we have proposed a detailed scenario to explain the main properties of and ii . in our model , this dsph , a satellite of m31 , is the result of a major merger between two gas - rich disky dwarf galaxies in the field at high redshift ( @xmath104 2 - 3 ) .
this major merger would have formed a new stellar population due to the gas compression during the fusion phase .
it would also have transformed the morphology of disky progenitors into a spheroidal remnant . due to the specific inclination of the angular momenta of the progenitors ,
the remnant galaxy would have kept some rotation , but around the major axis and not the minor one as in disky galaxies .
in addition to this major merger , our model requires and ii to fall into the potential well of m31 about 6 gyr ago and reach the pericentre of its orbit @xmath45 5 gyr ago . as a result of this close interaction , and ii s gas
would have been quickly removed mainly by ram pressure stripping but also by tidal stripping .
this quick stripping can explain why and ii has stopped its star formation @xmath45 5 gyr ago .
our simulations succeeded in reproducing the different properties of and ii : ellipticity , morphology , kinematics , sfh , and the properties of the stellar populations .
only the quick gas stripping is not properly reproduced .
it could be due to several reasons : ( 1 ) a problem in the initial conditions where we assume a too concentrated dark matter halo ; ( 2 ) an over - simplified recipe for star formation which does not properly take into account the effect of supernovae which could eject more gas from the dwarf galaxy ; ( 3 ) a simplified simulation of gas stripping which does not take into account the tidal force from m31 and star formation or ( 4 ) too small density of hot gas in m31 .
these effects alone or in combination may explain the lack of gas in and ii at present .
in general , this study of the particular case of and ii indicates , as @xcite have done for a cosmological simulation , that a major merger of gas - rich dwarf galaxies can be an important physical process to transform dirrs into dsphs and to explain their complicated sfhs .
this research was supported in part by the polish national science centre under grant 2013/10/a / st9/00023 and by the project rvo:6798581 .
we are grateful to l. widrow for providing procedures to generate @xmath0-body realizations of galaxies for initial conditions .
amorisco n. c. , evans n. w. , van de ven g. , 2014 , , 507 , 335 agertz o. et al .
, 2007 , , 380 , 963 bechtol k. et al . , 2015 , apj , 807 , 50 belokurov v. et al . , 2006 , apj , 647 , l111 belokurov v. et al . , 2007 , apj , 654 , 897 bentez - llambay a. , navarro j. f. , abadi m. g. , gottlber s. , yepes g. , hoffman y. , steinmetz m. , 2016 , , 456 , 1185 carlesi e. , hoffman y. , sorce j. g. , gottlber s. , yepes g. , courtois h. , tully r. b. , 2016 , mnras , 460 , l5 conn a. r. et al . , 2011 , apj , 740 ,
69 cox t. j. , jonsson p. , primack j. r. , somerville r. s. , 2006 , mnras , 373 , 1013 deason a. , wetzel a. , garrison - kimmel s. , 2014 , , 794 , 115 dekel a. , silk j. , 1986 , apj , 303 , 39 del pino a. , aparicio a. , hidalgo s. l. , 2015 , , 454 , 3996 del pino a. , aparicio a. , hidalgo s. l. , okas e. l. , 2016 , , submitted , arxiv:1605.09414 dressler a. , 1980 , , 236 , 351 ebrov i. , okas e. l. , 2015 , , 813 , 10 emerick a. , mac low m. , grcevich j. , gatto a. , 2016 , apj , 826 , 148 geha m. , guhathakurta p. , rich r. m. , cooper m. c. , 2006 , , 131 , 332 governato f. et al . , 2010 , nature , 463 , 203 governato f. et al . , 2012 , mnras , 422 , 1231 grcevich j. , putman m. e. , 2009 , , 696 , 385 gunn j. e. , gott j. r. , 1972 , , 176 , 1 hammer f. , yang y. b. , wang j. l. , puech m. , flores h. , fouquet s. , 2010 , apj , 725 , 542 hernquist l. , 1993 , , 86 , 389 ho n. et al . , 2012 , , 758 , 124 hunter d. a. et al . , 2012 , , 144 , 134 ibata r. , martin n. f. , irwin m. , chapman s. , ferguson a. m. n. , lewis g. f. , mcconnachie a. w. , 2007 , apj , 671 , 1591 kalirai j. s. et al . , 2010 , , 711 , 671 kazantzidis s. , okas e. l. , callegari s. , mayer l. , moustakas l. a. , 2011a , apj , 726 , 98 kazantzidis s. , okas e. l. , mayer l. , knebe a. , klimentowski j. , 2011b , apj , 740 , l24 kerp j. , kalberla p. m. w. , ben bekhti n. , fler l. , lenz d. , winkel b. , 2016 , a&a , 589 , a120 khochfar s. , burkert a. , 2006 , , 445 , 403 kirby e. n. , bullock j. s. , boylan - kolchin m. , kaplinghat m. , cohen j. g. , 2014 , , 439 , 1015 klimentowski j. , okas e. l. , kazantzidis s. , mayer l. , mamon g. a. , 2009 , mnras , 397 , 2015 klimentowski j. , okas e. l. , knebe a. , gottlber s. , martinez - vaquero l. a. , yepes g. , hoffman y. , 2010 , , 402 , 1899 koposov s. e. , belokurov v. , torrealba g. , evans n. w. , 2015 , apj , 805 , 130 laevens b. p. m. , et al .
2015 , apj , 813 , 44 okas e. l. , kazantzidis s. , mayer l. , 2011 , apj , 739 , 46 okas e. l. , majewski s. r. , kazantzidis s. , mayer l. , carlin j. l. , nidever d. l. , moustakas l. a. , 2012 , apj , 751 , 61 okas e. l. , ebrov i. , del pino a. , semczuk m. , 2014 , , 445 , l6 okas e. l. , semczuk m. , gajda g. , donghia e. , 2015 , apj , 810 , 100 martin n. f. et al . , 2009 , apj , 705 , 758 martin n. f. et al .
, 2013 , apj , 772 , 15 mayer l. , governato f. , colpi m. , moore b. , quinn t. , wadsley j. , stadel j. , lake g. , 2001 , apj , 559 , 754 mayer l. , kazantzidis s. , mastropietro c. , wadsley j. , 2007 , nature , 445 , 738 mcconnachie a. w. , irwin m. j. , 2006 , , 365 , 1263 mcconnachie a. w. , arimoto n. , irwin m. , 2007 , , 379 , 379 mcconnachie a. w. , 2012 , aj , 144 , 4 mcquinn k. b. w. et al .
, 2015 , , 812 , 158 miller m. j. , bregman j. n. , 2015 , , 800 , 14 navarro j. f. , frenk c. s. , white s. d. m. , 1996 , , 462 , 563 nichols m. , revaz y. , jablonka p. , 2015 ,
a&a , 582 , a23 oshea b. w. , nagamine k. , springel v. , hernquist l. , norman m. l. , 2005 , , 160 , 1 rodrigues m. , puech m. , hammer f. , rothberg b. , flores h. , 2012 , mnras , 421 , 2888 sadoun r. , mohayaee r. , colin j. , 2014 , , 442 , 160 sakamoto t. , hasegawa t. , 2006 , apj , 653 , l29 salomon j .- b .
, ibata r. a. , famaey b. , martin n. f. , lewis g. f. , 2016 , , 456 , 4432 skillman e. d. et al .
, 2016 , apj , submitted , arxiv:1606.01207 sofue y. , 2015 , , 198 sohn s. t. , anderson j. , van der marel r. p. , 2012
, , 753 , 7 spekkens k. , urbancic n. , mason b. s. , willman b. , aguirre j. e. , 2014 , apj , 795 , l5 springel v. , yoshida n. , white s. d. m. , 2001 , new astronomy , 6 , 79 springel v. , 2005 , mnras , 364 , 1105 van den bergh s. , 1972 , apj , 171 , l31
walsh s. m. , jerjen h. , willman b. , 2007 , apj , 662 , l83 wang j. , hammer f. , athanassoula e. , puech m. , yang y. , flores h. , 2012 , a&a , 538 , a121 watkins l. l. , evans n. w. , an j. h. , 2010 , mnras , 406 , 264 weisz d. r. , dolphin a. e. , skillman e. d. , holtzman j. , gilbert k. m. , dalcanton j. j. , williams b. f. , 2014 , apj , 789 , 147 widrow l. m. , perrett k. m. , suyu s. h. , 2003 , , 588 , 311 widrow l. m. , dubinski j. , 2005 , apj , 631 , 838 widrow l. m. , pym b. , dubinski j. , 2008 , apj , 679 , 1239 willman b. et al . , 2005 , apj , 626 , l85 young l. m. , lo k. y. , 1997 , , 476 , 127 zucker d. b. et al . , 2006 , apj , 643 , l103 | andromeda ii ( and ii ) has been known for a few decades but only recently observations have unveiled new properties of this dwarf spheroidal galaxy .
the presence of two stellar populations , the bimodal star formation history ( sfh ) and an unusual rotation velocity of and ii put strong constrains on its formation and evolution .
following @xcite , we propose a detailed model to explain the main properties of and ii involving ( 1 ) a gas - rich major merger between two dwarf galaxies at high redshift in the field and ( 2 ) a close interaction with m31 about 5 gyr ago .
the model is based on @xmath0-body / hydrodynamical simulations including gas dynamics , star formation and feedback .
one simulation is designed to reproduce the gas - rich major merger explaining the origin of stellar populations and the sfh .
other simulations are used to study the effects of tidal forces and the ram pressure stripping during the interaction between and ii and m31 .
the model successfully reproduces the sfh of and ii including the properties of stellar populations , its morphology , kinematics and the lack of gas .
further improvements to the model are possible via joint modelling of all processes and better treatment of baryonic physics .
[ firstpage ] galaxies : individual : andromeda ii galaxies : local group galaxies : dwarf galaxies : kinematics and dynamics galaxies : evolution galaxies : interactions |
a bayesian network or directed graphical model is a statistical model that uses a directed acyclic graph ( dag ) to represent the conditional independence structures between collections of random variables .
the word bayesian is used to describe these models because the nodes in the graph can be used to represent random variables that correspond to parameters or hyperparameters , though the basic models themselves are not a priori bayesian .
these models are used throughout computational statistics to model complex interactions between collections of random variables .
for instance , tree models are used in computational biology for sequence alignment @xcite and in phylogenetics @xcite .
special cases of bayesian networks include familiar models from statistics like factor analysis @xcite and the hidden markov model @xcite .
the dag that specifies the bayesian network specifies the model in two ways .
the first is through a recursive factorization of the parametrization , via restricted conditional distributions .
the second method is via the conditional independence statements implied by the graph .
the recursive factorization theorem ( * ? ? ?
* thm 3.27 ) says that these two methods for specifying a bayesian network yield the same family of probability density functions .
when the underlying random variables are gaussian or discrete , conditional independence statements can be interpreted as algebraic constraints on the parameter space of the global model . in the gaussian case , this means that conditional independence corresponds to algebraic constraints on the cone of positive definite matrices .
one of our main goals in this paper is to explore the recursive factorization theorem using algebraic techniques in the case of gaussian random variables , with a view towards the case of hidden random variables . in this sense ,
the current paper is a generalization of the work began in @xcite which concerned the special case of factor analysis .
some past work has been done on the algebraic geometry of bayesian networks in the discrete case in @xcite , but there are many open questions that remain in both the gaussian and the discrete case . in the next section ,
we describe a combinatorial parametrization of a bayesian network in the gaussian case . in statistics , this parametrization in known as the _ trek rule _ @xcite .
we also describe the algebraic interpretation of conditional independence in the gaussian case which leads us to our main problem : comparing the vanishing ideal of the model @xmath0 to the conditional independence ideal @xmath1 .
section [ sec : comp ] describes the results of computations regarding the ideals of bayesian networks , and some algebraic conjectures that these computations suggest .
in particular , we conjecture that the coordinate ring of a bayesian network is always normal and cohen - macaulay . as a first application of our algebraic perspective on gaussian bayesian networks , we provide a new and greatly simplified proof of the tetrad representation theorem ( * ? ? ?
* thm 6.10 ) in section [ sec : tetrad ] .
then in section [ sec : tree ] we provide an extensive study of trees in the fully observed case .
in particular , we prove that for any tree @xmath2 , the ideal @xmath3 is a toric ideal generated by linear forms and quadrics that correspond to conditional independence statements implied by @xmath2 .
techniques from polyhedral geometry are used to show that @xmath4/i_t$ ] is always normal and cohen - macaulay .
sections [ sec : hidtree ] and [ sec : alggeom ] are concerned with the study of hidden variable models . in section [ sec : hidtree ] we prove the upstream variables theorem ( theorem [ thm : upstream ] ) which shows that @xmath0 is homogeneous with respect to a two dimensional multigrading induced by upstream random variables . as a corollary ,
we deduce that hidden tree models are generated by tetrad constraints . finally in section [ sec : alggeom ] we show that models with hidden variables include , as special cases , a number of classical constructions from algebraic geometry .
these include toric degenerations of the grassmannian , matrix schubert varieties , and secant varieties .
i would like to thank mathias drton , thomas richardson , mike stillman , and bernd sturmfels for helpful comments and discussions about the results in this paper .
the i m a provided funding and computer equipment while i worked on parts of this project .
let @xmath5 be a directed acyclic graph ( dag ) with vertex set @xmath6 and edge set @xmath7 .
often , we will assume that @xmath8 : = \{1,2 , \ldots , n \}$ ] . to guarantee the acyclic assumption ,
we assume that the vertices are _ numerically ordered _ ; that is , @xmath9 only if @xmath10 . the bayesian network associated to this graph can be specified by either a recursive factorization formula or by conditional independence statements .
we focus first on the recursive factorization representation , and use it to derive an algebraic description of the parametrization .
then we introduce the conditional independence constraints that vanish on the model and the ideal that these constraints generate .
let @xmath11 be a random vector , and let @xmath12 denote the probability density function of this random vector .
bayes theorem says that this joint density can be factorized as a product @xmath13 where @xmath14 denotes the conditional density of @xmath15 given @xmath16 .
the recursive factorization property of the graphical model is that each of the conditional densities @xmath14 only depends on the parents @xmath17 \,\ , |\ , \ , j \to i \in e(g ) \}$ ] .
we can rewrite this representation as @xmath18 thus , a density function @xmath19 belongs to the bayesian network if it factorizes as @xmath20 to explore the consequences of this parametrization in the gaussian case , we first need to recall some basic facts about gaussian random variables .
each @xmath21-dimensional gaussian random variable @xmath22 is completely specified by its mean vector @xmath23 and its positive definite covariance matrix @xmath24 .
given these data , the joint density function of @xmath22 is given by @xmath25 where @xmath26 is the determinant of @xmath24 .
rather than writing out the density every time , the shorthand @xmath27 is used to indicate that @xmath22 is a gaussian random variable with mean @xmath23 and covariance matrix @xmath24 .
the multivariate gaussian generalizes the familiar `` bell curve '' of a univariate gaussian and is an important distribution in probability theory and multivariate statistics because of the central limit theorem @xcite .
given an @xmath21-dimensional random variable @xmath22 and @xmath28 $ ] , let @xmath29 .
similarly , if @xmath30 is a vector , then @xmath31 is the subvector indexed by @xmath32 . for a matrix @xmath24 , @xmath33 is the submatrix of @xmath24 with row index set @xmath32 and column index set @xmath34 .
among the nice properties of gaussian random variables are the fact that marginalization and conditioning both preserve the gaussian property ; see @xcite .
suppose that @xmath35 and let @xmath36 $ ] be disjoint
. then 1 .
@xmath37 and 2 .
@xmath38 to build the gaussian bayesian network associated to the dag @xmath5 , we allow any gaussian conditional distribution for the distribution @xmath39 . this conditional distribution is recovered by saying that @xmath40 where @xmath41 and is independent of the @xmath15 with @xmath10 , and the @xmath42 are the regression parameters .
linear transformations of gaussian random variables are gaussian , and thus @xmath22 is also a gaussian random variable . since @xmath22 is completely specified by its mean @xmath23 and covariance matrix @xmath24 , we must calculate these from the conditional distribution .
the recursive expression for the distribution of @xmath43 given the variables preceding it yields a straightforward and recursive expression for the mean and covariance .
namely @xmath44 and if @xmath45 the covariance is : @xmath46 and the variance satisfies : @xmath47 if there are no constraints on the vector @xmath48 , there will be no constraints on @xmath23 either .
thus , we will focus attention on the constraints on the covariance matrix @xmath24 .
if we further assume that the @xmath49 are completely unconstrained , this will imply that we can replace the messy expression for the covariance @xmath50 by a simple new parameter @xmath51 .
this leads us to the algebraic representation of our model , called the _ trek rule _ @xcite . for each edge
@xmath52 let @xmath42 be an indeterminate and for each vertex @xmath53 let @xmath54 be an indeterminate .
assume that the vertices are numerically ordered , that is @xmath55 only if @xmath10 .
a _ collider _ is a pair of edges @xmath56 , @xmath57 with the same head .
for each pair of vertices @xmath58 , let @xmath59 be the collection of simple paths @xmath60 in @xmath5 from @xmath61 to @xmath62 such that there is no collider in @xmath60 .
such a colliderless path is called a _ trek_. the name trek come from the fact that every colliderless path from @xmath61 to @xmath62 consists of a path from @xmath61 up to some topmost element @xmath63 and then from @xmath63 back down to @xmath62 .
we think of each trek as a sequence of edges @xmath64 . if @xmath65 , @xmath66 consists of a single empty trek from @xmath61 to itself .
let @xmath67 be the ring homomorphism @xmath68 \rightarrow { \mathbb{c}}[a_i , \lambda_{ij } \ , \ , | \ , \
, i , j \in [ n ] i \to j \in e(g ) ] \ ] ] @xmath69 when @xmath65 , we get @xmath70 . if there is no trek in @xmath59 , then @xmath71 .
let @xmath72 .
since @xmath0 is the kernel of a ring homomorphism , it is a prime ideal .
[ ex:4cycle ] let @xmath5 be the directed graph on four vertices with edges @xmath73 , @xmath74 , @xmath75 , and @xmath76 .
the ring homomorphism @xmath67 is given by @xmath77 the ideal @xmath0 is the complete intersection of a quadric and a cubic : @xmath78 dual to the ring homomorphism is the rational parametrization @xmath79 @xmath80 we will often write @xmath81 to denote the coordinate polynomial that represents this function
. let @xmath82 be the subset of parameter space satisfying the constraints : @xmath83 for all @xmath61 , where in the case that @xmath84 the sum is zero .
[ prop : goodfac][trek rule ] the set of covariance matrices in the gaussian bayesian network associated to @xmath5 is the image @xmath85 . in particular , @xmath0 is the vanishing ideal of the model .
the proof of the trek rule parametrization can also be found in @xcite .
the proof goes by induction .
first , we make the substitution @xmath86 which is valid because , given the @xmath42 s , @xmath49 can be recovered from @xmath51 and vice versa .
clearly @xmath87 .
by induction , suppose that the desired formula holds for all @xmath88 with @xmath89 .
we want to show that @xmath90 has the same formula . now from above ,
we have @xmath91 this last expression is a factorization of @xmath92 since every trek from @xmath61 to @xmath21 is the union of a trek @xmath93 and an edge @xmath94 where @xmath95 is some parent of @xmath21 .
the parameters used in the trek rule parametrization are a little unusual because they involve a mix of the natural parameters ( regression coefficients @xmath42 ) and coordinates on the image space ( variance parameters @xmath54 ) . while this mix might seem unusual from a statistical standpoint , we find that this parametrization is rather useful for exploring the algebraic structure of the covariance matrices that come from the model .
for instance : if @xmath2 is a tree , then @xmath3 is a toric ideal . for any pair of vertices @xmath96 in @xmath2
, there is at most one trek between @xmath61 and @xmath62 .
thus @xmath97 is a monomial and @xmath3 is a toric ideal .
in fact , as we will show in section [ sec : tree ] , when @xmath2 is a tree , @xmath3 is generated by linear forms and quadratic binomials that correspond to conditional independence statements implied by the graph . before getting to properties of conditional independence
, we first note that these models are identifiable .
that is , it is possible to recover the @xmath42 and @xmath54 parameters directly from @xmath24 .
this also allows us to determine the most basic invariant of @xmath0 , namely its dimension .
[ prop : dim ] the parametrization @xmath98 is birational . in other words , the model parameters @xmath42 and @xmath54 are identifiable and @xmath99 .
it suffices to prove that the parameters are identifiable via rational functions of the entries of @xmath24 , as all the other statements follow from this .
we have @xmath100 so the @xmath54 parameters are identifiable .
we also know that for @xmath10 @xmath101 thus , we have the matrix equation @xmath102 where @xmath103 is the vector @xmath104 .
since @xmath105 is invertible in the positive definite cone , we have the rational formula @xmath106 and the @xmath42 parameters are identifiable .
one of the problems we want to explore is the connection between the prime ideal defining the graphical model ( and thus the image of the parametrization ) and the relationship to the ideal determined by the independence statements induced by the model . to explain this connection , we need to recall some information about the algebraic nature of conditional independence .
recall the definition of conditional independence .
let @xmath32 , @xmath34 , and @xmath107 be disjoint subsets of @xmath108 $ ] , indexing subsets of the random vector @xmath22 .
the conditional independence statement @xmath109 ( @xmath32 is independent of @xmath34 given @xmath107 ) holds if and only if @xmath110 for all @xmath111 such that @xmath112 .
we refer to @xcite for a more extensive introduction to conditional independence . in the gaussian case ,
a conditional independence statement is equivalent to an algebraic restriction on the covariance matrix .
let @xmath113 be disjoint subsets of @xmath108 $ ] .
then @xmath114 satisfies the conditional independence constraint @xmath109 if and only if the submatrix @xmath115 has rank less than or equal to @xmath116 . if @xmath117 , then @xmath118 the ci statement @xmath109 holds if and only if @xmath119 . the @xmath120 submatrix of @xmath121 is easily seen to be @xmath122 which is the schur complement of the matrix @xmath123 since @xmath124 is always invertible ( it is positive definite ) , the schur complement is zero if and only if the matrix @xmath125 has rank less than or equal to @xmath116 . given a dag @xmath5 ,
a collection of conditional independence statements are forced on the joint distribution by the nature of the graph .
these independence statements are usually described via the notion of @xmath126-separation ( the @xmath126 stands for `` directed '' ) .
let @xmath32 , @xmath34 , and @xmath107 be disjoint subsets of @xmath108 $ ] .
the set @xmath107 _ d - separates _
@xmath32 and @xmath34 if every path in @xmath5 connecting a vertex @xmath127 and @xmath128 contains a vertex @xmath95 that is either 1 . a non - collider that belongs to @xmath107 or 2 . a collider that does not belong to @xmath107 and has no descendants that belong to @xmath107 .
note that @xmath107 might be empty in the definition of @xmath126-separation .
the conditional independence statement @xmath129 holds for the bayesian network associated to @xmath5 if and only if @xmath107 @xmath126-separates @xmath32 from @xmath34 in @xmath5 .
a joint probability distribution that satisfies all the conditional independence statements implied by the graph @xmath5 is said to satisfy the global markov property of @xmath5 .
the following theorem is a staple of the literature of graphical models , that holds with respect to any @xmath130-algebra .
* thm 3.27 ) a probability density has the recursive factorization property with respect to @xmath5 if and only if it satisfies the global markov property .
let @xmath131 $ ] be the ideal generated by the minors of @xmath24 corresponding to the conditional independence statements implied by @xmath5 ; that is , @xmath132 the ideal @xmath1 is called the _
conditional independence ideal _ of @xmath5 .
a direct geometric consequence of the recursive factorization theorem is the following for any dag @xmath5 , @xmath133 in the corollary @xmath134 is the cone of @xmath135 positive definite symmetric matrices .
it seems natural to ask whether or not @xmath136 for all dags @xmath5 .
for instance , this was true for the dag in example [ ex:4cycle ] .
the verma graph provides a natural counterexample .
[ ex : verma ] let @xmath5 be the dag on five vertices with edges @xmath74 , @xmath137 , @xmath138 , @xmath75 , @xmath76 , and @xmath139 .
this graph is often called the verma graph .
the conditional independence statements implied by the model are all implied by the three statements @xmath140 , @xmath141 , and @xmath142 .
thus , the conditional independence ideal @xmath1 is generated by one linear form and five determinantal cubics . in this case , we find that @xmath143 where @xmath19 is the degree four polynomial : @xmath144 we found that the primary decomposition of @xmath1 is @xmath145 so that @xmath19 is not even in the radical of @xmath1 .
thus , the zero set of @xmath1 inside the positive _ _
semi__definite cone contains singular covariance matrices that are not limits of distributions that belong to the model . note
that since none of the indices of the @xmath88 appearing in @xmath19 contain @xmath146 , @xmath19 vanishes on the marginal distribution for the random vector @xmath147 .
this is the gaussian version of what is often called the verma constraint .
note that this computation shows that the verma constraint is still needed as a generator of the unmarginalized verma model .
the rest of this paper is concerned with studying the ideals @xmath0 and investigating the circumstances that guarantee that @xmath148 .
we report on results of a computational study in the next section . towards the end of the paper ,
we study the ideals @xmath149 that arise when some of the random variables are hidden .
whenever approaching a new family of ideals , our first instinct is to compute as many examples as possible to gain some intuition about the structure of the ideals .
this section summarizes the results of our computational explorations .
we used macaulay2 @xcite to compute the generating sets of all ideals @xmath0 for all dags @xmath5 on @xmath150 vertices .
our computational results concerning the problem of when @xmath148 are summarized in the following proposition .
all dags on @xmath151 vertices satisfy @xmath148 . of the @xmath152 dags on @xmath153 vertices ,
exactly @xmath154 satisfy @xmath148 . of the @xmath155 dags on @xmath156 vertices
exactly @xmath157 satisfy @xmath148 .
on @xmath158 vertices , there were precisely nine graphs that fail to satisfy @xmath148 .
these nine exceptional graphs are listed below .
the numberings of the dags come from the atlas of graphs @xcite .
note that the verma graph from example [ ex : verma ] appears as @xmath159 after relabeling vertices .
1 . @xmath160 : @xmath161 , @xmath137 , @xmath75 , @xmath76 , @xmath162 .
@xmath163 : @xmath74 , @xmath138 , @xmath164 , @xmath76 , @xmath139 .
@xmath165 : @xmath73 , @xmath74 , @xmath137 , @xmath75 , @xmath76 , @xmath139 .
@xmath166 : @xmath167 , @xmath161 , @xmath138 , @xmath164 , @xmath76 , @xmath139 .
@xmath168 : @xmath74 , @xmath161 , @xmath75 , @xmath164 , @xmath169 , @xmath139
@xmath159 : @xmath74 , @xmath161 , @xmath138 , @xmath164 , @xmath76 , @xmath139
@xmath170 : @xmath73 , @xmath161 , @xmath137 , @xmath138 , @xmath164 , @xmath76 , @xmath139
@xmath171 : @xmath73 , @xmath74 , @xmath137 , @xmath75 , @xmath76 , @xmath172 , @xmath139 .
@xmath173 : @xmath73 , @xmath174 , @xmath138 , @xmath164 , @xmath76 , @xmath172 , @xmath139 .
the table below displays the numbers of minimal generators of different degrees for each of the ideals @xmath0 where @xmath5 is one of the nine graphs on five vertices such that @xmath175 .
the coincidences among rows in this table arise because sometimes two different graphs yield the same family of probability distributions .
this phenomenon is known as markov equivalence @xcite . [
cols="^,^,^,^,^,^",options="header " , ] let @xmath176 denote the polyhedron defined as the solution space to the given constraints .
first of all , @xmath176 is bounded . to see this , first note that because of the positive constraints and the equation @xmath177 , we have that @xmath178 is implied by the given constraints .
then , starting from the sources of the tree and working our way down the edges repeatedly using the inequalities @xmath179 , we see that the @xmath180 are also bounded .
now , we have @xmath181 , since every trek will satisfy any of the indicated constraints .
thus , we must show that @xmath182 . to do this
, it suffices to show that for any vector @xmath183 , there exists @xmath184 , @xmath185 and @xmath186 such that @xmath187 where @xmath185 is one of the @xmath188 vectors @xmath189 and @xmath190 . because @xmath176 is bounded
, this will imply that the extreme points of @xmath176 are a subset of the extreme points of @xmath191 , and hence @xmath182 . without loss of generality
we may suppose that all of the coordinates @xmath192 are positive , otherwise the problem reduces to a smaller tree or forest because the resulting inequalities that arise when @xmath193 are precisely those that are necessary for the smaller tree .
note that for a forest @xmath194 , the polytope @xmath195 is the direct join of polytopes @xmath191 as @xmath2 ranges over the connected components of @xmath194 , by proposition [ prop : disconnect ] .
for any fixed @xmath62 , there can not exist distinct values @xmath196 , @xmath197 , and @xmath198 such that all of @xmath199 @xmath200 @xmath201 hold .
if there were , we could add these three equations together to deduce that @xmath202 this in turn implies that @xmath203 with equality if and only if @xmath204 and @xmath205 .
this in turn implies that , for instance , @xmath206 contradicting our assumption that @xmath207 for all @xmath61 and @xmath62 . by a similar argument , if exactly two of these facet defining inequalities hold sharply , we see that @xmath208 which implies that @xmath62 has exactly two descendants and no parents .
now mark each edge @xmath57 in the tree @xmath2 such that @xmath209 by the preceding paragraph , we can find a trek @xmath60 from a sink in the tree to a source in the tree and ( possibly ) back to a different sink that has the property that for no @xmath61 in the trek there exists @xmath95 not in the path such that @xmath56 is a marked edge .
that is , the preceding paragraph shows that there can be at most @xmath210 marked edges incident to any given vertex .
given @xmath60 , let @xmath185 denote the corresponding @xmath188 vector .
we claim that there is a @xmath184 such that @xmath211 holds with @xmath212 .
take @xmath213 to be any very small number and define @xmath186 by the given equation .
note that by construction the inequalities @xmath214 and @xmath215 will be satisfied since for all the nonzero entries in @xmath185 , the corresponding inequality for @xmath216 must have been nonstrict and @xmath217 is small .
furthermore , the constraint @xmath218 is also automatically satisfied .
it is also easy to see that the last set of inequalities will also be satisfied since through each vertex the path will either have no edges , an incoming edge and an outgoing edge , or two outgoing edges and the top vertex , all of which do not change the value of the linear functional . finally to see that the inequalities of the form @xmath219 are still satisfied by @xmath220 , note that marked edges of @xmath2
are either contained in the path @xmath60 or not incident to the path @xmath60 .
thus , the strict inequalities remain strict ( since they will involve modifying by an incoming edge and an outgoing edge or an outgoing edge and the top vertex ) , and the nonstrict inequalities remain nonstrict since @xmath217 is small .
thus , we conclude that @xmath182 , which completes the proof .
[ cor : pull ] let @xmath221 be any reverse lexicographic term order such that @xmath222 for all @xmath61 and @xmath223
. then @xmath224 is squarefree .
in other words , the associated pulling triangulation of @xmath191 is unimodular .
the proof is purely polyhedral , and relies on the geometric connections between triangulations and initial ideals of toric ideals .
see chapter 8 in @xcite for background on this material including pulling triangulations .
let @xmath189 denote the vertex of @xmath191 corresponding to the monomial @xmath225 . for @xmath226 ,
each of the vertices @xmath189 has lattice distance at most one from any of the facets described by theorem [ thm : facets ] .
this is seen by evaluating each of the linear functionals at the @xmath188 vector corresponding to the trek between @xmath61 and @xmath62 .
if we pull from one of these vertices we get a unimodular triangulation provided that the induced pulling triangulation on each of the facets of @xmath191 not containing @xmath189 is unimodular .
this is because the normalized volume of a simplex is the volume of the base times the lattice distance from the base to the vertex not on the base .
the facet defining inequalities of any face of @xmath191 are obtained by taking an appropriate subset of the facet defining inequalities of @xmath191 .
thus , as we continue the pulling triangulation , if the current face contains a vertex @xmath189 with @xmath227 , we will pull from this vertex first and get a unimodular pulling triangulation provided the induced pulling triangulation of every face is unimodular .
thus , by induction , it suffices to show that the faces of @xmath191 that are the convex hull of vertices @xmath228 have unimodular pulling triangulations .
however , these faces are always unimodular simplices .
the ring @xmath229/i_t$ ] is normal and cohen - macaulay when @xmath2 is a tree .
since @xmath191 has a unimodular triangulation , it is a normal polytope and hence the semigroup ring @xmath229/i_t$ ] is normal .
theorem @xcite then implies that @xmath229/i_t$ ] is cohen - macaulay . while we know that @xmath229/i_t$ ] is always cohen - macaulay , it remains to determine how the cohen - macaulay type of @xmath3 depends on the underlying tree @xmath2 .
here is a concrete conjecture concerning the special case of gorenstein trees .
suppose that @xmath2 is a directed tree .
then @xmath229/i_t$ ] is gorenstein if and only if the degree of every vertex in @xmath2 is less than or equal to three . a _ downward directed tree _
is a tree all of whose edges point to the unique sink in the tree .
a leaf of such a downward directed tree is then a source of the tree . with a little more refined information about which inequalities defining @xmath191 are facet defining
, we can deduce results about the degrees of the ideals @xmath3 in some cases .
[ cor : treedegree ] let @xmath2 be a downward directed tree and let @xmath61 be any leaf of @xmath2 , @xmath230 the sink of @xmath2 , and @xmath60 the unique trek in @xmath231 .
then @xmath232 where @xmath233 denotes the forest obtained from @xmath2 by removing the edge @xmath64 .
first of all , note that in the case of a downward directed tree the inequalities of the form @xmath234 are redundant : since each vertex has at most one descendant , it is implied by the the other constraints . also , for any source @xmath235 , the inequality @xmath236 is redundant , because it is implied by the inequalities @xmath237 and @xmath238 where @xmath62 is the unique child of @xmath235 .
now we will compute the normalized volume of the polytope @xmath191 ( which is equal to the degree of the toric ideal @xmath3 ) by computing the pulling triangulation from corollary [ cor : pull ] and relating the volumes of the pieces to the associated subforests .
since the pulling triangulation of @xmath191 with @xmath239 pulled first is unimodular , the volume of @xmath191 is the sum of the volumes of the facets of @xmath191 that do not contain @xmath239 .
note that @xmath239 lies on all the facets of the form @xmath219 since through every vertex besides the source and sink , the trek has either zero or two edges incident to it .
thus , the only facets that @xmath239 does not lie on are of the form @xmath240 such that @xmath64 is an edge in the trek @xmath60 . however , the facet of @xmath191 obtained by setting @xmath241 is precisely the polytope @xmath242 , which follows from theorem [ thm : facets ] .
note that upon removing an edge in a tree we obtain a forest .
proposition [ prop : disconnect ] implies that the degree of such a forest is the product of the degrees of the associated trees . since the degree of the tree consisting of a single point is one , the formula from corollary [ cor : treedegree ] yields a recursive expression for the degree of a downward directed forest .
let @xmath243 be the directed chain with @xmath21 vertices .
then @xmath244 , the @xmath245st catalan number . in corollary
[ cor : treedegree ] we take the unique path from @xmath146 to @xmath21 .
the resulting forests obtained by removing an edge are the disjoint unions of two paths . by the product formula implied by proposition [ prop : disconnect ] we deduce that the degree of @xmath246 satisfies the recurrence : @xmath247 with initial condition @xmath248 .
this is precisely the recurrence and initial conditions for the catalan numbers @xcite .
now we want to prove the main result of this section , that the determinantal conditional independence statements actually generate the ideal @xmath3 when @xmath2 is a tree . to do this , we will exploit the underlying toric structure , introduce a tableau notation for working with monomials , and introduce an appropriate ordering of the variables .
each variable @xmath88 that is not zero can be identified with the unique trek in @xmath2 from @xmath61 to @xmath62 .
we associate to @xmath88 the tableau which records the elements of @xmath2 in this unique trek , which is represented like this : @xmath249\ ] ] where @xmath34 and @xmath107 are ( possibly empty ) strings . if , say , @xmath61 were at the top of the path , we would write the tableau as @xmath250.\ ] ] the tableau is in its standard form if @xmath251 is lexicographically earlier than @xmath252 .
we introduce a lexicographic total order on standard form tableau variables by declaring @xmath253 \prec [ \underline{c}c|\underline{c}d]$ ] if @xmath254 is lexicographically smaller that @xmath255 , or if @xmath256 and @xmath257 is lexicographically smaller than @xmath258 .
given a monomial , its tableau representation is the row - wise concatenation of the tableau forms of each of the variables appearing in the monomial .
let @xmath2 be the tree with edges @xmath74 , @xmath161 , @xmath75 , @xmath259 , @xmath260 , @xmath261 , and @xmath262 .
then the monomial @xmath263 has the standard form lexicographically ordered tableau : @xmath264.\ ] ] note that if a variable appears to the @xmath126-th power in a monomial , the representation for this variable is repeated as @xmath126 rows in the tableau .
when we write out general tableau , lower - case letters will always correspond to single characters ( possibly empty ) and upper case letters will always correspond to strings of characters ( also , possibly empty ) . for any tree @xmath2 ,
the conditional independence statements implied by @xmath2 generate @xmath3 .
in particular , @xmath3 is generated by linear polynomials @xmath88 and quadratic tetrad constraints .
first of all , we can ignore the linear polynomials as they always correspond to independence constraints and work modulo these linear constraints when working with the toric ideal @xmath3 .
in addition , every quadratic binomial of the form @xmath265 that belongs to @xmath3 is implied by a conditional independence statement .
this follows from proposition [ prop : treeind ] .
note that this holds even if the set @xmath266 does not have four elements .
thus , it suffices to show that @xmath3 modulo the linear constraints is generated by quadratic binomials . to show that @xmath3 is generated by quadratic binomials
, it suffices to show that any binomial in @xmath3 can be written as a polynomial linear combination of the quadratic binomials in @xmath3 .
this , in turn , will be achieved by showing that we can `` move '' from the tableau representation of one of the monomials to the other by making local changes that correspond to quadratic binomials . to show this last part
, we will define a sort of distance between two monomials and show that it is always possible to decrease this distance using these quadratic binomials/ moves .
this is a typical trick for dealing with toric ideals , illustrated , for instance , in @xcite . to this end
let @xmath19 be a binomial in @xmath3 . without loss of generality
, we may suppose the terms of @xmath19 have no common factors , because if @xmath267 then @xmath268 as well .
we will write @xmath19 as the difference of two tableaux , which are in standard form with their rows lexicographically ordered .
the first row in the two tableaux are different and they have a left - most place where they disagree
. we will show that we can always move this position further to the right .
eventually the top rows of the tableaux will agree and we can delete this row ( corresponding to the same variable ) and arrive at a polynomial of smaller degree . since @xmath268 , the treks associated to the top rows of the two tableaux must have the same top .
there are two cases to consider .
either the first disagreement is immediately after the top or not . in the first case
, this means that the binomial @xmath19 must have the form : @xmath269 - \left [ \begin{array}{l|l } \underline{a}bb & \underline{a}dd \\ & \end{array } \right].\ ] ] without loss of generality we may suppose that @xmath270 . since @xmath271 the string @xmath272 must appear somewhere on the right - hand monomial .
thus , @xmath19 must have the form : @xmath273 - \left [ \begin{array}{l|l } \underline{a}bb & \underline{a}dd \\
\underline{a}ee & \underline{a}cc ' \\
& \end{array } \right].\ ] ] if @xmath274 , we can apply the quadratic binomial @xmath275 - \left [ \begin{array}{l|l } \underline{a}bb & \underline{a}cc ' \\ \underline{a}ee & \underline{a}dd \\ \end{array } \right]\ ] ] to the second monomial to arrive at a monomial which has fewer disagreements with the left - hand tableau in the first row . on the other hand ,
if @xmath276 , we can not apply this move ( its application results in `` variables '' that do not belong to @xmath229 $ ] ) . keeping track of all the @xmath277 patterns that appear on the right - hand side , and the consequent @xmath277 patterns that appear on the left - hand side , we see that our binomial @xmath19 has the form @xmath278 - \left [ \begin{array}{l|l } \underline{a}bb & \underline{a}dd \\ \underline{a}dd ' & \underline{a}cc ' \\ \underline{a}d * & * \\ \vdots & \vdots \\ \underline{a}d * & * \end{array } \right].\ ] ] since there are the same number of @xmath277 s on both sides we see that there is at least one more @xmath279 on the right - hand side which has no @xmath126 s attached to it .
thus , omitting the excess @xmath277 s on both sides , our binomial @xmath19 contains : @xmath280 - \left [ \begin{array}{l|l } \underline{a}bb & \underline{a}dd \\
\underline{a}dd ' & \underline{a}cc ' \\ \underline{a}ee & \underline{a}gg \\ \end{array } \right].\ ] ] with @xmath274 or @xmath281 .
we can also assume that @xmath282 otherwise , we could apply a quadratic move as above .
thus we apply the quadratic binomials @xmath283 - \left [ \begin{array}{l|l } \underline{a}dd ' & \underline{a}gg\\ \underline{a}ee & \underline{a}cc'\\ \end{array } \right]\ ] ] and @xmath275 - \left [ \begin{array}{l|l } \underline{a}bb & \underline{a}cc ' \\ \underline{a}ee & \underline{a}dd \\
\end{array } \right]\ ] ] to reduce the number of disagreements in the first row .
this concludes the proof of the first case .
now suppose that the first disagreement does not occur immediately after the @xmath279 .
thus we may suppose that @xmath19 has the form : @xmath284 - \left [ \begin{array}{l|l } \underline{a}axdd & \underline{a}e \\ & \end{array } \right].\ ] ] note that it does not matter whether or not this disagreement appears on the left - hand or right - hand side of the tableaux .
since the string @xmath285 appears on right - hand monomial it must also appear somewhere on the left - hand monomial as well .
if @xmath30 is not the top in this occurrence , we can immediately apply a quadratic binomial to reduce the discrepancies in the first row .
so we may assume the @xmath19 has the form : @xmath286 - \left [ \begin{array}{l|l } \underline{a}axdd & \underline{a}e \\ & \\ & \\
\end{array } \right].\ ] ] if @xmath287 we can apply the quadratic binomial @xmath288 - \left [ \begin{array}{l|l } \underline{a}axdd ' & \underline{a}c \\
\underline{x}bb & \underline{x}gg \\ \end{array } \right]\ ] ] to the left - hand monomial to reduce the discrepancies in the first row .
so suppose that @xmath289 .
enumerating the @xmath290 pairs that can arise on the left and right hand monomials , we deduce , akin to our argument in the first case above , that @xmath19 has the form : @xmath291 - \left [ \begin{array}{l|l } \underline{a}axdd & \underline{a}e \\ & \\ & \\ & \\
\end{array } \right]\ ] ] where @xmath292 and @xmath95 are not equal to @xmath293 or @xmath126 .
then we can apply the two quadratic binomials : @xmath294 - \left [ \begin{array}{l|l } \underline{x}hh & \underline{x}bg \\
\underline{x}dd ' & \underline{x}kk \\ \end{array } \right]\ ] ] and @xmath295 - \left [ \begin{array}{l|l } \underline{a}axdd ' & \underline{a}c \\
\underline{x}bb & \underline{x}kk \\
\end{array } \right]\ ] ] to the left - hand monomial to produce a monomial with fewer discrepancies in the first row .
we have shown that no matter what type of discrepancy that can occur in the first row , we can always apply quadratic moves to produce fewer discrepancies .
this implies that @xmath3 is generated by quadrics . among the results in this section
were our proofs that @xmath3 has a squarefree initial ideal ( and hence @xmath229/i_t$ ] is normal and cohen - macaulay ) and that @xmath3 is generated by linear forms and quadrics .
it seems natural to wonder if there is a term order that realizes these two features simultaneously .
there exists a term order @xmath221 such that @xmath224 is generated by squarefree monomials of degree one and two .
this section and the next concern bayesian networks with hidden variables . a hidden or latent random
variable is one which we do not have direct access to .
these hidden variables might represent theoretical quantities that are directly unmeasurable ( e.g. a random variable representing intelligence ) , variables we can not have access to ( e.g. information about extinct species ) , or variables that have been censored ( e.g. for sensitive random variables in census data ) . if we are given a model over all the observed and hidden random variables , the partially observed model is the one obtained by marginalizing over the hidden random variables .
a number of interesting varieties arise in this hidden variable setting . for gaussian random variables ,
the marginalization is again gaussian , and the mean and covariance matrix are obtained by extracting the subvector and submatrix of the mean and covariance matrix corresponding to the observed random variables .
this immediately yields the following proposition .
[ prop : elim ] let @xmath296 $ ] be the vanishing ideal for a gaussian model .
let @xmath297 $ ] be a partition of the random variables into hidden and observed variables @xmath298 and @xmath299
. then @xmath300\ ] ] is the vanishing ideal for the partially observed model .
marginalization in the gaussian case corresponds to projection onto the subspace of pairs @xmath301 .
coordinate projection is equivalent to elimination @xcite .
in the case of a gaussian bayesian network , proposition [ prop : elim ] has a number of useful corollaries , of both a computational and theoretical nature .
first of all , it allows for the computation of the ideals defining a hidden variable model as an easy elimination step .
secondly , it can be used to explain the phenomenon we observed in example [ ex : verma ] , that the constraints defining a hidden variable model appeared as generators of the ideal of the fully observed model .
let @xmath302 be a partition of the nodes of the dag @xmath5 .
the hidden nodes @xmath298 are said to be _ upstream _ from the observed nodes @xmath299 in @xmath5 if there are no edges @xmath303 in @xmath5 with @xmath304 and @xmath305 .
if @xmath302 is an upstream partition of the nodes of @xmath5 , we introduce a grading on the ring @xmath306 $ ] which will , in turn , induce a grading on @xmath229 $ ] .
let @xmath307 for all @xmath308 , @xmath309 for all @xmath304 , @xmath310 if @xmath308 and @xmath304 , and @xmath311 otherwise .
suppose that @xmath312 $ ] is an upstream partition of the vertices of @xmath5 .
then each of the polynomials @xmath225 is homogeneous with respect to the upstream grading and @xmath313 thus , @xmath0 is homogeneous with respect to the induced grading on @xmath229 $ ] .
there are three cases to consider . if both @xmath314 , then every trek in @xmath59 has a top element in @xmath298 and no edges of the form @xmath315 . in this case
, the degree of each path is the vector @xmath316 . if @xmath317 and @xmath318 , every trek from @xmath61 to @xmath62 has a top in @xmath298 and exactly one edge of the form @xmath319 . thus , the degree of every monomial in @xmath320 is @xmath321 .
if both @xmath322 , then either each trek @xmath60 from @xmath61 to @xmath62 has a top in @xmath299 , or has a top in @xmath298 . in the first case there can be no edges in @xmath60 of the form @xmath319 , and in the second case there must be exactly two edges in @xmath60 of the form @xmath319 . in either case , the degree of the monomial corresponding to @xmath60 is @xmath323 .
note that the two dimensional grading we have described can be extended to an @xmath21 dimensional grading on the ring @xmath229 $ ] by considering all collections of upstream variables in @xmath5 simultaneously .
[ thm : upstream ] let @xmath302 be an upstream partition of the vertices of @xmath5 .
then every minimal generating set of @xmath0 that is homogeneous with respect to the upstream grading contains a minimal generating set of @xmath149 .
the set of indeterminates @xmath88 corresponding to the observed variables are precisely the variables whose degrees lie on the facet of the degree semigroup generated by the vector @xmath323 .
this implies that the subring generated by these indeterminates is a facial subring .
the upstream variables theorem is significant because any natural generating set of an ideal @xmath324 is homogeneous with respect to its largest homogeneous grading group .
for instance , every reduced grbner basis if @xmath0 will be homogeneous with respect to the upstream grading . for trees
, the upstream variables theorem immediately implies : [ thm : hidtree ] let @xmath2 be a rooted directed tree and @xmath299 consist of the leaves of @xmath2 .
then @xmath325 is generated by the quadratic tetrad constraints @xmath326 such that @xmath327 , and there is a choke point @xmath328 between @xmath329 and @xmath330 .
corollary [ thm : hidtree ] says that the ideal of a hidden tree model is generated by the tetrad constraints induced by the choke points in the tree .
sprites et al @xcite use these tetrad constraints as a tool for inferring dag models with hidden variables . given a sample covariance matrix ,
they test whether a collection of tetrad constraints is equal to zero . from the given tetrad constraints that are satisfied , together with the tetrad representation theorem
, they construct a dag that is consistent with these vanishing tetrads .
however , it is not clear from that work whether or not it is enough to consider only these tetrad constraints . indeed , as shown in @xcite , there are pairs of graphs with hidden nodes that have precisely the same set of tetrad constraints that do not yield the same family of covariance matrices .
theorem [ thm : hidtree ] can be seen as a mathematical justification of the tetrad procedure of spirtes , et al , in the case of hidden tree models , because it shows that the tetrad constraints are enough to distinguish between the covariance matrices coming from different trees .
in this section , we give families of examples to show how classical varieties from algebraic geometry arise in the study of gaussian bayesian networks .
in particular , we show how toric degenerations of the grassmannian , matrix schubert varieties , and secant varieties all arise as special cases of gaussian bayesian networks with hidden variables .
let @xmath331 be the grassmannian of @xmath210-planes in @xmath332 .
the grassmannian has the natural structure of an algebraic variety under the plcker embedding .
the ideal of the grassmannian is generated by the quadratic plcker relations : @xmath333.\ ] ] [ thm : grin ] let @xmath2 be a rooted directed binary tree with @xmath108 $ ] leaves and let @xmath299 be the set of leaves of @xmath2 . then there is a weight vector @xmath335 and a sign vector @xmath336 such that @xmath337 the proof idea is to show that the toric ideals @xmath325 have the same generators as the toric initial ideals of the grassmannian that have already been characterized in @xcite . without loss of generality
, we may suppose that the leaves of @xmath2 are labeled by @xmath108 $ ] , that the tree is drawn without edge crossings , and the leaves are labeled in increasing order from left to right .
these assumptions will allow us to ignore the sign vector @xmath338 in the proof .
the sign vector results from straightening the tree and permuting the columns in the steifel coordinates .
this results in sign changes in the plcker coordinates . in corollary
[ thm : hidtree ] , we saw that @xmath325 was generated by the quadratic relations @xmath326 such that there is a choke point in @xmath2 between @xmath329 and @xmath330 .
this is the same as saying that the induced subtree of @xmath2 on @xmath266 has the split @xmath340 .
these are precisely the generators of the toric initial ideals of the grassmannian @xmath341 identified in @xcite . in the preceding theorem , any weight vector @xmath342 that belongs to the relative interior of the cone of the tropical grassmannian corresponding to the tree @xmath2 will serve as the desired partial term order .
we refer to @xcite for background on the tropical grassmannian and toric degenerations of the grassmannian .
since and ideal and its initial ideals have the same hilbert function , we see catalan numbers emerging as degrees of bayesian networks yet again .
the fact that binary hidden tree models are toric degenerations of the grassmannian has potential use in phylogenetics .
namely , it suggests a family of new models , of the same dimension as the binary tree models , that could be used to interpolate between the various tree models .
that is , rather than choosing a weight vector in a full dimensional cone of the tropical grassmannian , we could choose a weight vector @xmath342 that sits inside of lower dimensional cone .
the varieties of the initial ideals @xmath345 then correspond to models that sit somewhere `` between '' models corresponding of the full dimensional trees of the maximal dimensional cones containing @xmath342 .
phylogenetic recovery algorithms could reference these in - between models to indicate some uncertainty about the relationships between a given collection of species or on a given branch of the tree .
these new models have the advantage that they have the same dimension as the tree models and so there is no need for dimension penalization in model selection . in this section
, we will describe how certain varieties called matrix schubert varieties arise as special cases of the varieties of hidden variable models for gaussian bayesian networks .
more precisely , the variety for the gaussian bayesian network will be the cone over one of these matrix schubert varieties . to do this
, we first need to recall some equivalent definitions of matrix schubert varieties .
let @xmath346 be a partial permutation matrix , which is an @xmath135 @xmath188 matrix with at most one @xmath146 in each row and column .
the matrix @xmath346 is in the affine space @xmath347 .
the borel group @xmath34 of upper triangular matrices acts on @xmath347 on the right by multiplication and on the left by multiplication by the transpose .
the matrix schubert variety @xmath351 , so we can identify its coordinate ring with a quotient of @xmath352 , j \in [ n']]$ ] . throughout this
section @xmath353 = \{1 ' , 2 ' , \ldots , n'\}$ ] , is a set of @xmath21 symbols that we use to distinguish from @xmath108 = \{1,2 , \ldots , n\}$ ] .
an equivalent definition of a matrix schubert variety comes as follows .
let @xmath354 be the index set of the ones in @xmath346 .
for each @xmath355 let @xmath356 be the variety of rank one matrices : @xmath357 then @xmath358 where the sum denotes the pointwise minkowski sum of the varieties . since @xmath356 are cones over projective varieties , this is the same as taking the join , defined in the next section .
let @xmath346 be the partial permutation matrix @xmath359 then @xmath348 consists of all @xmath360 matrices of rank @xmath361 and @xmath362 , [ 3']}| \right>$ ] . more generally , if @xmath346 is a partial permutation matrix of the form @xmath363 where @xmath364 is a @xmath365 identity matrix , then @xmath350 is the ideal of @xmath366 minors of a generic matrix .
the particular bayesian networks which yield the desired varieties come from taking certain partitions of the variables .
in particular , we assume that the observed variables come in two types labeled by @xmath108 = \{1,2 , \ldots , n\}$ ] and @xmath353 = \{1',2 ' , \ldots , n ' \}$ ] .
the hidden variables will be labeled by the set @xmath367 .
define the graph @xmath368 with vertex set @xmath369 \cup [ n ' ] \cup s(w)$ ] and edge set consisting of edges @xmath64 for all @xmath370 $ ] , @xmath371 for all @xmath372 $ ] , @xmath373 for all @xmath374 and @xmath375 and @xmath376 for all @xmath377 and @xmath378 .
[ thm : schubert ] the generators of the ideal @xmath350 defining the matrix schubert variety @xmath348 are the same as the generators of the ideal @xmath379 \cup [ n']}$ ] of the hidden variable bayesian network for the dag @xmath368 with observed variables @xmath108 \cup [ n']$ ] .
that is , @xmath380 \cup [ n ' ] ] = i_{g(w ) , [ n ] \cup [ n']}.\ ] ] the proof proceeds in a few steps .
first , we give a parametrization of a cone over the matrix schubert variety , whose ideal is naturally seen to be @xmath381 \cup [ n ' ] ] .$ ] then we describe a rational transformation @xmath382 on @xmath383 \cup [ n ' ] ] $ ] such that @xmath384 \cup [ n']}$ ] .
we then exploit that fact that this transformation is invertible and the elimination ideal @xmath379 \cup [ n ' ] } \cap { \mathbb{c}}[\sigma_{ij } \,\ , | \ , \ ,
i \in [ n ] , j \in [ n ' ] ] $ ] is fixed to deduce the desired equality .
first of all , we give our parametrization of the ideal @xmath350 . to do this
, we need to carefully identify all parameters involved in the representation .
first of all , we split the indeterminates in the ring @xmath385 \cup [ n']]$ ] into three classes of indeterminates : those with @xmath386 $ ] , those with @xmath387 $ ] , and those with @xmath388 $ ] and @xmath389 $ ] .
then we define a parametrization @xmath390 which is determined as follows : @xmath391 \rightarrow { \mathbb{c}}[\sigma_{ij } \ , \ , | \ , \ , i , j \in [ n ] \cup [ n']\ ] ] @xmath392 \\ \gamma_{ij } & \mbox { if } i , j \in [ n ' ] \\
{ ( k , l ) \in s(w ) : k \leq
i , l \leq j } a_{(k , l ) } \lambda_{(k , l),i } \lambda_{(k , l),j } & \mbox { if } i \in [ n ] , j \in [ n ' ] \end{array } \right.\ ] ] let @xmath393 .
since the @xmath338 , @xmath394 , @xmath217 , and @xmath279 parameters are all algebraically independent , we deduce that in @xmath395 , there will be no generators that involve combinations of the three types of indeterminates in @xmath396 \cup [ n']]$ ] .
furthermore , restricting to the first two types of indeterminates , there will not be any nontrivial relations involving these types of indeterminates .
thus , to determine @xmath395 , it suffices to restrict to the ideal among the indeterminates of the form @xmath88 such that @xmath388 $ ] and @xmath389 $ ] .
however , considering the parametrization in this case , we see that this is precisely the parametrization of the ideal @xmath350 , given as the minkowski sum of rank one matrices . thus , @xmath397 .
now we will define a map from @xmath398 \rightarrow { \mathbb{c}}[\sigma_{ij}]$ ] which sends @xmath395 to another ideal , closely related to @xmath379 \cup [ n ' ] } $ ] . to define this map ,
first , we use the fact that from the submatrix @xmath399 , [ n]}$ ] we can recover the @xmath42 and @xmath54 parameters associated to @xmath108 $ ] , when only considering the complete subgraph associated to graph @xmath400}$ ] ( and ignoring the treks that involve the vertices @xmath401 ) .
this follows because these parameters are identifiable by proposition [ prop : dim ] .
a similar fact holds when restricting to the subgraph @xmath402}$ ] .
the ideal @xmath395 we have defined thus far can be considered as the vanishing ideal of a parametrization which gives the complete graph parametrization for @xmath400}$ ] and @xmath402}$ ] and a parameterization of the matrix schubert variety @xmath348 on the @xmath88 with @xmath388 $ ] and @xmath403 $ ] .
so we can rationally recover the @xmath217 and @xmath279 parameters associated to the subgraphs @xmath400}$ ] and @xmath402}$ ] . for each @xmath404 pair in
@xmath108 $ ] or in @xmath353 $ ] , define the partial trek polynomial @xmath405 we fit these into two upper triangular matrices @xmath406 and @xmath407 where @xmath408 if @xmath409 with @xmath410 $ ] , @xmath411 and @xmath412 otherwise , with a similar definition for @xmath407 with @xmath108 $ ] replaced by @xmath353 $ ] .
now we are ready to define our map .
let @xmath382 be the rational map @xmath413
\to { \mathbb{c}}[\sigma]$ ] which leaves @xmath88 fixed if @xmath386 $ ] or @xmath387 $ ] , and maps @xmath88 with @xmath388 $ ] and @xmath389 $ ] by sending this is actually a rational map , because the @xmath42 that appear in the formula for @xmath415 are expressed as rational functions in terms of the @xmath88 by the rational parameter recovery formula of proposition [ prop : dim ] . since this map transforms @xmath399 , [ n']}$ ] by multiplying on the left and right but lower and upper triangular matrices , this leaves the ideal @xmath416 , j \in [ n']]$ ] fixed .
thus @xmath417 . on the other hand @xmath382
is invertible on @xmath395 so @xmath418 .
if we think about the formulas for the image @xmath419 , we see that the formulas for @xmath88 with @xmath388 $ ] and @xmath389 $ ] in terms of parameters are the correct formulas which we would see coming from the parametrization @xmath420 . on the other hand , the formulas for @xmath88 with
@xmath386 $ ] or @xmath387 $ ] are the formulas for the restricted graph @xmath421}$ ] and @xmath422}$ ] , respectively .
since every trek contained in @xmath421}$ ] or @xmath422}$ ] is a trek in @xmath368 , we see that the current parametrization of @xmath395 is only `` almost correct '' , in that it is only missing terms corresponding to treks that go outside of @xmath400}$ ] or @xmath402}$ ] .
denote this map by @xmath423 , and let @xmath420 be the actual parametrizing map of the model .
thus , we have , for each @xmath88 with @xmath386 $ ] or @xmath387 $ ] , @xmath424 , where @xmath425 is a polynomial remainder term that does not contain any @xmath54 with @xmath388 \cup [ n']$ ] , when @xmath426 $ ] or @xmath387 $ ] , and @xmath427 otherwise . on the other hand , every term of @xmath428 will involve exactly one @xmath54 with @xmath388 \cup [ n']$ ] , when @xmath386 $ ] or @xmath429 $ ] .
now we define a weight ordering @xmath221 on the ring @xmath430 $ ] that gives @xmath431 if @xmath432 \cup [ n']$ ] and @xmath433 otherwise and @xmath434 for all @xmath96 .
then , the largest degree term of @xmath435 with respect to this weight ordering is @xmath436 .
since highest weight terms must all cancel with each other , we see that @xmath437 \cup [ n']}$ ] , implies that @xmath438 .
thus , we deduce that @xmath379 \cup [ n ' ] } \subseteq j_w$ ] . on the other hand , @xmath439 \cup [ n ] ' } \cap { \mathbb{c}}[\sigma_{ij } \ , \ , | \ , \ , i \in [ n ] , j \in [ n ' ] ] = j_w \cap { \mathbb{c}}[\sigma_{ij } \ , \,| \ , \ ,
i \in [ n ] , j \in [ n ' ] ] \ ] ] and since the generators of @xmath440 , j \in [ n ' ] ] $ ] generate @xmath395 , we deduce that @xmath441 \cup [ n']}$ ] which completes the proof .
the significance of theorem [ thm : schubert ] comes from the work of knutson and miller @xcite .
they gave a complete description of antidiagonal grbner bases for the ideals @xmath350 .
indeed , these ideals are generated by certain subdeterminants of the matrix @xmath399,[n']}.$ ] these determinants can be interpretted combinatorially in terms of the graph @xmath368 .
@xcite the ideal @xmath350 defining the matrix schubert variety is generated by the conditional independence statements implied by the dag @xmath368 .
in particular , @xmath442 , b \subset [ n ' ] , c \subset s(w ) , \mbox { and } c \mbox { $ d$-separates } a \mbox { from } b \right>.\ ] ] in this section , we will show how joins and secant varieties arise as special cases of gaussian bayesian networks in the hidden variable case .
this , in turn , implies that techniques that have been developed for studying defining equations of joins and secant varieties ( e.g. @xcite ) might be useful for studying the equations defining these hidden variable models .
given two ideals @xmath324 and @xmath443 in a polynomial ring @xmath444 = { \mathbb{k}}[x_1 , \ldots , x_m]$ ] , their _ join _ is the new ideal @xmath445 \right > \right ) \bigcap { \mathbb{c}}[{\mathbf{x}}]\ ] ] where @xmath446 is the ideal obtained from @xmath324 by plugging in the variables @xmath447 for @xmath448 .
the secant ideal is the iterated join : @xmath449 with @xmath450 copies of @xmath324 .
if @xmath324 and @xmath443 are homogeneous radical ideals over an algebraically closed field , the join ideal @xmath451 is the vanishing ideal of the join variety which is defined geometrically by the rule @xmath452 where @xmath453 denotes the line spanned by @xmath279 and @xmath293 and the bar denotes the zariski closure .
suppose further that @xmath324 and @xmath443 are the vanishing ideals of parametrizations ; that is there are @xmath382 and @xmath454 such that @xmath455 \rightarrow { \mathbb{c}}[\theta ] \mbox { and } \psi : { \mathbb{c}}[{\mathbf{x } } ] \rightarrow { \mathbb{c}}[\eta]\ ] ] and @xmath456 and @xmath457 .
then @xmath451 is the kernel of the map @xmath458 \rightarrow { \mathbb{c}}[\theta , \eta]\ ] ] @xmath459 let @xmath5 be a dag and suppose that the vertices of @xmath5 are partitioned into @xmath463 where both @xmath464 and @xmath465 are hidden sets of variables .
suppose further that there are no edges of the form @xmath466 such that @xmath467 or edges of the form @xmath468 or @xmath469 with @xmath470 and @xmath471
. then @xmath472 the parametrization of the hidden variable model only involves the @xmath88 such that @xmath322 .
first , we restrict to the case where @xmath227 . since there are no edges between observed variables and no edges between @xmath464 and @xmath465 , every trek from @xmath61 to @xmath62 involves only edges in @xmath473 or only edges in @xmath474 .
this means that @xmath475 and these summands are in non - overlapping sets of indeterminates .
thus , by the comments preceding the proposition , the ideal only in the @xmath88 with @xmath476 is clearly a join .
however , the structure of this hidden variable model implies that there are no nontrivial relations that involve the diagonal elements @xmath477 with @xmath478 .
this implies that @xmath149 is a join .
let @xmath479 be the directed complete bipartite graph with bipartition @xmath480 $ ] and @xmath481 $ ] such that @xmath482 for all @xmath483 $ ] and @xmath484 $ ] .
then @xmath479 satisfies the conditions of the theorem recursively up to @xmath485 copies , and we see that : @xmath486 this particular hidden variable gaussian bayesian network is known as the factor analysis model .
this realization of the factor analysis model as a secant variety was studied extensively in @xcite . since in each case ,
the two subgraphs @xmath473 and @xmath487 are isomorphic , the ideals are secant ideals of the hidden tree models @xmath325 for the appropriate underlying trees . in both cases , the ideal @xmath488 is a principal ideal , generated by a single cubic . in the first case , the ideal is the determinantal ideal @xmath489 . in the second case ,
the ideal is generated by an eight term cubic in both of the cubic cases in the previous example , the ideals under questions were secant ideals of toric ideals that were initial ideals of the grassmann - plcker ideal , as we saw in theorem [ thm : grin ] .
note also that the secant ideals @xmath491 are , in fact , the initial terms of the @xmath492 pfaffian with respect to appropriate weight vectors .
we conjecture that this pattern holds in general .
let @xmath2 be a binary tree with @xmath21 leaves and @xmath299 the set of leaves of @xmath2 .
let @xmath334 be the grassmann - plucker ideal , let @xmath342 be a weight vector and @xmath338 a sign vector so that @xmath493 as in theorem [ thm : grin ] . then for each @xmath450 @xmath494 | conditional independence models in the gaussian case are algebraic varieties in the cone of positive definite covariance matrices .
we study these varieties in the case of bayesian networks , with a view towards generalizing the recursive factorization theorem to situations with hidden variables . in the case when the underlying graph is a tree , we show that the vanishing ideal of the model is generated by the conditional independence statements implied by graph . we also show that the ideal of any bayesian network is homogeneous with respect to a multigrading induced by a collection of upstream random variables .
this has a number of important consequences for hidden variable models .
finally , we relate the ideals of bayesian networks to a number of classical constructions in algebraic geometry including toric degenerations of the grassmannian , matrix schubert varieties , and secant varieties . |
let @xmath1 be a two sided standard brownian motion with drift @xmath2 .
given @xmath3 we say that @xmath1 admits an _ @xmath0minimum _ at @xmath4 , and that @xmath5 is a point of @xmath0minimum , if there exist @xmath6 such that @xmath7 for all @xmath8 $ ] , @xmath9 and @xmath10 .
similarly , we say that @xmath1 admits an _ @xmath0maximum _ at @xmath11 , and that @xmath5 is a point of @xmath0maximum , if there exist @xmath6 such that @xmath12 for all @xmath8 $ ] , @xmath13 and @xmath14 .
we say that @xmath1 admits an _ @xmath0extremum _ at @xmath4 , and that @xmath5 is a point of @xmath0extremum , if @xmath5 is a point of @xmath0minimum or a point of @xmath0maximum .
finally , the truncated trajectory @xmath1 going from an @xmath0minimum to an @xmath0maximum will be called upward @xmath0slope , while the truncated trajectory @xmath1 going from an @xmath0maximum to an @xmath0minimum will be called downward @xmath0slope .
our first object of investigation is the statistics of @xmath0slopes .
the non drifted case @xmath15 has been studied in @xcite .
here we assume @xmath16 and show ( see theorem [ napoleone ] ) that the statistics of @xmath0slopes is well described by a stationary alternating marked simple point process on @xmath17 whose points are the points of @xmath0extrema of the brownian motion , and each point @xmath5 is marked by the @xmath0slope going from @xmath5 to the subsequent point of @xmath0extremum .
we will show that the @xmath0slopes are independent and specify the laws @xmath18 , @xmath19 of upward @xmath0slopes and downward @xmath0slopes non covering the origin , respectively .
the @xmath0slope covering the origin shows a different distribution that can be derived by means of the palm
khinchin theory @xcite , @xcite .
our proof is based both on fluctuation theory for lvy processes , and on the theory of marked simple point processes .
the part of fluctuation theory follows strictly the scheme of @xcite and can be generalized to spectrally one sided lvy processes , i.e. real valued random processes with stationary independent increments and with no positive jumps or with no negative jumps @xcite[chapter vii ] . in fact , some of the identities of lemma [ davide ] and proposition [ calcoli ] below have already been obtained with more sophisticated methods for general spectrally one sided lvy processes ( see @xcite , @xcite , @xcite and references therein ) . on the other hand , the description of the @xmath0slopes as a stationary alternating marked simple point process allows to use the very powerful palm khinchin theory , which extends renewal theory and leads to a complete description of the @xmath0slope covering the origin .
this analysis can be easily extended to more general lvy processes , as the ones treated in @xcite .
as discussed in section [ rengroup ] , our results concerning the statistics of @xmath0extrema of drifted brownian motion correspond to the ones obtained in @xcite via a non rigorous real space renormalization group method applied to sinai random walk with a vanishing bias .
in addition of a rigorous derivation , we are able here to describe also the statistics of the @xmath0slopes , lacking in @xcite . in section [ autunno ] ( see theorem [ bonaparte ] ) ,
we analyze the behavior of the drifted brownian motion around its @xmath0extrema .
while in the non drifted case a generic @xmath0slope non covering the origin behaves in proximity of its extremes as a 3dimensional bessel process , in the drifted case it behaves as a process with a cothangent drift , satisfying the sde @xmath20 where @xmath21 is an independent standard brownian motion and the sign in the r.h.s .
depends on the kind of @xmath0slope ( downward or upward ) and on the kind of @xmath0extrema ( @xmath0minimum or @xmath0maximum ) .
in addition , we show that the process ( [ silurino ] ) is simply the brownian motion on @xmath22 , starting at the origin , with drift @xmath23 , doob conditioned to hit @xmath24 before @xmath25 .
the interest in the statistics of @xmath0slopes and their behavior near to the extremes comes also from the fact that , considering the diffusion in a drifted brownian potential , the piecewise linear path obtained by connecting the @xmath0extrema of the brownian potential is the effective potential for the diffusion at large times @xcite .
given @xmath26 we denote by @xmath27 the law on @xmath28 of the standard two sided brownian motion @xmath1 with drift @xmath2 having value @xmath5 at time zero , i.e. @xmath29 where @xmath30 denotes the two sided brownian motion s.t . @xmath31 has expectation zero and variance @xmath32 .
we denote the expectation w.r.t . @xmath33 by @xmath34 . if @xmath35 we simply write @xmath36 , @xmath37 .
recall the definitions of @xmath0maximum , @xmath0minimum and @xmath0extremum given in the introduction .
it is simple to verify that @xmath38a.s .
the set of points of @xmath0extrema is locally finite , unbounded from below and from above , and that points of @xmath0minima alternate with points of @xmath0maxima .
the @xmath0slope between two consecutive points of @xmath0extrema @xmath39 and @xmath40 is defined as the truncated trajectory @xmath41\bigr)$ ] .
we call it an _ upward _ slope if @xmath39 is a point of @xmath0minimum ( and consequently @xmath40 is a point of @xmath0maximum ) , otherwise we call it _ downward _ slope . the length @xmath42 and the height @xmath43 of the slope @xmath44 are defined as @xmath45 and @xmath46 , respectively . moreover , to the slope @xmath44 we associate the translated path @xmath47\bigr)$ ] . with some abuse of notation ( as in the introduction )
we call also @xmath48 the @xmath0slope between the points of @xmath0extrema @xmath39 and @xmath40
. when the context can cause some ambiguity , we will explicitly distinguish between the @xmath0slope @xmath44 and the translated path @xmath48 .
finally , we introduce the following notation : given @xmath49 , the constant @xmath50 is defined as @xmath51 .
given a two sided brownian @xmath1 with law @xmath52 we define the random variables @xmath53 as follows ( see figure [ coniglio2 ] ) : @xmath54 note that @xmath55a.s .
there exists a unique time @xmath56 $ ] with @xmath57 , which by definition coincides with @xmath58 .
[ l][l]@xmath59 [ l][l]@xmath60 [ l][l]@xmath61 [ l][l]@xmath0 , @xmath62 , @xmath63.,title="fig:",width=264 ] our analysis of the statistics of @xmath0slopes for the drifted brownian motion is based on the following lemma which extends to the drifted case the lemma in section 1 of @xcite : [ davide ] let @xmath64 , @xmath65 and @xmath66 . under @xmath55 ,
the two trajectories @xmath67 are independent ; in particular @xmath68 and @xmath69 are independent .
furthermore @xmath70 is exponentially distributed with mean @xmath71 and @xmath72= \exp \left\ { -x\left [ \sqrt{2\hat\a } \coth \left(\sqrt{2\hat \a } h\right ) -\mu \coth(\mu h ) \right ] \right\ } .\label{lapla1}\ ] ] in particular , @xmath73 is finite if and only if @xmath74 if ( [ mom_finito ] ) is fulfilled , then @xmath75 finally , it holds @xmath76 the proof of the above lemma is based on excursion theory and indentities concerning hitting times of the drifted brownian motion .
it will be given in section [ pierpa ] .
we define the path space @xmath79 as the set @xmath80)\,.\ ] ] given @xmath81 , we define @xmath42 as the nonnegative number such that @xmath82 $ ] and we define the path @xmath83 as @xmath84 then the space @xmath79 is a polish space endowed of the metric @xmath85 defined as @xmath86 on @xmath79 we define the borel probability measures @xmath18 , @xmath87 as follows .
let @xmath88 be independent brownian motions with law @xmath52 .
recall the definition ( [ esameinf ] ) of @xmath89 and define @xmath90 as ( see figure [ pezzobis ] ) : @xmath91 then @xmath18 is the law of the path @xmath44 , with @xmath92 , defined as @xmath93\ , , \\ b'_{t-(\t-\s)}+h \ , , &
\text { if } t \in [ \t-\s ,
\t-\s + \s']\ , , \end{cases}\ ] ] while @xmath87 is the law of the path @xmath44 , with @xmath94 , defined as @xmath95\,,\\ b_{t-(\t'-\s ' ) } - h \ , , & \text { if } t \in [ \t'-\s ' , \t'-\s'+\s]\ , .
\end{cases}\ ] ] note that @xmath87 equals the law of the path @xmath96 if @xmath97 is chosen with law @xmath98 .
[ l][l]@xmath99 [ l][l]@xmath100 [ l][l]@xmath101 [ l][l]@xmath0 , @xmath102 , @xmath103.,title="fig:",width=264 ] we introduce two disjoint subsets @xmath104 and @xmath105 of @xmath79 : @xmath106\}=0 \ , , \;\ ; \g_{\ell(\g ) } = \max \{\g_t\,:\ , t \in [ 0,\ell(\g)]\}\geq h \right\}\ , , \\ & { { \mathcal w}}_- = \left\ { \g \in{{\mathcal w}}\,:\ , \g_0 = \max \{\g_t\,:\ , t \in
[ 0,\ell(\g)]\}=0 \ , , \;\ ; \g_{\ell(\g ) } = \min \{\g_t\,:\ , t \in [ 0,\ell(\g)]\}\leq - h \right\}\,.\end{aligned}\ ] ] then the probability measure @xmath107 is concentrated on @xmath108 .
below we will prove that , given the two sided bm with law @xmath52 , @xmath77 is the law of the generic upward @xmath0slope not covering the origin , while @xmath109 is the law of the generic downward @xmath0slope not covering the origin .
we collect in what follows some results derived from lemma [ davide ] and straightforward computations , which will be useful in what follows : [ calcoli ] fix @xmath16 .
let @xmath110 be the random vector distributed as @xmath111 where @xmath44 is chosen with law @xmath77 and let @xmath112 be the random vector distributed as @xmath113 where @xmath44 is chosen with law @xmath109 .
then @xmath114 , @xmath115 are exponential variables of mean @xmath116 fix @xmath117 such that @xmath118 . then for all @xmath66 @xmath119 in particular , the expectation @xmath120 is finite if and only if @xmath121 if ( [ mom_condbis ] ) is fulfilled , then @xmath122 hence @xmath123 consider @xmath124 , where @xmath125 and @xmath126 are chosen independently .
then , @xmath127 given @xmath49 , @xmath128 is finite if and only if @xmath129 if ( [ mom_condtris ] ) is fulfilled , then @xmath130 due to the identity @xmath131 by straightforward computations one can check that for @xmath132 condition ( [ mom_condtris ] ) is fulfilled if and only if @xmath133 , where @xmath134 is the only positive solution of the equation @xmath135 . if @xmath136 then condition ( [ mom_condtris ] ) is fulfilled if and only if @xmath137 .
we denote by @xmath139 the space of sequences @xmath140 such that 1 ) @xmath141 , 2 ) @xmath142 , 3 ) @xmath143 and 4 ) @xmath144 . in what follows ,
@xmath145 will be often identified with the counting measure @xmath146 on @xmath147 .
@xmath139 is a measurable space with @xmath58-algebra of measurable sets generated by @xmath148 one can characterize the above @xmath58algebra as follows .
consider the space @xmath149 as a measurable space with @xmath58algebra of measurable sets given by the borel subsets associated to the product topology .
call @xmath150 the subset of @xmath151 given by the elements where the first entry is not larger than the entry with index @xmath25 of the factor space @xmath152 .
then by the same arguments leading to @xcite[proposition 7.1.x ] one can prove that the map @xmath153 is bijective and both ways measurable .
we note that the introduction of @xmath150 is due to the constrain @xmath154 .
let us define @xmath155 as the law of the sequences @xmath156 such that * @xmath157 are independent random paths , * @xmath158 are i.i.d .
random paths with law @xmath159 , * @xmath160 are i.i.d .
random paths with law @xmath161 , * @xmath162 , * @xmath163 . note that @xmath155 is concentrated on the measurable subset @xmath164 defined as @xmath165 finally we consider the convex combination @xmath166 let @xmath167 and , for all @xmath168 , let @xmath169 be the maps defined as @xmath170 we stress that the above translation map @xmath171 coincides with the map @xmath172 of @xcite and with the map @xmath173 of @xcite .
a probability measure @xmath174 on @xmath139 is called stationary if @xmath175 for all @xmath176 and all @xmath177 measurable , while it is called @xmath178invariant if @xmath179 for all @xmath177 measurable .
note that @xmath180 is @xmath178invariant . moreover , due to ( [ mangobis ] ) in lemma [ davide ] , @xmath181 hence , due to the palm
khinchin theory ( see theorem 1.3.1 and formula ( 1.2.15 ) in @xcite , and theorem 12.3.ii in @xcite ) there exists a unique stationary measure @xmath138 on @xmath139 such that @xmath182\,,\\ & = \frac { \mu^2 } { \sinh ^2 ( \mu h ) } { { \mathbf e}}_{{{\mathcal p}}^\mu _ 0 } \left [ \int _ 0 ^{x_1 } \chi \left ( t_{-t } \left ( \{x_i,\g_i\}_{i\in { { \mathbb z}}}\right ) \in a\ , \right ) dt \right]\ , , \end{split}\ ] ] where @xmath183 denotes the characteristic function .
we simply say that _
@xmath184 is the law of the stationary alternating marked simple point process on @xmath17 with alternating mark laws given by @xmath185_. the probability measure @xmath186 is the so called _ palm distribution _ associated to @xmath138 .
one can write @xmath187 from ( [ spaghetti ] ) we obtain that @xmath188 } { { { \mathbf e}}(\ell ) } = \frac{{{\mathbf e}}_{p^\mu _
\pm } \bigl ( \ell ( \g ) \bigr ) } { { { \mathbf e}}(\ell ) } = \frac { { { \mathbf e}}(\ell_\pm ) } { { { \mathbf e}}(\ell ) } \,.\ ] ] hence , by ( [ zeta1 ] ) and ( [ zeta2 ] ) of lemma [ davide ] , we can conclude that @xmath189 in order to describe the conditional probability measure @xmath190 we first observe that @xmath191 and @xmath192 univocally determined the set @xmath193 .
moreover from ( [ spaghetti ] ) we derive that , given borel subsets @xmath194 , @xmath195 for @xmath196 , it holds @xmath197 \bigl[\prod _ { \substack{-m\leq j \leq
n\\ j\not = 0 , \text { even } } } p ^\mu _ \pm ( b_j ) \bigr ] /{{\mathbf e}}(\ell _ \pm ) \,.\end{gathered}\ ] ] this identity implies that under @xmath198 the random paths @xmath199 , @xmath200 , are independent , the paths @xmath201 have common law @xmath202 while the paths @xmath203 have common law @xmath204 and that the path @xmath205 has law @xmath206 finally , we claim that under @xmath207 @xmath191 has probability density function on @xmath22 given by @xmath208 where @xmath209 indeed , given @xmath210 , due to ( [ spaghetti ] ) and ( [ parto1 ] ) we can write @xmath211 in particular , we point out that due to ( [ vito1 ] ) and ( [ parto1 ] ) under @xmath138 the random variable @xmath191 has probability density on @xmath22 given by @xmath212 .
we can finally prove the main theorem of this section : [ napoleone ] let @xmath1 be the brownian motion with law @xmath52 and let @xmath213 be the sequence of points of @xmath0extrema of @xmath1 , increasingly ordered , with @xmath214 .
for each @xmath215 define the @xmath0slope @xmath199 as @xmath216 then the marked point process @xmath217has law @xmath184 . in order to prove the above result we need to further elucidate the relation between @xmath138 and its palm distribution @xmath186 , benefiting of the palm khinchin theory . to this aim
we need the ergodicity of @xmath180 : [ saltelli ] the probability measure @xmath186 is @xmath178ergodic , i.e. if @xmath177 is a measurable set such that @xmath218 then @xmath219 .
let us suppose that @xmath177 is a borel set with @xmath220 . due to the characterization of the @xmath58algebra of measurable sets in @xmath139 given by the bijective and both ways measurable map ( [ pmassa ] ) and
since @xmath186 is concentrated on @xmath164 , for each @xmath221 we can find a measurable set @xmath222 and an integer @xmath223 such that @xmath224 depends only on the random variables @xmath225 with @xmath226 and @xmath227 . since @xmath228
we can conclude that @xmath229 since @xmath186 is @xmath178invariant and @xmath230 , we get for each positive integer @xmath231 that @xmath232 and therefore @xmath233 this implies that @xmath234 let us now write @xmath235 for any quantity which goes to @xmath25 as @xmath236 .
we note that for @xmath231 large enough and even it holds @xmath237 while for @xmath231 large enough and odd it holds @xmath238 let @xmath239 and @xmath240 . due to ( [ vegetale1]), ... ,([vegetale5 ] )
we can conclude that @xmath241 = \\ & \begin{cases } a+b - ( a^2+b^2 ) + o(1 ) & \text { if $ n$ is even and large}\ , , \\ a+b - 2 ab+o(1 ) & \text { if $ n$ is odd and large}\ , . \end{cases } \end{split}\ ] ] we conclude that @xmath242 hence , by subtraction , @xmath243 , i.e. @xmath244 .
it is simple to check that there are only two possible cases : ( i ) @xmath245 and @xmath246 , ( ii ) @xmath247 and @xmath248 .
since @xmath249 , in the first case we get @xmath250 while in the latter @xmath251 . due to the arbitrary of @xmath252 we conclude that @xmath253 .
let us now introduce the space @xmath254 given by the counting measures @xmath255 on @xmath147 , where @xmath256 and the set @xmath257 has finite intersection with sets of the form @xmath258\times { { \mathcal w}}$ ] .
note that if @xmath259 for all @xmath260 we can identity @xmath145 with its support .
hence we can think of @xmath139 as a subset of @xmath254 .
as discussed in @xcite[section 1.1.5 ] one defines on @xmath254 a suitable metric @xmath261 such that i ) @xmath254 is a polish space , ii ) @xmath139 is a borel subset of @xmath254 and iii ) the @xmath60algebra of borel subsets of @xmath254 is generated by the sets @xmath262 in particular , the @xmath58algebra of measurable subsets of @xmath139 introduced above coincides with the @xmath58algebra of borel subsets of @xmath139 and we can think of @xmath263 as borel probability measures on @xmath254 concentrated on @xmath139 . due to lemma [ saltelli ] and @xcite[theorem 1.3.13 ] we get [ balocco ] given @xmath16 , the probability measure @xmath264 weakly converges to @xmath265 as @xmath266 , i.e. for any continuous bounded function @xmath267 it holds @xmath268 let @xmath269 be the intensity measure associated to @xmath270 , i.e. @xmath269 is the probability measure on @xmath147 such that @xmath271 then due to @xcite[theorem 1.1.16 ] , the weak convergence of @xmath272 to @xmath138 stated in corollary [ balocco ] is equivalent to the following fact : given a finite family @xmath273 of disjoint sets @xmath274 satisfying @xmath275 where @xmath276 denotes the boundary of @xmath277 , it holds @xmath278 for all @xmath279 .
we have now the main tools in order to prove theorem [ napoleone ] .
[ l][l]@xmath280 [ l][l]@xmath281 [ l][l]@xmath282 [ l][l]@xmath283 [ l][l]@xmath284 [ l][l]@xmath285 [ l][l]@xmath0 [ l][l]@xmath286 [ l][l]@xmath287 [ l][l]@xmath288 [ l][l]@xmath289 , @xmath290.,title="fig:",width=453 ] _ proof of theorem [ napoleone]_. let @xmath1 be a two sided brownian motion with law @xmath52 . set @xmath291 and define the random variables @xmath292 inductively on @xmath293 as follows ( see figure [ tagliare ] ) : + for @xmath231 even set @xmath294 for @xmath231 odd set @xmath295 note that by construction @xmath296 is a point of @xmath0maximum for @xmath231 odd , while @xmath296 is a point of @xmath0minimum for @xmath297 even .
moreover , due to lemma [ davide ] and the strong markov property of brownian motion at the markov times @xmath298 , the slopes @xmath299 are independent , having law @xmath19 if @xmath231 is odd and law @xmath18 if @xmath231 is even .
in what follows we will use the independent random variables @xmath300 with the following laws : @xmath277 is distributed as @xmath301 , i.e. @xmath277 is distributed as @xmath302 where @xmath303 are independent copies of @xmath304 defined in ( [ esameinf ] ) , ( [ esameinfbis ] ) respectively .
@xmath305 is distributed as @xmath126 , i.e. as the length @xmath306 of the random path @xmath44 chosen with law @xmath307 .
given a realization of the two sided brownian motion @xmath1 with law @xmath308 , let @xmath309 be the associated marked simple point process defined in ( [ violino ] ) , while let @xmath145 denote a generic element of @xmath139 . fix a finite family @xmath273 of disjoint sets @xmath310 , with @xmath311 and @xmath312 borel , and consider the event @xmath313 for given @xmath314 .
finally , set @xmath315 due to the discussion after corollary [ balocco ] , we only need to show that @xmath316 to this aim , we set @xmath317 and restrict in what follows to the case @xmath318 . then our initial considerations imply that @xmath319 and therefore @xmath320 in what follows , we will frequently apply the above argument in order to get estimates from above without explicit mention .
let us consider the probability measure @xmath321 . by definition @xmath322
while @xmath323 hence we can estimate @xmath324 due to ( [ marzo ] ) , ( [ marzo3 ] ) and corollary [ balocco ] , in order to prove the theorem it is enough to show that @xmath325 we will derive from the local central limit theorem that , given a generic positive random variable @xmath326 having a ( bounded ) probability density and bounded third moment , it holds @xmath327 due to lemma [ uffina ] below , this result allows to derive ( [ marzo4 ] ) . in order to prove ( [ marzo5 ] )
define @xmath328 as the sum of @xmath231 independent copies of the random variable @xmath329 introduced in proposition [ calcoli ]
. moreover , let @xmath328 and @xmath326 be independent .
then @xmath330 hence in order to prove ( [ marzo5 ] ) it is enough to prove that @xmath331 in general , given a r.w .
@xmath332 we write @xmath333 for its probability density ( if it exists ) .
moreover , we denote by @xmath334 the norm in @xmath335 . setting @xmath336
, we can bound @xmath337 since @xmath338 by a change of variables we can conclude that @xmath339 let @xmath340 be the gaussian distribution with variance @xmath341 . due to ( [ brahms1 ] ) and ( [ brahms2 ] ) in order to conclude we only need to show that @xmath342 where @xmath343 denotes the convolution of @xmath344 and @xmath345 .
we note that ( [ centrale1 ] ) follows from ( [ centrale3 ] ) since @xmath346 and for generic functions @xmath347 in @xmath348 it holds @xmath349 . the limit ( [ centrale2 ] ) follows from straightforward computations while ( [ centrale3 ] ) corresponds to the @xmath350local central limit theorem for densities since @xmath326 has bounded probability density and bounded third moment ( see @xcite[page 193 ] or theorem 18 in @xcite[chapter vii ] where the boundedness of @xmath351 is required ) .
[ uffina ] the random variables @xmath277 and @xmath305 in the proof of theorem [ napoleone ] have bounded continuous probability densities .
moreover , they have finite n th moment for all @xmath352 . due to theorem 3 in @xcite[section xv.3 ] in order to prove that @xmath277 and @xmath305 have bounded continuous probability densities it is enough that the associated fourier transforms are in @xmath353 . since @xmath354 and @xmath355 where the random variables @xmath69 , @xmath58 and @xmath356 are independent , it is enough to prove that the fourier transform of @xmath69 is integrable . to this aim
we observe that due to lemma [ davide ] the expectation @xmath357 is finite for @xmath358 .
this implies that the complex laplace transform @xmath359 is well - defined ( i.e. the integrand is integrable ) and analytic on the complex halfplane @xmath360 .
indeed , integrability is stated in section 2.2 of @xcite and analyticity is stated in satz 1 ( proposition 1 ) in section 3.2 of @xcite .
we point out that in @xcite the author considers the complex laplace transform of functions , but all arguments and results can be easily extended to the complex laplace transform of probability measures .
+ in particular , we get that the fourier transform @xmath69 is given by @xmath361 where the square root is defined by analytic extension as @xmath362 on the simply connected set @xmath363 . from the above expression , one easily derives the integrability of the above fourier transform .
since the laplace transforms ( [ lapla1 ] ) and ( [ lapla2 ] ) are analytic in the origin , it follows from @xcite[section xiii.2 ] that @xmath58 , @xmath69 and @xmath356 have finite n th moments for all @xmath364 .
hence , the same holds for @xmath277 and @xmath305 .
in this section we give some comments on the results concerning the @xmath0extrema of drifted brownian motion obtained in @xcite via the non rigorous real space renormalization group ( rsrg ) method and applied for the analysis of 1d random walks in random environments .
we present the results obtained in @xcite in the formalism of sinai s random walk , keeping the discussion at a non rigorous level .
start with a sequence of i.i.d .
random variables @xmath365 such that @xmath366 and @xmath367 \in { { \mathbb r}}\setminus\{0\}\,,\qquad \text{var } \left [ \log \frac{1-\o_0}{\o_0 } \right ] = : 2\s \in ( 0,\infty).\ ] ] defining @xmath368 , the random variable @xmath369 corresponds then to the random variable @xmath344 in ( 27 ) of @xcite . without loss of generality we set @xmath370 as in @xcite , thus implying that @xmath371 .
the associated sinai s random walk is the nearest neighbor random walk on @xmath372 where @xmath373 is the probability to jump from @xmath5 to @xmath374 and @xmath375 is the probability to jump from @xmath5 to @xmath376 .
consider the function @xmath377 defined on @xmath372 as @xmath378 and extended to all @xmath17 by linear interpolation .
morally , the above sinai s random walk is well described by a diffusion in the potential @xmath305 . in @xcite
the authors obtain results on the statistics of the @xmath379extrema of @xmath305 taking the limits @xmath380 , @xmath381 with @xmath382 fixed . in what follows
, we show the link between their results and our analysis of the statistics of @xmath0extrema of drifted brownian motion . by the central limit theorem applied to @xmath383 ,
one concludes that for @xmath379 large @xmath384 where @xmath385 is the standard two sided brownian motion ( i.e. @xmath385 has law @xmath386 ) .
if we set @xmath387 and consider the limits @xmath388 and @xmath381 with @xmath389 fixed we get that the rescaled potential @xmath305 is well approximated by the brownian motion @xmath1 with law @xmath390 . in particular , for @xmath379 large one expects that the ordered family @xmath391 of @xmath379extrema of @xmath305 is well approximated by family @xmath392 where @xmath393 is the ordered sequence of points of @xmath0extrema of @xmath1 with @xmath394 ( we follow the convention that @xmath395 ) .
setting as in @xcite @xmath396 morally we get @xmath397 in @xcite the authors write @xmath398 , @xmath399 for the joint probability density of the random variables @xmath400 if @xmath401 is a point of @xmath379minimum or a point @xmath379maximum respectively , they set @xmath402 and derive via the rsrg method the limiting form of @xmath403 ( note that in @xcite the authors erroneously do not distinguish the @xmath379slope covering the origin from the other @xmath379slopes , but in order to have a correct result one must take @xmath404 ) .
it is simple to check that all the computations obtained in @xcite[section ii.c.2 ] equal the results obtained in proposition [ calcoli ] by approximating the random variables @xmath405 with distribution @xmath406 by means of the random variables @xmath407 of proposition [ calcoli ] , where @xmath408 , @xmath394 .
this confirms ( [ arrosticini ] ) .
given a path @xmath409 , define the hitting time of @xmath344 at @xmath5 as @xmath410 consider the process @xmath411 carrying the law @xmath55 and define ( see figure [ coniglio1 ] ) @xmath412 the process @xmath413 is the so called one sided drifted bm reflected at its last infimum .
it has the following properties : [ l][l]@xmath414 [ l][l]@xmath32 [ l][l]@xmath415 [ l][l]@xmath0 .,title="fig:",width=264 ] [ japan]@xcite[lemma vi.55.1 ] + the process @xmath416 is a diffusion and @xmath417 is a local time of @xmath413 at @xmath25 .
the transition density function of the process _ @xmath413 stopped at @xmath25 _
, i.e. @xmath418 , is @xmath419\,,\qquad x , y > 0\,.\ ] ] the entrance law @xmath420 , @xmath421 , associated with the excursions of @xmath413 from @xmath25 w.r.t the time @xmath422 is given by @xmath423 , where : @xmath424,\qquad y > 0.\ ] ] let us comment the above result and fix some notation ( for a general treatment of excursion theory see for example @xcite[chapter vi ] and @xcite[chapter iv ] ) .
we denote by @xmath425 the space of excursions from @xmath25 , i.e. continuous functions @xmath426 satisfying the coffin condition @xmath427 where @xmath428 is the life time of @xmath344 , namely @xmath429.\ ] ] the excursion space @xmath425 is endowed of the smallest @xmath58algebra which makes each evaluation map @xmath430 measurable .
one can prove that this @xmath58algebra coincides with the borel @xmath58algebra of the space @xmath425 endowed of the shohorod metric .
write @xmath431 for the right continuous inverse of @xmath432 , namely @xmath433 and define the excursion @xmath434 , @xmath421 , as @xmath435 then the random point process @xmath269 of excursions of @xmath413 from @xmath25 is defined as @xmath436 in what follows , we will often identify the random discrete set @xmath437 with the random measure @xmath438 on @xmath439 . decomposing @xmath425 as @xmath440 , where @xmath441 it theorem @xcite [ theorem vi.47.6 ] states that there exists a @xmath58finite measure @xmath231 on @xmath425 ( called it measure ) with @xmath442 such that , if @xmath443 is a poisson point process on @xmath444 with intensity measure @xmath445 and if @xmath446\times u_\infty \right)>0\right\},\ ] ] then the point process @xmath269 under @xmath308 has the same law of @xmath447\times u}$ ] : @xmath448\times u}.\ ] ] here and in what follows , given a measurable space @xmath277 with measure @xmath449 and a measurable subset @xmath450 we denote @xmath451 the measure on @xmath277 such that @xmath452 given @xmath421 the entrance law @xmath453 , with support in @xmath454 , is defined as @xmath455 since the process @xmath413 ( starting at @xmath25 ) defined via ( [ salgo ] ) is markov and visits each @xmath456 a.s .
, the definition of the process @xmath413 starting at @xmath457 is obvious .
due to lemma [ japan ] , the process @xmath413 starting at @xmath457 and stopped at @xmath25 is a strong markov process with transition probability @xmath458 . in
what follows we denote its law by @xmath459 .
then , given @xmath421 , measurable subsets @xmath460 with @xmath461 , it holds @xmath462 where @xmath463 . in particular ,
due to ( [ rino1 ] ) the transition density functions ( [ aquila ] ) and the entrance laws ( [ roma ] ) determine univocally the it measure @xmath231 . in order to get more information on the it measure @xmath231 of the point process of excursions of @xmath413 from @xmath25 , we give an alternative probabilistic interpretation of the transition density @xmath464 . to this aim recall
that girsanov formula implies that @xmath465 for each @xmath466measurable function @xmath467 , where @xmath468 . due to ( [ aperol ] )
, we get for all @xmath469 , @xmath470 in fact , the second identity follows from ( [ aperol ] ) while the last identity follows from ( [ aquila ] ) by computing @xmath471 via a reflection argument .
hence , given @xmath472 , @xmath458 is the transition density function of the process @xmath473 under @xmath474 , whose law equals @xmath475 . in particular , ( [ rino1 ] ) can be reformulated as @xmath476 the above identity will be frequently used in what follows .
we point out that , as stated in theorem 1 of @xcite[section vii.1 ] , the content of lemma [ japan ] is valid in more generality for spectrally positive lvy processes s.t .
the origin is a regular point , i.e. real valued processes starting at the origin with stationary independent increments , with no negative jumps and returning to the origin at arbitrarily small times .
moreover , defining @xmath477 it is simple to check that the process @xmath413 starting at @xmath66 has the same law of the process @xmath478 , where @xmath1 is chosen with law @xmath479 .
this implies that @xmath480 if @xmath481 , hence once gets again that @xmath482 as in ( [ islandina ] ) . in order to state our results
it is useful to fix some further notation . given @xmath3 we denote by @xmath483 the family of excursions with height at least @xmath0 and by @xmath484 the family of excursions with height less than @xmath0 , namely @xmath485 one of the main technical tools in order to extend the proof in @xcite to the drifted case is the following lemma , whose proof is postpone to section [ primavera ] . [ kuka ] if @xmath486 , @xmath16 , then @xmath487 finally , we can prove lemma [ davide ] : _ proof of lemma [ davide]_. one can recover the brownian motion from the point process @xmath269 of excursions of @xmath413 from @xmath25 by the formula @xmath488,\ ] ] which is valid for each couple @xmath489 by setting @xmath490 it is convenient to associate to @xmath491 , @xmath492 the measures @xmath493 , @xmath494 , @xmath495 and @xmath496 .
moreover , we set @xmath497 if @xmath498 is finite , let @xmath499 be the only excursion such that @xmath500 .
due to ( [ zitta ] ) , ( [ pizza ] ) and ( [ kuka4 ] ) @xmath501\times u\right)=0\right)= \exp \left\{-a \ ,
n^ * ( u ) \right\},\ ] ] therefore @xmath498 is an exponential variable with parameter @xmath502 ( in particular , @xmath498 is finite a.s . ) . due to the representation ( [ iaco ] ) , @xmath503 .
together with ( [ kuka1 ] ) this implies that @xmath504 is an exponential variable with mean ( [ lapla ] ) .
moreover , ( [ iaco ] ) implies that @xmath505 due to the above expression and the representation ( [ iaco ] ) , the trajectory @xmath506 depends only on @xmath507 and @xmath498 , while the trajectory @xmath508 coincides with the excursion @xmath499 stopped when it reaches level @xmath0 .
since @xmath507 and @xmath498 are independent from @xmath499 we get the independence of the trajectories . in order to prove ( [ lapla1 ] ) we observe that @xmath509 means that @xmath510 .
therefore , conditioning to @xmath511 , it holds @xmath512 thus implying that @xmath513= { { \mathbf e}}_0 ^\mu \left(\exp\left\ { -\a\int _ { ( 0,x)\times u } h(f ' ) \nu _ * ( da',df ' ) \right\ } \right)\,.\ ] ] note that , in order to derive the above identity , we have used that @xmath269 is the superposition of the independent point processes @xmath514 and @xmath507 .
we claim that @xmath515 in order to prove this claim we note that , due to ito theorem and ( [ kuka4 ] ) , the point process @xmath507 has the same distribution of the poisson point process on @xmath444 with intensity @xmath516 .
hence , for @xmath517 the above identity follows directly from the exponential formula for poisson point processes @xcite[section o.5 ] .
suppose now that @xmath518 and @xmath519 .
given @xmath520 and @xmath521 we define @xmath522 as @xmath523 if @xmath524 and as @xmath525 if @xmath526 .
due to ( [ saratoga ] ) and ( [ rino2 ] ) , we get the bound @xmath527 where the r.h.s .
is finite due to the form of @xmath528 and identity ( [ leone1 ] ) .
this allows to conclude that the integral @xmath529 is finite , and therefore the same holds for the smaller integral @xmath530 .
this last property allows to apply again the exponential formula for poisson point process and to deduce that @xmath531 taking the limit @xmath532 and applying the monotone convergence theorem we derive ( [ sale ] ) from the above identity . hence , ( [ lapla1 ] ) follows from ( [ kuka2 ] ) , ( [ miele ] ) and ( [ sale ] ) , while trivially ( [ lapla2 ] ) follows from ( [ lapla1 ] ) . finally , in order to prove ( [ lapla3 ] ) , we observe that @xmath533 . since the path @xmath499 has law @xmath534 , @xmath535 and the thesis follows from ( [ kuka3 ] ) .
[ piponebis ] as already remarked , the analogous of lemma [ davide ] ( restricted to @xmath536 ) has been proved for more general spectrally one sided lvy processes @xcite , @xcite and @xcite[proposition 1 ] , by means of more sophisticated arguments always based on fluctuation theory , excursion theory and the analysis of the hitting times of the process .
we have given a self contained and direct proof based on simple computations , which will be useful also for the proof of theorem [ bonaparte ] , but one can derive lemma [ davide ] from the cited references as follows .
the laplace exponent of the drifted bm with law @xmath308 is given by @xmath537 , i.e. @xmath538 for @xmath539 .
given @xmath536 we define the function @xmath540 as @xmath541 then it is simple to check that @xmath542 where the value @xmath543 is defined as the largest root of @xmath544 , i.e. @xmath545 . the function @xmath540 is related to the exit of the bm from a given interval .
more precisely , due to ( [ leone2 ] ) , it holds @xmath546 to the function @xmath540 one associates the function @xmath547 given by @xmath548 knowing the values of @xmath540 and @xmath547 one can compute the expressions in lemma [ davide ] for @xmath536 by applying for example proposition 1 in @xcite .
in this section we characterize the behavior of an @xmath0slope not covering the origin , near to its extremes . to this aim
we recall the definition of the drifted brownian motion doob conditioned to hit @xmath24 before @xmath25 , referring to @xcite[section vii.2 ] and references therein for a more detailful discussion .
first , we write @xmath549 for the function @xmath550 ( @xmath551 has been defined in remark [ piponebis ] ) . defining @xmath552 has the largest zero of @xmath553 ,
i.e. @xmath554 , the function @xmath326 is a positive increasing function with laplace transform @xmath555 satisfying the identity ( see ( [ resistenza1000 ] ) ) @xmath556 due to the above considerations , the function @xmath326 is the so called _ scale function _ of the drifted brownian motion with law @xmath52 . for each @xmath66 consider the new probability measure @xmath557 on the path space @xmath558 characterized by the identity @xmath559 where @xmath560 denotes a generic element of the path space @xmath558 and @xmath561 . as discussed in @xcite[section vii.3 ] , the above probability measure is well defined , the weak limit @xmath562 exists and the process @xmath563 is a feller process , hence strong markov ( we point out that in @xcite[section vii.3 ] the above results are proven in the skohorod path space @xmath564 , but one can adapt the proofs to @xmath565 ) . as explained in @xcite[section vii.3 ] , this process can be thought of as the brownian motion with drift @xmath2 doob conditioned to hit @xmath566 before @xmath25 . in the case of positive drift ,
i.e. @xmath567 , this can realized very easily by observing that due to ( [ resistenza ] ) @xmath568 and that this identity together with the markov property implies that @xmath569 for negative drift the event @xmath570 has zero probability , and a more subtle discussion is necessary .
[ black ] the process @xmath571 is a diffusion characterized by the sde @xmath572 where @xmath31 is the standard brownian motion . as already discussed , the above process has continuous paths and it is strong markov , i.e. it is a diffusion .
due to ( [ jonny ] ) and ( [ islandina ] ) , given @xmath573 , @xmath574 \,.\end{gathered}\ ] ] from the above expression , by direct computations one derives that @xmath575 hence , the generator of the process is given by @xmath576 and this implies the sde ( [ partigiani ] ) .
let us consider now the brownian motion with drift @xmath2 doob conditioned to hit @xmath0 before @xmath25 and killed when it reaches @xmath0 .
in order to precise its meaning when the brownian motion starts at the origin , given @xmath577 , we define @xmath578 as the conditioned law on @xmath579 @xmath580 note that the above definition is well posed since by ( [ resistenza ] ) @xmath581 .
[ block ] given @xmath577 , let @xmath582 and @xmath583 be the law of the path @xmath584 killed when level @xmath0 is reached , where @xmath277 is chosen with law @xmath585 and @xmath586 respectively
. then @xmath587 and the weak limit @xmath588 exists and equals @xmath589 .
given @xmath590 and times @xmath591 , we denote by @xmath592 the event @xmath593then , by definition of @xmath594 and the markov property of the brownian motion , for each @xmath577 we get that @xmath595 due to ( [ jonny ] ) , the last expression in the r.h.s .
equals the probability @xmath596 , hence we can conclude that @xmath597 .
hence @xmath598 for @xmath577 .
the last statement concerning @xmath599 follows from the fact that the weak limit @xmath600 exists and equals @xmath601 . due to the first part of the above lemma
, we can think of @xmath602 as the brownian motion with drift @xmath2 doob conditioned to hit @xmath0 before @xmath25 and killed when it hits @xmath0 .
we have now all the tools in order to describe the behavior of the @xmath0slopes not covering the origin , near to their extremes . in order to simplify the notation , in what follows we denote by @xmath603 the two
sided brownian motion with drift @xmath2 , starting at the origin .
moreover , given @xmath604 , we define @xmath605 [ bonaparte ] let @xmath606 and let @xmath607 be consecutive points of @xmath0extrema for the drifted brownian motion @xmath603 , both non negative or both non positive . if @xmath449 is a pont of @xmath0minimum and @xmath608 is a point of @xmath0maximum , then the processes @xmath609 have the same law of the brownian motion starting at the origin , with drift @xmath2 , doob conditioned to reach @xmath24 before @xmath25 and killed when it hits @xmath0 .
moreover , they have the same law of the brownian motion starting at the origin , with drift @xmath2 , doob conditioned to reach @xmath0 before @xmath25 and killed when it hits @xmath0 . in particular
, they satisfy the sde ( [ partigiani ] ) up to the killing time .
if @xmath449 is a point of @xmath0maximum and @xmath608 is a point of @xmath0minimum , then the processes @xmath610 have the same law of the brownian motion starting at the origin , with drift @xmath389 , doob conditioned to reach @xmath24 before @xmath25 and killed when it hits @xmath0 .
moreover , they have the same law of the brownian motion starting at the origin , with drift @xmath389 , doob conditioned to reach @xmath0 before @xmath25 and killed when it hits @xmath0 . in particular
, they satisfy the sde ( [ partigiani ] ) with @xmath389 replaced by @xmath2 , up to the killing time . the second part of the theorem follows from the first part by taking the reflection w.r.t . the coordinate axis .
as follows from the proof of lemma [ davide ] in section [ pierpa ] , the law of the process ( [ mela1 ] ) coincides with the law of the excursion @xmath344 killed when it reaches @xmath0 , where @xmath344 is chosen with probability measure @xmath611 due to proposition 15 in @xcite[section vii.3 ] , there exists a positive constant @xmath612 such that @xmath613 ( note that ( [ crock ] ) corresponds to ( [ kuka1 ] ) with @xmath614 ) .
hence , given numbers @xmath615 in @xmath616 and increasing times @xmath617 , we have @xmath618 ( the first identity follows from the markov property ( [ rino2 ] ) , while the latter follows from ( [ resistenza ] ) and ( [ crick ] ) ) . from ( [ crock ] ) and ( [ cruck ] ) one derives that @xmath619 this concludes the proof that the process ( [ mela1 ] ) has the same law of the brownian motion starting at the origin , with drift @xmath2 , doob conditioned to reach @xmath24 before @xmath25 and killed when it hits @xmath0 .
all the other statements follow from this property , lemmata [ black ] and [ block ] and by reflection arguments .
knowing the laplace transform of the hitting times of the non drifted brownian motion @xcite[chapter 2 ] , the following lemma follows by applying girsanov formula ( [ aperol ] ) : set @xmath622 .
we claim that @xmath623 indeed , the first identity follows from the monotone convergence theorem , the second one from ( [ aperol ] ) , and the last one by conditioning on @xmath624 and observing that @xmath625 is @xmath624 measurable . since , under @xmath386
, @xmath626 is a martingale and @xmath627 is a bounded stopping time , the optional sampling theorem implies that @xmath628 for @xmath629 , @xmath630 equals @xmath631 , hence @xmath632 since @xmath65 , the above identity together with formula ( 8.27 ) in @xcite[chapter 2 ] implies ( [ papera1 ] ) . due to the above lemma ,
given @xmath634 and @xmath635 , @xmath636 by taking the limit @xmath637 in ( [ leone1 ] ) we get for all @xmath638 and @xmath486 @xmath639 by considering the taylor expansion around @xmath640 in above identity , one can compute the expectation of @xmath641 .
in particular , for all @xmath642 it holds @xmath643 we collect some identities ( obtained by straightforward computations ) which will be very useful below .
first we observe that given @xmath644 and @xmath421 it holds @xmath645 in particular , fixed @xmath646 and @xmath647 , it holds as @xmath648 @xmath649 note that the last identity can be derived from ( [ pipolo ] ) by observing that @xmath650 since for a suitable positive constant @xmath653 it holds @xmath654 since excursions are continuous paths and @xmath655 pointwise , by the dominated convergence theorem in order to prove the above limit it is enough to show that @xmath656 by the monotone convergence theorem @xmath657 and due to ( [ rino2 ] ) @xmath658 due to ( [ kaka1 ] ) and ( [ kaka2 ] ) , the last term @xmath659 is negligible as @xmath660 while , due to ( [ leone6 ] ) , @xmath661 in order to conclude it is enough to observe that @xmath662 consider the measurable subsets @xmath663 .
then @xmath664 for @xmath665 and , by the continuity of excursions , @xmath666 this implies that @xmath667 . due to ( [ rino2 ] ) @xmath668 hence ( see also ( [ leone4 ] ) ) we get that @xmath669 where @xmath670 due to ( [ kaka2 ] ) @xmath671 . in order to treat the term
@xmath672 we write @xmath673 where @xmath674 due to ( [ kaka1 ] ) , @xmath675 and @xmath676 .
in particular , @xmath677 thus concluding the proof of ( [ kuka1 ] ) .
consider the subsets @xmath678 defined as @xmath679 then @xmath680 and , in order to prove ( [ kuka4 ] ) , it is enough to show that @xmath681 .
due to ( [ rino2 ] ) we can write @xmath682 note that if @xmath683 then @xmath684 for all @xmath638 , while if @xmath685 then @xmath686 for all @xmath687
. hence @xmath688 by ( [ kaka2 ] ) the last member above goes to @xmath25 as @xmath689 .
this implies that @xmath690 , thus concluding the proof of ( [ kuka4 ] ) .
we claim that @xmath691 indeed , the above identity follows from the dominated convergence theorem if we prove that @xmath692 to this aim , we observe that , for all @xmath421 , @xmath693 and by ( [ basilico1 ] ) @xmath694 let us prove ( [ mosca ] ) . by ( [ rino2 ] ) @xmath696 where @xmath697 due to the identities derived at the beginning of the proof of ( [ kuka1 ] ) we can write @xmath698 while due to ( [ kaka1 ] ) and ( [ kaka2 ] ) @xmath699 the above identities give @xmath700 due to ( [ leone1 ] ) @xmath701 hence , due to ( [ kaka3 ] ) , @xmath702 then ( [ mosca ] ) follows from ( [ russia ] ) , ( [ 15marzo ] ) and ( [ 15marzobis ] ) , thus concluding the proof of ( [ kuka2 ] ) . due to the monotone convergence theorem @xmath703
it is convenient to write the last integral as @xmath704 , where @xmath705 hence @xmath706 by ( [ rino2 ] ) @xmath707 where @xmath708 due to ( [ leone2 ] ) @xmath709 by applying ( [ kaka3 ] ) to the r.h.s .
we get that @xmath710 by ( [ leone5 ] ) @xmath711 and due to ( [ kaka2 ] ) @xmath712 .
this limit together with ( [ corno0 ] ) and ( [ corno1 ] ) implies that @xmath713 we claim that @xmath714 . note that this together with ( [ airberlin ] ) , ( [ grande1 ] ) and ( [ kuka1 ] ) implies ( [ kuka3 ] ) .
hence , in order to conclude we only need to prove our claim . to this aim we apply hlder inequality with exponents @xmath715 with @xmath716 and @xmath717 , deriving that @xmath718^{1/p } n ( t_h \leq t ) ^{1/q}\,.\ ] ] as @xmath719 the first factor in the r.h.s
. goes to @xmath720 , the second factor has finite limit due to ( [ grande1 ] ) where @xmath39 has to be replaced by @xmath721 ( here we use that @xmath722 ) .
hence we only need to prove that @xmath723 . by the monotone convergence theorem
it is enough to show that @xmath724 .
but @xmath725 which is bounded by ( [ kuka1 ] ) .
* acknowledgements*. the author thanks a. bovier , f. le gall and z. shi for useful discussions .
moreover , she kindly acknowledges the weierstrass institute for applied analysis and stochastics in berlin for the kind hospitality while part of this work was being done .
j. neveu , j. pitman , _
renewal property of the extrema and tree property of the excursion of a one dimensional brownian motion_. in _ sminaire de probabilits xxiii _ , 239246 .
lectures notes in math .
springer , 1989 .
prokhorov , yu.a .
rozanov _ probability theory ; basic concepts , limit theorems , random processes_. die grundlehren der mathematischen wissenschaften in einzeldarstellungen ; 157 .
berlin , springer - verlag , 1969 . | we show that the slopes between @xmath0extrema of the drifted 1d brownian motion form a stationary alternating marked point process , extending the result of j. neveu and j. pitman for the non drifted case .
our analysis covers the results on the statistics of @xmath0extrema obtained by p. le doussal , c. monthus and d. fisher via a renormalization group analysis and gives a complete description of the slope between @xmath0extrema covering the origin by means of the palm khinchin theory .
moreover , we analyze the behavior of the brownian motion near its @xmath0extrema . _
2000 mathematics subject classification : _
60j65 , 60g55 .
_ key words : _ brownian motion , marked point processes , palm
khinchin theory , fluctuation theory . |
let us remind the differential equations on time scales , proposed by hilger @xcite .
the main element of the equations is the time scale , which is understood as a nonempty closed subset , @xmath0 of the real numbers . on a time scale
@xmath1 the functions @xmath2 and @xmath3 are called the forward and backward jump operators , respectively .
the point @xmath4 is called right - scattered if @xmath5 and right - dense if @xmath6 similarly , it is called left - scattered if @xmath7 and left - dense if @xmath8 the @xmath9derivative of a continuous function @xmath10 at a right - scattered point is defined as @xmath11 and at a right - dense point it is defined as @xmath12 if the limit exists .
a differential equation @xmath13 is said to be a differential equation on time scale , where function @xmath14 in ( [ tsde ] ) is assumed to be rd - continuous on @xmath15 @xcite .
if a point @xmath16 is right - dense , then the @xmath17-derivative is the ordinary derivative . otherwise , @xmath4 is a right - scattered point , and the @xmath17-derivative is defined by means of the quotient .
let us discuss the second case more attentively . by the definition of the @xmath9derivative
, assuming that @xmath16 is not an isolated point , one has that @xmath18 now , compare this expression with the numerical approximation of the function on the interval @xmath19.$ ] setting @xmath20 and @xmath21 we have that @xmath22 thus , one can see that the idea of the @xmath9derivative , as well as of the @xmath23derivative is close to the basic one for the numerical analysis @xcite .
see figures [ fig1 ] and [ fig2 ] for illustration .
that is , in its heart the concept of the differential equations on time scales is closely adjoint with numerical methods .
however , for numerical methods , of course as far as the definition in @xmath24 is considered , smaller @xmath25 gives a better approximation . in the former theory of differential equations on time scales , the request was not considered , and the elements of linear change over the intervals near right - scattered ( left - scattered ) points are strong there . to return the non - linearity sense in the discussion , one needs to modify the equations . in our papers
@xcite , the concept of transition conditions were proposed , when we consider the simple version of the time scale as a union of disjoint sections . in @xcite ,
these systems emerge as an auxiliary instrument of a discontinuous dynamical system investigation . in section [ sec3 ]
, we generalize this idea for all possible time scales .
in this part we shall answer two questions : why the differential equations on time scales are needed by real world problems ? how the hilger s equations can be extended for these ?
consider a real world process , which can be symbolically given as a function @xmath26 where @xmath16 is from a time scale .
suppose for this moment that the time scale admits only non - isolated moments .
if the time is increasing , we have to prescribe what the value at @xmath27 is equal to for the right - scattered point @xmath28 in hilger s theory , the value is evaluated by using the derivative of @xmath29 at the moment @xmath30 and this is very restrictive , since the difference @xmath25 is not supposed to be small !
it is reasonable to assume that there exists more general law of transition for @xmath29 from the moment @xmath16 to @xmath31 in general , it can be given with a function @xmath32 we call this expression as the transition condition .
to illustrate , consider a population developing in time .
we may assume that the size satisfies a local law in some borders , say in summer time , and it is not under the law in winter time .
moreover , the local law in winter time is not known .
nonetheless , we can evaluate the change in the time , that is the size at the beginning of the season and that at the end of the season .
assume that the change satisfies a new formula , which is naturally a discrete one .
in that case , we come to the idea of the equations on time scales , but more general than those investigated before .
consider the time scale @xmath33 given above , and the same @xmath9derivative of a continuous function @xmath10 defined by ( [ dd1 ] ) and ( [ dd2 ] ) .
denote by @xmath34 the set of all right - scattered points of @xmath35 consider a function @xmath36 which is assumed to be rd - continuous on @xmath37 define a function @xmath38 now , on the time scale , we introduce a piecewise function @xmath39 which is equal to the function @xmath40 if @xmath16 is a right - dense point , and it is equal to @xmath41 if @xmath16 is a right - scattered point . we suggest to consider the following differential equation on the time scale @xmath42 condition that all solutions are continuous on the time scale should be assumed . to demonstrate the opportunities , which one can have with the newly introduced equations , consider the following theorem of existence and uniqueness .
let @xmath43 be a time scale , @xmath44 @xmath45 @xmath46 with @xmath47 and @xmath48 and put @xmath49 suppose that @xmath50 is rd - continuous , bounded with bound @xmath51 lipschitz continuous with lipschitz constant @xmath52 and @xmath53 for all @xmath54 and @xmath55 then the ivp @xmath56 has exactly one solution on @xmath57,$ ] where @xmath58 if @xmath59 and @xmath60 then the unique solution exists on the interval @xmath61.$ ] [ thmeu ] one could compare the last assertion with the following , from @xcite .
the condition on the transition function @xmath62 is added , in this particular case . in general
, properties of the function will generate new theoretical , and what is most important , application prospects for these equations .
one can see that many results in the theory , which have been obtained earlier can , now , be generalized , for newly proposed systems with additional interesting properties that transition conditions can admit .
another opportunity is to consider state - dependent time scales @xcite .
let a set @xmath63 in @xmath64 be given , such that for each @xmath65 the projection of the intersection , @xmath66 of the line @xmath67 with @xmath68 on the time axis is a time scale in the sense of hilger .
thus , one can define the function @xmath69 for each point @xmath70 we assume that the function @xmath71 is continuous in @xmath72 if the moment @xmath16 is such that @xmath73 is right - dense in the correspond time scale , @xmath74 then the @xmath9derivative is defined by ( [ dd2 ] ) .
otherwise it is equal to @xmath75 now , define a function @xmath76 which is equal to a continuous function @xmath77 at a right - dense point @xmath78 and to function @xmath79 if the point is a right - scattered one .
then , one can discuss the following equation @xmath80 it is obvious that the equations with state - dependent time scales can also play a certain role in this theory , and they will be used in the modeling of real world problems .
the analysis may be of great interest , since new opportunities related to the non - linear properties of the time scales will emerge .
finally , we want to mention that very profitable ways for differential equations on time scales is to investigate them together with different types of discontinuities .
these days there are several well extended theories such that systems with discontinuities in the right - hand - side , arguments , impulsive differential equations .
they are realized in very interesting application projects as well @xcite-@xcite .
we are sure that integration of the methods and results of these theories with above mentioned extensions of differential equations on time scales can provide really interesting horizons , which can give a light on new ways for modeling processes in mechanics , electronics , medicine , biology .
one of the particular results in this sense is our recent paper @xcite , where certain class of differential equations on time scales is embedded in differential equations with fixed moments of impulses .
a. katok , j .-
strelcyn , f. ledrappier and f. przytycki , _ invariant manifolds , entropy and billiards ; smooth maps with singularities _ , lecture notes in mathematics , 1222 , springer - verlag , berlin , 1986 . | we give a proposal to generalize the concept of the differential equations on time scales @xcite , such that they can be more appropriate for the analysis of real world problems , and give more opportunities to increase the theoretical depth of investigation .
differential equations on time scales ; transition condition ; modeling |
the era of gravitational wave ( gw ) astronomy has come since the laser interferometer gravitational - wave observatory ( ligo ) detected the first gw event , gw150914 , which is thought to be produced by the merging of a stellar mass binary black hole system @xcite . finding the electromagnetic ( em ) counterpart of the gw event is critically important in many aspects , such as identification of the gw event , study of the progenitor and its environment , fundamental physics and cosmology .
although many efforts have been made to search for the purported em counterpart of gw150914 @xcite , only the gamma - ray burst monitor ( gbm ) onboard fermi @xcite recorded an excess ( denoted as gw150914-gbm ) at 0.4 s after the ligo trigger time , which is claimed to be possibly associated with this gw event @xcite .
however , since there is no significant signal detected by the integral / spi - acs around the time of gw150914-gbm @xcite , a great debate has been raised about whether gw150914-gbm is of astrophysical origin and associated with the gw150914 .
@xcite reported that gw150914-gbm is likely associated with the gw150914 based on the following reasons : ( a ) its time is only 0.4 s after the ligo trigger time of the gw event .
( b ) its localization is not inconsistent with ligo locations of gw150914 , although quite poor due to the weakness and unfavorable incident angles to gbm detectors .
( c ) its duration and spectrum are consistent with typical short gamma - ray bursts ( grbs ) which is thought to be candidate of gw em counterpart .
( d ) it can not be attributed to other known sources , such as solar activity , terrestrial gamma - ray flashes @xcite or terrestrial electron beams @xcite produced on the earth , magnetosphereic activities and galactic sources .
however , @xcite argued that the non - detection of spi - acs suggests the gw150914-gbm is likely from magnetospheric activity or background fluctuation .
but @xcite argued that the fluence of gw150914-gbm is consistent with the upper limit estimated by spi - acs for most regions of ligo arc ; about half of weak gbm events are not found by spi - acs ; the origin of magnetosphere is very unlikely because gbm is on the low magnetic latitude and the duration of gw150914-gbm is too short for the magnetospheric events usually seen by gbm .
however , the reported gbm false alarm rate of 0.0022 @xcite is only of moderate significance to reject the possibility of background variation . given that there is no other measurement in a similar energy range as gbm and spi - acs to the gw150914 region around the ligo trigger time , and that no other em counterpart detection has been reported yet , the origin of gw150914-gbm and whether gw150914 has an em counterpart remain a big mystery , but of high importance .
here we focus on the observational properties of gw150914-gbm by carefully analyzing the gbm data using a straightforward but sophisticated method .
gbm employs the thallium - doped sodium iodide ( nai ) and bismuth germanate ( bgo ) scintillation detectors to cover a wide energy range from 8 kev to 40 mev @xcite . since a nai detector only has a good viewing to sources within @xmath0 of the detector s normal , the response of the detector decreases rapidly when the incident angle is greater than 60@xmath1 @xcite .
to form an all - sky field of view ( fov ) except the inevitable blocking by the earth in low earth orbit , gbm consists of 12 nai detectors with different orientations placed at the four corners of the spacecraft and two bgo detectors set in the opposite sides of the spacecraft ( see * ? ? ?
such configuration is also used to localize astrophysical transients based on the fact that these detectors will receive different number of photons from a distant source due to different incident angles of the source to different detectors .
such localization method @xcite has successfully helped gbm to localize @xmath2 2000 grbs @xcite .
the gbm team has reported an excess , gw150914-gbm , with false alarm rate of 0.0022 at t0 + 0.4 s , where t0 is the ligo trigger time 2015 - 09 - 14 09:50:45.391 utc , and this event is believed to be possibly linked to gw150914 @xcite .
as shown in the top panel in figure [ fig : f1 ] , we produced the summed light curve of all 14 detectors for gw150914-gbm in the same energy selection as used in @xcite , which is very consistent with the results ( bottom panel of fig .
4 ) in @xcite .
next we test whether gw150914-gbm is of astrophysical origin and associated with the gw event . unlike the search algorithm utilized in @xcite which is based on a weighted light curve of all 14 detectors of gbm
, here we choose to use a much more straightforward but sophisticated approach : we divide all 14 detectors into two groups with one group having good viewing angles to the ligo location of the gw event and the other group bad viewing angles . according to the placement of the detectors described above
, good - viewing detector group should have seen more photons than the bad - viewing group from the gw150914 , if there is any high energy emission from it .
we used the calculation of the incident angle for each detector in table 2 of @xcite to decide which detector belongs to which group .
it turns out that nai 2 , nai 4 , nai 5 and bgo 0 have smaller viewing angles than other detectors for all possible locations of the gw event , thus these four detectors are chosen to be the good - viewing detectors for gw150914 .
the incident angle to the good - viewing group ranges from 20@xmath1 to 80@xmath1 , depending on the real location of gw150914 .
we slightly relieved the usual threshold of good - viewing of 60@xmath1 to 80@xmath1 , because there are too few detectors with less than 60@xmath1 for a single location of the gw event .
the other ten detectors compose the bad - viewing group because their incident angles are well greater than 80@xmath1 , resulting in either very little exposure to the gw event or completely blocked by the spacecraft .
light curves for all 14 detectors , good - viewing and bad - viewing detectors are separately plotted in figure [ fig : f1 ] . in the duration of gw150914-gbm ( from t0 + 0.41
s to t0 + 1.43 s ) , there are 462 ( @xmath3 = 82.1 , significance of this excess is 5.63 @xmath3 ) net counts above the fitted background level in all 14 detectors , where 306 ( @xmath3 = 67.3 , significance of this excess is 4.55 @xmath3 ) of them come from bad - viewing detectors and 156 ( @xmath3 = 47.0 , significance of this excess is 3.32 @xmath3 ) from good - viewing ones .
we suggest this event is very unlikely from the gw150914 as following : * ( i. ) the majority of the excess ( @xmath4 ) of gw150914-gbm is from bad - viewing detectors , which is inconsistent with an astrophysical source incident from the ligo localizations visible to the gbm*. the ratio of net counts of good - viewing and bad - viewing detectors for gw150914-gbm is 0.51 @xmath5 0.19 . in principle
, one could do a detailed monte carlo simulation to estimate what this ratio should be for an astrophysical source ; however , such simulation is impossible without the accurate mass model of the fermi spacecraft which is unavailable for public .
alternatively , we estimate the expected ratio from gbm - swift jointly - detected grbs ( shown in table [ tab : grbs ] ) with similar incident angles in spacecraft coordinates as gw150914 , because it only depends sensitively on the incident angle in the spacecraft coordinates for the spectra dominated by low - energy photons , which is always true for grbs in gbm energy range . as an example
, the light curve of grb140716a is shown in figure [ fig : f2 ] .
the measured net counts ratio ( 0.51 @xmath5 0.19 ) of gw150914-gbm is far away from that of grbs in table [ tab : grbs ] , consistent with the fact that the best localization of gw150914-gbm is on the earth , rather than the gw150914 locations @xcite . assume the excess in bad - viewing detectors for gw150914-gbm is from ligo locations visible to gbm , and we use 1.4 as a conservative estimation of the net counts ratio for the gw150914 , then the expected net counts in good - viewing detectors should be 306 @xmath6 1.4 = 428.4 counts , with the error the same as original measurement of 67.3 , because it is dominated by the background variation .
thus this assumption is rejected in @xmath7 .
indeed , as shown in the table 1 and 2 in @xcite , the bad - viewing detectors ( nai 9 , bgo 1 ) with high incident angle for all possible ligo locations have the most high significance of excess while the good - viewing detector ( nai 2 ) with much smaller incident angle has the least significance of excess .
such distribution of excess among detectors is inconsistent with an astrophysical source from ligo locations .
it s very likely that gw150914-gbm will not be found out by the targeted search @xcite without the excess signal in the bad - viewing detectors .
the excess of gw150914-gbm in good - viewing detectors is consistent with background fluctuation*. to estimate the false alarm rate of the excess of gw150914-gbm in good - viewing detectors , we searched 216.6 ks of data of good - viewing detectors and found 56 events with higher significance , giving a false alarm rate of @xmath8 hz , much higher than that reported in @xcite . note that there are already several spikes with similar excess as gw150914-gbm within only 100 s of data in good - viewing detectors ,
see panel b in figure [ fig : f1 ] .
this suggests that the excess of gw150914-gbm in good - viewing detectors is consistent with background fluctuation .
in fact , the ratio of net counts between good - viewing and bad - viewing detectors for gw150914-gbm ( 0.51 @xmath5 0.19 ) is approximately equal to the ratio of background level ( red lines in panel b and c in figure [ fig : f1 ] ) and detector number of these two detector groups .
this may suggest that the gw150914-gbm is likely caused by a sudden increase of the local particles around the spacecraft .
ccccccc grb140716a & grb140716436 & 134.1 & 352.2 & 8.8 & 5/6/7/8 & 1.4 @xmath5 0.2 + grb140817a & grb140817293 & 131.5 & 10.7 & 9.5 & 7/8 & 1.7 @xmath5 0.2 + grb150607a & grb150607330 & 130.1 & 7.7 & 9.3 & 6/7/8 & [ 1.7 , 11.2 ] +
by using a straightforward but sophisticated method , we find that the excess of gw150914-gbm , which is suggested to be a possible counterpart of gw150914 @xcite , is mostly contributed by those gbm detectors that are not expected to see much radiation from the ligo localization region of gw150914 , whereas detectors ( nai 2 , nai 4 , nai 5 , bgo 0 ) with good exposures to the gw event did not see any significant excess beyond the background fluctuation .
the distribution of excess signals among detectors of gw150914-gbm is very inconsistent from the expected value that has been demonstrated by gbm - swift jointly - detected grbs with similar incident angles as gw150914 .
however , @xmath2 2000 grbs have been successfully localized based on such distribution @xcite .
therefore , we conclude that the gw150914-gbm is very unlikely associated with gw150914 , and the origin of the excess in bad - viewing detectors , which is not the focus of this paper though , is possibly due to some sudden increase in the local particles , while the relatively small excess in the good - viewing detectors is not distinguishable from the background fluctuation .
since gw150914-gbm is the only event found by gbm that is possibly associated with gw150914 in a comprehensive search , the rejection of it to be an astrophysical event in the ligo localizations means that gbm did not detect any significant em counterpart of gw150914 .
@xcite predicted that the merging of two astrophysical black holes will not produce any electromagnetic emission , based on the exact global solution of the dynamic and evolving metric for matter falling onto a pre - existing black hole in the frame of a distant external observer .
as of today , the non - detection of em counterpart of gw150914 by a series of telescopes @xcite including fermi / gbm is consistent with this prediction @xcite .
however , it is commonly believed that the in - falling matter to a black hole must approach asymptotically to but never cross the event horizon of the black hole in the frame of a distant external observer ( e.g. * ? ? ?
* ; * ? ? ?
* ) ; in this case the merging of two such objects will produce both gw and strong em radiations @xcite , through the well - known blandford - znajek ( bz ) mechanism @xcite .
because the relativistic jet produced through the bz mechanism for a gw150914-like event may not point to the observer , the non - detection of strong em radiation from just one gw150914-like event does not distinguish between the above two models ; therefore , a statistically significant sample of gw150914-like events with comprehensive em search is highly demanded .
thanks to the very wide instantaneous fov , broad energy coverage and moderate localization capability , fermi / gbm is one of the best telescopes to find the purported em counterparts of the gravitational wave events . as more joint observations are undergoing between ligo and fermi / gbm , it is very promising for fermi / gbm to either find some em counterparts or put a strong constraint on the em radiation of gw150914-like events .
i thank shuang - nan zhang , fangjun lu and liming song for very helpful discussions and encouragement .
i am grateful to the fermi / gbm team and nasa for the public data and software .
this work is partially supported by the hundred talents program funded by the chinese academy of sciences ( cas ) , the national basic research program ( `` 973 '' program ) of china ( grants 2014cb845800 ) and the strategic priority research program on space science , the chinese academy of sciences , grant no .
xda04010202 and xda04010300 . | finding the electromagnetic ( em ) counterpart is critically important for a gravitational wave event .
although many efforts have been made to search for the purported em counterpart of gw150914 , the first gravitational wave event detected by ligo , only fermi / gbm reported an excess above background ( i.e. gw150914-gbm ) at 0.4 s after the ligo trigger time , that is possibly associated with this gw event @xcite . however , since there is no significant detection by the integral / spi - acs around the time of gw150914-gbm , a great debate has been raised about whether gw150914-gbm is of astrophysical origin and associated with the gw150914 @xcite . in order to answer this question ,
we re - analyzed the gbm data with a straightforward but sophisticated method .
we find that the excess of gw150914-gbm mostly comes from those detectors with bad viewing angles to the gw event , whereas the good viewing detectors see nothing significant beyond background fluctuation around the trigger time of gw150914 .
this finding suggests that gw150914-gbm is very unlikely associated with the gw150914 .
given that gw150914-gbm is the only event found by gbm that is possibly associated with this gw event in a comprehensive search , we conclude that gbm did not detect any electromagnetic radiation from the gw150914 . |
the most recent results of , which measures the size of the direct cp violation in @xmath2 , reported by ktev @xcite and na48 @xcite lead to a world average of re = @xmath3 @xcite . this result is higher than the standard model ( sm ) predictions @xcite , opening the way to the interpretation that it may be a signal of new physics beyond the sm .
the sm predictions for suffer from large theoretical uncertainties @xcite such that one can not draw a definite conclusion if this observed high value of can be accommodated in the sm . in any case , one may wonder if the supersymmetry ( susy ) can be responsible for enhancing . in the minimal supersymmetric extension of the sm ( mssm )
there is no way of generating a siable susy contribution to even if one assume that the susy cp violating phases are large and the electric dipole moments ( edm ) of the electron and neutron are less than the experimental bounds due to the cancellation between the different contributions .
this is mainly due to the assumption of universal boundary conditions of the soft - breaking terms @xcite .
it has been shown that , without new flavor structure beyond the usual yukawa couplings , general susy models with phases of the soft terms of order @xmath4 ( but with a vanishing ckm phase @xmath5 ) can not give a sizeable contribution to the cp violating processes @xcite .
this means that the presence of non universal soft breaking terms besides large susy phases is crucial to enhance these cp violation effects . in agreement with this
, it has been explicitly shown that contributions to @xmath6 are small within the dilaton dominated susy breaking of the weakly coupled heterotic string model @xcite , where @xmath0terms as well as gaugino masses are universal . on the other hand ,
it is well known that the strict universality in the soft breaking sector is a strong assumption not realized in many supergravity and string inspired models @xcite .
all these arguments indicate not only that the presence of non universal soft terms can solve the problem of too large contributions to edms but also that it allows for large susy contributions in cp violation experiments .
hence , in this work we will follow this avenue and analyze the effects of non universal soft terms in cp violation in the @xmath7system @xcite .
it is well known that in susy models there are new possibilities for cp violation . in particular , the soft susy breaking terms contain several parameters that may be complex , as can also be the @xmath8-parameter . in the minimal supergravity model
there are only two new cp - violating phases .
this can be seen as follows .
the parameters @xmath9 and @xmath10 and @xmath8 can be complex .
but of these four phases only two are physical .
first , by an r - rotation with r - charge @xmath11 for lepton and quark superfields and @xmath12 for the vector and the higgs superfields , the gaugino mass parameter @xmath13 can be made real .
second , @xmath14 can be made real by a change of phase of the higgs superfield .
this ensures that the higgs vacuum expectation values are real .
the remaining phases can not be defined away and violate cp .
one is in @xmath15 and the other in @xmath16 .
the @xmath8 parameter then has a fixed phase @xmath17 . in any phase convention _
a= ( am^ * ) , _ b= ( bm^ * ) .
these phases can cause at one loop level an electric dipole moment ( edm ) for the quarks and leptons , and therefore also for the neutron .
it has been known for a long time that in the susy models the contributions to the neutron electric dipole moment are larger than the experimental limit @xmath18 e cm unless either the new ` susy phases ' are tuned to be of order @xmath19 , or the susy masses are of order a tev .
such small phases can not generate sizable cp violation .
also they constitute a fine tuning .
this is known as susy cp problem " .
however , in the last few it has been suggested that a natural cancellation mechanism exists whereby the electric dipole moment of the neutron may be made small without such fine - tuning . in this case
large susy phases are expected and still satisfy experimental bounds on the values of edm of the electron and neutron . in this section
we will study the effect of these phases in cp violation observables as and .
we assume that @xmath20 to maximize this effect .
the value of the indirect cp violation in the kaon decays , @xmath21 , is defined as @xmath22 where @xmath23 gev . the amplitude @xmath24 .
the relevant supersymmetric contributions to @xmath25 are the gluino and the chargino contributions , ( , the transition proceeds through box diagrams exchanging gluino - squarks and chargino - squarks ) .
it is usually expected that the gluino is the dominant contribution . however , as we will show , it is impossible in the case of degenerate @xmath0-terms that the gluino gives any significant contribution to @xmath21 when the ckm matrix is taken to be real even with large phase of @xmath0 .
the amplitude of the gluino contribution is given in terms of the mass insertion @xmath26 defined by @xmath27 where @xmath28 is an average sfermion mass and @xmath29 is off - diagonal terms in the sfermion mass matrices .
the mass insertion to accomplish the transition from @xmath30 to @xmath31 ( @xmath32 are flavor indices ) is given by @xmath33(3\tilde{m}^2+\vert x \vert^2 ) , \\
( \delta^d_{lr})_{ij}&\simeq&-\frac{1}{8\pi^2}\left[\frac{k^{\dag } ( m_u^{diag})^2 k\ m_d}{v^2 \sin^2\beta \cos\beta } \right ] \ln(\frac{m_{gut}}{m_w } ) x , \\ ( \delta^d_{rl})_{ij}&\simeq&-\frac{1}{8\pi^2}\left[\frac{m_d\ k^{\dag } ( m_u^{diag})^2 k } { v^2 \sin^2\beta \cos\beta } \right ] \ln(\frac{m_{gut}}{m_w } ) x , \\ ( \delta^d_{rr})_{ij}&=&0,\end{aligned}\ ] ] where @xmath34 .
it is clear that @xmath35 in general are complex due to the complexity of the ckm matrix , the trilinear coupling @xmath0 and @xmath8 parameter .
here we assume the vanishing of @xmath36 to analyze the effect of the susy phases .
we notice that @xmath37 is proportional to @xmath38 , it is real and does not contribute to @xmath21 whatever the phase of @xmath0 is
. moreover , the values of the @xmath39 and @xmath40 are proportional to @xmath41 and @xmath42 , hence they are quite small .
indeed in this case we find the gluino contribution to @xmath21 is of order @xmath43 . for the chargino contribution the amplitude
is given by @xcite @xmath44.\end{aligned}\ ] ] the complete expression for these function can be found in ref .
@xcite . for low and modurate values of @xmath45
the value of @xmath46 is much smaller than @xmath47 since it is suppressed by the ratio of @xmath41 to @xmath48 .
however , by neglecting the flavor mixing in the squark mass matrix @xmath47 turned out to be exactly real @xcite .
the imaginary part of @xmath47 is associated to the size of the intergenerational sfermion mixings , thus it is maximal for large @xmath45 .
in low @xmath45 case , that we consider , the imaginary part of @xmath47 is very small , and the gluino contribution is still the dominant contribution @xmath21 .
in particular , we found that is of order @xmath43 , which is less than the experimental value @xmath49 .
now we consider the effect of these two phases ( @xmath50 and @xmath51 ) on the direct cp violation parameter .
similar to the case of indirect cp violation parameter , in the gluino contribution the @xmath52@xmath52 transitions are almost real and the @xmath52@xmath53 transitions are suppressed by two up yukawa couplings and a down quark mass . moreover ,
the analysis of chargino contribution is also the same as in the indirect cp violation .
even the experimental bounds on the branching ratio of @xmath54 decay impose sever constraint on the lr transition .
therefore we do not find any significant susy cp violation effect in too .
in the previous section , we have shown that cp violation effects are always very small in susy models with universal soft susy breaking terms . recently
, it has been shown that the non universality of @xmath0terms is very effective to generate large cp violation effects @xcite .
in fact , the presence of non
degenerate @xmath0terms is essential for enhancing the gluino contributions to @xmath55 through large imaginary parts of the @xmath52@xmath53 mass insertions , @xmath56 and @xmath57 .
these susy contributions can , indeed , account for a sizeable part of the recently measured experimental value of @xmath55 @xcite . in the following
, we will present an explicit realization of such mechanism in the framework of a type i superstring inspired susy model . within this model ,
it is possible to obtain non universal soft breaking terms , i.e. scalar masses , gaugino masses and trilinear couplings .
type i string models contain nine branes and three types of five branes ( @xmath58 , @xmath59 ) .
if we assume that the gauge group @xmath60 is on one of the branes ( 9brane ) and the gauge group @xmath61 is on another brane ( @xmath62brane ) .
chiral matter fields correspond to open strings spanning between branes .
thus , they have non vanishing quantum numbers only for the gauge groups corresponding to the branes between which the open string spans . for example
, the chiral field corresponding to the open string between the @xmath63 and @xmath61 branes can have non trivial representations under both @xmath63 and @xmath61 , while the chiral field corresponding to the open string , which starts and ends on the @xmath63brane , should be an @xmath61singlet .
there is only one type of the open string which spans between the 9 and @xmath62branes , which we denote it as @xmath64 .
however , there are three types of open strings which could start and end on the 9brane , that is , the @xmath65 sectors ( i=1,2,3 ) , which corresponding to the @xmath66-th complex compact dimension among the three complex dimensions .
if we assign the three families to the different @xmath65 sectors we obtain non universality in the right handed sector .
in particular , we assign the @xmath67 sector to the third family and @xmath68 and @xmath69 to the first and second families respectively . under these assumption the soft susy breaking terms are obtained , following the formulae in ref . @xcite .
the gaugino masses are obtained @xmath70 while the @xmath0-terms are obtained as @xmath71 for the coupling including @xmath72 , i.e. the third family , @xmath73 for the coupling including @xmath74 , i.e. the second family and @xmath75 for the coupling including @xmath76 , i.e. the first family . here
@xmath77 is the gravitino mass , @xmath78 and @xmath79 are the cp phases of the f - terms of the dilaton field @xmath80 and the three moduli fields @xmath81 , and @xmath82 and @xmath83 are goldstino angles , and we have the constraint , @xmath84 .
thus , if quark fields correspond to different @xmath85 sectors , we have non universal a terms
. then we obtain the following a matrix for both of the up and down sectors , @xmath86 note that the non universality appears only for the right handed sector .
the trilinear susy breaking matrix , @xmath87 , itself is obtained @xmath88 in matrix notation .
in addition , soft scalar masses for quark doublets and the higgs fields are obtained , @xmath89 the soft scalar masses for quark singlets are obtained as @xmath90 if it corresponds to the @xmath91 sector . in models with non - degenerate @xmath0terms we have to fix the yukawa matrices to completely specify the model .
in fact , with universal @xmath0terms the textures of the yukawa matrices at gut scale affect the physics at ew scale only through the quark masses and usual ckm matrix , since the extra parameters contained in the yukawa matrices can be eliminated by unitary fields transformations . this is no longer true with non - degenerate @xmath0terms . here ,
we choose our yukawa texture to be y^u=
( m_u , m_c , m_t ) , y^d= k^ ( m_d , m_s , m_b ) k where @xmath7 is the ckm matrix . in this case one
find that the mass inserion @xmath92 can be written as @xcite @xmath93 where @xmath94 is an average squark mass and @xmath95 the quark mass .
this expression shows the main effects of the non
universal @xmath0terms . in the first place
, we can see that the diagonal elements are still very similar to the universal @xmath0terms situation .
apart of the usual scaling with the quark mass , these flavor diagonal mass insertions receive dominant contributions from the corresponding @xmath96 terms plus an approximately equal contribution from gluino to all three generations and an identical @xmath8 term contribution .
hence , given that the gluino rg effects are dominant , also the phases of these terms tend to align with the gluino phase , as in the minimal supergravity . therefore , edm bounds constrain mainly the relative phase between @xmath8 and gluino ( or chargino ) and give a relatively weaker constraint to the relative phase between @xmath97 ( the first generation @xmath0term ) and the relevant gaugino .
effects of different @xmath96 in these elements are suppressed by squared ckm mixing angles .
however , flavor off
diagonal elements are completely new in this model
. they do not receive significant contributions from gluino nor from @xmath8 and so their phases are still determined by the @xmath96 phases and , in principle , they do not directly contribute to edms .
= 11 cm in figure [ scat ] we show the allowed values for @xmath78 , @xmath98 and @xmath99 assuming @xmath100 . we have imposed the edm , and @xmath54 bounds with the usual bounds on susy masses .
we can see that , similarly to the minimal supergravity , @xmath101 is constrained to be very close to the gluino and chargino phases ( in the plot @xmath102 ) , but @xmath98 and @xmath99 are completely unconstrained . finally , in figure [ eps ] we show the values of @xmath103 versus the gluino mass in the same regions of parameter space and with the same constraints as in figure [ scat ] .
as we can see due to the effect of the off - diagonal phases a large percentage of points are above or close to @xmath104 , hence , sizeable supersymmetric contribution to @xmath105 can be expected in the presence of non - universal @xmath0terms .
non universal supersymmetry soft breaking terms are a natural consequence in many supergravity or string inspired susy models . moreover , the non universality of the @xmath0terms has a significant effect in the cp violation .
we have shown that in these models a sizeable supersymmetric contribution to cp observables and can be easily obtained .
i would like to thank s. barr , t. kobayashi , a. masiero and o. vives for their collaboration in this project .
i also would like to thank the organizers for such a nice and stimulating atmosphere in which the workshop took place .
+ a. alavi
( ktev coll . ) , .
( na48 coll . ) , .
g. dagostini , hep - ex/9910036 .
a. buras , m. jamin , and m.e : lautenbacher , ; m. ciuchini , e. franco , g. martinelli , and l. reina , ; s. bosh , a.j .
buras , m. gorbahn , s. jager , m. jamin , m.e .
lautenbacher , and l. silvestrini , ; m. ciuchini , e. franco , l. giusti , v. lubicz , and g. martinelli , hep - ph/9910237 ; m. jamin , hep - ph/9911390 . s. bertolini , m. fabbrichesi , and j.o .
eeg , _ rev .
mod . phys._*72 * ( 2000 ) 65 ; t. hambye , g.o .
kohler , e.a .
paschos , and p.h .
soldan , ; j.bijnens , and j.prades , jhep 01 , ( 1999 ) 023 ; e. pallante and a. pich , .
e. gabrielli and g.f .
giudice , ; erratum - ibid .
* b 507 * ( 1997 ) 549 .
s. abel and j. frere , .
s. khalil , t. kobayashi , and a. masiero , .
demir , a. masiero , and o. vives , ; d. a. demir , a. masiero , and o. vives , .
s. barr and s. khalil , .
a. brignole , l. e. , and c. , , erratum - ibid .
* b 436 * ( 1995 ) 747 .
l. e. , c. , and s. rigolin , .
s. khalil and t. kobayashi , .
s. khalil , t. kobayashi , and o. vives , .
e. gabrielli , s. khalil , e. torrente - lujan , hep - ph/0005303 .
branco , g.c .
cho , y. kizukuri and n. oshimo , .
a. masiero and h. murayama , .
r. barbieri , r. contino and a. strumia , ; + k. babu , b. dutta and r.n .
mohapatra , . | we study the predictions for the direct cp violation parameter in a class of string inspired model with non
universal soft @xmath0terms .
we show that the non universality of the @xmath0terms plays an important rule in enhancing the supersymmetric contributions to the cp violation . in particular , the supersymmetric contribution to can be of order the ktev result , @xmath1 .
# 1#2 # 1#2#3 _ ann .
phys .
( ny ) _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ ann .
rev .
nucl . part .
sci . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ nucl .
phys . _
* b # 1 * ( # 2 ) # 3 # 1#2#3 _ nucl .
phys . _
( proc .
suppl.)*b # 1 * ( # 2 ) # 3 # 1#2#3 _ phys .
lett . _
* b # 1 * ( # 2 ) # 3 # 1#2#3 _ phys .
rev . _
* d # 1 * ( # 2 ) # 3 # 1#2#3 _ phys .
rep . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ phys .
rev .
lett . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ prog .
theor .
phys . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ rev .
mod .
phys . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ zeit .
fr physik _ * c # 1 * ( # 2 ) # 3 # 1#2#3 _ mod .
phys .
lett . _ * a # 1 * ( # 2 ) # 3 # 1#2#3 _ sov .
j. nucl .
phys .
_ * # 1 * ( # 2 ) # 3 # 1#2#3 _ yad . fiz . _
* # 1 * ( # 2 ) # 3 # 1#2#3 _ nuovo cim . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ jetp lett . _ * # 1 * ( # 2 ) # 3 # 1#2#3 _ ibid .
_ * # 1 * ( # 2 ) # 3 |
nova explosions are caused by thermonuclear runaways on white dwarfs accreting hydrogen from a main sequence companion in a cataclysmic variable .
the explosive burning of hydrogen on the top of a co or an one degenerate core leads to the synthesis of some @xmath12 unstable nuclei , such as 13 , 14 , 15 , 17 and 18 .
these nuclei have short lifetimes ( @xmath51 - 2 minutes for 14 , 15 and 17 , @xmath515 min for 13 and @xmath53 hours for 18 ) , and they emit a positron when they decay . these positrons annihilate with electrons leading to the emission of photons with energy less or equal to 511 kev . furthermore , other medium- and long - lived radiocative nuclei are synthesized in classical novae : 7 ( @xmath13=77days ) , whichs emits a photon of 478 kev after an electron capture , 22 ( @xmath13=3.75yr ) and 26 ( @xmath13=1.04@xmath14 yr ) , which experience a @xmath12-decay emitting photons of 1275 and 1809 kev , respectively .
thus , classical novae are potential @xmath4-ray emitters , as was pointed out in previous works ( clayton & hoyle @xcite , clayton @xcite , leising & clayton @xcite )
. there are also some previous works concerning nucleosynthesis in classical novae ( see politano et al @xcite , prialnik & kovetz @xcite and references therein for nucleosynthesis in one and co novae , respectively , as well as hernanz et al @xcite , jos et al @xcite , jos & hernanz @xcite and @xcite ) .
but to our knowledge there are no previous works which follow all the phases ( from the accretion stage to the final explosion and mass ejection phases ) and couple them to the study of the @xmath4-ray emission of classical novae . in this paper
we have used realistic profiles of densities , velocities and chemical compositions ( obtained by means of a hydrodynamical code ) to determine the production and transfer of @xmath4-rays ( by means of a monte - carlo code ) during nova explosions . in this way , we are able to do a detailed analysis of the @xmath4-ray emission of classical novae and predict their detectability by instruments onboard the cgro ( see our previous paper by hernanz et al for our first results and their relation to the future mission integral ) .
we are mainly concerned by the emission and potential detectability of individual novae , related to the decay of the medium - lived nuclei 7 and 22 ( besides of the short - lived nuclei 13 and 18 ) .
in table 1 we present the main properties of the ejecta of some of the most representative models we have computed ( accretion rate [email protected]@xmath16 ) . there is an important difference between co and one novae : co novae are important producers of 7 ( thus line emission at 478 kev is expected during some days ) , whereas one novae are important producers of 22 ( thus line emission at 1275 kev during the first years after the explosion is expected ) .
this can clearly be seen by comparing the @xmath4-ray spectra of a typical co nova ( co1 ) and those of an one one ( one2 ) , shown in figure 1 for different epochs after the explosion . in all cases
a continuum component , mainly below 511 kev , appears as well as some lines ( 511 kev in all cases and 478 kev and 1275 kev in co and one novae , respectively ) .
two phases can be distinguished in the evolution of all models . during the early expansion , the ejected envelope is optically thick and there is an important contribution of the continuum below 511 kev , related to the comptonization of 511 kev photons .
later on , when the envelope becomes optically thin , absorption and comptonization become negligible ; therefore , the intensity of the lines is exclusively determined by the total mass of the radioactive nuclei 7 ( co novae ) and 22 ( one novae ) .
the main properties of the lines are shown in table 2 .
it is worth noticing that fluxes are rather low , due to the small ejected masses ( see table 1 ) .
this explains the null results of the observations of the 1275 kev emission by some novae made with comptel onboard cgro ( iyudin et al @xcite ) .
our predicted fluxes are fully compatible with the upper limits obtained in that work .
it is also worth noticing the important fluxes associated with the 511 kev line that we predict for all models ( see table 2 ) .
this flux should be detected by the current instruments onboard the cgro , provided that novae are observed very early after their explosion .
a retrospective analysis of the batse data would perhaps provide important clues to this issue ( see fishman et al @xcite ) .
we want to mention that there is a lack of agreement between all the theoretical models of novae ( including ours ) and some observations of ejected masses of classical novae , although large uncertainties affect sometimes the determination of observed ejected masses .
for instance , our model one1 fits quite well the observed abundances of the neon nova qu vul 1984 , but observations of this nova give an ejected mass that can reach @xmath17(saizar et al 1992 ) , almost two orders of magnitude larger than the theoretical one .
thus the flux of the 1275 kev line could be considerably larger , but no theoretical models are by now able to produce simultaneously such large ejected masses and neon in the ejecta .
.main properties of the ejecta one hour after peak temperature .
initial mass , total ejected mass and ejected mass of the most relevant radioactive nuclei are in and the mean kinetic energy of the ejecta , @xmath18 , is in erg.g@xmath16 [ cols="<,^,^,^,^,^,^,^ " , ] ( 15,8 ) ( 0,0)(7,8 ) ( 7.5,0)(7,8 ) in figure 2 we show the light curves of the 478 and 1275 kev lines for co and one models , respectively .
for the 478 kev line two different phases can be distinguished : during the first @xmath51.5 days , the line is completely dominated by the continuum generated by 13 and 18 decays , whereas later on the line follows the typical light curve of the 7 radioactive decay , with @xmath13=77 days .
the 1275 kev line has a rise phase lasting @xmath57days , followed by a decline related to the decay of 22 .
the small fluxes obtained make it impossible to detect this line with the current cgro instruments , unless a very close explosion occurs ( the same as for the future spi instrument onboard integral ) . concerning the light curve of the 511 kev line for all the models , as several isotopes , with different decay timescales ,
contribute to this emission , its temporal evolution is somewhat complex . besides one very early maximum appearing in the co novae ( related to 13 decay ) , one later maximum appears in all cases at @xmath5 6 hours ( see table 2 ) , which is related to 18 decay .
the further development of the light curves is different in the co and in the one cases : in the one case , as 22 emits a e@xmath19 when it decays , the emission at 511 kev lasts for a longer time ( as the 22 decay- timescale is much longer than the 18 one ) .
an early observation of a nova explosion , either of a co or an one one , would provide a positive detection of the 511 kev line ( an estimation of the detection distance of our novae by the future spi instrument onboard integral yields around 10 kpc for all of them except for the low - mass co1 model ) .
( 15,8 ) ( 0,0)(7,8 ) ( 7.5,0)(7,8 )
we have developed a monte - carlo code in order to treat the production and transfer of @xmath4-rays in nova envelopes , coupled to a hydrodynamical code that provides realistic profiles of all the relevant magnitudes .
thus a complete view of the nova explosion is obtained .
the most relevant features of the @xmath4-ray emission of classical novae are their intense 511 kev emission , lasting only some hours , the 478 kev emision in the case of co novae , related to 7 decay , and the 1275 kev emission in the case of one novae , related to 22 decay .
the 1275 kev emission is the most long - lived one , but the fluxes obtained are too low to be detected for novae at typical distances , thus explaining the negative detections of the novae observed up to now by comptel .
clayton , d.d . , and hoyle , f. _ apj _ * 187 * , l101 ( 1974 ) .
clayton , d.d .
, _ apj _ * 244 * , l97 ( 1981 ) .
leising , m. , and clayton , d.d . , _ apj _ * 323 * , 159 ( 1987 ) .
politano m. , starrfield s. , truran j.w .
, weiss a. , sparks w.m .
, _ apj _ * 448 * , 807 ( 1995 ) .
prialnik d. , kovetz a. , _ apj _ * 477 * , 356 ( 1997 ) .
hernanz m. , jos j. , coc a. , isern j. , _ apj letters _ * 465 * , l27 ( 1996 ) .
jos j. , hernanz m. , coc a. , _ apj letters _ * 479 * , l55 ( 1997 ) .
jos j. , hernanz m. , _ nuclear physics _ , in press ( 1997 ) .
jos j. , hernanz m. , in preparation .
hernanz m. , gmez - gomar , j. , jos , j. , and isern , j. , _ the transparent universe _ , esa sp-382 , noordwijk : esa publications division , 1997 , pp .
saizar , p. , et al ,
_ apj _ * 398 * , 651 ( 1992 ) .
iyudin a.f . ,
et al , _ a&a _ * 300 * , 422 ( 1995 ) .
fishman g.j . ,
et al , _ gamma - ray line astrophysics _ , new york : aip , 1991 , pp .
190 - 192 . | an implicit hydrodynamic code following the explosion of classical novae , from the accretion phase up to the final ejection of the envelope , has been coupled to a monte - carlo code able to simulate their gamma - ray emission .
carbon - oxygen ( co ) and oxygen - neon ( one ) novae have been studied and their gamma - ray spectra have been obtained , as well as the gamma - ray light curves for the important lines ( e@xmath0e@xmath1 annihilation line at 511 kev , @xmath2be decay - line at 478 kev and @xmath3na decay - line at 1275 kev ) .
the detectability of the emission by cgro instruments has been analyzed .
it is worth noticing that the @xmath4-ray signature of a co nova is different from that of an one one . in the co case
, the 478 kev line is very important , but lasts only for @xmath52 months . in the one case
, the 1275 kev line is the dominant one , lasting for @xmath5 4 years . in both cases ,
the 511 kev line is the most intense line at the beginning , but its short duration ( @xmath52 days ) makes it very difficult to be detected .
it is shown that the negative results from the observations made by comptel up to now are consistent with the theoretical predictions .
predictions of the future detectability by the integral mission are also made .
7@xmath2be 13@xmath6n 14@xmath7o 15@xmath8o 17@xmath9f 18@xmath10f 22@xmath3na 26@xmath11al |
during the last decade , the growing availability of large databases , the increasing computing powers , as well as the development of reliable data analysis tools , have constituted a better machinery to explore the topological properties of several complex networks from the real world @xcite .
this has allowed to study a large variety of systems as diverse as social , biological and technological .
the main outcome of this activity has been to reveal that , despite the inherent differences , most of the real networks are characterized by the same topological properties , as for instance relatively small characteristic path lengths and high clustering coefficients ( the so called _ small - world property _ ) @xcite , _ scale - free _ degree distributions @xcite , degree correlations @xcite , and the presence of motifs @xcite and community structures @xcite .
all such features make real networks radically different from regular lattices and random graphs , the standard topologies usually used in modeling and computer simulations .
this has led to a large attention towards the comprehension of the evolution mechanisms that have shaped the topology of a real network , and to the design of new models retaining the most significant properties observed empirically .
_ spatial networks _
are a special class of complex networks whose nodes are embedded in a two or three - dimensional euclidean space and whose edges do not define relations in an abstract space ( such as in networks of acquaintances or collaborations between individuals ) , but are real physical connections @xcite .
typical examples include neural networks @xcite , information / communication networks @xcite , electric power grids @xcite and transportation systems ranging from river @xcite , to airport @xcite and street @xcite networks .
most of the works in the literature , with a few relevant exceptions @xcite , have focused on the characterization of the topological properties of spatial networks , while the spatial aspect has received less attention , when not neglected at all .
however , it is not surprising that the topology of such systems is strongly constrained by their spatial embedding .
for instance , there is a cost to pay for the existence of long - range connections in a spatial network , this having important consequences on the possibility to observe a small - world behavior .
moreover , the number of edges that can be connected to a single node is often limited by the scarce availability of physical space , this imposing some constraints on the degree distributions . in few words ,
spatial networks are different from other complex networks and as such they need to be studied in a different way .
in this paper we focus on a particular class of spatial networks : _ networks of urban street patterns_. we consider a database of 1-square mile samples of different world cities and for each city we construct a spatial graph by associating nodes to street intersections and edges to streets . in this way , each of the nodes of the graph is given a location in a 2-dimensional square , and a real number , representing the length of the corresponding street , is associated to each edge . by construction ,
the resulting graphs are _ planar graphs _ , i.e. graphs forming nodes whenever two edges cross . after a previous work on the distribution of centrality measures @xcite , here
we present a comparative study of the basic properties of spatial networks of different city street patterns .
in particular we evaluate the characteristics of the graphs both at a global and at a local scale .
the main problem with spatial graphs is that , in most of the cases , the random graph or the complete graph are no more a good way to normalize the results .
in fact , the common procedure in relational ( non - spatial ) complex networks is to compare the properties of the original graph derived from the real system with those of some randomized versions of the graph , i.e. of graphs with the same number of nodes and links as the original one , but where the links are distributed at random .
this is , for instance , the standard way proposed by watts and strogatz in ref .
@xcite to assess whether a real system is a small world .
one quantifies the structural properties of the original graph by computing its characteristic path length @xmath0 and clustering coefficient @xmath1 , where @xmath0 measures the typical separation between two vertices in the graph ( a global property ) , whereas @xmath1 measures the cliquishness of a typical neighbourhood ( a local property ) .
then , the graph is a small - world if @xmath0 assume a value close to that obtained for the randomized version of the graph , @xmath2 , while the value of @xmath1 is much larger than @xmath3 .
similarly , in the efficiency - based formalism proposed in refs .
@xcite , a small - world network is defined as a system being extremely efficient in exchanging information both at a global and at a local scale . again , the values of global and local efficiency are compared with those obtained for a randomized version of the graph .
a similar method is used in the counting of short cycles or specific motifs in a graph representing a real system @xcite .
the research of the motifs and cycles is based on matching algorithms counting the total number of occurrences of each motif and each cycle in the original graph and in the randomized ones .
then , a motif or a cycle is statistically significant if it appears in the original graph at a number much higher than in the randomized versions of the graph . in a planar graph ,
as those describing urban street patterns , the randomized version of the graph is not significative because it is almost surely a non - planar graph due to the edge crossings induced by the random rewiring of the edges .
moreover , because of the presence of long - range links , a random graph corresponds to an extremely costly street pattern configuration , where the cost is defined as the sum of street lengths @xcite .
the alternative is to compare urban street patterns with grid - like structures .
following ref .
@xcite , we shall consider both _
minimum spanning trees _ and _ greedy triangulations _ induced by the real distribution of nodes in the square .
spanning trees are the planar graphs with the minimum number of links in order to assure connectedness , while greedy triangulations are graphs with the maximum number of links compatible with the planarity .
spanning trees and greedy triangulations will serve as the two extreme cases to normalize the structural measures we are going to compute .
the database we have studied consists of twenty 1-square mile samples of different world cities , selected from the book by allan jacobs @xcite .
we have imported the twenty maps into a gis ( geographic information system ) environment and constructed the correspondent spatial graphs of street networks by using a road - centerline - between - nodes format @xcite .
namely , each urban street pattern is trasformed into a undirected , valued ( weighted ) graph @xmath4 , embedded in the 2-dimensional unit square . in fig .
[ maps ] we show the case for the city of savannah : in the upper - left panel we report the original map , and in upper - right panel the obtained graph .
@xmath5 is the set of @xmath6 nodes representing street intersections and characterized by their positions @xmath7 in the square .
@xmath8 is the set of @xmath9 links representing streets .
the links follow the footprints of real streets and are associated a set of real positive numbers representing the street lengths , @xmath10 .
the graph is then described by the adjacency @xmath11 matrix @xmath12 , whose entry @xmath13 is equal to 1 when there is an edge between @xmath14 and @xmath15 and 0 otherwise , and by a @xmath11 matrix @xmath0 , whose entry @xmath16 is equal to the length of the street connecting node @xmath14 and node @xmath15 . in this way
both the topology and the geography ( metric distances ) of the system will be taken into account .
a list of the considered cities is reported in table [ table1 ] , together with the basic properties of the derived graphs .
the considered cases exhibit striking differences in terms of cultural , social , economic , religious and geographic context .
in particular , they can be roughly divided into two large classes : 1 ) patterns grown throughout a largely self - organized , fine - grained historical process , out of the control of any central agency ; 2 ) patterns realized over a short period of time as the result of a single plan , and usually exhibiting a regular grid - like , structure .
ahmedabad , cairo and venice are the most representative examples of self - organized patterns , while los angeles , richmond , and san francisco are typical examples of mostly - planned patterns .
we have selected two different parts of the city of irvine , ca , ( named irvine 1 and irvine 2 ) for two highly diverse kinds of urban fabrics : the first is a sample of an industrial area showing enormous blocks with few intersections while the second is a typical residential early sixties `` lollipop '' low density suburb based on a tree - like layout with a lot of dead - end streets .
the differences between cities are already evident from the basic properties of the derived graphs .
in fact , the number of nodes @xmath6 , the number of links @xmath9 , and the cost of the wiring , defined as the sum of street lenghts : @xmath17 and measured in meters , assume widely different values , notwithstanding the fact we have considered the same amount of land . notice that ahmedabad has 2870 street intersections and 4387 streets in a surface of 1-square mile , while irvine has only 32 intersections and 37 streets .
ahmedabad and cairo are the cities with the largest cost , while the cost is very small ( less than 40000 meters ) in barcelona , brasilia , irvine , los angeles , new delhi , new york , san francisco , washington and walnut creek .
a large difference is also present in the average edge ( street ) length @xmath18 , that assumes the smallest values in cities as ahmedabad , cairo and venice , and the largest value in san francisco , brasilia , walnut creek and los angeles . in ref .
@xcite we have studied the edges length distribution @xmath19 for the two different classes of cities , showing that self - organized cities show single peak distributions , while mostly planned cities exhibit a multimodal distribution , due to their grid pattern .
.basic properties of the planar graphs obtained from the twenty city samples considered .
@xmath6 is the number of nodes , @xmath9 is the number of edges , @xmath20 and @xmath18 are respectively the total length of edges and the average edge length ( both expressed in meters ) , @xmath21 is the box - counting fractal dimension . [ cols="<,<,^,^,^,^,^ " , ] in table [ table3 ] we report the values of efficiency obtained for each city and for the respective mst and the gt .
the values of @xmath22 and @xmath23 serve to normalize the results , being respectively the minimum and the maximum value of efficiency that can be obtained in a planar graph having the same number of nodes as in the original graph of the city .
notice that irvine 2 is the only case in which @xmath24 .
this is simply due to the fact that irvine 2 is the only city whose corresponding graph is not connected .
consequently , the mst has a smaller number of edges but a higher value of efficiency because it is , by definition , a connected graph .
the main result is that the cities considered , despite their inherent differences , achieve a relatively high value of efficiency , which is in most of the cases about @xmath25 of the maximum value of the efficiency in a planar graph , @xmath23 .
following ref .
@xcite we define the relative efficiency @xmath26 as : @xmath27 of course , the counterpart of an increase in efficiency is an increase in the cost of construction , i.e. an increase in the number and length of the edges .
the cost of construction can be quantified by using the measure @xmath20 defined in formula ( [ eq : cost ] ) . given a set of @xmath6 nodes , the shortest ( minimal cost ) planar graph that connects all nodes correspons to the mst , while a good approximation for the maximum cost planar graph is given by the gt .
we thus define a normalized cost measure , @xmath28 , as : @xmath29 by definition the mst has a relative cost @xmath30 , while gt has @xmath31 . in fig .
[ erel ] we plot for each city @xmath26 as a function of @xmath28 . , and relative efficiency , @xmath26 . each point in the plot
represent a city .
the point of coordinates ( 0,0 ) would correspond to the cost / efficiency of the mst while the point ( 1,1 ) would correspond to the gt network .
irvine 2 , having coordinates ( 0.175,-0.398 ) , i.e. a negative value of relative efficiency , has been plotted instead as having coordinates ( 0.175,0 ) . ]
the cities can be a - priori divided into different classes : 1 ) medieval fabrics , including both arabic ( ahmedabad and cairo ) and european ( bologna , london , venice , and vienna ) ; 2 ) grid - iron fabrics ( barcelona , los angeles , new york , richmond , savannah and san francisco ) ; 3 ) modernist fabrics ( brasilia and irvine 1 ) ; 4 ) baroque fabrics ( new delhi and washington ) ; 5 ) mixed fabrics ( paris and seoul ) ; 6 ) `` lollipop '' layouts ( irvine 2 and walnut creek ) .
we shall see that the plot @xmath26 vs. @xmath32 has a certain capacity to characterize the different classes of cities listed above .
the plot indicates an overall increasing behavior of @xmath26 as function of @xmath28 , with a saturation at @xmath33 for values of @xmath34 .
grid - iron patterns exhibit a high value of relative efficiency , about @xmath35 of the efficiency of the gt , with a relative cost which goes from @xmath36 to @xmath37 .
the three grid - iron cities ( new york , savannah and san francisco ) with the largest value of efficiency , @xmath33 , have respectively a cost equal to 0.342 , 0.354 and 0.383 .
medieval patterns have in general a lower cost and efficiency than grid - iron patterns althouh , in some cases as ahmedabad and cairo ( the two medieval cities with the largest efficiency ) , they can also reach a value of @xmath33 with a smaller cost equal to 0.29 .
modernist and `` lollipop '' layouts are those with the smallest value of cost but also with the smallest value of efficiency .
we have proposed a method to characterize both the local and the global properties of spatial graphs representing urban street patterns .
our results show that a comparative analysis on the structure of different world cities is possible by the introduction of two limiting auxiliary graphs , the mst and the gt , a certain level of structural similarities across cities as well as some difference are well captured by counting cycles and by measuring normalized efficiency and cost of the graphs .
the method can be applied to other planar graphs of different nature , as highway or railway networks .
r. albert and a .-
barabsi , rev .
mod . phys . * 74 * , 47 ( 2002 ) .
newman , siam review * 45 * , 167 ( 2003 ) .
r. pastor - satorras , a. vespignani , _ evolution and structure of the internet : a statistical physics approach _ , ( cambridge university press , 2004 ) .
s. boccaletti , v. latora , y. moreno , m. chavez and d .- u .
hwang , phys .
rep . in press .
d.j . watts and s.h .
strogatz , nature * 393 * , 440 ( 1998 ) .
barabsi and r. albert , science * 286 * , 509 ( 1999 ) .
r. pastor - satorras , a. vzquez , and a. vespignani , phys .
87 * , 258701 ( 2001 ) .
r. milo , s. shen - orr , s. itzkovitz , n. kashan , d. chklovskii , and u. alon , science * 298 * , 824 ( 2002 ) .
s. wasserman and k. faust , _ social networks analysis _ , ( cambridge university press , cambridge , 1994 ) .
o. sporns , complexity * 8 * , 56 ( 2003 ) . s .-
h . yook , h. jeong , and a
barabsi , proc .
sci . u.s.a . * 99 * , 13382 ( 2002 ) .
v. latora and m. marchiori , phys . rev . *
e71 * , 015103(r ) ( 2005 ) .
r. kinney , p. crucitti , r. albert and v. latora , eur .
j. * b46 * , 101 ( 2005 ) .
f. pitts , the professional geographer * 17 * , 15 ( 1965 ) .
r. guimer , s. mossa , a. turtschi , and l.a.n .
amaral , proc .
usa * 102 * 7794 ( 2005 ) .
a. barrat , m. barthlemy , r. pastor - satorras and a. vespignani .
usa * 101 * , 3747 ( 2004 ) .
s. porta , p. crucitti and v. latora , preprint physics/0506009 . in press in environment and planning
* b * j. buhl , j. gautrais , r.v .
sol , p. kuntz , s. valverde , j.l .
deneubourg , and g. theraulaz , eur .
j. * b42 * , 123 ( 2004 ) .
m. t. gastner and m. e. j. newman , cond - mat/0407680 .
p. crucitti , v. latora and s. porta , preprint physics/0504163 .
v. latora and m. marchiori , phys .
* 87 * , 198701 ( 2001 ) .
v. latora and m. marchiori , eur .
j. * b32 * , 249 ( 2003 ) .
a. jacobs , great streets ( mit press , boston , ma , 1993 ) .
n. dalton , j. peponis and r. dalton , to tame a tiger one has to know its nature : extending weighted angular integration analysis to the description of gis road - centerline data for large scale urban analysis , proceedings of the 4th international space syntax symposium , london , uk 2003 .
s. h. strogatz , _ nonlinear dynamics and chaos with applications to phsyics , biology , chemistry and engineering _ ( perseus , 1994 ) .
west , _ introduction to graph theory _ ,
( prentice hall , 1995 ) .
b. bollobs , _ modern graph theory _
( graduate texts in mathematics , springer - verlag , new york , 1998 ) .
dikerson , r.l.s .
drysdale , s.a .
mcelfresh , e.welzl , computational geometry * 8 * , 67 ( 1997 ) t.h .
cormen , c.e .
leiserson , r.l .
rivest , c. stein , _ introduction to algorithms _
( mit university press , cambrdige , 2001 ) . j. b. kruskal , proc .
* 2 * , 48 ( 1956 ) .
r. dalton , environment and behavior * 35 * 107 ( 2003 ) .
b. hillier and i. shinichi , paper presented at cosit 2005 , conference on spatial information theory , sept .
14 - 18 , ellicottville , new york .
m. southworth and e. ben - joseph , _ streets and the shaping of town and cities _ , ( island press , washington d.c . , 2003 ) .
s. marshall 2004 , _ streets and patterns _ , ( routledge , london , uk , 2004 ) . in the computed gt ,
the graphs we have obtained have less than @xmath38 edges .
this is due to the fact that we have considered all the edges as being straight - lines connecting the two end - nodes .
for such a reason some of edges connecting the external nodes ( the nodes on the border of the unit square ) can not be placed without causing edge crossings . however , the number of these edges is of the order of @xmath39 , so that in large gt graphs the meshedness is close to 1 .
n. alon , r. yuster and u. zwick , algorithmica * 17 * , 209 ( 1997 ) .
a. penn , environment and behavior , * 35 * , 30 ( 2003 ) .
teklemburg , h.j .
timmermans and a.f .
van wagemberg , environment and planning * b20 * 347 ( 1993 ) . | recent theoretical and empirical studies have focused on the structural properties of complex relational networks in social , biological and technological systems . here
we study the basic properties of twenty 1-square - mile samples of street patterns of different world cities .
samples are represented by spatial ( planar ) graphs , i.e. valued graphs defined by metric rather than topologic distance and where street intersections are turned into nodes and streets into edges .
we study the distribution of nodes in the 2-dimensional plane .
we then evaluate the local properties of the graphs by measuring the meshedness coefficient and counting short cycles ( of three , four and five edges ) , and the global properties by measuring global efficiency and cost . as normalization graphs , we consider both minimal spanning trees ( mst ) and greedy triangulations ( gt ) induced by the same spatial distribution of nodes .
the results indicate that most of the cities have evolved into networks as efficienct as gt , although their cost is closer to the one of a tree .
an analysis based on relative efficiency and cost is able to characterize different classes of cities . |
studying the helmholtz equation at the high - frequency regime is fundamental in both the theoretical understanding of the corresponding solutions and the derivation of appropriate numerical schemes .
the well - know asymptotic expansions developed by melrose and taylor @xcite have significantly contributed in this matter and were the key in the design of several high - frequency integral equation methods . indeed , integral equation methods are very efficient and widely used in the solution of acoustic scattering problems ( see e.g. @xcite and the references therein )
. however , the resulting linear systems are dense , ill - conditioned and with large size in particular when the frequency increases .
several effective strategies have been proposed to overcome these difficulties @xcite . despite this significant progress ,
integral formulations are limited at higher frequencies since the numerical resolution of field oscillations can easily lead to impractical computational times .
this is why hybrid numerical methods based on a combination of integral equations and asymptotic methods have found an increasing interest for the solution of high - frequency scattering problems .
indeed , the methodologies developed in this connection that specifically concern scattering from a smooth convex obstacle were first introduced in @xcite .
several other works followed these @xcite and mainly consist of improving and analyzing this kind of numerical algorithms in single and multiple scattering configurations .
all these methods are mainly based on construction of an appropriate ansatz for the solution of integral equations in the form of a highly oscillatory function of known phase modulated by an unknown slowly varying envelope , which is expected to generate linear systems quasi - independent of the frequency .
.,height=226 ] the high - frequency integral equation methods mentioned above use the asymptotic expansions developed in the well - known paper by melrose and taylor @xcite in the context of convex obstacles . from these expansions , an ansatz is derived and incorporated into integral equations to eliminate the highly oscillatory part of the unknown which usually corresponds to the physical density , normal derivative of the total field , computed on the surface of the obstacles .
this surface is decomposed into three regions , the illuminated and shadows regions in addition to the deep shadow one .
each region is then numerically treated differently and the ansatz is set in general on the illuminated one .
although carefully designed , the aforementioned high - frequency integral equation formulations result in ill - conditioned matrices that limit the numerical accuracy of the approximate solutions .
one explanation lies in the fact that the rapidly decaying behavior of the unknown density in the deep shadow regions is not incorporated into the approximation spaces as it is not intrinsic to the chosen ansatz . generally speaking , it is not clear how to extract all the information needed from the leading term in the expansion given in @xcite , which restricts the construction of the ansatz . in this paper
, we derive new expansions of the normal derivative of the total field using approximations of the dirichlet to neumann ( dtn ) operator .
the original expansions employed a pseudo - differential decomposition of the dtn operator , and the related analysis focuses on the behavior of this field around the shadow boundary which leads to a corrected formula for the kirchhoff approximation around this region @xcite .
however , it has been shown that these expansions are valid in the entire surface of the obstacles @xcite . here
, we choose first and second order approximations of the dtn operator of the bayliss - turkel type @xcite . to obtain these new expansions ,
we follow a similar procedure to the one given in @xcite .
briefly , it consists of first finding the kernel of a certain operator , which allows the computation of its amplitude , and then use the stationary phase method to get the final expansions around the shadow boundary . in this case , we can use some of the results derived by melrose and taylor @xcite in our analysis .
the resulting expansions can then be used to appropriately build an ansatz that contains the expected behavior of the solution in the three regions , namely , the illuminated and deep shadow regions in addition to the shadow boundaries .
this provides an improvement over the usual ansatz that behaves like kirchhoff approximations , meaning that the corresponding solutions are accurate mostly in the illuminated regions .
this paper is organized as follows . after reviewing the functional setting needed for this analysis , we state the problem and explain our choice , regarding the approximation of the dtn operator , in the second section .
the two following sections are , respectively , devoted to the derivation of asymptotic expansions in the context of first and second order approximations of the dtn operator .
the last section is reserved for some conclusions . in this work
, we will use the following functional spaces ( for more details , see for instance @xcite ) .
let @xmath0 be an open bounded set of @xmath1 .
* @xmath2 : space of smooth test functions with compact support , from @xmath0 to @xmath1 .
* @xmath3 : space of distributions . *
@xmath4 : schwartz space or space of rapidly decreasing functions on @xmath1 .
* @xmath5 : the space of tempered distributions , which is the dual space of @xmath6 .
* @xmath7 : space of compactly supported distributions .
* @xmath8 : space of pseudo - differential operators of order @xmath9 . * @xmath10 : space of fourier integral operators of order @xmath9
. we will also use symbols of h@xmath11rmander s classes @xcite , we say @xmath12 if and only if @xmath13 in particular we say @xmath14 if @xmath15 . note that each @xmath16 admits an asymptotic expansion of the form @xmath17 for @xmath18 large where the @xmath19 are homogeneous functions of degree @xmath20 in @xmath21 .
finally , if @xmath14 , we say @xmath22 , where @xmath23 is the corresponding pseudo - differential operator .
consider a convex obstacle @xmath24 such that @xmath25 is a hypersurface and let @xmath26 be the exterior domain given by @xmath27 .
we are interested in solutions of the following wave equation @xmath28 u(x , t)=-\delta(t - x \cdot \omega ) = - u^{i } ( x , t ) \quad \text{on}\ b \times \mathbb{r } , \end{array}\right.\ ] ] where @xmath29 is the incident wave , @xmath30 denotes the incidence direction , @xmath31 is the dirac function , and @xmath32 @xcite . defining the function @xmath33 by @xmath34 leads to the well - posed problem @xcite @xmath35 w(x , k)=-e^{ikx \cdot \omega } \quad \text{on}\ b \times \mathbb{r } , \\ [ 4pt ] w(x , k)=\mathcal{o } ( |x|^{-2n-1/2})\ \mbox{and}\ \displaystyle ( \partial_{|x|}-ik)w(x , k)=o ( |x|^{-2n-1/2})\ \mbox{for}\ |x|\rightarrow \infty .
\end{array } \right.\label{problem}\ ] ] for each @xmath36 , @xmath37 denotes the outgoing normal vector . in what follows ,
we use the notation @xmath38 indicating the total field , where @xmath39 is the scattered field solution of the problem and @xmath40 .
in addition , we write the normal derivative of the total field as @xmath41 = a_q ( x , k)e^{ik x \cdot \omega},\hspace{-0.2 cm } \end{aligned}\ ] ] that we express with respect to the problem , and using the same notation given in @xcite , by the operator @xmath42 where @xmath43 @xcite stands for the dirichlet to neumann operator @xmath44 and @xmath45 is a fourier integral operator to be defined later . using a pseudo - differential decomposition of this operator dtn , melrose and taylor derived the well - known expansion @xmath46 where @xmath47 as @xmath48 , and it is rapidly decreasing in the sense of schwartz as @xmath49 .
the real - valued function @xmath50 is positive on the illuminated region , negative on the shadow region and vanishes precisely to first order on the shadow boundary @xcite . here ,
@xmath51 result from the application of the stationary phase method and the expansion of the symbol of the operator @xmath52 @xcite .
as is mentioned in @xcite , the first term of the expansion represents the classical kirchhoff approximations . indeed ,
if @xmath53 is replaced by the leading term in its asymptotic expansion @xmath54 and taking @xmath55 in the illiminated region @xmath56 @xmath57 , we obtain @xmath58 generally speaking , the aforementioned high - frequency integral equation methods , based on an ansatz of the form @xmath59 , although delivering better accuracy than the kirchhoff approach in the illuminated region , they were designed replicating its behavior , and this explains why the decay in the deep shadow region is not observed but somehow forced .
this is partly due to the fact that an explicit form of the leading term in the asymptotic expansion is not available .
we propose in this work to use approximations of the dtn map to derive new asymptotic expressions of @xmath60 .
this can allow construction of a new ansatz in order to improve the behavior and the accuracy of the solution in the shadow and the deep shadow regions .
we use first and second order approximations of the dtn operator given by bayliss - turkel @xcite @xmath61 { \partial_{\bf n}}w^s(x , k ) = -ikw^i(x , k)+\frac{c(x)}{2}w^i(x , k)-\frac{c(x)^{2}}{8(c(x)-ik)}w^i(x , k)-\frac{1}{2(c(x)-ik)}\partial _ { x}^{2}w^i(x ,
k ) , \label{ordre2}\end{aligned}\ ] ] where @xmath62 represents the curvature .
although these conditions are approximations of the dtn operator , we use the sign `` = '' for the sake of the presentation . the motivation behind this choice of conditions and is summarized in the next example .
suppose that @xmath26 is a circle , in this case the exact solution of problem is given by bessel functions .
we then compute the quantity @xmath63 using , ( 1 ) the exact solution , ( 2 ) the first order approximation approximating @xmath64 , and ( 3 ) the kirchhoff approximations .
the resulting calculations are exhibited in figure [ fig : modes ] .
we can observe that the kirchhoff approach produces a good approximation in the illuminated region , but starts to degrade in the shadow boundaries to completely deteriorate in the deep shadow region in contrast with the solution , based on condition , where we can observe a satisfactory approximation in those regions .
[ rem1 ] the condition was derived in the case of two and three dimensions while the condition was derived only in the two dimensional case @xcite .
our computations do not distinguish between these two cases .
however , at the end of this analysis , we give an example of an expansion derived for a three dimensional second order approximation of the dtn operator .
the results produced in this paper are based on some of the results derived in the paper @xcite .
our analysis starts by determining the kernel associated with the operator in the case where the dtn operator is approximated by . for the sake of the presentation
, we use the notation @xmath65 . let @xmath66 be
a strictly convex bounded obstacle such that @xmath67 , where @xmath68 is @xmath69 hypersurface in @xmath70 .
suppose that @xmath71 is an open set of @xmath70 such that @xmath72 , and let @xmath39 be a solution of ( [ problem ] ) . using the approximation ( [ ordre1 ] ) , the operator @xmath52 can be written as @xmath73 where @xmath74 is its kernel given by @xmath75 and @xmath76 is the kernel of the fourier integral operator @xmath45 . using the approximation and the definition of the total field
, we can write @xmath77 to obtain the kernel of the operator @xmath52 , we compute the fourier transform with respect to the variable @xmath78 . let @xmath79 , we have then @xmath80 \nonumber = \int_\mathbb{r}\partial_{\bf n } w^t(x , k)\widehat{\varphi } ( x , k)dk \hspace{1 cm } \\[2pt ] \nonumber=\int_{\mathbb{r}\times \mathbb{r}}\partial_{\bf
n } w^t(x , k)\varphi ( x , t)e^{-ikt}dk dt \hspace{-0.5 cm } \\[2pt ] \nonumber=\int_{\mathbb{r}\times \mathbb{r}}\left[-ik(1-{\bf n } \cdot \omega)e^{ikx \cdot \omega}+\mathcal{c}(x ) e^{ikx\cdot \omega}\right ] \varphi(x , t)e^{-ikt}dkdt \hspace{-4.3cm}\\[2pt ] \nonumber = \langle ( \partial_t(1-{\bf n } \cdot \omega)+\mathcal{c}(x))\delta(t-\omega \cdot x),\varphi ( x , t ) \rangle_{s',s } \hspace{-2.6cm}\\[2pt ] = \left\langle\kappa_q(x , t),\varphi(x , t)\right\rangle_{s',s},\label{souad1 } \hspace{1.cm}\end{aligned}\ ] ] where @xmath81 and @xmath82 .
therefore @xmath83 and @xmath84 . in the following ,
we use the same decomposition of the operator @xmath45 given in @xcite ( equation ( 5.9 ) ) , that is , @xmath85 with @xmath86 , @xmath87 , and @xmath88 and @xmath89 are elliptic fourier integral operators of order 0 .
the operators @xmath90 and @xmath91 are fourier airy integral operators and are defined by @xmath92 @xmath93 @xmath94 where @xmath95 , @xmath96 , @xmath97 is an integer indicating the order of the derivative , @xmath98 is the airy function @xmath99 with @xmath100 and @xmath101 is its derivative . finally , we define the operator @xmath102 @xcite as follows @xmath103 the next theorem is concerned with the computation of the amplitude of the operator .
[ th4 ] let @xmath89 and @xmath88 be elliptic fourier integral operators of order 0 .
then the operator @xmath52 and its kernel @xmath104 can be respectively written as @xmath105 @xmath106 such that @xmath107 with a symbol @xmath108 , @xmath109 with a symbol @xmath110 , @xmath111 , @xmath112 , with @xmath113 , and @xmath102 is a pseudo - differential operator defined by @xcite .
in addition , its amplitude is given by @xmath114 with @xmath115 , @xmath116 and @xmath117 .
using ( [ f ] ) and equation we obtain @xmath118 where @xmath119 , @xmath120 , @xmath121 , and @xmath122 .
the operators @xmath90 and @xmath123 are airy operators given by .
using theorem 6.5 in @xcite , we can write @xmath124 with @xmath125 and @xmath126 . to compute the oscillatory integral related to @xmath52 , consider @xmath127 and the dirac delta function @xmath128 used here to find the kernel of @xmath52 at the base point @xcite , we have then @xmath129 where @xmath130 and @xmath131 with @xmath132 and @xmath133 such that @xmath134 and @xmath135 , and knowing that @xmath102 commute with @xmath136 and @xmath137 @xcite .
here , @xmath138 , @xmath139 indicates the dual of @xmath140 , and the phase function @xmath141 is defined in the three regions of the obstacle @xcite . in the illuminated region @xmath142 , it is given by @xmath143 while in the shadow region @xmath144 , we have @xmath145 finally , on the shadow boundary @xmath146 , the phase function is as follows @xmath147 since @xmath148 @xcite . here , @xmath149 , @xmath150 such that @xmath151 , @xmath152 , and @xmath153 .
the pseudo - differential operator @xmath136 is defined by @xmath154 where @xmath155 is the dual couple of @xmath156 and @xmath157 .
our objective is to compute @xmath158 .
first , using standard calculations on composition of operators @xcite , we have @xmath159 with @xmath160 thus @xmath161 where @xmath162 . to find @xmath163 we need to replace @xmath164 by @xmath165 in ( [ j0p ] ) where @xmath166 we get @xmath167 a similar approach for @xmath168 gives @xmath169 the equation becomes @xmath170\frac{1}{a_{+}}(k^{-1/3}\xi_{1 } ) dk d\xi , \label{q}\ ] ] where @xmath171 and @xmath172 .
this shows that the kernel @xmath104 is as follows @xmath173 d\xi dk.\ ] ] taking now the inverse fourier transform of @xmath104 , we obtain the following amplitude @xmath174 d\xi.\ ] ] applying the change of variable @xmath115 with @xmath175 , we find @xmath176 d\zeta , \label{aq}\ ] ] such that @xmath177 , @xmath178 and @xmath179 .
the remaining part of the computation of the asymptotic expression of @xmath180 consists of applying the stationary phase method .
first , we need the next lemma @xcite .
the function @xmath181 defined as follows @xmath182 is rapidly decreasing for @xmath183 , where @xmath184 . the asymptotic expansion of the amplitude @xmath180 is given by @xmath185 such that @xmath186 and where @xmath187 , @xmath188 , @xmath189 is a constant depending on @xmath190 and @xmath191 , @xmath30 is the incidence direction , and @xmath192 is a continuous real function that is positive on the illuminated region , negative on the shadow region , and vanishing on the shadow boundary .
the functions @xmath51 and @xmath193 result from the expansion of the symbols @xmath194 and @xmath195 ( see theorem [ th4 ] ) and the application of the stationary phase method . first , let us note that @xmath196 using and , @xmath180 becomes @xmath197 \psi(k^{1/3}t)dt d\zeta , \label{aq2}\end{aligned}\ ] ] where @xmath198 and @xmath199 are given in theorem [ th4 ] .
let us assume now that @xmath200 using properties of pseudo - differential operators @xcite , we can write @xmath201 with @xmath202 , and @xmath203 this shows that the integral can be rewritten as @xmath204 such that @xmath205 see @xcite for more details regarding the definition of @xmath206 . to get the asymptotic expansion of @xmath180 , it remains to apply the stationary phase method to @xmath207 the conditions for this application are satisfied and the computation of the critical points is done in @xcite . using standard calculations regarding the stationary phase method @xcite , we obtain @xmath208 where @xmath209 , and @xmath210 with @xmath211 a constant . using ( [ a_pl ] ) in ( [ aq2 - 2 ] ) , we obtain @xmath212 with @xmath186 and @xmath188 and @xmath191 is a real number .
the next theorem establishes a relation between the functions @xmath51 and @xmath193 found in .
[ thab ] the functions @xmath51 and @xmath193 in , @xmath213 , satisfy the equation @xmath214 from equation and some operators used in theorem [ th4 ] ( @xmath45 , @xmath88 , @xmath137 and @xmath136 ) , we have @xmath215 similar calculations to the equation give @xmath216 with @xmath217 , where @xmath218 and @xmath219 are given in theorem [ th4 ] . knowing that @xmath220 and @xmath221 implies that @xmath222 .
we know that @xmath223 , and then usual pseudo - differential calculus results in @xmath224 .
this allows us to conclude that @xmath225 .
two estimates are established in this subsection .
for the completion of the paper , we recall the next lemma @xcite .
let @xmath226 , @xmath227 is rapidly decreasing for @xmath183 and @xmath228 the next result compares the asymptotic expansion with @xmath60 .
if @xmath180 is the amplitude given by ( [ sum.aq ] ) , then @xmath229 where @xmath230 is a real constant and @xmath231 we know that @xmath232 using theorem [ thab ] , the expansion can be written as @xmath233 from the preceding lemma , we have @xmath234 for @xmath235 and taking @xmath236 @xcite , we obtain @xmath237 knowing that @xmath238 for each @xmath239 and @xmath240 , we get @xmath241 and @xmath230 is a real constant .
the following result estimates near the shadow boundary .
if @xmath180 is the amplitude given by , then @xmath242 is a real constant . from the definition of @xmath243
, it follows that @xmath244 with @xmath245 and @xmath246 .
we know that in the shadow boundary @xmath247 .
we assume that the curvature is constant , so @xmath248 , then @xmath249 and all its derivatives are bounded , @xmath250 for all @xmath251 , where @xmath252 is a real constant .
thus @xmath253 and @xmath254 .
taking @xmath255 , we obtain
we are now interested in computing the asymptotic expansion in the case where we have @xmath256 = -ikw^i(x , k)+\frac{c(x)}{2}w^i(x , k)-\frac{1}{2(c(x)^{2}+k^{2})}(c(x)+ik)(\frac{c(x)^{2}}{4}+ \partial^{2}_{x})w^i(x , k ) .
\label{appr02}\ ] ] as for the first order case , we compute first the kernel associated to the operator @xmath52 in the case where the dtn is approximated by . for the sake of simplicity , we denote the operator by @xmath257 .
let @xmath66 be a strictly convex bounded obstacle such that @xmath67 , where @xmath68 is a @xmath69 hypersurface in @xmath70 .
suppose that @xmath71 is an open set of @xmath70 such that @xmath72 .
let @xmath39 be a solution of ( [ problem ] ) . using the approximation ( [ appr02 ] )
, the operator @xmath52 can be written as @xmath258 where @xmath259 is its kernel given by @xmath260 and @xmath261 denotes the convolution operator of @xmath262 , @xmath263 , and @xmath264 where @xmath76 . using the approximation and the definition of the total field ,
we can write @xmath265 with @xmath266 @xmath267 to obtain the kernel of the operator @xmath52 , we compute the fourier transform of the amplitude @xmath268 with respect to @xmath269 .
let @xmath270 , thus @xmath271 the quantity @xmath272 is given by .
for the one regarding @xmath273 , we have @xmath274 \nonumber = -\int_{\mathbb{r}\times\mathbb{r}}\left[\frac{1}{2(c(x)^{2}+k^{2})}(c(x)+ik)(\frac{c(x)^{2}}{4}+\partial_{x}^{2})e^{ikx \cdot \omega}\right ] \varphi_x(t)e^{-ikt}dtdk \hspace{-0.7cm}\\ [ 2pt ] \nonumber = -\int _ { \mathbb{r}}\left[\int_{\mathbb{r}}\frac{1}{2(c(x)^{2}+k^{2})}e^{-ikt}dk \ast \int_{\mathbb{r}}(c(x)+ik)(\frac{c(x)^{2}}{4}+\partial_{x}^{2})e^{ikx \cdot \omega}e^{-ikt}dk \right]\varphi_x(t)dt \hspace{-2.7 cm } \\[2pt ] = -\frac{\pi}{2c(x ) } \langle e^{-c(x)|t|}\ast \left(c(x)-\partial_t\right)(\frac{c(x)^{2}}{4}+\partial_{x}^{2 } ) \delta(t-\omega x),\varphi \rangle_{s',s } \nonumber\hspace{0.6cm}\\[2pt ] = \left\langle \kappa _ { { q}_2}(x , t),\varphi\right\rangle_{s',s}.\hspace{6.8cm}\end{aligned}\ ] ] taking @xmath275 we get @xmath276 where @xmath277 indicates the kernel of @xmath257 .
this allows us to write @xmath278 where @xmath261 is the convolution operator of @xmath262 and @xmath279 .
the next theorem is concerned with the computation of the amplitude of the operator .
let @xmath89 and @xmath88 be elliptic fourier integral operators of order 0 . then the operator @xmath52 and its kernel @xmath104 can respectively be written as @xmath280 @xmath281 where @xmath277 is the kernel of @xmath257 , and @xmath282 is defined by @xmath283\frac{1}{a_+}(k^{-1/3}\xi_1 ) d\xi dk dr .
\nonumber \hspace{-2 cm } \end{aligned}\ ] ] here @xmath284 , @xmath285 , @xmath286 , and @xmath287 where @xmath288 and @xmath289 are defined in theorem [ th4 ] .
furthermore , the amplitude @xmath180 is given by @xmath290 where @xmath291 is the amplitude and @xmath292\frac{1}{a_{+}}(k^{2/3}\zeta_{1 } ) d\zeta , \nonumber \end{aligned}\ ] ] such that @xmath293 , @xmath294 , and @xmath295 is given in theorem [ th4 ] . using ( [ f ] ) and
( [ qapp2 ] ) we obtain @xmath296 where @xmath297 , @xmath298 , @xmath299 , and @xmath300 . using theorem 6.5 in @xcite
, we get @xmath280 such that @xmath284 with symbol @xmath301 and @xmath302 with symbol @xmath303 where @xmath304 and @xmath305 are described in theorem [ th4 ] . as in the first order case , we use now the dirac delta function @xmath306 to find the kernel of @xmath52 at the base point .
let @xmath307 , thus we have @xmath308\delta _ { ( 0,t)},\varphi \rangle_{s',s}\nonumber \hspace{-2.5cm}\\=\langle q_1\delta _ { ( 0,t)},\varphi\rangle_{s',s } + \langle q_2\delta _ { ( 0,t)},\varphi\rangle_{s',s } , \hspace{3.5cm}\label{q2int}\end{aligned}\ ] ] with @xmath309 and @xmath310 . the term @xmath311 is already computed .
therefore , we only need @xmath312 with @xmath313 on the other hand , we can write @xmath314\frac{1}{a_+ } ( k^{-1/3}\xi_1 ) d\xi dk , \end{aligned}\ ] ] where @xmath315 and @xmath316 .
this leads to @xmath317\frac{1}{a_+ } ( k^{-1/3}\xi_1 ) d\xi dk dr .
\end{aligned}\ ] ] finally , we find @xmath318\frac{1}{a_+}(k^{-1/3}\xi_1)d\xi dk dr .
\nonumber \hspace{-2 cm } \label{q_2}\end{aligned}\ ] ] this shows that the kernel @xmath104 is as follows @xmath319 and @xmath320\frac{1}{a_+}(k^{-1/3}\xi_1)d\xi dk dr , \nonumber \hspace{-2 cm } \end{aligned}\ ] ] with @xmath286 and @xmath287 . to obtain the amplitude @xmath180
, we take the inverse fourier transform of @xmath104 . the one related to @xmath321
is given by .
first , we can write that @xmath322 where @xmath323 is the kernel of the operator . using the inverse fourier transform , we find @xmath324 knowing that @xmath325 , we obtain @xmath326\frac{1}{a_{+}}(k^{-1/3}\xi_{1})d\xi .
\nonumber\end{aligned}\ ] ] applying the change of variable @xmath327 with @xmath328 , we get @xmath292\frac{1}{a_{+}}(k^{2/3}\zeta_{1})d\zeta , \label{aq222}\end{aligned}\ ] ] such that @xmath329 , @xmath330 , and @xmath295 is described in theorem [ th4 ] .
the next theorem gives the asymptotic expansion of @xmath331 .
the asymptotic expression of the amplitude is given by @xmath332 -\frac{1}{2}\frac{c(x)}{c^2(x)+k^2}\sum_{p , l=0}^{p , l}k^{2/3-p-2l/3}\left[b^{\#}_{p , l}(\omega , x)- \frac{1}{c(x)}a^{\#}_{p , l}(\omega , x)\right]\psi^{(l)}(k^{1/3}z)e^{ikx\omega } + r_{p , l}(k ) , \label{2020}\end{aligned}\ ] ] where @xmath333 here @xmath334 , @xmath335 , @xmath187 , @xmath188 , @xmath30 is the incidence direction , and @xmath336 is the curvature .
in addition , @xmath192 is a continuous real function that is positive on the illuminated region , negative on the shadow region , and vanishing on the shadow boundary .
the functions @xmath51 , @xmath193 , @xmath337 , and @xmath338 result from the expansion of the symbol and the application of the stationary phase method .
the derivation of is based on the application of the stationary phase method to the amplitude @xmath339 where @xmath340 \psi(k^{1/3}t)dt d\zeta,\end{aligned}\ ] ] see , and @xmath341\psi(k^{1/3}t ) d\zeta dt , \end{aligned}\ ] ] obtained using in .
the critical points are the same as the ones given in the paper @xcite and used when applying the stationary phase method to to derive .
therefore , the latter method for the amplitude @xmath180 leads to the asymptotic expression @xmath342\psi^{(l)}(k^{1/3}z(\omega , x))e^{ikx \cdot \omega } \\
+ r_{p , l}(k),\nonumber \end{aligned}\ ] ] such that @xmath343 is ( first order expansion ) , @xmath334 , and @xmath335 .
the remainder @xmath344 satisfies @xmath345 and @xmath189 is a constant depending on @xmath191 and @xmath190 .
as is mentioned in remark [ rem1 ] , the second order condition is derived in two dimensions .
if the three dimensional absorbing boundary condition @xmath346 is used ( see condition ( 29 ) in@xcite ) , then we obtain @xmath347\psi^{(l)}(k^{1/3}z(\omega , x ) ) , \end{aligned}\ ] ] where @xmath343 is given by with @xmath348 , @xmath349 , and @xmath350 .
in this paper we have derived some new expansions of the normal derivative of the total field solution of the helmholtz equation .
the original expansions are based on a pseudo - differential decomposition of the dirichlet to neumann operator . in this work ,
we have used approximations of this operator to derive new expansions .
one of the goals is to facilitate construction of a new ansatz class that can be used in the development of numerical solvers that can produce more accurate solutions .
y. boubendir gratefully acknowledges support from nsf through grant no .
dms-1319720 .
antoine x. and darbas m. , _ generalized combined field integral equations for the iterative solution of the three - dimensional helmholtz equation _
, mathematical modeling and numerical analysis , 41 , 2007 , pp .
147167 .
x. antoine , a. bendali and m. darbas .
_ analytic preconditioners for the boundary integral solution of the scattering of acoustic waves by open surfaces
_ , 2005 , journal of computational acoustics , volume 13(3 ) , pages 477498 .
anand , a. , ovall , j. , turc , c. , _ well - conditioned boundary integral equations for two - dimensional sound - hard scattering problems in domains with corners _ , to appear in journal integral equations and applications , 2011 .
tong , m.s . , chew , w.c .
: multilevel fast multipole acceleration in the nystrm discretization of surface electromagnetic integral equations for composite objects , ieee trans .
antennas and propagation * 58 * ( 2010 ) , no .
10 , 34113416 .
y. boubendir , c. turc . _ wavenumbers estimates for regularized combined field boundary integral operators in acoustic scattering problems with neumann boundary conditions , _
i m a numerical analysis , volume 33 issue 4 october 2013 .
h@xmath11rmander , l. _ pseudo - differential operators and hypoelliptic equations .
singular integrals_. ( proc
pure math .
x , chicago , ill . , 1966 ) , 138183 ; amer . math . soc . , providence , r.i .
( 1967 ) .
a. anand , y. boubendir , f. ecevit and f. reitich , _ analysis of multiple scattering iterations for high - frequency scattering problems .
ii : the three - dimensional scalar case_. numerische mathematik , 114:373 - 427 , 2010 .
o. p. bruno , c. geuzaine , j - a . monro and f. reitich , _ prescribed error tolerance within fixed computational times for scattering problems of arbitrarily high frequency : the convex case_. phil .
london 362 ( 2004 ) , 629 - 645 .
abboud , t. , ndlec , j .- c .
, zhou , b. _ improvement of the integral equation method for high frequency problems _ , in mathematical and numerical aspects of wave propagation : mandelieu - la napoule , siam , ( 1995 ) , 178187 . | this paper is concerned with the asymptotic expansions of the amplitude of the solution of the helmholtz equation .
the original expansions were obtained using a pseudo - differential decomposition of the dirichlet to neumann operator .
this work uses first and second order approximations of this operator to derive new asymptotic expressions of the normal derivative of the total field .
the resulting expansions can be used to appropriately choose the ansatz in the design of high - frequency numerical solvers , such as those based on integral equations , in order to produce more accurate approximation of the solutions around the shadow and deep shadow regions than the ones based on the usual ansatz . |
it has been said that we entered the era of precision cosmology .
this is mainly due to observations of the cosmic microwave background radiation ( cmb ) .
as the cmb power spectrum is very sensitive to cosmological parameters it provides very tight constraints .
currently the errors are at the level of a few percent @xcite .
the cmb power spectrum is shaped by several processes that can be divided into two groups .
the first group involves processes that occurred before and during the generation of the cmb .
this part is well understood as the universe at that times was very close to homogeneous and contribution from the spatial curvature and dark energy was negligible .
the constraints on cosmological parameters coming just from this group of processes ware recently presented and discussed in @xcite .
the second group of processes takes into account what happens with photons between the last scattering surface and the observer .
this includes the distance to the last scattering surface , the isw effect , the impact of reionization on the cmb , the sunyaev - zeldovich and the lensing effects .
the analysis of this second group of effects is a subject to systematics , as the geometry of the late time universe does not have to be close to homogeneous and isotropic robertson - walker geometry ( these processes are most often studied within the framework of linear perturbations around the homogeneous friedmann models ) . on small scales ( say @xmath4 , i.e. @xmath5 ) , expect for the distance , these processes should not significantly modify the overall shape of the cmb power spectrum .
the distance to the last scattering surface affects the mapping of cmb temperature fluctuations into the angular power spectrum @xmath1 .
this effect can be modelled by introducing the shift parameter @xmath6 , which is proportional to the distance to the last scattering surface @xcite .
therefore , there is a number of papers where the constraints from the cmb are just limited to the constraints on the shift parameter .
this is usually done when testing different models of dark energy @xcite , alternative cosmological models @xcite , or gpc - scale inhomogeneous models @xcite . as discussed in @xcite this type of analysis
can be improved but taking into account both the scale of the sound horizon at the last scattering and @xmath6 .
this paper studies the effect of the distance but does not consider any alternative cosmologies .
here we just focus on the effect of small - scale inhomogeneities on the distance to the last scattering .
the presence of inhomogeneities is know to affects the distance - redshift relation @xcite .
some studies claim that the effect is large @xcite , others that it is small @xcite . even if the change of the distance just at the level of a few percent @xcite this analysis is still important as a change of the distance by a few percent leads to a similar change of the inferred values of cosmological parameters .
one may think that the effect of inhomogeneities is taken into account in the standard analysis of the cmb via the isw or lensing effects .
however , in both cases ( in the standard approach ) this is done by using the matter power spectrum .
thus , this analysis is insensitive to the change of the mean , i.e. the uniform change of the distance to the last scattering surface .
for example the lensing analysis deals with the change of the distance but is only sensitive to the change of the variance , not the mean ( cf .
if the change of distance is relatively uniform ( i.e. with a negligible variance ) then no additional temperature anisotropies are generated .
since the change of distance affects the mapping of the physical position of the peaks to the peaks in the angular power spectrum @xmath1 this effect needs to be considered also at the level of the mean . therefore , if inhomogeneities are not taken into account when calculating the distance , then they may become a major source of systematics . with the increasing precision of cmb experiments , in particular planck ,
a proper understanding of systematics is important . without taking into account all systematics
the precision cosmology will not be an accurate cosmology .
therefore , this paper aims to study how the presence of inhomogeneities affects the distance - redshift relation , in particular the distance to the last scattering surface .
several models are considered and it is shown that the change of distance due to inhomogeneities is not negligible . the structure of this paper is as follows : sec .
[ distance ] discusses different methods for the distance calculation , sec .
[ dissec ] applies these methods to calculate the distance to the last scattering surface , sec .
[ labdc ] explores the implication of differences in the distance , and sec .
[ conk ] concludes the results .
as shown by sachs @xcite , the equation for the angular diameter distance @xmath7 is @xmath8 where @xmath9 is a parameter along the geodesic , @xmath10 is a null vector that is tangent to the geodesic , @xmath11 is the shear of the light bundle , @xmath12 is the ricci tensor and @xmath13 . in the comoving and synchronous
coordinates , for pressure - less matter , @xmath14 . as seen , the distance depends on density fluctuations , and secondly , even if the perturbations vanish after averaging ( i.e. @xmath15 , where @xmath16 is the background density ) they do modify the distance relation and the final result deviates from the homogeneous solution .
equation ( [ dsr ] ) is the differential equation so it is sensitive to any change of initial conditions .
therefore , it is sensitive to @xmath17 as this term acts as a potential ( for a detailed discussion see @xcite ) . as the parameter @xmath9 is not directly measured ( unless the geodesic parametrization is chosen in a particular way ) , it is convenient to express it in terms of a redshift .
the redshift is given by the following relation @xcite @xmath18 where @xmath19 is the 4-velocity , and subscripts @xmath20 and @xmath21 refer to emission and observation instants respectively .
in the comoving and synchronous coordinates @xmath22 and @xmath23 where @xmath24 . thus , for a given model and using the above relation the parameter @xmath9 can be replaced with @xmath25 .
when homogeneity is assumed , i.e. @xmath26 , and @xmath27 ( where @xmath28 is the scale factor ) then @xmath29 and ( [ dsr ] ) reduces to @xmath30 where @xmath31 is the hubble constant , @xmath32 , @xmath33 , and @xmath34 . the dyer
roeder approach also assumes homogeneity but takes into account that light propagates through vacuum .
therefore , @xmath35 which photons ` feel ' is different than true @xmath35 .
this is modelled by a constant parameter @xmath36 ( of value between 0 and 1 ) that multiplies @xmath35 . in this case
( [ dsr ] ) reduces to @xcite @xmath37 where @xmath38 in order to include the evolution of density fluctuations along the line of sight , ref .
@xcite suggests the following modification of the dyer
roeder equation @xmath39 where @xmath40 is the mean of the present - day density fluctuations along the line of sight .
initial conditions needed to solve ( [ ddre ] ) and ( [ mddre ] ) are : @xmath41 and @xmath42 . within the linear perturbative scheme
the distance is @xmath43 where @xmath44 the distance in the homogeneous universe and is given by ( [ dlcdm ] ) .
the most general form for @xmath45 is presented in @xcite . neglecting the contribution from the observer s and source s motion , and taking the leading term
, @xmath45 can be expressed as @xmath46 where @xmath47 is the comoving coordinate @xmath48 , @xmath49 is the gravitational potential that relates to the density perturbations @xmath50 via the poisson equation @xmath51 .
the above expression is equivalent to the convergence in the lensing approximation and is known as the born approximation .
as seen , propagation through voids ( @xmath52 ) increases the distance while propagation through overdense regions ( @xmath53 ) decreases the distance . in order to solve ( [ ddba ] )
one needs to know exactly what the density fluctuations along the line of sight are .
however , within the weak lensing analysis one does not solve ( [ ddba ] ) directly .
instead the variance is calculated , thus instead of @xmath50 one has @xmath54 which can be expressed in terms of the matter power spectrum . in this paper
we are interested in the overall ( uniform ) change of the distance .
thus , we will solve ( [ ddba ] ) directly , this requires knowledge of density fluctuations along the line of sight .
the algorithm is described below , but see also @xcite for an alternative approach .
to test what is the effect of inhomogeneities on the distance to the last scattering surface and the accuracy of the cmb analysis , let us consider 7 different ways of calculating the distance to the last scattering surface 1 . the @xmath3cdm model : + distance to the last scattering surface is calculated from ( [ dlcdm ] ) .
2 . dyer roeder approximation ( model dr ) : + based on the cosmological observations , yu et al .
@xcite suggest that @xmath55 .
therefore , the distance to the last scattering surface is calculated using ( [ ddre ] ) with @xmath55 .
3 . modified dyer roeder relation ( model mdr ) : + the modified version of the dyer
roeder equation ( [ mddre ] ) is solved with @xmath56 .
lensing approximation ( model len 1 ) : + in order to solve ( [ ddba ] ) , one needs to know @xmath57 along the line of sight . using the poisson equation the gravitational potential
is related to density fluctuations .
the density fluctuations along the line of slight are generated from the log - normal pdf the algorithm is discussed in details in [ alen1 ] .
lensing approximation ( model len 2 ) : + in model len 1 the mean of density fluctuations along the line of sight is negligible .
however , as pointed out in @xcite the mean of density fluctuations along the line of sight does not have to vanish , as @xmath58 ( 3d volume average ) does not necessarily imply @xmath59 ( 1d along the line of sight ) . in model len 2 ,
@xmath60 although @xmath58 .
the algorithm is discussed in details in [ alen2 ] .
lensing approximation ( model len 3 ) : + model len 3 uses a more flexible ( than len 1 & 2 ) method of generating density fluctuations along the line of sight . the algorithm is presented in details in [ alen3 ] . 7 .
fully non - linear model ( model sc ) : + the above approaches were based either on the assumption of homogeneity ( the @xmath3cdm and dyer roeder relation ) or on the linear approximation ( models len 13 ) .
in fact , the lensing approximation is not even fully linear , as it takes into account just the leading linear term .
also , when generating fluctuations along the line of sight , the matching conditions were not taken into account , i.e. the structures were not properly matched .
in addition , the light rays were not properly match as well , i.e. when light ray exits one structure and enters another one , the null geodesics need to be properly matched .
therefore , a fully non - linear model is considered here , where the distance is calculated directly from ( [ dsr ] ) and all matching conditions are handled properly .
the model is based on the swiss - cheese model .
each inhomogeneous patch is of spherical symmetry and is described using the lematre - tolman ( lt ) model @xcite .
thus , density within each inhomogeneous patch changes continuously , unlike in the lensing approximation , where within each generated structures , matter density is constant .
also the evolution of matter is calculated using the exact method , not the linear approximation .
the properties of the model are chosen so that the mean of density fluctuations along the line of sight is zero , i.e. @xmath61 .
thus , this model allows us to test how large non - linear contributions are , in contrast to @xmath62 contributions . the detailed description of this model is presented in [ asc ] .
in the vicinity of the last scattering surface . ] in this section it will be assumed that the large - scale universe is exactly the @xmath3cdm , so @xmath63 km s@xmath64 mpc@xmath64 , @xmath65 , @xmath66 .
a sample of density fluctuations along the line of sight is presented in figure [ fig1 ] .
the results , in the form of @xmath67 , are presented in figure [ fig2 ] .
the distance to the last scatting surface is presented in the inset , and in the form of the shift parameter @xmath6 in the 2nd column of table [ tab1 ] .
as seen , the distance to the last scattering surface depends on the choice of the method . even within the most optimistic case , model len 1
, the change is of order of @xmath68 .
however , the non - linear contributions ( even if @xmath61 ) gives @xmath69 change of the distance . when @xmath62 , even within the lensing approximation the change is of order of @xmath70 .
another important fact is that the variance within each method is small .
models len 13 allow for random ( within the limits of the method ) density fluctuations along the line of sight . based on 100,000 runs ( each with a different distribution of density fluctuations along the line of sight ) the results are len 1 : @xmath71 , len 2 : @xmath72 , len 3 : @xmath73 .
this shows that the change of distance is relatively uniform .
below , when referring to models len 1 - 3 it will be referred to the mean of @xmath45 . for model len 1 changes to @xmath74 and @xmath75 when parameters are as in table [ tab1 ] and table [ tab2 ] respectively . ] in the next section we will study the implications of the change of the distance to the last scattering surface and the cmb constraints when this change is taken into account .
in this section we assume that the physical densities of baryonic and cold dark matter , and primordial power spectrum are the same as in the standard cosmological model , and only implications of the change of the distance to the last scattering surface are going to be examined .
the change of the distance - redshift relation leads to a different mapping of the cmb temperature fluctuations into the angular power spectrum @xmath1 .
thus , if the distance - redshift relation changes , then the standard analysis of the cmb needs to be adjusted . the way to do this is following : first the cmb power spectrum is calculated using standard codes ( for example camb @xcite ) then the output requires following scaling @xmath76 thus , the @xmath77 axis changes while the amplitude @xmath78 remains relatively unchanged ( for a detailed discussion see @xcite ) . to present how it works
let us consider the following example : let us change the value of @xmath3 in such a way so that the distance to the last scattering instant , for studied models , is the same as in the standard cosmological model .
@xmath79 needed to fit the distance , for studied models , is presented in the 3rd column of table [ tab1 ] .
apart from model len 1 , @xmath79 is much larger than when homogeneity is assumed . if the power spectrum was calculated for a homogeneous friedmann model ( i.e. with the distance - redshift relation the same as in the friedmann model ) the result would look like the one presented in figure [ fn1 ] .
however , after applying scaling ( [ cmbscaling ] ) the angular power spectrum is as in figure [ fig3 ] .
as seen the cmb power spectrum is very similar except for low-@xmath77 where the isw effect is different .
the low-@xmath77 power spectrum is presented in the inset , where the error - bars are enlarged to include the cosmic variance @xmath80 .
another example : let us correct the distance by changing the cosmological parameters but keeping the spatial curvature zero , i.e. @xmath81 .
thus , the change of @xmath35 automatically implies the change of @xmath79 . also to keep the shape of the cmb power spectrum unchanged , @xmath82 must not change , so the change of @xmath35 automatically implies the change of @xmath31 .
the change of cosmological parameters needed to fit the distance is presented in table [ tab2 ] . as seen all models except for model len 1
can be ruled out because of too high amplitude of @xmath31 . even within model len 1
, the change of parameters is large compared to the current precision for example in model len 1 , @xmath35 is around @xmath83 not @xmath84 so the difference is around 7% , which is twice as large as the error estimated by the wmap team @xcite . for completeness the cmb power spectrum for len 1 and the @xmath3cdm model are presented in figure [ fig4 ] .
the above examples show why it is important to have an accurate method for calculating the distance . as the universe is inhomogeneous , the distance - redshift relation is different than the distance - redshift relation in the friedmann model .
as shown , even a small change of the distance can have a large impact on the accuracy of the cmb analysis . as seen from table [ tab1 ]
a change of the distance by @xmath68 leads to the 0.5% change of the cosmological constant .
although this change is small , in the era of precision cosmology this may be important . especially that as shown in table [ tab2 ] if one puts by hand the assumption of spatial flatness , then systematics are much larger .
.the effect of inhomogeneities on the distance to the last scattering surface
2nd column ( @xmath85 ) ; the accuracy of estimation of @xmath3 3rd and 4th column . in all models @xmath63 km s@xmath64 mpc@xmath64 , @xmath86 , @xmath87 . [ cols="<,^,^,^ " , ] ) employed ) .
the inset presents the power spectrum for low-@xmath77 with errors enlarged to include the cosmic variance . ]
let us now focus only on models len 1 , len 2 , and sc .
we exclude the dyer roeder relation as
this is only a rough approximation ( as the shear accumulates along geodesics this approximation is expected to be inaccurate for large distances ) . also as model len 3 is rather of arbitrary matter fluctuations ( it was considered only to check how arbitrariness of matter fluctuations changes the results ) it is excluded from the further and more detailed analysis . in this section
we will only focus on len 1 , len 2 , and sc models .
len 1 model is a conservative approach
it is based on the linear approach with random density fluctuations along the line of sight .
len 2 model is also based on linear approximation , but the mean of density fluctuations along the line of sight is non - zero ( as discussed in @xcite if structures in the universe are organized , then density fluctuations are no longer purely random , hence even though the 3d average vanish , the average of density fluctuations over the the line of sight is not zero ) .
sc model is a fully non - linear model with vanishing density fluctuations along the line of sight .
the results of the cmb analysis within these 3 models is presented in fig .
now we allow @xmath35 and @xmath79 to vary freely ( however , we keep @xmath88 and @xmath89 unchanged , so the change of @xmath35 implies the change of @xmath31 ) .
as seen the constraints obtained within len 1 model are very similar to the constraints obtained within the standard cosmological model .
however , model len 1 does not include non - linear corrections .
if these are taken into account or if only the mean of density fluctuations along the line of sight does not vanish then spatial curvature can not be zero .
as seen the presence of small - scale non - linear structures changes the distance - redshift relation and now the cmb data combined with local value of @xmath31 no longer favors spatially flat models .
this shows that the flat friedmann solution is not preferred by the data , and when inhomogeneities are included into the distance - redshift relation then @xmath90 models with larger amount of dark energy are favored .
let us discuss for a moment the possibility that in fact @xmath91 .
what are the implications ? in this case , the age of the universe increases to approximately @xmath92 gyr .
this change is acceptable .
an increased amount of dark energy should be detectable by other cosmological observations .
the supernova data alone are not very precise , but still homogeneous friedmann models that fit the data best are the elliptic , not flat , models with @xmath93 @xcite .
also , the bao analysis is in the agreement with larger amount dark energy .
however , it should be noted that one of the elements of the bao analysis is a conversion from the redshift to distance . this conversion is obtained using the friedmann models .
therefore , if the distance - redshift relation is changed then the result of the bao analysis can also be different .
this should be kept in mind when analyzing the bao data , especially that as shown in @xcite ignoring inhomogeneities also leads to additional systematics .
this paper investigated how the presence of inhomogeneities ( small - scale inhomogeneities observed in the universe ) affects the distance to the surface of last scattering , and how via the change of the distance the inhomogeneities influence the cmb data . in the standard approach
, the distance is calculated within a framework of homogeneous friedmann models .
there are several reasons for using this approach .
first of all , there is an argument used by weinberg @xcite who pointed out that although for a single case the distance is modified by the inhomogeneities , but due to photon conservation , when averaged over large enough angular scales the overall effect is zero ( however see @xcite for an argument why weinberg s reasoning should not apply ) .
second of all , as seen from ( [ ddba ] ) if the amount of voids is the same as amount of overdensities then @xmath45 should be zero and the distance should be exactly the same as in the friedmann model . as density fluctuations should vanish after averaging over sufficiently large scales it seems reasonable to expect that the friedmann distance - redshift relation correctly describes the reality
. however , there is a difference between vanishing 3d average and vanishing of the average of the density fluctuations over the line of sight . if density fluctuations in the universe are purely random then vanishing 3d average implies vanishing average over the line of sight .
if density fluctuations are not purely random , i.e. if there is some degree of organization , then vanishing 3d average does not necessarily imply vanishing density fluctuations over the line of sight ( for a discussion see @xcite ) . as the large - scale matter distribution in the universe has a form of the cosmic web ,
there is some kind of organization , and thus the result of ( [ ddba ] ) does not have to be zero , even if 3d average of density fluctuations vanishes after averaging over sufficiently large scales .
this is the case of model len 2 .
also , relation ( [ ddba ] ) does not apply when structures are non - linear .
the calculations within model sc were carried out within the non - linear regime .
in addition matter distribution was chosen so that the density fluctuations along the line of sight vanish after averaging ( i.e. @xmath94 ) . in this case , again , the distance was not the same as within the friedmann model . even if the change of the distance is small , like in len 1 model , still the effect is important , as we have already entered the era of precision cosmology as seen from table [ tab1 ] and table [ tab2 ] the change of the inferred cosmological parameters , even for len 1 model , can be of order of a percent or more .
if the change of the distance is larger , for example as in models len 2 and sc , then the inferred parameters are significantly different . in this case
spatially flat models are ruled out and the data favors positively curved models with with @xmath95 . as discussed in sec .
( [ dissec ] ) , within a chosen method , the distance to the last scattering surface is relatively uniform with only small variance .
the presented , in this paper , results show that the change of the mean is important , and the proper handling of this effect is essential , otherwise the systematics may be larger than the precision of measurements . the only problem , however , is to have a good method for the calculation of the distance .
all models presented in this paper have their limitations , and should rather be treated as toy models , i.e. as examples to show the significance of the problem .
the results show that the accuracy of the cmb analysis strongly depends on the accuracy of the calculation of the distance corrections .
peebles described the importance of accuracy and precision in cosmology with the following example _ a digital scale may read out the weight of an object to many significant figures , in a precise measurement .
but if the scale is not well calibrated the measurement may not be very accurate _ @xcite . in our case , the ` calibration ' requires an inhomogeneous framework that will allow us to estimate the distance with inhomogeneities taken into account .
as the non - linear corrections are important this can not be done within the lensing approximation ( [ ddba ] ) but requires a fully inhomogeneous framework .
only then we will be able to accurately predict the distance and say : ` the distance is larger by @xmath96 ' or ` @xmath97 than when homogeneity is assumed ' . with the increasing precision of cmb experiments , in particular planck , atacama cosmology telescope , and south pole telescope
this is important .
otherwise , what would be the meaning of a measurement with very high precision ( @xmath98 statistical errors ) if the accuracy is low ( @xmath69 systematics ) ?
i would like to thank richard bond , chris clarkson , timothy clifton , ruth durrer , pedro ferreira , valerio marra , teppo mattsson , syksy rsnen , marco regis , david wiltshire for useful discussions and suggestions .
this research of was supported by the marie curie fellowship ( pief - ga-2009 - 252950 ) .
10 komatsu e , smith k m , dunkley j , bennett c l , gold b , et al 2010 , arxiv:1001.4538 vonlanthen m , rsnen s , durrer r 2010 _ j. cosmol .
_ jcap08(2010)023 bond j r , efstathiou g , tegmark m 1997 _ mon . not .
r. astron .
soc . _ * 291 * l33 efstathiou g , bond j r , 1999 _ mon . not .
r. astron .
soc . _ * 304 * 75 melchiorri a , mersini l , dman c j , trodden m 2003 _ phys .
_ d * 68 * 043509 kurek a , szydowski m 2008 _ astrophys .
j. _ * 675 * 1 lazkoz r , maartens r , majerotto e 2006 _ phys .
_ d * 74 * 083510 rydbeck s , fairbairn m , goobar a 2007 _ j. cosmol .
_ jcap05(2007)003 alnes h , amarzguioui m , grn 2006 _ phys .
_ d * 73 * 08351 alnes h , amarzguioui m 2007 _ phys .
_ d * 75 * 023506 bolejko k , wyithe j s b 2009 _ j. cosmol . astropart .
_ jcap02(2009)020 clifton t , ferreira p g , zuntz j 2009 _ j. cosmol .
_ jcap07(2009)029 alexander s , biswas t , notari a , vaid d 2009 _ j. cosmol .
_ jcap09(2009)025 zibin j p , moss a , scott d 2008 _ phys . rev . lett . _
* 101 * 251303 clarkson c , regis m 2010 the cosmic microwave background in an inhomogeneous universe - why void models of dark energy are only weakly constrained by the cmb _ preprint _
yoo c - m , nakao k - i , sasaki m 2010 _ j. cosmol .
_ jcap07(2010)012 moss a , zibin j p , scott d 2010 precision cosmology defeats void models for acceleration _ preprint _ arxiv:1007.3725 biswas t , notari a , valkenburg w 2010 _ j. cosmol .
_ jcap11(2010)030 wang y , mukherjee p 2007 _ phys .
d * 76 * 103533 elgary , multamki t 2007 _ astron .
astrophys . _ * 471 * 65 sachs r k 1961 _ proc .
london _ a * 264 * 309 kristian j , sachs r k 1966 _ astrophys .
j. _ * 143 * 379 kantowski r , vaughan t , branch d 1995 _ astrophys .
j. _ * 447 * 35 marra v , kolb e w , matarrese s , riotto a 2007 _ phys .
_ d * 76 * 123004 marra v , kolb e w , matarrese s 2008 _ phys .
_ d * 77 * 023003 kainulainen k , marra v 2009 _ phys . rev .
_ d * 80 * 127301 brouzakis n , tetradis n , tzavara e 2007 _ j. cosmol .
_ jcap02(2007)013 brouzakis n , tetradis n , tzavara e 2008 _ j. cosmol .
_ jcap04(2008)008 vanderveld r a , flanagan , wasserman i 2008 _ phys .
_ d * 78 * 083511 valkenburg w 2009 _ j. cosmol .
_ jcap06(2009)010 szybka s j 2010 arxiv:1012.5239 clifton t , zuntz j 2009 _ mon . not
. soc . _ * 400 * 2185 clifton t , ferreira p g 2009 _ phys .
_ d * 80 * 103503 bolejko k 2011 _ astron . astroph . _ * 525 * a49 bolejko k 2010 _ mon . not .
r. astron .
_ in press , _ preprint _
lewis a , challinor a 2006 _ phys .
* 429 * 1 hanson d , challinor a , lewis a 2010 _ gen .
grav . _ * 42 * 2197 ellis g f r 1971 in _ proceedings of the international school of physics ` enrico fermi ' , course 47 : general relativity and cosmology _ , ed .
r. k. sachs . academic press ,
new york and london , pp . 104 182 ; reprinted , with historical comments , in 2009 _ gen .
_ * 41 * 581 dyer c c , roeder r c , 1972 , _ astrophys .
* l115 dyer c c , roeder r c 1973 _ astrophys . j_. * 180 * l31 mattsson t 2010 _ gen . rel . grav . _ * 42 * 567 pyne t , birkinshaw m 2004 _ mon . not .
r. astron .
soc . _ * 348 * 581 bonvin c , durrer r , gasparini m a 2006 _ phys .
_ d * 73 * 023523 hui l , greene p b 2006 _ phys .
_ d * 73 * 123526 enqvist k , mattsson m , rigopoulos g , 2009 _ j. cosmol .
_ jcap09(2009)022 kainulainen k , marra v 2009 _ phys .
_ d * 80 * 123020 kainulainen k , marra v 2011 arxiv:1011.0732 yu h - r , lan t , wan h - y , zhang t - j , wang b - q 2010 _ research astron . astrophys .
_ in press ; arxiv:1008.1935 lematre g 1933 _ ann .
bruxelles _ a * 53 * 51 ( english translation , with historical comments : 1997 _ gen . rel .
* 29 * 637 ) tolman r c 1934 _ proc .
usa _ * 20 * 169 ( reprinted , with historical comments : 1997 _ gen .
_ * 29 * 931 ) lewis a , challinor a , lasenby a 2000 _ astrophys . j. _ * 538 * 473 amanullah r , lidman c , rubin d , aldering g , astier p , et al 2010 _ astrophys .
j. _ * 716 * 712 weinberg s 1976 _ astrophys .
j. _ * 208 * l1
ellis g f r , bassett b a c c , dunsby p k s 1998 _ class .
quantum grav . _
* 15 * 2345 peebles p j e 2010 _ aip conference proceedings _ *
1241 * 175 kayo i , taruya a , suto y 2001 _ astrophys . j. _ * 561 * 22 lahav o , suto y 2004 _ living rev .
relat . _ * 7 * 8 [ online article ] : cited on 27/09/2010 , http://www.livingreviews.org/lrr-2004-8 peebles , p. j. e. ( 1980 ) . _ the large - scale structure of the universe_. princeton : princton university press krasiski a 1997 _ inhomogeneous cosmological models _ ( cambridge : cambridge university press ) plebaski j , krasiski a 2006 _ an introduction to general relativity and cosmology _ ( cambridge : cambridge university press ) bolejko k , krasiski a , hellaby c , clrier m n 2009 _ structures in the universe by exact methods : formation , evolution , interactions _ ( cambridge : cambridge university press )
in order to solve ( [ dlen ] ) one needs to know @xmath57 along the line of sight . using the poisson equation the gravitational potential
is related to density fluctuations .
therefore , a method of generating density fluctuations is needed . in the non - linear regime ,
the density fluctuations are not gaussian ( there is no symmetry as @xmath99 ) .
however , in the non - linear regime the distribution of density fluctuations can be approximated with the one - point log - normal pdf @xcite @xmath100 \frac{1}{1+\delta},\ ] ] where @xmath101 , \quad { \rm ~and~ } \quad \sigma_r^2 = \frac{1}{2\pi^2 } \int\limits_0^\infty { \rm d } k { \cal p}(k ) w^2(kr ) k^2,\ ] ] and @xmath102 is the matter power spectrum .
therefore , the algorithm for @xmath103 along the line of sight is as follows : 1 .
the radius of a structure @xmath104 is randomly generated from a uniform distribution , from 0 to 10 mpc .
2 . then @xmath105 is calculated using ( [ nlvar ] ) .
an initial value of a density fluctuation @xmath106 is generated from the log - normal distribution ( [ nlpdf ] ) .
the evolution of @xmath103 ( at a fixed point ) is calculated using the linear approximations @xcite @xmath107 5 . using the poisson equation ,
@xmath57 is calculated and inserted into ( [ ddba ] ) which is solved from @xmath108 to @xmath109 .
6 . steps ( i)(v ) are repeated so ( [ ddba ] ) is solved from @xmath110 to @xmath111 .
a sample of density fluctuations along the line of sight is presented in figure [ fig1 ] .
instead of using directly the log - normal distribution to generate density fluctuations along the line of sight an alternative approach is used . in @xcite
it was shown that compensated structure can also have a log - normal pdf .
thus , len 2 generates density fluctuations along the line of sight using a method discussed in @xcite .
the algorithm is as follows : 1 .
first the radius of a void @xmath112 is generated from the gaussian distribution with the mean of 12 mpc and the standard deviation of 2 mpc .
2 . density within the void @xmath113 is generated from the gaussian distribution with the mean @xmath114 and @xmath115 .
if a generated in this way @xmath113 is lower than @xmath116 then the generation is repeated .
if after 6 times it is still less than @xmath116 then @xmath113 is generated for a uniform distribution between @xmath117 and @xmath116 .
if @xmath118 then @xmath113 is generated one more time . if after 6 times @xmath118 then its value is chosen for a uniform distribution from @xmath119 to @xmath35 .
density contrast is then @xmath120 .
3 . density of the surrounding shell @xmath121 is generated from the gaussian distribution with the mean of @xmath122 and @xmath123 .
if @xmath124 then its value is generated again . if after 6 times @xmath124 then its value is generated for a uniform distribution between @xmath122 and @xmath125 .
density contrast is then @xmath126 .
4 . the condition , that the structure is compensated implies that the radius of the whole structure is + @xmath127 5 .
the angle at which the light ray enters the structure is generated from a uniform distribution between @xmath117 and @xmath128 .
the evolution of @xmath103 ( at a fixed point ) was calculated using ( [ deltaev ] ) .
integral ( [ ddba ] ) is solved from @xmath108 to @xmath129 ( where @xmath108 is the comoving coordinate of the entry point and @xmath129 the point where the light ray exits the structure ) .
8 . steps ( i)(vii ) are repeated so ( [ ddba ] ) is solved from @xmath110 to @xmath111 .
a sample of density fluctuations along the line of sight is presented in figure [ fig1 ] .
the reason for defining the profile as above is to have a profile that allows for some kind of randomness ( the size and the density of voids and shells is random ) , but still because we use spheres ( which suppose to correspond to real wall that surround real voids ) the fluctuations are not purely random the pure randomness is gone once we decided to use spheres . still because of the above construction the 3d density pdf of this model is almost log - normal .
( see @xcite for a discussion ) . if instead of the gaussian , the uniform pdf is used ( steps ( i)(iii ) )
then the pdf of @xmath103 is not log - normal pdf .
first the radius of a void @xmath112 is generated from a uniform distribution between 0 and 30 mpc .
2 . density contrast of the void @xmath130 is generated from a uniform distribution between @xmath131 and @xmath132 .
3 . integral ( [ ddba ] )
is solved from @xmath108 to @xmath129 ( where @xmath108 is the comoving coordinate of the entry point and @xmath129 the point where the light ray exits the void ) .
then the length of a filament @xmath133 is generated from a uniform distribution between 0 and 5 mpc .
the density contrast of the filament @xmath134 is generated from a uniform distribution between @xmath135 and @xmath136 .
integral ( [ ddba ] ) is solved from @xmath108 to @xmath137 .
7 . steps ( i)(vi ) are repeated so ( [ ddba ] ) is solved from @xmath110 to @xmath111 .
a sample of density fluctuations along the line of sight is presented in figure [ fig1 ] .
the swiss cheese model considered here is based on the lematre - tolman ( lt ) model @xcite , which is the simples generalization of the friedmann models .
the metric of the lt model is where @xmath139 is an arbitrary function . in the friedmann limit @xmath140 ( in a general lt model the ` curvature index ' is position dependent ) .
@xmath141 is the areal distance and in the friedmann limit @xmath142 ( in a general lt model the scale factor is position dependent ) .
the evolution ( i.e. the generalized friedmann equation ) is @xmath144 is an arbitrary function defined by the mass density : @xmath145 from ( [ vel ] ) the age of the universe , @xmath146 , is @xmath147 where @xmath148 is an arbitrary function and in the friedmann model @xmath149 const ( in a general lt model the age of the universe is position dependent ) .
there are three arbitrary functions in the lt model @xmath150 , and @xmath139 , however only 2 are physically independent . for a review of different applications of the lt models
see @xcite . here
we will assume that the model is defined by the assumption that the age of the universe is everywhere the same , @xmath151 and the function @xmath144 is @xmath152 & { \rm ~for~ } x_a \leqslant \ell \leqslant 3x_a \\
-m_1 ( \ell - 4x_a)^3 & { \rm ~for~ } 3x_a \leqslant \ell \leqslant 4x_a , \\ 0 & { \rm ~for~ } \ell \geqslant 4x_a , \end{array } \right.\ ] ] where @xmath153/kpc , @xmath154 , @xmath155 , @xmath65 , @xmath156 km s@xmath64 mpc@xmath64 , @xmath157 , @xmath158 kpc and @xmath159 for regions a and b respectively , @xmath160 and @xmath161 , for regions a and b respectively .
region a and b are matches alternately at @xmath162 .
the above profile was chosen for the following reasons : it behaves like a flrw model for @xmath163 but with lower density than outside , then for @xmath164 we have a transition region , and a cubic behavior for @xmath165 allows for a smooth matching to the background values .
density fluctuations along the line of sight are presented in figure [ fig1 ] .
it should now be clear why we have selected the above set of functions .
these functions allow us to obtain the present - day density profile as presented in figure [ fig1 ] .
this model has thus 2 advantages : 1 ) it allows to calculate the evolution and the distance exactly , without additional approximations , 2 ) the mean of the density fluctuations along the line of sight is zero , i.e. as in len 1 model and unlike in len 2 and len 3 models .
when constructing a swiss cheese model , the junction conditions need to be satisfied . here
we match the lt inhomogeneity ( holes ) to the friedmann background ( cheese ) .
the lt patches are placed so that their boundaries touch wherever a light ray exits one inhomogeneous patch .
thus , the ray immediately enters another lt patch and does not spend any time in the friedmann background . to match the lt patch to the friedmann background across a comoving spherical surface , @xmath166 constant , the conditions are : 1 ) the mass inside the junction surface in the lt patch is equal to the mass that would be inside that surface in the homogeneous background ; 2 ) the spatial curvature at the junction surface is the same in both the lt and friedmann models , which implies that @xmath167 and @xmath168 ; 3 ) the bang time and @xmath3 must be continuous across the junction
. as seen , by the construction these conditions are satisfied . only radial geodesic ( i.e. the geodesics that propagate through the origin ) will be considered .
in such a case the shear of light bundle is zero @xcite .
thus , the junction of null geodesics only requires the matching of one of the components of @xmath10 as the other one is given from @xmath170 . | the standard analysis of the cmb data assumes that the distance to the last scattering surface can be calculated using the distance - redshift relation as in the friedmann model .
however , in the inhomogeneous universe , even if @xmath0 , the distance relation is not the same as in the unperturbed universe .
this can be of serious consequences as a change of distance affects the mapping of cmb temperature fluctuations into the angular power spectrum @xmath1 .
in addition , if the change of distance is relatively uniform no new temperature fluctuations are generated .
it is therefore a different effect than the lensing or isw effects which introduce additional cmb anisotropies .
this paper shows that the accuracy of the cmb analysis can be impaired by the accuracy of calculation of the distance within the cosmological models .
since this effect has not been fully explored before , to test how the inhomogeneities affect the distance - redshift relation , several methods are examined : the dyer roeder relation , lensing approximation , and non - linear swiss - cheese model . in all cases ,
the distance to the last scattering surface is different than when homogeneity is assumed
. the difference can be as low as 1% and as high as 80% . excluding extreme cases , the distance changes by about 20 30% .
since the distance to the last scattering surface is set by the position of the cmb peaks , in order to have a good fit , the distance needs to be adjusted . after correcting the distance , the cosmological parameters change .
therefore , a not properly estimated distance to the last scattering surface can be a major source of systematics .
this paper shows that if inhomogeneities are taken into account when calculating the distance then models with positive spatial curvature and with @xmath2 are preferred . the @xmath3cdm model ( i.e. a flat friedmann solution with the cosmological constant ) , in most cases , is at odds with the current data . |
in classical dynamical systems , one may study how long it takes a system to return to the phase space region it occupied at time zero .
in general , this time will increase when the phase space area under consideration is decreased .
m. kac has shown that , for an ergodic dynamical system , the average return time , assuming that the system starts in a set @xmath0 , is given by @xmath1 where the average is taken over the ergodic measure @xmath2 .
related to this are results about asymptotically rare events . consider a sequence @xmath3 of subsets of phase space such that @xmath4 .
if the time for a point @xmath5 in phase space to visit @xmath3 is denoted by @xmath6 , then the question arises how the product @xmath7 behaves .
this turns out , for a large class of models , to have an exponential distribution @xcite .
one may wonder whether a similar situation occurs in quantum mechanics . in finite dimensional systems ,
looking with increasing precision ( corresponding to the @xmath8 limit of above ) means increasing the dimension of the system , which amounts to taking its classical limit .
here , the inverse dimension plays the rle of planck s constant .
return times were studied for various quantum dynamical systems .
a natural way of measuring returns is provided by the notion of transition probability : for any two vectors @xmath9 and @xmath10 in a hilbert space it is defined as @xmath11 .
if a system starts from an initial state @xmath12 , one can consider @xmath13 where @xmath14 is the state of the system after @xmath15 time steps . for finite dimensional systems
this function is almost periodic .
interesting behaviour is displayed only when looking at the correct time scale .
it is known that , for classically chaotic systems , the quantum system follows its classical counterpart in the sense of coherent states up to a certain breaking time of the order of @xmath16 , where @xmath17 is the dimension of its hilbert space of states .
we propose to analyse the function ( [ overlap ] ) through the spectrum of the gram matrix associated with a sequence of normalised vectors @xmath18 : @xmath19 @xmath20 is a positive semi - definite matrix and its eigenvalues are independent of the order of the vectors and of any extra phase added to any of them .
information about the amount and the frequency of vectors that return to a small neighbourhood in hilbert space lies encoded in the spectrum of this matrix .
this fact is best appreciated by examining a simple example .
consider the sequence @xmath21 , where the @xmath22 s are normalised and mutually orthogonal .
this amounts to specifying a sequence of letters belonging to a given alphabet ( the classical case ) .
the gram matrix consists of only zeroes and ones ( because of the orthonormality ) and its spectrum is @xmath23 .
an eigenvalue different from zero indicates the number of times a certain vector is repeated and its multiplicity tells how many vectors are repeated that often . in our example , only two vectors are repeated , the vector @xmath24 twice and @xmath25 once , summing up to three , which is exactly the multiplicity of zero .
this example provides the link with the return times mentioned in the introduction : one considers a partition of phase space and keeps track of the different sets a particle visits during its evolution .
the associated gram matrix has a one for its @xmath26th entry if the visited sets at times @xmath27 and @xmath28 are the same and a zero otherwise . hence calculating the spectrum of the gram matrix
immediately gives all the return times .
in general , for a true quantum system , the vectors in a sequence will not be orthogonal and the overlaps will be complex numbers with absolute value smaller than or equal to one .
however , a similar interpretation of the spectrum can be obtained remembering that overlaps with absolute value close to 1 indicate vectors close to each other , while overlaps close to zero indicate almost orthogonal vectors .
generically , the non - zero eigenvalues of the gram matrix are non - degenerate and the notion of multiplicity of above should be replaced by the spectral density , i.e. a degenerate eigenvalue is now replaced by different eigenvalues close to each other .
the multiplicity of zero is equal to the number of linear dependencies in the given sequence .
two rather extreme cases will clarify this .
a large concentration of eigenvalues around 1 indicates that a large amount of vectors have small mutual overlaps .
this can be compared with the classical situation of above , with a spectrum consisting mainly of ones and only a few zeroes , indicating very few repetitions .
a presence of many eigenvalues close to zero suggests that many vectors lie close to one another , leading to many large overlaps .
this compares to a classical situation where some letters are repeated many times , giving rise to many zeroes in the spectrum and only a few large integers .
a point that has not been stressed yet is the number of time steps to consider .
the spectrum of the gram matrix will surely contain an eigenvalue zero when the number @xmath29 of vectors exceeds the dimension @xmath17 of the hilbert space . including too many time steps
causes a large degeneracy of zero .
we shall demonstrate , using numerical simulations , that also @xmath29 has to be taken large such that the ratio @xmath30 is a fixed number @xmath31 : time measured on the heisenberg scale .
the same scaling is relevant for a random model that can be treated analytically .
a natural choice for the initial state is a coherent state centred around a point in classical phase space .
there are several reasons for this .
taking the large @xmath17 limit immediately implies the comparison of quantum systems with different dimensions .
this also entails that initial states for different dimensions have to be comparable .
coherent states exist for every @xmath17 , occupy a volume @xmath32 of phase space and have a clear limit , namely a single point in phase space .
also , a coherent state is typically not a eigenvector of the evolution which would lead to a completely trivial spectrum of the gram matrix .
the precise object we shall investigate is the empirical eigenvalue distribution of the gram matrix .
if a matrix has eigenvalues @xmath33 , then its empirical eigenvalue distribution is @xmath34 in this way , the number of eigenvalues in a set @xmath35 is given by @xmath36 .
we shall illustrate this approach by example .
we consider two simple quantum dynamical systems frequently encountered in the quantum chaos literature .
they have clearly defined classical limits .
the first example is the quantum baker map and the second the kicked top .
the quantum baker map is a dynamics on @xmath37 with @xmath17 even @xcite .
defining the discrete fourier transform on @xmath38 @xmath39 the unitary evolution operator of the baker map is @xmath40 its classical limit can be shown to be @xmath41/2\\ \end{pmatrix}\ ] ] where @xmath42 is a point in the phase space @xmath43\times[0,1]$ ] .
the position and momentum @xmath44 and @xmath45 are the classical limits of @xmath46 the baker map is an example of a chaotic system : it has a positive lyapunov exponent equal to @xmath47 .
first , we numerically construct a coherent state in an @xmath17-dimensional hilbert space centred around a chosen classical point in phase space .
one way of doing this is by calculating the ground state of an appropriate harper operator and subsequently shifting it over to the desired classical point .
then , we apply the operator @xmath48 a number of times to calculate the overlaps needed in the gram matrix .
note that when the first row of the gram matrix is calculated , all other matrix elements are known because of the unitarity of the evolution .
unitarity has another interesting consequence : the precision of the calculation does not exponentially decrease with the number of time steps .
one way to see this , is applying the unitary evolution to an initial state many times , followed by applying an equal number of times the adjoint of the evolution .
the resulting state is remarkably close to the original state , despite the presence of numerical round - off errors in the components of the vectors and the unitary matrix .
this is in sharp contrast with numerical simulations of chaotic classical systems where positive lyapunov exponents cause the numerical errors to quickly exceed the desired precision .
the eigenvalues of the gram matrix are calculated and the empirical eigenvalue distribution is represented by the bar diagrams of figure [ bakerfig ] .
the solid black line will be commented on later in this letter .
lccc & @xmath49 & @xmath50 & @xmath51 + @xmath52 & & & + @xmath53 & & & + @xmath54 & & & + the first thing that figure [ bakerfig ] reveals is the time scale @xmath17 on which a non - trivial eigenvalue density is present . for a fixed time on the heisenberg time scale ,
i.e. a fixed ratio @xmath55 , the eigenvalue densities are very similar . for comparison ,
choosing @xmath29 of the order of @xmath16 produces a gram matrix close to the identity matrix and therefore a spectrum which is greatly concentrated around one . for times much larger than @xmath17 the linear dependencies dominate the picture and the empirical eigenvalue distribution tends to a dirac distribution at zero .
secondly , we note that we did not specify which coherent state we began with .
our simulations show that the initial state does not affect the form of the eigenvalue distribution
. diagrams , very similar to those of figure [ bakerfig ] , were obtained for different initial states .
the same procedure was carried out for a second example : the quantum kicked top .
this is a quantum spin-@xmath28 system with basic observables @xmath56 and @xmath57 obeying the usual commutation relations @xmath58=ij_z\ , + \ , cycl.\ perm.$ ] .
the unitary evolution is given by @xmath59 the classical limit of this system is a dynamics on the two dimensional sphere given by @xmath60 with @xmath61 a unit vector in @xmath62 which is the classical limit of @xmath63 .
this map is a composition of a rotation around the y - axis over an angle @xmath45 and a torsion around the @xmath64-axis over an angle proportional to the @xmath64-component .
numerical studies suggest the following behaviour . for small values of the parameters @xmath65 and @xmath45 , this system displays regular behaviour in large parts of the sphere , separated by chaotic regions .
the regular islands shrink as the parameters grow , leading eventually to almost global chaos ( see @xcite ) .
this is an interesting feature of this model : varying the control parameters @xmath65 and @xmath45 changes the behaviour of the classical model considerably .
it can be concluded from figure [ kickfig ] that this difference in behaviour is also present in the quantum kicked top . for small values of the control parameters @xmath65 and @xmath45 there are a lot of approximate returns during the first @xmath66 time steps .
this is clear from the large number of eigenvalues close to zero .
we expect that in the limit @xmath67 this distribution tends to a degenerate one in zero as the system visits only a small fraction of the available states . enlarging the control parameters places more weight on higher eigenvalues and a totally different eigenvalue distribution appears .
these distributions appear to be very similar to the ones for the baker map for the same value of @xmath31 .
lcc & @xmath49 & @xmath50 + @xmath68 and @xmath69 & & + @xmath70 and @xmath71 & & + one unexplained feature in the diagrams is the solid line .
this is actually the graph of an analytical function obtained in the following way . instead of trying to calculate the limiting empirical eigenvalue distribution for a true quantum mechanical model , like the baker map or the kicked top
, one can try to simplify the calculations by considering a random vector model . for an @xmath17-dimensional hilbert space , a sequence of @xmath29 normalised vectors
is chosen randomly and independently according to the uniform distribution on the unit sphere . in this way
, one can try to mimic the evolution of a system which wanders wildly in hilbert space .
the limiting eigenvalue distribution ( @xmath72 and @xmath73 ) of the gram matrix associated with this sequence of random vectors can be calculated explicitly and is given by the marchenko - pastur distribution @xcite @xmath74 this density is the solid line depicted in figures [ bakerfig ] and [ kickfig ] .
a possible explanation for the similarity between the eigenvalue distributions generated by the baker map and the kicked top and the marchenko - pastur distribution is that , when the number of time steps is linear in the dimension of the hilbert space , subsequent vectors in a chaotic quantum evolution are almost random and visit all of the hilbert space without preference for one direction or region .
the eigenvalue distribution of the gram matrix appears to be a meaningful object in the so - called deep quantum regime where the system is allowed to visit a finite fraction of the total number of available states .
moreover , for systems with a chaotic classical limit , numerical simulations indicate that the limiting distribution is universal and coincides with that of random vectors , even if the latter model is not a unitary dynamics .
if the classical limit is regular , we find a degenerate distribution .
linking the limiting spectrum of the gram matrix with the distribution of energy or quasi - energy level separations @xcite definitely deserves further research .
another interesting question , that is also relevant in quantum information theory , is what happens when the dynamical system is no longer isolated but weakly coupled to an environment .
in such a situation , density matrices instead of vector states must be used to describe the decoherence induced by the environment .
wat n.l .
balazs and a. voros , the quantized baker s transformation , _ ann . phys . _
* 190 * , 131 ( 1989 ) m.v .
berry and m. tabor , level clustering in the regular spectrum , _ proc .
a _ * 356 * , 375394 ( 1977 ) m. de cock , m. fannes , and p. spincemaille , on quantum dynamics and statistics of vectors , _ j. phys .
a : math .
gen . _ * 32 * , 65476571 ( 1999 ) f. haake , m ku , and r. scharf , classical and quantum chaos for a kicked top , _ zeitschrift fr physik _
b*65 * , 381395 ( 1987 ) m. hirata , b. sausol and s. vaienti , statistics of return times : a general framework and new applications , _ commun . math .
* 206 * , 3355 ( 1999 ) | for chaotic classical systems , the distribution of return times to a small region of phase space is universal .
we propose a simple tool to investigate multiple returns in quantum systems .
numerical evidence for the baker map and kicked top points , also in the quantum case , at a universal distribution .
+ _ pacs _ : 05.45.mt,05.45.pq + _ keywords _ : return times , baker map , kicked top |
pinning and sliding of lattice systems , which can form periodic structures in the absence of perturbations , are subjects of considerable interest . in the presence of pinning disorder and driving forces , they can exhibit a wide variety of interesting equilibrium and non - equilibrium behavior with partially ordered and glassy structures .
important examples in two dimensions include vortex lattices in superconducting films @xcite , atomic layers adsorbed between sliding surfaces @xcite or on oscillating substrates @xcite , and colloidal crystals on a rough substrate @xcite .
although in three dimensions a topological ordered phase , the bragg glass @xcite , with quasi - long range positional order is possible in the weak disorder regime , the two - dimensional limit is qualitatively different due to the proliferation of thermally and disorder - induced dislocations even in the weak - disorder regime .
analytical and numerical studies for the equilibrium behavior @xcite have shown that , in two dimensions , positional and orientational order are both destroyed by weak pinning disorder and topological defects at any nonzero temperature leading to a liquid - like phase in the thermodynamic limit . in the absence of thermal fluctuations ( zero temperature ) an amorphous glass is expected @xcite . for the system under a driving force moving at high velocities ,
it has been shown analytically that some components of the disorder remain static in the co - moving reference frame , leading to a moving glass phase @xcite . in three - dimensional systems ,
such state can show topological order as a moving bragg glass @xcite . in two dimensions
, however , this phase corresponds to a moving smectic glass , which retains quasi long - range order in the direction transverse to the driving force but only exponential correlations in the parallel direction @xcite . in modeling such lattice systems , both for static and dynamic properties ,
it is essential to include the periodicity of the lattice and allow for topological defects ( dislocations and disclinations ) .
these topological defects are not contained in pure elastic models that have completely different properties , specially in two dimensions .
a phase - field - crystal ( pfc ) model was introduced recently @xcite that allows for both elastic deformations and topological defects within an effective continuous description of the lattice system while still retaining information on short length scales . by extending the pfc model to take into account the effect of an external periodic pinning potential @xcite ,
a two - dimensional version of the model has been used to describe commensurate - incommensurate transitions in the presence of thermal fluctuations @xcite and the driven response @xcite including inertial effects @xcite . however , in order to study the statics and dynamics of the disordered system , quenched pinning disorder needs to be included in the pfc modeling . in this work ,
we study the structural correlations and nonlinear response to a driving force of a two - dimensional pfc model with random pinning .
the model provides an effective continuous description of lattice systems in presence of disordered external pinning centers , allowing for both elastic and plastic deformations .
we show that in the presence of disorder , the phase - field crystal assume an amorphous glassy ground state , with only short - ranged positional and orientational correlations even in the limit of weak disorder . under increasing driving force
, the pinned amorphous - glass phase evolves into a moving plastic - flow phase and then finally a smectic phase .
the transverse response of the moving smectic phase shows a vanishing transverse critical force for increasing system sizes .
the effective hamiltonian of the pfc model with an external pinning potential @xcite can be written in a dimensionless form as @xmath0 \psi + \frac{\psi^4}{4 } + v(\vec x ) \psi ( \vec x ) \ } , \label{pfcp}\ ] ] where @xmath1 is a continuous field at position @xmath2 in two dimensions and @xmath3 is the pinning potential .
the field @xmath1 is conserved and its average value , @xmath4 , together with @xmath5 are the relevant parameters in the model . in the absence of the pinning potential , @xmath6 , the energy functional of eq .
( [ pfcp ] ) can be minimized by a configuration of the field @xmath1 forming a hexagonal pattern of peaks with wave vector of magnitude @xmath7 , when the values of the parameters @xmath8 and @xmath4 are chosen appropriately .
this structure of peaks can be regarded as the ground state of a lattice system with perfect crystalline order , where the phase field @xmath1 represents the deviation of the particle number density @xmath9 from a reference value @xmath10 , @xmath11 .
the energy functional of eq .
( [ pfcp ] ) can then be used to describe both elastic as well plastic properties of the lattice system @xcite , including the effects of thermal fluctuations @xcite .
an external driving force @xmath12 in the lattice system , can be represented by an additional contribution to the effective hamiltonian as @xmath13 which will lead to nonequilibrium behavior .
the effects of the driving force have been recently investigated in some detail @xcite , in absence of disorder .
inertial effects of the lattice system can also be described within the pfc model , by including an additional kinetic energy contribution to the effective hamiltonian as @xmath14 where @xmath15 is the momentum density field .
the dynamical equations describing the time evolution of the lattice system in presence of thermal fluctuations and the external force can be written as @xcite @xmath16 where @xmath17 is the spatially uniform external force and @xmath18 is a thermal noise satisfying the fluctuation - dissipation relation corresponding to a temperature @xmath19 and damping coefficient @xmath20 . in the present work ,
we only consider the limit of very large @xmath20 , when inertial effects are negligible , leading to the overdamped dynamical equations @xmath21 the above coupled equations for @xmath22 and @xmath23 can also be combined into a single equation for @xmath22 , giving @xmath24 the generalization of the pfc model with an external periodic pinning potential studied previously @xcite to the case of a quenched random pinning potential considered here is straightforward .
such a model is relevant for studying diverse systems such as , adsorbate layers with quenched impurities or on substrates with disorder , and vortex lattice in superconductors in the presence of pinning centers . to this end
we model the quenched potential in the simplest way by defining at every spatial location @xmath25 , where @xmath26 is a @xmath27-correlated distribution @xmath28 and @xmath29 is an amplitude characterizing the strength of the disorder . since , there is no characteristic length scale in such pinning potential , it is particularly suitable for the investigation of finite - size effects using small system sizes as done in this work . in this
_ dense _ pinning model , the separation between pinning centers can be much smaller than the distance between the density peaks in the phase - field crystal .
this corresponds to a physical system where the length scale of the varying pinning potential is smaller than the average lattice spacing of the system . such a scenario could be realized _
e.g. _ in the case of colloidal particles confined near a rough glass plate @xcite . for the numerical calculations ,
the phase field @xmath1 is defined on a space grid @xmath30 with periodic boundary conditions .
the simulations in presence of thermal fluctuations and the driving force were performed using eq .
( [ coveq ] ) , using euler s method with the laplacians and gradients evaluated by finite differences . in the absence of external force and thermal fluctuations , the equation of motion ( [ soveq ] ) was used in most simulations , and solved using a semi - implicit algorithm .
the linear part is treated implicitly , while the non - linear term is treated explicitly @xcite .
the field at time @xmath31 is obtained according to @xmath32 where , @xmath33 is the fourier transform of @xmath34 , @xmath35 is the equivalent in inverse space of the laplacian @xmath36 and @xmath37 is the inverse space equivalent of the linear operator @xmath38 .
the term @xmath39 is the fourier transform of the nonlinear part @xmath40 .
to investigate the influence of disorder on the structural properties of the phase - field crystal , we study the behavior of the static correlation functions from calculations of the structure factor @xmath41 and orientational susceptibility @xmath42 , which measure translational order and orientational order @xcite , respectively .
@xmath41 can be obtained from the positions @xmath43 of the peaks in the phase - field pattern as @xmath44 where @xmath45 is the number of peaks , while @xmath42 can be obtained from the local orientational order parameter @xmath46 as @xmath47 the orientational order parameter @xmath46 is a measure of the six - fold orientational symmetry of the crystalline order and is defined as @xmath48 where the summation is taken over the @xmath49 nearest neighbors of the peak at position @xmath50 and @xmath51 is the angle of the bond @xmath52 with an arbitrary axis .
the disorder averaged @xmath53 and @xmath42 are obtained by averaging over different realizations of the disorder configurations @xmath3 . in the absence of disorder , solutions of the pfc model form a periodic hexagonal array of peaks in the ground state with wavector @xmath54 , which has both long range positional and six - fold orientational order .
positional order can be characterized by the scaling of the structure factor @xmath55 with system size @xmath56 , which behaves as @xmath57 for a perfectly ordered state .
disorder can destroy positional long range order and lead to quasi long - range or short - range positional order , which corresponds to correlations that decay with distance @xmath8 as a power law @xmath58 or exponentially @xmath59 , respectively .
for these distinct types of correlations the corresponding structure factor is expected to behave as @xmath60 and @xmath61 , for large system sizes .
we find it convenient to define an _ effective exponent _
@xmath62 to characterized the positional order in different regimes , from a power law fit of the normalized structure peak as @xmath63 with this definition , since @xmath64 , @xmath65 indicates long range order , @xmath66 quasi long - range order and @xmath67 short - range order .
an analogous effective exponent @xmath68 characterizing the orientational correlations , can be defined from the finite - size dependence of the orientational susceptibility @xmath42 as @xmath69
in this section , we present our numerical results for the static behavior and the driven response of the phase - field crystal at zero temperature ( without thermal noise ) , obtained in the absence and presence of the external force , respectively . the dynamical equations , eqs .
( [ coveq ] ) and ( [ soveq ] ) , were integrated numerically on a uniform square grid with @xmath70 .
the total system size @xmath71 accommodates approximately @xmath72 density maxima .
the pfc parameters are set to @xmath73 , where the ground state of the model is a perfectly ordered hexagonal phase in the absence of disorder , and @xmath74 . for the case without external force and thermal fluctuations we used a rectangular box of size
@xmath75 with @xmath76 and @xmath77 . in order to check the size effect we performed calculations for several values of @xmath78 from @xmath79 to @xmath80 .
we first search for the the lowest - energy state of the phase - field crystal in the absence of the external driving force . for this purpose
, we have used the simulated annealing method to avoid trapping in higher energy metastable states .
first , simulations of equilibrium states at some initial temperature corresponding to finite noise terms in the dynamical equation ( eq .
( [ coveq ] ) ) are performed using arbitrary initial configurations .
then the initial temperature is slowly decreased to zero leading to a final configuration with a minimum energy .
this annealing procedure is repeated for different initial configurations and the zero - temperature configuration with the lowest energy is identified as the ground state .
[ sfanneal ] shows the behavior of the structure factor peak and orientational susceptibility as function of the system size obtained by simulated annealing , in the weak disorder regime , @xmath81 . the effective exponent for both positional and orientational correlations obtained from a power law fit of the size dependence in fig .
[ sfanneal ] is consistent with two , indicating that correlations are short ranged .
larger disorder strength gives the same result .
this behavior suggests that the ground state of the phase - field crystal in presence of pinning disorder is an amorphous glass .
this is expected on theoretical grounds .
in fact , analytical studies by renormalization - group methods @xcite and computer simulations @xcite indicate that , in two dimensions , quasi long - range positional and orientational order are destroyed by weak disorder or thermal fluctuations , leading to a liquid - like phase in the thermodynamic limit . in the absence of thermal fluctuations ( zero temperature ) an amorphous - glass state is expected @xcite . the length scale for the crossover to such state increases with decreasing disorder strength .
long - range order in presence of weak disorder has been found at low temperatures in some other models @xcite , which however describe structural ( internal ) disorder @xcite rather than external pinning disorder as considered here . on the other hand , in some molecular - dynamics simulations of particle models of colloidal crystals on a disordered substrate @xcite ,
quasi long - range order was observed at low temperatures in a weak disorder regime , although no detailed finite - size analysis was carried out .
quasi long - range order has also been observed in experiments on colloidal crystals on a rough substrate @xcite .
however , the apparent positional and orientational order in such cases can also be explained as a finite - size effect due to a large but finite crossover length scale , which increases with decreasing disorder strength @xcite .
another possible explanation is the effect of slow dynamics , which requires a much longer time scale to observe the true equilibrium state , leading to partially - ordered metastable states . in sec .
ivc , we examine similar states for the pfc model , which were obtained using the dynamical equations in absence of thermal annealing . to obtain the driven response of solutions of the phase - field crystal model ,
we determine the steady state velocity of the peaks under a constant external force as described by eq .
( [ coveq ] ) .
the velocity is measured by tracking the position of the peaks @xmath82 in the phase field @xmath1 during the simulation @xcite . from the velocities of the peak positions @xmath83 ,
the steady state drift velocity is obtained as @xmath84 where @xmath85 is the number of peaks and @xmath86 denotes time average .
the disorder averaged drift velocity is then obtained by averaging over different realizations of the disorder configurations .
the calculations were done without thermal noise ( zero temperature ) .
the initial state at @xmath87 is obtained from simulated annealing as described in sec .
the velocity response of the phase - field crystal to the applied driving force is shown fig .
[ vxf ] for strong disorder @xmath88 .
the velocity response is nonlinear , with different behavior at low and large driving forces , due to the effects of pinning .
this property is observed in different lattice systems with quenched disorder @xcite . at sufficiently low drive ,
the drift velocity is negligibly small and the phase - field crystal remains pinned in the amorphous glassy phase . as the force increases beyond a critical value , @xmath89 , corresponding to a depinning transition , the phase - field crystal moves with increasing steady - state velocity but with different patterns at low and high velocities as can be seen from the corresponding configurations in fig .
[ patf02f08 ] . at low velocities above the critical force @xmath90 ,
the pattern of the phase field has a liquid - like structure as shown in fig .
[ patf02f08]a for @xmath91 .
the trajectories of the local maxima in the phase field @xmath1 correspond to a plastic flow due to pinned and unpinned regions . for this moving state ,
the corresponding structure factor ( fig .
[ sff02f08 ] ) , averaged over time and disorder , shows small and broad peaks . for larger values of the driving force above a characteristic value @xmath92 ,
there is a dynamical reordering transition to a moving partially ordered phase as shown in fig .
[ patf02f08]c for @xmath93 .
the trajectories of the local maxima form well defined static channels parallel to the driving force .
the structure factor now become larger and sharper as shown in fig .
[ sff02f08 ] .
rough estimates of critical values @xmath90 and @xmath92 obtained for different disorder strengths were used to construct the qualitative dynamical phase diagram as a function of disorder strength and applied force shown in fig .
[ phadiag ] .
the nature of the moving phase at high velocities is particularly interesting . since the pinning potential acts as a time oscillating perturbation in the co - moving reference frame
, one would expects that the disorder effects on the initial hexagonal structure should decrease with increasing velocity .
however , it has been shown that some components of the disorder remain static @xcite , leading to a moving glass phase . in two dimensions ,
such a phase corresponds to a moving transverse glass characterized by a smectic - like structure , which retains quasi long - range order in the direction perpendicular to the applied force but only exponential correlations in the direction parallel to the force @xcite .
this moving smectic state has been observed in many different driven systems with disorder @xcite .
indeed , for the present driven pfc model , the pattern for the large driving force in fig .
[ patf02f08]d also shows a smectic - like structure . to verify this behavior more quantitatively ,
we have studied the finite - size dependence of the structure factor peaks ( fig .
[ sff02f08 ] ) in the transverse and longitudinal directions for the sliding state at @xmath93 .
the results shown in a log - log plot in fig .
[ sff08 ] are consistent with a smectic phase in the thermodynamic limit , where ordering only occurs in one spatial direction .
the height of the peaks in the transverse direction decreases slowly with system size as a power law @xmath94 with @xmath95 , consistent with quasi long - range order , while in the longitudinal direction it decreases much faster with @xmath96 as expected for short - range order .
this estimate of the critical exponent @xmath97 for the transverse peaks in the driven phase - field crystal is comparable to the result , @xmath98 , obtained for a particle model of driven vortex lattices in disordered superconductors @xcite .
another important property of the moving glass state at large velocities is the existence of barriers to transverse motion .
it was predicted @xcite that this should lead to a transverse critical force at zero temperature for an additional external force applied perpendicular to the initially driving force .
however , it has also been argued @xcite that the transverse response at zero temperature is nonuniversal , and qualitatively different behavior is possible for different physical systems .
the former scenario of transverse depinning has already been observed for driven vortex lattices @xcite . to find out which scenario is realized in the present version of the pfc model
, we have studied the velocity response in the transverse direction @xmath99 at a large longitudinal force @xmath100 for different realizations of the disorder configurations .
we find that the transverse depinning force is very sensitive to the disorder configurations and seems to vanish in some cases as can be seen in fig .
[ trans ] , where the transverse velocity response is shown for different realizations of the disorder configuration for a system size @xmath101 .
to investigate if the transverse critical force is still nonzero in the large system limit , we performed calculations for increasing system sizes and the results were averaged over different disorder configurations . fig .
[ avtrans ] shows that the critical value @xmath102 for the onset of transverse depinning decreases with increasing system size and therefore may actually vanish in the thermodynamic limit .
although such behavior is possible according to analytical arguments @xcite , it is unclear at the present which distinct feature of the pfc model is responsible for it .
since a transverse critical force has been found for the same model in the case of periodic pinning @xcite , we can only speculate that the presence of disorder allows some defects to be generated even by small transverse forces .
in the previous section iva we have determined the lowest energy state in absence of a driving force ( ground state ) using simulated annealing methods and found that the configuration corresponds to an amorphous glass state , with only short - ranged positional and orientational correlations . in this section
, we will discuss some metastable states .
these are states corresponding to local minima rather than the global minimum for the ground state .
nonetheless , they may still be relevant under suitable experimental conditions .
metastable states were obtained by starting with the perfect hexagonal ground state in the absence of disorder and then allowing the system to evolve according to the dynamical equation for a relatively short period ( @xmath103 ) , for different disorder strengths in the absence of thermal noise .
the disorder strength @xmath29 is incremented in steps of @xmath104 .
[ deta ] shows the behavior of the effective exponents @xmath62 and @xmath68 for positional and orientational correlations of the final static configuration as a function of the disorder strength @xmath29 .
it appears that there are three different thresholds values of @xmath29 : @xmath105 , @xmath106 and @xmath107 .
below @xmath108 , the _ weak - disorder _
regime , the system has long - range orientational order and quasi - long range translational order . between @xmath108 and @xmath109
the system exhibits both quasi - long range orientational and translation order . between @xmath109 and @xmath110 a hexatic
ordering occurs which corresponds to quasi - long range orientational order and short - range translational order .
this phase ordering is analogous to such a phase that occurs in two dimensional crystals induced by thermal flucuations @xcite . finally above @xmath110 , which is beyond the range of @xmath29 investigated
, we expect that only short - range orientational and translational order should remain .
as described in sec .
iva , the lowest - energy state obtained by simulated annealing is an amorphous glass rather than a partially - ordered state observed in such netastable states .
in fact , the inset in fig . [ deta ]
indicates that the phase - field crystal would remain essentially ordered for weak pinning if no simulated annealing was performed .
however , even in this weak disorder regime correlations are actually short ranged if the system is allowed to reach its true ground state , as can be seen from the inset in fig .
[ sfanneal ] .
although the ground state is an amorphous glass , the thresholds @xmath111 @xmath112 and @xmath110 in fig .
[ deta ] can represent some characteristic values for non - equilibrium behavior .
we have also investigated the metastable states for a different model of disorder . in this _ sparse _ pinning model , the pinning sites are separated by a minimum distance @xmath113 , _
i.e. _ they can not come closer to each other than the particles in the ideal hexagonal ground state of the pfc model with wave vector @xmath54 .
such model may be realized _
e.g. _ in the case of adsorbed atomic layers on a substrate with impurities @xcite .
a finite density @xmath114 of randomly distributed pinning centers is assigned with pinning strength ( amplitude ) @xmath115 .
thus , unlike the _ dense _ pinning model discussed above , there are now two relevant parameters , namely @xmath114 and @xmath115 .
metastable states were obtained again starting from the hexagonal state and incrementing @xmath115 in steps of @xmath116 , while the density of pinning sites @xmath114 was fixed to @xmath117 . for this density of the pinning centers the ratio between the pinning centers and
the number of maxima of the ideal triangular lattice is @xmath118 .
[ seta ] shows the behavior of the effective exponents @xmath62 and @xmath68 for positional and orientational correlations of the final static configuration as a function of the disorder strength @xmath115 obtained in this sparse disorder model . although the errorbars are relatively large , there is again three distinct thresholds values , @xmath119 , @xmath120 and presumably @xmath121 .
the results are qualitatively the same as that for the dense pinning model .
we have studied a two - dimensional pfc model with random pinning .
the model provides an effective continuous description of lattice systems in the presence of disordered external pinning centers , allowing for both elastic and plastic deformations .
the structural correlations and nonlinear driven response are determined from numerical simulations of the dynamical equations .
we find that in the absence of a driving force , the phase - field crystal assumes an amorphous glassy ground state , with short - ranged positional and orientational correlations even in the limit of weak disorder . under increasing driving force
, this pinned amorphous - glass phase evolves first into a moving plastic - flow phase then followed by a moving smectic phase .
these results are largely in agreement with previous analytical works @xcite .
an additional new feature is that the transverse response of the moving smectic phase shows a vanishing transverse critical force for increasing system sizes .
finally , we have identified interesting quasi - long range ordered metastable states by evolving initially ordered states in the presence of disorder .
the nature of these metastable states are insensitive to the details of disorder and they may be relevant under suitable experimental conditions .
e.g. was supported by fundao de amparo pesquisa do estado de so paulo - fapesp ( grant no . 07/08492 - 9 ) .
k.r.e . acknowledges support from the nsf under grant no .
this work was also supported in part by computer facilities from the centro nacional de processamento de alto desempenho ( cenapad - sp and cenapad - ufpe ) .
g. tegze , g. bansel , g. i. t ' othb , t. pusztai , z. fan , and l. gr ' ansy , j. comput . phys .
* 228 * , 1612 ( 2009 ) ; b. p. vollmayr - lee and andrew d. rutenberg , phys .
e * 68 * , 066703 ( 2003 ) ; j. zhu , l .- q .
chen , j. shen , and v. tikare , ibid . * 60 * , 3564 ( 1999 ) . | we study the structural correlations and the nonlinear response to a driving force of a two - dimensional phase - field - crystal model with random pinning .
the model provides an effective continuous description of lattice systems in the presence of disordered external pinning centers , allowing for both elastic and plastic deformations .
we find that the phase - field crystal with disorder assumes an amorphous glassy ground state , with only short - ranged positional and orientational correlations even in the limit of weak disorder . under increasing driving force
, the pinned amorphous - glass phase evolves into a moving plastic - flow phase and then finally a moving smectic phase .
the transverse response of the moving smectic phase shows a vanishing transverse critical force for increasing system sizes . |
in 1993 the discovery of an inorganic spin - peierls ( sp ) compound , cugeo@xmath2 , was first reported by hase _
@xcite this material has chains of cu@xmath4 ( @xmath5 ) along its @xmath6-axis which distort into dimers below the sp transition temperature , @xmath7k.@xcite the dimerized structure has a singlet ground state and a triplet excited state ; singlet - to - triplet excitations were demonstrated by nishi _
et al._@xcite using inelastic neutron scattering .
cugeo@xmath2 is interesting not only because it is the first known inorganic spin - peierls system , but also because evidence is mounting that it does not behave like a typical sp system .
it has been believed that good one - dimensionality was needed for a material to undergo a spin - peierls transition .
however nearest - neighbor exchange parameters , estimated from the dispersions of magnetic excitations@xcite imply that the intrachain coupling is only 10 times stronger than the interchain coupling , which is far smaller than is found for other one - dimensional systems , typically 10@xmath8 to 10@xmath9 [ , ] .
recent neutron - scattering measurements have revealed that the softening of a zone - boundary phonon as expected in ordinary sp systems does not exist.@xcite in addition , high - pressure neutron scattering studies@xcite have shown that as the pressure increases , @xmath10 and the sp energy gap increase while the lattice dimerization decreases , implying an additional mechanism , such as a spin - only effect , is at work .
the susceptibility above @xmath10@xcite and the critical exponent @xmath11 of the sp energy gap@xcite both deviate substantially from the theoretical calculations of bonner and fisher@xcite and cross and fisher@xcite respectively , which work well for organic sp systems .
et al._@xcite have recently proposed one possible theoretical model where the magnetic properties of cugeo@xmath12 can be well described by the one - dimensional heisenberg model with competition between nearest - neighbor and next - nearest neighbor antiferromagnetic interactions . shortly after the discovery@xcite of cugeo@xmath2 , a series of extensive studies were begun on samples with small amounts of dopant .
investigations have concentrated on systems where cu atoms were replaced with zn @xcite or ge was replaced with si.@xcite it is now well established that a new antiferromagnetic ( af ) ordered phase appears at low temperatures in both zn - doped@xcite and si - doped@xcite samples .
a preliminary phase diagram was determined from susceptibility measurements on powder samples@xcite which showed that @xmath10 decreases with increasing zn concentration while the nel temperature increases to @xmath13 4k .
this phase diagram was recently confirmed using single crystals@xcite and in figure [ phase_diagram ] we present the phase diagram as determined from neutron and susceptibility measurements on all single crystals produced for the present study .
the most intriguing feature reported so far in these doped cugeo@xmath2 systems is the coexistence of the sp and af states below the nel temperature .
this was first demonstrated by regnault _
et al._@xcite who used neutron scattering to observe the sp order parameter , i.e. the lattice dimerization , become smaller but still finite below @xmath14 . sasago _ et al._@xcite showed that the sp and af states also coexist for samples having a nominal range of 2 6% zn . the coexistence of a sp singlet state and af ordering was quite a surprise since it was thought that these two ordered states were mutually exclusive , and it adds another puzzle to the already unusual sp state in cugeo@xmath2 .
recently , fukuyama _
et al._@xcite have proposed a theoretical model for disorder - induced antiferromagnetic long range order within a spin - peierls system .
this model concludes that an infinitesimal amount of impurity can induce a long - range af state in a spin - gap system ( which includes spin - ladders and two dimensional systems , as well as sp systems ) .
the comparison of this theory with our experimental results will be presented . in the present paper comprehensive neutron scattering results
are presented studying zn - doped cugeo@xmath2 single crystals grown by the floating - zone ( fz ) method .
the amount of zn - dopant , @xmath3 , was carefully determined using electron probe micro - analysis ( epma ) for each crystal .
the sp dimerization and the af magnetic moment are surveyed as a function of @xmath3 .
magnetic excitations in both the sp and the sp - af coexistence states were studied in detail as a function of temperature , @xmath15 , and @xmath3 yielding some important conclusions about the interplay of the af and sp states .
a series of relatively large ( @xmath16 @xmath17 ) cu@xmath0zn@xmath1geo@xmath2 single crystals having _ nominal _ zn concentrations of 1 , 2 , 4 , 6 , and 8% were grown using the floating - zone ( fz ) method .
the crystals grow along the @xmath6-axis ( the chain direction ) and typically had mosaic spreads of less than 0.3 degrees . as shown in figure [ samples ] , the crystals were cut into several parts for characterization using electron probe micro - analysis ( epma ) and measurements of neutron scattering and susceptibility .
typical susceptibilities along the @xmath6 direction for several values of @xmath3 are shown in the lower part of figure [ samples ] .
the epma analysis revealed that the true zn concentrations were consistently lower than the nominal values : the average measured values of @xmath3 are 0.42 , 0.9 , 3.2 , 2.1 , and 4.7% corresponding to the nominal values of 1 , 2 , 4 , 6 , and 8% , respectively .
as the zn concentration increases , it becomes more difficult to grow a single crystal with a single domain . in order to obtain high quality single crystals with small mosaic spreads
, the speed at which the zone is moved must be kept slow . on the other hand , impurities ( zn atoms )
tend to be purged from the melted zone if the growth process is too slow .
it is thus difficult to fabricate crystals both of high quality and with high zn concentration .
we were able to grow crystals with a fairly wide range of zn concentrations enabling a systematic study of the effects of zn doping .
the homogeneity of the zn concentrations was checked by measuring various regions ( @xmath18m@xmath8 each ) of the samples by epma .
these homogeneities are quoted as uncertainties in table [ tab ] which also presents the crystal volumes and measured transition temperatures .
both sp and af transitions are quite sharp , consistent with small error bars in the actual zn concentrations .
neutron scattering measurements were carried out using triple - axis spectrometers on the h7 and h8 ( 14.7 mev incident energy neutrons ) , and h9a ( 5 mev ) beamlines of the high flux beam reactor at brookhaven national laboratory .
the crystals were mounted in helium filled aluminum cans which were subsequently attached to the cold finger of a cryostat .
the samples were aligned so as to place the @xmath19 or @xmath20 ) zones in the experimental scattering plane .
incident neutrons with energies of 14.7 or 5.0 mev were selected by a pyrolytic graphite ( pg ) @xmath21 monochromator ; pg @xmath21 was also used for an analyzer . to eliminate higher - order harmonics ,
pg filters were placed before the sample and after the analyzer .
the beam was horizontally collimated , typically 40@xmath22 - 40@xmath22-sample-40@xmath22 - 80@xmath22 in sequence from the reactor core to the detector .
tighter collimations were used for some of the higher resolution measurements presented . when comparing the intensities of dimerization and magnetic superlattice peaks between the various crystals , bragg peaks unaffected by extinction are required for use as normalization .
the weak @xmath23 and @xmath24 peaks fill this need and are used for our data analysis .
the transition temperatures @xmath10 and @xmath14 , where the sample enters the sp and af states respectively , were determined for each crystal using both magnetic susceptibility measurements and observing critical scattering peaks in the neutron studies .
figure [ order_parameters ] plots the intensity of the sp and af superlattice peaks as a function of temperature for 3.2 and 0.42% zn - doped samples . @xmath10 and @xmath25 for each crystal are summarized in table [ tab ] .
the full phase diagram including susceptibility data from numerous crystals is presented in figure [ phase_diagram ] .
the most significant result here is that af ordering is observed for even the smallest available amounts of zn , @xmath26 .
as shown in figure [ order_parameters](b ) , the af superlattice peak is nearly resolution limited demonstrating that a long - ranged nel ordering is induced by very small amounts of zn impurities .
a second interesting result is that @xmath10 and @xmath14 change as @xmath3 is increased to about 2% , but then do not change much for zn concentrations exceeding 2% .
hirota _ et al._@xcite determined that as the dimerization occurs below @xmath10 , the @xmath27 superlattice peaks , with @xmath28 and @xmath29 odd and @xmath30 any integer , appear .
the intensities of the sp superlattice peaks decrease continuously as a function of increasing zn content indicating a weakening sp dimerization .
the transition is less sharp in the zn - doped crystals compared to pure cugeo@xmath2 , and it can be estimated by fitting the data with a power law model which includes a gaussian distribution of transition temperatures . using the maximum intensity of these sp superlattice peaks ( at about 6 k ) , the dimerizing atomic displacement , @xmath31 , can be calculated as compared to the displacement in pure cugeo@xmath2 , which we define as @xmath32 . to normalize with respect to crystal volume , the extinction - free (
1 2 1 ) bragg peak is used as a reference for each sample . the structural factor @xmath33
is calculated for each peak using the relation @xmath34 , where @xmath35 is the measured intensity and @xmath36 is the scattering angle .
the approximate atomic displacements are now calculated for each dopant level by comparing the ratio @xmath37 for each zn - doped sample to the known ratio in the undoped case .
the numerical results are presented in table [ tab ] . as the zn - doping increases ,
the average atomic distortion is reduced dramatically ; by adding only 1% zn the dimerizing atomic shift loses 1/3 of its magnitude .
et al._@xcite found that the af ordering is associated with a superlattice reflection at @xmath38 .
the magnetic moment of the af state can be calculated from the observed intensity of this peak .
using the ( 0 2 1 ) peak as an extinction - free normalization , we compare the @xmath39 ratio to the ratio previously measured by hase _
et al._@xcite for a different 3.4% zn - doped crystal .
calculated t he magnetic moment , @xmath40 , extrapolated to zero temperature for that sample to be @xmath41 . by comparing the normalized superlattice peak intensities , values for the magnetic moments of each of the zn - doped samples
are obtained , and are summarized in table [ tab ] . in all zn - doped samples
studied the sp dimer superlattice peak loses some intensity below @xmath14 .
since the superlattice peak intensity is directly related to the atomic displacements ( @xmath42 ) this observation shows that the dimers are indeed adversely affected by the onset of a nel ordering .
figure [ delta_mu ] shows the sp atomic dimerization @xmath31 at 6 k and the af magnetic moment @xmath40 at 0 k as a function of zn concentration @xmath3 .
@xmath40 is found to scale with @xmath25 very well , showing a linear relation , as demonstrated in the bottom panel of figure [ delta_mu ] . as mentioned previously , the resolution - limited peak width shown in fig .
[ order_parameters](b ) indicates a long - range af ordering ; this is true for all samples .
however , the sp reflections exhibit line broadening for 3.4 and 4.7% zn - doped samples .
in the sp state of cugeo@xmath2 excitations were observed to have an energy gap of 2.1 mev at @xmath43 [ ] .
et al._@xcite reported that the magnetic excitations in a 3.4% zn - doped cugeo@xmath2 crystal are overdamped .
magnetic excitations in si - doped cugeo@xmath2 have also been recently reported by regnault _
et al._@xcite where , in contrast to hase _
_ , the excitations remain well defined ( are underdamped ) .
extensive measurements of the magnetic excitation spectra were obtained in all of the zn - doped crystals of the present study .
some preliminary results were presented by sasago _
et al._@xcite in the subsequent examination of all other compositions , well - defined ( underdamped ) magnetic excitations were found .
therefore the 3.4% zn - doped crystal used by hase _
et al._@xcite is the only sample in which overdamped excitations were observed .
the measurements were extended to a wider range of @xmath15 within the ( 0 @xmath30 @xmath29 ) zone to determine the dispersion along @xmath44 and @xmath45 .
a summary of the measured dispersions is presented in fig .
[ sp_dispersion ] . as in pure cugeo@xmath2 ,
the dispersion is much steeper in the @xmath45 direction compared to @xmath44 .
a small decrease in the sp energy gap magnitude is observed throughout the zone upon increasing zn concentration .
however the excitations remain well defined and are only somewhat broadened .
this broadening is demonstrated in fig .
[ sp_excitation ] , where an excitation profile at @xmath46 for 3.2% zn is compared with those of pure and 0.9% zn samples . by convolving the proper resolution function with a dispersion surface , the intrinsic line widths @xmath47 are extracted for 0 , 0.9% and 3.2% zn to be @xmath48 , 0.31 , and 1.09 mev , respectively .
the solid lines in the figure are the resultant fits . to try to clear up the discrepancy between the present results and those of hase _
et al._@xcite , the 3.4% zn sample used in that earlier study was re - examined .
this sample was found to have a reasonably well defined peak only at higher energy transfer .
thus the overdamped modes observed in the hase sample are limited to low energy sp magnetic excitations .
we speculate that the af modes may be underdamped in hase s 3.4% zn sample .
regnault and co - workers@xcite reported finding sharp af magnetic excitations at energies of 0.2 mev at the magnetic zone center in their 0.7% si doped sample . using electron spin resonance ( esr ) hase _ et al._@xcite recently observed an antiferromagnetic resonance below @xmath49k in a cu@xmath50zn@xmath51geo@xmath2 crystal , which indicates spin - wave excitations ( af modes ) at the magnetic - zone center at energies of 0.11 and 0.17 mev.@xcite we therefore searched for these af excitation modes in our neutron studies of our best characterized higher doped crystal which has 3.4% zn . in our search for af excitations , we encountered considerable difficulties mostly due to various kinds of spurious peaks appearing through the resolution function . after experimenting with two incident neutron energies and three different scattering zones ,
the proper experimental window was found in which the af excitations can be characterized in a wide enough @xmath15@xmath52 range .
figure [ sp_af_excitations_lowe ] depicts how the af mode is sharp at low @xmath15 values , but becomes significantly broadened at larger values of @xmath44 .
figure [ af_excitation_a ] demonstrates the dispersion and broadening which occurs as a function of @xmath15 along @xmath53 .
the af mode along @xmath45 appears to be broad even for small @xmath15 .
it seems that the broadening occurs when the transfer energy , rather than @xmath15 , exceeds a certain value .
it is interesting to note that the broadening at higher @xmath15 values is exactly opposite to the sp modes which are sharp at high @xmath15 and broaden dramatically at smaller @xmath15 .
the af excitation data are combined into full dispersion relations which are presented in figure [ sp_af_dispersion_bc ] , along with @xmath15 maps showing where the measurements were carried out .
as we have already emphasized , the af ordering in the cu@xmath0zn@xmath1geo@xmath2 system takes place for as little as 0.42% zn which has a @xmath14 = 0.6 k. it is now established that a very small zn concentration is sufficient for long range af ordering and that this long - range af order is present in all zn concentrations studied .
these remarkable results have immediate consequences on theoretical models .
percolation type descriptions are ruled out as they would predict the onset of af only after a certain amount of impurity islands have been formed ; when the amount of zn is too small , long range order could not exist in such a system .
the present results instead lend weight to a theoretical model recently proposed by fukuyama _
et al._@xcite which predicts a standing wave long range order for arbitrarily small amounts of an impurity within a sp system , probably even at @xmath54 for perfect crystals .
our data of af ordering in a 0.42% zn crystal strongly support this theoretical prediction .
in contrast to the long - range correlations in the af state , the sp phase exhibits increasing disorder with increasing zn concentration ; the sp superlattice reflection at ( 1.5 , 1 , 1.5 ) shows considerable line broadening with increasing @xmath3 .
two recent x - ray studies@xcite generally agree with our results , one measures the line widths , and the other takes a new approach using critical scattering
. it will be interesting to learn how the sp phase is gradually destroyed at high @xmath3 values
. it may be in this aspect where the distinction between site substitution of zn for cu is different from the bond disorder in substitution of si for ge .
the ability to grow uniform , large @xmath3 single crystals of cu@xmath0zn@xmath1geo@xmath2 is necessary for further quantitative studies .
as discussed in the sample preparation description above , this ability is still forthcoming . in the present study , evidence for af and sp orderings
have been presented along with detailed quantitative results for af magnetic moment @xmath40 and dimerization @xmath31 as a function of zn concentration @xmath3 .
one important finding is that @xmath40 scales well with @xmath14 .
in addition , key characteristics of sp and af magnetic excitations have been established .
contrasting behaviors are found in these excitations .
the af excitations show a distinctive peak in the small @xmath15 and energy region .
as the energy transfer increases , the af excitation rapidly broadens , and above 1 mev it becomes indistinguishable from the background . on the other hand ,
the sp excitations show considerable line - broadening as the zn concentration is increased , and this peak broadens at smaller @xmath15 as well .
these results will serve as critical tests for forthcoming theoretical models of this fascinating compound .
so far the problem has been treated as if the zn content @xmath3 is the only parameter with which to characterize the cu@xmath0zn@xmath1geo@xmath2 system
. there may be other issues at play , such as exact oxygen stoichiometry ( which can have dramatic effects in other well - known oxide systems ) .
closer inspection of table i reveals immediately that @xmath14 , @xmath40 , and @xmath31 are not smooth functions of @xmath3 .
it may , in part , be due to inaccuracies in the experiments , but the variation is too large to be completely assigned to experimental errors . and the fact that the hase 3.4% sample does not quite fit into the overall scheme remains a puzzle .
another example of how much is still to be learned about the cugeo@xmath2 system is the variance of @xmath10 even for @xmath55 .
an undoped crystal which is our most perfect crystalographically has a @xmath10 of 13.2 k , one full degree lower than the accepted value of 14.2 k. after writing this paper we became aware of a manuscript by coad _
et al._@xcite which discusses elastic neutron measurements of cugeo@xmath2 where cu has been replaced by zn or ni .
they similarly report a degradation of the sp state with the onset of af ordering and their resultant phase diagram is close , although not identical , to the one reported here .
overall their work and the present results lend support to each other . we would like to thank r.j .
birgeneau , guillermo castilla , and vic emery for stimulating discussions .
thanks to s. coad , h. fukuyama , and l.p .
regnault for providing their manuscripts prior to publication and for informative private discussions .
we also are pleased to acknowledge the expert technical help of j. biancarosa , n. donahue , b. liegel , k. mohanty , and r. rothe of the bnl hfbr physics staff .
this work was supported in part by the u.s.- japan cooperative research program on neutron scattering , and a nedo ( new energy and industrial technology development organization ) international joint research grant .
research at brookhaven national laboratory was supported by the division of materials research at the u.s .
department of energy , under contract no .
de - ac02 - 76ch00016 .
m. hase , i. terasaki , and k. uchinokura , phys .
* 70 * , 3651 ( 1993 ) . k. hirota , d.e .
cox , j.e .
lorenzo , g. shirane , j.m .
tranquada , m. hase , k. uchinokura , h. kojima , y. shibuya , and i. tanaka , phys .
lett . * 73 * , 736 ( 1994 ) .
m. nishi , o. fujita , and j. akimitsu , phys .
b * 50 * , 6508 ( 1994 ) . k. hirota , g. shirane , q. j. harris , q. feng , r. j. birgeneau , m. hase , and k. uchinokura , phys .
b * 52 * , 15412 ( 1995 ) . s. katano , o. fujita , j. akimitsu and m. nishi , phys . rev .
b * 52 * , 15364 ( 1995 )
. m. nishi , o. fujita , j. akimitsu , k. kakurai , and y. fujii , phys .
b * 52 * , r6959 ( 1995 ) .
et al . _ ,
56 * , 371 ( 1986 ) .
_ , europhys . lett .
* 3 * , 945 ( 1987 ) .
michael c. martin , g. shirane , y. fujii , m. nishi , o. fujita , j. akimitsu , m. hase , and k. uchinokura , phys .
b * 53 * , r14713 ( 1996 ) .
lussier , s.m .
coad , d.f .
mcmorrow , and d. mck .
paul , j. phys . condens .
matter * 8 * , l59 ( 1996 ) .
bonner and m.e .
fisher , phys . rev . * 135 * , a640 ( 1964 ) . m.c .
cross and d.s .
fisher , phys .
b * 19 * , 402 ( 1979 ) ; m.c .
cross , phys .
b * 20 * , 4606 ( 1979 ) .
m. hase , i. terasaki , y. sasago , k. uchinokura and h. obara , phys .
* 71 * , 4059 ( 1993 ) .
m. hase , n. koide , k. manabe , y. sasago , k. uchinokura , and a. sawa , physica b * 215 * , 164 ( 1995 ) .
oseroff , s .- w .
cheong , b. aktas , m.f .
hundley , z. fisk , and l.w .
rupp , jr .
lett . * 30 * , 74 ( 1995 ) . j.g .
lussier , s.m .
coad , d.f .
mcmorrow and d. mck .
paul , j. phys .
matter * 7 * , l325 ( 1995 ) .
m. hase , k. uchinokura , r.j .
birgeneau , k. hirota and g. shirane , j. phys .
japan * 65 * , 1392 ( 1996 ) .
y. sasago , n. koide , k. uchinokura , michael c. martin , m. hase , k. hirota , and g. shirane , phys .
b * 54 * , r6835 ( 1996 ) .
renard , k. ledang , p. veillet , g. dhalenne , a. revcolevschi , and l.p .
regnault , europhys . lett . * 30 * , 475 ( 1995 ) .
m. poirier , r. beaudry , m. castonguay , m.l .
plumer , g. quirion , f.s .
razavi , a. revcolevschi and g. dkalenne , phys .
b * 52 * , r6971 ( 1995 ) .
regnault , j.p .
renard , g. dhalenne , and a. revcolevschi , europhys . lett . * 32 * , 579 ( 1995 )
. h. fukuyama , t. tanimoto , and m. saito , j. phys .
japan * 65 * , 1182 ( 1996 ) .
m. hase , m. hagiwara , and k. katsumata , phys . rev .
b * 54 * , r3722 ( 1996 ) .
j. p. schoeffel , j. p. pouget , g. dhalenne , and a. revcolevschi , phys .
b * 53 * , 14971 ( 1996 ) . v. kiryukhin , b. keimer , j. p. hill , s. m. coad , and d. mck .
paul , phys .
b * 54 * , 7269 ( 1996 ) .
s. coad , j .- g .
lussier , d.f .
mcmorrow , and d. mck .
paul , j. phys .
matter * 8 * , 6251 ( 1996 ) .
c|c|d|dd|dd identification&zn&crystal volume & & + & @xmath3 ( % ) & ( @xmath17)&@xmath10 ( k)&@xmath31(6k)&@xmath14 ( k)&@xmath40(0k ) + n 1 & [email protected]&0.03&13.6&0.72&0.6&0.06 + & 0.74 & 0.01&12.5&0.60&2.4&0.14 +
n 2 & [email protected]&0.4 & 12.5 & & 2.0&0.10 + n 6 & 2.1 @xmath560.3 & 0.4 & 10.2&0.26&4.2&0.24 + n 4 & 3.2 @xmath560.4 & 0.4 & 10.3&0.40&4.1&0.18 + hase 4 & 3.4 & 0.04&10.0 & & 4.2&0.22 + n 8 & 4.7 & 0.1 & 10.0 & & 4.3&0.19 | comprehensive neutron scattering studies were carried out on a series of high - quality single crystals of cu@xmath0zn@xmath1geo@xmath2 .
the zn concentration , @xmath3 , was determined for each sample using electron probe micro - analysis .
the measured zn concentrations were found to be 40 - 80% lower than the nominal values .
nevertheless the measured concentrations cover a wide range which enables a systematic study of the effects due to zn - doping .
we have confirmed the coexistence of spin - peierls ( sp ) and antiferromagnetic ( af ) orderings at low temperatures and the measured phase diagram is presented .
most surprisingly , long - range af ordering occurs even in the lowest available zn concentration , x=0.42% , which places important constraints on theoretical models of the af - sp coexistence .
magnetic excitations are also examined in detail .
the af excitations are sharp at low energies and show no considerable broadening as @xmath3 increases indicating that the af ordering remains long ranged for @xmath3 up to 4.7% .
on the other hand , the sp phase exhibits increasing disorder as @xmath3 increases , as shown from the broadening of the sp excitations as well as the dimer reflection peaks . |
in recent years frustrated quantum spin systems on regular two - dimensional ( 2d ) lattices have aroused a great deal of research interest .
@xcite in particular the interplay of magnetic frustration and quantum fluctuations has been seen to be a very effective route to destabilize or destroy magnetic order and thereby to create new quantum phases .
such 2d magnetic systems can thus in turn develop a diverse array of phases with widely different ordering properties , such as antiferromagnets with quasiclassical nel ordering , quantum `` spirals '' , valence - bond crystals / solids , phases with nematic ordering , and spin liquids .
other factors that influence the ground - state ( gs ) phase structures are the nature of the underlying crystallographic lattice , the number and nature of the bonds on this lattice , and the spin quantum numbers of the atoms localized to the sites on the lattice .
the theoretical investigation of these models has proceeded hand in hand with the discovery and experimental investigation of ever more quasi-2d magnetic materials with novel properties .
one of the most intensively studied of all of the frustrated 2d models is the spin-@xmath0 @xmath1@xmath2 heisenberg antiferromagnet ( haf ) on the square lattice with nearest - neighbor ( nn ) bonds ( of strength @xmath19 ) competing with next - nearest - neighbor ( nnn ) bonds ( of strength @xmath20 ) .
this quantum system has two different quasiclassical phases with collinear magnetic long - range order ( lro ) at small ( @xmath21 ) and large ( @xmath22 ) values of the frustration strength parameter @xmath23 , separated by an intermediate quantum paramagnetic phase with no magnetic lro in the regime @xmath24 .
interest in this model has been greatly stimulated recently by its experimental realization in such layered magnetic materials as li@xmath25vosio@xmath26,@xcite li@xmath25vogeo@xmath26,@xcite and vomoo@xmath26.@xcite the syntheses of such layered quasi-2d materials has stirred up a great deal of renewed interest in the model ( and see also , e.g. , refs .
[ ] ) . amongst several methods
that have been very successfully applied to the @xmath1@xmath2 model has been the coupled cluster method ( ccm),@xcite which has also been applied to many similar strongly - interacting and highly frustrated spin - lattice models with comparable success .
other frustrated 2d models that have similarly engendered great recent interest include the spin-@xmath0 hafs on the triangular@xcite and kagome lattices.@xcite there has been a large amount of recent experimental investigation of the properties of quasi-2d magnetic materials with a ferromagnetic ( fm ) nn coupling ( @xmath27 ) and an antiferromagnetic ( afm ) nnn coupling ( @xmath28 ) .
examples include pb@xmath29vo(po@xmath30)@xmath29,@xcite ( cucl)lanb@xmath29o@xmath31,@xcite srznvo(po@xmath30)@xmath29,@xcite bacdvo(po@xmath30)@xmath29,@xcite pbznvo(po@xmath30)@xmath29,@xcite and ( cubr)lanb@xmath25o@xmath32.@xcite these experimental studies have also served to reignite interest in the theoretical investigation of the gs and thermodynamic properties of the fm @xmath1@xmath2 model , i.e. , the model with fm nn exchange ( @xmath27 ) and frustrating afm nnn exchange ( @xmath5 ) .
@xcite interestingly , arguments for the existence of a spin - nematic phase between two quasiclassical magnetically - ordered phases were presented.@xcite on the other hand , the existence of such a non - classical magnetically - disordered phase was also questioned in ref .
other systems that have grown in importance in the last few years are various spin-@xmath0 magnetic models defined on the 2d honeycomb lattice .
several such systems have been both theoretically and experimentally studied @xcite intensively , partly because of their special properties and partly due to the recent syntheses of various quasi-2d honeycomb - lattice materials .
one reason for the theoretical interest in such models on the 2d honeycomb lattice is that a spin - liquid phase has been found for the exactly solvable kitaev model,@xcite in which the spin-@xmath0 particles reside on just such a lattice .
furthermore , the honeycomb lattice is obviously germane to the very active research field of graphene , where the relevant physics may well be described by hubbard - like models on this lattice.@xcite interestingly , meng _
et al._@xcite found that for the hubbard model on the honeycomb lattice with moderate values of the coulomb repulsion parameter @xmath33 , strong quantum fluctuations lead to an insulating spin - liquid phase between the non - magnetic metallic phase and the afm mott insulator phase . from the experimental side recent observations on the spin-@xmath34 honeycomb - lattice haf bi@xmath35mn@xmath26o@xmath36(no@xmath35 ) demonstrate a spin - liquid - like behavior at temperatures much lower than the curie - weiss temperature.@xcite we have recently studied@xcite the afm @xmath1@xmath2@xmath3 honeycomb model for the case where the spin quantum number @xmath37 of each of the spins on every lattice site is @xmath38 , and with afm nearest - neighbor exchange bonds ( @xmath39 ) in the presence of frustration caused by afm nnn bonds ( @xmath5 ) and with next - next - nearest neighbor ( nnnn ) bonds of strength @xmath3 also present , for the special case where @xmath7 .
we found@xcite that the scenario of deconfined criticality may hold for this model ( and see also ref .
to date there exist only limited studies of the corresponding fm @xmath1@xmath2@xmath3 model ( namely where @xmath27 ) . in this paper , we further the investigation into the fm @xmath1@xmath2@xmath3 honeycomb model with fm nn bonds ( of strength @xmath27 ) in the presence of frustrating afm nnn bonds ( of strength @xmath5 ) and nnnn bonds ( of strength @xmath40 ) .
once again we consider only the interesting special case where @xmath7 .
we focus our attention in the present study particularly on the detection and characterization of the gs phases of the quantum model .
bearing in mind the controversial discussion of the corresponding @xmath1@xmath2 square - lattice model with fm nn exchange bonds ( @xmath27 ) , the question naturally arises as to whether any indications for a non - classical magnetically - disordered phase might now be found for the honeycomb model .
to determine the relevant gs phases and their properties we calculate the gs energy , the spin - spin correlation function , and the magnetic order parameter for the stripe - ordered state discussed below that is present as a gs phase in the corresponding classical version ( equivalent to the @xmath41 limit ) of the model . in view of its proven past ability to give results of high accuracy for a wide variety of highly frustrated 2d spin - lattice models
, we again use the coupled cluster method ( ccm ) as our main computational tool in this paper .
additionally , we use the exact diagonalization ( ed ) method for a large finite lattice of @xmath42 spins as a validity check of our ccm results .
since at the classical level the model now under consideration also exhibits some similarities with the corresponding model with afm nn bonds ( @xmath39 ) , we compare our results for the quantum model of the fm case ( @xmath43 ) with those of the corresponding afm case ( @xmath44 ) .
the rest of the paper is organized as follows . after describing the model in sec .
[ model_section ] , we apply the ccm to investigate its gs properties .
the ccm itself is very briefly described in sec .
[ ccm ] , before presenting and discussing our ccm and ed results in sec .
[ results ] .
we conclude in sec .
[ conclusions ] with a summary of the main results .
the hamiltonian of the spin-@xmath0 @xmath1@xmath2@xmath3 heisenberg model on the honeycomb lattice , which we studied recently@xcite for the afm case ( @xmath39 ) is defined as @xmath45 where @xmath46 runs over all lattice sites on the lattice , and where @xmath47 runs over all nn sites connected to site @xmath46 by @xmath1 bonds , @xmath48 runs over all nnn sites connected to site @xmath46 by @xmath2 bonds , and @xmath49 runs over all nnnn sites connected to site @xmath46 by @xmath3 bonds , but counting each bond once and once only in the three sums .
each site @xmath46 of the lattice carries a spin-@xmath0 particle with spin operator @xmath50 .
we note that precisely the same model has also been studied recently on the square lattice , both in the case where all the bonds are afm in nature,@xcite and in the fm case where @xmath27 and @xmath51.@xcite the aim of the present work is now to study further the spin-@xmath0 @xmath1@xmath2@xmath3 fm model ( namely the above model in the case @xmath27 ) on the honeycomb lattice.@xcite the lattice and the exchange bonds are illustrated in fig .
[ model](a ) . the classical gs phase diagram for the @xmath1@xmath2@xmath3 afm model ( with @xmath39 ) on the honeycomb lattice model displays collinear nel and striped phases , both afm in nature , as well as a spiral phase .
these phases meet in a triple point at @xmath52 ( and for more details see , e.g. , ref .
[ ] ) . for the remainder of this paper we again focus on the case where @xmath53 , but now where the nn exchange bond is fm in nature ( @xmath27 ) .
the gs energies of the only two corresponding classical collinear states are then given by @xmath54 for the fm state and collinear striped afm state shown in figs .
[ model](a ) and ( b ) respectively . if these were the only gs phases in this @xmath27 regime we would thus have a classical transition between the fm state and the striped afm state at @xmath55 ( @xmath10 for @xmath4 ) and a classical energy per site at this point of @xmath56 for the @xmath38 system with @xmath4 . for the corresponding afm model with @xmath39 such a striped afm state also exists as stated above , but the classical transition between the afm nel state and the striped afm state is at @xmath57 .
the reason why the corresponding phase transition in the fm model occurs at a smaller value of the frustration parameter , @xmath58 than the value @xmath59 for the afm model is due to the @xmath3(@xmath60 ) nnnn exchange bonds that act to frustrate the fully polarized fm state , whereas they reinforce the afm nel state .
by contrast , the @xmath2(@xmath60 ) nnn exchange bonds act to frustrate the @xmath61 bonds for both the fm state of the fm model and the nel state of the afm model .
we note that the classical fm state is also an eigenstate of the hamiltonian , with energy eigenvalue equal to the energy of its classical fm counterpart .
we note , however , that in fact the classical @xmath1@xmath2@xmath3 heisenberg model on the honeycomb lattice with @xmath27 , @xmath5 , and @xmath6 also has a spiral phase that intervenes in a very narrow strip between the fm phase and the collinear striped afm phase .
( in refs .
[ ] this is referred to as phase v. ) the region in the @xmath62-@xmath63 plane ( where @xmath64 and @xmath65 ) in which it is the stable gs phase in the case @xmath27 is bounded by the three curves ( i ) @xmath66 , @xmath67 , ( ii ) @xmath68 , @xmath69 , and ( iii ) @xmath70 $ ] , @xmath71 .
the point @xmath72 is a classical tetracritical point at which the spiral phase v meets the fm phase , the striped collinear afm phase , and the afm nel phase ( and see fig . 3 of ref .
[ ] for further details ) .
thus , in our case , where @xmath7 and @xmath27 , the classical spiral phase v exists in the narrow region @xmath73 .
naturally this includes the point @xmath58 discussed above at which the fm and striped collinear afm phases would meet in the absence of the spiral phase v as a stable gs phase intervening between them . in all of our results below for the fm @xmath1@xmath2@xmath3 honeycomb system we henceforth set @xmath43 with no loss of generality , since this simply sets the overall scale of the hamiltonian , and we consider the case where @xmath53 , such that both the nnn and nnnn bonds act to frustrate the ferromagnetism
the ccm ( see , e.g. , refs . [ ] and references cited therein ) that we use here is one of the most powerful and most versatile modern techniques in quantum many - body theory .
it has been used to study various quantum magnets ( see , e.g. , refs . [ ] and references cited therein ) very successfully .
the method is particularly suitable for investigating frustrated systems , due to the fact that some of the main alternative methods are restricted by certain problems that arise in such cases .
for instance , quantum monte carlo ( qmc ) techniques suffer from the infamous and well - known `` sign problem '' for such systems .
the exact ed method is also usually restricted by available computational power to relatively small finite - sized lattices .
nevertheless it can often be used , as here , to provide a handy tool to check and validate the results of other numerical or approximate methods .
we briefly describe here some of the important features of the ccm as applied to spin - lattice problems ( and see , e.g. , refs . [ ] and references cited therein for further details ) .
the starting point for any ccm calculation is to select a normalized state @xmath74 as a reference or model state against which to incorporate in a systematic and potentially exact fashion the correlations present in the exact ground state .
we often use a relevant classical ground state as the model state for spin systems for the sake of convenience , but other appropriate states may certainly also be used . in order to treat each site equivalently , a mathematical rotation of the local axes of the spins is conveniently performed in such a way that all spins in the reference state align along the same direction , say the negative @xmath75-axis .
clearly , such rotations leave unchanged the su(2 ) commutation relations between components of the spin operators .
the exact ket and bra gs energy eigenstates , @xmath76 and @xmath77 , of the many - body system are then parametrized in the ccm form as : 0.2 cm @xmath78 @xmath79 where @xmath80 are the schrdinger gs ket and bra equations respectively .
the multiconfigurational creation operators @xmath81 are defined so that @xmath82 , and where we have defined @xmath83 .
they are required to form a complete set of mutually commuting many - body creation operators in the hilbert space , defined with respect to @xmath74 as a cyclic vector .
clearly the states are normalized such that @xmath84 . for spin - lattice systems
they take the form of multi - spin raising operators and are written as products of single - spin raising operators , @xmath85 , where @xmath86 .
the gs energy is calculated in terms of the correlation coefficients @xmath87 as @xmath88 ; and the average on - site magnetization @xmath89 in the rotated spin coordinates is calculated equivalently in terms of the coefficients @xmath90 as @xmath91 , which now plays the role of the order parameter . finally , the complete set of unknown ket- and bra - state correlation coefficients @xmath92 is calculated by requiring the expectation value @xmath93 to be a minimum with respect to all parameters @xmath94 .
this readily leads to the coupled set of nonlinear equations for the ket - state creation correlation operators @xmath87 , @xmath95 , and to the coupled set of linear equations , @xmath96 , which can then be solved for the bra - state destruction correlation operators @xmath97 . when all many - body configurations @xmath98 are included in the expansions of the correlation expansions operators @xmath99 and @xmath100 , the ccm formalism is exact .
however , it is necessary of course in practice to use approximation schemes to truncate the sets of configurations @xmath98 contained in the expansions of eqs .
( [ eq : ket_eq ] ) and ( [ eq : bra_eq ] ) for the ccm correlation operators . for systems defined on a regular periodic spatial lattice as here , it is convenient to use the well - established lsub@xmath101 approximation scheme in which all possible multi - spin - flip correlations over different locales on the ( here , honeycomb ) lattice defined by @xmath101 or few contiguous lattice sites are retained .
clusters are defined to be contiguous in this sense if every site in the cluster is adjacent ( as a nearest neighbor ) to at least one other site in the cluster .
this is the scheme we use for all our results presented below .
the number @xmath102 of independent fundamental clusters ( i.e. , those that are inequivalent under the symmetries of the hamiltonian and of the model state ) increases rapidly with the truncation index @xmath101 , as shown in table [ table_fundconfig ] .number of fundamental lsub@xmath101 configurations ( @xmath102 ) for the collinear striped afm state of the spin-@xmath0 @xmath1@xmath2@xmath3 honeycomb model . [ cols="^,^",options="header " , ] for the present spin-@xmath0 @xmath1@xmath2@xmath3 model on the honeycomb lattice , where we use the natural lattice geometry itself to define the notion of adjacency inherent in the definition of the lsub@xmath101 scheme .
we see , for example , that the number @xmath102 of such fundamental clusters ( and hence the number of simultaneous nonlinear equations we need to solve for the retained correlation coefficients @xmath87 ) for the striped model state is 250891 at the highest lsub12 level of approximation that we utilize here .
the corresponding numbers , @xmath103 , of fundamental configurations are appreciably higher at a given lsub@xmath101 level when the spiral phase v is used as the ccm model state , due to the considerably reduced symmetry .
it is necessary to use massive parallelization and supercomputing resources in order to perform the ccm calculations at such high level of approximation.@xcite thus , for example , to obtain a single data point ( i.e. , for a given value of @xmath7 ) for the striped model state at the lsub12 level typically requires about 0.5 h computing time using 1000 processors simultaneously .
we present ccm results below based on the striped collinear afm state as model state , at various lsub@xmath101 levels of approximation with @xmath104 , and also in the corresponding @xmath105 extrapolated limits ( lsub@xmath106 ) based on the well - tested extrapolation schemes described below and in more detail elsewhere.@xcite we have also performed extrapolations for the data set with @xmath107 .
both sets of results agree well with one another , which gives added credence to our results .
note that we do not use the lsub@xmath101 approximation scheme for values @xmath108 of the truncation index , since these low - order approximations will not capture the natural hexagonal structure of the lattice .
we remark that , as always , the ccm exactly obeys the goldstone linked - cluster theorem at every lsub@xmath101 level of approximation .
hence we work from the outset in the limit @xmath109 , where @xmath110 is the number of sites on the honeycomb lattice , and extensive quantities such as the gs energy are always guaranteed to be linearly proportional to @xmath110 in this limit .
we clearly do not need to perform any finite - size scaling of our results , as all ccm approximations are automatically performed from the outset in the infinite - lattice limit , @xmath111 , as discussed above .
it is , however , necessary to extrapolate to the exact @xmath105 limit in the lsub@xmath101 truncation index @xmath101 , in which limit the complete ( infinite ) hilbert space is reached .
for the gs energy per spin , @xmath112 , a well - tested and very accurate extrapolation ansatz ( and see , e.g. , refs . ,
[ ] ) is @xmath113 while for the magnetic order parameter , @xmath89 , different schemes have been employed for different situations . for models showing no or only relatively small amounts of frustration , a well - tested and accurate rule ( and see , e.g. , refs . [ ] ) is @xmath114 for highly frustrated systems , particularly those showing a gs order - disorder transition , a more appropriate extrapolation rule with fixed exponents that has been found to give good results ( and see , e.g. , refs .
[ ] ) is @xmath115 we give illustrations here of the use of each of these schemes , wherever and whenever possible .
we now present and discuss our ccm results . in order to have an independent check on the accuracy and consistency of our ccm results ,
we have also performed additional computations of selected gs properties of the present models using the ed technique that is a well - established and successful tool for studying frustrated quantum spin systems ( and see , e.g. , refs . [ ] ) . in fig .
[ e](a ) we show the ccm results for the gs energy per spin , @xmath112 , in various lsub@xmath101 approximations based on the striped state as ccm model state , as well as the exact gs energy for a finite lattice of size @xmath42 .
we also show separately , in fig .
[ e](b ) , the extrapolated ( @xmath105 ) results obtained from eq .
( [ e_extrapo ] ) using the data set @xmath104 .
comparison is made with the results for the corresponding afm version of the model with @xmath12 .
the ccm lsub@xmath101 data displayed in fig . [ e](a ) show that the gs energy results converge extremely rapidly as the truncation index @xmath101 is increased , such that the difference between the lsub12 results and the extrapolated ( @xmath105 ) results obtained from eq .
( [ e_extrapo ] ) is very small indeed .
we note too that , just as in the corresponding afm case of the model with @xmath12 , the various ccm lsub@xmath101 solutions based on the striped model state now also terminate at some lower termination point @xmath116 as @xmath2 is decreased .
such terminations of ccm solutions are very common and have been very well documented.@xcite in all such cases a termination point always arises due to the solution of the ccm equations becoming complex at this point , beyond which there exist two branches of entirely unphysical complex conjugate solutions.@xcite in the region where the solution reflecting the true physical solution is real there actually also exists another ( unstable ) real solution . however , only the shown branch of these two solutions reflects the true ( stable ) physical ground state , whereas the other branch does not .
the physical branch is usually easily identified in practice as the one which becomes exact in some known ( e.g. , perturbative ) limit .
this physical branch then meets the corresponding unphysical branch at some termination point with infinite slope , beyond which no real solutions exist .
the lsub@xmath101 termination points are themselves also reflections of the quantum phase transitions in the real system and may be used to estimate the position of the phase boundary,@xcite although we do not do so for this critical point in the fm model , since we have more accurate criteria that we now discuss .
we note first from fig .
[ e](a ) that the lsub@xmath101 termination points using the striped state as the ccm model state for the present fm version of the model with @xmath4 , lie very close indeed to the points where the curves cross ( or nearly cross ) the corresponding curve for the fm state given by eq .
( [ h_classical ] ) .
this gives us our first evidence that either there is no intermediate phase between the quantum striped phase and the fm phase for the case @xmath4 , or , if one exists , it can occur only over a very narrow regime indeed .
this situation may be contrasted with that of the afm version of the model ( @xmath12),@xcite where the lsub@xmath101 results for the gs energy using the striped model state terminate before they meet the corresponding results using the nel state as model state ( which themselves also terminated at some upper termination points that were lower in value than the lower termination points for the striped state ) . in the latter case there is an intermediate plaquette valence - bond crystal ( pvbc ) phase . at the classical level
the difference in the values of the gs energy per spin of the collinear striped states between the two @xmath38 cases ( i.e. , for positive and negative values of @xmath1 with @xmath117 ) is 0.25 , independent of @xmath2 and @xmath3 .
the quantum versions follow this pattern for larger values of @xmath7 , as seen from fig .
[ e ] , but the constancy in the difference breaks down at around @xmath118 , where the afm case ( @xmath12 ) exhibits a critical point marking a transition to the pvbc phase , which then in turn undergoes a further phase transition to the nel phase at another lower critical value .
the corresponding best available ccm estimates for those two critical values for the afm case of the model with @xmath119 are @xmath14 and @xmath17 respectively.@xcite in the present fm case of the model with @xmath120 we see no evidence ( apart from the seeming termination of the solutions to the equations for the lsub@xmath121 approximation based on the striped state as ccm model state very slightly before the gs energy crossing point with the fm state ) of any similar intermediate state between the fm state and the collinear striped afm state . if any such intermediate state exists at all , however , it must be confined to a very narrow region indeed around @xmath122 , probably confined to @xmath18 .
we return to a more detailed discussion of this region later . for the moment
we note only that it is much reduced from the region @xmath123 in which the corresponding classical version of the model has the spiral phase v as its stable gs phase .
for the present fm case with @xmath4 the ccm lsub@xmath101 gs energy curves using the striped model state cross the corresponding gs energy curve for the fm state from eq .
( [ h_classical ] ) for @xmath107 at corresponding critical values @xmath124 ( where @xmath125 ) , @xmath126 ( where @xmath127 ) , and @xmath128 ( where @xmath129 ) .
the corresponding lsub@xmath121 result for the gs energy using the striped state as the ccm model state appears to terminate just before meeting the gs energy curve for the fm phase .
however we note that for such very high - order ccm calculations it becomes very computationally expensive to determine the termination point with high accuracy .
if we use the extrapolated lsub@xmath106 results for the gs energy for the striped phase by making use of eq .
( [ e_extrapo ] ) and employing the whole data set @xmath104 , we thus need to perform a further very small extrapolation of the ccm results to lower values of @xmath2 to find the presumed crossing point of the energies of the striped and fm phases , in the scenario in which these two phases meet at a first - order transition with no intermediate phase ( that would itself be confined to the very narrow intervening region @xmath18 , as discussed above ) .
as expected , simple power - law expansions give very accurate fits , and give crossing points very close to those above .
putting all the energy data together , our best estimate for the critical point of the first - order phase transition from the collinear striped phase to the fm phase ( in the scenario where this transition occurs directly , with no intermediate phase confined to the narrow region @xmath18 ) for the spin-@xmath0 @xmath1@xmath2@xmath3 heisenberg ferromagnet ( with @xmath4 ) on the honeycomb lattice , and with @xmath53 , is @xmath130 , at which point the gs energy per spin is @xmath131 .
we see from fig .
[ e](a ) that the agreement between the ed ( @xmath42 ) and the ccm energies is very satisfactory .
moreover , due to the finite - size scaling of the gs energy , @xmath132 with @xmath133 ( and see , e.g. refs . [ ] and [ ] ) , the difference between the ccm and the ed gs energies would become even smaller if finite lattices of larger size could be considered .
the ed turnover point in the energy curve that marks the termination of the fm phase occurs at a value of about 0.1003 for the @xmath42 lattice used , and for the same reasons as above this value will increase as @xmath110 is increased .
thus , in summary , while the ccm estimates for the gs energy per spin for the spin-@xmath0 @xmath1@xmath2@xmath3 heisenberg model on the extended infinite honeycomb lattice are much more accurate than the ed results , the latter do serve as an independent check on the former . the hypothetical phase transition ( i.e. , when the existence of the intervening spiral phase v is momentarily ignored ) from fm order to collinear striped afm order for the classical version of the fm model with @xmath4 occurs at a value @xmath134 , compared with the corresponding value @xmath135 found here . thus quantum fluctuations act to stabilize the collinear afm order , at the expense of the fm order , to higher values of frustration than in the classical case . it is interesting to note that a similar situation was found in the fm version ( @xmath136 ) of the spin-@xmath0 @xmath1@xmath2 model on the square lattice,@xcite where a quantum critical point exists at a value @xmath137 for a similar transition from fm order to collinear striped order , compared with a corresponding classical value of @xmath138 .
it is well known , from many cases studied , that quantum fluctuations almost always favor collinear states over noncollinear ones ( e.g. , spiral or canted states ) .
what is interesting in both the present case and the spin-@xmath0 @xmath1@xmath2 model on the square lattice cited above , is that quantum fluctuations seem also to favor one collinear state ( namely the collinear striped afm state in these two cases ) where the quantum fluctuations are present , over another collinear state ( namely the fm state in these two cases ) where quantum fluctuations are absent .
it is intriguing to wonder whether these are examples of a more general rule .
we present results in fig .
[ m ] for the ccm collinear stripe order parameter @xmath89 , as defined in sec .
figure [ m](a ) shows lsub@xmath101 results with @xmath104 , while fig .
[ m](b ) shows the corresponding extrapolated ccm lsub@xmath106 ( @xmath139 ) results using both eqs .
( [ m_extrapo_standard ] ) and ( [ m_extrapo_frustrated ] ) .
we note firstly that the ccm lsub@xmath101 order parameter results depend on the approximation level @xmath101 much more strongly than those for the gs energy .
it is clear that the order parameter behaves similarly for large values of @xmath140 for both the fm model ( @xmath27 ) and the afm model ( @xmath39 ) .
however , once again there are considerable differences in the behavior of @xmath89 between the two models for values of the frustration parameter @xmath13 . the extrapolated ccm results for @xmath89 for the afm model in fig . [ m](b )
clearly show the breakdown of the quasiclassical collinear magnetic lro near the critical value of @xmath118 , i.e. , significantly above the classical transition point @xmath141 ( and see , e.g. , refs . [ ] ) .
indeed , the ccm estimate for the critical value of the frustration parameter in the afm case for the disappearance of collinear striped order is @xmath14 from the point at which @xmath89 becomes zero , using the extrapolation scheme of eq .
( [ m_extrapo_frustrated]).@xcite by contrast , the order parameter for the fm model stays almost constant over the whole parameter region shown in fig .
we do not observe any indication of the breakdown of the collinear striped magnetic lro until @xmath122 for the fm model , which is below the hypothetical classical transition point @xmath142 , as we observed previously in the results for the gs energy .
lastly , we present results for various spin - spin correlation functions for the fm as well as for the afm model in fig .
[ corrfunctn ] .
figure [ corrfunctn](a ) shows the ccm lsub10 results and fig .
[ corrfunctn](b ) shows the corresponding ed results .
once again we note that for large values of the frustration parameter @xmath140 the corresponding spin - spin correlations functions for both the fm ( @xmath4 ) and afm ( @xmath12 ) models agree remarkably well with one another for both the ccm and ed calculations .
furthermore , for the fm model the agreement of the ccm correlation functions with the ed data is excellent . for the afm model
the agreement between the ccm and ed results is again excellent for values of @xmath140 above the transition point at which the afm collinear striped order disappears , namely @xmath143 , but around and below this value there are noticeable differences . in particular ,
the very steep change in the correlation functions at @xmath144 present in the ed ( @xmath42 ) data for the afm model is not observed in the ccm lsub10 data .
instead the ccm data show a smoother change in that region . however , we have argued@xcite that for @xmath145 no striped magnetic lro order exists .
indeed we argued that no magnetic lro order exists at all for the afm model in the regime @xmath146 , where instead we have a pvbc state .
hence , it is not surprising that the ccm solution in a finite order of lsub@xmath101 approximation based on the collinear stripe reference state does not provide such accurate results for the correlation functions inside this magnetically disordered phase . to conclude , we return to examine more closely the very narrow region @xmath18 for which our ccm results based on the striped afm state as model state could not exclude the possibility of an intervening phase between the striped afm and the fm phases . in fig . [ corrfunctn_zoom ] shown in fig .
[ corrfunctn](b ) for the spin-@xmath0 @xmath1@xmath2@xmath3 model on the honeycomb lattice for the fm case with @xmath4 and @xmath7 using the ed method on a lattice of size @xmath42.,width=226 ] we show a more detailed view of the ed results for the same spin - spin correlation functions shown in fig . [
corrfunctn](b ) in this narrow region just above the fm transition point .
the ed data does definitely indicate the existence of a phase in precisely the region @xmath18 .
it is difficult from this data to say with any certainty whether or not the state is the quantum - mechanical remnant of the classical spiral phase v that exists in the classical regime @xmath123 .
furthermore , without ed calculations on larger lattices , for which the computational cost would be prohibitive , it is also not possible to say whether these results over such a narrow region are an artefact of the finite lattice size .
our results are summarized in sec .
[ conclusions ] .
in this paper we have presented results on the gs properties of the spin-@xmath0 @xmath1@xmath2@xmath3 heisenberg model with fm nn ( @xmath4 ) exchange bonds in the presence of frustrating afm nnn ( @xmath5 ) and nnnn ( @xmath6 ) exchange bonds of equal strength ( @xmath7 ) on the honeycomb lattice , using both the ccm and lanczos ed . by comparison with previous studies for the afm ( @xmath12 ) version of the model,@xcite we find similar behavior for both models for values @xmath147 , but for values of @xmath148
the models differ markedly . the results of the present paper for the fm version of the model and that of the previous paper @xcite for the afm version may conveniently be combined and summarised in the phase diagram shown in fig .
[ phase - diagram ] .
@xmath1@xmath2@xmath3 honeycomb model in the @xmath61-@xmath140 plane , for the case @xmath7 .
the continuous transition between the afm nel and pvbc phases at @xmath149 is shown by a broken line , while the first - order transition between the pvbc and afm striped phases at @xmath150 is shown by a solid line .
our results indicate that the transition between the striped afm and fm phases is either a first - order one at @xmath151 or occurs via an intermediate phase , probably with noncollinear spiral order , which exists in the region @xmath152 .
the region between the fm and afm nel phases with @xmath153 has not been investigated by us.,width=415 ] we note that , by contrast with the corresponding model with afm nn exchange ( @xmath12 ) we do not find indications for a non - classical magnetically - disordered phase for the model with fm nn exchange ( @xmath4 ) .
if such a phase exists at all it must be confined to a very small range of the frustration parameter around @xmath18 . however , any such phase is much more likely to be a quasiclassical remnant of the spiral phase v that exists in the corresponding classical model ( with @xmath4 ) in the parameter regime @xmath123 .
as expected , quantum fluctuations then usually favor a collinear phase over a noncollinear one , and the extent of any spiral phase is smaller in the quantum spin-@xmath0 case than in the classical ( @xmath154 ) case . in one scenario the results presented here for the case @xmath4 indicate a direct first - order transition between the two magnetically ordered phases , namely the fm ground state at small values of the frustration parameter @xmath140 and the striped collinear afm ground state at larger values of @xmath140 . our best estimate of the phase transition point is
then @xmath130 .
although in this scenario a quasiclassical gs phase ( viz .
, the collinear striped afm state ) exists in the whole parameter space down to the fm gs phase , the frustration might still have a strong effect on the low - temperature thermodynamics near the transition point at @xmath130.@xcite for values @xmath155 the fm multiplet becomes a low - lying excitation , and this might lead to an additional low - temperature peak in the specific heat @xmath156.@xcite we note that indications for such an additional low - temperature peak in @xmath156 were also found on the fm side near such a transition@xcite ( i.e. , at @xmath157 ) in other frustrated spin models .
in an alternative scenario our results also indicate the possibility of an intervening phase between the collinear fm and striped afm phases .
any such phase , however , is limited to lie within the very narrow range @xmath18 , as shown in fig . [ phase - diagram ] . in principle
we could more accurately establish the existence of such a phase as a quasiclassical remnant of the classical spiral phase v , and thence also more accurately establish its phase boundaries , by performing another comparable set of ccm calculations to those performed here with the striped afm state as model state , but using instead the spiral state v as model state .
such calculations would be much more onerous and computationally expensive , however , since on the one hand the number @xmath103 of fundamental ccm configurations at a given lsub@xmath101 level is greater for the spiral model state than for the striped model state and , furthermore , the ccm results would need to be optimized at a given lsub@xmath101 level with respect to the spiral pitch angle parameters by minimizing the corresponding result for the energy per spin separately for each set of values for the bond strength parameters .
we note finally that we have not yet investigated the present model in the case where @xmath153 . for the fm version of the model when @xmath27 also , the fm phase is then obviously the stable ground state .
conversely , when @xmath39 and frustration occurs , there is a direct first - order transition in the classical version of the model between the fm and nel afm states at a value @xmath158 . following the discussion in sec .
[ results ] we might expect that quantum fluctuations could again act either ( a ) to retain the direct transition but to stabilize the collinear afm order in preference to the fm order , thus pushing the phase boundary to a somewhat lower value , @xmath159 , for the spin-@xmath0 case ; or ( b ) to permit an intervening state with no classical counterpart . indeed , very preliminary ccm calculations indicate that scenario ( a ) is realized and that this corresponding critical point may be pushed to a value @xmath160 .
we hope to report in more detail on this region and to give a more accurate value of this phase boundary in a future paper . as discussed briefly in sec .
[ intro ] , it has been proposed @xcite that the competition between fm heisenberg interactions between nn pairs of spins and afm interactions between other spins in frustrated spin-@xmath0 systems on the square lattice could lead to gapless spin - liquid states with multipolar order ( e.g. , spin - nematic states ) adjacent to the fm state . similar states
have also been proposed to arise in frustrated multiple cyclic spin - exchange models on the triangular lattice with fm nn pairwise interactions , @xcite either in the presence of a magnetic field ( where octupolar order occurs ) or in its absence ( where quadratic or nematic ordering occurs in a state bordering the fm state ) . in the case of the frustrated honeycomb - lattice ferromagnet considered here
we have found no evidence for such states .
however , the multipolar - ordering phenomenon in the zero - field case considered here is evidently rather fragile , and in the square - lattice case for the spin-@xmath0 fm version of the @xmath61@xmath140 model ( i.e. , with @xmath27 ) even their existence has been questioned in recent rather accurate work @xcite that also employed both high - order ccm and ed techniques .
no evidence was found for such states either in a very recent schwinger boson study on the square lattice,@xcite using the same fm version of the spin-@xmath0 @xmath1@xmath2@xmath3 heisenberg model that we studied here on ther honeycomb lattice .
nevertheless , the history of the study of quantum magnets has shown us that the detection of phases with novel quantum ordering , such as nematic states of various kinds , is extremely subtle .
in particular , the present honeycomb - lattice model surely warrants further investigation before the absence of nematic states in the fm case discussed here is considered definite .
finally we mention that frustrated ferromagnets are also interesting with respect to multi - magnon bound states appearing in high magnetic fields ( and see , e.g. , refs .
the present model also warrants further investigation when the coupling to an external magnetic field is included .
we thank the university of minnesota supercomputing institute for digital simulation and advanced computation for the grant of supercomputing facilities , on which we relied for the numerical calculations reported here .
one of the authors ( djjf ) acknowledges and thanks the european science foundation for financial support under the research network program `` highly frustrated magnetism '' ( short visit grant number 3858 ) .
r. shindou , s. yunoki , and t. momoi , phys . rev .
b * 84 * , 134414 ( 2011 ) .
f. wang , phys .
b * 82 * , 024419 ( 2010 ) .
s. okumura , h. kawamuro , t. okubo , and y. motome , j. phys .
jpn . * 79 * , 114705 ( 2010 ) | we study the ground - state ( gs ) properties of the frustrated spin-@xmath0 @xmath1@xmath2@xmath3 heisenberg model on the two - dimensional honeycomb lattice with ferromagnetic nearest - neighbor ( @xmath4 ) exchange and frustrating antiferromagnetic next - nearest - neighbor ( @xmath5 ) and next - next - nearest - neighbor ( @xmath6 ) exchanges , for the case @xmath7 .
we use the coupled - cluster method implemented to high orders of approximation , complemented by the lanczos exact diagonalization of a large finite lattice with 32 sites , in order to calculate the gs energy , magnetic order parameter , and spin - spin correlation functions . in one scenario
we find a quantum phase transition point between regions characterized by ferromagnetic order and a form of antiferromagnetic ( `` striped '' ) collinear order at @xmath8 , which is below the corresponding hypothetical transition point at @xmath9 ( @xmath10 ) for the classical version of the model , in which we momentarily ignore the intervening noncollinear spiral phase in the region @xmath11 .
hence we see that quantum fluctuations appear to stabilize somewhat the collinear antiferromagnetic order in preference to the ferromagnetic order in this model .
we compare results for the present ferromagnetic case ( with @xmath4 ) to previous results for the corresponding antiferromagnetic case ( with @xmath12 ) .
the magnetic order parameter is found to behave similarly for the ferromagnetic and the antiferromagnetic models for large values of the frustration parameter @xmath2 . however , there are considerable differences in the behavior of the order parameters for the two models for @xmath13 . for example
, the quasiclassical collinear magnetic long - range order for the antiferromagnetic model ( with @xmath12 ) breaks down at @xmath14 , whereas the `` equivalent '' point for the ferromagnetic model ( with @xmath4 ) occurs at @xmath15 .
unlike in the antiferromagnetic model ( with @xmath12 ) , where a plaquette valence - bond crystal phase intrudes between the two corresponding quasiclassical antiferrmagnetic phases ( with nel and striped order ) for @xmath16 , with @xmath17 , we find no clear indications at all in the ferromagnetic model for an intermediate magnetically disordered phase between the corresponding phases exhibiting ferromagnetic and striped order .
instead the evidence for the ferromagnetic model ( with @xmath4 ) points to one of two scenarios : either there is a direct first - order transition between the two magnetically ordered phases , as mentioned above ; or there exists an intervening phase between them in the very narrow range @xmath18 , which is probably a remnant of the spiral phase that exists in the classical counterpart of the model over the larger range @xmath11 . |
theorists seek to capture the essence of protein folding with simple models of a self - interacting polymer chain @xcite .
there are two distinct limits pertaining to the nature of the interactions in this minimalist approach .
one is that of purely random interactions , and is considered too frustrated to describe real proteins .
another is the g model @xcite , where the polymer is self - attractive only for those parts of it in their native configurations .
this is considered too unfrustrated to describe real proteins , and also impossible to achieve in practice . as these two models bracket the behavior of real proteins , we consider perturbing from the g model towards real protein interactions by adding some non - native heterogeneity .
some of the effects of adding non - native interactions on the folding mechanism for the honeycutt - thirumalai @xmath0-barrel model @xcite were investigated in @xcite . at first glance
one would expect that adding frustration begins to slow the rate at the transition temperature , or at best has initially no effect .
what follows is a derivation of the somewhat counterintuitive result that in general the folding rate initially increases as ruggedness is increased from zero .
eventually of course the rate decreases drastically , so a plot of the folding rate _ vs. _ the amount of non - native heterogeneity should look like fig .
( [ rate ] ) . then the question of where real protein interactions reside on this plot may be addressed .
for some fast - folding proteins , it is possible that non - native noise in the system may actually assist folding .
consider first the thermodynamics of a protein obtained from a statistical analysis of a correlated landscape @xcite .
the energy , entropy , and free energy as functions of the fraction of native contacts @xmath1 , are given by in this paper , so temperatures have units of energy , and entropies are in units of @xmath2 . ]
@xmath3 these quantities are shown in figure ( [ s ] ) , and the parameters used in them are given in table i. @xmath4 in eq .
( [ eq : s])is the configurational entropy in the system _ vs. _ @xmath1 , @xmath5 is the extra internal energy in the native state ( the stability gap ) , and @xmath6 is the non - native variance which is a measure of the overall ruggedness of the energy landscape ( see eq . [ eqv ] ) .
the variance @xmath7 as @xmath8 . in obtaining the functional form of the non - native ruggedness
, it is assumed here that all the native contacts have roughly the same strength . , where @xmath9 is the native variance @xcite . if the set of native energies has variance but the distribution is fully specified , then the ruggedness terms are again proportional to @xmath6 @xcite . ]
the native energy is the number of native contacts @xmath10 times the mean native attraction energy @xmath11 ( @xmath12 ) . if @xmath13 is the number of interacting residues in the polymer chain and @xmath14 is the number of effective bonds per residue , = m = z n .
[ eqen ] the scale for the overall non - native ruggedness @xmath15 is given by ^2(q ) = m b^2 ( q ) , [ eqv ] where @xmath16 is the non - native packing density ( @xmath17 ) , @xmath18 is the intrinsic variance per non - native interaction , and @xmath10 is the total possible number of ( non - native ) interactions , i.e. the native state is assumed to be fully collapsed with the maximal number of contacts , and this is the maximal number of total contacts of any state . the density @xmath16 tends to increase upon folding ( see section [ sect : collapse ] ) , hence the ruggedness scale @xmath19 increases as well .
the strength @xmath20 of non - native interactions is taken to be weak : < < 1 , [ lesss ] therefore the ratio of folding transition temperature @xmath21 to thermodynamic glass temperature @xmath22 is large > > 1 , [ tftg ] i.e. the proteins we consider are strongly ( but not infinitely ) unfrustrated- we are perturbing away from the g model .
the configurational entropy @xmath4 has the property that entropy loss on folding is more rapid initially than in later stages .
we approximate this effect here by assuming the form s_c(q ) = s_o ( 1-q ) - ( q ) [ eqso ] where @xmath23 is the total conformational entropy in the unfolded ( @xmath24 ) state ( @xmath25 is the log number of conformational states per residue ) , and @xmath26 is a tent function : ( q ) = \ { ll 2 q & q<1/2 + 2 ( 1-q ) & q > 1/2 . .
[ eq : tent ] we ve let the barrier be at @xmath27 for simplicity of argument . at the transition temperature @xmath21 ,
the free energy of the folded and unfolded states are equal : f(0 ) & & f(1 ) + -s_o - & & .
[ eqtf ] using this relation in eq .
( [ eq : f ] ) gives .
| _ = ( q ) - . [ fattf ]
when @xmath28 the free energy at @xmath21 is the tent function ( see fig .
[ s ] ) , and so @xmath29 in eq .
( [ eq : tent ] ) is thus @xmath30 . then from eq .
( [ fattf ] ) the free energy barrier at @xmath21 is given by - ^ , [ eq : bar ] where @xmath31 is the change in non - native density between the barrier peak and unfolded state ( @xmath32 ) .
so long as @xmath33 , the barrier height decreases as non - native heterogeneity ( @xmath18 ) increases , as shown in fig .
( [ s ] ) .
we now show that this is nearly always the case , by considering the physics of collapse for our problem in question . in this section
we investigate the coupling of non - native density with the amount of native structure present in a protein , by showing that native topological constraints can induce a collapse transition on the non - native parts of the protein .
then the trend in eq .
( [ eq : bar ] ) of adding non - native ruggedness would be to lower the folding barrier .
collapse occurs below a temperature @xmath34 , defined as the temperature where the free energy of the coil and collapsed molten globule phases ( both at @xmath35 ) are equal : f_coil ( ) = f_mg ( ) . again using eq .
( [ eq : f ] ) , but now noting that the conformational entropies are different in the coil and globule phases , and that @xmath36 in the globule and @xmath37 in the coil phase , we have - s_coil = - s_mg - - m a .
note we have now allowed for a mean homopolymer attraction @xmath38 in general , for reasons which will become clear below . using the reduction in entropy for collapsed _ vs. _ coil chain statistics @xcite s_coil - s_mg = n - n ( /e ) =
n gives for the collapse temperature = ( 1 + ) . [ tc1 ] note from ( [ tc1 ] ) that when @xmath28 , @xmath39 , i.e. the collapse temperature is the mean homopolymer attraction times the number of contacts per residue , and when @xmath40 , @xmath41 , i.e. non - native heterogeneity can drive collapse , with the collapse temperature now scaling with the root number of contacts per residue times the width of interactions .
now we note that in our model ( g perturbed by weak heterogeneity , with @xmath40 ) collapse and folding will tend to occur together , with folding driving the collapse through native structural constraints .
so the total density increases from zero to one as @xmath42 , and the non - native density @xmath16 should increase as well since non - native polymer is more strongly constrained by larger native cores , see figure [ fig : corehalo ] .
the simplest approximation to capture this increase in density upon folding is to replace the mean homopolymer field @xmath38 by the native energy scale @xmath11 times the fraction of native bonds made @xmath1 : a(q ) = q .
[ eq : a ] this is the effective homopolymer field for the ensemble of states with fraction @xmath1 of native structure . using ( [ eq : a ] ) in ( [ tc1 ] ) and noting that the glass temperature = gives = ( + ) [ tc2 ] where @xmath25 and @xmath43 are the entropy per residue in the coil and globule state respectively .
note that in eq .
( [ tc2 ] ) @xmath44 as long as the term in parentheses is greater than one .
this gives a critical value @xmath45 where collapse occurs during folding , i.e. when @xmath46 , @xmath47 and when @xmath48 , @xmath49 .
this is sketched in fig ( [ eta]a ) below .
solving for @xmath45 gives & = & - ( ) ^2 = - [ eq : qc ] + ( q ) & & q s_o = q z [ eq : qc2 ] since , by construction of the problem , eqn .
( [ lesss ] ) holds , e.g. say @xmath50 is about @xmath51 .
then the second term in ( [ eq : qc ] ) is of order @xmath52 and can be neglected .
equation ( [ eq : qc ] ) says that the more chain entropy the polymer has ( the more flexible it is ) the _ sooner _ it collapses when folding at the transition temperature .
calorimetric measurements of the conformational entropy change per residue in unfolding to the coil state for say barnase give @xmath53 per residue @xcite .
this entropy also counts side chain conformational entropy , which is estimated to be about @xmath54 per residue @xcite , giving a net chain conformational entropy of about @xmath55 per residue in the coil state , and therefore @xmath56 .
for typical off - lattice simulations @xcite @xmath57 , therefore @xmath58 .
so collapse typically occurs before the barrier is reached ( see fig .
( [ eta])c ) . for lattice simulations @xmath59 , which is around @xmath60 . in any event
, @xmath61 will tend to be greater than @xmath62 as long as the system is large enough and bulk thermodynamics can be used ( however see caveats in appendix [ appa ] ) , see figure [ eta ] .
the values of non - native packing density obtained from simulations appear to be smaller than one , probably because of finite size and stiffness effects . applying bulk thermodynamics to the residual segments of non - native polymer may not be an accurate approximation in some cases .
more complete treatments of the coupling of density with native similarity can be made within the energy landscape framework @xcite .
it is fairly straightforward to write an approximate free energy as a function of both @xmath63 and @xmath1 and then minimize with respect to @xmath63 to obtain the density as a function of @xmath1 .
we have taken the simplest approach here to illustrate the coupling of collapse with the thermodynamics .
some cautionary notes are made in appendix a regarding a possible reversal of the trend on barrier height in small , stiff proteins , or proteins with a significant amount of generic self - attraction .
what about the rate ? the question is now whether the increase in prefactor is larger than the decrease in barrier height , as we add non - native heterogeneity .
since the ruggedness is weak , the kinetics are single exponential ( there is a single dominant folding barrier ) , and a kramers law holds for the rate : k ^-1(b ) ^-(b)/t , [ rate1 ] where the prefactor is proportional to the reciprocal of the reconfiguration time scale @xcite .
if we were to follow the argument for the dependency of the prefactor on ruggedness for an uncorrelated landscape @xcite , or for a correlated landscape at low temperature with activated dynamics @xcite , we would find that the ratio of rates .
| _ = ( ^ - ) [ eq : k1 ] where @xmath64 is a function of @xmath65 on the uncorrelated landscape , and on the correlated landscape is a function of both @xmath65 and structural entropic factors having to do with the density of states of given similarity to a trap .
so by inspection of eq .
( [ eq : k1 ] ) , in this low temperature regime the rate may go up or may go down with non - native interaction strength .
however an important result arising from energetic correlations in the landscape is the existence of a critical temperature @xmath66 where the dynamics becomes unactivated @xcite . above this temperature
the dynamics is similar to reconfigurations in a normal liquid rather than the hopping dynamics of trap escape in a supercooled liquid or glassy system @xcite , i.e. above @xmath66 , the prefactor @xmath67 remains nearly constant with ruggedness , since at these temperatures the rouse modes depend much more weakly on the ruggedness introduced .
the existence of such a temperature scale can be seen from the following simple argument @xcite .
we can think of escape from a trap as a mini - unfolding event : escape is driven by entropy and is opposed by the putative trap s low energy , say @xmath68 . then , as in unfolding , the escape barrier arises from a mismatch between entropy gains and energy losses as the system reconfigures out of the trap , so we can rewrite equation ( [ eq : f ] ) for the free energy relative to the state @xmath69 as f(q ) = q e_i - t s_c ( q ) - ( 1-q ) [ eq : trap ] where @xmath5 in ( [ eq : f ] ) is replaced by @xmath68 , @xmath1 in ( [ eq : f ] ) is replaced by @xmath70 , defined as the fraction of contacts shared with state @xmath69 , and density changes during untrapping are not particularly important since @xmath71 ( energetic trapping occurs only when the polymer is collapsed @xcite ) .
the transition to unactivated dynamics occurs when the states typically occupied at that temperature have zero escape barrier . setting @xmath68 in ( [ eq : trap ] ) to the thermal energy of states at temperature @xmath72 , ) at @xmath24 . for strata of states with @xmath73
the larger ruggedness scale @xmath74 increases @xmath66 and @xmath22 for that stratum of states . ]
e_i ( t ) = - [ etherm ] we note that @xmath66 occurs when the free energy profile is downhill away from the trap at @xmath75 , i.e when @xmath76 in our model , where the subscript @xmath77 indicates the high @xmath70 region of the piecewise free energy function ( [ eq : trap ] ) ( eq .
( [ eqso ] ) for the configurational entropy has a piecewise structure ) . using equations ( [ eq : tent ] ) , ( [ eq : trap ] ) , and ( [ etherm ] )
, this gives = ( 1 - ) ^-1/2 [ ta ] for the transition temperature to activated dynamics . from ( [ ta ] )
we see that @xmath78 by an amount which depends on the deviation from linear entropy loss over the total unconstrained entropy , i.e. by the entropic contribution to the barrier .
there is no energetic contribution since we have used @xmath70 as the order parameter and assumed pairwise interactions . ) . in mean - field theory
, @xmath29 is extensive and @xmath78 in the thermodynamic limit .
but in the capillarity theory the entropic deviation @xmath29 comes from surface entropy and should scale as @xmath79 ( or even with a smaller power if the interface is roughened ) .
one might argue that since @xmath80 , @xmath66 approaches @xmath22 as @xmath81 , but matching the theories in this fashion is incorrect since eq .
( [ eq : trap ] ) is not valid in the capillarity limit . in the capillarity theory a dynamical transition can only be seen by investigating where the intensive surface tension vanishes . ] for typical size proteins of @xmath82 , @xmath83 .
a plot of the escape time on a correlated landscape is given below ( see fig .
[ tau ] ) . in the regime
we are interested in , @xmath84 ( c.f .
( [ tftg ] ) ) , so it is also true that < < 1 and so the characteristic temperatures where folding occurs ( @xmath85 ) are way above the transition temperature for activated diffusion by construction of the problem ; see figure ( [ tau ] ) . expanding eq .
( [ rate1 ] ) around @xmath86 using eq .
( [ eq : bar ] ) gives the rate for weak non - native heterogeneity : 1 + m ^ ( ) , where @xmath87 and we ve used @xmath88 .
the increase in rate occurs until around @xmath89 , where @xmath90 is where @xmath91 ( about @xmath92 here ) , and @xmath93 is the finite - size fluctuation of @xmath90 due to temperature fluctuations which round the transition @xcite .
that is , = . |
| _ where eq .
( [ eq : s ] ) was used for the entropy at @xmath24 .
this gives a value for @xmath94 .
realistic values of @xmath95 for a typical protein can be obtained from the ratio of folding to glass temperature , given by /= + where @xmath96 @xcite .
commonly accepted values of @xmath97 for proteins are about @xmath98 .
using the values in table i for @xmath25 and @xmath14 gives .| _ ( 0.1 - 0.15 ) which is above the rate enhancement regime ( see fig . [ rate ] ) .
if the effects of non - native rate enhancement are observable , they will be seen possibly in only the fastest folding proteins . on the other hand
such an observation would support the existence of a dynamic glass transition in protein systems .
as non - native heterogeneity is increased from zero , the folding rate initially increases by a factor of about @xmath99 , then eventually drops rapidly towards zero ( see fig
. [ rate ] ) .
there is a regime near the g hamiltonian where the rate is relatively robust to changes in non - native interaction strength .
the density of non - native polymer must be greater at the barrier peak than in the unfolded state for rate enhancement to be observed .
why does the rate initially increase ?
the upshot is as follows .
first , it follows from energy landscape theory that if there is no change in density upon folding , then changes in ruggedness do not affect the barrier height for a well - designed protein ( c.f .
( [ eq : f ] ) ) , that is , effects on the barrier must come from the coupling of non - native density to the degree of nativeness . moreover , the strength of the ruggedness per residue increases with non - native density since there are simply more interactions , and non - native density tends to increase with nativeness , at least for weak non - native interaction strength .
then , since ruggedness lowers the free energy , the free energy at the barrier position is lowered more than the unfolded free energy .
so as non - native interaction strength is increased from zero , the barrier lowers and the rate increases if the effect on the prefactor is weaker .
but the prefactor is related to the reconfigurational diffusion time @xcite , and since a dynamic glass transition is expected in such systems , there will be a window for weak ruggedness within which the diffusion time is relatively unaffected as ruggedness is increased .
thus the rate initially increases , as shown in fig
. [ rate ] .
this phenomena provides a good example of how energy landscape theory can be applied to the physics of protein folding to reveal and explain a counter - intuitive result .
we thank jose onuchic and cecilia clementi for many enjoyable and insightful discussions , and peter wolynes specifically for helpful discussions on the collapse transition .
we are also grateful to hugh nymeyer for providing the off - lattice simulation data and constructive advice .
this work was supported by national science foundation ( nsf ) bio - informatics fellowship dbi9974199 and by nsf grant 0084797 .
the derivation leading to eq .
( [ eq : bar ] ) assumed mean - field theory could be applied , that the protein could could be treated as a bulk system , and that properties arising from chain connectivity would not alter the results arising from the energetics in the problem .
in particular we have assumed that the polymer persistence length or kuhn length @xmath100 is much less than the length of a typical piece of disordered protein @xmath101 near the barrier peak , so that a dangling piece of disordered polymer may interact with itself .
if the protein under study is particularly stiff and/or small , the return length of polymer fragments may be comparable to the length of the disordered pieces , reducing the number of non - native interactions near the barrier peak relative to the number in the unfolded state . then in eq .
( [ fattf ] for the @xmath103-mer lattice model @xmath104 ; these models are relatively stiff compared to their total length . for typical off - lattice models on the other hand , @xmath105
but they are considerably longer , e.g. for sh3 , @xmath106 .
if all the non - native polymer is in one strand , @xmath107 at the barrier peak
. then @xmath108 for the lattice model , and @xmath109 for the sh3 off - lattice model .
if the non - native polymer is distributed among a number of disordered strands that can dress the native core , roughly @xmath110 @xcite , then @xmath111 for the lattice model , and @xmath108 for the off - lattice model .
neither of these numbers are very small indicating that the collapse transitions are quite rounded , and the effect on folding rate will be mild if it exists .
in fact the discrepancy of @xmath112 between the off - lattice and on - lattice models , although fairly small here , leads to different behavior of the non - native density @xmath16 , as shown in fig .
[ eta ] . in the lattice system @xmath113 , but in the off - lattice system @xmath114 .
hence we anticipate the rate - enhancement effects will be seen in off - lattice models , but probably not in at least the shorter on - lattice models @xcite . real proteins may tend to have some net homopolymer attraction inducing generic collapse .
this decreases the change in density upon folding and would further attenuate any rate enhancement effect present
. however at least some proteins are sufficiently stable that collapse and folding are concomitant @xcite .
moreover g models , for which collapse and folding are concomitant by construction , give reasonably accurate predictions of @xmath29-values and barrier heights @xcite .
collapse accompanies folding when the folding transition temperature @xmath21 given by eq .
( [ eqtf ] ) is comparable to the collapse temperature @xmath34 in eq .
( [ tc1 ] ) . for weak ruggedness , ( 1 - - ) .
so the effects of generic collapse are not important in the problem as long as @xmath115 , to the first approximation .
several additional features may affect the folding rate .
in finite - sized systems the unfolded state tends to have partial order .
moreover its position may drift , along with that of the transition state , as non - native variance is increased .
this modifies the barrier height .
additionally the density @xmath16 in figure [ eta ] is exact in the limit @xmath116 , but for nonzero @xmath20 , @xmath16 may begin to alter in structure . accounting for these effects will modify the folding rate , but should nt alter the general trend of figure [ rate ] .
fig . [ rate ] @xmath117 rate _ vs. _ non - native heterogeneity is split up into two regimes , one where it assists folding , the other where it hinders .
the rate plotted here is the folding rate at @xmath21 which is itself a function of @xmath20 ( c.f . eq .
( [ eqtf ] ) , however @xmath21 changes by only @xmath118 over the range of this plot .
( a ) schematic depicting the two regimes , ( b ) result of the theoretical model introduced in the text ( see equations [ eq : bar ] , [ rate1 ] , and figure [ tau ] ) for a system with parameters given in table i. initially the rate rises as @xmath119 , then strongly decreases for larger @xmath20 as non - native interactions slow conformational transitions .
realistic protein interactions are believed to have typical values of @xmath120 @xcite , which is above the rate enhancement regime .
the inset of ( b ) shows a semi - log plot of the same rate ; it can be seen that there is a regime where the rate is roughly constant as ruggedness is increased from zero , then a turnover where the rate drastically decreases .
[ s ] @xmath117 energy , entropy , and free energy _ vs. _ @xmath1 in the model , for the g model with @xmath121 ( solid line ) , and when ruggedness is introduced , when @xmath122 ( dashed line ) .
we took @xmath123 ( see eq .
[ eqv ] ) , where the folding rate is maximal ( see fig . [ rate ] ) .
parameters used in the model are given in table [ table1 ] .
a bilinear approximation for the configurational entropy is used here , giving a tent functional form for the g free energy at the transition temperature @xmath21 .
the non - native density function @xmath16 used here is a fit to the off - lattice data in fig .
the folding free energy barrier at @xmath21 is _ lowered _ by non - native heterogeneity , because the energy is lowered twice as much as the entropy ( see eq . 1 ) .
[ fig : corehalo ] @xmath117 as folding progresses , the non - native polymer halo surrounding the native core ( central shaded globule ) has more topological constraints placed upon it .
therefore the non - native packing density , given by the total number of non - native contacts divided by the characteristic volume of non - native polymer ( open spheres ) , tends to increase .
[ eta ] @xmath117 free energy profiles @xmath124 and non - native polymer density @xmath16 in the model for ( a ) the simple bulk - mean - field model used in the derivation , ( b ) a 27-mer lattice model , ( c ) an off - lattice model for the @xmath125 residue fragment corresponding to the @xmath126-spectrin _ sh3 _ domain ( pdb code 1bk2 ) .
( a ) collapse occurs at @xmath45 before the barrier peak at @xmath60 .
the transition is rounded for typical sized proteins , as in ( b ) and ( c ) . in ( b ) , the non - native density is overall fairly small , and is comparable in the unfolded and transition state , @xmath127 .
thus eq .
( [ fattf ] ) gives a barrier height roughly independent of @xmath20 at least for small @xmath20 . in ( c ) on the other hand , the non - native density rises to values larger in overall magnitude , and is monotonically increasing until the barrier peak : @xmath128 and @xmath129 . for the parameter values in table
i , eq . ( [ fattf ] ) then gives a barrier height decreasing with @xmath20 as @xmath130 .
the drop - off at high @xmath1 in ( b ) and ( c ) is most probably due to stiffness effects on the small pieces of residual non - native polymer in this regime .
[ tau ] @xmath117 log of the reconfiguration time _ vs. _ reciprocal temperature in units of @xmath22 , for a system of size @xmath131 , adapted from reference @xcite .
this is used in equation ( [ rate1 ] ) to produce the rate curve in figure [ rate ] . at temperatures above @xmath66
the time to reconfigure is @xmath132 , below @xmath66 the time increases exponentially as @xmath133 with @xmath134 for @xmath135 .
the width of the transition @xmath136 as @xmath137 and the value of @xmath138 as @xmath137 for the mean field correlated landscape .
when @xmath139 for a finite system , the deepest trap tends to dominate the kinetics , and the relaxation rate turns over to an arrhenius law with slope corresponding to the barrier height for escape from that trap ( shown schematically by the dashed line ) . c. clementi , h. nymeyer , and j. n. onuchic .
topological and energetic factors : what determines the structural details of the transition state ensemble and en - route intermediates for protein folding ?
an investigation for small globular proteins .
, 298:937953 , 2000 . | perturbing a g model towards a realistic protein hamiltonian by adding non - native interactions , we find that the folding rate is in general enhanced as ruggedness is initially increased , as long as the protein is sufficiently large and flexible .
eventually the rate drops rapidly towards zero when ruggedness significantly slows conformational transitions .
energy landscape arguments for thermodynamics and kinetics are coupled with a treatment of non - native collapse to elucidate this effect .
= -1.6 cm = -0.05 in = 0.4 cm = 6.6 in 0 _ * abstract * |
dao white dwarfs , the prototype of which is hz34 @xcite , are a group of white dwarfs that are observed to have both hydrogen and helium lines in their optical spectra , in contrast to the more common das , which exhibit only hydrogen absorption .
hence , the presence of detectable he must arise from the existence of a thin overlying h envelope or there must be a mixing process , dredging up he from the deeper layers of the stellar photosphere .
radiative forces appear to be too low to support sufficient quantities of helium to be able to produce the observed lines @xcite .
a thin h envelope might arise through float up of residual h in a he - rich do atmosphere as objects transfer between the helium and hydrogen cooling channels @xcite .
however , the discovery by @xcite that the heii line in the optical spectra of one dao was better reproduced by a homogeneous rather than a layered atmospheric model was contrary to this view .
@xcite performed a similar comparison on a sample of 14 daos , but found only one object for which a stratified model was preferred .
of the remaining objects , most were found to be comparatively hot and low gravity , implying they have low mass .
therefore , the progenitors of some of these daos are unlikely to have had sufficient mass to ascend the asymptotic giant branch ( agb ) , and instead they may have evolved from extended horizonal branch stars .
it was suggested that , in these stars , weak mass loss may be occurring , a process that might be able to support the observed quantities of helium in the line forming regions of the daos @xcite .
three objects ( re1016 - 053 , pg1413 + 015 and re2013 + 400 ) were in close binary systems and have m dwarf ( dm ) companions .
these have ` normal ' temperatures and gravities , yet still have detectable helium lines in their optical spectra .
this may be because mass is lost as the progenitor star passes through the common envelope phase , leading to the star being hydrogen poor , and allowing a weak process , such as mass loss , to mix helium into the line forming region of the white dwarf .
alternatively , these daos might be accreting from the wind of their companions - re0720 - 318 , another dao with a dm companion , has been shown to have changing helium abundance over time , which might be due to episodic accretion @xcite .
in addition , @xcite identified likely spatial non - uniformities in the surface abundance of helium , which are consistent with models of accretion .
finally , for one star in the @xcite sample ( pg1210 + 533 ) , the heii line profile was not reproduced satisfactorily by either chemically homogeneous or stratified models .
in addition , its helium line strengths have been observed to change over a time period of @xmath015 years @xcite .
this is not a high temperature , low gravity object @xcite , and has no infrared excess that might indicate a companion star , so it does not appear to fit in with any of the other daos .
therefore , dao white dwarfs form a relatively heterogeneous group . for some ,
the existence of cool , non - degenerate companions may play a significant role in the presence of photospheric he , either from prior common envelope evolution or from ongoing mass transfer .
we investigate a sample of 16 dao white dwarfs to determine how many exhibit evidence of short term radial velocity variations and thus may be members of close binary systems , using _ far - ultraviolet spectroscopic explorer _
( _ fuse _ ) data .
_ fuse _ spectra are built up from a number of exposures that last up to 90 minutes , which need to be combined to achieve the desired signal - to - noise of the proposer . although the signal - to - noise of these exposures is far worse than in the final combined spectrum , strong interstellar and photospheric absorption features can still be identified .
this investigation exploits this fact to search for radial velocity shifts in photospheric lines relative to the stationary interstellar medium lines , which might indicate the presence of short period binary systems .
far - ultraviolet ( far - uv ) data for all the objects were recorded by the _ _ fuse__spectrographs .
table [ fuseobs ] lists the observations that were used , which were downloaded by us from the multimission archive ( http://archive.stsci.edu/mast.html ) , hosted by the space telescope science institute .
overviews of the mission and in - orbit performance have been published by @xcite and @xcite respectively ; below is a brief description of the instrument . [
cols="<,<,^,^,^,^,^",options="header " , ] of the daos with comparitively high mass , evidence for binarity was found for re0720 - 318 and re2013 + 400 ( both of which are dao+dm binaries ) , but not for pg1210 + 533 and lb2 .
only two exposures were obtained for pg1210 + 533 , while lb2 has 5 , and it might be possible that radial velocity shifts were missed because the separation in time of exposures was insufficient , because the system was viewed at an unfavourable phase , or if the radial velocity measurement technique were insufficiently sensitive .
in addition , if a system is at high inclination , no radial velocity variations would be detectable
. this may indeed be the case for lb2 , which was identified by @xcite as having an infrared excess .
it should also be noted that for this object the mass determined from the lyman line results is lower than that obtained from the balmer line results and using the lower mass would place lb2 in the lower mass group . as shown in figure [ probs ] , since there are fewer exposures for pg1210 + 533 , the method is less sensitive than for lb2 .
however , it is still 80% likely that radial velocity shifts would be detected by this experiment if it were in a binary system with period less than almost a day , unless its period happened to match the separation of exposures in time , or a multiple of that separation .
this indicates that this method is quite a sensitive way of searching for white dwarfs in binary systems , particularly if a number of exposures have been made , and hence it is unlikely , although not impossible , that pg1210 + 533 is in a short period binary system , hence the explanation for helium in the spectrum of pg1210 + 533 is still elusive . in the observations of the remaining 12 white dwarfs , evidence of binarity
was found for only ton320 , pg0834 + 500 , hs1136 + 6646 and feige55 , which are all known or suspected binary stars .
although low mass white dwarfs can be formed within in binary systems , these results suggest that not all of these have companions . therefore , for most of the high temperature , low gravity , low mass objects , evolution within a binary system , or accretion from a companion ,
can not be used to explain their low mass or the presence of helium absorption lines in their optical spectra , and therefore evolution as single objects from the extended horizontal branch is still the most likely explanation for their existence .
_ fuse _ data have been used to measure the radial velocity variations in 16 dao white dwarfs , from the position of the photospheric heavy element absorption lines in their spectra .
the technique was found to be successful in detecting radial velocity variability in all the known binaries , and even with only 2 exposures there is better than a 80% chance of observing velocity changes if the period of the system is less than a day , apart from where the period is a multiple of the temporal separation of exposures .
radial velocity variations were only detected for those objects already known or suspected to be binaries , although it is possible that some of the others possess binary companions that are too highly inclined , or which were observed at unfavourable phases or have orbital periods too long with respect to the time span covered by our observations to be detected .
of the low mass daos , significant changes in radial velocity were observed for only four of the twelve objects , suggesting that the majority of this type of dao do not exist in binary systems and may therefore have evolved as single stars from the extended horizontal branch .
no evidence of binarity was found for two of the four higher mass white dwarfs in the sample .
one of these objects , lb2 , has an infrared excess and hence may in fact have a companion .
the presence of helium in the spectrum of the other , pg1210 + 533 , is still unexplained .
based on observations made with the nasa - cnes - csa far ultraviolet spectroscopic explorer .
_ fuse _ is operated for nasa by the johns hopkins university under nasa contract nas5 - 32985 .
sag , mab , mrb and pdd were supported by pparc , uk ; mrb acknowledges the support of a pparc advanced fellowship .
jbh wishes to acknowledge support from nasa grants nag5 - 10700 and nag5 - 13213 .
barstow m. a. , et al . , 1995 , mnras , 272 , 531 barstow m. a. , odonoghue d. , kilkenny d. , burleigh m. r. , fleming t. a. , 1995 , mnras , 273 , 711 bergeron p. , wesemael f. , beauchamp a. , wood m. a. , lamontagne r. , fontaine g. , liebert j. , 1994 , apj , 432 , 305 bloecker t. , 1995 , a&a , 297 , 727 dobbie p. d. , barstow m. a. , burleigh m. r. , hubeny i. , 1999 , a&a , 346 , 163 driebe t. , schoenberner d. , bloecker t. , herwig f. , 1998 , a&a , 339 , 123 finley d. s. , koester d. , basri g. , 1997 , apj , 488 , 375 fontaine g. , wesemael f. , 1987 , iau colloq .
95 : second conference on faint blue stars , 319 - 326 good s. a. , barstow m. a. , holberg j. b. , sing d. k. , burleigh m. r. , dobbie p. d. , 2004 , mnras , 355 , 1031 heber u. , dreizler s. , hagen h .- j . , 1996 , a&a , 311 ,
l17 hillwig t. c. , honeycutt r. k. , robertson j. w. , 2000 , aj , 120 , 1113 holberg , j.b . and magargle , 2005 , in proceedings of the nato advanced workshop on white dwarfs 2004 holberg j. b. , saffer r. a. , tweedy r. w. , barstow m. a. , 1995 , apj , 452 , l133 koester d. , weidemann v. , schulz h. , 1979 , a&a , 76 , 262 kruk j. w. , chayer p. , dupuis j. , 2003 , in proceedings of the nato avanced workshop on white dwarfs 2002 , de martino , d. , silvotti , r. , solheim , j .- e . and kalytis , r. ( eds ) , kluwer moos h. w. et al . ,
2000 , apj , 538 , l1 morton d. c. , 1991 , apjs , 77 , 119 napiwotzki r. , schberner d. , 1993 , in : barstow m. ( ed ) 8th european workshop on white dwarfs , kluwer academic publishers , p.99 pounds k. a. , et al . , 1993 , mnras , 260 , 77 saffer r. a. , livio m. , yungelson l. r. , 1998 , apj , 502 , 394 sahnow d. j. et al . ,
2000 , apj , 538 , l7 sing , d. k. , holberg , j. b. , burleigh , m. r. , good , s. a. , barstow , m. b. , oswalt , t. d. , howell , s. b. , brinkworth , c. s. , rudkin , m. , johnston , k. , rafferty , s. , 2004 , aj , 127 , 2936 unglaub k. , bues i. , 1998 , a&a , 338 , 75 unglaub k. , bues i. , 2000 , a&a , 359 , 1042 vennes s. , pelletier c. , fontaine g. , wesemael f. , 1988 , apj , 331 , 876 vennes s. , thorstensen j. r. , 1994 , apj , 433 , l29 vennes s. , thorstensen j. r. , polomski e. f. , 1999 , apj , 523 , 386 wesemael f. , greenstein j. l. , liebert j. , lamontagne r. , fontaine g. , bergeron p. , glaspey j. w. , 1993 , pasp , 105 , 761 | we report on a search for evidence of binarity in _ far - ultraviolet spectroscopic explorer _ ( _ fuse _ ) observations of dao white dwarfs .
spectra recorded by _ fuse _ are built up from a number of separate exposures .
observation of changes in the position of photospheric heavy element absorption lines between exposures , with respect to the stationary interstellar medium lines , would reveal radial velocity changes - evidence of the presence of a binary system .
this technique is successful in picking out all the white dwarfs already known to be binaries , which comprise 5 out of the sample of 16 , but significant radial velocity shifts were found for only one additional star , ton320 .
this object is also known to have an infrared excess @xcite .
daos can be separated broadly into low or normal mass objects .
low mass white dwarfs can be formed as a result of binary evolution , but it has been suggested that the lower mass daos evolve as single stars from the extended horizontal branch @xcite , and we find no evidence of binarity for 8 out of the 12 white dwarfs with relatively low mass . the existence of higher mass daos can also be explained if they are within binary systems , but of the four higher mass stars in the sample studied , pg1210 + 533 and lb2 do not exhibit significant radial velocity shifts , although there were only two exposures for the former object and the latter has an infrared excess @xcite .
[ firstpage ] stars : atmospheres - white dwarfs - ultraviolet : stars . |
a common misconception outside of the gravitational - wave research community is that the primary purpose of observatories such as ligo is to detect directly gravitational waves .
although the first unambiguous direct detection will certainly be a celebrated event , the real excitement will come when gravitational - wave detection can be used as an observational tool for astronomy . because the processes which drive gravitational - wave emission are fundamentally different from processes that radiate electromagnetically
, gravitational - wave astronomy will provide a view of the universe that is rather different from our `` usual '' views . because they arise from fundamentally different physical processes , the information carried by gravitational radiation is `` orthogonal '' to
that carried by electromagnetic radiation .
consider the following differences : * electromagnetic waves are oscillations of electric and magnetic fields that propagate through spacetime .
gravitational waves are oscillations in spacetime itself . *
astrophysical electromagnetic radiation typically arises from the incoherent superposition of emissions from individual electrons , atoms , and molecules .
they often provide direct information about the thermodynamic state of a system or environment .
gravitational waves are coherent superpositions of radiation that arise from the bulk dynamics of a dense source of mass - energy ( matter or highly curved spacetime ) .
they provide direct information about the system s dynamics . *
the wavelength of electromagnetic radiation is typically smaller than the radiating system .
they can thus be used to form an image of that system ; any good public lecture on astronomy includes a number of `` pretty pictures '' . by contrast , the wavelength of gravitational radiation is typically of order or larger than the size of the radiating source .
such waves _ can not _ be used to image the source ; instead , the two gravitational - wave polarizations are akin to sound , carrying a stereophonic description of the source s dynamics .
indeed , it is becoming common for workers in gravitational radiation to play audio encodings of expected gravitational - wave events , and the notion of `` pretty sounds '' ( or at least `` interesting sounds '' ) may become widely accepted as gravitational - wave astronomy matures . * with a few exceptions ,
electromagnetic astronomy is based upon deep imaging of narrow , small - angle fields of view : observers obtain a large amount of information about sources on a small piece of the sky .
gravitational - wave astronomy will be an all - sky affair : observatories such as ligo and lisa have nearly @xmath0 steradian sensitivity to events over the sky .
a consequence of this is that their ability to localize a source s sky position is not good by usual astronomical standards ; but , it means that any source on the sky will be detectable , not just sources towards which the instrument is `` pointed '' .
the contrast between the all - sky sensitivity , relatively poor angular resolution of gravitational - wave observatories , and the pointed , high angular resolution of telescopes is very similar to the angular resolution contrast of hearing and sight , strengthening the useful analogy of gravitational waves with sound . * electromagnetic waves
interact strongly with matter ; gravitational - waves do not .
this is both blessing and curse : it means that gravitational waves propagate from emission to earth with essentially zero absorption , making it possible to probe astrophysics that is hidden or dark ( for example , the coalescence and merger of black holes ; the collapse of a stellar core ; the dynamics of the early universe very soon after the big bang ) .
it also means that the waves interact very weakly with gravitational - wave detectors , requiring enormous experimental effort for assured detection . *
the directly observable gravitational waveform @xmath1 is a quantity that falls off as @xmath2 .
most electromagnetic observables are some kind of energy flux , and so fall off with distance with a @xmath3 law .
because of this , relatively small improvements in the sensitivity of gravitational - wave detectors have an enormous impact on their ability to make measurements : doubling a detector s sensitivity doubles the distance to which sources can be detected , increasing the volume of the universe in which sources are measurable by a factor of eight .
every factor of two improvement in the sensitivity of a gravitational - wave observatory increases the number of observable sources by nearly an order of magnitude .
these differences motivate the efforts to inaugurate gravitational - wave astronomy as a vital field . in this article
, we will describe current efforts to develop ground - based detectors ( sec .
, focusing in particular on the american ligo project ) as well as the joint american - european space - based lisa project ( sec .
[ sec : spacebased ] ) .
we also discuss in some detail the science that can be expected from gravitational - wave astronomy ( sec .
[ sec : sciencereach ] ) .
first , we briefly discuss the properties of gravitational waves and their likely strength , and sketch how a gravitational - wave detector is able to measure their tiny effects .
gravitational radiation is a natural consequence of general relativity ( indeed , of any causal theory of gravitation ; see @xcite for a fascinating discussion of early work by laplace on gravitation with a finite speed of propagation ) .
it was first described more - or - less correctly by albert einstein in 1918 @xcite .
( in an earlier work @xcite , einstein predicted gravitational radiation , though with a rather incorrect description . )
general relativity describes the radiation as a tensor perturbation to the metric of spacetime ( so that it can be thought of as a spin-2 boson ) , propagating at the speed of light , and with two independent polarizations . as electromagnetic radiation
is generated by the acceleration of charges , gravitational radiation arises from the acceleration of masses . in particular ,
electromagnetic waves are created ( at lowest order ) by the time changing charge dipole moment , and are thus dipole waves .
monopole em radiation would violate charge conservation . at lowest order , gravitational waves come from the time changing quadrupolar distribution of mass and energy ; monopole gravitational waves would violate mass - energy conservation , and dipole waves violate momentum conservation .
gravitational waves act tidally , stretching and squeezing any object that they pass through . because the waves arise from quadrupolar oscillations , they are themselves quadrupolar in character , squeezing along one axis while stretching along the other .
when the size of the object that the wave acts upon is small compared to the wavelength ( as is the case for ligo ) , forces that arise from the two gw polarizations act as in fig .
[ forcelines ] .
the polarizations are named `` @xmath4 '' ( plus ) and `` @xmath5 '' ( cross ) because of the orientation of the axes associated with their force lines .
interferometric gravitational - wave detectors measure this tidal field by measuring their action upon a widely - separated set of test masses . in ground - based interferometers , these masses are arranged as in fig .. the space - based detector lisa arranges its test masses in a large equilateral triangle that orbits the sun , illustrated in fig .
[ fig : lisa_orbit ] . on the ground
, each mass is suspended with a sophisticated pendular isolation system to eliminate the effect of local ground noise . above the resonant frequency of the pendulum ( typically of order @xmath6 ) , the mass moves freely .
( in space , the masses are actually free floating . ) in the absence of a gravitational wave , the sides @xmath7 and @xmath8 shown in fig .
[ fig : interferometer ] are about the same length @xmath9 .
suppose the interferometer in fig .
[ fig : interferometer ] is arranged such that its arms lie along the @xmath10 and @xmath11 axes of fig .. suppose further that a wave impinges on the detector down the @xmath12 axis , and the axes of the @xmath4 polarization are aligned with the detector .
the tidal force of this wave will stretch one arm while squeezing the other ; each arm oscillates between stretch and squeeze as the wave itself oscillates .
the wave is thus detectable by measuring the separation between the test masses in each arm and watching for this oscillation . in particular , since one arm is always stretched while the other is squeezed , we can monitor the difference in length of the two arms : @xmath13 for the case discussed above , this change in length turns out to be the length of the arm times the @xmath4 polarization amplitude : @xmath14 the gravitational wave acts as a strain in the detector ; @xmath1 is often referred to as the `` wave strain '' .
note that it is a dimensionless quantity .
aficionados of general relativity can easily derive eq.([eq : h_simple ] ) by applying the equation of geodesic deviation to the separation of the test masses and using a gravitational - wave tensor on a flat background spacetime to develop the curvature tensor ; see ref .
@xcite , sec .
9.2.2 for details .
we obviously do not expect astrophysical gravitational - wave sources to align themselves in as convenient a manner as described above .
generally , both polarizations of the wave influence the test masses : @xmath15 the antenna response functions @xmath16 and @xmath17 weight the two polarizations in a quadrupolar manner as a function of a source s position and orientation relative to the detector ; see @xcite , eqs .
( 104a , b ) and associated text . in ground - based interferometers ,
the test masses at the ends of each arm are made of a highly transparent material ( fused silica in present designs ; perhaps sapphire in future upgrades ) .
the mirrors at the far end of each arm have amplitude reflectivities approaching unity .
the mirrors at the corner are less reflective since they must couple the light into the fabry - perot cavity arms .
the corner mirrors multilayer dielectric coatings have power reflectivities @xmath18 .
a very stable laser beam is divided at the beamsplitter , directing light into the two arm cavities .
if the finesse of the cavity is @xmath19 and the amplitude reflectivity of the corner mirrors is @xmath20 , then each photon makes on average @xmath21 bounces .
the light from the two arms then recombines at the beamsplitter .
the mirrors are positioned so that , in the absence of a gravitational wave , all of the light goes back towards the laser and the photodiode reads no signal . if a signal is present , the relative phase @xmath22 of the two beams changes by an amount proportional to @xmath1 , changing the light s interference pattern . without any intervention ,
this would cause light to leak into the photodiode . in principle , the wave strain @xmath1 could be read from the photointensity of this light . in practice , a system of servo loops controls the system such that destructive interference is guaranteed the photodiode is kept dark , and so is called the `` dark port '' .
the wave strain @xmath1 is then encoded in the servo signals used to keep the dark port dark .
the wave strain must fall off with distance as a @xmath2 law to conserve the total energy flowing through large spheres .
we have already argued that the lowest order contribution to the waves is due to the changing quadrupole moment of the source .
to order of magnitude , this moment is given by @xmath23 .
by dimensional analysis , we then know that the wave strain must have the form @xmath24 the second time derivative of the quadrupole moment is given approximately by @xmath25 ; @xmath26 is the source s internal velocity , and @xmath27 is the nonspherical part of its internal kinetic energy .
strong sources of gravitational radiation are sources that have strong non - spherical dynamics for example , compact binaries ( containing white dwarfs , neutron stars , and black holes ) , mass motions in neutron stars and collapsing stellar cores , the dynamics of the early universe . in order to have an interesting rate of observable events
, we need to be sensitive to sources at rather large distances .
for example , when binary neutron stars coalesce , we need to reach out to several hundred mpc ( _ i.e. _ , a substantial fraction of @xmath28 light years ) @xcite .
in such a case @xmath29 . plugging these numbers into eq .
( [ eq : h_ordermag ] ) yields the estimate @xmath30 this sets the sensitivity required to measure gravitational waves . combining this scale with eq .
( [ eq : h_def ] ) says that for every kilometer of baseline @xmath9 , we need to be able to measure a distance shift @xmath31 of better than @xmath32 centimeters .
the prospect of achieving such a stringent displacement sensitivity often strikes people as insane .
how can light whose wavelength @xmath33 is @xmath34 times larger than the typical displacement be used to measure that displacement ?
for that matter , how is it possible that thermal motions do not wash out such a tiny effect ? that such measurement is possible with laser interferometry was analyzed thoroughly and published by rainer weiss in 1972 @xcite .
( it should be noted that the possibility of detecting gravitational waves with laser interferometers has an even longer history , reaching back to pirani in 1956 , and has been independently invented by several workers : gertsenshtein and pustovoit in 1962 , weber in the 1960s , and weiss c. 1970 .
see sec .
9.5.3 of ref .
@xcite for further discussion and references . )
examine first how the 1 micron laser can measure a @xmath32 cm effect . as mentioned above , the light bounces roughly 100 times before leaving the arm cavity ( corresponding to about half a cycle of a 100 hz gravitational wave ) .
the light s acquired phase shift during those 100 round trips is @xmath35 this phase shift can be measured provided that the photon shot noise at the photodiode , @xmath36 , is less than @xmath37 .
@xmath38 is the number of photons accumulated over the measurement ; @xmath39 is the magnitude of phase fluctuation in a coherent state , appropriate for describing a laser .
we therefore must accumulate @xmath40 photons over the roughly @xmath41 second measurement , which translates to a laser power of about 100 watts .
in fact , as was pointed out by ron drever @xcite , one can use a much less powerful laser : even in the presence of a gravitational wave , only a tiny portion of the light that comes out of the interferometer s arms goes to the photodiode .
the vast majority of the laser power is sent back to the laser .
an appropriately placed mirror bounces this light back into the arms , _ recycling _ the laser light .
the recycling mirror is shown in fig . [ fig : interferometer ] , labeled `` r '' . with it , a laser of @xmath42 watts drives several hundred watts to circulate in the `` recycling cavity '' ( the optical cavity between the recycling mirror and the arms ) , and @xmath42 kilowatts to circulate in the arms .
thermal excitations are overcome by averaging over many many vibrations .
for example , the atoms on the surface of the interferometers test mass mirrors oscillate with an amplitude @xmath43 at room temperature @xmath44 , with @xmath45 the atomic mass , and with a vibrational frequency @xmath46 .
this amplitude is huge relative to the effect of the gravitational wave how can we possibly hope to measure the wave ?
the answer is that atomic vibrations are random and incoherent .
the @xmath47 cm wide laser beam averages over about @xmath48 atoms and at least @xmath49 vibrations in a typical measurement .
atomic vibrations are irrelevant compared to the coherent effect of a gravitational wave . other thermal vibrations , however ,
are not irrelevant and in fact dominate the noise spectrum of ligo in certain frequency bands .
for example , the test masses normal modes are thermally excited .
the typical frequency of these modes is @xmath50 , and they have mass @xmath51 , so @xmath52 .
this , again , is much larger than the effect we wish to observe .
however , the modes are very high frequency , and so can be averaged away provided the test mass is made from material with a very high quality factor @xmath53 .
understanding the physical nature of noise in gravitational - wave detectors is an active field of current research ; see refs . and references therein for a glimpse of recent work . in all cases ,
the fundamental fact to keep in mind is that a gravitational wave acts _ coherently _ , whereas noise acts _ incoherently _ , and thus can be beaten provided one is able to average away the incoherent noise sources .
the first generation of long baseline , kilometer - scale interferometric gravitational - wave detectors are being constructed and commissioned at several sites around the world . briefly , the major ground - based interferometric gravitational - wave projects are as follows : * * ligo*. three ligo interferometers are in the commissioning phase : two in hanford , washington ( with 2 and 4 km arms , sharing the vacuum system ) , and one in livingston , louisiana ( 4 km arms ) . an aerial view of the hanford site is included in fig . [
fig : ligo_optical ] .
the ligo detectors are designed to operate in the power recycled michelson configuration with arms acting as fabry - perot cavities . the large distance between sites ( about @xmath54 km ) and
differing arm lengths are designed to support coincidence analysis .
much current research and development is focused on advanced ligo detector design .
the new generation of detectors will provide a broader frequency band and a @xmath55 10-fold increase in range for inspiral sources via the lowered noise floor . * * virgo*. virgo , the italian / french long baseline gravitational - wave detector , is under construction near pisa , italy @xcite .
it has 3 km arms and advanced passive seismic isolation systems . in most respects , virgo is very similar to ligo ; a major difference is that it should achieve better low frequency sensitivity from the beginning due to its advanced seismic isolation system .
virgo will very usefully complement the ligo detectors , strengthening coincidence analysis and making source position determination possible . * * geo600*. geo600 is a 600 meter interferometer being built by a british - german collaboration in the vicinity of hannover , germany @xcite .
it will use advanced interferometry and advanced low noise multiple pendulum suspensions , serving as a testbed for advanced detector technology and allowing it to achieve sensitivity comparable to the multi - kilometer instruments . * * tama300*. the tama detector near tokyo , japan can already claim significant observation time with more than 1000 hours of operation @xcite .
it has achieved a peak strain sensitivity of @xmath56 @xcite at frequencies near 1000 hz .
tama has 300 meter arms and is operated in the recombined michelson configuration with fabry - perot arms .
a much improved 3 km detector is currently under design @xcite . * * aciga*. the australian consortium for interferometric gravitational astronomy plans to build an observatory near perth , australia @xcite .
they are presently engaged in the construction of an 80 meter research interferometer , which can be extended to kilometer scale .
they are studying advanced detection methods and technologies which could lead to much decreased noise floors in advanced ground based interferometers .
since interferometric gravitational - wave detectors are sensitive to nearly all directions , it is nearly impossible to deduce pointing information from the signal of a single detector . to get accurate information about the source direction
it is necessary to make use of the detection time difference among detectors .
these detectors must be widely spaced and not collinear . at the very minimum ,
three sites are needed for acceptable pointing .
a fourth detector , widely removed from the plane of the other three , is particularly valuable for improving directional information .
thus a detector in australia would greatly add to the science output of the gravitational - wave observatory network , which is otherwise confined entirely to the northern hemisphere .
ligo is a good example to illustrate the design and operation principles of kilometer - scale interferometers ; we shall focus on it for the remainder of our discussion .
construction of both ligo facilities and vacuum systems was completed by early 2000 .
the 2 km interferometer at the hanford site was installed in mid 2000 ; the livingston site 4 km interferometer was finished by late 2000 .
installation of the hanford 4 km interferometer was delayed due to repairs needed following the february 28 , 2001 earthquake ; as this document is being written , installation is well underway .
both observatories are concentrating on detector commissioning and a series of `` engineering runs '' .
these runs have provided excellent real life experience for the participating members of the ligo scientific community , paving the way for the approaching `` science run '' scheduled to begin at the end of 2001 .
this will be followed by more frequent and longer science runs until 2006 , when major detector upgrades are scheduled . in parallel to the commissioning effort and engineering / science runs , the collaboration also focuses on the research and development of advanced ligo detectors , promising a wider detector band and greatly improved sensitivity .
the ligo detectors operate as power recycled michelson interferometers with fabry - perot arms ; see fig .
[ fig : ligo_optical ] .
very high duty cycle is needed for each interferometer in order to effectively use the full network for coincidence analysis , which is necessary to achieve a low false detection rate and have high confidence in observations .
the wide ( 3000 km ) separation between the ligo sites is large enough that the chance of environmentally induced coincidence events is negligible .
both sites are heavily equipped with environmental sensors that thoroughly cover a wide range of possible disturbances that could cause false detections .
for example , ligo monitors the local seismic background , electromagnetic fluctuations , acoustic noise , cosmic radiation , dust , vacuum status , weather , records power line transients , and uses ultra - sensitive magnetometers at several locations at each observatory .
we now briefly describe the operating principles of ligo optics , the major sources of noise that limit sensitivity , the data analysis system that will be used during operations , and plans for future upgrades .
the basic optical layout of the ligo detectors is shown in figure [ fig : ligo_optical ] .
ligo uses a nd : yag near infrared laser ( wavelength 1064 nm ) with peak power @xmath55 10w as the light source .
various electro - optical components and servo loops are used to stabilize both the frequency and power of the laser .
the light from the pre - stabilized laser passes through the input optics and is coupled into the 12 meter , triangular mode cleaner cavity . the mode cleaner passes only the tem@xmath57 mode , eliminating higher order modes . starting with the mode cleaner ,
every major optical component is within a large vacuum system , capable of reaching @xmath58 torr .
after conditioning by the mode cleaner , the light enters the interferometer .
all major optics in the interferometer are suspended on a single steel wire loop , mechanically isolated from the ground by vibration isolators and controlled by multiple servo loops .
the mirrors are made of fused silica with extremely high mechanical @xmath53 and are polished to within @xmath55 1 nanometer rms .
they have high homogeneity , low bulk loss and multi - layer coatings with less than 50 ppm scattering loss
. each mirror is actuated by four precision coils , each positioned around a permanent magnet glued to the back side of the test masses .
the coil assembly also features a sensitive shadow sensor for local control .
additional optical levers and wavefront sensors provide more precise sensing .
the laser beam is coupled into the arms by a beam splitter .
each arm is a fabry - perot optical cavity , increasing the effective length of the arm to magnify the phase shift ( proportional to cavity finesse ) caused by the wave .
the stored power within the interferometer is built up by the partially transmitting recycling mirror .
an operating interferometer tries to keep the dark port perfectly dark , holding the various optical components such that light coming out of the arms destructive interferes and no light goes to the photodetector .
when this is achieved , the interferometer is on _ resonance _ , with maximum power circulating in the arms .
several interconnected control loops are used to achieve and then maintain resonance .
an interferometer on resonance is usually described as _
locked_. keeping lock must be highly automated , requiring minimal operator interaction for high uptime .
the gravitational - wave signal is extracted from the servo signals used to maintain the lock and correct for the length difference between the arms .
gravitational - wave interferometers have their sensitivity limited by a number of noise sources .
we list here some of the most important and interesting fundamental noise sources ; many of these were recognized and had their magnitude estimated by rainer weiss in 1972 @xcite . * * seismic noise*. ambient or culturally induced seismic waves continuously pass under the test masses of the detector .
the natural motion of the surface peaks around 150 mhz ; this is called the `` microseismic peak '' .
cultural noises tend to be at higher frequencies , near several hz .
the test masses must be carefully isolated from the ground to effectively mitigate the seismic noise .
seismic noise will limit the low frequency sensitivity of first generation ground - based gravitational - wave detectors . * * thermal noise*. thermally excited vibrational modes of the test mass or the suspension system will couple into the resonances of the system . by improving the @xmath53 of the components one can decrease
the thermally induced noise between resonances .
* * shot noise*. the number of photons in the input laser beam fluctuates ; this surfaces as noise at the dark port .
the strain noise due to this effect is proportional to @xmath59 .
thus , increasing the recycling gain and/or increasing the laser power lowers the shot noise .
unfortunately , high power in the cavities induces other unwanted effects such as radiation pressure noise ( discussed in the next item ) or thermal lensing ( local deformation of the optical surfaces of the cavity ) .
the right choice of laser power and recycling gain is a compromise . * * radiation pressure noise*. fluctuating number of photons bouncing from the mirrors will introduce a fluctuating force on the mirror .
this effect is proportional to @xmath60 the inverse of the proportionality entering the shot noise .
one can not just increase the laser power without penalty . reducing shot noise and
radiation pressure noise in tandem is a topic of advanced detector r&d ; techniques for doing so require specially prepared laser states .
see @xcite and references therein for further discussion . *
* gravity gradient noise*. when seismic waves , atmospheric pressure fluctuations , cars , animals , tumbleweeds , etc .
, pass near a gravitational - wave detector , they act as density perturbations in the neighboring region .
this in turn can produce significant fluctuating gravitational forces on the interferometer s test masses @xcite .
this is expected to become the limiting factor at low frequencies for advanced ground - based detectors with high quality seismic isolation systems . * * laser intensity and frequency noise*.
the laser itself inevitably is somewhat noisy , with fluctuations in both intensity and frequency . this noise will not cancel out perfectly when the signal from the two arms destructively recombines , and so some noise can propagate into the signal on the dark port . *
* scattered light*. some laser light can scatter out of the main beam , and then be scattered back , coupling into the interferometer s signal .
this light will carry information about its scattering surface and be out of phase with the beam , and thus contaminate the desired signal .
a dense baffling system has been installed to greatly reduce this source of noise . * * residual gas*. any vacuum system contains some trace amount of gas that is extremely difficult to reduce ; in ligo , these traces ( mostly hydrogen ) are at roughly @xmath58 torr .
density fluctuations from these traces in the beam path will induce index of refraction fluctuations in the arms .
residual gas particles bouncing off the mirrors can also increase the displacement noise . * * beam jitter*. jitter in the optics will cause the beam position and angle to fluctuate slightly , which causes some noise at the dark port .
* * electric fields*. fluctuations in the electric field around the test masses can couple into the interferometric signal via interaction between the field and the induced or parasitic surface charge on the mirror surface . *
* magnetic fields*. fluctuations in the local magnetic field can affect the test masses when interacting with the actuator magnets bonded to the surface of the mirrors . * * cosmic showers*. high energy penetrating muons can be stopped by the test masses and induce a random transient due to the recoil .
the impact of the most important of these noise sources is shown in fig .
[ fig : ligo_sens ] . the initial ligo detectors will be limited by seismic noise at low frequencies ( @xmath61 ) , by thermal noise in the test mass suspensions at intermediate frequencies ( @xmath62 , and by shot noise at high frequencies ( @xmath63 ) .
the present detector noise is well above this target , particularly at low and intermediate frequencies .
as these noise sources are reduced , new challenges emerge .
eventually , any given instrument will be limited by a set of noise floors .
the curve labeled `` facility limit '' in fig .
sets the ultimate sensitivity that is likely to be attained even with all other noise sources controlled perfectly . at low frequencies , gravity gradient noise ( which will be extremely difficult , if not impossible , to mitigate ) will be the dominant source of noise ; beyond that , optical path fluctuations from residual gas limit the sensitivity .
as will be discussed further below , the gravitational - wave datastream will be filtered by a large number ( @xmath64 ) of model waveforms ( `` templates '' ) to search for sources such as coalescing compact binaries .
these templates must be applied in near real time . besides the gravitational - wave signal , ligo detectors have a large number of data channels for auxiliary data sources like the environmental monitors discussed above .
a significant portion of these signals must be digitized and recorded .
ligo thus creates a huge amount of data ( @xmath42 mb per second , gathered around the clock , equivalent to about 0.25 terabytes per day per detector ) which must be stored and managed in near real time . to this end
, ligo has developed its own , state - of - the - art data management system called ldas ( ligo data analysis system ) .
ldas is a distributed computing environment , mixing remote process control on servers with message passing on distributed clusters .
it provides a framework for conducting scientific studies of ligo data , allowing data access and conditioning , signal reconstruction , coincidence analysis , database management , data caching and archiving .
large beowulf - type clusters will be used for analysis at ligo s sites and at the campuses . the digitized data ( at rates from 16 to 16384 samples per second )
are organized and stored in `` frames '' , an international standard format for gravitational - wave observatories .
the data are buffered on local disks and archived on tape in the observatories .
the data include accurate ( sub microsecond ) timestamps to facilitate coincidence analyses between various different gravitational - wave detectors , and also with other astrophysics experiments , such as neutrino or gamma - ray detectors .
parallel to ldas , the data is digested by the global diagnostic system , which is responsible for the real time monitoring of environmental problems and the interferometer s state of state .
all important local or externally generated events , from lock losses to gamma ray bursts , are recorded in the ldas metadata database tables .
these records are available to analysis processes running under ldas .
a major focus of research within the gravitational - wave experimental community right now is on developing technologies for improving the sensitivity of ligo and other ground - based detectors .
the first stage detectors being commissioned right now are somewhat conservatively designed , ensuring that they can be operated without needing to develop too much new technology .
the price for this conservative design is limited astrophysical reach : their sensitivity is such that detection of sources is plausible , but not necessarily probable , based on our current understanding of sources . to broaden the astrophysical reach of these instruments ,
major upgrades are planned for roughly the year 2006 .
these upgrades will push the lower frequency `` wall '' to lower frequencies , and push the noise level down by a factor @xmath42 across the band .
this will increase the distance to which sources can be detected by a factor of @xmath65 , and the volume of the universe which ligo samples by a factor of @xmath66 .
this will dramatically boost the rate at which events are measured .
discussion of how these plans will be implemented is given in ref .. major changes will include a redesigned seismic isolation system ( pushing the wall down to about 10 hz ) , a more powerful laser ( pushing the shot noise down by about a factor of 10 ) , and replacement of the optical and suspension components with improved materials to reduce the impact of thermal noise .
in addition , the system will allow `` tunable '' noise curves experimenters will be able to shape the noise curve to some extent to chase after particularly interesting sources .
ground - based gravitational - wave detectors , particularly in their earliest generations , are primarily sensitive only to extremely violent astrophysical processes . in the frequency band of interest
( roughly @xmath67 hz for early designs , and roughly @xmath68 hz in future upgrades ) , these processes include the coalescence of compact binary systems and stellar core collapse in supernova .
these are extremely energetic but short - lived events ; for example , near the end of binary black hole coalescence , the binary s gravitational - wave luminosity approaches the theoretical maximum @xmath69 erg / sec for @xmath70 seconds , brighter than any other other source in the sky .
sources that should be of interest in later generations are continuous gravitational - wave emitters , such as pulsars or accreting neutron stars , and stochastic backgrounds , perhaps relics left from the big bang .
joint analyses with data from neutrino and gamma - ray detectors may prove particularly valuable , providing views of events through multiple radiation channels . in the remainder of these section , we discuss certain reasonably well - understood sources of gravitational waves , focusing in particular on those that promise to be useful in providing new tests of physics .
it is worth emphasizing at this point that , due to the pioneering nature of this field , there is great hope that we will see significant signals from unknown or unexpected sources .
figure [ fig : ligo_srcs ] illustrates several important gravitational - wave sources compared with ligo noise curves .
the noise curves shown are the amplitude spectrum @xmath71 , with units @xmath72 , for first generatation detectors , and for advanced detectors in both wide and narrow band configurations @xcite . for each source ,
the plotted strain @xmath73 is defined such that the ratio is equal to the ratio of signal strength @xmath74 to noise threshold @xmath44 , rms averaged over all source directions and orientations : @xmath75 .
the threshold is set assuming that the best practical data analysis techniques have been applied , and that the probability of a false alarm is one percent .
thus , a source that touches the noise in this figure is at the threshold for detection .
details of how the threshold is computed for each source are discussed in the following subsections .
coalescing compact binary star systems are currently the best understood sources of gravitational waves
. members of such binaries are compact , collapsed stellar remnants neutron stars or black holes .
double neutron star binaries are observed in our galaxy ; detailed studies of these systems @xcite provide what is currently our best data on gravitational - wave emission and led to joseph taylor and russell hulse winning the nobel prize in 1993 .
the galactic binary neutron star systems have orbital periods of several hours , radiating at @xmath76 , far beyond the range of ground - based detectors .
gravitational waves carry energy and angular momentum away from the system , driving the neutron stars to inspiral towards one another .
the emitted waves sweep upwards in frequency and amplitude as the neutron stars come closer together . in several hundred million years
, the gravitational waves from a galactic binary will enter the frequency band of ligo detectors ; about 10 minutes later , their neutron stars will violently collide and merge into a single object .
a few hundred million years is a bit long to run an experiment , so gravitational - wave detectors must be sensitive to a large volume of the universe in order to obtain an interesting event rate . extrapolating from the observed population to the universe at large ( see ,
e.g. , refs .
@xcite ) suggests that in order to measure an interesting rate of events ( several per year ) , detectors must be sensitive to coalescences hundreds of megaparsecs away .
extrapolation can only predict the rate of binary neutron star systems , since those are the only compact binaries that have been observed to date .
another method is population synthesis : modeling stellar evolution to predict the rate of compact binary formation from a population of main sequence stars ( see , e.g. , refs . ) .
population synthesis predictions span a rather wide range not surprising , since much of the underlying physics is rather uncertain .
most calculations agree at least to within an order of magnitude with the extrapolated predictions for binary neutron star coalescence .
this is by design : the calculations are `` tuned '' to match the data for binary neutron stars .
predictions for other compact binary systems , neutron star - black hole and black hole - black hole , vary quite a bit @xcite .
good data from gravitational - wave detectors will have a large impact on our understanding of stellar evolution and compact binary formation .
the entire coalescence process can be usefully ( albeit somewhat crudely ) divided into three broad epochs @xcite : the _ inspiral _ , in which the binary s members are widely separated and spiral inward due to gravitational - wave backreaction ; the _ merger _ , in which the binary s orbit becomes dynamically unstable , and the two bodies merge into a single object ; and ( possibly ) the _ ringdown _ , in which the merged remnant settles down to a stationary , rotating ( kerr ) black hole .
the ringdown only occurs if the binary leaves a black hole behind after coalescence .
this broad characterization is rather oversimplified , but useful , providing a qualitative description of the system s dynamics and wave emission .
further discussion of these epochs in the context of current research on coalescing binaries is given in sec .
[ sec : sciencereach ] . dividing the system s evolution into three epochs likewise divides the gravitational - wave signal into three broad frequency bands .
this is one reason that this crude characterization is useful : it gives some sense of what source dynamics are accessible to the observatories .
consider first the inspiral . very roughly speaking ( cf .
discussion in sec .
iiib of ref .
@xcite ) , @xmath77\;. \label{eq : insp_freq}\ ] ] ( here , @xmath12 is the cosmological redshift . )
likewise , we can with a fair degree of confidence characterize the frequency band of the ringdown waves .
using black hole perturbation theory @xcite , we find that the mode that is most likely to dominate after binary coalescence has @xmath78\nonumber\\ & \sim & ( 1200 - 3200)\,{\rm hz}\left[10\,m_\odot\over(1 + z)m\right]\;. \label{eq : ring_freq}\end{aligned}\ ] ] the kerr spin parameter @xmath79 varies from 0 to @xmath80 ; the span in frequency reflects this range .
merger waves are then all waves emitted at frequencies between these two extremes .
note that all frequencies scale inversely with the redshifted mass , @xmath81 .
the redshift factor should be fairly obvious , since @xmath82 is precisely the factor by which radiation s wavelength changes due to cosmological evolution .
the inverse mass dependence enters because the mass sets all timescales relevant to the radiation .
since frequency is an inverse timescale , all frequencies are proportional to inverse mass .
as mentioned in the discussion of the ldas architecture , some systems will be analyzed using _
matched filtering _
, correlating the experimental data with theoretical waveform models , known as templates @xcite . matching a template to the signal boosts the signal - to - noise ratio ( snr ) by a factor that is roughly @xmath83 , where @xmath38 is the number of measured gravitational - wave cycles . because the ringdown and early inspiral are rather well - understood ,
we are confident in our templates for those coalescence epochs ; the late inspiral and merger epochs are rather more poorly understood .
this motivates much research in relativity theory today . the tracks illustrating compact body coalescence in fig .
assume that matched filtering is applied .
the thresholds are computed assuming that matched filters are used to integrate the signal at each frequency over bands of width @xmath84 .
we assume that data from all three ligo interferometers is combined . for neutron star - neutron star binaries ,
only the inspiral waves are easily accessible to ground - based detectors . as will be mentioned in sec .
[ subsec : dense_nuclear ] , narrow banded interferometers and acoustic detectors will be able to provide some information about the high frequency binary neutron star merger signal .
the binary neutron star signal is fairly weak ; for assured detection by first detectors ( amplitude snr @xmath85 in all ligo interferometers ) , the binary can be no further than about 20 mpc .
we must go much farther out to obtain an interesting rate of neutron star coalescences . as a consequence ,
current wisdom is that neutron star - neutron star detection by first detectors , though plausible , is not likely .
however , detection after upgrade is quite likely by figure [ fig : ligo_srcs ] , neutron star - neutron star coalescence should be seen out to about 300 mpc .
non - detection of binary neutron star coalescence by advanced interferometers would be rather surprising .
neutron star - neutron star waves are weak because of the system s relatively small mass . increasing the mass increases the signal , but shifts all frequency bands downward .
the shift in frequency causes much of the inspiral to shift out of band , leaving the very late stage of inspiral and the merger in ligo s most sensitive band .
figure [ fig : ligo_srcs ] shows that for the more massive binary black hole systems , the late inspiral and merger waves are likely to be the most relevant for detection .
the fact that these rather poorly understood epochs of binary black hole coalescence are right in the most sensitive band of ground - based detectors greatly motivates theoretical work to understand this waves better .
if we succeed in modeling the late inspiral and merger accurately , these signals should be detectable at cosmological distances , with redshift @xmath86 .
stochastic backgrounds are `` random '' gravitational waves , arising from a large number of independent , uncorrelated sources that are not individually resolvable . a particularly interesting source of backgrounds is the dynamics of the early universe an all sky gravitational - wave background , similar to the cosmic microwave background .
backgrounds can arise from amplification of primordial fluctuations in the universe s geometry , phase transitions as previously unified interactions separated , or the condensation of a brane from a higher dimensional space
. mechanisms of this kind and their connections to unification physics are discussed in sec . ; here , we briefly discuss how these backgrounds are characterized and the levels of sensitivity ligo should achieve .
stochastic backgrounds are always discussed in terms of their contribution to the universe s energy density , @xmath87 . in particular , one is interested in the energy density as a fraction of the density needed to close the universe , over some frequency band : @xmath88 different cosmological sources produce different levels of @xmath89 , centered in different bands . amplified primordial fluctuations surely exist , but are likely to be rather weak
: estimates suggest that the spectrum will be flat across ligo s band , with magnitude @xmath90 @xcite .
waves from phase transitions can be significantly stronger , but are typically peaked around a frequency that depends on the temperature @xmath44 of the phase transition : @xmath91 we note here that the temperatures required to enter the lisa band , @xmath92 hz , is @xmath93 , nicely corresponding to the electroweak phase transition .
waves arising from extradimensional dynamics should peak at a frequency given by the scale @xmath94 of the extra dimensions @xcite : @xmath95 for the waves to be in ligo s band , the extra dimensions must be rather small , @xmath96 meters .
lisa s band is accessible for a scale similar to those discussed in modern brane - world work @xcite .
the detectable magnitude of @xmath89 and the sensitive frequency band depend on the particular instrument .
ligo will measure stochastic backgrounds by comparing data at the two sites and looking for correlated `` noise '' @xcite . for comparing to a detector s noise
, one should construct the characteristic stochastic wave strain , @xmath97 ( for further discussion and the proportionality constants , see @xcite . )
stochastic backgrounds are illustrated in fig . as the downward sloping dotted lines , labeled @xmath98 , @xmath99 , etc .
these curves assume @xmath100 .
early detectors will have fairly poor sensitivity , sensitive to a background at levels @xmath101 in a band from about 100 hz to 1000 hz .
this is barely more sensitive than known limits from cosmic nucleosynthesis @xcite .
later upgrades will be significantly more sensitive , able to detect waves with @xmath102 , which is good enough to place interesting limits on cosmological backgrounds .
another promising source of radiation for ground - based detectors is the core collapse of massive stars . when massive stars die their most dense central regions catastrophically collapse , driving the star to explode in a supernova and the innermost matter concentrations to form a neutron star or black hole .
the conditions for strong gravitational wave emission highly energetic dense matter dynamics are clearly met . however , because stellar collapse is not very well understood , our understanding of gravitational - wave emission from such collapse is rather uncertain .
one of the major goals of gravitational - wave observatories is to coordinate gravitational - wave detection with external observations
. this will be particularly important for stellar core collapse due to our poor understanding of the likely gravitational waves emitted in that case coincident detection with neutrino , gamma ray , and/or optical observatories will greatly increase confidence in the measurements , as the combined network of instruments provides triggers and cross - checks for each other .
coincident detection and measurement will produce great science payoff aside from increased confidence and cross - checks .
for example , a supernova in our galaxy would be easily detected by both neutrino and gravitational - wave observatories .
measurement of such events will paint a far more solid picture of collapse dynamics than our current view , tracking the relative evolution of processes that generate neutrinos with the dense matter dynamics .
an event in our galaxy would have very large snr , so such studies could be done with high precision .
such nearby events will be very rare ( the supernova rate is estimated to be several events per galaxy per century ) , but would provide enormous scientific payoff .
a review of gravitational waves from stellar collapse , focusing on their relevance to gravitational - wave observatories , has recently appeared @xcite ; our discussion is largely based on that paper .
the strongest core collapse gravitational waves come from instabilities that can develop in the matter dynamics .
one of the most - discussed instabilities is _ bar formation _ : the tendency of rotating matter to go into a non - axisymmetric , rotating , bar - shaped mode .
a bar mode is potentially a strong source of radiation .
many numerical simulations @xcite have shown that bar modes are promising sources , though usually with rather simplified models . ref .
@xcite looked at the potential for bar mode instability in a variety of realistic stellar collapse scenarios , and found that it still remains a promising source of waves for ground - based observatories , though much work remains to be done to clarify the waves characteristics in different circumstances .
for example , bar modes will only be unstable if the precollapse progenitor is rotating sufficiently rapidly .
this appears likely for many supernova progenitors .
it _ may _ also be the case in `` accretion induced collapse '' ( aic ) of some white dwarf stars , though the conditions for this to occur and its likely rate make aic much less promising as a source @xcite . in some star collapses , the distribution of density and angular momentum is such that the dense inner material may fragment into pieces , forming `` chunks '' that rapidly orbit for some time before settling into an equilibrium
. these orbiting chunks would be copious gravitational - wave radiators .
van putten has recently argued that such a matter distribution is likely to form in the progenitor to a gamma - ray burst ( cf .
@xcite and references therein ) .
gamma - ray bursts may thus be accompanied by an extraordinarily strong burst of gravitational radiation .
fryer , holz , and hughes @xcite find that a fragmentation instability may allow ground - based detectors to observe the collapse of population iii stars a putative population of extremely massive ( @xmath103 ) stars that formed and died early in the universe s history .
this is an example of new astrophysics that gravitational - wave astronomy may discover .
periodic sources of gravitational waves are emitters which radiate at constant frequency ( or nearly constant frequency ) , much like radio pulsars . at first sight , this suggests they might be easy to detect it should be simple to coherently follow a periodic source s phase evolution , allowing us to build up enough power for an intrinsically weak signal to stand above noise . however , the signal is strongly modulated by the earth s rotation and orbital motion , `` smearing '' the wave across multiple frequency bands and greatly degrading the source s strength . searching for periodic gravitational waves means demodulating the motion of the detector .
this is computationally arduous the modulation is different for every sky position .
unless one knows in advance the position of the source , one needs to search over a huge number of sky position `` error boxes '' , perhaps as many as @xmath104 .
one rapidly becomes computationally limited .
for further discussion , see @xcite ; for ideas about doing hierarchical searches , requiring less computational horsepower , see @xcite .
the prototypical source of continuous gravitational waves is a rotating neutron star .
if the neutron star is non - axisymmetric ( for example , it has a crust that is somewhat oblate and misaligned with the star s spin axis ) , it will radiate gravitational waves with characteristic amplitude @xmath105 where @xmath106 is the star s moment of inertia , @xmath107 characterizes the degree of distortion ( see @xcite and references therein for further discussion ) , @xmath108 is the wave frequency , and @xmath109 is the distance to the source .
realistic choices for these parameters suggest that @xmath110 or smaller . to measure these waves ,
we need to coherently track the signal for a large number of wave cycles . coherently tracking @xmath38
cycles boosts the signal strength by a factor @xmath111 . the characteristic amplitude of several possible pulsars
are illustrated in fig .
[ fig : ligo_srcs ] .
the points labeled `` crab '' and `` vela '' illustrate the maximum possible gravitational - wave strength that the well - known crab and vela pulsars could emit .
this pulsars are known to `` spin down '' ; that is , their rotation periods are observed to decrease over time .
figure [ fig : ligo_srcs ] assumes that _ all _ of this spindown is due to gravitational - wave backreaction , an assumption that is known to be wrong .
the waves from these pulsars will be much weaker ; however , since their sky position and frequencies are known in advance , it will not be too difficult to search for any waves they might emit .
the upward sloping lines labeled `` known pulsar , @xmath112 , @xmath113 '' illustrate the waves produced by a source whose position is known but frequency is unknown , if the distance and deformation are as indicated . in all of these cases , it is assumed that the signal can be coherently tracked for several months , building up power .
a nearly periodic source of gravitational waves from rotating neutron stars is the r - mode instability , a fluid instability that may be active in hot , fluid neutron stars @xcite .
this mode is not perfectly periodic , since radiative backreaction rapidly reduces the star s spin and gravitational - wave frequency ; but it is nearly so , and search techniques for r - mode active stars are likely to be very similar to searches for periodic sources @xcite .
finally , we note that neutron stars accreting matter from a close companion may be periodic sources of gravitational radiation @xcite .
this would explain why most neutron stars in such low mass x - ray binary ( lmxb ) systems appear to have a maximum spin frequency : rather than spinning up indefinitely by the matter accreting onto them , they eventually are braked by gravitational - wave backreaction .
such sources are very promising targets for observation since their sky positions are known very accurately from x - ray observation . since they are in binaries , though , their waveforms will be even more strongly modulated than `` normal '' radiating pulsars .
the points on fig .
[ fig : ligo_srcs ] labeled `` sco x-1 '' and `` lmxbs '' show waves that are possible from known lmxb systems
. many of these systems will only be detectable in the narrow - banded configuration , demonstrating the science impact possible with tunable interferometer configurations .
many promising sources of gravitational waves radiate at low frequencies , from a few microhz to about a hz , where earth - based detectors have very poor sensitivity due to geophysical noise .
such sources include massive black hole systems , compact binaries , very close normal binaries , and cosmological backgrounds . to get away from the geophysical background
, the detector must be located in space . in this section ,
we describe the space - based detector lisa which will target these low - frequency sources .
we first describe the lisa mission , and then discuss certain sources that are of particular interest for lisa science .
more extensive information on the mission and additional references are given in refs .. the laser interferometer space antenna ( lisa ) is designed to detect and study in detail gravitational - wave signals at frequencies from roughly 10 microhz to 1 hz .
once the conclusion to build a low - frequency , space - based detector has been reached , it becomes attractive to make the size of the antenna quite large .
using laser heterodyne measurements between widely separated spacecraft , even with 1 w power levels , antenna sizes of millions of kilometers are feasible .
the measurements are made between freely floating proof masses , inside the spacecraft , that are very carefully shielded from both internal and external disturbances . with this approach
, very desirable sensitivity levels can be achieved throughout the planned frequency range .
the lisa mission is planned as a joint mission of the european space agency and nasa , for launch in about 2010 .
the antenna will be based on laser measurements between three spacecraft located at the corners of an equilateral triangle that is 5 million km on a side .
a one year orbit around the sun for each spacecraft has been found such that : ( a ) the spacecraft separation stays constant to better than @xmath114 over more than 10 years ; ( b ) the center of the antenna is on a circular orbit in the ecliptic plane and about 50 million km ( @xmath115 ) behind the earth ; ( c ) the plane of the antenna is tipped at @xmath116 to the ecliptic , and that plane rotates around the pole of the ecliptic with a one year period ; and ( d ) the antenna rotates within that plane with a one year period .
the orbits each have an eccentricity of about 0.01 , and an inclination that is the square root of 3 times the eccentricity .
the main components of the payload for each spacecraft are located within a y - shaped thermal shield made of carbon fiber reinforced plastic , which has a low coefficient of thermal expansion .
there are two cylindrical optical assemblies about 1 m long and 0.35 m in diameter that point out along two of the arms of the y and toward the distant spacecraft . a schematic drawing of one optical assembly is shown in fig . [ fig : opt_assembly ] .
only a brief description of its operation will be given here .
the rectangular structure located near the center of the optical assembly is the optical bench .
it is supported by a wheatstone bridge mounting from the optical assembly support cylinder so that temperature gradients along that cylinder do nt get transmitted conductively to the optical bench .
the gravitational sensor is mounted at the center of the optical bench , and contains the freely floating proof mass and the housing around it .
capacitive bridge measurements between a number of pairs of capacitive plates on the inside of the housing and the proof mass determine even small translations and rotations of the proof mass with respect to the housing .
the laser measurements of changes in the distances between the proof masses in different spacecraft are made to the front face of each proof mass through a small window in the outer enclosure of the gravitational sensor .
light from the 1.06 micron nd : yag laser ( not shown ) is brought onto the optical bench through a fiber , and almost all of it is sent by a polarizing beamsplitter to the 0.3 m diameter transmit / receive telescope .
the transmitted beam has a diameter of about 20 km when it reaches the distant spacecraft , and a small portion of it is collected by the telescope there .
this light goes through the beamsplitter , is reflected from the face of the proof mass , and is then deflected by the beamsplitter to go to a photodiode detector .
some light from the on - board laser also goes to the photodiode , and the beat between the received beam and this local laser beam is one of the six main signals generated by the antenna .
each beam propagating in one of the two directions along one of the three sides of the antenna is used in generating one of the six main signals . because of the choice of spacecraft orbits , the lengths of the sides of the triangle stay constant in length to better than @xmath114 over 10 years or longer , even including planetary perturbations
. however , the resulting relative velocities are still on the order of 10 m / s , and cause doppler shifts of up to roughly 10 mhz at some times .
these doppler shifts vary typically with periods of several months or longer and in an extremely smooth fashion , so that they can be fitted out from the observations .
the gravitational waves will cause phase variations in the lisa signals that can be analyzed after the huge but much lower frequency doppler shifts have been fitted out . to understand how the lisa antenna works , it is useful to consider only two sides of the triangle , with the spacecraft common to both sides being the master spacecraft .
the lasers at the two distant spacecraft are assumed to be phase locked to the received laser beams , so that the return beams have the same phase as if the beams were just reflected by mirrors .
thus the system acts like a michelson interferometer , but with two signals generated back at the master spacecraft that give the variations in the length of each of the interferometer arms separately . to the extent that the proof masses at the ends of the arms really are undisturbed by spurious accelerations , the apparent variations in the sum of the lengths of the two arms will almost all be due to phase noise in the laser .
this information then can be used to correct the difference in length of the two arms in software for the laser phase noise , which otherwise would swamp the gravitational wave signals .
an industrial phase a study of the lisa mission was carried out for the european space agency and presented to their scientific community in september 2000 .
the spacecraft , payload and mission designs from this study , which started from the results of earlier studies in both europe and the us , essentially form the current baseline plan for the mission . under this plan , each spacecraft is 2.7 m across and 0.56 m high , with the sides slanted in at @xmath117 so that sunlight only hits the top surface , where the solar cells to provide power are mounted .
the optical assemblies point out through the slanted sides , and the direction to the sun makes a constant @xmath117 angle with respect to the top of the spacecraft .
this keeps the temperature and temperature distribution in the spacecraft extremely constant .
this is a major reason why the time variations in spurious forces acting on the carefully shielded proof masses deep in the spacecraft can be kept extremely low .
each spacecraft initially has a propulsion module attached to it , and the three combined units are stacked up and launched by a single launch vehicle into roughly 13 month period elliptical orbits around the sun . after leaving the earth , the units separate and their propulsion units carry them to their desired orbits over a period of roughly a year .
the orbits are checked for a few weeks by tracking with nasa s deep space network , and minor orbit corrections are made as needed .
the propulsion modules then separate gently from the spacecraft and drift away .
after separation , the spacecraft are oriented properly with their optical assemblies pointing at the other spacecraft forming the antenna .
this is done using continuously operating and electrically controllable micronewton thrusters mounted on the spacecraft .
next , the proof masses are released from the clamping mechanisms that have held them during launch and electrical forces are applied to move them to the centers of their housings .
then , the control laws for the micronewton thrusters are changed so that they also take on the job of making the spacecraft follow the average position of the two proof masses .
the motion of the spacecraft is then determined by the forces acting on the proof masses , and the effect of non - gravitational forces on the spacecraft can be essentially eliminated .
this is called a drag - free system , since such systems were first used to greatly reduce the drag on earth satellites due to the residual atmosphere . for lisa , since there are two proof masses , weak electrical forces perpendicular to the laser beam directions are still applied to the proof masses to keep them centered in their housings .
the capacitive position measurements are sensitive enough and the noise in the thrusters low enough that the relative motion of the spacecraft with respect to the proof masses can be kept very low .
this is another reason why the variations in spurious forces on the proof masses can be kept very low .
the total thrust needed is mainly the roughly 20 micronewtons required to buck out the solar radiation pressure , and varies only slightly .
finally , a beam acquisition procedure is started to successively turn on each laser and adjust its pointing direction accurately .
when all six main signals are obtained , the scientific part of the mission can begin
. the nominal scientific mission lifetime may be as short as two or three years , but it is hoped that the observations can continue for about a decade . the thrust required from the micronewton thrusters is low enough that the reaction mass needed even for a decade of operation is small .
the instrumental sensitivity of the lisa antenna is determined by two factors .
one is the noise in measuring changes in the distances between the different proof masses .
the second is the level of spurious accelerations of the proof masses . from noise budgets for these two different types of noise , a threshold sensitivity curve for lisa
can be determined .
such a curve is shown in fig .
, along with some information about expected signals from binary stars in our galaxy that will be discussed later .
the sensitivity curve shown is for 1 year of observations and an amplitude snr of 5 .
nearly all of the resolvable galactic binary sources expected for lisa will be thousands to millions of years from coalescence , and thus their frequencies will change little over a few years .
each can be represented by a single point giving the rms gravitational - wave amplitude and frequency ( twice the orbital frequency ) .
if that point lies above the sensitivity curve , then there is enough signal to overcome the instrumental noise with 1 year of observations and a snr of 5 .
this snr is needed to be confident of detection for a source whose sky position and frequency are not known .
there is of course an error allocation budget for each of the two types of noise .
in such a budget , allowable levels of error are assigned to each of the expected or possible sources of error in the measurements .
for measuring changes in the distances between proof masses , the total error budget level for the round - trip arm length difference of two antenna arms is 40 picometers per root hz ( pm / rthz ) , independent of frequency ( white noise ) down to below 1 millihertz ( 1 mhz ) .
this is called the spectral amplitude of the noise , and is defined as the square root of the power spectral density .
the noise power in a narrow bandwidth is proportional to the bandwidth , with units of [ ( pm)@xmath118/hz , so the spectral amplitude has units of pm / rthz .
most of the expected distance measurement noise comes from shot noise in the detected laser photons .
the rate of photon detections for the received signal is a few times @xmath119 per second , so measurements can be made to roughly @xmath120 of a laser wavelength in 1 second , or about @xmath121 in 1 second for the fractional change in the round - trip distance .
there is some additional contribution to the error due to jitter in the pointing of the laser beam toward the distant target and other noise sources , but they are relatively small
. the lisa sensitivity curve at frequencies substantially above 3 mhz is determined by the distance measurement noise . at lower frequencies ,
the instrumental sensitivity is limited mainly by spurious accelerations of the proof masses .
there are roughly half a dozen sources of such acceleration noise that dominate the error budget , and they are assigned equal error levels . with allowances for other smaller acceleration noise sources ,
the rms total is @xmath122 meters sec@xmath123/rthz down to 0.1 mhz , and somewhat higher at still lower frequencies .
a space validation flight for the subsystem consisting of the proof mass , its housing , and the associated electronics currently is planned in order to verify that they perform according to their design specifications . -2
cm with the sensitivity curve shown in fig .
[ fig : lisa_sens ] , lisa will be able to observe signals from many binaries in our galaxy .
in order for the two members of a binary to be close enough together to have gravitational wave frequencies above about 0.1 mhz , both must be compact objects white dwarf stars , neutron stars , or black holes .
the number of such binaries in our galaxy is roughly @xmath119 ; most will be close white dwarf binaries .
below 1 or 2 mhz there will be more than one in each 1 cycle / year frequency resolution bin , so that individual ones can not be resolved .
the curve labeled `` confusion noise estimate '' in fig .
[ fig : lisa_sens ] represents the expected noise level due to the random superposition of such signals . at higher frequencies ,
the number per bin drops rapidly with increasing frequency , as they lose energy more and more rapidly by gravitational - wave emission .
about 3000 will be resolvable above about 3 mhz .
roughly @xmath124 of them will have amplitudes between the curves labeled @xmath125 and @xmath126 in the figure .
the direction and frequency of only a few of them will be known ahead of time , and some of these are shown .
there are at least three types of events involving massive black holes ( mbhs ) that appear likely to produce signals observable by lisa .
however , the event rates can only be estimated very crudely , and this situation seems unlikely to change much in the near future .
the primary objective of the lisa mission is to detect such signals and study them in detail . but
none of these signals are guaranteed to be present , so the present design of the lisa antenna has been chosen to make as many of the types of mbh signals observable as possible . the first type of signal of strong interest is associated with the origin of massive black holes . an important question is how the seed mbhs that later grew to be the massive and supermassive black holes observed today were formed .
we label black holes with masses @xmath127 `` stellar mass black holes '' ( potentially formed by the evolution of massive stars ) ; holes with mass up to about @xmath128 are labeled `` massive black holes '' ( mbhs ) ; and still larger ones `` supermassive black holes '' .
there are several theories for how seed mbhs form . in one ,
stellar mass black holes sink to the center in a dense galactic core by dynamical friction , and collisions lead to the formation of higher mass objects .
the largest black hole grows faster than all others , swallowing up holes comparable size .
this becomes the seed for growth of an mbh of perhaps @xmath129 or larger .
when it reaches a mass of roughly @xmath130 , it would be able to continue growing fairly rapidly by absorption of gas in the galactic nucleus and by tidal disruption of stars @xcite .
an alternate scenario postulates a dense cloud of gas and dust , evolving to the point where it becomes optically thick .
radiation pressure plus magnetic fields then prevents further fragmentation of the cloud to form stars @xcite . at that point
, if energy and angular momentum can be dissipated fairly rapidly , several things might happen .
one possibility is that a supermassive `` star '' with mass @xmath131 forms .
this would be subject to a pulsational instability @xcite that drives it to collapse into a mbh .
another possibility is that the cloud may become dense enough to reach the point of relativistic instability and collapse directly to a mbh without going through the supermassive star stage . from the standpoint of lisa observations ,
the major issue of the collisional growth scenario is whether some of the individual black holes become large enough before merging with the largest hole that their coalescence is detectable .
an example mass pair that would just be detectable during the last year before coalescence at @xmath132 @xmath133 ; another is @xmath134 , @xmath135 .
the lowest straight curve in fig . [ fig : lisa_bbh ] shows the signal strength and frequency evolution during the year before coalescence of a binary with @xmath136 at @xmath137 .
the ticks along this line represent times 0.0 , 0.2 , 0.4 , 0.6 , 0.8 , and 0.99 year from the beginning of the last year .
such a signal would be observable even from a coalescence at @xmath138 .
the other curves in the figure will be discussed later .
a second important signal comes from black holes of roughly @xmath139 , neutron stars , or white dwarfs in orbit around mbhs in galactic nuclei .
such highly unequal mass systems will result from compact objects in the cusp around the mbh being scattered in close enough to start losing energy by gravitational radiation , becoming progressively more tightly bound .
the orbits remain elliptical throughout in this case , and during the last year the orbital speed near periapsis can reach almost half the speed of light .
the period for periapsis precession is similar to that for radial motion ; rapid frame - dragging may also be present .
thus the orbits are extremely non - newtonian , as discussed later . from the astrophysics point of view , the number and type of systems of this kind observed will give valuable information about a combination of the distribution of mbh masses in galactic nuclei and the space density of compact objects in the cusps around mbhs . the third and strongest type of mbh signal expected for lisa is from the coalescence of binaries consisting of two mbhs , where both are substantially more massive than @xmath140 .
such events can follow the mergers of galaxies or pre - galactic structures , provided that two conditions are fulfilled .
first , the pre - merger structures must already contain mbhs ; second , the mbhs must form a close binary in substantially less than the age of the universe .
present estimates are that these conditions have been fulfilled frequently enough to give one or more mbh coalescences per year , but even this is uncertain .
it seems likely that coalescences involving mbhs of roughly @xmath141 will be the most common ; mergers of structures with @xmath142 or more massive mbhs may well be less frequent , and less massive mbhs may have difficulty forming close binaries . the signal strength during the last year for equal - mass mbh coalescences at @xmath132 are shown in fig .
[ fig : lisa_bbh ] .
since the snr ratios are high , such events would be easily observable anywhere in the universe .
the waves from these events will be measured sufficiently well that detailed parameter fits will be possible @xcite , making possible much astrophysical analysis
. however , since the event rate is definitely not guaranteed , efforts will be made to keep the acceleration noise level at frequencies below 0.1 mhz as low as possible .
this will increase the chances of seeing coalescences or @xmath143 or @xmath144 mbhs a number of years before coalescence , and thus improve the statistical information on them . by their very nature ,
stochastic gravitational waves are difficult to distinguish from noise . as described in sec .
, ground - based detectors can find them by coordinated measurements , essentially comparing the outputs of multiple detectors to find sources of correlated noise .
provided there are no common noise sources ( which , due to their wide separation , should be the case ) , this should work well .
lisa is a single space - based antenna and so can not use this technique , but has other tools that can be employed instead . in particular , it is possible to combine the signals from lisa s spacecraft in such a way that the detector is _ insensitive _ to gravitational waves @xcite .
observables of this kind are called `` sagnac '' observables , since they correspond to observables constructed from phase information that propagates around the lisa antenna , as in a sagnac interferometer .
since the sagnac observable is insensitive to gravitational waves , it measures noise alone . in this way , one can detangle the noise - like signature of a stochastic background from the actual interferometer noise @xcite .
lisa s stochastic background sensitivity should then be limited by the galactic white dwarf background ( a somewhat prosaic stochastic background ) .
this will allow it to set strong limits on cosmological backgrounds in its band , corresponding to an energy density about a million times smaller than the cosmic microwave background .
lisa s band is fortuitously placed to measure potentially interesting cosmological sources , as has been mentioned already in sec .. an electroweak phase transition , occuring when the temperature of the universe was @xmath145 gev , generates waves peaked at a frequency @xmath146 hz .
such waves could be directly detected by lisa @xcite .
likewise , waves from extradimensional dynamics are likely to be in lisa s band for extra dimensions on the scale of millimeters to microns @xcite .
as discussed in the next section , lisa is well situated to at least put stringent limits on many early universe scenarios , and perhaps directly detect stochastic waves .
as should be clear from the discussion of sources given in secs . and [ sec : spacebased ] , the routine detection of gravitational waves will certainly transform the field of astronomy .
the rarest and most obscure cosmic events , such as black hole mergers , will suddenly achieve a phenomenological prominence appropriate to their huge power output .
populations barely detected at all today , such as white dwarf binaries , will produce a din that overwhelms even the lisa instrumental noise sources .
the current orders - of - magnitude uncertainties in rates and intensities of all kinds of gravitational - wave sources clearly indicates our ignorance of the dominant mass - energy flows
. we will learn about details of many things that astrophysicists regard as fundamental : the dynamics of massive black hole formation in the centers of galaxies , the evolution of binary star systems , perhaps the equation of state of nuclear matter and the mechanism of neutrino - driven supernova explosions .
readers of these proceedings will be more interested in new physics than new astrophysics .
these observatories should not be regarded as `` merely '' tools of astronomers ; they will probe phenomena that will either constrain or reveal something new about the laws of nature .
we discuss these possibilities in the subsections below .
we first describe how gravitational - wave observatories will probe the nature of extremely strong gravity .
section [ subsec : sf_dynamic ] discusses the gravitational - wave probes of highly dynamic gravity that will be studied in the merger of binary black hole systems ; sec .
focuses on measurements that will map in detail the structure of black hole spacetimes .
it is worth noting at this point that it is still not conclusively known whether the massive objects usually described as `` black holes '' in fact are black holes as described by einstein gravity ( though the evidence from sources such as x - ray telescopes is getting better and more conclusive all the time ) .
section [ subsec : dense_nuclear ] discusses how tracking gravitational waves from systems with neutron stars may be used to study the properties of dense nuclear matter , probing its equation of state . finally , in sec .
[ subsec : new_physics ] we examine what kinds of `` new physics '' may be probed , focusing in particular on how gravitational waves may be used to test new ideas for unification physics beyond the standard model . for the general relativity theorist ,
some of the greatest promise of gravitational - wave astronomy comes from the stringent tests of classical gravity that it will provide .
the most spectacular such tests are likely to come from observing the mergers of black holes .
comparing the waveforms measured by ground- and space - based observatories to those computed from theory will test the strongest and most dynamically varying gravitational fields since the big bang .
such comparisons obviously require us to have theoretical waveforms in hand .
for the most interesting , extremely dynamical , strong - field phase of binary black hole coalescences , this is unfortunately where theory is currently stuck . from the standpoint of mathematical physics
, this problem may appear deceptively simple : black holes contain no matter , so one `` merely '' needs to find an appropriate solution to the vacuum einstein equations , @xmath147 ( see ref . or @xcite for a definition of the einstein tensor @xmath148 ) .
such an appropriate solution is one which : ( a ) consists of two widely separated black holes in the asymptotic past ; ( b ) consists of a single kerr black hole in the asymptotic future ; ( c ) allows only outgoing radiation to radiate `` out to infinity '' ; and ( d ) allows only ingoing radiation to propagate down event horizons .
anyone with experience solving nonlinear partial differential equations subject to bizarre boundary conditions will appreciate that this is a tall order . at the moment , only the asymptotic past and the asymptotic future solutions are very well understood .
when the black holes are widely separated , one can study the binary using the post - newtonian approximation to general relativity .
this approximation treats the dynamics of the system as newtonian at lowest order , and finds corrections to the dynamics as a power series in @xmath149 ( where @xmath109 is orbital radius ) .
this dimensionless parameter roughly describes the gravitational field strength . for more detailed discussion ( particularly of
the many subtleties which this discussion has ignored ) , see @xcite .
tests of general relativity in this early `` inspiral '' epoch are essentially generalizations of current tests that use binary pulsar systems @xcite .
the binary pulsars are relatively weak - field , however , with hundreds of millions of years remaining until the neutron stars merge .
more stringent limits on general relativity s validity are likely to be set examining the final few minutes of inspiral , as the members of the binary spiral through extremely strong gravitational fields . in the asymptotic future of the system ,
the holes will have merged and formed an isolated spinning black hole .
the final waves to come from the system will be the `` ringdown '' waves emitted as the merged remnant settles down to the kerr rotating black hole solution .
the ringdown waves are those emitted by a black hole that is distorted from its quiescent kerr form , and can be modelled with black hole perturbation theory @xcite .
general relativity predicts that the frequency and damping times of black hole modes depend only on the mass and spin of the hole .
the longest lived mode correspond to a mode with spherical - harmonic - like indices @xmath150 a bar - like mode , rotating in the same sense as the hole s spin .
the frequency and damping times of those modes are well - understood @xcite , and so matching observation and theory should be relatively simple . as the inspiraling black holes come close together ,
their gravitational interaction becomes progressively stronger .
eventually , the number of terms that must be kept in a post - newtonian expansion becomes large enough to be impractical .
this is the vanguard of current research into binary systems in general relativity : understanding how to model the dynamics and gravitational - wave emission of extremely close binary black hole systems , particularly the `` merger '' epoch in which the two black holes fuse into a single object . as discussed in sec .
[ sec : groundbased ] , it is likely that for some systems the most poorly understood , strong - field late inspiral and merger regime will produce waves that are ideally suited for detection .
this further motivates theoretical work to understand these waves in detail .
most efforts to model strong - field binary black hole dynamics have focused on _ numerical relativity _
@xcite : large - scale computational attempts to solve einstein s equations , subject to the constraints and conditions discussed above .
this has proven to be an extraordinarily difficult problem . solving the 10 coupled ,
nonlinear partial differential equations of the tensor equation @xmath147 is difficult enough for starters .
most efforts begin by splitting the 4 dimensions of spacetime into 3 space plus 1 time dimension .
( not all efforts focus on this 3 + 1 split .
there is a substantial body of work that splits spacetime into 2 spatial directions and 2 null directions , along which radiation propagates ; see ref . and sec .
3.2 of @xcite for detailed discussion . )
einstein s equations then divide into constraints which the data must satisfy on each time `` slice '' , and evolution equations which step the data forward from slice to slice .
there is a great deal of freedom in how one can perform the space - time split , and it is not obvious how it can be done in a manner that is maximally useful for both modeling a binary s dynamics and extracting the gravitational wave information . in addition , there are many different ways of actually writing the einstein equations down in preparation for a numerical computation .
it is now clear that some forms of einstein s equations are particularly bad , being susceptible to the growth of unphysical instabilities that poison a computation .
much of the effort in numerical relativity is now focused on finding `` good '' equations ( see , for example , ref .
@xcite and references therein ) .
progress in the field has been rapid recently , with large advances in the ability to evolve systems with two black holes and large amounts of radiation without making too many simplifying assumptions ; see refs .
@xcite for further discussion
. relatively recent work by thibault damour and colleagues has developed an alternative analytical approach to understanding the strong - field dynamics of binary black hole systems ; see @xcite and references therein .
this work is based in part on `` resummation methods '' for improving the analytic post - newtonian description of gravitational - wave emission , coupled with a novel recasting of a binary s two - body dynamics in terms of the motion of a single body in an `` effective one - body metric '' , usefully thought of as a deformed black hole .
the resummation methods are , essentially , pad approximants , rewriting the ( poorly convergent ) post - newtonian description of the body s dynamics as a function with improved convergence properties .
the effective one - body remapping technique is inspired by tools that have been developed to describe two - body problems in quantum electrodynamics ; see ref . for details .
both the new analytic techniques and numerical relativity show great promise and will hopefully work together to elucidate strong - field binary dynamics and their gravitational - wave generation .
the great difficulties and uncertainties in theoretical modeling apply to _ comparable mass _ binary black hole systems .
these systems are also among the most interesting from the standpoint of testing relativity , since there may be high event rates for both ground- and space - based detectors , and because the gravitational dynamics are the most extreme .
a special case that can be modeled with less difficulty is that of _ extreme mass ratio _ binary black holes : one black hole far more massive than the other . as discussed in sec . , these are sources that are targets for lisa .
the spacetime of the binary is describable in this case as that of a single rotating black hole plus a perturbation arising from the small black hole .
the system s dynamics are well - described as an adiabatically evolving orbit of the small body .
a substantial body of literature on such systems has developed over the past decade ( examples of which are refs . ) ; all work on this subject is based , at heart , on teukolsky s 1973 description of radiative perturbations to rotating black holes @xcite .
these extreme mass ratio inspirals promise to be wonderful tools for precision tests of gravitation . rather than probing the violent strong - field dynamical properties found when comparable mass holes merge ,
extreme mass ratio inspirals will make it possible to _ map _ , with high accuracy , the structure of the large black hole s spacetime .
einstein s theory of gravitation predicts that black holes have event horizons and have a structure that depends only on the mass and spin of the black hole .
extreme mass ratio inspirals promise an extremely clean test of this structure , demonstrating once and for all that the massive objects seen in many galaxies are `` einstein black holes '' , or in fact are something even more bizarre . for an astrophysically interesting range of black hole masses ( about @xmath151
black hole masses commonly encountered in the centers of galaxies @xcite ) the gravitational waves generated during inspiral are at frequencies ideally suited for measurement by lisa .
these waves come from the small body spiraling through the black hole s very strong field regions very close to its event horizon .
the number of orbits executed as the small body spirals inwards is very large it orbits about @xmath152 times before reaching a dynamical instability and plunging into the hole ( depending on the specific details of the system ) . by tracking the gravitational wave s phase evolution over these many orbits
, we will be able to `` weigh '' the spacetime s multipoles .
rotating black holes have a family of multipole moments set entirely by the black hole s mass and spin : @xmath153 in this equation , @xmath154 is a spherical - harmonic - like index , @xmath155 is the @xmath154-th `` mass moment '' , and @xmath156 is the @xmath154-th `` current moment '' .
( the index @xmath45 does not enter because quiescent black holes are axisymmetric : all multipoles with @xmath157 are zero . ) for a fluid body , with density @xmath158 and velocity distribution @xmath159 , the moments would be @xmath160 a rotating black hole has a family of mass multipoles that is purely even , a family of current multipoles that is purely odd , and is determined entirely by the first two moments : @xmath161 and @xmath162 . that these two moments the black hole s mass and spin determine the _ entire _ family of multipole moments is a restatement of the no - hair theorem : _ a black hole s properties are entirely determined by its mass and spin_. ( we ignore charged black holes , which are of little astrophysical relevance . )
measuring three multipoles is sufficient to falsify whether the `` large central object '' is in fact an einstein black hole : if the third moment is inconsistent with the first two according to eq.([eq : kerr_multipoles ] ) then that object _ is not _ a black hole as described by einstein s theory of gravity . on the other hand ,
if all multipoles that we measure beyond the first two are consistent , this is very strong evidence that the `` large central object '' is in fact described by the kerr black hole solution .
mapping a planet s gravitational multipoles from satellite orbits is a well - established science called _
geodesy_. for example , consider a planet that is mostly spherical , with a small quadrupolar distortion .
orbits of this body will precess : the time to cover its full range of @xmath163 motion is incommensurate with the time it takes to move through the full range of @xmath164 . measuring this precession , we measure the quadrupolar distortion . mapping the multipoles of a black hole s spacetime with small body orbits
is very similar ; in recognition of this similarity , it has been suggested that the mapping of black hole spacetimes be called
_ bothrodesy _ @xcite .
a proof - of - principle analysis by ryan @xcite has shown that bothrodesy can be done with gravitational waves , showing that a massive body s multipole moments are encoded in the waves emitted as a small body spirals in .
his analysis is in somewhat crude , however .
he restricts the inspiraling body to circular , equatorial orbits , losing information about the body s multipoles that are encoded in an inclined orbit s precession .
instead , he `` weighs '' the multipoles by the fact that their radial dependence is different , and so affect the orbit at different rates as the body spirals inward . even with this excessively restricted setup , ryan finds that at least three and in some cases four or five multipoles can be distinguished in gravitational - wave data .
it is almost certain we will find that multipoles are measured even better when realistic orbits are used to generate the waves .
the black hole nature of a massive object influences the inspiral of a small body in other ways .
[ fig : inspiral_trajs ] shows the `` trajectories '' in radius @xmath109 and inclination angle @xmath165 followed by a @xmath166 body that spirals into a @xmath144 black hole .
these trajectories were computed by studying how the energy and angular momentum of the orbit evolve due to losses from gravitational wave emission ; see @xcite for details . in the top panels ,
gravitational waves carry energy and angular momentum both out to infinity and down the hole s event horizon . in the lower panels ,
the horizon flux is ignored .
when the hole is rapidly spinning , the horizon flux has an enormous influence on the inspiral , increasing the inspiral by several weeks and adding thousands of additional orbits . that the horizon flux
_ increases _ the inspiral duration is at first sight counterintuitive one would guess that the horizon is a sink for the orbit s energy , and so its influence should decrease the inspiral duration .
in fact , the hole _ supplies _ orbital energy : if it were not for the loss of energy in radiation carried to infinity , the body would spiral outward , rather than inward !
this surprising behavior can be understood as either a tidal interaction between the small body and the event horizon , or as due to radiation that superradiantly scatters from the ergosphere of the black hole ; see refs . for further discussion .
when the hole does not rotate rapidly , the effect is far smaller , and is more intuitively acceptable : orbital energy is lost to the black hole . rapidly rotating black holes , if supplied to us by nature ,
will be particularly interesting laboratories for probing strong field gravity .
black holes are purely vacuum solutions to einstein s equations .
other gravitational - wave sources contain dense matter concentrations .
in particular , neutron star sources are among the most studied for gravitational - wave detectors .
a typical neutron star consists of approximately @xmath167 of material with mean density @xmath168 grams per cubic centimeter ; such a star has a radius of 10 15 kilometers .
the properties of that matter can have a large influence on the gravitational waves that neutron star systems generate .
thus gravitational - wave measurements from such systems may be used to study highly dense matter .
just a few measurements are needed to strongly constrain the equation of state of neutron star matter @xcite . in this subsection
, we briefly describe some examples of how this may be done . when a neutron star is a member of a binary , the details of its nuclear matter structure are entirely irrelevant to gravitational - wave generation over much of the inspiral .
it is only when the members of the binary are close enough that their tidal gravitational fields begin to distort one another that neutron star structure becomes potentially important . for double neutron star binaries
, this means that the very late inspiral and merger waves ( when the neutron stars collide and fuse into a single body ) will depend quite strongly upon the properties of dense nuclear matter .
unfortunately , these waves are at very high frequencies ( @xmath169 hz ) where gravitational - wave detectors do not have good sensitivity .
special high frequency laser interferometers @xcite and acoustic detectors @xcite can be used to improve our ability to measure these waves @xcite .
better accuracy may be achieved by measuring waves from the coalescence of a black hole - neutron star binary .
if the black hole is more massive than roughly @xmath139 , the neutron star will essentially be swallowed whole by the black hole .
this is because the radii of the innermost orbits scale with the hole s mass , and because the tidal field of the black hole scales as @xmath170 .
hence when the hole is large , its tides are relatively gentle and the neutron star can survive inspiral all the way through the hole s event horizon . when the hole is smaller , however , the tides rip the neutron star apart before any matter falls down the horizon .
this drastically changes the binary s mass distribution and significantly affects its gravitational - wave emission . because a black hole - neutron star system is significantly more massive than a double neutron star system ,
the waves that depend upon the neutron star s matter are emitted at lower frequencies where laser interferometers have better sensitivity .
vallisneri @xcite has shown that planned upgrades to ligo will make it possible to extract a fair amount of information about neutron star properties by measuring such disruptions .
however , he cautions that better theoretical modeling of the waves characteristics will be needed to fully understand what the detectors measure .
neutron stars may undergo phase transitions as dense material changes state , perhaps forming a condensation of strange matter or free quark matter @xcite .
if this occurs , the star s density profile will rapidly change ; any gravitational waves that it might be emitting will rapidly change character as well .
this may be seen by following the waves that are emitted by an accreting neutron star @xcite : matter dumped onto the neutron star increases its mass beyond the critical point , until the phase transition occurs .
the star then changes radius rapidly , exciting radial vibrational modes in the star .
other work suggests that such a phase transition may be see in the waves that are emitted as a neutron star forms during supernova collapse @xcite , changing the character of the star s various oscillational modes .
finally , we note that a transition to such a form of nuclear matter has a drastic effect on the viscosity of material by changing the kinds of interactions that occur between nucleons in the star .
this may cause various instabilities , particularly that due to r - modes , to damp out rapidly @xcite .
gravitational - wave generation by neutron star systems present great opportunities for studying extremely dense matter , but as vallisneri has emphasized , many theoretical uncertainties must be hammered out before we will be able to do so well . finally , gravitational waves will test current ideas for new unification physics beyond the standard - model - plus - gr paradigm . the best models for fundamental unification , based on string theory , lead to the idea that gravity propagates in more than three spatial dimensions .
this has led to the investigation of `` brane worlds '' , in which standard model fields live on a 3d `` brane '' or boundary of a larger space ( `` bulk '' ) with one or more small or highly - curved extra dimensions @xcite . in these worlds
, gravitons can have massive `` kaluza - klein '' modes that might be produced in accelerators , or survive from the early universe as dark matter .
even the massless modes , classical gravitational waves , no longer necessarily obey the same causal structure as electromagnetic waves ; they can propagate either above or below the speed of light @xcite .
the possibility that a graviton might travel slower than a photon appears to be very tightly constrained by the fact that ultra high energy cosmic rays do not appear to lose energy to gravitational erenkov radiation @xcite .
however , in brane worlds , gravitons could also propagate faster than the speed of light
. one can think of the extra dimensions as a kind of waveguide for the gravitational waves , so that the propagation is nondispersive ; however , the curvature of the extra dimensions can also allow for faster propagation .
we have little empirical information about this possibility at present , since we know only that the right amount of gravitational wave power is being emitted from binary pulsars . with lisa we will have coherent streams of gravitational waves from identified compact white - dwarf binaries that also have optically measured orbital parameters and orbital phases .
the optical and gravitational wave data can be compared over many orbits , yielding a direct test of the propagation speed of gravitational waves that should be very precise .
these data will also constrain the possibility that the graviton is a composite particle , with components of more than one ( possibly nonzero , but very small ) mass .
classically emitted waves may , after propagating a large distance , separate out into two or more components .
we might see a `` standard '' waveform of reduced amplitude , corresponding to the massless mode , with the rest of the energy appearing in some other , more dispersive waveform ( or perhaps so highly delayed that it is not seen at all ) . while we have discussed these programs in the context of concrete brane world models , it is possible to motivate them in more general language , as tests of lorentz symmetry @xcite .
the proliferation of unification ideas certainly justifies such an empirical approach ; the detection of a cosmological constant has opened up consideration of an enormous range of mass scales so it has become interesting to constrain even what used to seem outrageously small masses .
cosmology provides another context in which gravitational wave detectors will constrain fundamental physics .
generally , the connection comes through the possibility of generating a stochastic background of gravitational waves in the early universe that can be detected today @xcite .
gravitational - wave observatories will have an interesting sensitivity to these backgrounds ; both ligo and lisa will be able to detect broad - band stochastic backgrounds with energy densities well below the microwave background @xcite .
the source most often considered for such a background is cosmological inflation . we now think that parametrically - amplified quantum fluctuations of the inflaton field are responsible for large scale structure in the universe , and also for most of the fluctuation power in the microwave background anisotropy .
the model predicts that similar fluctuations in the graviton field will produce a gravitational wave background .
it is possible that these tensor waves are already contributing to the detected microwave anisotropy on the largest scales , and indeed there are experiments to test this possibility using the different polarization signatures of the anisotropy produced by the two kinds of modes .
these experiments are clearly important , but it will not be surprising if the tensor component turns out to be negligible .
there is no reason for it to be comparable to the scalar ( inflaton produced ) component , as it depends on different parameters of the inflation model .
( indeed , the pure - scalar model with a scale - invariant spectrum fits the anisotropy data so well we would be disappointed to spoil it by adding another component with extra parameters . )
the likely outcome is that the polarization experiments will constrain the tensor component , thereby constraining directly the expansion rate during inflation .
even if the tensor component is detectable on the largest scales , the corresponding gravitational wave energy density for a scale - free spectrum ( about @xmath171 of the critical density at low frequencies , changing to @xmath171 of the microwave energy density at high frequencies ) is far too low for ligo or lisa to detect directly . in some models , where the expansion rate increased to nearly the planck scale during the course of inflation
, the spectrum can be tilted enough to produce a detectable background , but these models are already ruled out by millisecond pulsar timing ( which limits stochastic backgrounds to less than @xmath172 of the microwave background density at frequencies of the order of inverse years ) @xcite . therefore it seems unlikely that the quantum graviton background will be seen by these observatories .
some models however predict an entirely different source of primordial backgrounds .
after inflation , the early universe may have hosted violent classical events that gave rise to a stochastic background .
these gravitational waves could be much more intense than the inflationary quantum graviton production ; the maximal energy density , limited by equipartition arguments , is approximately the same as the microwave background energy density ten orders of magnitude above the typical inflationary level .
there are a number of circumstances which might approach within a few orders of magnitude of the maximal level .
for example , a strongly first - order phase transition would lead to bubble nucleation and relativistic flows on the hubble scale @xcite ; condensation of brane in a higher - dimensional space would lead to kibble excitation of tensor modes on the hubble scale @xcite .
strong effects of this kind are not required in any well established theory , but they are generic behaviors of relativistic systems that are worth constraining
. backgrounds from these events are not scale - free ; they tend to be broad band , but they have a characteristic frequency given by the redshifted hubble rate . for detection at lisa and ligo frequencies ,
the corresponding cosmic temperature turns out to be upwards of tev energies a range of interest for unification and supersymmetry breaking .
this is a period of cosmic history about which we have almost no empirical information at present , so even upper limits from gravitational wave observatories will convey new information about physics at these energies .
a detection would open up a new period of cosmic history to detailed scrutiny ; we might for example find much more detailed information about the process of cosmic baryogenesis .
the `` last scattering surface '' for gravitational waves is just their production epoch ( that is , there is almost no significant absorption or large - angle scattering since the planck time ) so they provide a direct window into the early universe much deeper than the microwave background , and down to much smaller comoving scales .
if a detectable stochastic background does exist , future generations of detectors may one day measure anisotropy in the background that contains new information about the early universe a directly - imaged view of events invisible ( even in principle ) in any other way .
sah is supported by nsf grant phy9907949 ; his travel to the snowmass meeting was in addition supported in part by the lisa project and the jet propulsion laboratory , california institute of technology under contract to nasa .
ligo is funded by the national science foundation under cooperative agreement phy9210038 .
this work was supported at the university of washington by nsf grant ast0098557 , and at the university of colorado by nasa grants nag54095 and nag510259 .
we would like to thank colleagues who helped us and gave excellent talks at snowmass 2001 p4.6 : keith riles , nergis mavalvala , rana adhikari of ligo .
david shoemaker kindly provided us with excellent ligo posters and support of our efforts , robin stebbins provided much of the lisa graphics used here , and kip thorne provided us with the figure of ligo sources .
we thank yuk tung liu for helpful comments on gravitational waves from accretion induced collapse of white dwarfs .
we appreciate the support , help and encouragement of barry barish , stan whitcomb , and tom prince .
gravitational - wave strain sensitivies are usually quoted with units @xmath173 .
this is because the squared strain noise in a band @xmath174 is given by @xmath175 ; the final strain noise is thus dimensionless . the noise spectrum @xmath176 is often denoted @xmath177 .
b. allen , in _ proceedings of the les hourches school on astrophysical sources of gravitational waves _ , edited by j .- a .
marck and j .- p .
lasota ( cambridge university press , cambridge , 1996 ) ; gr - qc/9504033 .
g. ushomirsky , l. bildsten , and c. cutler , in _ gravitational waves : third edoardo amaldi conference _ , edited by sydney meshkov , aip conference proceedings no .
523 ( american institute of physics , melville , new york , 2000 ) , p. 65 .
l. bender , _ lisa : a proposed joint esa - nasa gravitational wave mission _ , in gravitational waves , edited by i.
ciufolini , v. gorini , v. moschella , and p. fre ( institute of physics publishing , bristol , uk ) , 115151 ( 2001 ) .
the term `` bothrodesy '' has been coined by sterl phinney .
it descends from the use of the suffix `` -bothron '' introduced by brandon carter to describe properties of black holes .
this suffix , in turn , descends from the greek word `` @xmath178 '' , meaning `` garbage dump '' .
s. a. hughes , in preparation ; preliminary version available as chapter 4 of s. a. hughes , `` gravitational - wave astronomy : aspects of the theory of binary sources and interferometric detectors '' , unpublished ph .
d. thesis , california institute of technology ( 1998 ) . | gravitational - wave detectors with sensitivities sufficient to measure the radiation from astrophysical sources are rapidly coming into existence . by the end of this decade
, there will exist several ground - based instruments in north america , europe , and japan , and the joint american - european space - based antenna lisa should be either approaching orbit or in final commissioning in preparation for launch .
the goal of these instruments will be to open the field of _ gravitational - wave astronomy _ : using gravitational radiation as an observational window on astrophysics and the universe . in this article , we summarize the current status of the various detectors currently being developed , as well as future plans . we also discuss the scientific reach of these instruments , outlining what gravitational - wave astronomy is likely to teach us about the universe . |
mixtures of nonadsorbing polymers and colloidal particles often exhibit rich phase behavior @xcite . for certain polymer - to - colloid size ratios @xmath0 ( where @xmath1 is the radius of gyration of the polymer and @xmath2 is the radius of colloid ) entropy - driven effective interactions may lead to stable colloidal gas , liquid and solid phases even if all bare interactions are purely repulsive @xcite .
a simple theoretical model giving an insight into this phenomenon is the asakura - oosawa - vrij ( aov ) model of colloid - polymer mixtures in which the polymers ( modeled as spheres ) are ideal and can overlap freely , but the polymer - colloid and colloid - colloid interactions are of the hard sphere type @xcite .
effective attractive interactions in such systems arise due to a tendency to decrease the volume excluded to the polymer coils @xcite . the aov model has attracted much attention due to its simplicity @xcite , however when comparing to experiments the agreement is only qualitative .
this is due to the fact that the real polymers are nonideal .
the incorporation of the polymer nonideality can be tackled at the monomer - resolved @xcite or coarse - grained @xcite levels of description .
the latter technique is particularly useful in the so - called `` colloid limit '' ( @xmath3 ) , in which the full mixture is well described by invoking the effective , pairwise depletion potentials @xcite . however ,
when the polymer dimensions are larger than the size of the colloid particles , the resulting two - body effective potentials may be insufficient to correctly describe the underlying mixture and the incorporation of the higher - order , many - body terms becomes necessary @xcite .
in addition to many industrial applications colloid - polymer mixtures offer a convenient tool to study important fundamental concepts that are often difficult to investigate in simple fluids , such as real space observation of the thermal capillary waves @xcite , the capillary length @xcite , the interfacial width @xcite and the interfacial tension of the fluid - fluid interface @xcite . wijting _
et al_. @xcite studied the behavior of colloid - polymer mixtures close to a wall .
they found a wetting transition to a state in which the colloid - rich phase wets completely a nonadsorbing planar hard wall .
these findings qualitatively agree with the results of density functional theory @xcite and computer simulations @xcite carried out for the aov model .
et al_. @xcite used the cahn - fisher - nakanishi approach to study the wetting transition in mixtures of colloids and excluded volume polymers and pointed out some subtleties associated with the precise measurements of the contact angle @xcite .
recently paricaud _
et al_. @xcite examined the bulk phase behavior of mixtures of hard - sphere colloids and excluded - volume polymers within the framework of wertheim s first - order thermodynamic perturbation theory ( tpt1 ) @xcite .
they found a demixing transition into the colloid - rich and polymer - rich phases provided that the ratio of the colloid diameter to polymer segment diameter is large enough .
the tpt1 description employs the microscopic description of the polymer , consequently it is treated on equal footing with the colloid particles .
this theory provides thus a straightforward procedure of taking the attractive / repulsive interactions into account .
however , the wertheim tpt1 approach is not free of its own deficiencies .
we note here that incorporation of the solid - fluid equilibrium into the framework , while in principle possible @xcite , is difficult to execute for the colloid - polymer mixtures
. moreover , boublik _
al_.@xcite pointed out that within the tpt1 theory the second virial coefficient scales quadratically with the chain length @xmath4 instead of @xmath5 , although it is possible to improve this deficiency @xcite .
finally , the value of the flory exponent , @xmath6 , resulting from the tpt1 approach is 0.5 rather than 0.588 @xcite . despite these shortcomings ,
the tpt1 approach to colloid - polymer mixtures provides a welcome departure from the well - studied aov model . in this work we focus on the surface phase behavior of model colloid - polymer mixtures .
we employ nonlocal density functional theory that was used previously to investigate liquid - liquid interfaces of athermal mixtures of colloids and excluded - volume polymers @xcite .
the density functional framework , proposed originally by yu and wu @xcite , uses the tpt1 ideas and yields in the limit of bulk systems an equation of state identical to that from ref . @xcite .
since the dft framework has already been described in detail @xcite we recall only the basic points of this approach .
bulk phase diagram for a colloid polymer mixture with @xmath7 and @xmath8 .
the main figure shows the binodal ( solid line ) , the spinodal ( dashed line ) , and the bulk critical point ( open circle ) evaluated in the polymer segment mole fraction - total density plane .
the open circle denotes the bulk critical point .
the inset shows the binodals resulting from the tpt1 ( solid line ) and free volume ( dash - dotted line ) theories evaluated in the colloid - polymer packing fraction representation .
, width=302 ]
we model the colloids as hard spheres of diameter @xmath9 and the polymers as chains composed from @xmath4 tangentially bonded hard - sphere segments of diameter @xmath10 .
the hard - sphere monomers forming the chains are freely jointed i.e. they can adopt any configuration as long as it is free of the intermolecular and intramolecular overlap . within the dft approach the grand potential of the system
, @xmath11 is a functional of the local densities of polymers , @xmath12 and colloids , @xmath13 @xmath14=
f[\rho_p({\bf r}),\rho_{c}({\bf r})]+}\nonumber\\ & & \int\!\!d{\bf r}\rho_{p}({\bf r})(v_{ext}^{(p)}({\bf r})-\mu_{p})\nonumber\\ & + & \int\!\!d{\bf r}\rho_{c}({\bf r } ) ( v_{ext}^{(c)}({\bf r})-\mu_{c})\;.\end{aligned}\ ] ] in the above @xmath15 , @xmath16 , @xmath17 and @xmath18 are the external and the chemical potentials for colloids and polymers , respectively .
@xmath19 denotes a set of monomer coordinates .
the excess polymer segment adsorption isotherm @xmath20 for a colloid - polymer mixture with @xmath7 and @xmath8 , calculated along path i from fig . [ fig:1 ] ( upper panel ) and along path ii from fig .
[ fig:1 ] ( lower panel ) .
the dotted lines denote bulk coexistence .
, width=302 ] the free energy of the system @xmath21 is a sum of the ideal and excess contributions , @xmath22 .
the ideal part of the free energy is known exactly , @xmath23=
\beta\int\!\!d{\bf r}\rho_{p}({\bf r})v_{b}({\bf r } ) + \int\!\!d{\bf r}\rho_{p}({\bf r})[\ln(\rho_{p}({\bf r}))-1 ] + \int\!\!d{\bf r } \rho_{c}({\bf r})[\ln(\rho_{c}({\bf r}))-1]$ ] . the total bonding potential , @xmath24 , is represented as a sum of the bonding potentials @xmath25 between the monomers , @xmath26 , and satisfies @xmath27\propto \prod_{i=1}^{m-1}\delta(|{\bf r}_{i+1}-{\bf r}_{i}|-\sigma_{ps})$ ] .
furthermore , it is assumed that the excess free energy is a functional of the local density of colloids and average segment local density defined as @xmath28 .
the excess polymer segment adsorption isotherm @xmath20 for a colloid - polymer mixture with @xmath7 and @xmath8 , calculated along a path of constant polymer segment packing fraction @xmath290.944 727 94 .
the dashed line marks the prewetting transition , while the dotted line denotes bulk coexistence .
the inset illustrates the nature of the divergence of the excess adsorption ( open circles ) upon approaching bulk coexistence .
@xmath20 increases linearly as a function of the logarithm of the undersaturation @xmath30 .
the straight solid line is a guide to the eye .
, width=302 ] within the approach of yu and wu @xcite the excess free energy is a volume integral over the excess free energy density , @xmath31 where @xmath32 is the excess free energy density of the reference mixture of hard spheres and @xmath33 is a perturbation contribution due to the chain connectivity .
@xmath32 is evaluated using the elegant and inspiring white bear version @xcite of rosenfeld s fundamental measure theory ( fmt ) @xcite .
likewise , fmt - style weighted densities are used for the polymer contribution , @xmath33 .
we refer the reader to earlier papers for explicit formulae @xcite .
the equilibrium density profiles were found from the condition @xmath34 } { \delta \rho_p(\mathbf{r})}= \frac{\delta\omega [ \rho_p({\bf r}),\rho_{c}({\bf r } ) ] } { \delta \rho_c({\bf r})}=0 \;.\ ] ]
paricaud _ et al_. @xcite shown that within the wertheim approach a mixture of colloids and excluded volume polymers undergoes a demixing transition into the colloid - rich ( polymer - poor ) and colloid - poor ( polymer - rich ) phases .
the properties of the fluid - fluid interface resulting from this theory were reported in ref .
@xcite . in the present paper
we investigate the wettability of hard structureless walls by such mixtures .
the calculations were carried out for constant size ratio @xmath35 and for chain lengths @xmath36 , 100 , 120 and 200 .
this corresponds to @xmath37 , 1 , 1.095 and 1.414 , and in the above we assumed that @xmath38 with @xmath39 and @xmath40@xcite .
thus we consider both regimes @xmath41 and @xmath42 . in fig .
[ fig:1 ] we recall the bulk phase diagram for the system with @xmath8 plotted in the polymer segment mole fraction , @xmath43 , - total density , @xmath44 , plane .
this representation proved to be useful in carrying out numerical calculations presented below .
if we disregarded the chain connectivity @xmath45 would correspond to the mole fraction of the small spheres in a mixture of big and small spheres , while @xmath46 and @xmath47 would correspond to the bulk number densities of small and big hard spheres , respectively . the bulk densities @xmath46 and @xmath47 serve as an input to the dft calculations .
we note that the critical polymer segment mole fraction for the system depicted in fig .
[ fig:1 ] is 0.976 763 34 , thus subcritical state points with lower mole fractions correspond to the polymer - poor ( colloid - rich ) side of the phase diagram while state points with @xmath45 higher than this value correspond to the the polymer - rich ( colloid - poor ) side .
the numerical procedure of the evaluation of the surface phase diagrams relied on monitoring excess adsorption isotherms , @xmath48 , @xmath49c , ps , and grand potentials ( eq . [ eq:1 ] ) calculated along paths of constant total density ( an example of such path labelled as `` i '' is shown in fig . [ fig:1 ] ) , or along paths of constant polymer segment mole fraction ( the paths marked as `` ii '' and `` iii '' in fig .
[ fig:1 ] ) .
the inset to fig .
[ fig:1 ] shows a comparison of the binodal evaluated for @xmath8 resulting from the werthiem tpt1 description with the binodal resulting from the free - volume theory for the aov model @xcite for @xmath50 .
the diagrams are plotted in the colloid - polymer packing fraction representation . for such comparison to be possible one has to transform the microscopic polymer segment density @xmath46 into the polymer packing fraction defined as @xmath51 . when compared with the aov model the wertheim tpt1 approach leads to a shift of the colloid and polymer critical packing fractions similar to the perturbative treatment of warren _
et al_. @xcite and the geometrical - based approach @xcite .
we note that the polymer packing fractions are too low when compared with experimental values @xcite .
this is in part due to the fact that , as mentioned earlier , the wertheim tpt1 theory leads to the incorrect flory exponent ( @xmath40 rather than @xmath52 ) .
we start the presentation of the results of dft calculations by analyzing the colloid - poor side of the bulk phase diagram .
the upper panel of fig .
[ fig:2 ] shows the excess polymer segment adsorption isotherm evaluated along path i , i.e. along a path of constant total density .
we find the adsorption isotherm to be smooth and small in magnitude .
we have carefully inspected adsorption isotherms calculated along paths similar to the path i for the system with @xmath8 and other chain lengths and have always found similar behavior .
therefore we conclude that within the tpt1 formalism mixtures with the state points located on the colloid - poor side of the phase diagram always stays in the partial wetting / drying regime .
this is the opposite of what is found in the wetting studies of the aov model @xcite .
average polymer segment ( a ) and colloid ( b ) density profiles for a colloid - polymer mixture with @xmath7 and @xmath8 near a hard wall .
the profiles are evaluated for constant polymer segment packing fraction @xmath290.944 727 94 and for total bulk densities @xmath53 0.012 3625 ( the dashed and solid lines correspond to the coexisting thin and thick film phases of the prewetting transition , respectively ) , 0.0126 ( dotted lines ) , and 0.012 617 ( dash - dotted lines ) .
, width=302 ] we turn now to the wetting properties of the state points located on the polymer - poor side of the phase diagram .
the lower panel of fig .
[ fig:2 ] shows the excess polymer segment adsorption isotherm calculated along path ii , i.e. along a path of constant polymer segment mole fraction @xmath290.928 889 85 .
similar to the path i we find the adsorption isotherm to be smooth , small in magnitude , and finite at bulk coexistence .
however , when paths located closer to the bulk critical point are considered , a drastic change in the behavior of the inhomogeneous colloid - polymer mixture is observed .
this is illustrated in fig .
[ fig:3 ] , where we show the adsorption isotherm , calculated along path iii i.e. along a path of constant @xmath290.944 727 94 . at lower total bulk densities
the excess adsorption is very similar to that evaluated along path ii .
however , as @xmath54 increases the adsorption isotherm makes a sudden jump ( marked by the dashed line ) to a large positive value and diverges upon approaching bulk coexistence ( denoted by the vertical dotted line ) .
the nature of this divergence is investigated further in the inset to fig .
[ fig:3 ] .
we find that the excess adsorption diverges logarithmically and this behavior is characteristic of complete wetting for short range forces @xcite .
we therefore conclude that in the tpt1 description the colloid - poor demixed phase wets completely the hard wall at coexistence .
the wetting transition is of the first order and is accompanied by the prewetting ( or thin - thick film ) transition .
morphology of the demixed mixture close to the hard wall is investigated further in fig .
[ fig:4 ] where we show examples of the density profiles evaluated at fixed @xmath290.944 727 94 and for several total bulk densities .
the dashed and solid lines correspond to the coexisting thin and thick film profiles ( these states are marked as black dots in fig . [ fig:3 ] ) . the polymer segment profile ( cf .
[ fig:4]a , dashed line ) corresponding to the thin film phase exhibits oscillations in the direct proximity of the wall , and the oscillation period approximately equals the colloid diameter .
however the excess adsorption is small in magnitude and negative .
the colloid profile ( cf .
[ fig:4]b , dashed line ) also exhibits oscillations but the structure is much more pronounced and the profile exhibits a sharp and high first peak with a large contact value .
this suggests that in the thin film phase the polymers are depleted from the wall , whereas the colloids are attracted to the wall and form a highly packed first layer . on the other hand , in the thick film phase of the polymer segment density profile ( cf .
[ fig:4]a , solid line ) one observes formation of a thick polymer layer next to the wall . as the total density increases ( cf .
[ fig:4]a , dotted and dash - dotted lines ) the polymer - rich layer grows and extends over the distance of many colloid diameters .
the structure of the colloids in the thick film phase is a reverse of the structure of the polymer .
the contact value of the colloid profile ( see the inset to fig . [
fig:4]b ) is reduced by more than a half at the prewetting transition @xcite suggesting that the colloids are depleted from the region adjoining the wall .
the colloid - poor region broadens upon approaching bulk coexistence ( fig .
[ fig:4]b , dotted and dash - dotted lines ) . surface phase diagram for a colloid polymer mixture with @xmath7 and @xmath8 in the polymer segment mole fraction - total density plane .
the binodal and the spinodal are denoted by the solid and dashed lines , respectively .
the open circle denotes the bulk critical point .
the filled circle denotes the wetting point whereas the triangle denotes the surface critical point .
thick solid line denotes the prewetting line .
the inset shows the surface phase diagram in the colloid - polymer packing fraction representation .
, width=302 ] figures [ fig:5 ] and [ fig:6 ] show the surface phase diagram for colloid - polymer mixtures with @xmath8 and @xmath55 , respectively .
as in simple fluids the prewetting line tangentially meets the binodal in the wetting transition point ( marked by the filled circle ) and ends at the surface critical point ( marked by the triangle ) . in the polymer segment mole fraction - total density representation
the prewetting line is located very close to the binodal and the surface critical point is well removed from the bulk critical point ( marked by the open circle ) . however , when the colloid - polymer packing fraction ( cf .
the inset to fig .
[ fig:5 ] ) or the polymer segment mole fraction - pressure representation is chosen ( cf .
the inset to fig .
[ fig:6 ] ) , the prewetting line is more distant from the binodal .
we note that for all chain lengths considered the surface critical point pressure is lower than the pressure of the bulk demixing transition .
this curious feature was also found in a recent study of the wetting behavior of polymer solutions @xcite .
however in the present paper we do not find the lower wetting points and , despite considerable effort , we do not find any hints for the layering transitions in the partial wetting regime .
a likely explanation to the last observation is that forsman and woodward @xcite considered a mixture of polymers and solvent molecules with diameter ratio @xmath56 .
consequently , the individual polymer segments were effectively attracted to the wall due to the depletion forces induced by the overlap of the excluded volumes of the polymer segments and the wall . in the present study this effect does not come into play .
surface phase diagram for a colloid polymer mixture with @xmath7 and @xmath55 in the polymer segment mole fraction - total density plane .
the binodal and the spinodal are denoted by the solid and dashed lines , respectively .
the open circle denotes the bulk critical point .
the filled circle denotes the wetting point whereas the triangle denotes the surface critical point .
thick solid line denotes the prewetting line .
the inset shows the surface phase diagram in the polymer segment mole fraction - pressure representation .
, width=302 ] in a recent study of wettability of solid surfaces by short - chain fluids @xcite it was suggested that the wetting temperature for a given surface , when divided by the critical temperature , is almost independent on @xmath4 for longer chains .
although in the athermal systems considered in the present paper the temperature does not have any significant role ( is a simple scaling parameter ) , other parameters like the reciprocal pressure play the role similar to the temperature in thermal systems . therefore the wetting point pressure rescaled by the critical pressure @xmath57 should barely depend on the chain length .
the estimated wetting point pressure @xmath58 for systems @xmath36 , 100 , 120 , and 200 is 9.75 , 7.76 , 6.51 , and 4.25 , respectively .
the corresponding @xmath59 thus are 0.814 , 0.818 , 0.824 , and 0.835 .
this agrees nicely with the prediction @xcite .
that the rescaled @xmath59 barely depends on @xmath4 has a simple explanation . as demonstrated in refs .
@xcite , in the limit of an infinite chain length @xmath60 , and for fixed @xmath61 , the bulk critical pressure ( and also the colloid critical packing fraction ) tend to a finite , nonzero value with the leading order correction @xmath62 .
this is the so - called `` protein limit '' of colloid - polymer mixtures @xcite .
but then the wetting point pressure also should tend to a constant in that limit .
therefore the rescaled wetting point pressure should barely depend on m for longer chains .
we also note that the systematic deviations of @xmath59 from a constant suggest that the leading order chain - length dependent correction to @xmath63 is of different order than that for the bulk critical point .
however the precise form of these corrections is rather difficult to estimate and we leave this problem for future work .
wetting in the aov model for @xmath41 can be readily explained by invoking depletion interactions .
the overlap of excluded volumes of a colloid and a wall is larger than the overlap between two colloidal particles and this gives rise to net colloid - wall attraction .
consequently the colloid - rich phase favors the hard wall .
our results are in opposition to the aov model results and show some features that are characteristic of wetting in athermal polymer solutions .
the suggestion that the polymer - rich phase should wet the hard wall seems reasonable when @xmath42 .
however , we find this behavior also for @xmath41 i.e. for @xmath7 , @xmath36 ( in order to be completely assured that our calculations are representative of the tpt1 description and are not a result of e.g. an artifact connected with the incorrect scaling of the polymer gyration radius , we checked that even for colloid - to - polymer - segment size ratios as large as @xmath64 and for chains as short as @xmath65 the polymer - rich phase wets completely the hard wall ) .
this is at variance with experimental results .
we connect this discrepancy with the inaccurate description of the bulk phase behavior resulting from the tpt1 theory , and identify two possible sources of errors .
the first issue is connected with the incorrect scaling of the gyration radius and of the polymer second virial coefficient with @xmath4 .
the consequence is that the dilute and semidilute regimes of long chains are not described accurately .
this problem has already been recognized in the literature .
proposed a simple correction that significantly improves the accuracy of the tpt1 description in the low density limit @xcite .
a similar treatment , which combines the field - theoretic approach with liquid - state theory , was proposed by lue @xcite . however ,
at present it is not clear how these ideas can be extended to mixtures . to explain the second deficiency in the tpt1 description of colloid - polymer mixtures we recall that the wertheim approach , as any other perturbation theory , relies on the good description of the reference system . in this case
it is a mixture of big and small hard spheres and we used the usual boublik - mansoori - carnahan - starling - leland ( bmcsl ) equation of state @xcite .
however , it was argued @xcite that for highly asymmetric hard sphere mixtures a fluid - fluid phase separation occurs , which is metastable to a broad fluid - solid phase transition .
consequently the bmcsl equation of state is not very accurate .
the reference system in the tpt1 approach to colloid - polymer mixtures consists of a very low packing fraction of small spheres , therefore the potential impact of this problem seems to be moderate .
we have used a microscopic density functional theory based on wertheim s first order thermodynamic perturbation theory to study the wetting behavior of athermal mixtures of colloids and excluded - volume polymers . in opposition to the wetting behavior in the aov model
we have found the polymer - rich phase to wet a hard wall .
the wetting transition is of first order and is accompanied by the prewetting transition .
we have not found any hints for the layering transitions in the partial wetting regime .
the rescaled wetting point pressure is very similar for mixtures with different chain lengths and the surface critical point pressure is lower than the pressure of the bulk demixing transition .
we have found our results to resemble the wetting behavior in athermal polymer solutions .
our considerations show that it is certainly a challenge to devise a monomer - resolved theory for colloid - polymer mixtures that is accurate in both the colloid , and protein limits
. however , we feel that the wertheim approach could be a good candidate but this theory should incorporate the correct scaling behavior in the dilute polymer regime and an accurate description of the reference system . | we use a microscopic density functional theory based on wertheim s first order thermodynamic perturbation theory to study wetting behavior of athermal mixtures of colloids and excluded - volume polymers . in opposition to the wetting behavior of the asakura - oosawa - vrij model we find the polymer - rich phase to wet a hard wall .
the wetting transition is of the first order and is accompanied by the prewetting transition .
we do not find any hints for the layering transitions in the partial wetting regime .
our results resemble the wetting behavior in athermal polymer solutions .
we point out that an accurate , monomer - resolved theory for colloid - polymer mixtures should incorporate the correct scaling behavior in the dilute polymer regime and an accurate description of the reference system . |
quantum information processing can potentially provide a considerable speedup over classical information processing for certain problems @xcite along with the ability to efficiently simulate physical systems that can not be done classically @xcite . in view of this , much work has been done on finding a viable physical implementation of a quantum computer and many different realizations have been proposed @xcite , some of which explored experimentally on a small scale @xcite .
most large scale architectures rely on protocols enabling the transport of quantum systems or distributed schemes to perform logical gates on isolated qubits .
such schemes make use of ` flying ' qubits @xcite or more generally of quantum bus concepts @xcite .
decoherence effects on the bus are crucial and being able to overcome them will bring us one step closer to true scalability .
most of the results in quantum information theory were developed in a discrete setting , making use of qubits .
however even though single qubit operations are not so much of an issue , the level of control required to physically implement entangling gates between individual qubits and measure them within realistic coherence times is tremendous , limiting the experimental realizations .
the initial theoretical proposals were rapidly adapted to a continuous variable ( cv ) setting , where cv quantum information processing was shown to be possible @xcite .
cv implementations may be more accessible in some respects , with simple measurements and entangling operations . despite these advantages , this framework
is limited by the nonlinearity available experimentally , making single system operations difficult . combining the exactness of discrete variables and
the robustness of continuous variables is therefore a judicious route to take .
the term ` hybrid ' was first coined by lloyd @xcite to describe quantum information processes featuring both discrete and continuous variables .
the ability to switch on and off particular interaction hamiltonians enables one to simulate interactions and quantum logical gates on discrete systems .
multi - qubit extensions were rapidly undertaken in generalizations @xcite .
the initial observation came at the same time as milburn s proposal to simulate interactions between trapped ions by coupling them to a common vibrational mode @xcite , constituting a direct physical realization of a hybrid quantum computer .
this approach was used to entangle up to four ions experimentally @xcite . in recent years other physical implementations
have been explored making use of matter qubits @xcite .
a hybrid quantum computer could potentially be more versatile than its strictly discrete counterpart , providing simple algorithms to compute eigenvectors and eigenvalues @xcite or implementing grover s search algorithm in a direct way @xcite . a cv bus ( qubus )
can also be used to achieve quantum non - demolition ( qnd ) measurements and parity gates @xcite , which for example can be combined with an ancilla qubit to implement a near deterministic cnot gate @xcite . in these cases the limited strength of the nonlinearity is compensated for by an increase in the bus amplitude , leading to entangling gates based on homodyne measurements functioning at greater success probabilities than in single photon mediated applications .
this is particularly relevant to the generation of cluster states @xcite and entanglement distribution for repeater applications @xcite .
the physical circumstances in which one can envisage a hybrid coupling between a cv and discrete variable system have been extensively investigated .
the jaynes - cummings model @xcite is very successful at accounting for the interaction of radiation with an atom in a cavity quantum electro - dynamics ( cqed ) setting .
based on this , one possible realization for the qubus scheme is that of an atomic qubit interacting dispersively with a cavity mode .
however superconducting charge qubits are also a good candidate for a physical realization @xcite , for which a dispersive coupling with a microwave bus mode has already been experimentally demonstrated @xcite .
decoherence effects during such an interaction have been explored in the past , for example in the case of a two - level atom interacting dispersively with an optical mode in a dissipative cavity @xcite .
this dispersive interaction forms the basis for many qubus schemes and dissipation effects during the interaction on an entangling gate between two qubits can potentially be overcome @xcite .
a symmetrization technique to develop resilience to both dissipation and thermal fluctuations was also proposed @xcite . there
the authors took advantage of the invariance under time reversal of the action of the gate , noticing that the combination of an interaction sequence with its time reversed version canceled out dissipation effects . even though they are effective , these methods focus on a pair of qubits within the same cavity or trap and decoherence due to inter - cavity communication remains to be addressed
in addition to this they propose interaction sequences which have to be iterated many times before any significant improvement in gate fidelity can be appreciated .
the cross - kerr effect is also well known and used in several different areas of optics . in this context
, decoherence effects have been approximated along with their impact on optical quantum information processing using weak nonlinearities @xcite .
it was shown that dissipation effects in a two - qubit parity gate can be minimized if one can implement a photon number measurement on the bus . in the present work we extend previous results by solving exactly the master equation during the interaction for arbitrary input states and observing a trade off between the entanglement generated and the required precision in interaction time .
we also provide the resulting quantum operations incurred by qubits involved in a cz gate assuming sequential interactions .
these operations are critical when quantum error correction techniques have to be introduced as the system is scaled up .
we note here that direct decoherence effects incured by the qubits , independently from the bus , are being neglected . in so doing we obtain a simplified picture , which can become a good approximation if bus dissipation is the primary source of noise .
the eventual optimization of a full architecture will nonetheless need to factor in additional sources of noise .
the paper is structured as follows . in the first section of this paper we derive closed expressions for the effects of dissipation during cross - phase and dispersive jaynes - cummings interactions .
we follow the entanglement and coherence dynamics of a qubit and the continuous variable as they interact with each other . in section
ii we carry our attention over to the hybrid gates themselves starting with the conditional displacement gate followed by the cz gate in section iii . by adding dissipation between interactions we obtain the quantum operations undergone by the qubits and discuss gate fidelity .
finally we provide a simple iteration scheme to simplify the operation down to a perfect cz gate followed by independent single qubit dephasing .
as explained in the introduction we focus here on two types of interactions .
the first consists in the cross phase modulation undergone by probe and signal modes as they travel through a kerr medium .
the second type of interaction we consider takes place between an off - resonant optical probe mode and a single atom in a cavity .
their effects are described by interaction hamiltonians of the form @xmath0 where @xmath1(@xmath2 ) are the annihilation ( creation ) operators acting on the probe mode . in the case of a cross - kerr interaction , @xmath3 the number operator acting on the signal mode ( @xmath4 , the states @xmath5 corresponding to the foch state basis ) and @xmath6 is proportional to the third order nonlinear susceptibility of the medium . for the cqed
setting , the interaction is described by the dispersive limit of the jaynes - cummings hamiltonian @xcite in which @xmath7 where @xmath8 is the pauli operator acting on the atomic qubit ( @xmath9 , @xmath10 ) and @xmath6 is the atom - light coupling strength @xcite . in both the cross phase and the cqed settings
we have a hybrid interaction between the continuous quadrature variables of the probe field and the discrete degrees of freedom of the subsystem . initiating the probe in a coherent state @xmath11 and applying the interaction for a time @xmath12 yields @xmath13 [ ] and the continuous variable prepared in the coherent state @xmath11 . ]
given these interaction hamiltonians , we can evaluate the effects of dissipation in the probe mode during the interaction ( see fig .
1 ) by solving the optical linblad master equation @xcite @xmath14+\gamma(2a\rho(t ) a^{\dag}-a^{\dag}a\rho(t)-\rho(t ) a^{\dag}a ) \label{}\ ] ] where we have assumed a zero temperature bath ( a good approximation in the visible light regime ) .
the damping factor @xmath15 quantifies the dissipation in the probe mode .
if we consider a general input density matrix element @xmath16 in which the probe and the signal are disentangled , we can find the equation of motion for this particular element by looking at the operator @xmath17 . due to the disentangled form of the initial state we have @xmath18 for all @xmath19 and @xmath20 .
the equation of motion for each element is given by @xmath21 following the method used in @xcite , we use the super - operators @xmath22 , @xmath23 and @xmath24 to rewrite the above equation as @xmath25 the formal solution to ( 5 ) is then @xmath26 .
the super - operators realize an algebra obeying the commutation relations @xmath27=\left[\mathcal{j},\mathcal{p}\right]=\mathcal{j}$ ] for which decomposition theorems have been derived @xcite , leading to @xmath28 & = & \mathrm{exp}\left[\frac{2\gamma(e^{(2\gamma - i(\lambda_n-\lambda_m)\chi)t}-1)}{2\gamma-(\lambda_n-\lambda_m)\chi}\mathcal{j}\right ] \nonumber\\ & \times & \mathrm{exp}\left[(i\lambda_n\chi-\gamma)\mathcal{m}t\right ] \nonumber\\ & \times & \mathrm{exp}\left[(-i\lambda_m\chi-\gamma)\mathcal{p}t\right ] .
\label{}\end{aligned}\ ] ] now applying this result to our initial element @xmath18 we obtain : @xmath29 \nonumber \\ &
\times & |\alpha e^{(-\gamma+i\lambda_n\chi ) t}\rangle\langle\alpha e^{(-\gamma+i\lambda_m\chi ) t}| .
\label{}\end{aligned}\ ] ] the coefficient derived above is the closed expression for the ` coherence parameter ' given in @xcite .
if we denote this coefficient by @xmath30 then we have @xmath31 and so @xmath32 as well as @xmath33 .
we also quickly notice from ( 7 ) that this coherence parameter does not tend to 0 as @xmath12 tends to infinity for a fixed @xmath34 .
we have @xmath35 .
\label{}\ ] ]
one way of understanding this is that the probe undergoes loss as it couples to the signal , thus reducing the coherence in the signal .
but eventually the probe returns to the vacuum state ( the time it takes depends on the initial amplitude and the damping factor ) , disentangling itself and leaving some coherence in the signal .
however the larger the amplitude , the larger the effective interaction time , the less coherence remains in the signal .
as time progresses , the process can be viewed as a unitary operation between the signal and the damped probe in addition to a dephasing effect on the signal . to view this more clearly ,
let us write the output density matrix using @xmath7 , @xmath36 and defining @xmath37 , @xmath38 writing @xmath39 , the factor @xmath40}$ ] characterizes the decoherence and takes the form @xmath41 ( see the appendix for the full expression ) . applying this type of operation to a qubit density matrix @xmath42 yields
directly the phase flip channel @xcite @xmath43}\rho & = & e^{-\epsilon(1-z_nz_m)}\rho \nonumber\\ & = & e^{-\epsilon}(\mathrm{cosh}\epsilon+z_nz_m\mathrm{sinh}\epsilon)\rho \nonumber\\ & = & \frac{1+e^{-2\epsilon}}{2}\rho+\frac{1-e^{-2\epsilon}}{2}z\rho z. \label{}\end{aligned}\ ] ] the additional phase @xmath44}$ ] acquired in the process ( see appendix ) is known and can be corrected for if needed ; it is not an intrinsic source of noise .
one expects this issue of coherence to be intimately linked to the entanglement shared between the signal and the probe systems . in order to observe the dynamics of this entanglement we restrict the signal to being a qubit and
will continue to use @xmath7 .
we also take @xmath34 real for simplicity . equation ( 7 )
provides us with a time dependent density matrix and having our input signal in the state @xmath45 , it reads @xmath46 with @xmath47 and @xmath48 .
the entanglement being invariant under local unitary operations we allow ourselves for simplicity to apply the conditional phase @xmath49}|1\rangle\langle 1|$ ] on the qubit .
then we redefine the bus probe states as @xmath50 and @xmath51}|\alpha_1\rangle$ ] so that the overlap between the two is real : @xmath52 .
this allows us to express them in an orthogonal basis @xmath53 as @xcite @xmath54 taking @xmath1 and @xmath55 real without loss of generality , normalization leads to @xmath56 with @xmath57 . at this point we can write our locally equivalent density matrix in the orthonormal basis @xmath58 as follows @xmath59 we have now managed to express the qubit and continuous variable composite state in the form of a two qubit state .
given the resulting two - qubit density matrix , there are several entanglement measures to choose from , including the logarithmic negativity and the relative entropy of entanglement @xcite . here
we will work with the concurrence as defined by wooters @xcite , which we plot as a function of the scaled time @xmath60 for particular choices of parameters @xmath34 and @xmath61 .
as the qubit and the field initially start in a product state , and eventually for large times should return to a product state when the probe field does nt contain anymore photons , we would expect the entanglement to peak at some point in time .
[ fig2 ] ( solid ) and the von neumann entropy @xmath62 ( dashed ) of the combined state of the continuous variable mode and qubit as a function of the scaled time @xmath60 .
( a ) from left to right @xmath34=200,100 and 50 with a fixed ratio of damping rate to nonlinearity @xmath63 .
( b ) the amplitude @xmath34 is fixed to 100 and from lowest to highest peaking curves @xmath61=1 , 7 and 21 . ]
this is verified in the plots of fig .
2 . in fig .
2(a ) we can observe how the amplitude of the field @xmath34 affects the behavior of entanglement in time .
as expected the larger @xmath34 is , the larger the maximum entanglement .
this is simply explained by the fact that the separation between the possible states of the field in phase space increases with @xmath34 , for a same interaction strength , thus making them more distinguishable . for large @xmath34
the maximum concurrence tends naturally to 1 , however the peaking of the entanglement also becomes sharper .
it is quickly generated , but also quickly destroyed , as illustrated by the von neumann entropy @xmath64 characterizing the decoherence which is a function of the squared distance in phase space between the two field states .
2(b ) shows us how the maximum achievable entanglement depends on the ratio @xmath61 , as did the limit of the coherence parameter in the previous section .
the larger the relative damping @xmath15 , the lower and the quicker the entanglement peaks in time . in both plots
@xmath64 tends to 1 , meaning the qubit is left in a maximally mixed state , disentangled from the probe .
however this is not always the case and in general the smaller the ratio @xmath65 is , the lower the final entropy of the qubit becomes . in view of a qnd measurement on a single qubit @xcite , only the entanglement with the probe needs to be taken into account , as decoherence in the process will not affect the measurement statistics .
however , when the application becomes cat state generation or multiqubit gates @xcite , decoherence becomes a crucial issue which has to be weighted against the entanglement .
in such applications one wishes to produce coherent superpositions of single or multiple quantum systems .
thus it is important to view the behavior of the entropy of the combined state at the time at which the entanglement is maximized .
this behavior is illustrated in fig .
3(a ) , showing the expected limiting behavior of the maximum entanglement as @xmath34 increases . in fig .
3(b ) the corresponding entropy of the combined outgoing state is seen to decrease asymptotically for all choices of the parameter @xmath61 . in consequence
one can simply reduce the amount of decoherence by increasing the strength of the probe .
this is in part due to the fact that the interaction time becomes shorter , reducing the effective decoherence time .
the success of such an approach to minimize the decoherence will then depend on the loss incurred in between interactions .
the reason being that the larger the amplitude @xmath34 is , the larger the amount of dephasing incured by the qubits coupled to the probe mode during these time intervals will be .
this will become clear in the next section . in consequence
we observe a similar trade - off of as that encountered in schemes such as the hybrid quantum repeater proposed in @xcite .
if the transit time and conditions are appropriate , the approach is indeed effective .
for example taking @xmath66 and a reasonable amplitude @xmath67 we obtain a maximum concurrence of 0.998 for a von neumann entropy of @xmath68 .
[ fig3 ] of the probe with @xmath63 , 3 , 5 , 10 and 15 from top to bottom .
( b ) the von neumann entropy of the combined state at the entanglement peaking time with the same values for @xmath61 in decreasing order from top to bottom . ]
so the higher the entanglement we want to measure or couple out of the cavity if we are dealing with cavity qed systems , the larger the probe amplitude and the more precise the timing of the interaction will have to be .
clearly , these issues of coherence and entanglement will have to be combined in order to optimize quantum gates in which different qubits interact with the same probe mode .
many qubus gate proposals @xcite rely directly on the dispersive jaynes - cummings interaction . in those schemes different qubits in separate cavities interact sequentially with the same probe mode and subsequent homodyne measurements can project the qubits to entangled states in a heralded fashion .
alternatively photon number resolved detection can be used , potentially leading to deterministic gates .
dissipation in the probe mode during such gates will reduce the fidelity of the post - selected entangled states and has been investigated in some detail @xcite .
an interesting trade - off arises between the distinguishability of the measurement outcomes and the amount of dephasing incured by the qubits .
more in line with the initial hybrid proposals @xcite , measurement - free quantum gates can be implemented with the use of what we will refer to as the _ conditional displacement _ interaction .
a qubit conditionally displaces a continuous variable bus with the following operation : @xmath69 $ ] where @xmath70 is the amount by which the field is displaced .
thus we can see that after the operation , the position of the probe in phase space is dependent on the state of the qubit ( they have become entangled ) .
there are different ways of arriving at such a coupling between a qubit and a continuous variable .
the physical system may for example be naturally described by an interaction hamiltonian directly producing the conditional displacement as defined above .
a flux qubit in a cavity for example could displace the cavity mode in such a way @xcite . in cavity qed systems ,
a particularly shaped pulse entering and leaking out of the cavity could also provide us with a very similar interaction hamiltonian @xcite .
however , based on the dispersive jaynes - cummings interaction , it was shown lately that a conditional displacement as defined above can be exactly simulated by using unconditional displacements @xcite . defining the _ conditional rotation _
operation encountered in the previous section as @xmath71 $ ] , a possible sequence of interactions is the following @xmath72 with @xmath34 real ( see fig .
4 ) . surprisingly all the phases cancel out to give us an exact conditional displacement , independent of the initial probe position in phase space .
making @xmath34 pure imaginary will lead to a displacement in the orthogonal ( real ) direction .
consecutive displacements induce phases as @xmath73 and consequently @xmath74 as @xmath75 .
displacements and rotations acting on the density matrix we will denote as @xmath76 and @xmath77 . [ fig4 ] and @xmath78 respectively . here as in the text , we assume @xmath34 to be real .
the ` r ' and ` i ' exponents denote the real and imaginary parts of the probe amplitude @xmath79 at each time step . ] in order to characterize loss in an interaction sequence , we will introduce dissipation in the probe mode between and during each interaction .
dissipation during the interactions is due to loss in the nonlinear material or cavity system used to to mediate the interaction , whereas dissipation in between each interaction is due to fiber loss , mode mismatch and other effects .
consequently a different loss parameter should arise .
however , both types of loss result in a dephasing of the qubit(s ) and their effects can thus be combined into a single loss parameter .
we will also assume the amount of loss is the same in each segment .
dissipation affects a coherent state matrix element as @xmath80 with @xmath81 @xcite .
we fix the amount of loss between each interaction @xmath82 and represent the attenuated coherent state by @xmath83 such that the number of dots will determine the number of attenuations i.e. @xmath84 .
the quantum operations on the qubits will be obtained by calculating these state - dependent overlaps .
now let us consider the effects of dissipation in the whole interaction sequence ( 15 ) .
for generality we will keep the amplitudes of the three displacements as free real variables @xmath85 , @xmath86 and @xmath87 . the first step in the sequence is a displacement so taking our probe initially in the vacuum state we have @xmath88 where @xmath89 represents the basis states of the qubit and the state of the probe mode is kept to the right .
loss in the probe mode at this point will decrease the amplitude of the coherent state to @xmath90 , without affecting the qubit .
proceeding to the second step which is a conditional rotation we have @xmath91 now we introduce loss in the probe mode , leading to @xmath92 continuing in this fashion and completing the sequence @xmath93 \times ( \xi_1\xi_2\xi_3)^{\eta } , \label{}\end{aligned}\ ] ] with the @xmath94 s representing the three loss terms ( overlaps ) .
the notation @xmath95 means the contents of the bra are the same as in the previous ket replacing @xmath96 with @xmath97 . in order to simulate a conditional displacement ( with dephasing on the qubit )
, we require the state of the probe mode to be of the form @xmath98 .
this is achieved by setting @xmath99 . combining the loss terms
we obtain @xmath100\times\mathrm{exp}[it(z_a - z_{a ' } ) ] , \label{}\ ] ] with @xmath101 and @xmath102 . here as in the evaluation of the coherence parameter ( 7 ) in section i , the exponent can be separated into real and imaginary parts . the former is characterized by @xmath103 , representing the amount of dephasing incurred by the qubit , which can be decomposed into a phase flip channel as in ( 10 ) .
the latter constitutes the known phase acquired by the qubit in the process , characterized by @xmath104 .
interestingly , this overall conditional phase can be tuned at will by adapting the amplitudes of the displacements , leading to the exact simulation of a conditional displacement . in other words ,
in this sequence we can limit the effects of dissipation to a dephasing on the qubit , producing nonetheless the correct combined output state .
previously also , loss during the interaction had led to single qubit dephasing effects which can be factored in here .
we now move on to examine a two qubit gate in the presence of probe loss .
based on four of these conditional displacements induced by a pair of qubits on the same probe mode , a gate locally equivalent to the cz gate can be built .
an interaction sequence leading to this unitary operation is @xmath105 with @xmath106 where @xmath107 and @xmath108 act on qubits @xmath1 and @xmath55 respectively .
here we use the same method as before , introducing dissipation between each interaction , as illustrated in fig .
we will assume the probe starts off in the vacuum state , but the same result is obtained if we initialize it in any coherent state .
the initial density matrix reads @xmath109 at this point it is worth noting that even though the probe undergoes amplitude damping , it can nonetheless be perfectly disentangled from the qubits coupled to it whatever the amplitude of the coherent state it starts off in ( it may not be the case for other optical states ) .
this is done by tuning the second ( opposite ) conditional displacement , in function of the known loss parameter @xmath110 into which loss during the interaction has also been factored .
thus the amplitude of the second conditional displacement will be reduced by a factor of @xmath111 , the damping undergone by the probe since the last coupling . resolving the whole gate sequence ( see fig . 5 ) and choosing @xmath112 and @xmath113 to be real we obtain [ fig4 ] @xmath114 . \label{}\end{aligned}\ ] ] first we notice the geometrical phase represented by the last term .
this is precisely the form of a two - qubit conditional phase having been applied to the density matrix . ignoring the other terms ,
if we can set @xmath115 then we have simulated a cz gate .
however the dephasing effects are included in the three @xmath94 overlaps . the first and third lead to single qubit dephasing @xmath116,\nonumber\\ \nonumber\\ \xi_3=\langle i\dot{\beta}_bz_{b'}|i\dot{\beta}_b z_b\rangle=\mathrm{exp}[-\dot{\beta}_b^2(1-z_bz_{b ' } ) ] , \label{}\end{aligned}\ ] ] while the second overlap corresponds to loss in the probe mode when it holds information on both qubits @xmath117.\nonumber\\ \label{}\end{aligned}\ ] ] in order to be able to express the resulting quantum operation in a closed form , we first arrange the terms in the exponential such that when writing the expansion we obtain an accessible closed algebra .
symmetrizing the terms we have [ fig5 ] and @xmath118 respectively , insuring that the probe disentangles from the qubits . ]
@xmath119 , \label{}\ ] ] with @xmath120 , @xmath121 , @xmath122 and @xmath123 .
focusing first on the @xmath124 term , we can write out the expansion of that particular exponential acting on the density matrix as @xmath125 now we define the two - qubit operators @xmath126 then grouping the terms in the expansion we obtain the following unnormalized operation @xmath127 with @xmath128 now we have to factor in the other terms @xmath129 , @xmath130 and @xmath131 of @xmath132 . further identifying @xmath133 and @xmath134
we obtain the final normalized operation @xmath135 on the two - qubit state .
setting @xmath136 removes the single qubit terms @xmath107 and @xmath108 and also the @xmath137 terms , yielding the operation @xmath138 with @xmath139 . focusing on the low loss regime
, we can assume a small @xmath140 between each interaction .
truncating to second order in @xmath140 the expansion ( 26 ) leading to the above operation we have @xmath141 let us assume we are in a quantum error correction ( qec ) setting where ancilla qubits are being used .
once the ancilla systems have undergone projective measurements , syndromes are extracted for each logical qubit , indicating whether or not it has been subject to a @xmath8 error .
the last two terms however will lead to cross terms of syndrome states which can not be observed in the measurement process and thus they are removed from the resulting density matrix @xcite .
this leaves us with single qubit errors on @xmath1 and @xmath55 with equal probability ( they constitute the observable part of the map ) .
so for small loss we are only observing single qubit errors throughout the gate ( at each one of the three dissipation stages ) which can be corrected for via qec .
correlated errors represented by the operator @xmath142 appearing in higher order terms are quantified by the coefficient @xmath143 in the normalized operation ( 32 ) . in fig .
7(a ) we can see how this part scales with loss in comparison to uncorrelated errors .
the general quantum operations we have obtained are of great importance for error correction .
they provide us with the error syndromes and their associated probabilities which in turn can be directly fed into fault tolerance calculations .
this will ultimately allow us to compare the qubus scheme with other proposed implementations of quantum information processing .
[ fig3 ] ) and uncorrelated errors ( @xmath144 ) against the loss @xmath145 as defined in the text .
dissipation occurs three times , once between each interaction , in equal amounts .
( b ) the fidelity ( dashed ) @xmath146 of the two qubit output state , where @xmath147 constitutes the ideal output state and the concurrence ( solid ) , both for the single and the iterated sequences .
we observe the same fidelity for both of them while the outputted concurrence is lower for the iterated sequence . ] to appreciate the effect of loss on the whole gate we look at the fidelity of the output state with regards to the ideal output , for a two qubit input state @xmath148 .
such an equally weighted superposition input state provides a good general indication as to the gates performance in addition to the fact that the ideal output is a useful resource , locally equivalent to a bell state or two - qubit cluster state . a plot of the fidelity against the relative intensity decrease of the probe through dissipation defined as @xmath149)^2 $ ] is shown in fig .
this can be understood in that if the probe is initiated in a coherent state with amplitude @xmath34 , then at the end of the gate it disentangles from the qubits and is left with an amplitude @xmath150 . in computing the fidelity
we use @xmath151 so as to make the phase represented by the last term in ( 22 ) that of an ideal cz gate .
we find that even for an intensity decrease of up to 80% in the probe mode ( corresponding to 8 db loss in total ) , the output fidelity remains above 0.5 , allowing for purification . in the moderate loss regime , both the fidelity and the entanglement remain high ; for example taking @xmath152 corresponding to a decrease of @xmath153 in probe intensity ( 0.22 db loss ) results in an output with @xmath154 and a concurrence of @xmath155 .
finally we now show how a simple repetition scheme can significantly simplify full gate operation .
the first point to notice is similar in spirit to the observation made in @xcite that the ideal ( loss - free ) operation is invariant under time reversal .
that is if we reverse the order of the interactions , we obtain the same conditional geometric phase .
however in the case of separate cavities , a time reversed iteration does not help fight decoherence for a coherent state probe .
the reason for this is that the single qubit error linked to the transfer between each cavity scales in the same way as the geometrical phase ( of order @xmath156 ) .
the observation we make here is that the gate is also invariant under a swapping of the displacement directions .
that is the same geometrical phase is obtained , again in the loss free case , if now qubit @xmath1 conditionally displaces the probe in the imaginary direction in ( 22 ) and qubit @xmath55 displaces it in the real direction .
let us denote the two different sequences in a dissipative setting as @xmath157 we have @xmath158 with @xmath159 .
now we notice that the effects of the two sequences differ only in the central overlap terms @xmath160 and @xmath161 which contain the multiple correlated error terms .
it is straightforward to see that their combined effect yields @xmath162 which are just single qubit phase flip channels .
thus the combination of the two sequences gives the operation @xmath163 where the first term is the unitary conditional two - qubit phase and the next two are single qubit dephasing terms .
the corresponding operation undergone by the qubits , omitting the conditional phase is @xmath164 with @xmath165 the probability that each qubit incured a @xmath8 error .
clearly these dephasing processes are independent , leading to a very simple operation , in contrast with the partial operations ( 31 ) or even ( 32 ) obtained in the single interaction sequence .
due to the fact that the geometrical phase and the qubit dephasing both double for a single iteration of the gate , there is no advantage to further reducing the sizes of the displacements and increasing the number of iterations .
the output state fidelity and entanglement of this sequence are plotted in fig .
7(b ) , setting @xmath166 . the fidelity is very similar while the concurrence is slightly reduced , compared to the single sequence .
so for entanglement distribution in view of communication applications , which only follow the purification of an entangled state up to an acceptable level , the iterated scheme is penalizing , as it requires twice the amount of time and reduces the entanglement .
but in view of full blown quantum computation with quantum error correction , the iterated scheme may present a serious advantage , suppressing correlated errors at the gate level .
in this work we set out to examine the vital issue of dissipation in a cv quantum bus being used to realize quantum information processing tasks over discrete subsystems .
we first looked at dissipation in the probe during interactions .
we considered a general interaction hamiltonian between continuous and discrete subsystems which appeared in two different physical settings : all optical cross - phase and dispersive cqed settings . solving the master equation we derived the closed form of the coherence parameter ( 7 ) discussed in some previous work @xcite .
this parameter can be expressed as a function of @xmath34 , @xmath61 and @xmath60 and has a non - zero limiting behavior for large interaction times . as expected , it has the effect of a phase flip channel ( 10 ) in the case that the subsystem is a qubit .
more interestingly , loss on the probe mode induces a conditional phase operation on the qubit , which needs to be undone in the case that the interaction is part of a more elaborate gate . the resulting time dependent density matrix we were able to rewrite as a two - qubit density matrix . from this
we observed the evolution of the entropy and the entanglement between the continuous variable and the qubit during the interaction .
increasing the probe amplitude results in higher entanglement with a sharper peaking , meaning the interaction time has to be controlled more and more precisely . we also found that increasing the probe amplitude increases the coherence of the combined state at entanglement peaking time .
this could be a possible approach to minimizing decoherence effects in qubus gates , however it has to be balanced against loss in between interactions and the conditional phase discussed above which depends strongly on the amplitude . as we saw in sections iii and iv the dephasing effect on the qubit
is related to the overlap between the possible states of the probe in phase space , which is a function of the squared distance between them . thus increasing @xmath34 in the interactions
will augment the effects of loss in the transmission .
the solution to this trade - off will depend on the physical system at hand .
having solved the effects of loss on interactions , we combined them with displacements to simulate conditional displacements .
these couplings are fundamental to hybrid schemes making use of the geometrical phase acquired by the probe in phase space .
we managed to compensate for the combined dephasing effects during and in between interactions , obtaining the desired coupling up to a single qubit dephasing .
this result is crucial to the analysis of the full cz gate undertaken in section iv , in which we compounded the conditional displacement operations and added loss in between them too .
the first and last loss terms constitute single qubit errors whereas the middle term contains correlated errors .
we calculated the resulting quantum operation on the pair of qubits and found that in the low loss regime it only contained single qubit dephasing errors , given a qec setting with syndrome measurements . to gain more insight into the gate s performance
, we looked at the output state fidelity , for equally weighted input superposition states .
we found that the fidelity remained high and that even for losses of up to 8 db in probe signal , the output states could be purified .
a simple iteration procedure then allowed us to remove extra correlated errors altogether , presenting a serious advantage in view of a full - blown quantum computing scheme .
this full - blown scheme will also have to factor other sources of errors such direct qubit decoherence processes .
the method we use here is very broad and could be applied to general bus mediated quantum information processing schemes .
loss of coherence in the bus will automatically result in dephasing on the qubits coupled to it at that time .
if several qubits are simultaneously coupled then correlated errors will arise .
these will have to be minimized in order to efficiently correct for errors .
however the peculiarity of the qubus scheme is that we are using non - orthogonal states of the bus to encode information held by the subsystems .
surprisingly , in the absence of loss , perfect multiqubit gates can be implemented this way . in a dissipative
setting , this non - orthogonality leads to overlap calculations which are very likely to arise in general non - orthogonal bus schemes . following the results presented in this paper ,
there are two main directions for future work .
the first is to find a way of dealing with the single qubit phases accumulated throughout qubus gates .
simply undoing them would require considerable precision and this may not be a realistic option .
the second is to investigate possible schemes to reduce dephasing effects , possibly by engineering the probe itself .
we would like to thank s. devitt , t.d .
ladd , p. van loock , n. ltkenhaus and t. karasawa for useful insights and discussions .
sgrl acknowledges the support of a monbukagakusho scholarship .
this work was supported in part by mext in japan and the eu project qap .
the exponent @xmath167 in the coherence parameter @xmath39 can be written as @xmath168+i\mathrm{im}[f_{nm}]$ ] . here
we give expressions for the real and imaginary parts of the exponent with @xmath7 . beginning with the real part
we obtain @xmath43}&=&\mathrm{exp}[-\frac{|\alpha|^2}{2(\gamma^2+\chi^2)}(\chi^2(1-e^{-2\gamma t})\nonumber\\ & -&2\gamma^2e^{-2\gamma t}\mathrm{sin}^2\chi t-\chi\gamma
e^{-2\gamma t}\mathrm{sin}2\chi t)\nonumber\\ & \times & ( 1-z_nz_m ) ] , \label{}\end{aligned}\ ] ] representing the decay of the off - diagonal components of the density matrix . in the limit of large interaction times
this term leads to a fixed dephasing effect @xmath169}_{t\rightarrow\infty}=\mathrm{exp}\left[-\frac{|\alpha|^2}{2(1+(\gamma/\chi)^2)}(1-z_nz_m)\right ] .
\label{}\ ] ] then the imaginary part of the exponent is @xmath170}&=&\mathrm{exp}[\frac{i\gamma|\alpha|^2}{2(\gamma^2+\chi^2)}(\chi(1-e^{-2\gamma t}\mathrm{cos}2\chi t)\nonumber\\ & -&\gamma\mathrm{sin}2\chi t)(z_n - z_m ) ] , \label{}\end{aligned}\ ] ] corresponding to a known single qubit phase term . 99 a. ekert and r. jozsa , rev .
phys . * 68 * , 733 ( 1996 ) .
s. lloyd , science * 273 * , 1073 ( 1996 ) .
abrams and s. lloyd , phys .
lett . * 79 * , 2586 ( 1997 ) .
cirac and p. zoller , nature ( london ) * 404 * , 579 ( 2000 ) .
e. knill , r. laflamme and g. milburn , nature ( london ) * 409 * , 46 ( 2001 ) .
kane , nature ( london ) * 393 * , 133 ( 1998 ) .
ladd , j.r .
goldman , f. yamaguchi , y. yamamoto , e. abe and k.m .
itoh , phys .
89 * , 017901 ( 2002 ) .
briegel and r. raussendorf , phys .
lett . * 86 * , 910 ( 2001 ) .
vandersypen et al . ,
nature * 414 * , 883 ( 2001 ) .
sackett et al . ,
nature * 404 * , 256 ( 2000 ) .
pittman , m.j .
fitch , b.c jacobs , j.d .
franson , rev .
a * 68 * , 032316 ( 2003 ) .
p.walther et al . ,
nature * 434 * , 169 ( 2005 ) .
divincenzo , fortschr . phys . * 48 * , 771 ( 2000 ) . m.b .
plenio , s.f .
huelga , a. beige , and p.l .
knight , phys .
a * 59 * , 2468 ( 1999 ) .
greentree , j.h .
cole , a.r .
hamilton , and l.c.l .
hollenberg , phys .
b * 70 * , 235317 ( 2004 ) . t.p .
spiller , k. nemoto , s.l .
braunstein , w.j .
munro , p. van loock and g.j .
milburn , new j. phys .
* 8 * , 30 ( 2006 ) .
s. lloyd and s.l .
braunstein , phys .
* 82 * , 1784 ( 1999 ) .
s.l . braunstein and p. van loock , rev .
phys . * 77 * , 513 ( 2005 ) .
s. lloyd , quant - ph/0008057 ( 1999 ) .
x. wang , a. srensen and k. mlmer , phys .
* 86 * , 3907 ( 2001 ) .
x. wang and p. zanardi , phys .
a * 65 * , 032327 ( 2002 ) . g.j .
milburn , quant - ph/9908037 ( 1999 ) . g.j .
milburn and d.f .
walls , phys .
a * 30 * , 56 ( 1984 ) .
n. imoto , h.a .
haus , and y. yamamoto , phys .
a * 32 * , 2287 ( 1985 ) .
munro , k. nemoto , t.p .
spiller , s.d .
barrett , p. kok , and r.g .
beausoleil , j. opt .
b : quantum semiclass . opt .
* 7 * s135 ( 2005 ) .
munro , k. nemoto and t.p .
spiller , new j. phys .
* 7 * , 137 ( 2005 ) .
louis , k. nemoto , w.j . munro and t.p .
spiller , phys .
rev . a * 75 * 042323 ( 2006 ) .
k. nemoto and w.j .
munro , phys .
lett . * 93 * , 250502 ( 2004 ) .
louis , k. nemoto , w.j .
munro and t.p .
spiller , new j. phys .
* 9 * , 193 ( 2007 ) .
van loock , t.d .
ladd , k. sanaka , f. yamaguchi , k. nemoto , w.j .
munro , and y. yamamoto , phys .
* 96 * , 240501 ( 2006 ) .
ladd , p. van loock , k. nemoto , w.j .
munro , y. yamamoto , new j. phys . * 8 * , 164 ( 2006 ) .
jaynes and f.w .
cummings , proc .
ieee * 51 * , 89 ( 1963 ) .
rodrigues , c.e .
a. jarvis , b.l .
gyrffy , t.p .
spiller , and j.f .
annett , j. phys . : condens .
matter * 20 * , 075211 ( 2008 ) .
a. wallraff , d.i .
schuster , a. blais , l. frunzio , r .- s .
huang , j. majer , s. kumar , s.m .
girvin , and r.j .
schoelkopf , nature * 431 * , 162 ( 2004 ) .
a. wallraff , d.i .
schuster , a. blais , l. frunzio , j. majer , m.h .
devoret , s.m .
girvin , and r.j .
schoelkopf , phys .
* 95 * , 060501 ( 2005 ) .
j.g.p . de faria and m.c .
nemes , phys .
a * 59 * , 3918 ( 1999 ) .
k.m.f . romero and m. c. nemes ,
physica a * 325 * , 333 ( 2003 ) .
s.d . barrett and g.j .
milburn , phys .
a * 74 * , 060302(r ) ( 2006 ) .
x . cen and p. zanardi , phys .
a * 71 * , 060307(r ) ( 2005 ) .
h. jeong , phys .
a * 72 * , 034305 ( 2005 ) .
h. jeong , phys .
a * 73 * , 052320 ( 2006 ) .
gerry and p.l .
knight , _ introductory quantum optics _ , cambride university press ( 2005 ) .
gardiner and p. zoller , _ quantum noise _ , springer ( 2000 ) .
w. witschel , int .
j. quantum chem . * 20 * , 1233 ( 1981 ) .
nielsen and i.l .
chuang , _ quantum computation and quantum information _ , cambride university press ( 2000 ) .
n. ltkenhaus and p. van loock , private communication .
plenio and s. virmani , quant .
inf . comp .
* 7 * , 1 ( 2007 ) .
bennet , d. p. divincenzo , j. smolin and w. k. wooters , phys .
a * 54 * , 3824 ( 1996 ) .
wooters , phys .
lett . * 80 * 2245 ( 1997 ) .
milburn and d.f .
walls , phys .
a * 30 * , 56 ( 1984 ) .
n. imoto , h.a .
haus , and y. yamamoto , phys .
a * 32 * , 2287 ( 1985 ) .
a. blais , r. huang , a. wallraff , s.m .
girvin and r.j .
schoelkopf , phys .
a * 69 * , 062320 ( 2004 ) .
munro , in preparation .
p. van loock , w.j .
munro , k. nemoto , t.p .
spiller , t.d .
ladd , s.l .
braunstein and g.j .
milburn , quant - ph/0701057 ( 2007 ) .
d.f . walls and g.j .
milburn , phys .
a * 31 * , 2403 ( 1985 ) .
s. devitt , private communication . | we provide a characterization and analysis of the effects of dissipation on oscillator assisted ( qubus ) quantum gates . the effects can be understood and minimized by looking at the dynamics of the signal coherence and its entanglement with the continuous variable probe . adding loss in between successive interactions we obtain the effective quantum operations , providing a novel approach to loss analysis in such hybrid settings .
we find that in the presence of moderate dissipation the gate can operate with a high fidelity .
we also show how a simple iteration scheme leads to independent single qubit dephasing , while retaining the conditional phase operation regardless of the amount of loss incurred by the probe . |
a large literature has investigated whether economic variables such as , e.g. , the price - dividend ratio , proxies of labour income , or the interest rate can predict stock returns . the econometric approach to test for predictability is mostly based on a predictive regression of stock returns onto a set of lagged financial variables ; see , e.g. , stambaugh ( 1999 ) .
important differences between testing approaches in the literature arise because of the different test statistics , asymptotic theories or resampling approaches used to test the null hypothesis of no predictability .
these differences lead in a number of cases to diverging results and conclusions .
mankiw and shapiro ( 1986 ) and stambaugh ( 1986 ) note that in a setting with endogenous predictor and correlated innovations standard asymptotic theory causes small sample biases that may imply an overrejection of the hypothesis of no predictability . to mitigate the problem ,
recent studies propose tests based on bias - corrected estimators of predictive regressions .
for instance , stambaugh ( 1999 ) , and amihud , hurvich and wang ( 2008 ) introduce bias - corrected ols estimators for the univariate and the multi - predictor setting , respectively .
recent work has also considered the issue of endogenous integrated or nearly integrated predictors , following the evidence in torous , valkanov and yan ( 2004 ) that various variables assumed to predict stock returns follow a local - to - unit root autoregressive process .
lewellen ( 2004 ) , torous , valkanov and yan ( 2004 ) , and campbell and yogo ( 2006 ) introduce testing procedures and more accurate unit - root and local - to - unit root asymptotics for predictive regression models with a single persistent predictor and correlated innovations .
more recently , kostakis , magdalinos and stamatogiannis ( 2015 ) propose a new class of test statistics , by extending the instrumental variables approach in magdalinos and phillips ( 2009 ) to predictive regressions . a general approach to obtain tests that are less susceptible to finite sample biases or assumptions on the form of their asymptotic distribution relies on nonparametric monte carlo simulation methods , such as the bootstrap or the subsampling .
ang and bekaert ( 2007 ) use the bootstrap to quantify the bias of parameter estimation in a regression of stock returns on the lagged dividend yield and the interest rate . in a multi - predictor setting with nearly integrated regressors , amihud , hurvich and wang ( 2008 ) compare the results of bootstrap tests to bias - corrected procedures and find the latter to have accurate size and good power properties .
wolf ( 2000 ) introduces subsampling tests of stock return predictability in single - predictor models . as shown in hall and horowitz ( 1996 ) and andrews ( 2002 ) , among others , a desirable property of bootstrap tests
is that they may provide asymptotic refinements of the sampling distribution of standard @xmath0-test statistics for testing the hypothesis of no predictability .- test statistic are of lower order with respect to the sample size than those implied by the conventional asymptotics .
] moreover , as shown in romano and wolf ( 2001 ) , choi and chue ( 2007 ) , and andrews and guggenberger ( 2009 , 2010 ) , subsampling methods produce reliable inference also in predictive regression models with multiple nearly integrated predictors .
a common feature of all above approaches to test predictability hypotheses is their reliance on procedures that can be heavily influenced by a small fraction of anomalous observations in the data . for standard ols estimators and @xmath0-test statistics ,
this problem is well - known since a long time ; see , e.g. , huber ( 1981 ) for a review .
more recent research has also shown that inference provided by bootstrap and subsampling tests may be easily inflated by a small fraction of anomalous observations .
intuitively , we explain this feature by the too high fraction of anomalous observations that is often simulated by conventional bootstrap and subsampling procedures , when compared to the actual fraction of outliers in the original data .
it is not possible to mitigate this problem simply by applying conventional bootstrap or subsampling methods to more robust estimators or test statistics .
resampling trimmed or winsorized estimators does not yield a robust resampling method ( see singh , 1988 , camponovo , trojani and scaillet , 2012 , for detailed examples ) .
hence , we consider in this paper a new robust resample methodology for time series , which allows us to develop more robust tests of predictability hypotheses in predictive regression settings .
our robust predictive regression approach relies on robust weighted least - squares procedures that are data - driven and easily manageable .
this can be motivated economically with the presence of a time - varying ambiguity about predictive relations , which is consistently addressed by ambiguity averse investors using robust estimators that bound the effects of anomalous data features .
wrampelmeyer , wiehenkamp and trojani ( 2015 ) show that different specifications of aversion to ambiguity in the literature imply robust optimal estimator choices related to robust weighted least - squares . in this sense ,
our robust predictive regression testing approach is consistent with the preferences of investors that dislike a time - varying ambiguity in the data - generating processes for returns .
the data - driven weights in the procedure dampen , where necessary , the few data points that are estimated as anomalous with respect to the postulated predictive link .
this feature automatically avoids , e.g. , arguing ex ante that a large value of the predicted or the predictive variables is per se an anomalous observation , which is not in general the case .
indeed , observations linked to large values of both the predictive and the predicted variables might be very informative about a potential predictability structure and discarding them in an ad hoc way might bias the inference . in a multivariate predictive regression setting , it is even more difficult to determine with an informal approach which subset of observations is potentially anomalous , for example by eyeballing the data . a useful property of our methodology it that it embeds a formal data - driven identification of observations that can be excessively influential for the resulting inference on predictive relations .
the more detailed contributions to the literature are as follows .
first , using monte carlo simulations we find that the size and power of conventional hypothesis testing methods for predictive regressions , including bias - corrected tests , tests implied by local - to - unity asymptotics , and conventional bootstrap and subsampling tests , are dramatically nonresistant to even small fractions of anomalous observations in the data
. even though the test probability of rejecting a null by chance alone features some degree of resistance in our monte carlo experiments , the test ability to reject the null of no predictability when it is violated is in most cases drastically reduced .
second , we quantify theoretically the robustness properties of subsampling and bootstrap tests in a time series context , borrowing from the concept of breakdown point , which is a measure of the degree of resistance of a testing procedure to outliers ; see , e.g. , hampel ( 1971 ) , donoho and huber ( 1983 ) , and hampel , ronchetti , rousseeuw and stahel ( 1986 ) . in section 3.3 below , theorem 1 ( and its proof ) for subsampling differs from theorem 2 ( and its proof ) in camponovo , scaillet and trojani ( 2012 ) valid for the i.i.d .
case since subsamples in a time series context are not generated in the same way in order to avoid breaking serial dependences .
theorem 1 ( and its proof ) for block bootstrap is also new , and not a straightforward extension of the results in singh ( 1998 ) for the i.i.d . bootstrap .
third , we develop a novel class of resampling tests of predictability , which are resistant to anomalous observations and consistent with nearly integrated regressors at sustainable computational costs .
we confirm by monte carlo simulations that these tests successfully limit the damaging effect of outliers , by preserving desirable finite sample properties in presence of anomalous observations .
finally , we provide a robust analysis of the recent empirical evidence on stock return predictability for us stock market data . following cochrane ( 2008 )
, our main purpose is not to determine the best return - forecasting specification , but rather to study the amount of predictability resulting from very simple specifications , motivated by economic theory for the vast majority for the data .
we study single - predictor and multi - predictor models , using several well - known predictive variables suggested in the literature , such as the lagged dividend yield , the difference between option - implied volatility and realized volatility ( bollerslev , tauchen and zhou , 2009 ) , the interest rate , and the share of labor income to consumption ( santos and veronesi , 2006 ) .
our robust tests of predictability produce the following novel empirical evidence .
first , we find that the dividend yield is a robust predictive variable of market returns , which is significant at the 5% significance in all our regressions , for each subperiod , sampling frequency and forecasting horizon considered . in univariate regressions with monthly data , the lagged dividend yield is significant at the 5% level according to the robust tests , in each window of 180 monthly observations from january 1980 to december 2010 . in contrast ,
bias - corrected methods , local - to - unity asymptotics and conventional sabsampling tests produce a weaker and more ambiguous evidence overall , e.g. , by not rejecting the null of no predictability at the 10% significance level in the subperiod from january 1995 to december 2010 .
multi - predictor regressions including variance risk premium and labor income proxies confirm the significant predictive power of the dividend yield . while the dividend yield is again significant at the 5% level in all cases using the robust tests , it is significant only at the 10% level using the conventional tests in the sample period 1994 - 2009 , within monthly predictive regressions including the difference of implied and realized volatility as a predictive variable .
it is not significant using conventional tests in the sample period 1955 - 2010 within quarterly predictive regression based on the share of labor income to consumption .
a weak evidence of predictability for future returns and cash flows is at odd with the fundamental present - value relations linking expected returns , expected cash - flows and the price - dividend ratio ; see also cochrane ( 2011 ) . therefore , our robust findings of market return predictability are consistent with the main logic of basic present - value models .
second , we find that the difference between option - implied volatility and realized volatility is a robust predictive variable of future market returns at quarterly forecasting horizons . it is always significant at the 5% significance level in each window of 180 observations , using both robust and nonrobust testing approaches .
this finding supports the remarkable market return forecasting ability of the variance risk premium , noted in bollerslev , tauchen and zhou ( 2009 ) and confirmed in bollerslev , marrone and zhou ( 2014 ) in an international context .
third , using conventional testing approaches , we find that the evidence of return predictability associated with the ratio of labor income to consumption is either absent or weak in the sample periods 1955 - 2000 and 1965 - 2010 , respectively .
in contrast , the null of no predictability is always rejected at the 5% significance level by our robust testing method , indicating that the weak and ambiguous evidence produced by the conventional tests is likely a consequence of their low power in presence of anomalous observations .
fourth , we exploit the properties of our robust testing method to identify observations that might excessively influence the diverging conclusions of conventional testing approaches .
we find a fraction of less than about 5% of influential observations in the data , which tend to be more frequent during the nasdaq bubble and the more recent financial crisis .
such influential data points , including the lehman brothers default on september 2008 , the terrorist attack of september 2001 , the black monday on october 1987 , and the dot - com bubble collapse in august 2002 , are largely responsible for the failure of conventional testing methods in uncovering the hidden predictability structures .
finally , we find that our results can not be substantially improved by specifying predictive relations with time - varying parameters , as we do not find evidence of structural breaks in our sample period after controlling for the impact of anomalous observations .
motivated by the findings in goyal and welch ( 2003 ) and campbell and thompson ( 2008 ) , we also show that the predictive relations detected by our robust approach generate incremental out - of - sample predictive power over a monthly forecasting horizon , improving on both the predictive relations estimated by conventional methods or those of a simple forecast based on the sample mean of market returns .
the rest of the paper is organized as follows . in section 2 ,
we introduce the usual predictive regression model , and we illustrate by simulation the robustness problem of some of the recent tests of predictability proposed in the literature . in section 3 ,
we study theoretically the robustness properties of bootstrap and subsampling approximations . in section 4 ,
we introduce our robust approach , and develop robust bootstrap and subsampling tests of predictability . in section 5
, we apply our robust testing procedure to us equity data and reconsider some of the recent empirical evidence on market return predictability .
section 6 concludes .
in this section , we introduce the benchmark predictive regression model and a number of recent methods proposed for testing the predictability of stock returns . through monte carlo simulations , we study the finite - sample properties of these testing procedures both in presence and absence of anomalous observations . in section [ pm ] , we first introduce the model . in section [ ri ] ,
we focus on bias - corrected methods and testing procedures based on local - to - unity asymptotics .
finally , in section [ rm ] , we consider testing approaches based on resampling methods .
we consider the predictive regression model , @xmath1 where , @xmath2 denotes the stock return at time @xmath3 and @xmath4 is an economic variable observed at time @xmath5 , predicting @xmath2 . the parameters @xmath6 and @xmath7 are the unknown intercepts of the linear regression model and the autoregressive model , respectively , @xmath8 is the unknown parameter of interest , @xmath9 is the unknown autoregressive coefficient , @xmath10 , @xmath11 are error terms with @xmath12 , @xmath13 , and @xmath14 is a scalar random variable . in this setting , it is well - known that inference based on standard asymptotic theory suffers from small sample biases , which may imply an overrejection of the hypothesis of no predictability , @xmath15 , where @xmath16 denotes the true value of the unknown parameter @xmath17 ; see mankiw and shapiro ( 1986 ) , and stambaugh ( 1986 ) , among others .
moreover , as emphasized in torous , valkanov , and yan ( 2004 ) , various state variables considered as predictors in model ( [ pregmodel1])-([pregmodel2 ] ) might be well approximated by a nearly integrated process , which might motivate a local - to - unity framework @xmath18 , @xmath19 , for the autoregressive coefficient of model ( [ pregmodel2 ] ) , implying a nonstandard asymptotic distribution for the ols estimator @xmath20 of parameter @xmath17 .
several recent testing procedures have been proposed in order to overcome these problems .
stambaugh ( 1999 ) , lewellen ( 2004 ) , amihud and hurvich ( 2004 ) , polk , thompson and vuolteenaho ( 2006 ) , and amihud , hurvich and wang ( 2008 , 2010 ) , among others , propose bias - corrected procedures that correct the bias implied by the ols estimator @xmath20 of parameter @xmath17 .
cavanagh , elliott and stock ( 1995 ) , torous , valkanov and yan ( 2004 ) , and campbell and yogo ( 2006 ) , among others , introduce testing procedures based on local - to - unity asymptotics that provide more accurate approximations of the sampling distribution of the @xmath0-statistic @xmath21 in nearly integrated settings , where @xmath22 is an estimate of the standard deviation of the ols estimator @xmath20 .
kostakis , magdalinos and stamatogiannis ( 2015 ) also propose a new class of test statistics by extending the instrumental variables approach develop in magdalinos and phillips ( 2009 ) to predictive regressions .
a common feature of bias - corrected methods and inference based on local - to - unity asymptotics is a nonresistance to anomalous observations , which may lead to conclusions determined by the particular features of a small subfraction of the data .
intuitively , this feature emerges because these approaches exploit statistical tools that can be sensitive to small deviations from the predictive regression model ( [ pregmodel1])-([pregmodel2 ] ) . consequently , despite the good accuracy under the strict model assumptions , these testing procedures may become less efficient or biased even with a small fraction of anomalous observations in the data . to illustrate the lack of robustness of this class of tests , we analyze through monte carlo simulation the bias - corrected method proposed in amihud , hurvich and wang ( 2008 ) and the bonferroni approach for the local - to - unity asymptotic theory introduced in campbell and yogo ( 2006 ) .
we first generate @xmath23 samples @xmath24 , where @xmath25 , of size @xmath26 according to model ( [ pregmodel1])-([pregmodel2 ] ) , with @xmath27 , @xmath28 , @xmath29 , @xmath30 , @xmath31 , and @xmath32 .
are qualitatively very similar . ] in a second step , to study the robustness of the methods under investigation , we consider replacement outliers random samples @xmath33 , where @xmath34 is generated according to , @xmath35 with @xmath36 and @xmath37 is an i.i.d . @xmath38
random sequence , independent of process ( [ pregmodel1])-([pregmodel2 ] ) such that @xmath39=\eta$ ] . the probability of contamination by outliers is set to @xmath40 , which is a small contamination of the original sample , compatible with the features of the real data set analyzed in the empirical study in section [ spm1 ] . , the estimated fraction of anomalous observations in the sample period 1980 - 2010 is less than about @xmath41 . ]
we study the finite sample properties of tests of the null hypothesis @xmath42 in the predictive regression model . in the first two rows of figure [ poweracy ]
, we plot the empirical frequency of rejection of null hypothesis @xmath43 for the bias - corrected method proposed in amihud , hurvich and wang ( 2008 ) and the bonferroni approach for the local - to - unity asymptotic theory introduced in campbell and yogo ( 2006 ) , respectively , with respect to different values of the alternative hypothesis @xmath32 , and different degree of persistence of the predictors @xmath31 .
the nominal significance level of the test is @xmath44 .
the results for different degree of persistence are qualitatively very similar . in the monte carlo simulation with noncontaminated samples ( straight line ) , we find that the fraction of null hypothesis rejections of all procedures is quite close to the nominal level @xmath44 when @xmath45 .
as expected , the power of the tests increases for increasing values of @xmath46 . in the simulation with contaminated samples ( dashed line ) ,
the size of all tests remains quite close to the nominal significance level .
in contrast , the presence of anomalous observations dramatically deteriorates the power of both procedures .
indeed , for @xmath47 , the frequency of rejection of the null hypothesis for both tests is much lower than in the noncontaminated case .
unreported monte carlo results for the instrumental variable approach proposed in kostakis , magdalinos and stamatogiannis ( 2015 ) produce similar findings . the results in figure [ poweracy ] highlight the lack of resistance to anomalous data of bias - corrected methods and inference based on local - to - unity asymptotics .
because of a small fraction of anomalous observations , the testing procedures become unreliable , and are unable to reject the null hypothesis of no predictability , even for large values of @xmath16 .
this is a relevant aspect for applications , in which typically the statistical evidence of predictability is weak . to overcome this robustness problem ,
a natural approach is to develop more resistant versions of the nonrobust tests considered in our monte carlo exercise .
however , this task may be hard to achieve in general .
a more general approach to obtain tests that are less susceptible to finite sample biases or assumptions on their asymptotic distribution can rely on nonparametric monte carlo simulation methods .
we address these methods in the sequel .
nonparametric monte carlo simulation methods , such as the bootstrap and the subsampling , may provide improved inferences in predictive regression model ( [ pregmodel1])-([pregmodel2 ] ) both in stationary or nearly integrated settings .
as shown in hall and horowitz ( 1996 ) and andrews ( 2002 ) , for stationary data the block bootstrap may yield improved approximations to the sampling distribution of the standard @xmath0-statistics for testing predictability , having asymptotic errors of lower order in sample size .
moreover , as shown in choi and chue ( 2007 ) and andrews and guggenberger ( 2010 ) , we can use the subsampling to produce correct inferences in nearly integrated settings .
we first introduce block bootstrap and subsampling procedures .
we then focus on predictive regression model ( [ pregmodel1])-([pregmodel2 ] ) and study by monte carlo simulation the degree of resistance to anomalous observations of bootstrap and subsampling tests of predictability , both in stationary and nearly integrated settings .
consider a random sample @xmath48 from a time series of random vectors @xmath49 , @xmath50 , and a general statistic @xmath51 .
block bootstrap procedures split the original sample @xmath52 into overlapping blocks of size @xmath53 . from these blocks , bootstrap samples @xmath54 of size @xmath55
are randomly generated .
finally , the empirical distribution of statistic @xmath56 is used to estimate the sampling distribution of @xmath57 .
similarly , the more recent subsampling method applies statistic @xmath58 directly to overlapping random blocks @xmath59 of size @xmath60 strictly less than @xmath55 .
then , the empirical distribution of statistic @xmath61 is used to estimate the sampling distribution of @xmath57 , under the assumption that the impact of the block size is asymptotically negligible ( @xmath62 ) . in the predictive regression model ( [ pregmodel1])-([pregmodel2 ] ) , the usual @xmath0-test statistic for testing the null of no predictability is @xmath21 .
therefore , we can define a block bootstrap test of the null hypothesis with the block bootstrap statistic @xmath63 , where @xmath64 is an estimate of the standard deviation of the ols estimator @xmath65 in a random bootstrap sample of size @xmath55 , constructed using blocks of size @xmath60 .
similarly , we can define a subsampling test of the same null hypothesis with the subsampling statistic @xmath66 , where @xmath67 is now an estimator of the standard deviation of the ols estimator @xmath68 in a random overlapping block of size @xmath53 .
it is well - known that ols estimators and empirical averages are very sensitive to even small fractions of anomalous observations in the data ; see , e.g. , huber ( 1981 ) .
since bootstrap and subsampling tests rely on such statistics , inference based on these methods may inherit the lack of robustness . to verify this intuition
, we study the finite - sample properties of bootstrap and subsampling tests of predictability in presence of anomalous observations through monte carlo simulations .
first we consider stationary settings . to this end
, we generate @xmath23 samples @xmath24 , where @xmath25 , of size @xmath26 according to model ( [ pregmodel1])-([pregmodel2 ] ) , with @xmath27 , @xmath28 , @xmath29 , @xmath30 , @xmath69 , and @xmath70 .
we then consider also contaminated samples @xmath33 according to ( [ conta ] ) .
we test the null hypothesis @xmath42 , using symmetric bootstrap and subsampling confidence intervals for parameter @xmath17 under different values of the alternative hypothesis @xmath70 .
shows how to construct symmetric confidence intervals for the parameter of interest , based on resampling distributions . for the selection of the block size @xmath60
, we use the standard data - driven method proposed in romano and wolf ( 2001 ) . ] in the first and second rows of figure [ powerstat ] , we plot the empirical frequencies of rejection of null hypothesis @xmath43 , using the subsampling and the bootstrap , respectively .
the nominal significance level of the test is @xmath44 . for @xmath71 , the size of the tests is close to the nominal level @xmath44 , while for @xmath72 the power increases . on the other hand ,
in presence of contamination the power of the tests dramatically decreases . indeed , for @xmath47 ,
also in this case the frequency of rejection of the null hypothesis for both tests is much lower than in the noncontaminated case .
we also study the robustness properties of the subsampling in nearly integrated settings , under the same simulation setting of the previous section . in the third row of figure [ poweracy ] , we plot the empirical frequencies of rejection of null hypothesis @xmath43 for different values of the alternative hypothesis @xmath32 .
the nominal significance level of the test is @xmath44 , as before . with noncontaminated samples ( straight line ) ,
we find for all values of @xmath32 that the frequency of rejection of subsampling tests is close to the one of the bias - corrected method and the bonferroni approach in the previous section .
for @xmath71 , the size of the tests is close to the nominal level @xmath44 , while for @xmath73 the power increases . on the contrary , also in this case
, contaminations with anomalous observations strongly deteriorate the power of the tests . in summary ,
the results in figures [ poweracy ] and [ powerstat ] show that bootstrap and subsampling tests inherit , and to some extent exacerbate , the lack of robustness of ols estimators for predictive regressions . to robustify the inference produced by resampling methods , a natural idea is to apply conventional bootstrap and subsampling simulation schemes to a more robust statistic , such as , e.g. , a robust estimator of linear regression .
unfortunately , as shown in singh ( 1998 ) , salibian - barrera and zamar ( 2002 ) , and camponovo , scaillet and trojani ( 2012 ) for i.i.d .
settings , resampling a robust statistic does not yield a robust inference , because conventional bootstrap and subsampling procedures have an intrinsic nonresistance to outliers . intuitively , this problem arises because the fraction of anomalous observations generated in bootstrap and subsampling blocks is often much higher than the fraction of outliers in the data . to solve this problem ,
it is necessary to address more systematically the robustness of resampling methods for time series .
we characterize theoretically the robustness of bootstrap and subsampling tests in predictive regression settings .
section [ bpq ] introduces the notion of a quantile breakdown point , which is a measure of the global resistance of a resampling method to anomalous observations .
section [ bpointmc ] quantifies and illustrates the quantile breakdown point of conventional bootstrap and subsampling tests in predictive regression models .
finally , section [ tlor ] derives explicit bounds for quantile breakdown points , which quantify the degree of resistance to outliers of bootstrap and subsampling tests for predictability , before applying them to the data . given a random sample @xmath52 from a sequence of random vectors @xmath49 , @xmath50 , let @xmath74 denote a block bootstrap sample , constructed using overlapping blocks of size @xmath60 .
similarly , let @xmath75 denote an overlapping subsampling block .
the construction of blocks is a key difference with respect to the i.i.d . setting , and implies a different extension of available results on breakdown properties for i.i.d
we denote by @xmath76 , @xmath77 , the corresponding block bootstrap and subsampling statistics , respectively . for @xmath78 , the quantile @xmath79 of @xmath76
is defined by @xmath80 where @xmath81 is the probability measure induced by the block bootstrap or the subsampling method and , by definition , @xmath82 .
quantile @xmath79 is effectively a useful nonparametric estimator of the corresponding finite - sample quantile of statistic @xmath57 .
we characterize the robustness properties of block bootstrap and subsampling by the breakdown point @xmath83 of the quantile ( [ quantile ] ) , which is defined as the smallest fraction of outliers in the original sample such that @xmath79 diverges to infinity . borrowing the notation in genton and lucas ( 2003 ) , we formally define the breakdown point of the @xmath0-quantile @xmath84 as , @xmath85,\ ] ] where @xmath86 , and @xmath87 is the set of all @xmath55-samples @xmath88 with exactly @xmath89 nonzero components that are @xmath90-dimensional outliers of size @xmath91 .
, we do not necessarily assume outliers @xmath92 to be all equal to @xmath93 , but we rather assume existence of constants @xmath94 , such that @xmath95 . to better capture the presence of outliers in predictive regression models , our definitions for the breakdown point and the set @xmath87 of all @xmath55-components outlier samples are slightly different from those proposed in genton and lucas ( 2003 ) for general settings
. however , we can modify our results to cover alternative definitions of breakdown point and outlier sets @xmath87 . ] literally , @xmath96 is the smallest fraction of anomalous observations of arbitrary size , in a generic outlier - contaminated sample @xmath97 , such that quantile @xmath98 , estimated by bootstrap or subsampling monte carlo simulation schemes , can become meaningless .
intuitively , when a breakdown occurs , inference about the distribution of @xmath57 based on bootstrap or subsampling tests becomes pointless .
estimated test critical values may be arbitrarily large and confidence intervals be arbitrarily wide . in these cases ,
the size and power of bootstrap and subsampling tests can collapse to zero or one in presence of anomalous observations , making these inference procedures useless .
therefore , providing theory ( see theorem [ bsubboot ] below ) for quantifying @xmath99 in general for bootstrap and subsampling tests of predictability , in dependence of the statistics and testing approaches used , is key in order to understand which approaches ensure some resistance to anomalous observations and which do not , even before looking at the data .
the quantile breakdown point of conventional block bootstrap and subsampling tests for predictability in section [ rm ] depends directly on the breakdown properties of ols estimator @xmath20 .
the breakdown point @xmath100 of a statistics @xmath101 is simply the smallest fraction of outliers in the original sample such that the statistic @xmath102 diverges to infinity ; see , e.g. , donoho and huber ( 1983 ) for the formal definition .
we know @xmath100 explicitly in some cases and we can gauge its value most of the time , for instance by means of simulations and sensitivity analysis .
most nonrobust statistics , like ols estimators for linear regression , have a breakdown point @xmath103 .
therefore , the breakdown point of conventional block bootstrap and subsampling quantiles in predictive regression settings also equals @xmath104 . in other words ,
a single anomalous observation in the original data is sufficient to produce a meaningless inference implied by bootstrap or subsampling quantiles in standard tests of predictability . it is straightforward to illustrate these features in a monte carlo simulation that quantifies the sensitivity of block bootstrap and subsampling quantiles to data contaminations by a single outlier , where the size of the outlier is increasing .
we first simulate @xmath23 random samples @xmath24 of size
@xmath105 , where @xmath25 follows model ( [ pregmodel1])-([pregmodel2 ] ) , @xmath27 , @xmath28 , @xmath29 , @xmath30 , @xmath106 , and @xmath45 .
for each monte carlo sample , we define in a second step @xmath107 and we modify @xmath108 over the interval @xmath109 $ ] .
this means that we contaminate the predictability relationship by an anomalous observation for only one single data point in the full sample .
we study the sensitivity of the monte carlo average length of confidence intervals for parameter @xmath17 , estimated by the standard block bootstrap and the subsampling .
this is a natural exercise , as the length of the confidence interval for parameter @xmath17 is in a one - to - one relation with the critical value of the test of the null of no predictability ( @xmath110 : @xmath71 ) . for the sake of comparison , we also consider confidence intervals implied by the bias - corrected testing method in amihud , hurvich and wang ( 2008 ) and the bonferroni approach proposed in campbell and yogo ( 2006 ) . for all tests under investigation , in the first 2 rows of figure [ sensacy ] ,
we plot the relative increase of the average confidence interval length in our monte carlo simulations , under contamination by a single outlier of increasing size .
we find that all sensitivities are basically linear in the size of the outlier , confirming that a single anomalous observation can have an arbitrarily large impact on the critical values of those tests and make the test results potentially useless , as implied by their quantile breakdown point of @xmath104 . to obtain bootstrap and subsampling tests with more favorable breakdown properties , it is necessary to apply resampling procedures to a robust statistic with nontrivial breakdown point ( @xmath111 ) , such as , e.g. , a robust estimator of linear regression . without loss of generality ,
let @xmath101 be a statistic with breakdown point @xmath112 . in the next theorem , we compute explicit quantile breakdown point bounds , which characterize the resistance of bootstrap and subsampling tests to anomalous observations , in dependence of relevant parameters , such as @xmath55 , @xmath60 , @xmath0 , and @xmath100 .
[ bsubboot ] let @xmath100 be the breakdown point of @xmath113 , @xmath78 , and @xmath114 .
the quantile breakdown points @xmath115 and @xmath116 satisfy the following bounds , @xmath117,\label{fibps}\\ & & \frac{\lceil mb\rceil}{n}\le b^{b}_{t , n , m } \le \frac{1}{n}\cdot\bigg[\inf_{\{p_{1},p_{2}\}}\bigg\{p = p_{1}\cdot p_{2}\bigg\vert p\bigg ( bin\bigg(r,\frac{mp_{2}-p_{1}+1}{n - m+1}\bigg)\ge\bigg\lceil\frac{nb}{p_{1}}\bigg\rceil\bigg)>1-t\bigg\}\bigg],\phantom{p}\label{fibpb}\end{aligned}\ ] ] where @xmath118 , with @xmath119 , and @xmath120 denotes a binomially distributed variable with parameters @xmath121 and @xmath122 . in theorem [ bsubboot ] ,
the term @xmath123 represents the number of degenerated subsampling statistics necessary in order to cause the breakdown of @xmath124 , while @xmath125 is the fraction of outliers which is sufficient to cause the breakdown of statistic @xmath58 in a block of size @xmath60 . the breakdown point formula for the i.i.d .
bootstrap derived in singh ( 1998 ) emerges as a special case of the second inequality in theorem [ bsubboot ] .
we quantify the implications of theorem [ bsubboot ] by computing in table [ table1 ] lower and upper bounds for the breakdown point of subsampling and bootstrap quantiles , using a sample size @xmath105 , and a maximal statistic breakdown point ( @xmath126 ) .
we find that even for a highly robust statistic with maximal breakdown point , the subsampling implies a very low quantile breakdown point , which increases with the block size but is also very far from the maximal value @xmath126 .
for instance , for a block size @xmath127 , the @xmath128-quantile breakdown point is between @xmath129 and @xmath130 . in other words , even though a statistic is resistant to large fractions of anomalous observations , the implied subsampling quantile can collapse with just 5 outliers out of 100 observations .
, the @xmath128-quantile breakdown point of the overlapping subsampling is 0.23 in i.i.d . settings . since in a time series setting the number of possible subsampling blocks of size @xmath60 is typically lower than the number of i.i.d .
subsamples of size @xmath60 , the breakdown of a statistic in one random block tends to have a larger impact on the subsampling quantile than in the i.i.d . case . ]
similar results arise for the bootstrap quantiles . even though the bounds are less sharp than for the subsampling , quantile breakdown points are again clearly smaller than the breakdown point of the statistic used . , the @xmath128-quantile breakdown point for time series is less than @xmath131 , but it is @xmath132 for i.i.d .
settings , from the results in camponovo , scaillet and trojani ( 2012 ) . ]
overall , the results in theorem [ bsubboot ] imply that subsampling and bootstrap tests for time series feature an intrinsic non - resistance to anomalous observations , which can not be avoided , simply by applying conventional resampling approaches to more robust statistics .
when using a robust statistic with large breakdown point , the bootstrap and subsampling still imply an important nonresistance to anomalous observations , which is consistent with our monte carlo results in the predictive regression model . to overcome the problem ,
it is necessary to introduce a novel class of more robust resampling tests in the time series context .
section [ rpr ] introduces our robust bootstrap and subsampling approaches , and section [ robb ] demonstrates theoretically their favorable breakdown properties .
section [ near ] characterizes the asymptotic validity of the robust subsampling in both stationary and nonstationary settings .
finally , in section [ mcrs ] we study the accuracy of our approach through monte carlo simulations .
we develop a new class of easily applicable robust resampling tests for the null hypothesis of no predictability in predictive regression models . to this end , first we focus on robust estimators with nontrivial breakdown point @xmath111 .
several such estimators are available in the literature . among those estimators ,
a convenient choice is the huber estimator of regression , which ensures together good robustness properties and moderate computational costs .
let @xmath133 and @xmath134 , given a positive constant @xmath135 , the robust huber estimator @xmath136 is the @xmath137-estimator that solves the equation @xmath138 where the functions @xmath139 and @xmath140 are defined as @xmath141 in equation ( [ huber1 ] ) , we can write the huber estimator @xmath142 as a weighted least squares estimator with data - driven weights @xmath140 defined by ( [ huberw ] ) . by design , the huber weight @xmath143 reduces the influence of potential anomalous observations on the estimation results . equation ( [ huber1 ] ) is an estimating function and not the way we define the predictive relationship .
weights below one indicate a potentially anomalous data - point , while weights equal to one indicate unproblematic observations for the postulated model .
therefore , the value of weight ( [ huberw ] ) provides a useful way for highlighting potential anomalous observations that might be excessively influential for the fit of the predictive regression model ; see , e.g. , hampel , ronchetti , rousseeuw and stahel ( 1986 ) .
constant @xmath144 is useful in order to tune the degree of resistance to anomalous data of estimator @xmath136 in relevant applications , and it can be determined in a fully data - driven way . and the degree of robustness @xmath135 . ]
note that the norm of function @xmath145 in equation ( [ huber1 ] ) is bounded ( by constant @xmath135 ) , and the breakdown point of estimator @xmath136 is maximal ( @xmath126 , see , e.g. , huber , 1981 ) . conventional bootstrap and subsampling
solve equations @xmath146 , with @xmath147 ( bootstrap ) and @xmath148 ( subsampling ) for each random bootstrap sample @xmath149 and subsampling random sample @xmath150 , respectively , which can be a computationally demanding task .
instead , we consider a standard taylor expansion of ( [ huber1 ] ) around the true parameter @xmath151 , @xmath152^{-1}\psi_{n}(z_{(n)},\theta_{0})+o_{p}(1),\ ] ] where @xmath153 is the derivative of function @xmath145 with respect to parameter @xmath154 . based on this expansion , we can use @xmath155^{-1}\psi_{k}(z_{(k)}^{\ast},\hat{\theta}_{n}^r)$ ] as an approximation of @xmath156 in the definition of the resampling scheme estimating the sampling distribution of @xmath157 .
this approach avoids computing @xmath158 in random samples , which is a markable computational advantage that produces a fast numerical procedure .
this is an important improvement over conventional resampling schemes , which can easily become unfeasible when applied to robust statistics .
let @xmath159^{-1 } \left(\frac{1}{n}\sum_{t=1}^{n } g(z_t,\hat{\theta}_n^r)g(z_t,\hat{\theta}_n^r)'\cdot h_c(z_t,\hat{\theta}_n^r)^2\right ) [ \nabla_{\theta}\psi_{n}(z_{(n)},\hat{\theta}_{n}^r)]^{-1 } , \\
\hat{\sigma}_{k}^{r\ast } & = & [ \nabla_{\theta}\psi_{k}(z_{(k)}^{\ast},\hat{\theta}_{n}^r)]^{-1 } \left(\frac{1}{k}\sum_{t=1}^{k } g(z_t^{\ast},\hat{\theta}_n^r)g(z_t^{\ast},\hat{\theta}_n^r)'\cdot h_c(z_t^{\ast},\hat{\theta}_n^r)^2\right ) [ \nabla_{\theta}\psi_{k}(z_{(k)}^{\ast},\hat{\theta}_{n}^r)]^{-1},\end{aligned}\ ] ] with @xmath160 .
following this fast resampling approach , we can finally estimate the sampling distribution of @xmath161^{-1/2}(\hat{\theta}_n^r-\theta_0)$ ] with the distribution @xmath162^{-1/2}\bigg(-[\nabla_{\theta}\psi_{k}(z_{(k),i}^{\ast},\hat{\theta}_{n}^r)]^{-1}\psi_{k}(z_{(k),i}^{\ast},\hat{\theta}_{n}^r)\bigg)\le x\bigg),\ ] ] where @xmath163 denote the number of possible random samples . in the next section ,
we analyze the breakdown properties of the robust fast resampling procedure .
a closer inspection of quantity @xmath164^{-1/2}[\nabla_{\theta}\psi_{k}(z_{(k),i}^{\ast},\hat{\theta}_{n}^r)]^{-1}$ ] @xmath165 in equation ( [ fastpr1 ] ) reveals important implications for the breakdown properties of the robust fast resampling distribution ( @xmath166 ) .
indeed , this quantity can degenerate only when either ( i ) matrix @xmath167 is singular , ( ii ) matrix @xmath168 is singular or ( iii ) estimating function @xmath165 is not bounded .
however , since we are making use of a robust ( bounded ) estimating function , situation ( iii ) can not arise .
therefore , we intuitively expect the breakdown of the quantiles of robust subsampling distribution ( @xmath166 ) to arise only when conditions ( i ) or ( ii ) are realized .-
estimator with nonrobust ( unbounded ) estimating function does not solve the robustness problem , consistent with our theoretical results in section [ tlor ] .
] we borrow from this intuition and in the next theorem , we compute the quantile breakdown point of resampling distribution ( [ fastpr1 ] ) .
[ ftheo ] for simplicity , let @xmath169 .
the @xmath0-quantile breakdown points @xmath170 and @xmath171 of the robust block bootstrap and robust subsampling distributions , respectively , are given by @xmath172,\label{fbps}\\ b_{t , n , m}^{rb\ast } & = & \frac{1}{n}\bigg[\inf_{\{p\in\mathbb{n},p\le n - m+1\ } } \bigg\{m+p\bigg\vert p\bigg(bin\bigg(r,\frac{p+1}{n - m+1}\bigg)=r\bigg)>1-t\bigg\}\bigg],\label{fbpb}\end{aligned}\ ] ] where @xmath173 is a binomial random variable with parameters @xmath163 and @xmath122 .
the quantile breakdown points of the robust bootstrap and subsampling approach are often much higher than the one of conventional bootstrap and subsampling
. table [ table111 ] quantifies these differences , confirming that the robust bootstrap and subsampling quantile breakdown points in table [ table111 ] are considerably larger than those in table [ table1 ] for conventional bootstrap and subsampling methods . in this section , we show that the robust subsampling can provide an asymptotically valid method for introducing inference in both stationary and nonstationary predictive regression models .
however , to achieve this objective , the definition of robust subsampling tests requires some cares .
indeed , only tests based on symmetric robust subsampling confidence intervals ensure a correct inference . to this end ,
we consider the predictive regression model ( [ pregmodel1])-([pregmodel2 ] ) under the following assumptions .
[ asso ] @xmath174 $ ] , and @xmath12 , @xmath13 , where @xmath175 and @xmath14 are independent .
@xmath175 and @xmath14 are strictly stationary with @xmath176=0 $ ] , and @xmath177=0 $ ] . furthermore , for @xmath178 , @xmath179<\infty$ ] , and @xmath180<\infty$ ] .
@xmath175 and @xmath14 are strong mixing with mixing coefficients @xmath181 and @xmath182 , respectively , that satisfy @xmath183 , @xmath178 , @xmath184 .
@xmath185 is the unique solution of @xmath186=0 $ ] , and the set @xmath187 is compact .
assumption [ asso ] provides a set of conditions also adopted in choi and chue ( 2007 ) to prove the validity of the subsampling in nearly integrated settings , and in lucas ( 1995 ) to derive the limit distribution of robust m - estimators in integrated settings .
consider the statistic @xmath188 , where @xmath189 is the square - root of the second diagonal component of @xmath190 .
when @xmath191 , then @xmath192 converges in distribution to a standard normal , see choi and chue ( 2007 ) . on the other hand ,
when @xmath193 , the limit distribution of @xmath192 is nonstandard and depends on nuisance parameters that have to be simulated , see lucas ( 1995 ) .
because of this discontinuity in the limit distribution , conventional bootstrap methods are inconsistent .
also subsampling approximations may suffer from a lack of uniform convergence , see andrews and guggenberger ( 2009 ) . to verify the uniform validity of inference based on the robust subsampling
, we follow the same approach adopted in andrews and guggenberger ( 2009 ) , and focus on the quantiles of statistics @xmath192 , @xmath194 , and @xmath195 . more precisely , in figure [ asize ]
, we simulate the @xmath128-quantiles of the limit distribution of these statistics for different values of the degree of persistence @xmath196 , with @xmath197 $ ] , and covariance parameter of the error terms @xmath198 . in figure [ asize ]
, we can observe that the graphs of the 0.95quantile for different values of @xmath199 have similar shapes , and are monotone in @xmath135 .
in particular , the @xmath128quantiles of the limit distribution of @xmath192 and @xmath195 are decreasing , while those of @xmath194 are increasing .
therefore , using the same arguments adopted in section 7 in andrews and guggenberger ( 2009 ) , we can conclude that upper and symmetric conventional subsampling confidence intervals have correct asymptotic size , while lower and equal - tailed conventional subsampling confidence intervals have incorrect size asymptotically .
more precisely , given @xmath78 let @xmath200 denote a @xmath0-confidence interval obtained by inverting the conventional subsampling approximation of the sampling distribution of statistic @xmath195 .
then , @xmath201 , i.e. , conventional symmetric confidence intervals ensure a correct asymptotic size uniformly in the degree of persistence @xmath202 .
importantly , because of the negligible remainder term in the taylor approximation ( [ taylor ] ) , these results hold also for our symmetric robust subsampling confidence intervals .
to quantify the implications of theorem [ ftheo ] , we can study the sensitivity of confidence intervals estimated by the robust bootstrap and subsampling , with respect to contaminations by anomalous observations of increasing size . to this end
, we consider the same monte carlo setting of section [ bpointmc ] . in the last row of figure
[ sensacy ] , we plot the percentage increase of the length in the average estimated confidence interval , with respect to contaminations of the available data by a single anomalous observation of increasing size for the robust bootstrap and subsampling , respectively . in
evident contrast to the findings for conventional testing procedures , figure [ sensacy ] shows that the inference implied by our robust approach is largely insensitive to outliers , with a percentage increase in the average confidence interval length that is less than @xmath203 , even for an outliers of size @xmath204 .
the robustness of our approach has favorable implications for the power of bootstrap and subsampling tests in presence of anomalous observations .
for the same monte carlo setting of sections [ ri ] and [ rm ] , figures [ poweracy ] and [ powerstat ] show that in presence of noncontaminated samples ( straight line ) the frequencies of null hypothesis rejections of robust bootstrap and subsampling tests are again very close to those observed for nonrobust methods .
this means that the asymptotic efficiency loss of robust estimators in the absence of anomalous observations do not seem to reduce the performance of the robust bootstrap and subsampling with respect to nonrobust procedures .
however , in presence of anomalous observations ( dashed line ) , robust bootstrap and subsampling tests still provide an accurate empirical size close to the actual nominal level , as well as a power curve that is close to the one obtained in the noncontaminated monte carlo simulation .
using our robust resampling tests , we revisit the recent empirical evidence on return predictability for us stock market data from a robustness perspective .
we study single - predictor and multi - predictor settings , using several well - known predictive variables suggested in the literature , such as the lagged dividend yield , the difference between option - implied volatility and realized volatility ( bollerslev , tauchen and zhou , 2009 ) , and the share of labor income to consumption ( santos and veronesi , 2006 ) . because of the high persistence of dividend yields , in this empirical analysis we do not consider bootstrap procedures .
we compare the evidence produced by our robust subsampling tests of predictability with the results of recent testing methods proposed in the literature , including the bias - corrected method in amihud , hurvich and wang ( 2008 ) , the bonferroni approach for local - to - unity asymptotics in campbell and yogo ( 2006 ) , and conventional subsampling tests . the dividend yield is the most common predictor of future stock returns , as suggested by a simple present - value logic .
however , its forecasting ability has been called into question , e.g. , by the ambiguous empirical evidence of studies not rejecting the null of no predictability for a number of forecasting horizons and sample periods ; see , e.g. , goyal and welch ( 2003 ) , and ang and bekaert ( 2007 ) , among others . whether these ambiguous results are related to the weakness of conventional tests in detecting predictability structures masked by anomalous observations ,
is an empirical question that we can analyze using our robust testing method .
the empirical study is articulated in three parts .
section [ spm1 ] studies the forecast ability of the lagged dividend yield for explaining monthly s@xmath205p 500 index returns , in a predictive regression model with a single predictor .
this study allows us to compare the results of our methodology with those of the bonferroni approach for local - to - unity asymptotics , which is applicable to univariate regression settings .
instead , section [ tpm2 ] considers models with several predictive variables . in section [ dyvrp ] ,
we test the predictive power of the dividend yield and the variance risk premium , for quarterly s@xmath205p 500 index returns sampled at a monthly frequency in periods marked by a financial bubble and a financial crisis .
section [ dyli ] tests the predictive power of the dividend yield and the ratio of labor income to consumption for predicting quarterly value - weighted crsp index returns .
we consider monthly s@xmath205p 500 index returns from shiller ( 2000 ) , @xmath206 , where @xmath207 is the end of month real stock price and @xmath208 the real dividend paid during month @xmath0 .
consistent with the literature , the annualized dividend series @xmath209 is defined as , @xmath210 where @xmath211 is the one - month maturity treasury - bill rate .
we estimate the predictive regression model @xmath212 and test the null of no predictability , @xmath42 .
we collect monthly observations in the sample period 1980 - 2010 and estimate the predictive regression model using rolling windows of 180 observations .
table [ tablef1 ] reports the detailed point estimates and test results for the different testing procedures in the four subperiods 1980 - 1995 , 1985 - 2000 , 1990 - 2005 , 1995 - 2010 .
we find that while the robust subsampling tests always clearly reject the hypothesis of no predictability at the 5%-significance level , the conventional testing approaches produce a weaker and more ambiguous predictability evidence .
for instance , the subsampling tests can not reject @xmath110 at the @xmath44 significance level in subperiod 1985 - 2000 , while the bias - corrected method and the bonferroni approach fail to reject @xmath110 at the @xmath44 significance level in the subperiod 1995 - 2010 .
it is interesting to study to which extent anomalous observations in sample periods 1985 - 2000 and 1995 - 2010 might have caused the diverging conclusions of robust and nonrobust testing methods .
we exploit the properties of our robust testing method to identify such data points .
figure [ huberweights ] plots the time series of huber weights estimated by the robust estimator ( [ huber1 ] ) of the predictive regression model ( [ pregmodel31 ] ) .
we find that subperiod 1998 - 2002 is characterized by a cluster of infrequent anomalous observations , which are likely related to the abnormal stock market performance during the nasdaq bubble in the second half of the 1990s .
similarly , we find a second cluster of anomalous observations in subperiod 2008 - 2010 , which is linked to the extraordinary events of the recent financial crisis .
overall , anomalous observations are less than 4.2% of the whole data sample , and they explain the failure of conventional testing methods in uncovering hidden predictability structures in these sample periods .
we find that the most influential observation before 1995 is november 1987 , following the black monday on october 19 1987 . during the subperiod 1998 - 2002
, the most influential observation is october 2001 , reflecting the impact on financial markets of the terrorist attack on september 11 2001 . finally , the most anomalous observation in the whole sample period 1980 - 2010 is october 2008 , following the lehman brothers default on september 15 2008 . to investigate the potential presence of time - varying parameters in the predictive regression model ( [ pregmodel31 ] ) , we test formally for the presence of structural breaks .
we apply both the standard wald test statistic proposed by andrews ( 1993 ) , and its robust version introduced in gagliardini , trojani and urga ( 2005 ) .
asymptotic critical values of these test statistics are provided in andrews ( 1993 ) . to improve on the inference of asymptotic tests
, we also follow diebold and chen ( 1996 ) and gagliardini , trojani and urga ( 2005 ) , and implement nonrobust and robust resampling tests of structural breaks for our predictive regression model . using all methods , we never reject the null hypothesis of no structural break at the @xmath44 significance level in our sample period . therefore , the lack of predictability produced in some cases by the standard approach can not be explained by a structural break in a significant subset of the the data .
this evidence supports the presence of a small subset of influential anomalous observations as a plausible explanation for the diverging conclusions of classical and robust predictive regression methods .
finally , we study the out - of - sample accuracy of predictive regressions estimated by nonrobust and robust methods .
borrowing from goyal and welsh ( 2003 ) and campbell and thompson ( 2008 ) , we introduce the out - of - sample @xmath213 statistics , defined as @xmath214 where @xmath215 and @xmath216 are the fitted values from a predictive regression estimated up to period @xmath5 , using the robust huber estimator and the ols estimator , respectively . whenever statistic @xmath213 is positive
, the robust approach yields a lower average mean squared prediction error than the nonrobust method , providing more accurate out - of - sample forecasts .
as reported in table [ tableos ] , we obtain @xmath217 . therefore , besides the more robust in - sample results , our robust approach also yields better out - of - sample predictions . to compare the out - of - sample accuracy of the nonrobust and robust approaches with respect to the simple forecast based on the sample mean of market returns , we consider also the out - of - sample @xmath218 statistic , defined as @xmath219 where @xmath220 is the historical average return estimated through period @xmath5 , and @xmath221 .
as reported in table [ tableos ] , we obtain @xmath222 , and @xmath223 . therefore ,
both nonrobust and robust methods provide more accurate out - of - sample predictions than simple forecast based on the sample mean of market returns .
we extend our empirical study to two - predictor regression models .
this approach has several purposes .
first , we can assess the incremental predictive ability of the dividend yield , in relation to other well - known competing predictive variables .
second , we can verify the power properties of robust subsampling tests in settings with several predictive variables .
section [ dyvrp ] borrows from bollerslev , tauchen and zhou ( 2009 ) and studies the joint predictive ability of the dividend yield and the variance risk premium .
section [ dyli ] follows the two - predictor model in santos and veronesi ( 2006 ) , which considers the ratio of labor income to consumption as an additional predictive variable to the dividend yield .
we consider again monthly s@xmath205p 500 index and dividend data between january 1990 and december 2010 , and test the predictive regression model : @xmath224 where @xmath225 and the variance risk premium @xmath226 is defined by the difference of the s&p 500 index option - implied volatility at time @xmath0 , for one month maturity options , and the ex - post realized return variation over the period @xmath227 $ ] .
bollerslev , tauchen and zhou ( 2009 ) show that the variance risk premium is the most significant predictive variable of market returns over a quarterly horizon .
therefore , we test the predictive regression model ( [ pregmodel3f ] ) for @xmath228 .
let @xmath229 and @xmath230 denote the true values of parameters @xmath231 and @xmath232 , respectively .
using the subsampling tests , as well as our robust subsampling tests , we first test the null hypothesis of no return predictability by the dividend yield , @xmath233 . table [ tablef23 ] collects the detailed point estimates and testing results .
we find again that the robust tests always clearly reject the null of no predictability at the 5%-significance level .
in contrast , the conventional subsampling tests produce weaker and more ambiguous results , with uniformly lower @xmath89-values ( larger confidence intervals ) and a nonrejection of the null of no predictability at the @xmath234level in period 1994 - 2009 .
since the bonferroni approach in campbell and yogo ( 2006 ) is defined for single - predictor models , we can not apply this method in model ( [ pregmodel3f ] ) .
unreported results for the multi - predictor testing method in amihud , hurvich and wang ( 2008 ) show that for data windows following window 1993 - 2008 the bias - corrected method can not reject null hypothesis @xmath235 at the 10% significance level . by inspecting the huber weights ( [ huberw ] ) , implied by the robust estimation of the predictive regression model ( [ pregmodel3f ] )
, we find again a cluster of infrequent anomalous observations , both during the nasdaq bubble and the recent financial crisis . in this setting , the most influential observation is still october 2008 , reflecting the lehman brothers default on september 15 2008 .
table [ tablef23 ] reports the estimates and testing results for parameter @xmath230 .
in contrast to the previous evidence , we find that all tests under investigation clearly reject @xmath236 at the 5%-significance level , thus confirming the remarkable return forecasting ability of the variance risk premium noticed in bollerslev , tauchen and zhou ( 2009 ) , as well as the international evidence reported in bollerslev , marrone , xu and zhou ( 2014 ) . )
, we also consider the three - predictor model @xmath237 where @xmath238 is the detrended long - term yield , defined as the ten - year treasury yield minus its trailing twelve - month moving averages . again , using the standard subsampling and the robust subsampling , we find evidence in favor of predictability at @xmath239 significance level for the variance risk premium for the sample period 1990 - 2010 .
in contrast , all tests do not reject the null hypothesis of no predictability at @xmath44 significance level for the detrended long - term yield .
finally , both conventional and robust tests reject the null hypothesis of no predictability at the @xmath239 significance level for the dividend yield .
the comparison of these empirical results with those obtained in the two - predictor model ( [ pregmodel3f ] ) again confirms the reliability of our robust tests and the ( possible ) failure of nonrobust procedures in uncovering predictability structures in presence of anomalous observations . ] finally , also for this predictive regression model , we do not find evidence of structural breaks at the @xmath44 significance level .
moreover , we obtain out - of - sample statistics @xmath240 and @xmath241 , indicating again an improved out - of - sample predictive power for our robust approach .
we finally focus on the two - predictor regression model proposed in santos and veronesi ( 2006 ) : @xmath242 where @xmath243 is the share of labor income to consumption .
we make use of quarterly returns on the value weighted crsp index , which includes nyse , amex , and nasdaq stocks , in the sample period q1,1955-q4,2010 . the dividend time - series is also obtained from crsp , while the risk free rate is the three - months treasury bill rate .
labor income and consumption are obtained from the bureau of economic analysis .
let @xmath229 and @xmath230 denote the true values of parameters @xmath231 and @xmath232 , respectively .
using subsampling tests , as well as our robust testing method , we first test the null hypothesis of no predictability by the dividend yield , @xmath233 .
table [ tablef45 ] collects detailed point estimates and test results for the four subperiods 1950 - 1995 , 1955 - 2000 , 1960 - 2005 , 1965 - 2010 .
we find again that our robust tests always clearly reject @xmath244 at the @xmath239-significance level .
in contrast , conventional tests produce more ambiguous results , and can not reject at the 10%-significance level the null hypothesis @xmath244 for subperiod 1955 - 2000 .
table [ tablef45 ] summarizes estimation and testing results for parameter @xmath230 .
while the conventional tests produce a weak and mixed evidence of return predictability using labor income proxies , e.g. , by not rejecting @xmath245 at the 10%-level in subperiod 1950 - 1995 , the robust tests produce once more a clear and consistent predictability evidence for all sample periods .
the clusters of anomalous observations ( less than @xmath246 of the data in the full sample ) , highlighted by the estimated weights in figure [ huberweights2 ] , further indicate that conventional tests might fail to uncover hidden predictability structures using samples of data that include observations from the nasdaq bubble or the recent financial crisis , a feature that was noted in santos and veronesi ( 2006 ) and lettau and van nieuwerburgh ( 2007 ) from a completely different angle . in such contexts ,
the robust subsampling tests are again found to control well the potential damaging effects of anomalous observations , by providing a way to consistently uncover hidden predictability features also when the data may only approximately follow the given predictive regression model .
we do not find evidence of structural breaks in the predictive relation at the @xmath44 significance level , while we obtain an out - of - sample statistic @xmath247 , indicating that our robust approach improves the out - of - sample predictions of classical predictive regression methods .
however , in this case the out - of - sample statistic @xmath248 shows no improvement over quarterly forecasts provided by standard sample mean of market returns .
a large literature studies the predictive ability of a variety of economic variables for future market returns .
several useful testing approaches for testing the null of no predictability in predictive regressions with correlated errors and nearly integrated regressors have been proposed , including tests that rely on nonparametric monte carlo simulation methods , such as the bootstrap and subsampling .
all these methods improve on the conventional asymptotic tests under the ideal assumption of an exact predictive regression model . however , we find by monte carlo evidence that even small violations of such assumptions , generated by a small fraction of anomalous observations , can result in large deteriorations in the reliability of all these tests . to systematically understand the problem ,
we characterize theoretically the robustness properties of resampling tests of predictability in a time series context , using the concept of quantile breakdown point , which is a measure of the global resistance of a testing procedure to outliers .
we obtain general quantile breakdown point formulas , which highlight an important nonresistance of these tests to anomalous observations that might infrequently contaminate the predictive regression model , thus confirming the fragility detected in our monte carlo study .
we propose a more robust testing method for predictive regressions with correlated errors and nearly integrated regressors , by introducing a novel general class of fast and robust resampling procedures for predictive regression models at sustainable computational costs .
the new resampling tests are resistant to anomalous observations in the data and imply more robust confidence intervals and inference results .
we demonstrate by monte carlo simulations their good resistance to outliers and their improved finite - sample properties in presence of anomalous observations . in our empirical study for us stock market data , we study single - predictor and multi - predictor models , using well - known predictive variables in the literature , such as the market dividend yield , the difference between index option - implied volatility and realized volatility ( bollerslev , tauchen and zhou , 2009 ) , and the share of labor income to consumption ( santos and veronesi , 2006 ) .
first , using the robust tests we find clear - cut evidence that the dividend yield is a robust predictive variable for market returns , in each subperiod and for each sampling frequency and forecasting horizon considered .
in contrast , tests based on bias - corrections , local - to - unity asymptotics , or standard subsampling procedures provide more ambiguous findings , by not rejecting the null of no predictability in a number of cases .
second , we find that the difference between option - implied volatility and realized volatility is a robust predictive variable of future market returns at quarterly forecasting horizons , both using robust and nonrobust testing methods .
this finding confirms the remarkable return forecasting ability of the variance risk premium , first noticed in bollerslev , tauchen and zhou ( 2009 ) .
third , we find that conventional testing approaches deliver an ambiguous evidence of return predictability by proxies of labor income , which is either absent or weak in the sample periods 1955 - 2000 and 1965 - 2010 , respectively .
in contrast , the null of no predictability is always clearly rejected using the robust testing approach , indicating that the weak findings of the conventional tests are likely deriving from their low ability to detect predictability structures in presence of small sets of anomalous observations .
fourth , we exploit the properties of our robust tests to identify potential anomalous observations that might explain the diverging conclusions of robust and nonrobust methods .
we find a fraction of less than about 5% of anomalous observations in the data , which tend to cluster during the nasdaq bubble and the more recent financal crisis .
anomalous data points , including the lehman brothers default on september 2008 , the terrorist attack of september 2001 , the black monday on october 1987 , and the dot - com bubble collapse in august 2002 , are responsible for the failure of conventional testing methods in uncovering the hidden predictability structures for these sample periods .
fifth , we find that the different conclusions of our robust approach with respect to conventional methods can not be explained by the presence of time - varying predictive regression parameters , as we do not find any evidence of structural breaks in predictive relations over our sample period .
moreover , we find that the out - of - sample predictions for monthly market returns of our robust approach are more accurate than the ones given by conventional predictive regressions and sample mean market return forecasts . finally ,
while our subsampling tests have been developed in the context of standard predictive systems with autocorrelated regressors , our approach is extendable also to more general settings , including potential nonlinear predictive relations or unobserved state variables .
for instance , van binsbergen and koijen ( 2010 ) propose a latent - variable approach and a kalman filter to estimate a present value model with hidden and persistent expected return and dividend growth , in order to formulate powerful tests for the joint predictability of stock returns and dividend growth .
the application of our robust subsampling tests in the context of such present value models is an interesting avenue for future research .
we first consider the subsampling .
the value @xmath125 is the smallest fraction of outliers , that causes the breakdown of statistic @xmath58 in a block of size @xmath60 .
therefore , the first inequality is satisfied . for the second inequality ,
we denote by @xmath249 , @xmath250 and @xmath251 , @xmath252 , the nonoverlapping and overlapping blocks of size @xmath60 , respectively .
given the original sample @xmath52 , for the first nonoverlapping block @xmath253 , consider the following type of contamination : @xmath254 where @xmath255 , @xmath256 and @xmath257 , @xmath258 , denote the noncontaminated and contaminated points , respectively . by construction , the first @xmath259 overlapping blocks @xmath260 , @xmath261 , contain @xmath262 outliers .
consequently , @xmath263 , @xmath264 .
assume that the first @xmath265 nonoverlapping blocks @xmath266 , @xmath267 , have the same contamination as in ( [ co ] ) . because of this contamination , the number of statistics @xmath268 which diverge to infinity is @xmath269 .
consider @xmath266 , @xmath250 .
assume that @xmath274 of these nonoverlapping blocks are contaminated with exactly @xmath275 outliers for each block , while the remaining @xmath276 are noncontaminated ( @xmath277 outlier ) , where @xmath118 and @xmath278 , @xmath279 .
moreover , also assume that the contamination of the @xmath274 contaminated blocks has the structure defined in ( [ co ] ) .
the block bootstrap constructs a @xmath55-sample randomly selecting with replacement @xmath280 overlapping blocks of size @xmath60 .
let @xmath281 be the random variable which denotes the number of contaminated blocks in the random bootstrap sample .
it follows that @xmath282 . by equation ( [ break2 ] ) ,
@xmath283 , when the proportion of statistics @xmath284 with @xmath285 is larger than @xmath272 .
the smallest number of outliers such that @xmath285 is by definition @xmath286 .
let @xmath287 . consequently ,
@xmath288.\ ] ] this concludes the proof of theorem [ bsubboot ] .
since the estimating function @xmath145 is bounded , it turns out that @xmath289^{-1/2}[\nabla_{\theta}\psi_{k}(z_{(k),i}^{\ast},\hat{\theta}_{n}^r)]^{-1 } \psi_{k}(z_{(k),i}^{\ast},\hat{\theta}_{n}^r),\ ] ] may degenerate only when ( i ) @xmath290 or ( ii ) @xmath291 .
consider the function @xmath292 using some algebra , we can show that @xmath293 where @xmath294 denotes the @xmath295 null matrix .
it turns out that by construction the matrix @xmath296 is semi - positive definite , and in particular @xmath297 , only when @xmath298 , for all the observations @xmath299 in the random sample @xmath300 . for the original sample ,
consider following type of contamination @xmath301 where @xmath255 , @xmath302 and @xmath303 and @xmath304 , @xmath305 , denote the noncontaminated and contaminated points , respectively , where @xmath306 .
it turns out that all the @xmath307 overlapping blocks of size @xmath60 @xmath308 @xmath309 contain only outliers .
therefore , for these @xmath307 blocks we have that @xmath310 , i.e. , some components of vector ( [ term ] ) may degenerate to infinity .
moreover , @xmath311 when the proportion of statistics @xmath312 with @xmath313 is larger than @xmath272 . therefore , @xmath314 , which proves the result in equation ( [ fbps ] ) . for the result in equation ( [ fbpb ] ) , note that because of the contamination defined in ( [ cof ] ) , by construction we have @xmath307 overlapping blocks of size @xmath60 with exactly @xmath60 outliers , and @xmath315 blocks with less than @xmath60 outliers .
let @xmath281 be the random variable which denotes the number of full contaminated blocks in the random bootstrap sample .
it follows that @xmath316 . to imply ( i ) or ( ii )
, all the random observations @xmath317 have to be outliers , i.e. , @xmath318 . by equation ( [ break2 ] ) ,
@xmath319 , when the proportion of statistics @xmath320 with @xmath321 is larger than @xmath272 . consequently , @xmath322.\ ] ] this concludes the proof .
bollerslev , t. , marrone , j. , xu , l. , and h. zhou , 2014 . stock return predictability and variance risk premia : statistical inference and international evidence .
journal of financial and quantitative analysis , 49 , 633 - 661 .
donoho , d. l. and p. j. huber ( 1983 ) the notion of breakdown point , in _ a festschrift for erich l. lehmann _ by bickel , p. j. , doksum , k. a. and j. l. hodges jr .
( eds . ) , 157 - 184 , wadsworth , belmont , california .
lettau , m. , and s. van nieuwerburgh , 2007 .
reconciling the return predictability evidence : in - sample forecasts , out - of - sample forecasts , and parameter instability .
review of financial studies , 21 , 16071652 .
mancini , l. , ronchetti , e. , and f. trojani , 2005 .
optimal conditionally unbiased bounded - influence inference in dynamic location and scale models .
journal of the american statistical association , 100 , 628 - 641 .
salibian - barrera , m. , van aelst , s. , and g. willems , 2006 .
principal components analysis based on multivariate mm estimators with fast and robust bootstrap .
journal of the american statistical association , 101 , 11981211 .
@xmath323{bias90.pdf } & \includegraphics[height=1.5in , width=2in]{bias95.pdf } & \includegraphics[height=1.5in , width=2in]{bias99.pdf } \\
\includegraphics[height=1.5in , width=2in]{cy90.pdf } & \includegraphics[height=1.5in , width=2in]{cy95.pdf } & \includegraphics[height=1.5in , width=2in]{cy99.pdf } \\
\includegraphics[height=1.5in , width=2in]{sub90.pdf } & \includegraphics[height=1.5in , width=2in]{sub95.pdf } & \includegraphics[height=1.5in , width=2in]{sub99.pdf } \\
\includegraphics[height=1.5in , width=2in]{rsub90.pdf } & \includegraphics[height=1.5in , width=2in]{rsub95.pdf } & \includegraphics[height=1.5in , width=2in]{rsub99.pdf } \end{array}$ ] @xmath323{sub30.pdf } & \includegraphics[height=1.5in , width=2in]{sub50.pdf } & \includegraphics[height=1.5in , width=2in]{sub70.pdf } \\
\includegraphics[height=1.5in , width=2in]{boot30.pdf } & \includegraphics[height=1.5in , width=2in]{boot50.pdf } & \includegraphics[height=1.5in , width=2in]{boot70.pdf } \\
\includegraphics[height=1.5in , width=2in]{rsub30.pdf } & \includegraphics[height=1.5in , width=2in]{rsub50.pdf } & \includegraphics[height=1.5in , width=2in]{rsub70.pdf } \\
\includegraphics[height=1.5in , width=2in]{rboot30.pdf } & \includegraphics[height=1.5in , width=2in]{rboot50.pdf } & \includegraphics[height=1.5in , width=2in]{rboot70.pdf } \end{array}$ ] @xmath324{bias.pdf } & \includegraphics[height=1.75in , width=2.5in]{cy.pdf}\\ & \\ & \\ \includegraphics[height=1.75in , width=2.5in]{sub.pdf } & \includegraphics[height=1.75in , width=2.5in]{boot.pdf}\\ & \\ & \\ \includegraphics[height=1.75in , width=2.5in]{rsub.pdf } & \includegraphics[height=1.75in , width=2.5in]{rboot.pdf}\\ \end{array}$ ] | testing procedures for predictive regressions with lagged autoregressive variables imply a suboptimal inference in presence of small violations of ideal assumptions . we propose a novel testing framework resistant to such violations , which is consistent with nearly integrated regressors and applicable to multi - predictor settings , when the data may only approximately follow a predictive regression model .
the monte carlo evidence demonstrates large improvements of our approach , while the empirical analysis produces a strong robust evidence of market return predictability hidden by anomalous observations , both in- and out - of - sample , using predictive variables such as the dividend yield or the volatility risk premium .
* keywords : * predictive regression , stock return predictability , bootstrap , subsampling , robustness .
* jel : * c12 , c13 , g1 . |
in the late 80 s , inspired by witten s ideas @xcite , reshetikhin and turaev @xcite came up with a construction of new 3manifold invariants , known as wrt quantum invariants .
few years later , turaev @xcite formalized this construction by introducing the notion of _ modular category_. his main result is that any modular category gives rise to a quantum @xmath0manifold invariant .
a modular category is a special kind of ribbon category which has a finite set of simple objects @xmath1 , including the unit object @xmath2 , satisfying duality , domination and non - degeneracy axioms .
a ribbon category is a monoidal category with braiding , twist and compatible duality .
ribbon categories are universal receivers for invariants of ribbon graphs @xcite .
examples of modular categories arise from representation theory of quantum groups , when the quantum parameter @xmath3 is a root of unity , or can be constructed skein theoretically @xcite .
many authors @xcite , @xcite , @xcite observed independently that for some special values of @xmath3 , the @xmath4 wrt invariants admit spin and cohomological refinements .
cohomological refinements give rise to homotopy quantum field theories ( hqft s ) , constructed by turaev in @xcite .
however , spin refinements do not fit in the framework of hqft s .
the main aim of this paper is to provide an algebraic setting for spin type refinements of quantum 3manifold invariants . before explaining our results
let us recall few definitions . given a group @xmath5 ,
a @xmath5category was defined in ( * ? ? ?
* section vi ) as an additive monoidal category @xmath6 with left duality and unit object @xmath7 that splits as a disjoint union of full subcategories @xmath8 such that * @xmath9 if @xmath10 ; * for @xmath11 and @xmath12 , @xmath13 ; * @xmath14 , and for @xmath11 , @xmath15 .
we call an object @xmath16 of a modular category _ invertible _ if there exists an object @xmath17 , such that @xmath18 .
it is easy to see that isomorphism classes of invertible objects form a finite abelian group under tensor multiplication .
let us denote by @xmath19 the group of isomorphism classes of invertible objects in the modular category @xmath20 . in section 5
we show that the braiding ( or monodromy ) coefficients of @xmath21 with @xmath22 define a map @xmath23 .
this map induces on @xmath6 the structure of a @xmath24category .
note the braiding matrix defines a bilinear form on @xmath5 , and the twist coefficients extend it to a quadratic form .
a special role in our approach will play a subgroup @xmath25 of @xmath5 , such that the bilinear form restricted to @xmath25 is trivial while the quadratic extension is not .
let us state our main definition .
[ def1 ] let @xmath20 be a modular category with a group @xmath5 of invertible objects .
for any subgroup @xmath26 , we call @xmath20 @xmath25refinable if @xmath27 .
moreover , an @xmath25refinable modular category @xmath6 is called
_ @xmath25spin _ if the twist quadratic form restricted to @xmath25 is non - trivial , or equivalently if @xmath25 has at least one element with twist coefficient @xmath28 .
when @xmath25 is cyclic of order @xmath29 , we will use shorthand @xmath29spin and @xmath29refinable .
for example , the @xmath4 modular category at the @xmath30th root of unity @xmath3 is @xmath31refinable for @xmath32 and it is @xmath31spin if @xmath33 .
the group @xmath34 is generated by the @xmath35dimensional representation .
we say that @xmath22 has order @xmath36 if @xmath36 is the minimal integer such that @xmath37 .
we will see that the order of an element with twist @xmath28 has to be even .
@xmath25refinable modular categories which are non spin give rise to invariants of pairs @xmath38 for any compact orientable 3manifold @xmath39 and a cohomology class @xmath40 .
they also fit in the setting of modular group - categories introduced in @xcite . in the spin case ,
the formalism of @xcite does not strictly apply since the subcategory in trivial degree is not modular .
let us concentrate on the spin case .
if we have a @xmath25spin modular category , then the twist coefficients define an order @xmath31 element @xmath41 , which we will call the spin character . we will extend the definition of generalized spin structures with modulo @xmath29 coefficients given in @xcite to this situation and define @xmath42 generalized spin structures .
one of the results of this paper is the following .
[ main ] any @xmath25spin modular category @xmath6 with associated spin character @xmath43 provides a topological invariant @xmath44 of a pair @xmath45 , where @xmath46 is a @xmath42 generalized spin structure on @xmath39 .
moreover , @xmath47 we expect that theorem [ main ] extends naturally to a spin type tqft . in section 2 we define @xmath29_complex _ spin structures .
let us denote the set of such structures on @xmath39 by @xmath48 .
generalizing results of @xcite , we identify @xmath48 with the set of modulo @xmath29 chern vectors which are further used for constructing extensions of wrt invariants .
[ main_c ] suppose @xmath29 is an even positive integer .
for any @xmath49spin modular category @xmath6 , there exists a topological invariant @xmath44 of a pair @xmath45 , where @xmath50 is a @xmath29complex spin structure on @xmath39 .
we call a modular category @xmath6 _ reduced _ if it is @xmath51refinable
. assume the group of invertible objects @xmath52 of our modular category @xmath6 is cyclic , but @xmath53 with @xmath54 .
if @xmath55 , then there is a positive integer @xmath56 , such that @xmath57 .
clearly , @xmath6 is not reduced , but @xmath25refinable , where the subgroup @xmath25 of order @xmath58 is generated by @xmath59 .
given these data , one way to construct refined invariants is by using the @xmath58refinable structure on @xmath6 .
in section 7 we show that there is a more efficient way to compute this invariant .
namely , there always exists a smaller reduced category @xmath60 , which leads to the same invariant up to a correction term fully determined by the linking matrix . if @xmath61 , @xmath62 is particularly simple and coincides with @xmath63 where @xmath64 .
[ decomp ] let @xmath6 be a modular category with a cyclic group of invertible objects @xmath52 .
assume @xmath65 , @xmath66 and @xmath57 .
then , there exists a reduced @xmath58refinable category @xmath67 , a positive integer @xmath68 and a root of unity @xmath69 such that for any closed oriented 3manifold @xmath39 we have @xmath70 where @xmath71 is the refined murakami ohtsuki okada invariant .
we have either @xmath72 or @xmath73 , and @xmath74 is the first betti number . in the particular case , when @xmath75 , we have @xmath76 , @xmath69 is a root of unity of order @xmath56 and @xmath77 does not depend on @xmath46 so that we have @xmath78 the murakami ohtsuki
okada invariant defined in @xcite depends only on the homological information which can be obtained from the linking matrix of the surgery link .
theorem 3 generalizes the well - known decompositon results for quantum invariants stated in @xcite .
we expect that these decomposition results extend to refined tqfts .
* organization of the paper .
* after recalling the basic definitions of @xmath79structures , we give homotopy theoretical and also combinatorial descriptions of their reduction modulo @xmath29 .
further we recall the definitions of ribbon and modular categories and of quantum 3manifold invariants .
section [ graded ] deals with modular categories containing invertible objects .
the refined invariants are studied in sections [ refined ] and 7 .
the last section is devoted to the proof of theorem [ decomp ] .
in this section , given a compact orientable 3manifold @xmath39 , we define the set @xmath48 . throughout this paper
all manifolds are assumed to be compact and oriented ; all ( co)homology groups are computed with integer coefficients , unless otherwise is specified ; @xmath80 denotes the cyclic group of integers modulo @xmath81 .
complex spin structures are additional structures some manifolds can be endowed with and just like the more common spin structures , they can be seen as a generalization of orientations .
we recall basic facts and equivalent ways to define @xmath79structures following the lines of @xcite and @xcite .
let @xmath82 be an integer .
the group @xmath83 is defined as the non - trivial double cover of the special orthogonal group @xmath84 : @xmath85 @xmath86 , @xmath87 and @xmath88 the complex spin group is defined as the quotient @xmath89 where @xmath90 is generated by @xmath91 .
it follows that the map @xmath92 , defined as @xmath93)=\lambda(a)$ ] is a principal @xmath94fibration .
@xmath95 .
let @xmath96 be an @xmath36dimensional riemannian manifold and denote by @xmath97 the principal bundle of oriented orthonormal frames of @xmath96 .
[ definition ] a @xmath79structure on the manifold @xmath96 is a principal @xmath98bundle @xmath99 , together with a map @xmath100 that restricted to the fibers is @xmath101 , i.e , makes the following diagram commute , i.e , @xmath102^{(\pi , \rho ) } \ar[rr ] & & p_{{{\operatorname{spin}}}^c(n ) } \ar[dd]^{\pi}\ar[dr]^{s } \\ & & & x\\ p_{\so(n ) } \times \so(n ) \ar[rr ] & & p_{\so(n ) } \ar[ur]\\ } \ ] ] since @xmath103 is the @xmath94bundle of the unique non trivial line bundle over @xmath84 ( @xcite ) and isomorphism classes of principal @xmath94bundles over @xmath96 are in one - to - one correspondence with elements of @xmath104 , complex spin structures can be defined as cohomology classes : a @xmath79structure on the manifold @xmath96 is an element @xmath105 whose restriction to every fiber of @xmath106 is the unique non trivial element of @xmath107 , i.e , @xmath108 let @xmath109 denote the grassman manifold of oriented @xmath36-planes in @xmath110 and let @xmath111 be the universal @xmath36-dimensional oriented vector bundle over @xmath109 .
note that for discrete topological groups @xmath5 , the classifying space @xmath112 is eilenberg - maclane of type @xmath113 .
let @xmath114 be the unique ( up to homotopy ) non - homotopically trivial map from @xmath109 to the eilenberg
maclane space @xmath115 .
we fix @xmath114 in its homotopy class and we define the fibration @xmath116 as the pull - back under @xmath114 of the path - space fibration over @xmath115 . if we set @xmath117 , then : + a @xmath79structure on the manifold @xmath96 is a homotopy class of bundle maps between the ( stable ) tangent bundle @xmath118 of @xmath96 and @xmath119 .
let us denote by @xmath120 the homomorphism @xmath121)=z^2 $ ] .
then , to any @xmath79 structure @xmath46 on an @xmath36dimensional manifold @xmath96 ( in the sense of definition [ definition ] ) , one can associate a complex line bundle as follows : using the map @xmath68 , the action of the group @xmath98 on the space @xmath122 extends to an action on the product @xmath123 and we consider its orbit space @xmath124 called the _ determinant line bundle _ of @xmath46 . the chern map @xmath125 is defined as @xmath126 where @xmath127 is the first chern class of the bundle @xmath128 , and it is affine over the doubling map @xmath129 see @xcite for more details . in this subsection
we define @xmath29complex spin structures ( short @xmath130structures ) on @xmath36dimensional manifolds and we describe some of their properties .
then , we focus on dimension three and we present a set of refined kirby s moves for @xmath130manifolds obtained by surgery along links in @xmath131 .
let @xmath132 be the bockstein homomorphism associated to the exact sequence of groups : @xmath133 the following lemma will help justify our construction of @xmath130structures .
the group @xmath134 is cyclic of order @xmath31 generated by @xmath135 , where @xmath136 is the second stiefel
whitney class .
since the elements of the group @xmath137 are in one - to - one correspondence with homotopy classes of maps in @xmath138 $ ] , there is a unique ( up to homotopy ) non - homotopically trivial map
@xmath139 we fix @xmath140 in its homotopy class and define the fibration @xmath141 as the pull - back of the path space fibration of @xmath142 under the map @xmath140 : @xmath143^{\pi_d } \ar[r ] & \pk(\z_d,3 ) \ar[d ] \\
\bso(n)\ar[r]^{g } & \k(\z_d,3 ) \\ } \ ] ] we set @xmath144 .
note that another choice of the map @xmath140 in its homotopy class yields a different , but homotopy equivalent , space @xmath145 .
let @xmath96 be an @xmath36dimensional riemannian manifold .
[ bundlemap ] a @xmath29complex spin structure on @xmath96 is a homotopy class of a bundle map between the ( stable ) tangent bundle @xmath118 of @xmath96 and @xmath146 .
[ lift ] a @xmath29complex spin structure on @xmath96 is a homotopy class of a lift @xmath147 of @xmath148 to @xmath145 , where @xmath149 is a classifying map for the bundle @xmath118 .
@xmath150^{\pi_d } \\ x \ar[r]^{f } \ar@{.>}[ru]^{\overline{f } } & \bso(n ) } \\\ ] ] since the fiber of @xmath151 is the eilenberg - maclane space @xmath152 , there is a unique obstruction @xmath153 to the existence of lifts @xmath147 and this obstruction lies in the group @xmath154 . note that the universal obstruction @xmath155 ( obtained from @xmath153 by setting @xmath156 and @xmath157 ) is non - zero , therefore not all manifolds can admit @xmath130structures . to see this ,
assume the contrary .
then , the fibration @xmath141 has a section @xmath158 and the map @xmath159 lifts @xmath140 to the contractible space @xmath160
. therefore @xmath140 must be null - homotopic which contradicts the choice we made for @xmath140 .
@xmath161^{k}\ar[d]^{\pi_d } \ar[r ] & \pk(\z_d,3 ) \ar[d]^{p } \\
\bso(n)\ar@/^/[u]^{s}\ar[r]^{g } & \k(\z_d,3 ) } \\\ ] ] as a consequence , the universal obstruction @xmath162 is the generator @xmath135 of @xmath134 .
the set @xmath163 of @xmath29complex spin structures on a manifold @xmath96 is non - empty , if and only if , @xmath164 . if non - empty , the set @xmath163 is affine over @xmath165 ( @xmath166 is the second stiefel
whitney class of @xmath96 ) .
@xmath167^{\pi_d } \\ x \ar[r]^{f } \ar@{.>}[ru]^{\overline{f } } & \bso(n ) } \\\ ] ] let @xmath148 be the classifying map of the ( stable ) tangent bundle of @xmath96 .
then , the lifts @xmath147 of @xmath148 to @xmath168 are in one - to - one correspondence with the sections of the bundle @xmath169 .
the obstruction @xmath153 to the existence of such sections is a characteristic cohomology class so by its naturality property we have @xmath170 the above relation together with the naturality of bockstein homomorphisms imply the result .
the second part of the theorem follows by standard arguments of obstruction theory .
let us consider a manifold @xmath96 with boundary .
then , any section of @xmath171 transverse to @xmath172 and oriented outwards gives rise to a homotopy class of isomorphisms between the oriented vector bundles @xmath173 and @xmath171 .
therefore , there is a well - defined restriction map @xmath174 affine over the map @xmath175 induced by the inclusion @xmath176 .
we have seen in section [ equiv ] that the fibration @xmath116 is defined as the pull - back under @xmath114 of the path - space fibration over @xmath115 .
the modulo @xmath29 restriction morphism @xmath177 induces a map @xmath178 on the level of eilenberg - maclane spaces and further , via @xmath179 , a natural map @xmath180 . as a result
, there exists a well - defined natural map @xmath181 affine over @xmath182 induced by @xmath69 . @xmath183\ar[rr]\ar[dd]^{\pi } & & \pk(\z,3)\ar @{->}[dd]|!{[d];[d]}\hole \ar[dr ] & \\ & & & & \bspin^c_d(n)\ar[dd]^(.65){\pi_d}\ar[rr ] & & \pk(\z_d,3)\ar[dd]\\ x\ar[rrr]^{f } & & & \bso(n ) \ar @{=}[dr ] \ar @{->}[rr]^(.35){h}|!{[r];[r]}\hole & & \k(\z,3)\ar @{->}[dr]^{\xi _ * } & \\ & & & & \bso(n ) \ar[rr]^{g } & & \k(\z_d,3)}\ ] ] let @xmath184 be an oriented framed link in @xmath131 with linking matrix @xmath185 and denote by @xmath186 the @xmath0manifold obtained by surgery .
the manifold @xmath186 is the boundary of a @xmath187manifold @xmath188 , constructed from a @xmath187ball @xmath189 by attaching @xmath36 @xmath31handles @xmath190 along embeddings of @xmath191 in concordance with the orientation and framing of each component @xmath192 .
the @xmath187manifold @xmath188 is sometimes called _ trace of surgery_. let us define the set @xmath193 whose elements will be called _ modulo _ @xmath29 _
chern vectors_. [ scdk ] there is a canonical bijection @xmath194 we have seen in the previous section that @xmath79structures induce @xmath29complex spin structures .
in particular , the map @xmath195 is surjective since it is affine over the surjective map @xmath196 induced by restriction modulo @xmath29 of coefficients and similarly , the restriction map @xmath197 is surjective since it is affine over the surjective map @xmath198 induced by inclusion . with the help of @xmath30 and @xmath199
, we define the map @xmath200 as follows : any @xmath201 is the image of an element @xmath202 under the composition @xmath203 .
the value @xmath204 is characteristic for @xmath205 ( see @xcite ) , therefore @xmath206 with @xmath207 .
we set @xmath208 to be the image of @xmath209 @xmath210 in @xmath211 .
* @xmath208 is well - defined : let us assume that @xmath212 contains two different elements @xmath213 and @xmath214 .
then , they differ by @xmath215 whose modulo @xmath29 reduction belongs to @xmath216 in @xmath217 .
since @xmath218 @xmath219 belongs to @xmath220 in @xmath221 , we get that @xmath222=[c(\tilde{\sigma_1})+2y]=[c(\tilde{\sigma_1})]$ ] ( the first equality follows from the fact that the chern map is affine over the doubling map ) . * @xmath223 is injective : let us assume that @xmath224 . then , the preimages @xmath213 and @xmath214 ( of @xmath225 and @xmath226 , respectively ) under @xmath203 differ by @xmath215 such that @xmath218 @xmath227 belongs to @xmath220 in @xmath221 .
this implies that @xmath228 @xmath229 belongs to @xmath216 in @xmath217 and therefore @xmath230 . the sets @xmath231 and @xmath211 have the same cardinality ,
hence @xmath223 is bijective . from now on , we will reffer to the set @xmath211 as the combinatorial description of @xmath29complex spin structures on the surgered manifold @xmath186 . a celebrated theorem of kirby @xcite states that two ( oriented ) framed links in @xmath131 produce the same manifold by surgery , if and only if , they are related by a finite sequence of local geometric transformations called _
kirby moves_. in what follows , we present a refined version of the original kirby theorem for manifolds equipped with @xmath130structures .
kirby moves ( a ) stabilization ( b ) handle slide ( c ) orientation reversal .
note that we use the blackboard framing and the labels refer to modulo @xmath29 chern vectors . ]
[ kirby ] let @xmath232 and @xmath233 be two oriented framed links with chern vectors @xmath46 , @xmath234 . then
, the manifolds @xmath235 and @xmath236 are @xmath130homeomorphic , if and only if , the pairs @xmath232 and @xmath233 are related by a finite sequence of the moves in figure [ k1 ] or their inverses . we must check that for any of the kirby moves ( stabilization , handle slide and orientation reversal ) @xmath237 , the chern vectors change under the homeomorphism @xmath238 in the way described by figure [ k1 ] . to do this , note that the map @xmath239 defined in the proof of theorem [ scdk ] is surjective and the following diagram commutes : @xmath240^{r\circ r_d } \ar[rr]^{k } & & { { \operatorname{spin}}}^c(w_{l ' } ) \ar[d]^ { r'\circ r'_d}\\ { { \operatorname{spin}}}^c_d(s^3(l))\ar[rr]^{k } & & { { \operatorname{spin}}}^c_d(s^3(l ' ) ) . \\ } \ ] ] since @xmath79structures on @xmath188 are combinatorially described as the elements of the set @xmath241 and their change under kirby moves is known ( @xcite ) , the conclusion follows .
in @xcite the second author introduced spin structures modulo an even integer @xmath29 ( short @xmath242structures ) . in dimension three , he gave a combinatorial description of such structures as well as a refined set of kirby moves . in this section
we recall his results using the notations of section [ combinatorial ] and extend the definition to a possibly non cyclic group of coefficient @xmath243 with distinguished order @xmath31 element @xmath244 .
consider the set @xmath245 the elements of @xmath246 are called _ modulo _ @xmath29 _ characteristic solutions of _ @xmath205 .
there is a canonical bijection @xmath247 given a @xmath242structure @xmath46 on @xmath186 it can be extended to @xmath188 if and only if a certain cohomology class in @xmath248 vanishes .
we denote this class by @xmath249 and we call it relative obstruction . to any @xmath250
there is associated a relative obstruction @xmath249 in @xmath251 . since the group @xmath251 is free of rank @xmath36 , taking the coefficients of the relative obstruction we obtain a map @xmath252 .
we will show that this map is injective and its image coincides with @xmath246 .
let us consider the embedding @xmath253 .
then , the relative obstruction @xmath254 , where @xmath255 is the induced map on the level of cohomology . given an integral @xmath31cycle @xmath81 , we denote by @xmath256 its self - intersection number and by @xmath257_m$ ] its modulo @xmath56 restriction . the second stiefel
whitney class @xmath258 is defined by the following equation : @xmath259_2\ra = x\cdot x \pmod 2 , \forall x\ ] ] therefore , the relative obstruction is defined by @xmath260_d\ra=\frac{d}{2 } x\cdot x \pmod d , \forall x.\ ] ] using functoriality and writing @xmath249 in the preferred basis of @xmath261 , the result follows .
the set @xmath246 will be referred to as a combinatorial description of @xmath242structures on the surgered manifold @xmath186 .
let @xmath243 be a finite abelian group and @xmath263 a non trivial element in the @xmath31-torsion subgroup @xmath264 , then we can define @xmath262 generalized spin structures . for @xmath265 , an @xmath262 generalized spin structure on an @xmath266 principal bundle
@xmath267 is a cohomology class @xmath268 whose restriction to the fiber is equal to @xmath263 in @xmath269 .
a @xmath262 generalized spin structure on an oriented manifold of dimension @xmath270 is a @xmath262 spin structure on its oriented framed bundle . as usual this definition
can be extended to dimensions less than 3 by using stabilization .
if @xmath243 is decomposed as a product of cyclic groups , @xmath271 , then a @xmath262 generalized spin structure is a sequence @xmath272 where @xmath273 is either a @xmath274spin structure or a mod @xmath274 cohomology class , depending on the twist coefficient of the corresponding generator
. refined @xmath242kirby moves ( a ) stabilization ( b ) handle slide ( c ) orientation reversal .
note that we use the blackboard framing and the labels refer to modulo @xmath29 characteristic solutions of @xmath205 . ]
the second author proved the following result : @xcite let @xmath275 and @xmath276 be two oriented framed links with characteristic solutions @xmath277 .
then , the manifolds @xmath278 and @xmath279 are @xmath242homeomorphic , if and only if , the pairs @xmath275 and @xmath276 are related by a finite sequence of the moves in figure [ k2 ] or their inverses .
we introduce basic notions about ribbon categories following @xcite . without loss of generality , we work with monoidal categories that are strict ( according to mac lane s coherence theorem @xcite , every monoidal category is equivalent to a strict one ) , i.e , categories with tensor product and unit object @xmath7 satisfying @xmath280 for any objects @xmath281 and morphisms @xmath282 of the category .
braiding _ in a monoidal category @xmath6 is a family of natural isomorphisms @xmath283 where @xmath284 run over all the objects of @xmath6 , such that @xmath285 and @xmath286 .
let us consider the category @xmath287 of finite dimensional vector spaces over a field @xmath288 .
the family of maps @xmath289 for all @xmath290 defines a braiding .
a _ twist _ in a monoidal category @xmath6 with braiding @xmath291 is a family of natural isomorphisms @xmath292 where @xmath293 runs over all the objects of @xmath6 , such that for any two objects @xmath284 of @xmath6 , we have @xmath294 in the category @xmath295 , the family @xmath296 of maps @xmath297 defines a twist .
let @xmath6 be a monoidal category .
assume that to each object @xmath293 of @xmath6 there are associated an object @xmath298 and two morphisms @xmath299 .
the rule @xmath300 is called _ duality _ in @xmath6 if @xmath301 and @xmath302 for all objects @xmath293 of @xmath6 .
the duality is called _ compatible _ with the braiding @xmath291 and the twist @xmath303 if in addition , for any object @xmath293 of @xmath6 we have @xmath304 for example , in @xmath287 , if we set @xmath305 , fix a basis @xmath306 of @xmath293 with @xmath307 , and denote by @xmath308 its dual , we can define the map @xmath309 by @xmath310 and @xmath311 by @xmath312 then , the rule @xmath300 defines a duality in @xmath313 , compatible with the braiding @xmath291 and twist @xmath303 defined above . a monoidal category @xmath6 equipped with braiding @xmath291 , twist @xmath303 and a compatible duality @xmath314
is called _ ribbon_. 1 .
the category @xmath315 defined above is ribbon ; 2 .
let @xmath316 be the quantum group corresponding to a semi - simple lie algebra @xmath317 , then the category of its finite dimensional representations @xmath318 is ribbon @xcite . in ribbon categories
we can define traces of morphisms and dimensions of objects as follows .
[ qtrace ] 1 .
the _ quantum trace _ of an endomorphism @xmath319 is the composition @xmath320 2 .
the _ quantum dimension _ of an object @xmath293 is @xmath321 in any ribbon category @xmath6
a. @xmath322 for any morphisms @xmath323 and @xmath324 ; b. @xmath325 for any morphisms @xmath323 and @xmath326 ; c. @xmath327 for any two objects @xmath284 ; d. @xmath328 for any object @xmath293 .
a ribbon graph is a compact , oriented surface in @xmath329 $ ] decomposed into bands , annuli and coupons .
bands are homeomorphic images of the square @xmath330\times [ 0,1]$ ] , annuli are homeomorphic images of the cylinder @xmath331 $ ] and coupons are bands with a distinguished base , called bottom ( the oposite base is called top ) .
isotopy of ribbon graphs means isotopy in the strip @xmath329 $ ] , constant on the boundary intervals and preserving the decomposition into annuli , bands and coupons as well as preserving orientations .
given a ribbon category @xmath6 , a ribbon graph is @xmath6_colored _ if each band and each annulus is equipped with an object of @xmath6 , called _ color _ , and each coupon is colored by a morphism in @xmath6 .
isotopy of colored ribbon graphs means color - preserving isotopy . starting with a ribbon category @xmath6
, one can construct the category @xmath332 of @xmath6colored ribbon graphs as follows : the objects are finite sequences of end points of ribbon graphs colored with objects of @xmath6 or their duals , according to the orientation of strands , and the morphisms of @xmath332 are isotopy classes of @xmath6colored ribbon graphs ( see figure [ mf ] ) .
the category @xmath332 can be given a tensor product , a natural braiding , twist and compatible duality and it becomes in this way ribbon .
[ mf ] ] ribbon categories turn out to play an important role in the theory of link invariants .
the following theorem , due to turaev , describes a way to associate to any ribbon category , invariants of colored ribbon graphs and in particular , of colored ribbon tangles ( i.e. graphs without coupons ) .
[ functor ] let @xmath6 be a ribbon category with a braiding @xmath291 , a twist @xmath303 and a compatible duality @xmath314 .
then there exists a unique covariant functor @xmath333 preserving the tensor product such that ( see figure [ mf ] ) : 1 .
@xmath334 , @xmath335 ; 2 . for any objects @xmath284 of @xmath6
, we have @xmath336 3 . for any elementary @xmath6colored ribbon graph ( ribbon graph with one coupon )
@xmath337 , we have @xmath338 where @xmath148 is the color of the only coupon of @xmath337 .
the functor @xmath339 has the following properties : @xmath340 @xmath341 @xmath342 in what follows we will use pictures to illustrate identities in a ribbon category . when drawing a pictorial identity , we always mean the corresponding morphisms in the ribbon category ( see figure [ formula ] for an example ) . [ formula ] ] [ or ] let @xmath343 be a @xmath6-colored ribbon graph that contains an annulus @xmath344
. if @xmath345 is the ribbon graph obtained from @xmath343 by reversing the orientation of @xmath344 and replacing the color of @xmath344 with its dual object , then @xmath346 let @xmath288 be a field of zero characteristic .
a ribbon category is called @xmath288linear or pre - additive if the hom sets are @xmath288vector spaces , composition and tensor product are bilinear and @xmath347 .
an object @xmath17 of the category is called _ simple _ if the map @xmath348 from @xmath349 to @xmath350 is an isomorphism
. we will denote by @xmath351 the additive closure of a pre - additive ribbon category @xmath6 , which admits direct sums of objects and compositions of morphisms are modelled on the matrix multiplication ( @xcite ) .
a modular category over the field @xmath288 is a @xmath288linear ribbon category @xmath6 with a finite set of simple objects @xmath1 that satifies the following axioms : 1 . _
normalization _ : the trivial object @xmath7 is in @xmath1 ; 2 .
_ duality _ : for any object @xmath352 , its dual @xmath353 is isomorphic to an object in @xmath1 ; 3 .
_ domination _ : for any object @xmath293 of the category , there exists a finite decomposition @xmath354 with @xmath355 ; 4 . _
non - degeneracy _ : the matrix @xmath356 is invertible over @xmath288 , where @xmath357 is the endomorphism of the trivial object associated with the @xmath358colored , @xmath359framed hopf link with linking @xmath360 . and colors @xmath361 and @xmath362 the category is called pre - modular if we remove the last axiom . replacing @xmath6 with its additive closure
, we can reformulate the domination axiom as follows : for any @xmath363 there exist positive integers @xmath364 called _ structure constants _ , such that @xmath365 .
the domination axiom says that any object decomposes into a direct sum of simple ones .
let @xmath366 and @xmath3 be an @xmath367th root of unity .
then , there exists an associated modular category @xmath368 ( @xmath369 ) with simple objects given by partitions @xmath17 from @xmath370 see e.g. @xcite for more details .
[ mainex1 ] in what follows , we describe basic properties of pre - modular categories that will be relevant to the rest of the paper .
we follow the lines of @xcite . unless otherwise stated , @xmath6 is a pre - modular category over a field @xmath288 with zero characteristic and @xmath337 is the set of representatives of simple objects .
a. an object @xmath17 of @xmath6 is called _ transparent _ if for any object @xmath371 the following morphisms in @xmath6 are equal + b. a morphism @xmath372 is called
_ negligible _ if @xmath373 for any @xmath374 . c. the _ braiding coefficients _ between two objects @xmath375 are defined as a collection @xmath376 for all @xmath377 such that + d. for any object @xmath352 the _ twist coefficient _
@xmath378 is defined by the equality : + [ twist ] + we discuss in detail a formula that will be used extensively throughout the paper .
let us start by fixing some notations . given objects @xmath379 , we denote by @xmath380 the @xmath288module of morphisms @xmath381 the modules @xmath382 , @xmath383 , @xmath384 , @xmath385 , @xmath386 and @xmath387
are mutually isomorphic , as well as the modules @xmath388 , @xmath389 and all obtained from them by a cyclic permutation of colors .
identifying these modules along the isomorphisms , we get a symmetrized multiplicity module @xmath390 for which only the cyclic order of colors is important
. the elements of @xmath390 are represented by a round coupon with one incoming line ( colored with @xmath391 ) and two outgoing ones ( colored by @xmath17 and @xmath371 ) .
the pairing @xmath392 defined as @xmath393 is non - degenerate since the category @xmath6 can be assumed without negligible morphisms ( if any , they can be quotiened out ) .
the symmetrized modules @xmath390 and @xmath394 are dual to each other , therefore we can choose bases @xmath395 for @xmath394 and @xmath396 for the module @xmath390 that are dual with respect to @xmath397 note that the composition @xmath398 is an endomorphism of the simple object @xmath391 , so it is of the form @xmath399 , @xmath400 . comparing the traces
gives @xmath401 so @xmath402 .
for any simple objects @xmath403 , the domination axiom applied to @xmath404 yields the following relation , known as the _ fusion formula _ : @xmath405 for any pre - modular category @xmath6 , let us define the _
kirby color _ @xmath406 as an element of @xmath407 , where @xmath408 denotes the grothendieck ring of the category @xmath409 .
[ slide ] in any pre - modular category the following handle sliding identity holds @xmath410 let @xmath6 be a pre - modular category with @xmath411 and @xmath412 . then , the morphism is non - trivial if and only if @xmath17 is transparent .
if @xmath17 is transparent , then the map @xmath413 conversely , if we assume that for some @xmath414 , @xmath415 we have @xmath416 then , by sliding any @xmath371colored strand along the @xmath417colored one , we obtain the following equalities of morphisms in @xmath6 @xmath418 and therefore , @xmath361 is transparent . [ twisted ] a pre - modular category with @xmath411 and with no non - trivial transparent objects
is modular .
let @xmath419 be a matrix with entries given by the relation @xmath420 we want to prove that the product @xmath421 , where @xmath422 is the identity matrix of size @xmath423 .
we have @xmath424 where @xmath425 is the kronecker index .
the second equality holds by isotopy , the third equality holds by the _ fusion formula _ , while the fourth equality is a consequence of the _ killing property_. the last equality can be proved using the structure of the modules @xmath426 and @xmath427 .
if @xmath6 is a modular category , then @xmath428 is invertible in @xmath288 , and hence @xmath429 . given a modular category @xmath6 and a closed @xmath0manifold @xmath430 obtained by surgery on @xmath131 along a framed link @xmath205 , whose linking matrix has @xmath431 positive and @xmath432 negative eigenvalues , we define @xmath433 where @xmath434 denotes the @xmath435framed unknot and @xmath436 is the natural ribbon functor .
[ t ] for any modular category @xmath6 , @xmath437 defines a topological invariant of @xmath39 , independent on the choice of the link @xmath205 .
we need to show that @xmath437 is well - defined and does not change under kirby moves .
the fact that @xmath438 are non - zero follows from the properties of the ribbon functor @xmath339 and the non degeneracy axiom for @xmath6 .
the first kirby move is easy to establish .
the invariance under the second kirby move is provided by proposition [ slide ] . in order to prove invariance under orientation reversal ,
let us consider the link @xmath439 obtained from @xmath205 by reversing the orientation of a component @xmath440 . without loss of generality , we may assume @xmath441 .
thanks to corollary [ or ] we have @xmath442 the linking matrix @xmath443 with @xmath444 hence the matrices @xmath445 and @xmath446 have the same eigenvalues . in particular , @xmath447 so @xmath448 .
this concludes the proof .
in this section , we show that for any modular category @xmath6 , its group of invertible objects @xmath5 defines the structure of @xmath449category on @xmath6 .
we identify an object with its isomorphism class and hence equality between objects means an isomorphism .
an object @xmath17 in @xmath1 is called _ invertible _ if there exists another object @xmath450 such that @xmath451 .
isomorphism classes of invertible simple objects form a finite abelian group under tensor multiplication .
let us denote this group by @xmath452 and let @xmath453 be the group of its characters .
we say that the _ tensor order _ of @xmath17 is @xmath29 if @xmath454 with @xmath29 minimal . for any @xmath352 and any invertible @xmath455 , @xmath456 . by the domination axiom we have @xmath457 .
multiplying this identity by the inverse of @xmath140 , we get @xmath458 . thus , the right hand is simple , and so @xmath459 has to be simple too .
in example [ mainex1 ] the object @xmath243 ( the longest row ) is invertible of order @xmath460 .
[ mainex2 ] assume @xmath6 is a modular category with @xmath5 the group of invertible simple objects .
we can define a @xmath449structure on @xmath6 as follows : given @xmath352 , the braiding coefficient of @xmath17 with elements of @xmath5 defines a map @xmath461 which associates to each @xmath17 a character @xmath462 defined by the equality : @xmath463 indeed , the previous lemma implies that the braiding operator acts on @xmath17 and @xmath140 as a multiplication by @xmath464 , i.e. only one braiding coefficient is non - zero .
using the fact that @xmath140 is of finite order , we deduce that this coefficient is a root of unity of that order .
observe that @xmath465 defines a group multiplication on @xmath24 . by taking a logarithm of the character
, we can identify @xmath5 with its dual @xmath24 and write the group operation on @xmath5 additively .
clearly , @xmath6 splits into a disjoint union of subcategories @xmath466 for @xmath467 .
moreover , for any @xmath468 , @xmath469 is either zero if @xmath470 or @xmath288 otherwise ( by the assumption these objects are simple ) .
hence , we just proved the following .
[ gcategory ] a modular category @xmath6 with a group @xmath5 of invertible objects splits as a disjoint union of subcategories @xmath471 such that * each @xmath472 is a full subcategory of @xmath6 ; * each object of @xmath6 belongs to @xmath472 for a unique @xmath473 ; * @xmath9 if @xmath10 ; * for @xmath11 and @xmath12 , @xmath474 ; * @xmath475 , and for @xmath11 , @xmath476 . note
that this is a special case of group - categories defined by turaev in @xcite . if @xmath477 , then we will call @xmath473 _ degree _ of @xmath17 and denote by by @xmath478 .
if @xmath5 is cyclic of order @xmath29 , we call @xmath6 a _ modular @xmath29category_. fixing a generator @xmath16 of @xmath5 and a primitive @xmath29th root of unity @xmath479 , we have the decomposition @xmath480 where @xmath481 .
let us assume that @xmath482 in examples [ mainex1 ] and [ mainex2 ] .
then @xmath483 is generated by @xmath243 .
hence , the category @xmath368 is a modular @xmath460-category with @xmath484 .
we will need the following lemma .
[ lem21 ] let @xmath6 be a modular @xmath29category such that the group @xmath485 .
then @xmath486 if @xmath29 is odd , and @xmath487 if @xmath29 is even . let @xmath488 be the braiding coefficient , such that using we have @xmath489 .
on the other hand , from definition [ qtrace ] we get @xmath490 , where @xmath491 , so @xmath492 .
since @xmath493 , it follows that @xmath487 .
if @xmath29 i d odd , then @xmath494 .
let us recall the following definition which will play a central role in our exposition .
[ hspin ] let @xmath26 be a subgroup .
a modular category with a group @xmath5 of invertible objects is called _ @xmath25refinable _ if @xmath495 .
if , in addition , @xmath25 has at least one element with twist coefficient @xmath28 , we call the category _ @xmath25spin_. in the case @xmath25 is cyclic of order @xmath29 , we will use the shorthand @xmath29refinable and @xmath29spin . given an even integer @xmath496 , such that @xmath497 for @xmath498 and @xmath499 , the second author constructed in @xcite a category @xmath500 with @xmath501 and @xmath502 generated by @xmath503 . here
@xmath504 denotes the transpose partition .
the second author proved that the category @xmath500 is @xmath29spin modular .
assume @xmath6 is a pre - modular @xmath29category .
for any @xmath505 , let us define the refined kirby colors @xmath506 as objects in the additive closure of @xmath6 . in any pre - modular @xmath29category @xmath6
we have the following equality of morphisms : @xmath507 with @xmath508 .
the proof is the same as in the non - graded case , using the fact that @xmath509 is zero unless @xmath510 .
hence , the sum over @xmath391 can be restricted to @xmath511 .
@xmath512 in the second and the fourth equalities we use the _ fusion formula _
, the third equality holds by isotopy . for any @xmath513 , and a primitive @xmath29th root of unity @xmath514
, we define @xmath515 the dual refined kirby color . note that @xmath516 , however other dual kirby colors depend on the choice of @xmath514 .
we use the graded sliding identity to prove the following lemma .
[ dual sliding property ] in any pre - modular @xmath29category @xmath6 we have the following equality of morphisms : a. @xmath517 b. @xmath518 \(a ) we have the following equalities : @xmath519 the second and fourth equalities follow from the _ fusion formula _ , the third equality holds by isotopy .
+ ( b ) according to corollary [ or ] , given a @xmath6colored ribbon graph @xmath343 with an annulus component @xmath344 colored with the dual kirby color @xmath520 , if we consider the ribbon graph @xmath345 obtained from @xmath343 by reversing the orientation of @xmath344 and changing its color to @xmath521 , then @xmath522 .
in particular , we have @xmath523 where the second equality is a straightforward application of ( a ) .
this section is devoted to the proof of theorem 1 .
motivated by the known examples we will first give the proof in the cyclic case . for @xmath29spin modular categories we obtain invariants of @xmath0manifolds equipped with @xmath242 structure . in cohomological case
we get refined invariants for @xmath0manifolds with modulo @xmath29 @xmath524dimensional cohomology classes .
for the rest of this section @xmath430 is a closed @xmath0manifold obtained by surgery on @xmath131 along a framed link @xmath184 whose linking matrix @xmath446 has @xmath525 positive and @xmath526 negative eigenvalues .
let @xmath29 be an even , positive integer and @xmath6 be a @xmath29spin modular category .
for any @xmath527 let us define @xmath528 [ main - spin ] for any @xmath29spin modular category @xmath6 , @xmath529 is a topological invariant of the pair @xmath530 .
moreover , @xmath531 to prove the first statement we need to show that @xmath532 is well - defined and does not change under refined kirby moves .
invariance under the first two @xmath242 kirby moves follows immediately from the _ graded sliding property _ and the next lemma which implies that @xmath533 is invertible for @xmath534 . in order to prove invariance under orientation reversal ,
let us consider the link @xmath439 obtained from @xmath205 by reversing the orientation of a component @xmath440 . without loss of generality , we may assume that @xmath441 .
we have seen in the proof of theorem [ t ] that the linking matrices @xmath445 and @xmath446 have the same eigenvalues so @xmath535 , @xmath536 . + since @xmath537 applying corollary [ or ] gives @xmath538 and therefore @xmath539 .
it remains to prove that @xmath540 if @xmath541 do not solve @xmath542 .
the proof is in three steps .
assume that the first component @xmath543 of our link is the @xmath435framed unknot .
then it can be unlinked from the rest of @xmath205 by applying the fenn - rourke move .
the graded sliding identity and lemma [ cancel ] tell us that @xmath544 should solve the above equations .
assume @xmath543 is the @xmath344framed unknot .
then we add a @xmath435framed unknot to our link ( with an invertible invariant ) and slide it along @xmath543 ( perform the inverse fenn - rourke move ) .
this changes the framing on @xmath543 by @xmath545 and allows to reduce this case to the previous one . finally , assume @xmath543 is arbitrary
. then we can unknot @xmath543 by adding @xmath435framed unknots to our link in such a way , that their linking number with @xmath543 is zero .
this again reduces the situation to the previous case .
[ cancel ] for any @xmath29spin modular category @xmath6 , @xmath546 is zero unless @xmath547 .
recall that invertible objects form an abelian group under tensor multiplication , which acts on @xmath1 .
in particular , its cyclic subgroup @xmath548 acts on each @xmath549 .
let us denote by @xmath550 the set of orbits under this action and by @xmath551 the corresponding reduced kirby color .
note that @xmath552 .
let @xmath553 be the @xmath554framed hopf link with linking matrix @xmath28 . .
] after sliding the second component of @xmath555 along the first one we get @xmath556 which is zero unless @xmath557 .
here we used that @xmath558 and @xmath559 , by lemma [ lem21 ] . since @xmath560 and @xmath561 are complex conjugate to each other , the result follows . in this subsection
we assume that @xmath6 is a non - spin @xmath29refinable modular category . the elements @xmath562 are combinatorially given by solutions of @xmath563 let @xmath205 be an oriented framed link and @xmath564 .
the usual kirby moves admit refinements for manifolds equipped with such structures as follows : * stabilization : @xmath565 ; * handle slide : @xmath566 where @xmath439 is obtained from @xmath205 by sliding component @xmath192 along @xmath567 and @xmath568 if @xmath569 and @xmath570 . here
the sign depends on whether the orientations of @xmath192 and @xmath567 match or not , respectively ; * orientation reversal : @xmath566 where @xmath439 is obtained from @xmath205 by reversing the orientation of component @xmath192 and @xmath571 if @xmath572 and @xmath573 . for any @xmath574 let us define @xmath575 [ main - coho ] for any non - spin @xmath29refinable modular category , @xmath576 is a topological invariant of the pair @xmath38 .
moreover , @xmath577 the proof is based on the following lemma .
[ vanish ] for any non - spin @xmath29refinable modular category @xmath6 , @xmath546 is zero unless @xmath578 . just as in the proof of lemma
[ cancel ] , we consider the hopf link @xmath579 with linking number @xmath28 and we slide the second component along the first one .
we get @xmath580 which is zero , unless @xmath578 .
the proof of theorem 1 follows from the two previous cases .
an @xmath42 generalized spin structure @xmath46 on @xmath430 is described by a sequence of coefficients @xmath581
satisfying a characteristic equation : @xmath582 indeed , the kirby element decomposes using @xmath583-grading , and the formula for the refined invariant given in still holds .
the condition for non vanishing in lemma [ cancel ] is @xmath584 .
the solution of these equations is a sequence of @xmath585 where the index @xmath586 runs over cyclic components @xmath587 of @xmath25 .
moreover , either @xmath588 or @xmath589 , depending on the twist coefficient of the corresponding generator . combining the two previous theorems
we get @xmath590
this section is devoted to the proof of theorem 2 . again , according to the twist coefficients , we will either get an extension of wrt invariants for @xmath0manifolds equipped with modulo @xmath29 complex spin structures or with @xmath31dimensional cohomology classes . throughout this section
@xmath6 is a @xmath49spin modular category with @xmath29 even . for any @xmath591 ,
let us define @xmath592 where the shorthand @xmath593 means that the summation is taken over all elements of @xmath221 in the equivalence class of @xmath46 .
[ main - spinc ] let @xmath29 be an even integer .
for any @xmath594spin modular category @xmath6 , @xmath595 is a topological invariant of the pair @xmath596 . in order to prove that @xmath595 is a topological invariant of the @xmath130manifold @xmath45 , we have to check invariance under the @xmath597 kirby moves of theorem [ kirby ] .
we start by checking invariance under the first kirby move .
let @xmath233 be obtained from @xmath598 by a positive stabilization .
we have that @xmath599 to compute the sum @xmath600 , we write the dual kirby color @xmath601 in terms of the refined ( graded ) kirby colors @xmath602 as follows : @xmath603 lemma [ cancel ] together with the identity @xmath604 gives @xmath605 so @xmath606 .
analogously , @xmath607 is invariant under a negative stabilization .
the invariance under the second kirby move is provided by the _
dual sliding property_. finally , we must check invariance under orientation reversal . for that , let @xmath608 be obtained from @xmath232 by changing the orientation of a component @xmath440 . without any loss of generality , we may assume that @xmath441 and , just like in the proof of theorem [ t ] , we get @xmath535 , @xmath536 .
we have that @xmath609 the first equality above is a consequence of the fact that @xmath443 , for @xmath444 while the third equality is an immediate application of corollary [ or ] .
this concludes the proof .
let @xmath29 be a positive integer and @xmath20 be a non
spin @xmath29refinable modular category .
the group @xmath610 is described combinatorially as the set @xmath611 .
the kirby moves for the pair @xmath38 where @xmath39 is obtained by surgery on a link @xmath205 and @xmath612 can be described as follows : * stabilization : @xmath613 ; * handle slide : @xmath566 where @xmath439 is obtained from @xmath205 by sliding component @xmath192 along @xmath567 and @xmath568 if @xmath614 and @xmath615 . here
the sign depends on whether the orientations of @xmath192 and @xmath567 match or not , respectively ; * orientation reversal : @xmath566 where @xmath439 is obtained from @xmath205 by changing the orientation of component @xmath192 and @xmath571 if @xmath572 and @xmath573 . for any @xmath616 let us define @xmath617 where the shorthand @xmath618 means that the summation is taken over all elements of @xmath217 in the equivalence class of @xmath114 .
[ main - cohom2 ] for any non
spin @xmath29refinable modular category @xmath6 , @xmath576 is a topological invariant of the pair @xmath38 . in order to prove that @xmath576 is a topological invariant of the manifold @xmath619 , we have to check invariance under the kirby moves listed above .
we start by checking invariance under the first kirby move .
let @xmath620 be obtained from @xmath621 by a positive stabilization .
@xmath622 we compute @xmath623 since @xmath624 unless @xmath625 . using lemma [ vanish ]
we get that @xmath626 and therefore @xmath627 .
analogously , @xmath628 is invariant under a negative stabilization .
the invariance under the second kirby move is provided by the _
dual sliding property_. finally , we must check invariance under orientation reversal . for that , let @xmath620 be obtained from @xmath621 by changing the orientation of a component @xmath440 . without any loss of generality , we may assume that @xmath441 and , just like in the proof of theorem [ t ] , we get @xmath535 , @xmath536 . since @xmath629 , it follows that @xmath630 the first equality above is a consequence of the fact that @xmath443 , for @xmath444 while the third equality is an immediate application of corollary [ or ] . this concludes the proof .
this section is devoted to the proof of theorem 3 .
reader interested in the case @xmath631 only can skip this section and consult an easy direct argument in appendix . throughout this section
@xmath29 is any positive integer , it needs not to be even .
let us recall the setting .
we assume that @xmath6 be a modular category with cyclic group of invertible objects @xmath52 .
let @xmath632 and @xmath633 .
moreover , let us split @xmath634 , such that @xmath635 , @xmath636 @xmath637 . for any @xmath638 we can choose @xmath639 where @xmath640 is a primitive @xmath29th root of unity .
let us fix the generator @xmath16 , so that @xmath641 .
the twist coefficient for @xmath16 is @xmath642 and satisfies @xmath643 .
we consider the subgroup of invertible objects @xmath644 . clearly , @xmath495 , so the modular category @xmath20 is @xmath58refinable .
the twist coefficient for the generator @xmath645 is @xmath646 .
it is equal to @xmath28 if @xmath58 is even , @xmath56 is odd and either @xmath486 and @xmath647 , or @xmath648 and @xmath649 ; it is equal to @xmath524 in all other cases . the modular category @xmath20 is @xmath58spin if @xmath58 is even , @xmath56 is odd and @xmath650 has order @xmath651 , and @xmath58cohomological in all other cases .
we now present the idea of the proof of the decomposition statement .
we define a tensor category @xmath652 with simple objects represented by @xmath653 .
the tensor product in @xmath652 mimics central extension of groups using @xmath31-cocycles .
we lift the map @xmath654 into a map @xmath655 , which plays the role of a section .
further , we extend @xmath148 into a map @xmath656 , such that @xmath657 given two elements @xmath658 and @xmath659 of @xmath660 we define their tensor product as @xmath661 .
we allow in @xmath662 direct sums of objects with homogenous @xmath148 value . for @xmath663 and @xmath664 we set @xmath665 the category
@xmath652 is a tensor category over @xmath288 with unit object @xmath666 and compatible duality .
note that @xmath667 is a left and right dual for @xmath658 if @xmath668 is choosen so that @xmath669 .
the category @xmath652 is semisimple with @xmath670 as representative set of simple objects .
the group of invertible objects is @xmath671 , generated by @xmath672 and @xmath673 .
the object @xmath674 is invertible in @xmath675 if and only if @xmath293 is invertible in @xmath20 .
we deduce the last statement . to prove semisimplicity , it is enough to decompose the tensor product of two objects objects @xmath674 , @xmath676 in @xmath677 .
we have in the category @xmath20 a decomposition @xmath678 we set @xmath679
. then we have in the category @xmath675 the following decomposition @xmath680 further , let us give @xmath662 a ribbon structure which twists the one given on @xmath20 .
the braiding is given by a formula @xmath681 with appropriate choice of a root of unity @xmath69 whose order @xmath682 if @xmath29 is even and @xmath683 if @xmath29 is odd .
using duality , the twist is then given by @xmath684 the corresponding colored link invariants @xmath685 and @xmath686 are equal up to a power of @xmath69 which is computed from map @xmath148 and linking numbers .
note that @xmath69 is choosen such that @xmath687 elements of @xmath688 become transparent .
the @xmath688category @xmath689 is premodular and can be modularized as described in @xcite .
simple objects in the modularization @xmath690 are obtained from those of @xmath689 quotienting by a free action .
the set @xmath691 of simple objects in @xmath60 has cardinality @xmath692 . below we give a detailed proof of the decomposition formula in the spin case , the cohomological cases can be proven similarly .
we consider here the spin case , which means that @xmath29 is even , @xmath56 is odd and the twist @xmath650 has order @xmath651 . the generator @xmath16 can be choosen so that @xmath693
. let @xmath694 with @xmath695 mod @xmath696 .
note that @xmath697 so that the modified braiding is well defined .
the braiding coefficients for the generators of @xmath688 are : @xmath698 @xmath699 the twist coefficients are @xmath700 @xmath701 it follows that the group of transparent objects is generated by @xmath702 which has trivial twist and quantum dimension @xmath524 . applying the results @xcite we see that the category @xmath675 is modularizable , i.e. that there exists a modular category @xmath690 and a dominant ribbon functor @xmath703 . here
the group of transparent objects acts freely on the set @xmath670 of simple objects in @xmath675 .
this is proved using the map @xmath148 and the fact that @xmath704 has order @xmath687 which is the order of @xmath702 .
hence , the simple objects @xmath705 in @xmath690 are represented by cosets in @xmath660 under this free action . in the category @xmath690
the group of invertible objects is @xmath706 .
the twist coefficient for @xmath672 is equal to @xmath28 , so the category @xmath690 is @xmath58spin .
denote by @xmath707 the kirby element in @xmath675 which represents @xmath687 times the kirby element @xmath708 in @xmath690 .
we write the graded decomposition @xmath709 so that @xmath710 moreover , for any @xmath711 mod @xmath58 , the kirby color @xmath712 in @xmath67 is represented by @xmath713 , i.e. @xmath714 indeed , the set @xmath713 consists of all @xmath715 such that @xmath716 mod @xmath717 .
there are @xmath718 such elements . acting with @xmath719
we can shift the degree of solutions by @xmath58 . in this way
we obtain all @xmath715 with @xmath720 mod @xmath58 . taking the quotient by this action we get @xmath721 .
it makes sense to evaluate both reshetikhin - turaev ribbon functors @xmath722 , and @xmath723 on @xmath675 colored links .
let @xmath430 be a @xmath0manifold given by surgery on the @xmath36-component link @xmath205 with signature @xmath724 , and @xmath73 represented by coefficients @xmath725 , @xmath726 . for objects
@xmath727 , @xmath728 , we have @xmath729 where @xmath730 .
note that the left hand side is invariant under action of @xmath702 on objects and can be used for the evaluation of the reduced invariant @xmath731 which we want to compare with @xmath732 .
we have @xmath733 @xmath734 after normalization we get @xmath735 and @xmath736 is the complex conjugate .
one can check , following the graded construction in section [ refined ] that the formula @xmath737 defines an invariant of @xmath45 .
we conclude @xmath738 in the case @xmath739 , we have @xmath76 and @xmath740 mod @xmath58 . furthermore , we can assume that our surgery presentation has even linking matrix ( the obstruction given by the spin cobordism group vanishes ) , so that @xmath741 mod @xmath742 .
decomposing @xmath743 with @xmath744 and @xmath745 we see that @xmath746 does not depend on @xmath747 .
summing over @xmath46 we get @xmath748 @xmath749 +
here we give a simple direct proof of theorem 3 in the case when @xmath631 . for readers convenience , we repeat the statement .
[ decomposition ] let @xmath6 be a modular @xmath29category with the group @xmath52 of invertible objects , such that @xmath750 , @xmath751 . then there are exists a subcategory @xmath752 and a root of unity @xmath69 , such that for any closed orientable @xmath0manifold @xmath39 @xmath753 let @xmath56 be such that @xmath754 and we set @xmath755 let @xmath67 be the full ribbon subcategory of @xmath6 generated by @xmath756 and @xmath708 be the corresponding kirby color .
let @xmath757 in this situation @xmath243 has order @xmath58 and @xmath758 moreover , consider @xmath759 such that clearly , @xmath767 since @xmath768 .
duality axiom holds since for @xmath769 , @xmath770 and hence @xmath771 .
domination follows trivially from the same property of @xmath6 .
it remains to prove the non - degeneracy .
observe that the kirby color @xmath417 decomposes as the sum @xmath772 since @xmath773 .
decomposing @xmath417 as above and using proposition [ crossing ] ( a ) we get for any @xmath774 the following equalities @xmath775 the _ killing property _ combined with proposition [ twisted ] implies non - degeneracy .
let @xmath39 be presented by an oriented framed link @xmath184 with all components @xmath192 unknotted ( such a link always exists , see @xcite ) .
then , the invariant @xmath776 if we replace @xmath777 in @xmath778 and apply proposition ( a ) we obtain : @xmath779 in particular , for @xmath534 , we have @xmath780 given the link @xmath205 with components colored by @xmath781 , @xmath782 can be computed as follows : first we make each component of @xmath205 zero framed , then we unlink the components ( using proposition ( c ) and ( d ) as many times as necessary ) .
finally , we obtain @xmath36 disjoint and unlinked copies of the zero framed unknot with colors @xmath781 and the relation : @xmath783 where @xmath446 is the linking matrix of @xmath205 .
similarly @xmath784 and the reshetikin - turaev invariant decomposes as @xmath785 note that for @xmath29 even , @xmath69 is an @xmath56th root of unity if @xmath56 is odd and a @xmath651th root of unity if @xmath56 is even and according to @xcite , @xmath786 is a topological invariant of @xmath39 independent on the choice of @xmath205 , known as the murakami ohtsuki
okada invariant .
@xmath749 c. blanchet .
_ hecke algebras , modular categories and @xmath0 manifolds quantum invariants _ , ( 2000 ) , 193223 .
c. blanchet .
_ a spin decomposition of the verlinde formulas for type a modular categories _ , , ( 2005 ) , 1 28 . | modular categories are a well - known source of quantum 3manifold invariants . in this paper
we study structures on modular categories which allow to define refinements of quantum 3manifold invariants involving cohomology classes or generalized spin and complex spin structures . a crucial role in our construction is played by objects which are invertible under tensor product .
all known examples of cohomological or spin type refinements of the witten - reshetikhin - turaev 3manifold invariants are special cases of our construction . in addition
, we establish a splitting formula for the refined invariants , generalizing the well - known product decomposition of quantum invariants into projective ones and those determined by the linking matrix . |
in recent years , experimental studies of biological macromolecules have motivated intense research in the field of statistical mechanics of single semiflexible polymer chains .
indeed , studies like force - elongation measurements of different biological ( dna , titin , tenascin ) and synthetic ( polyethylene glycol , polyvinyl alcohol ) polymers using atomic force microscopes , optical tweezers and other recently developed tools are abundant@xcite .
the measured force - elongation curve is generally fitted to the prediction of the wormlike chain model ( wcm ) of semiflexible chains@xcite , originally proposed by kratky and porod@xcite . from this fit ,
parameters like the persistence length of the biopolymer are extracted .
another kind of experiments has targeted the mechanical properties of eukaryotic cells@xcite .
these properties are determined by an assembly of protein fibers called the cytoskeleton .
this three dimensional assembly is made of the cytoskeletal polymers ( microtubules , actin filaments , etc . ) .
all these polymers are semiflexible polymers at the relevant length scales ( a few microns at most ) .
thus , the predictions of the wcm are very relevant for the understanding of the physical behavior of cytoskeletal polymers .
the wcm was originally proposed by kratky and porod in 1949@xcite and reformulated using field theoretic methods by saito _
et al_.@xcite in 1967 . in this model
, the polymer chain displays resistance to bending deformations .
this resistance is modeled using a free energy that penalizes bending the polymer backbone .
the free energy depends on parameters ( elastic constants ) that are a consequence of many short - range monomer - monomer interactions .
explicitly , the free energy is @xmath3 where @xmath4 is the vectorial field that represents the polymer chain , @xmath5 is the arc of length parameter , @xmath6 is the contour length of the polymer and @xmath7 is the bending modulus . in addition
, the local inextensibility constraint @xmath8 must be satisfied . as a consequence of the bending rigidity ,
a wormlike chain is characterized by a persistence length ( proportional to the bending modulus ) such that , if the length scale is shorter than the persistence length , then the chain behaves like a rod while , if the length scale is larger than the persistence length , then the chain is governed by the configurational entropy that favors the random - walk conformations .
the local inextensibility constraint has not allowed researchers to find an exact solution to the wcm .
indeed , the constraint @xmath8 is written using a dirac delta distribution in infinite dimensions .
depending on how the constraint is written , @xmath8 or @xmath9 , we get an edwards hamiltonian that is non - analytic or non - linear , respectively .
consequently , there is no exact solution of this model at present .
however , a few properties like the first few moments of the distribution of the end - to - end distance@xcite are known exactly .
the aforementioned complexity of the wcm has motivated many approximate treatments of semiflexible polymers .
for example , fixman and kovac@xcite developed a modified gaussian model for stiff polymer chains under an external field ( external force ) . in this approach , they computed an approximate distribution for the bond vectors from which they were able to compute the partition function and average end - to - end vector .
an alternative approach was proposed by harris and hearst@xcite who developed a distribution for the continuous model from which they were able to compute the two - point correlation function and , consequently , the mean - square end - to - end distance and radius of gyration .
another statistical property of interest has been the distribution function of the end - to - end distance or its fourier transform .
many approximations for this function have been proposed .
for example , different expansions of the distribution in inverse powers of the number of segments have been developed@xcite .
similarly , perturbations with respect to the rodlike limit have been derived@xcite .
other approaches to the distribution function of the end - to - end distance of semiflexible polymers have led to modified gaussian functions@xcite .
finally , many approximations have been proposed for the structure factor@xcite .
as stated in the previous paragraph , most of the approximate treatments of the distribution function of the end - to - end distance have been perturbative in nature .
indeed , these approximations were perturbation expansions with respect to the flexible or rigid chain limits .
a different approach to semiflexible polymers was taken by kholodenko@xcite . in this model , the euclidean version of the dirac propagator
is used to predict the conformational properties of semiflexible polymers .
in particular , the single chain structure factor has been used to describe experimental data quantitatively@xcite .
recently , winkler has proposed another treatment of semiflexible polymers@xcite . in this work ,
an approximate expression for the distribution function valid for any value of the stiffness of the polymer backbone was developed using the maximum entropy principle .
motivated by the experimental studies done on semiflexible polymers and our incomplete understanding of the properties of the wcm , we have developed a new approach that captures many physical properties of the model , like the limits of flexible and rigid polymers , exactly and provides approximate crossover behaviors for all the distribution functions . to accomplish this goal ,
we have employed a computational technique called the generalized borel transform ( gbt ) which was taken from quantum mechanics and quantum field theories@xcite .
this method computes mellin / laplace transforms exactly .
we provide a brief summary of this technique in appendix a. this paper is organized as follows . in section
ii , we evaluate the characteristic function of the wcm approximately such that some physical constraints are satisfied exactly . furthermore , we evaluate the distribution function ( polymer propagator ) using the gbt and compute the single chain structure factor . in section iii
we discuss the results of our calculations which are valid for any value of the semiflexibility of the polymer .
section iv contains the conclusions of our work and some speculations about applications to polymer physics of possible extensions of the gbt .
the details of the mathematical calculations are presented in the appendices .
consider a polymer chain modeled as a sequence of @xmath10 bond vectors @xmath11 connected in a sequential manner .
in addition , let us assume that the length of each bond vector is @xmath12 ( = kuhn length ) and that pairs of consecutive bond vectors try to be parallel to each other .
this preferential orientation is modeled with a boltzmann weight given by the following expression @xcite @xmath13 where @xmath14 is the strength of the interaction in units of thermal energy @xmath15 .
inserting eq .
( [ eq : potential_energy ] ) into the expression for the propagator of the random flight model@xcite , we obtain the following expression for the polymer propagator of semiflexible chains @xmath16 where @xmath17 is the end - to - end vector and @xmath18 is given by the formula @xmath19 the propagator , eq .
( [ eq : definition ] ) , is not normalized .
we proceed to express the delta function using its fourier representation@xcite then , eq .
( [ eq : definition ] ) becomes @xmath20},\end{array}\label{eq : propagator_fourier}\ ] ] which can be used to define the characteristic function , @xmath21 , as follows @xmath22 the mathematical expression of the characteristic function is @xmath23 note that since the polymer propagator , eq .
( [ eq : definition ] ) , is not normalized then the characteristic function , eq .
( [ eq : caracprim ] ) , does not approach one when the wave vector goes to zero . instead , it approaches the canonical partition function of the model . furthermore , note that the characteristic function is a fourier transform in a @xmath24-dimensional space . as stated before by yamakawa@xcite ,
the exact evaluation of the characteristic function ( or the polymer propagator ) for semiflexible chains is not possible at present .
therefore , we have developed a new approximation to evaluate this function .
this new mathematical approach was developed in such a way that the most relevant physics of the problem is not altered by the approximation .
specifically , the proposed approach keeps the thermodynamics ( partition function ) of the model exact
. moreover , all the properties of fully flexible chains @xmath25 and infinitely stiff chains @xmath26 are preserved exactly .
consequently , this approach captures both asymptotic limits exactly and provides an approximate description of the crossover behavior .
in addition , our treatment of the problem uses the exact expression of the mean - square end - to - end distance .
consequently , this quantity and the mean - square radius of gyration are exact . another important property of our approach is that it keeps the local inextensibility constraint intact .
therefore , our chains have finite extensibility and this model can be used to compute the force - elongation relationship of semiflexible polymers .
we describe our approximation hereafter .
let us start by computing the following class of integrals @xmath27 which are present in the characteristic function .
the wave vector @xmath28 is constant and can be chosen in the direction of the versor @xmath29 .
writing all the vectors in spherical coordinates we can express @xmath30 as follows @xmath31 where @xmath32 and @xmath33 are defined as follows @xmath34 the @xmath35-integrals can be done exactly . the result is @xmath36 where @xmath37 is the bessel function of second class@xcite .
after we integrate the delta function , the function @xmath30 becomes @xmath38\cos\left(\theta_{j}\right)\right\ } i_{0}\left(\alpha l^{2}\left|\sin\left(\theta_{j}\right)\sin\left(\theta_{j+1}\right)\right|\right).}\label{eq : gj3}\ ] ] we now replace this expression into eq .
( [ eq : caracprim ] ) and obtain the characteristic function @xmath39\cos\left(\theta_{j}\right)\right\ } i_{0}\left(\alpha l^{2}\left|\sin\left(\theta_{j}\right)\sin\left(\theta_{j+1}\right)\right|\right)}.\end{array}\label{eq : characti}\ ] ] the evaluation of @xmath21 is done by iterations .
first , we take the term @xmath40 , redefine @xmath41 and @xmath42 , and remove the factor @xmath43 from the definition of @xmath44 in eq .
( [ eq : gj3 ] ) ; then , we can write @xmath45\cos\left(\theta_{1}\right)\right\ } i_{0}\left(\alpha\left|\sin\left(\theta_{1}\right)\sin\left(\theta_{2}\right)\right|\right)},\label{eq : geuno}\ ] ] which is exactly doable@xcite .
the result is @xmath46 the next step in the iterative process is the evaluation of @xmath47 given by @xmath48\cos\left(\theta_{2}\right)\right\ } i_{0}\left(\alpha\sin\left(\theta_{2}\right)\sin\left(\theta_{3}\right)\right)g_{1}\left(\theta_{2}\right)}.\label{eq : g2}\ ] ] this integral is not exactly doable .
consequently , we proceed to approximate it such that the asymptotic limits of flexible and stiff polymers are captured exactly .
thus , the expression that we obtain will give an approximate crossover behavior between the aforementioned limiting regimes .
note that in the limit of very stiff chains , @xmath49 , all the segments will be parallel to each other .
in other words , when @xmath49 , @xmath50 .
then , in this limit we can say that @xmath51^{2}}.\label{eq : g2aprox}\ ] ] in the other limit , @xmath52 , @xmath53 is independent of @xmath54 .
therefore , eq . ( [ eq : g2aprox ] ) is also valid in the limit of flexible chains .
thus , we conclude that eq .
( [ eq : g2aprox ] ) is a good approximation for @xmath47 since it captures the asymptotic limits exactly and provides an approximate crossover behavior for @xmath47 .
the iteration of the aforementioned approximation @xmath55 times leads to the following expression for the characteristic function @xmath56^{n-1}}.\label{eq : chac}\ ] ] note that this expression gives the exact canonical partition function of the model @xmath57 thus , this first part of the approximation preserves both asymptotic behaviors and the thermodynamics of the problem intact .
let us now proceed to evaluate the approximate expression of the characteristic function , @xmath58 .
the integral in eq .
( [ eq : chac ] ) is not exactly doable thus , we evaluate it using a variational procedure .
let us introduce the following anzats @xmath59 where the parameters @xmath60 and @xmath61 are determined from the constraints imposed by the physics of the problem as described below .
one of the requirements is that the flexible and rigid limits are captured exactly by the model .
this requires that the parameters must behave in the following way @xmath62 using eq .
( [ eq : anzats ] ) , we can approximate the characteristic function as follows @xmath63\right\ } } } \\ { \displaystyle \times\frac{\left[\exp\left\ { \sqrt{\alpha^{2}-k^{2}\nu_{\alpha , n}^{2}}\right\ } -\exp\left\ { -\sqrt{\alpha^{2}-k^{2}\nu_{\alpha , n}^{2}}\left[1-\frac{{\displaystyle 2ikg_{\alpha , n}\cos\left(\theta_{n}\right)}}{{\displaystyle \sqrt{\alpha^{2}-k^{2}\nu_{\alpha , n}^{2}}}}\right]\right\ } \right]^{n-1}}{2^{n}\left[\sqrt{\alpha^{2}-k^{2}\nu_{\alpha , n}^{2}}\right]^{n-1}\left[1+\frac{{\displaystyle ikg_{\alpha , n}\cos\left(\theta_{n}\right)}}{{\displaystyle \sqrt{\alpha^{2}-k^{2}\nu_{\alpha , n}^{2}}}}\right]^{n-1}}}\end{array}\label{eq : characapprox}\ ] ] note that the term @xmath64 goes to zero as @xmath65 in the limit of @xmath49 and , when @xmath52 , it also approaches zero because @xmath66 .
thus , neglecting this term does not alter the predictions of the model for the flexible and stiff limits .
consequently , we approximate eq .
( [ eq : characapprox ] ) as follows @xmath67^{n-1}}{2\left[\sqrt{\alpha^{2}-k^{2}\nu_{\alpha , n}^{2}}\right]^{n-1}}}\\ { \displaystyle \times\int_{0}^{\pi}d\theta_{n}\sin\left(\theta_{n}\right)\exp\left\ { ik\cos\left(\theta_{n}\right)\left[1+\left(n-1\right)g_{\alpha , n}\right]\right\ } } , \end{array}\label{eq : characapprox1}\ ] ] which is exactly doable .
the final expression for the characteristic function is @xmath68^{n-1}}{\left[\sqrt{\alpha^{2}-k^{2}\nu_{\alpha , n}^{2}}\right]^{n-1}k\left[1+\left(n-1\right)g_{\alpha , n}\right]}\sin\left\ { k\left[1+\left(n-1\right)g_{\alpha , n}\right]\right\ } .}\label{eq : chacfinal}\ ] ] we note that the expression of the characteristic function given by eq .
( [ eq : chacfinal ] ) recovers the exact expression of the canonical partition function of the model , eq .
( [ eq : cano ] ) , in the limit of @xmath69 .
let us now proceed to determine the values of the parameters @xmath61 and @xmath60 from the physics of the problem .
we first look at the force - elongation behavior predicted by this model .
this curve is given by the following mathematical expression @xmath70\right\ } } { \partial f}=\frac{1}{k\left(if,\alpha , n\right)}\frac{\partial\left\ { k\left(if,\alpha , n\right)\right\ } } { \partial f},}\label{eq : elongacionsemiflex}\ ] ] where @xmath71 is the applied force and @xmath6 is the average end - to - end distance of the polymer chain in the direction of the force .
the physics of the problem imposes the following constraint @xmath72 which represents the finite extensibility of the polymer chain . in other words ,
the polymer chain can not be stretched more than its total contour length .
this constraint , as expressed by eq .
( [ eq : constrain ] ) , results in the following relationship between the parameters @xmath61 and @xmath60 @xmath73 which is in perfect agreement with the required asymptotic behaviors given by eq .
( [ eq : conditionasympt ] ) . equation ( [ eq : secondfit ] ) gives one of the two equations required to determine the parameters @xmath61 and @xmath60 completely .
the second equation is obtained from the mean - square end - to - end distance , @xmath74 .
we require that our approximation reproduce this statistical quantity exactly .
the exact mathematical expression of this average is @xcite , @xmath75},\label{eq : rcuadrado}\ ] ] where @xmath76 is the langevin function @xcite . in order to derive the second relationship between @xmath61 and @xmath60
, we divide the characteristic function , eq .
( [ eq : chacfinal ] ) , by the canonical partition function and expand this ratio in powers of the wave vector @xmath77 to second order .
the result is the following @xmath78^{2}+\frac{3\left(n-1\right)\nu_{\alpha , n}^{2}}{\alpha}\mathcal{l}\left(\alpha\right),}\label{eq : firstfit}\ ] ] which completes our approximation .
equations ( [ eq : secondfit ] ) , ( [ eq : rcuadrado ] ) and ( [ eq : firstfit ] ) determine @xmath61 and @xmath60 completely .
furthermore , the use of the exact expression for @xmath74 assures that our approximation predicts not only @xmath74 exactly , but also @xmath79 since they are related by the equation@xcite @xmath80 the final expression for @xmath61 is @xmath81\left(n^{2}-\left\langle r^{2}\right\rangle _ { \alpha , n}\right)}}{2\left[\left(n-1\right)^{2}+\frac{3\left(n-1\right)}{\alpha}\mathcal{l}\left(\alpha\right)\right]},}\label{eq : nu}\ ] ] and our approximation is complete . replacing the expression given by eq .
( [ eq : chacfinal ] ) into eq .
( [ eq : distri ] ) we obtain the following approximate expression for the polymer propagator @xmath82\right\ } \left[\sinh\left\ { \sqrt{\alpha^{2}-k^{2}\nu_{\alpha , n}^{2}}\right\ } \right]^{n-1}}{2\pi^{2}r\left[1+\left(n-1\right)g_{\alpha , n}\right]\left[\sqrt{\alpha^{2}-k^{2}\nu_{\alpha , n}^{2}}\right]^{n-1}}}.\label{eq : dis}\ ] ] in the limit of @xmath52 , eq . ( [ eq : dis ] ) becomes @xmath83^{n}}{k^{n-1}},}\label{eq : flex}\ ] ] which is the exact expression for the polymer propagator of the random flight model@xcite .
similarly , in the limit of @xmath49 , we can perform the following expansion valid for large values of @xmath14 @xmath84^{n-1}}{\left[\sqrt{\alpha^{2}-k^{2}\nu_{\alpha , n}^{2}}\right]^{n-1}}\simeq\frac{\exp\alpha\left(n-1\right)\exp\left[-\left(n-1\right)k^{2}\nu_{\alpha , n}^{2}/\alpha\right]}{\left[2\alpha\right]^{n-1}},}\ ] ] and compute the polymer propagator .
the result is @xmath85^{n-1}}\delta\left(r - n\right),}\label{eq : rigido}\ ] ] which is the polymer propagator of an infinitely stiff polymer chain@xcite .
this propagator is not normalized .
we now combine the approximate expression of the characteristic function , eq . ( [ eq : chacfinal ] ) , with the generalized borel transform to compute the polymer propagator and the single chain structure factor of the model .
we start the evaluation of the polymer propagator by rewriting eq .
( [ eq : dis ] ) as follows @xmath86}},}\label{eq : propa}\ ] ] where @xmath87 is defined by the mathematical expression @xmath88\left(\frac{{\displaystyle \sinh\left(\sqrt{\alpha^{2}-w^{2}\nu_{\alpha , n}^{2}}\right)}}{{\displaystyle \sqrt{\alpha^{2}-k^{2}\nu_{\alpha , n}^{2}}}}\right)^{n-1}\right].}}\label{eq : pimf}\ ] ] this integral is evaluated exactly using gbt@xcite .
a brief summary of the gbt technique can be found in appendix a. we present the most important steps of the calculation hereafter and leave all the mathematical details for appendix b. we first define an auxiliary function @xmath89 as follows @xmath90,\quad b\geq0}\label{eq : gb}\ ] ] from which the function @xmath87 can be computed as the analytic continuation to the complex plane @xmath91 consequently , @xmath87 can be evaluated from the laplace transform given by eq .
( [ eq : gb ] ) . following the technique we can write @xmath92
the @xmath93 integrations are computed using well - established properties and expansions of the functions @xmath94 and @xmath95^{-\mu}$]@xcite . after some straightforward algebra , we can write the analytical solution of @xmath89 for any even number of segments as follows @xmath96,\left[\beta+\frac{n+1}{2},\beta+\frac{n}{2}\right],-\frac{\left[\left(n-2k-1\right)\nu_{\alpha , n}\right]^{2}}{b^{2}}\right)}}{{\displaystyle \left(n-2k-1\right)^{-n-2\beta+1}2^{n-2}\gamma\left(2\beta+n\right)b}}}},\end{array}\label{eq : gbfinal}\ ] ] where @xmath97,\left[,\right],x\right)$ ] and @xmath98 are the generalized hypergeometric@xcite and gamma@xcite functions , respectively . replacing eq.([eq : gbfinal ] ) into eq.([eq : disg ] ) and computing the analytic continuation to the complex plane through the substitution @xmath99 we obtain @xmath100,\left[\beta+\frac{n+1}{2},\beta+\frac{n}{2}\right],\frac{\left[\left(n-2k-1\right)\nu_{\alpha , n}\right]^{2}}{r^{2}}\right)\right\ } } .\end{array}\label{eq : disim}\ ] ] the imaginary part of the generalized hypergeometric function is calculated using its well - known analytical properties@xcite .
this function is an analytic function for values of the modulus of the argument @xmath101 less than one and its continuation to the rest of complex plane generates a cut on the positive real axis starting at @xmath102 .
this implies that only values of the argument , @xmath103^{2}}{r^{2}}$ ] , larger or equal to one will contribute to the imaginary part of @xmath104 .
consequently , this condition reduces the number of terms in the @xmath105sum such that the last term of eq.([eq : gbfinal ] ) is @xmath106 $ ] .
the explicit evaluation of @xmath107 can be found in appendix c. finally , we replace eq.([eq : hyperim ] ) into eq.([eq : disim ] ) to obtain the exact expression for @xmath87 @xmath108}\left(-\right)^{k}\left(\begin{array}{c } n-1\\ k\end{array}\right)\sum_{\beta=0}^{\infty}\frac{\left(\alpha\right)^{2\beta}}{\left[\beta!\right]^{2}}\sum_{l=0}^{\beta}\left(\begin{array}{c } \beta\\ l\end{array}\right)}\\[0.2 in ] { \displaystyle \times\left(\frac{r}{\nu_{\alpha , n}}\right)^{l}\left(\frac{n}{2}-k-\frac{r}{2\nu_{\alpha , n}}-\frac{1}{2}\right)^{2\beta+n-2-l}\frac{{\displaystyle \left(2\beta - l\right)!}}{{\displaystyle \left(2\beta+n-2-l\right)!}}}.\end{array}\label{eq : disfin}\ ] ] note that eq .
( [ eq : disfin ] ) is just a sum of polynomials in @xmath109 .
the sum over the index @xmath77 is the one obtained for the random flight model@xcite and imposes the finite extensibility of the polymer chain .
the sums over the indexes @xmath33 and @xmath6 are a consequence of the stiffness of the polymer backbone .
the expression of @xmath87 can be rewritten as follows @xmath108}\frac{\left(-\right)^{k}}{\left(n-3\right)!}\left(\begin{array}{c } n-1\\ k\end{array}\right)\left(\frac{n}{2}-k-\frac{r}{2\nu_{\alpha , n}}-\frac{1}{2}\right)^{n-2}}\\ { \displaystyle \times\int_{0}^{1}dz\,\left(1-z\right)^{n-3}\ , i_{0}\left(2\alpha\left(\frac{n}{2}-k-\frac{r}{2\nu_{\alpha , n}}-\frac{1}{2}\right)\sqrt{z^{2}+z\frac{\frac{r}{\nu_{\alpha , n}}}{\left(\frac{n}{2}-k-\frac{r}{2\nu_{\alpha , n}}-\frac{1}{2}\right)}}\right),}\end{array}\label{eq : disfin2}\ ] ] which can be used for further approximation if so desired . the expression for @xmath87 given by eq . ( [ eq : disfin ] ) was derived for even number of segments but , its validity for odd number of segments larger than two can be proved by analytic continuation . finally , in order to obtain the polymer propagator for semiflexible chains we have to replace eq .
( [ eq : disfin ] ) into eq .
( [ eq : propa ] ) .
observe that , after the replacement , eq .
( [ eq : propa ] ) recovers the known exact solution of the random flight model@xcite when the limit @xmath110 is taken .
indeed , the only term different from zero is the one for which @xmath111 .
finally , we conclude our calculations of the wcm by computing the single chain structure factor which is defined by the following formula @xcite @xmath112
figures 1 , 2 and 3 show the prediction of the normalized polymer propagator , eq .
( [ eq : propa ] ) , as function of the end - to - end distance for chains with 5 , 10 and 30 kuhn segments , and different values of the semiflexibility parameter @xmath14 .
the numerical evaluation of the propagator was done using eq .
( [ eq : disfin ] ) .
the sum over the index @xmath33 converges quickly even for large values of the semiflexibility parameter @xmath14 . indeed , even with less than 80 terms in the sum we obtained a relative precision of @xmath113 .
the figures clearly show that the location of the peak in the polymer propagator ( multiplied by @xmath114 ) moves toward larger values of @xmath115 when the stiffness of the polymer backbone increases .
this behavior is in good qualitative agreement with previous results arising from computer simulation studies@xcite and theoretical approaches based on the maximum entropy principle@xcite .
this is the correct result because the stiffer the polymer backbone , the higher the energetic penalty to bend the chain .
consequently , those configurations of the macromolecule with small end - to - end distance will be more and more hindered as the stiffness increases while those configurations with large end - to - end distance should be more and more favored .
therefore , the peak should shift toward larger values of @xmath115 when the stiffness increases .
figure 4 shows the polymer propagator for polymer chains with 5 , 10 and 30 kuhn segments and a fixed value of the semiflexibility parameter @xmath116 .
this figure shows that the longer the polymer is , the more it behaves like a flexible chain since the location of peak ( = end - to - end distance divided by the contour length ) moves toward smaller values . in other words , the longer the polymer is , the less relevant the stiffness of the backbone becomes .
figures 5 , 6 and 7 show the behavior of @xmath117 , eq . ( [ eq : structuralfactor ] ) , as a function of @xmath77 for different values of @xmath14 and three values of @xmath10 ( 5 , 10 and 30 ) .
the figures clearly show a decrease of the single chain structure factor with increasing @xmath77 until it reaches a plateau at infinite @xmath77 .
note that our computations predict that the decrease of the single chain structure factor for small values of @xmath77 should be faster in the case of stiff polymers than in the case of flexible ones .
this is a consequence of the fact that rigid polymers have a larger radius of gyration than flexible ones for a fixed chain length .
in addition , the decrease of the structure factor for large values of @xmath77 is faster for flexible polymers than for stiff ones . indeed
, our computations predict that the structure factor goes as @xmath1 for large values of @xmath77 just before the plateau is reached . for values of @xmath10=5 , 10 and 30 , the values of @xmath2 that we got were 1 , 1.08 and 1.3 for @xmath14=10 ( rigid ) and 1.64 , 1.9 and 2 for @xmath14=0.33 ( flexible ) .
these results are in good agreement with the fact that the structure factor of polymers with large chain length should scale as @xmath118 for large @xmath77 values where @xmath119 is the fractal dimension of the object ( 2 for a flexible polymer and 1 for a rigid polymer ) .
consequently , these results imply that for short polymer chains ( @xmath120 ) , a value of @xmath121 is high enough to make this polymer behave like stiff rod , @xmath122 . on the other hand , as the chain length increases we observe that @xmath121 is not high enough to make the polymer behave as a rigid rod and deviations from the power law @xmath122 are observed . in the case of @xmath123
our calculations predict that a value of @xmath10 equal to five is not high enough to recover the scaling behavior of flexible chains , @xmath124 .
but , as the number of segments increases , the exponent approaches the value of two , indicating that the polymer chain behaves more and more like a flexible one .
this result was also showed in figure 4 .
the figures also show that the approximation we developed in the previous section gives a smooth crossover behavior from the rigid to the flexible limit . in order to make the presentation of our work more balanced and objective ,
we proceed to compare our results with the predictions of two other models .
we start with our prediction for the single chain structure factor , @xmath125 , and compare it with the expression obtained by kholodenko@xcite .
it has been shown that kholodenko s result can describe experimental data quantitatively@xcite .
thus , a comparison between our expression for the single chain structure factor and kholodenko s will help us gauge the quality of the approximations used in our treatment of the wcm .
figures 8 and 9 show the comparison for polymers with @xmath126 and two different values of the semiflexibility parameter ( shown in the plots ) .
we have checked that other values of the parameters @xmath14 or @xmath127 and @xmath10 give quantitative agreements of similar quality . in all the cases studied we found that the relationship @xmath128 always gives excellent quantitative agreement between the predictions of both models .
thus , our single chain structure factor should agree very well with the experimental data of ballauff and coworkers .
still , since both models have different origins , some very small differences can be observed in the case of flexible chains ( figure 8) .
we now proceed to compare our expression of the polymer propagator with the one computed by wilhelm and frey@xcite .
figure 10 shows this comparison for the case of polymers with @xmath120 .
the continuous curves are the results of our calculation with @xmath129 and @xmath130 .
the dashed curves were constructed based on the work of wilhelm and frey where we adjusted the bending modulus such that the location of the peak in the propagator matched our results .
this gives a better picture of the differences and similarities between both results .
figure 10 shows quantitative agreement between both results when the stiffness is low . as the stiffness increases fig .
10 shows that the qualitative behavior of both propagators is still the same .
for example , both results predict that the location of the peak moves toward larger values of the end - to - end distance and that the distribution becomes narrower . the main difference between both results is quantitative in nature .
our distribution becomes narrower than the one predicted by wilhelm and frey s work by a factor of two approximately which , in turn , generates a higher peak ( the distributions are normalized ) .
finally , let us conclude this section by rationalizing the origin of the discrepancy between the propagators and the reason why this does not affect the structure factor significantly .
we first note that differences in the propagators are to be expected because both , ours and frey s , calculations are based on different approximations and versions of the model ( continuum or discrete ) .
let us now proceed to rationalize the origin of the discrepancy between both predictions .
equations ( [ eq : distri ] ) and ( [ eq : structuralfactor ] ) define the polymer propagator and single chain structure factor in terms of the characteristic function .
observe that the propagator is a fourier transform of the characteristic function .
consequently , the oscillatory nature of the complex exponential generates partial cancellation of the contributions to the integral arising from different parts of the interval of integration .
this cancellation magnifies any inaccuracies made in the approximation of the characteristic function .
moreover , the consequences of this cancellation become more and more important as the polymer becomes stiffer because the characteristic function itself adopts an oscillatory behavior which , in principle , is out of phase with respect to the complex exponential .
therefore , the stiffer the polymer is , the more important the consequences of the approximation become . on the other hand
, the evaluation of the structure factor from the characteristic function does not involve any oscillatory function .
consequently , any small inaccuracy made in the approximation of the characteristic function will remain small in the expression of the structure factor , as shown previously .
the results obtained in this paper show that the generalized borel transform is a very useful computational tool for the statistical mechanics of single semiflexible polymer chains . indeed
, the results presented in this paper clearly show that gbt is able to compute polymer propagators for single chain problems exactly .
this capability of the technique is a direct consequence of its mathematical simplicity ( the gbt requires elements of basic calculus and some fundamental knowledge of complex variables ) .
consequently , it does not add any mathematical complexity to the physics of the starting model .
our analysis of the wormlike chain model was based on an approximate expression of the characteristic function .
the exact evaluation of this function is not possible at present .
therefore , we developed a new approximation that preserves the most relevant physical characteristics of the model intact . specifically , our approach keeps the thermodynamics of the model , the flexible and rigid limits , the mean square end - to - end distance and the finite extensibility of the model intact while providing an approximate expression for the characteristic function for intermediate values of the stiffness of the polymer chain . the polymer propagator was obtained exactly from the approximate characteristic function using gbt .
note that the propagator is approximate not because of the gbt , which computes this quantity exactly , but because of the approximate nature of the characteristic function .
our expression for the propagator shows a peak that shifts toward larger values of the end - to - end distance as the stiffness of the polymer backbone is increased , in agreement with other theoretical and computational treatments of the model .
we also found that , in the low wave vector region , the structure factor decreases faster with increasing wave vector when the stiffness increases .
this was rationalized in terms of the behavior of the radius of gyration .
similarly , we found that , in the large wave vector region , the structure factor of flexible chains decreases faster than the one of rigid polymers .
this was compared with the behavior of very long polymer chains whose behavior for large wave vectors is known exactly .
we also compared of the predictions of our calculation with established results for the single chain structure factor and polymer propagator .
excellent quantitative agreement was observed between our prediction for the single chain structure factor and the one predicted by kholodenko s model .
the polymer propagator was compared with the prediction of wilhelm and frey .
very good quantitative agreement was observed between the predictions of both models for low values of the stiffness . for stiff polymers , quantitative deviations
were observed and the origin of the deviations was rationalized .
the proposed approach to semiflexible polymers can also address semiflexible polymers with other topologies like ring and m - arm star polymers .
the procedure should be similar to the one presented in this paper but , the characteristic function will have a different mathematical expression .
we conclude this section with a discussion of the characteristic function of the wcm which limits our ability to solve this model exactly .
let us rewrite this function .
the expression is @xmath131 where we have replaced the wave vector @xmath28 by a group of wave vectors @xmath132 .
this expression is the one of the characteristic function when all the @xmath132 are equal to @xmath28 .
observe that if we replace @xmath132 by @xmath133 where @xmath134 is the imaginary unit , then @xmath135 which has the form of a laplace transform in @xmath24 dimensions .
but , the gbt computes laplace transforms very accurately or even exactly .
consequently , a generalization of gbt to many dimensions might lead to an exact or very accurate expression of the characteristic function of the wcm .
this expression can be further used to compute the polymer propagator using gbt .
thus , we speculate that such extension of the gbt technique might allow us to solve very accurately or even exactly the wormlike chain model . in general , such extension of gbt might allow us to solve other models of polymer chains of the form @xmath136\right)},\ ] ] where @xmath137 $ ] is the hamiltonian of the model .
thus , helical wormlike polymers and other models might be mathematically tractable with this generalization of the gbt .
we acknowledge the national science foundation , grant # che-0132278 ( career ) , the ohio board of regents action fund , proposal # r566 and the university of akron for financial support .
let us briefly present the mathematical aspects of the gbt in connection with the computation of the laplace - mellin transform@xcite .
we start with the expression @xmath138 where we have explicitly extracted a factor @xmath139 from the function to be transformed . defining the generalized borel transform ( gbt ) of @xmath140 as @xmath141\left[\frac{1}{\lambda\eta}+1\right]^{-\lambda s}s\left(g , a , n\right)d\left(1/\eta\right),\quad re\left(s\right)<0\label{borel}\ ] ] where @xmath142 is any real positive non - zero value and @xmath143 , we can invert eq .
( [ borel ] ) in the following way @xmath144dwdt,\label{doble}\ ] ] where the explicit expression of @xmath145 is not important for our present purposes ( for more details see ref .
@xcite ) .
the expression given by eq .
( [ doble ] ) is valid for any non - zero , real and positive value of the parameter @xmath142 .
but , the resulting expression for @xmath146 does not depend on @xmath142 explicitly .
thus , we can choose the value of this parameter in such a way that it allows us to solve eq .
( [ doble ] ) . the dominant contribution to the double integral
is obtained using steepest descent @xcite in the combined variables @xmath147 $ ] . in doing so , one first computes the saddle point @xmath148 and @xmath149 in the limit @xmath150 and then checks the positivity condition @xcite ( the hessian of @xmath151 at this point should be positive ) obtaining @xmath152=t_{o}\left(g , a , n\right)\quad,\quad w_{o}=\ln\left[x_{o}\left(g , a , n\right)\right]=w_{o}\left(g , a , n\right),\label{tw}\ ] ] where @xmath153 is the real and positive solution of the implicit equation coming from the extremes of the function @xmath151 in the asymptotic limit in @xmath142 .
therefore , one obtains the following equation @xmath154,\label{gx}\ ] ] where @xmath155}{dx_{o}}.\label{fx}\ ] ] in the range of the parameters where @xmath156 which is fulfilled when @xmath157 , and assuming that there is only one saddle point , we can retain the first order in the expansion of @xmath151 around the saddle point .
finally , we obtain the approximate expression for the starting function @xmath146 @xmath158 + 1}}{\sqrt{d\left[x_{o},a , n\right]}}\left[x_{o}\right]^{n+1}h\left[x_{o},a\right]\exp\left[-f\left[x_{o},a , n\right]\right],\label{eq : res}\ ] ] where @xmath159+f\left(x_{o},a , n\right)\left[1+f\left(x_{o},a , n\right)\right].\label{dx}\ ] ] note that the expression given by eq .
( [ eq : res ] ) is valid for functions @xmath160 that fulfill the following general conditions : \1 ) the relation given by eq .
( [ gx ] ) must be biunivocal .
\2 ) @xmath161 has to be positive and @xmath162 $ ] has to be negative in @xmath163 .
\3 ) @xmath156 .
in particular , this condition is fulfilled when @xmath164 these conditions provide the range of values of the parameters where the approximate solution , eq .
( [ eq : res ] ) , is valid . in summary
, the gbt provides an approximate solution , eq .
( [ eq : res ] ) , to amplitudes with the mathematical form given by eq .
( [ int ] ) .
the calculation consists of solving the implicit equation eq .
( [ gx ] ) for @xmath157 to obtain the saddle point and replace it into eq .
( [ eq : res ] ) .
let us now focus on amplitudes with the mathematical form of a laplace transform @xmath165 this kind of amplitudes can be mapped onto expressions of the form given by eq .
( [ int ] ) . in order to use the gbt on eq .
( [ eq : fi ] ) , we use the following relationship between eq .
( [ eq : fi ] ) and eq .
( [ int ] ) @xmath166 which can be inverted to give @xmath167 the finite sum comes from the indefinite integrations .
note that all the coefficients vanish whenever the laplace transform , eq .
( [ eq : fi ] ) , fulfills the following asymptotic behavior @xmath168 in addition , the expression , given by eq .
( [ fiapro ] ) , is valid for any value of @xmath169 in particular for @xmath164 consequently , if eq .
( [ eq : asymp ] ) , is fulfilled , then the analytical solution reads @xmath170 where , for @xmath157 , we can use the expression given by eq .
( [ eq : res ] ) .
it is important to note that it is the limit @xmath171 that makes the saddle point solution , eq .
( [ eq : res ] ) , an exact solution for eq .
( [ doble ] ) .
thus , as long as the @xmath10 indefinite integrals can be done without approximations , as it is in our case , the result for @xmath140 is exact .
we start the evaluation of the polymer propagator by rewriting eq .
( [ eq : gb ] ) as follows @xmath172\right\ } _ { c=0}}\\ { \displaystyle = \frac{\partial^{n-1}}{\partial c^{n-1}}\left\ { ga\left(b,\alpha , c\right)\right\ } _ { c=0}},\end{array}\label{eq : gb2}\ ] ] where @xmath173 is @xmath174},\label{eq : ga}\ ] ] and @xmath175 is given by the expression @xmath176 the integral expressed by eq.([eq : ga ] ) satisfies all the requirements of the gbt technique .
then , we evaluate it in the following way @xmath177 where @xmath178}.\label{eq : gaproxeneuno}\ ] ] in the asymptotic limit of @xmath179 the gbt provides an analytical solution for eq.([eq : gaproxeneuno ] ) . following the technique ,
we solve the implicit equation , eq .
( [ gx ] ) , for the saddle point @xmath180 .
the asymptotic solution is @xmath181 replacing this expression for @xmath180 in the expression provided by the gbt , eq .
( a7 ) , we obtain @xmath182 furthermore , we replace eq.([eq : gan ] ) into eq.([eq : gaprox ] ) and the resulting expression into eq.([eq : gb2 ] ) , and we exchange the order of the operators in the resulting expression . in other words , we first evaluate the @xmath10 derivatives with respect to @xmath183 and , afterward , we take the limit of @xmath184 to obtain the expression given by eq .
( [ eq : gelimite ] ) .
next , we solve the @xmath93 integrations using standard properties and expansions of the functions @xmath94 and @xmath95^{-\mu}$]@xcite , and write @xmath89 as follows @xmath185 where @xmath10 is even and @xmath186^{r - n+1}}{r!}}\\ { \displaystyle \times\sum_{\beta=0}^{\infty}l_{\beta}^{-\frac{n - r-1}{2}-\beta}\left(0\right)\left(-\frac{\alpha^{2}}{n^{2}}\right)^{\beta}\gamma\left(n+1\right)\int db\cdots\int db\frac{1}{b^{2+n - n+r-2\beta}},}\end{array}\label{eq : eme}\ ] ] where @xmath187 are the laguerre polynomials @xcite .
note that the only powers on @xmath188 in eq .
( [ eq : eme ] ) that fulfill the asymptotic behavior of the function @xmath89 are those for which the condition @xmath189 is satisfied
. therefore , the @xmath93 indefinite integrations are exactly doable .
the result is @xmath190 replacing eq .
( [ eq : nint ] ) into eq .
( [ eq : eme ] ) and after the change of variables @xmath191 we can write @xmath192 using the asymptotic properties of the gamma function@xcite @xmath193 eq .
( [ eq : eme2 ] ) finally reads @xmath194 the sum over @xmath195 is exactly doable .
the result is @xmath196 where @xmath197,\left[\beta+\frac{n+1}{2},\beta+\frac{n}{2}\right],-\frac{\left[\left(n-2k-1\right)\nu_{\alpha , n}\right]^{2}}{b^{2}}\right)}{\gamma\left(2\beta+n\right)}}\\ { \displaystyle + \frac{2i\left(n-2k-1\right)\nu_{\alpha , n}}{b}\frac{\gamma\left(\beta+\frac{3}{2}\right)\,_{3}f_{2}\left(\left[\beta+\frac{3}{2},1,1\right],\left[\beta+\frac{n}{2}+1,\beta+\frac{n}{2}+\frac{1}{2}\right],-\frac{\left[\left(n-2k-1\right)\nu_{\alpha , n}\right]^{2}}{b^{2}}\right)}{\gamma\left(2\beta+n+1\right)\sqrt{\pi}}.}\end{array}\label{eq : fd}\ ] ] equation ( [ eq : eme3 ] ) clearly shows that the imaginary part affects only the real part of the function @xmath198 .
thus , the final expression for @xmath89 is given by eq .
( [ eq : gbfinal ] ) .
we evaluate @xmath107 using the following integral representation of the hypergeometric function@xcite @xmath199,\left[\beta+\frac{n+1}{2},\beta+\frac{n}{2}\right],-\frac{\left(n-2k-1\right)^{2}}{b^{2}}\right)=\frac{b^{2\left(\beta+1\right)}}{b\left(1,2\beta+n-1\right)}}\\ { \displaystyle \times\frac{1}{\left(n-2k-1\right)^{2\beta+n-1}}\int_{0}^{n-2k-1}\left[n-2k-1-x\right]^{2\beta+n-2}\left[x^{2}+b^{2}\right]^{-\beta-1}dx},\end{array}\label{eq : otravez}\ ] ] which is valid for values of @xmath10 equal or larger than two . the analytic continuation to the complex plane
is done as before through the replacement @xmath99 .
then , the imaginary part of the hypergeometric function is @xmath200,\left[\beta+\frac{n+1}{2},\beta+\frac{n}{2}\right],\frac{\left(n-2k-1\right)^{2}}{r^{2}}\right)\right\ } = } \\ { \displaystyle \frac{r^{2\left(\beta+1\right)}\left(-\right)^{\beta+1}}{b\left(1,2\beta+n-1\right)\left(n-2k-1\right)^{2\beta+n-1}}}\\ { \displaystyle \times im\int_{0}^{n-2k-1}\left[n-2k-1-x\right]^{2\beta+n-2}\left[x^{2}-r^{2}\right]^{-\beta-1}dx}.\end{array}\end{array}\label{eq : imaginarypart}\ ] ] the analytical behavior of the function @xmath204 is well known@xcite .
it is an analytic function for @xmath205 but , its analytic continuation to the complex plane generates a cut on the real axis in the range @xmath206 which provides its imaginary part . writing @xmath222^{2}}\sum_{l=0}^{\beta}\left(\begin{array}{c } \beta\\ l\end{array}\right)\frac{\left(2\beta - l\right)!}{\left(2\right)^{2\beta+1-l}}}\\ { \displaystyle \frac{r^{l}\left[n-2k - r-1\right]^{2\beta+n-2-l}}{\left(2\beta+n-2-l\right)!}\quad n\geq2}.\end{array}\label{eq : hyperim}\ ] ] 10 m. g. poirier _
_ , _ phys .
_ 86 , * 360 * ( 2001 ) ; j. f. leger _ et at .
_ , _ phys .
lett . _ * 83 * , 1066 ( 1999 ) ; v. parpura and j. m. fernandez , _ biophys .
j. _ * 71 * , 2356 ( 1996 ) ; m. d. wang _ et al .
_ , _ biophys
. j. _ * 72 * , 1335 ( 1997 ) ; a. d. mehta , k. a. pullen and a. spudich , _ febs lett . _ * 430 * , 23 ( 1998 ) ; a. d. mehta , m. rief and j. a. spudich , _ j. biol
. chem . _ * 274 * , 14517 ( 1999),t
. e. fisher , p. e. marszalek and j. m. fernandez , _ nat .
biol . _ * 7 * , 719 ( 2000 ) ; t. e. fisher _ et al .
_ , _ trends biochem .
* 24 * , 379 ( 1999 ) ; m. carrion - vazquez _ et al .
. biophys .
biol . _ * 74 * , 63 ( 2000 ) ; t. t. perkins , s. r. quake , d. e. smith and s. chu , _ science _ * 264 * , 822 ( 1994 ) ; s. b. smith , y. cui and c. bustamante , _ science _ * 271 * , 795 ( 1996 ) ; t. t. perkins , d. e. smith and s. chu , _ science _ * 276 * , 2016 ( 1997 ) ; a. f. oberhauser , p. e. marszalek , h. p. erickson and j. m. fernandez , _ nature ( london ) _ * 393 * , 181 ( 1998 ) ; f. oesterhelt , m. rief and h. e. gaub , _ new j. phys . _ * 1 * , 6.1 ( 1999 ) ; h. b. li , w. k. zhang , w. q. xu and x. zhang , _ macromolecules _ * 33 * , 465 ( 2000 ) . j. f. marko and e. d. siggia , _ macromolecules _ * 28 * , 8759 ( 1995 ) ; k. kroy and e. frey , _ phys . rev . lett . _ * 77 * , 306 ( 1996 ) .
g. porod , _ monatsh .
chem . _ * 80 * , 251 ( 1949 ) ; o. kratky and g. porod , _ rec .
chim . _ * 68 * ,
1106 ( 1949 ) . l. eichinger _
et al_. , _ biophys .
j. _ * 70 * , 1054 ( 1996 ) ; p. janmey _ cell membranes and the cytoskeleton _ , vol .
1a of _ handbook of biological physics _ , chapter 17 , page 805 ( north holland , amsterdam , 1995 ) ; p. a. janmey , _ curr
. op . cell .
* 2 * , 4 ( 1991 ) .
n. saito , k. takahashi and y. yunoki , _ journal of the physical society of japan _ , * 22 * , 219 ( 1967 ) .
h. yamakawa , _ helical worm - like chains in polymer solutions _
( springer , berlin , 1997 ) .
m. fixman and j. kovac , _ j. chem .
phys . _ * 58 * , 1564(1973 ) .
r. a. harris and j. e. hearst , _ j. chem .
phys . _ * 44 * , 2595 ( 1966 ) .
h. e. daniels , _ proc .
edinburgh _ * a63 * , 290 ( 1952 ) ; w. gobush , h. yamakawa , w. h. stockmayer and w. s. magee , _
_ * 57 * , 2839 ( 1972 ) .
h. yamakawa , _ pure and appl . chem . _
* 46 * , 135 ( 1976 ) .
j. wilhelm and e. frey , _ phys .
lett . _ * 77 * , 2581 ( 1996 ) ; p. l. hansen and r. podgornik ,
_ j. chem .
phys . _ * 114 * , 8637 ( 2001 ) .
m. g. bawendi and k. f. freed , _ j. chem .
phys . _ * 83 * , 2491 ( 1985 ) ; j. b. lagowski , j. noolandi and b. nickel , _ j. chem
. phys . _ * 95 * , 1266 ( 1991 ) ; g. a. carri and m. muthukumar , _ j. chem . phys . _ * 109 * , 11117 ( 1998 ) .
r. g. winkler , p. reineker and l. harnau , _ j. chem .
* 101 * , 8119 ( 1994 ) . c. m. marques and g. h. fredrickson , _
ii france _ * 7 * , 1805 ( 1997 ) . l. harnau , r. g. winkler and p. reineker , _ j. chem . phys . _ * 104 * , 6355 ( 1996 ) ; k. kroy and e. frey , _ phys . rev
. e _ * 55 * , 3092 ( 1997 ) .
a. l. kholodenko , _ ann .
* 202 * , 186 ( 1990 ) ; a. l. kholodenko _
faraday trans . _ * 91 * , 2473 ( 1995 ) ; a. l. kholodenko , _ phys .
a _ * 141 * , 351 ( 1989 ) ; a. l. kholodenko , _ phys .
lett . a _ * 178 * , 180 ( 1993 ) ; a. kholodenko and t. vilgis _ phys . rev
. e _ * 50 * , 1257 ( 1994 ) , a. l. kholodenko and t. a. vilgis _ phys . rev .
e _ * 52 * , 3973 ( 1995 ) .
a. l. kholodenko , _ macromolecules _ * 26 * , 4179 ( 1993 ) . a. kholodenko , m. ballauff and m. aguero granados , _ physica a _ * 260 * , 267 ( 1998 ) . m. ballauff _ et al .
_ , _ macromolecules _ * 30 * , 273 ( 1997 ) .
r. g. winkler , _
phys . _ * 118 * , 2919 ( 2003 ) .
l.n.epele , h. fanchiotti , c.a .
garcia canal and m. marucho , _ phys .
lett . _ * b523 * , 102 ( 2001 ) .
l.n.epele , h. fanchiotti , c.a .
garcia canal and m. marucho , _ phys .
lett . _ * b556 * , 87 ( 2003 ) .
l.n.epele , h. fanchiotti , c.a
. garcia canal and m. marucho , _ nucl .
* b583 * , 454 ( 2000 ) . s. m. bhattacharjee and m. muthukumar , _ j. chem . phys . _ * 86 * , 411(1987 ) .
k. f. freed in _ advances in chemical physics _ , vol .
22 , pages 1 - 128 , i. prigogine and stuart a. rice editors ( john wiley and sons , new york , 1972 ) .
g. arfken , _ mathematical methods for physicists _
( academic press , inc . , new york , 1985 ) .
m. abramowitz and i. stegun , _ handbook of mathematical functions _ ( dover , new york , 1970 ) .
i.s . gradshteyn and i.m .
ryzhik , _ table of integrals , series , and products _
( academic press , new york , 2000 ) . h. benoit and p. doty , _ j. phys .
chem . _ * 57 * , 958 ( 1953 ) .
k. f. freed , _ renormalization group theory of macromolecules _ ( john wiley & sons , new york , 1987 ) .
p. g. de gennes , _ biopolymers _ * 6 * , 715 ( 1968 ) .
h. bateman , _ higher transcendental functions _ ( mc graw hill , new york , 1953 ) .
m. c. wang and e. guth , _ j. chem .
_ * 20 * , 1144 ( 1952 ) ; c. hsiung , h. hsiung and a. gordus , _ j. chem . phys . _
* 34 * , 535 ( 1961 ) . w. l. mattice and u. w. suter , _ conformational theory of large macromolecules : the rotational isomeric state model in macromolecular systems _
( john wiley & sons , new york , 1994 ) .
j. wilhelm and e. frey , _ phys .
* 77 * , 2581 ( 1996 ) .
n. bleistein and r. handelsman , _
asymptotic expansion of integrals _
( dover , new york , 1986 ) .
e. t. copson , _ asymptotic expansions _ ( cambridge university press , 1965 ) .
h. jefrey and b. s. jeffrey , _ methods of mathematical physics _
( cambridge university press , 1966 ) . * normalized polymer propagator @xmath223 versus @xmath224 for @xmath225 .
continuous line @xmath226 , dotted line @xmath227 , dashed line @xmath228 , long dashed line @xmath229 and dashed - dotted line @xmath230 . *
normalized polymer propagator @xmath223 versus @xmath224 for @xmath231 .
continuous line @xmath226 , dotted line @xmath227 , dashed line @xmath228 , long dashed line @xmath229 and dashed - dotted line @xmath230 . *
normalized polymer propagator @xmath223 versus @xmath224 for @xmath232
. continuous line @xmath226 , dotted line @xmath227 , dashed line @xmath228 , long dashed line @xmath229 and dashed - dotted line @xmath230 .
* normalized polymer propagator @xmath223 versus @xmath224 for @xmath233 . continuous line @xmath234 , dashed line @xmath235 and dashed - dotted line @xmath236 . *
single chain structure factor @xmath237 versus wave vector @xmath238 for @xmath120 .
continuous line @xmath239 , dotted line @xmath240 , dashed line @xmath241 , dashed - dotted line @xmath242 and circles ( best fits to the power law in the range @xmath243 for @xmath244 and @xmath245 ) . *
single chain structure factor @xmath237 versus wave vector @xmath238 for @xmath246 .
continuous line @xmath239 , dotted line @xmath240 , dashed line @xmath241 , dashed - dotted line @xmath242 and circles ( best fits to the power law in the range @xmath243 for @xmath244 and @xmath245 ) . *
single chain structure factor @xmath237 versus wave vector @xmath238 for @xmath126 .
continuous line @xmath239 , dotted line @xmath240 , dashed line @xmath241 , dashed - dotted line @xmath242 and circles ( best fits to the power law in the range @xmath243 for @xmath244 and @xmath245 ) . *
single chain structure factor @xmath117 versus wave vector @xmath77 for @xmath126 . (
line ) kholodenko s model with @xmath247 , ( points ) this work with @xmath248 .
* single chain structure factor @xmath117 versus wave vector @xmath77 for @xmath126 . ( line ) kholodenko s model with @xmath249 , ( points ) this work with @xmath250 .
* normalized polymer propagator @xmath251 versus end - to - end distance @xmath115 in units of kuhn length for @xmath120 .
( continuous lines ) this work , ( dashed lines ) wilhelm and frey s results .
normalized polymer propagator @xmath252 versus @xmath253 for @xmath120 .
continuous line @xmath254 , dotted line @xmath255 , dashed line @xmath256 , long dashed line @xmath257 and dashed - dotted line @xmath258 .
normalized polymer propagator @xmath252 versus @xmath253 for @xmath246
. continuous line @xmath254 , dotted line @xmath255 , dashed line @xmath256 , long dashed line @xmath257 and dashed - dotted line @xmath258 .
normalized polymer propagator @xmath252 versus @xmath253 for @xmath126 .
continuous line @xmath254 , dotted line @xmath255 , dashed line @xmath256 , long dashed line @xmath257 and dashed - dotted line @xmath258 .
single chain structure factor @xmath117 versus wave vector @xmath77 for @xmath120 .
continuous line @xmath254 , dotted line @xmath255 , dashed line @xmath256 , dashed - dotted line @xmath258 and circles ( best fits to the power law in the range @xmath263 for @xmath123 and @xmath264 ) .
single chain structure factor @xmath117 versus wave vector @xmath77 for @xmath246 . continuous line @xmath254 , dotted line @xmath255 , dashed line @xmath256 , dashed - dotted line @xmath258 and circles ( best fits to the power law in the range @xmath263 for @xmath123 and @xmath264 ) .
single chain structure factor @xmath117 versus wave vector @xmath77 for @xmath126 .
continuous line @xmath254 , dotted line @xmath255 , dashed line @xmath256 , dashed - dotted line @xmath258 and circles ( best fits to the power law in the range @xmath263 for @xmath123 and @xmath264 ) . | in this paper , we present a new approach to the discrete version of the wormlike chain model ( wcm ) of semiflexible polymers . our solution to the model
is based on a new computational technique called the generalized borel transform ( gbt ) which we use to study the statistical mechanics of semiflexible polymer chains .
specifically , we evaluate the characteristic function of the model approximately .
afterward , we compute the polymer propagator of the model using the gbt and find an expression valid for polymers with any number of segments and values of the semiflexibility parameter .
this expression captures the limits of flexible and infinitely stiff polymers exactly . in between ,
a smooth and approximate crossover behavior is predicted .
another property of our propagator is that it fulfills the condition of finite extensibility of the polymer chain .
we have also calculated the single chain structure factor .
this property is a decreasing function of the wave vector , @xmath0 until a plateau is reached .
our computations clearly show that the structure factor decreases faster with increasing wave vector when the semiflexibility parameter is increased .
furthermore , when the wave vector is large enough , there is a regime where the structure factor follows an approximate power law of the form @xmath1 even for short polymer chains .
@xmath2 is equal to two for flexible polymers and to one for rigid chains .
we also compare our results to the predictions of other models . |
a striking feature of the atmosphere of titan is that no heavy noble gases other than argon were detected by the gas chromatograph mass spectrometer ( gcms ) aboard the huygens probe during its descent to titan s surface in january 2005 ( niemann et al .
2005 , 2010 ) .
the detected argon includes primordial @xmath0ar , present in subsolar abundance in titan s atmosphere ( @xmath0ar/@xmath1n is found to be about six orders of magnitude lower than the solar value ) , and the radiogenic isotope @xmath2ar , which is a decay product of @xmath2k ( niemann et al .
the other primordial noble gases @xmath3ar , kr and xe were not detected by the gcms instrument , yielding upper limits of 10@xmath4 for their atmospheric mole fractions ( niemann et al .
2005 , 2010 ) . in order to interpret this deficiency
, it has been proposed that the atmospheric depletion of these species could be explained by their dissolution at ambient temperature in the hydrocarbon lakes and seas present on titan s surface ( cordier et al .
however , the fractions of argon and krypton that would dissolve in these liquids were found to be negligible ( cordier et al . 2010 ) .
another interpretation of the lack of heavy noble gas abundances above 10 ppb in titan s atmosphere is based on laboratory results suggesting that the haze present in titan s atmosphere could efficiently trap argon , krypton and xenon during its formation ( jacovi & bar - nun 2008 ) . in this mechanism
, the open structure of the small aerosol particles would allow the noble gas atoms to fill their pores .
however , even if this trapping mechanism is effective , the erosion of sedimented aerosols might induce the progressive release of the trapped noble gases in titan s atmosphere .
the presence of clathrate hydrates on titan s surface has also been proposed to be the origin of the heavy noble gas deficiency measured in its atmosphere ( osegovic & max 2005 ) .
a series of theoretical investigations showed that the trapping efficiency of these ice structures is high enough to significantly decrease the atmospheric concentrations of xenon and krypton , but failed to explain the observed argon atmospheric deficiency ( thomas et al .
2007 , 2008 ) .
it was then argued that titan s building blocks were partly devolatilized within a subnebula around the forming saturn , implying the loss of argon initially incorporated as pure condensate prior to satellite formation ( mousis et al .
however , more recent work showed that argon could have been retrapped in primordial clathrates so as to remain abundant in the planetesimals incorporated in titan ( mousis et al .
2010 ) . given the evidence from @xmath2ar that titan is extensively outgassed ( niemann et al .
2010 ) , it is difficult to explain the observed @xmath0ar rarity in the atmosphere in the absence of a sink over geologic time .
here we show that trapping over time of atmospheric argon in clathrates could indeed provide such a sink .
this conclusion is based on the use of recent intermolecular potential parameters in a statistical thermodynamic model describing clathrate composition . the fresh water ice needed for formation of these clathrates
could be delivered by successive episodes of cryovolcanic lavas hypothesized to periodically cover the surface of titan .
the formation of clathrates in the porous cryo - lavas and their propensity for trapping ar , kr and xe would progressively remove these species from the atmosphere of titan over its history .
a global clathrate crust with an average thickness not exceeding a few meters could be sufficient on titan for a complete removal of the heavy noble gases from the atmosphere .
to calculate the relative abundances of guest species incorporated in a clathrate from a coexisting gas of specified composition at given temperature and pressure , we use a model applying classical statistical mechanics that relates the macroscopic thermodynamic properties of clathrates to the molecular structure and interaction energies ( van der waals & platteuw 1959 ; lunine & stevenson 1985 ) .
it is based on the original ideas of van der waals and platteeuw for clathrate formation , which assume that trapping of guest molecules into cages corresponds to the three - dimensional generalization of ideal localized adsorption . in this work , the availability of recent intermolecular potential parameters retrieved from experiments ( sloan & koh 2008 ; hereafter sk08 ) allows us to revise the relative abundances of guest species incorporated in a clathrate from a coexisting gas of specified composition at temperature and pressure relevant to titan s surface conditions . in this formalism , the fractional occupancy of a guest molecule @xmath5 for a given type @xmath6 ( @xmath6 = small or large ) of cage can be written as @xmath7 where the sum in the denominator includes all the species which are present in the initial gas phase .
@xmath8 is the langmuir constant of species @xmath5 in the cage of type @xmath6 , and @xmath9 is the partial pressure of species @xmath5 .
this partial pressure is given by @xmath10 ( we assume that the sample behaves as an ideal gas ) , with @xmath11 the mole fraction of species @xmath5 in the initial gas , and @xmath12 the total atmospheric gas pressure , which is dominated by n@xmath13 .
the langmuir constant depends on the strength of the interaction between each guest species and each type of cage , and can be determined by integrating the molecular potential within the cavity as @xmath14 where @xmath15 represents the radius of the cavity assumed to be spherical , @xmath16 the boltzmann constant , and @xmath17 is the spherically averaged kihara potential representing the interactions between the guest molecules @xmath5 and the h@xmath13o molecules forming the surrounding cage @xmath6 .
this potential @xmath18 can be written for a spherical guest molecule , as ( mckoy & sinanolu 1963 ) @xmath19,\end{aligned}\ ] ] with @xmath20.\ ] ] in eq .
( [ pot_kihara ] ) , @xmath21 is the coordination number of the cell .
this parameter depends on the structure of the clathrate ( i or ii ) and on the type of the cage ( small or large ) .
the kihara parameters @xmath22 , @xmath23 and @xmath24 for the molecule - water interactions , given in table [ kihara ] , have been taken from the recent work of sk08 when available and from parrish & prausnitz ( 1972 ) ( hereafter pp72 ) for the remaining species .
finally , the mole fraction @xmath25 of a guest molecule @xmath5 in a clathrate can be calculated with respect to the whole set of species considered in the system as @xmath26 where @xmath27 and @xmath28 are the number of small and large cages per unit cell respectively , for the clathrate structure under consideration , and with @xmath29 .
values of @xmath15 , @xmath21 , @xmath27 and @xmath28 are taken from pp72 .
table [ atm ] gives the composition of titan s atmosphere used in our calculations .
we made the conservative assumption that all noble gases were initially present in the atmosphere of titan , with ar /
n , kr / n and xe / n ratios assigned to be solar ( asplund et al . 2009 ) . because our composition calculations are only valid along the equilibrium curve of the clathrate of interest , they were performed at 165.8 k , i.e. the equilibrium temperature of a multiple guest clathrate formed from the atmosphere of titan possessing a pressure of 1.46 bar at the surface ( see appendix ) .
composition calculations have been performed in the cases of structures i and ii clathrates .
we have also taken into account the influence of variation of the cage sizes on the mole fractions of guests encaged in clathrates by modifying the values of their initial radius by up to 3% . indeed , thermal expansion or contraction measured in the temperature range of 90270 k have been found to significantly affect the composition of clathrates as well as their dissociation pressures that define the ( p , t ) loci of stability ( mousis et al .
2010 ; belosludov et al .
2002 ) .
figure [ comp1 ] gives the composition of structures i and ii clathrates expected to form on the surface of titan from the nitrogen - dominated atmosphere in the cases of our nominal model ( no cage variation ) and a 2% contraction of the cages ( belosludov et al .
2002 ) . assuming that noble gases are in solar abundances relative to elemental nitrogen ( asplund et al .
2009 ) , the total number of ar , kr and xe atoms that would exist in the atmosphere of titan is order of 1.06 @xmath30 10@xmath31 , 3.23 @xmath30 10@xmath32 and 2.02 @xmath30 10@xmath3 , respectively .
the equivalent volume of clathrate needed to fully trap argon , the most abundant noble gas , would be then about 2.29 x 10@xmath33 m@xmath34 in clathrates , irrespective of the considered structure 10@xmath35 m@xmath34 while the unit cell of structure ii consists of 24 cages for a unit volume of 5.18 @xmath30 10@xmath35 m@xmath34 . ] .
on the other hand , since the mole fraction of ar has been found to be about 5.01 @xmath30 10@xmath36 and 1.78 @xmath30 10@xmath37 in structures
i and ii clathrates of our nominal model , we then obtain a total clathrate volume of 4.55 @xmath30 10@xmath38 and 1.28 @xmath30 10@xmath38 m@xmath34 needed to trap simultaneously the three noble gases in these two structures , respectively . in the case of a 2% contraction of the cages ,
the mole fraction of ar becomes 9.24 @xmath30 10@xmath36 and 4.02 @xmath30 10@xmath37 in structures i and ii clathrates , corresponding to a total clathrate volume of 2.47 @xmath30 10@xmath38 and 5.67 @xmath30 10@xmath33 m@xmath34 in these two structures , respectively .
interestingly , irrespective of the considered case , ar remains the less efficiently trapped of the three considered noble gases ( respectively to their initial atmospheric abundances ) so its complete enclathration also implies the full trapping of krypton and xenon .
figure [ thick ] represents the thickness of the clathrate layer needed to trap the three noble gases as a function of the variation of the cage sizes in the cases of structures i and ii clathrates . in the case of the nominal model
, the total clathrate volume translates into a multiple guest clathrate layer with an equivalent thickness of about 549.6 and 154.7 m globally averaged in structures i and ii , respectively .
if one considers the case corresponding to a 2% contraction of the clathrates cages , then the equivalent thickness of the clathrate layer would be reduced to about 298.0 and 68.6 m over the surface of titan in structures i and ii , respectively .
formation scenarios predict that titan only accreted elemental nitrogen in the form of ammonia during its formation ( mousis et al .
2002 , 2009 ) , this latter being converted into molecular nitrogen via photolysis ( atreya et al . 1978 ) or shock chemistry ( mckay et al . 1988 ) in the atmosphere .
if 5% of solar elemental nitrogen was accreted by titan in the form of ammonia ( mousis et al .
2002 ) , then the equivalent thickness of the clathrate layer would reduce to about 27.4 and 7.7 m in the nominal model and 14.8 and 3.4 m in the case of a 2% contraction of the cages in structures i and ii , respectively .
our model then suggests that formation scenarios favoring the accretion of titan from nh@xmath39-rich and n@xmath13-poor planetesimals are preferable because the amount of clathrates needed to explain the noble gas atmospheric deficiency is much less important than in the case of satellite accretion from n@xmath13-rich planetesimals .
it should be noted that , because i ) the atmosphere of titan is by far dominated by n@xmath13 and ii ) n@xmath13 clathrate is of structure ii ( lunine & stevenson 1985 ; sk08 ) , we should expect the presence of structure ii clathrates on the satellite s surface . in any case , future laboratory experiments aiming at investigating the formation of clathrates formed from a mixture analogous to that of titan s atmosphere could help to disentangle between the two possible structures .
in this context , assuming that titan s clathrates are of structure ii , the average thickness of the clathrate layer needed on titan for the complete removal of the heavy noble gases from the atmosphere should then range between a few tens of meters to a few hundreds of meters .
our calculations imply that noble gas - rich clathrates either formed ( i ) when the surface of the satellite of saturn was warmer than today ( thomas et al .
2007 ) and/or ( ii ) during more recent release events of hot and porous cryolava possibly associated with methane delivery towards the surface ( mousis & schmitt 2008 ; tobie et al . 2006 ) . in the first case , titan s accretional heating would have resulted in an initially warm surface ( @xmath40300500 k ) , implying the melting of the ice ( kuramoto & matsui 1994 ; lunine et al .
2010 ) . with time , crustal freezing would have allowed formation of multiple guest clathrates at the surface / atmosphere interface ( tobie et al . 2006 ; lunine et al .
2010 ) . in the second case
, a highly porous icy material should be generated from the decomposition of methane clathrates ( mousis & schmitt 2008 ) and the formation of multiple guest clathrates at the surface - atmosphere interface would take place during the cooling of the cryolava , which needs about 1 earth yr to decrease to titan s surface temperature ( lorenz 1996 ) . in both cases , the clathrates composition is assumed to remain fixed during the cooling of titan s surface . because the volume expansion induced by clathrate formation ( @xmath4020% compared to that of water ice ( mousis & schmitt 2008 ) ) may well cause its self - isolation by closing the pores of the ice that is not in direct contact with the atmosphere , the thickness of the clathrate layer formed during the satellite cooling should be very limited , unless events such as impacts ( sekine et al . 2011 ) expose clean ice .
cryovolcanic activity , whose features are visible in some regions of titan ( wall et al .
2009 ) , is probably required to create substantial amounts of porous water ice spread across the surface , in a way similar to basaltic lava flows on earth .
long periods of active cryovolcanism on titan ( tobie et al .
2006 ) should lead to the continuous build up of consecutive layers of cryolava ( mousis & schmitt 2008 ) and thus favor the formation of noble gas rich clathrate layers .
the gcms detection of @xmath2ar remains consistent with our model and implies that it is the most abundant argon isotope trapped in the clathrates layers .
its release into the atmosphere would then result from outgassing events ( waite et al .
2005 ) that occurred later than the removal of all the atmospheric noble gases .
however , the formation of noble gas - rich clathrate layers from water ice delivered via cryovolcanism might meet two limitations .
first , once in contact with the atmosphere , the cryolava flows might be too hot to directly form noble gas - rich clathrate layers .
second , cryolava flows could act as sources of noble gases by liberating them from underlying clathrates through melting .
the volume ( temporarily ) melted by the flow may well exceed the volume of the flow itself that is available for sequestering , resulting in net release of noble gas , instead of net sequestration . on the other hand ,
clathrate formation should remain possible during the rapid cooling of the cryolava .
indeed , the equilibrium pressure of a n@xmath13-dominated clathrate ( similar to our structure ii nominal model ) is about 10@xmath41 bars for a surface temperature of 94 k. with a value of @xmath40 1.3 bars , the partial pressure of n@xmath13 in the atmosphere greatly exceeds the equilibrium pressure of n@xmath13-dominated clathrate and then favors the formation of this structure at the current surface temperature . as such , the newly formed clathrate layers could at least partially absorb the noble gases recently released into the atmosphere by hot cryolava flows .
we thank an anonymous referee whose comments have helped us improve our manuscript .
in the present study , the temperature dependence of the dissociation pressure of the multiple guest clathrate is determined from available experimental data and from a combination rule due to lipenkov & istomin ( 2001 ) .
thus , the dissociation pressure @xmath42 of a multiple guest clathrate is calculated from the dissociation pressure @xmath43 of a pure clathrate of guest species @xmath5 as where @xmath43 and @xmath46 are expressed in pa and k , respectively .
the constants @xmath47 and @xmath48 used in the present study have been fitted to the experimental data used by thomas et al .
( 2007 ) and derived from lunine & stevenson ( 1985 ) and sk08 .
parrish , w. r. , prausnitz , j. m. 1972 , dissociation pressures of gas hydrates formed by gas mixtures .
process design dev .
11 ( 1 ) , 26 ( erratum : parrish , w. r. , prausnitz , j. m. 1972 , ind .
chem . process design dev .
11(3 ) , 462 ) ( pp72 ) | a striking feature of the atmosphere of titan is that no heavy noble gases other than argon were detected by the gas chromatograph mass spectrometer ( gcms ) aboard the huygens probe during its descent to titan s surface in january 2005 .
here we provide an explanation of the mysterious absence or rarity of these noble gases in titan s atmosphere : the thermodynamic conditions prevailing at the surface - atmosphere interface of the satellite allow the formation of multiple guest clathrates that preferentially store some species , including all heavy noble gases , over others . the clean water ice needed for formation of these clathrates
could be delivered by successive episodes of cryovolcanic lavas that have been hypothesized to regularly cover the surface of titan .
the formation of clathrates in the porous lavas and their propensity for trapping ar , kr and xe would progressively remove these species from the atmosphere of titan over its history . in some circumstances ,
a global clathrate crust with an average thickness not exceeding a few meters could be sufficient on titan for a complete removal of the heavy noble gases from the atmosphere . |
recently , the atlas and the cms collaborations at the large hadron collider ( lhc ) discovered a new boson @xmath0 in the mass range 125126 gev @xcite .
in addition , the standard model ( sm ) higgs boson has been excluded at 95% c.l . in the mass range 110149 gev , except for the narrow region 122.8127.8 gev @xcite . the mass range from 127 to 600 gev
was previously excluded @xcite .
it is also noticeable that the mass @xmath1@xmath2 gev perfectly agrees with the lep precision measurements @xcite . on the other hand ,
the contact interactions in the processes of @xmath3 and @xmath4 are severely constrained , i.e. , the compositeness scale @xmath5 should be larger than , say , 10 tev @xcite . against this situation , is there still a room for some strong dynamics responsible for the electroweak symmetry breaking ( ewsb ) ?
we give an overview of typical models with composite higgs bosons .
it might give some hint for the origin of the ewsb .
we will also introduce the isospin symmetric higgs model , which is recently proposed in ref .
@xcite , as an example of the dynamical ewsb scenario .
the earliest idea of the dynamical ewsb is technicolor ( tc ) @xcite : the chiral condensate of ( techni- ) fermions is dynamically generated by the technicolor gauge interaction and it breaks the electroweak gauge symmetry , as in low - energy qcd .
the would - be nambu - goldstone ( ng ) bosons are eaten by the weak gauge bosons . then @xmath6 and @xmath7 acquire their masses proportional to the technipion decay constant ( @xmath8 100 gev ) , which is analogous to the pion decay constant @xmath9 mev in qcd . in order to produce the masses of the sm fermions ,
the extended technicolor ( etc ) has been proposed @xcite .
although it was beautiful , this old - fashioned tc has been already excluded by several reasons @xcite .
for example , the constraint of the @xmath10-parameter @xcite rules out this qcd - like tc with many weak doublets of the technifermions @xcite .
a modern version of tc is the walking tc where the gauge coupling of the tc gauge group runs very slowly , or `` walks '' @xcite .
the walking tc resolves the difficulties of the old - fashioned tc .
the estimate of the @xmath10-parameter expected from qcd is not applicable to walking tc . even in walking tc
, however , it is difficult to generate the observed mass of the top quark from the etc interactions without producing unacceptably large isospin breaking .
also , the walking tc usually predicts a heavy composite higgs boson . noticing that the @xmath10-parameter constraint requires a heavy technirho bound state @xmath11
, it is quite nontrivial to get a light composite higgs , @xmath12 gev , and simultaneously to obtain the heavy @xmath11 , say , @xmath13 few tev . to generate the mass of the top quark , the topcolor dynamics is useful @xcite . in the topcolor scenario , we assume that the new topcolor interaction strongly couples to the third generation of quarks and then the top quark condensate mainly yields the top quark mass . in the simplest four dimensional model that the top condensate is responsible both for the ewsb and @xmath14 , and that there appears only one ( composite ) higgs doublet , too large top quark mass is predicted , however . in a model with extra dimensions ,
this difficulty is relaxed @xcite .
another approach to avoid too large top mass is to assume that the top quark condensate is responsible only for @xmath14 and the ewsb takes place by some other mechanism . in topcolor
assisted technicolor ( tc2 ) , the tc interaction causes the ewsb @xcite .
because the chiral symmetry is extended in tc2 , there appears the extra ng bosons , so - called top - pions .
the masses of the top - pions are generally light and thus they are severely constrained . in a model in ref .
@xcite , we employed a subcritical dynamics ( although nearcritical , i.e. , strong ) for the topcolor interaction , so that the mass of the scalar bound state of the top and the anti - top quarks is naturally heavy .
this mechanism is used in the isospin symmetric ( is ) higgs model recently proposed in ref .
@xcite .
we introduce the is higgs model in the next section .
the atlas and cms experiments did not only announced the mass of the new discovered boson @xmath0 , but they also reported the nature of @xmath0 : while the decay channels of @xmath15 and @xmath16 are fairly consistent with the sm , the diphoton branching ratio br@xmath17 is about 1.6 times larger than the sm value @xcite . in the latest results , the atlas collaboration confirmed the similar enhancement in the diphoton channel @xcite .
on the other hand , the cms group changed their previous results of the signal strength from @xmath18 to @xmath19 for the mass - fit - mva analysis and @xmath20 for the cut - based analysis @xcite .
the situation thus becomes unclear at present . in any case , the deviation from the sm in the diphoton channel , if established , would be an indication of a new physics beyond the sm .
let us introduce the is higgs boson model .
the main characteristics of the is higgs boson model are as follows @xcite : a ) it is assumed that the dynamics primarily responsible for the ewsb leads to the mass spectrum of quarks with no ( or weak ) isospin violation .
_ moreover , it is assumed that the values of these masses are of the order of the observed masses of the down - type quarks . _
b ) the second ( central ) assumption is introducing the horizontal interactions for the quarks in the three families . as a first step , a
_ subcritical _
( although nearcritical , i.e. , strong ) diagonal horizontal interactions for the top quark is utilized which lead to the observed ratio @xmath21 @xcite .
the second step is introducing _ equal _ strength horizontal flavor - changing - neutral ( fcn ) interactions between the @xmath22 and @xmath23 quarks and the @xmath24 and @xmath25 ones . as was shown in ref .
@xcite , these interactions naturally provide the observed ratio @xmath26 in the second family @xcite . as to the mild isospin violation in the first family
, it was studied together with the effects of the family mixing , reflected in the cabibbo - kobayashi - maskawa ( ckm ) matrix @xcite . in this scenario ,
the main source of the isospin violation is only the strong top quark interactions .
however , because these interactions are subcritical , the top quark plays a minor role in ewsb .
this distinguishes the is higgs scenario from the top quark condensate model @xcite .
one of the signatures of this scenario is the appearance of a composite top - higgs boson @xmath27 composed of the quarks and antiquarks of the third family @xcite .
note that unlike tc2 @xcite , this class of models utilizes subcritical dynamics for the top quark , so that the top - higgs @xmath27 is heavy in general .
here we also emphasize that while the top - higgs boson @xmath27 has a large top - yukawa coupling , the is higgs boson @xmath0 does not , @xmath28 . on the other hand , the @xmath29 and @xmath30 coupling constants are close to those in the sm .
also , the mixing between @xmath0 and much heavier @xmath27 should be small .
we now describe the decay processes of the is higgs @xmath0 .
it is well known that the @xmath6-loop contribution to @xmath31 is dominant in the sm , while the top - loop effect is destructive against the @xmath6-loop . in
the is higgs model , however , the yukawa coupling between the top and the is higgs @xmath0 is as small as the bottom yukawa coupling , so that the top - loop contribution is strongly suppressed .
the partial decay width of @xmath32 is thus enhanced without changing essentially @xmath15 and @xmath16 . a rough estimate taking the isospin symmetric top and bottom yukawa couplings @xmath33 is as follows : @xmath34 here using the pagels - stokar formula @xcite , we estimated the vacuum expectation value ( vev ) of the top - higgs @xmath27 as @xmath35 gev , and the vev @xmath36 of the is higgs @xmath0 is given by the relation @xmath37 with @xmath38 gev .
note that the values of the ratios in eq .
( [ gamma ] ) are not very sensitive to the value of @xmath39 , e.g. , for @xmath40@xmath41 gev , the suppression factor in the pair decay modes to @xmath42 and @xmath43 is @xmath44@xmath45 and the vev @xmath36 of the is higgs @xmath0 is given by the relation @xmath37 with @xmath38 gev .
note that the values of the ratios in eq .
( [ gamma ] ) are not very sensitive to the value of @xmath39 , e.g. , for @xmath40@xmath41 gev , the suppression factor in the pair decay modes to @xmath42 and @xmath43 is @xmath44@xmath45 and the enhancement factor in the diphoton channel is @xmath46@xmath47 . for the decay mode of @xmath48 ,
this model yields @xmath49 the values in eq .
( [ gamma ] ) agree well with the data in the atlas and cms experiments . however , obviously , the main production mechanism of the higgs boson , the gluon fusion process @xmath50 , is now in trouble .
the presence of new chargeless colored particles , which considered by several authors @xcite can help to resolve this problem .
we pursue this possibility in the next section .
as a benchmark model , we may introduce a real scalar field @xmath10 in the adjoint representation of the color @xmath51 : @xmath52 where @xmath53 represents the is higgs doublet .
the effective lagrangian @xmath54 also contains the is higgs quartic couplings @xmath55 ,
@xmath56 . the is higgs mass is @xmath57 , and we will take it to be equal to @xmath58 gev .
the mass - squared term for the scalar @xmath10 is @xmath59 , and should be positive in order to avoid the color symmetry breaking .
typically , @xmath60 gev .
taking into account the @xmath10 contribution to @xmath50 , we find appropriate values of the higgs - portal coupling , @xmath61 as a typical value , we may take @xmath62 for @xmath63 gev and @xmath64 gev . a comment concerning the is higgs quartic coupling @xmath55 is in order . in the sm ,
the higgs mass 125 gev suggests that the theory is perturbative up to an extremely high energy scale @xcite . on the contrary , in the present model , when we take a large higgs - portal coupling @xmath65 that reproduces @xmath50 correctly , the quartic coupling @xmath55 will grow because the @xmath66-function for @xmath55 contains the @xmath67 term . also , there is no large negative contribution to the @xmath66-function for @xmath55 from the top - yukawa coupling @xmath68
. one can demonstrate such a behavior more explicitly by using the renormalization group equations . in fig .
[ lam ] , the running of the coupling @xmath55 is shown . taking a large higgs - portal coupling @xmath69 and the @xmath70-coupling @xmath71
, it turns out that the coupling @xmath55 rapidly grows .
the blowup scale strongly depends on the initial values of @xmath65 and @xmath72 .
a detailed analysis will be performed elsewhere .
last but not least , we would like to mention that other realizations of the enhancement of the @xmath0 production are also possible . .
the solid and dashed lines correspond to @xmath55 and the sm higgs quartic coupling , respectively .
we fixed the is higgs mass @xmath73 gev and took @xmath69 and @xmath71 .
unlike the sm , the is higgs quartic coupling grows up due to a large higgs - portal coupling @xmath65 and a small top - yukawa coupling @xmath74 .
[ lam],width=283 ]
we gave the overview of the typical scenarios with the dynamical ewsb and also introduced the is higgs boson model .
in particular , the is higgs model can explain the enhanced higgs diphoton decay rate observed at the lhc , and also makes several predictions .
the most important of them is that the value of the top - yukawa coupling @xmath0-@xmath22-@xmath75 should be close to the bottom - yukawa one .
another prediction relates to the decay mode @xmath48 , which is enhanced only slightly , @xmath76 , unlike @xmath32 .
last but not least , the lhc might potentially discover the top - higgs resonance @xmath27 , if lucky . for details , see ref .
@xcite .
s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
d * 87 * , 052017 ( 2013 ) [ arxiv:1301.5023 [ hep - ex ] ] ; phys .
d * 87 * , 032001 ( 2013 ) [ arxiv:1212.4563 [ hep - ex ] ] ; g. aad _ et al .
_ [ atlas collaboration ] , phys .
d * 87 * , 015010 ( 2013 ) [ arxiv:1211.1150 [ hep - ex ] ] .
s. weinberg , phys .
d * 19 * , 1277 ( 1979 ) ; l. susskind , phys .
d * 20 * , 2619 ( 1979 ) . s. dimopoulos and l. susskind , nucl .
b * 155 * , 237 ( 1979 ) .
e. eichten and k. d. lane , phys .
b * 90 * , 125 ( 1980 ) . for a textbook , see , e.g. , v. a. miransky , _ dynamical symmetry breaking in quantum field theories _ , ( world scientific pub .
co. , singapore , 1993 ) . for a comprehensive review , see c. t. hill and e. h. simmons , phys .
* 381 * , 235 ( 2003 ) ; erratum - ibid .
* 390 * , 553 ( 2004 ) . m. e. peskin and t. takeuchi , phys .
* 65 * , 964 ( 1990 ) ; phys .
d * 46 * , 381 ( 1992 ) .
j. beringer _
[ particle data group ] , phys .
d * 86 * , 010001 ( 2012 ) .
m. hashimoto , m. tanabashi and k. yamawaki , phys .
d * 64 * , 056003 ( 2001 ) ; _ ibid _ d * 69 * , 076004 ( 2004 ) ; v. gusynin , m. hashimoto , m. tanabashi and k. yamawaki , phys .
d * 65 * , 116008 ( 2002 ) . c. t. hill , phys .
b * 345 * , 483 ( 1995 ) .
m. hashimoto and v. a. miransky , phys .
d * 80 * , 013004 ( 2009 ) ; phys .
d * 81 * , 055014 ( 2010 ) .
the atlas collaboration , atlas - conf-2013 - 014 .
h. pagels and s. stokar , phys .
d * 20 * , 2947 ( 1979 ) .
r. boughezal and f. petriello , phys .
d * 81 * , 114033 ( 2010 ) ; b. batell , s. gori and l. -t .
wang , jhep * 1206 * , 172 ( 2012 ) ; b. a. dobrescu , g. d. kribs and a. martin , phys .
d * 85 * , 074031 ( 2012 ) ; g. d. kribs and a. martin , arxiv:1207.4496 [ hep - ph ] . j. a. casas , j. r. espinosa and m. quiros , phys . lett . b * 342 * , 171 ( 1995 ) ; _ ibid _ b * 382 * , 374 ( 1996 ) ; g. isidori , g. ridolfi and a. strumia , nucl . phys .
b * 609 * , 387 ( 2001 ) ; j. elias - miro , j. r. espinosa , g. f. giudice , g. isidori , a. riotto and a. strumia , phys .
b * 709 * , 222 ( 2012 ) ; s. alekhin , a. djouadi and s. moch , phys .
b * 716 * , 214 ( 2012 ) . | typical models with composite higgs bosons are briefly reviewed .
we also introduce the isospin symmetric higgs model recently proposed in ref . @xcite . |
in this paper we study the dynamics of the front separating a spatio - temporally chaotic state from a stable steady ordered state .
such situations occur in many experimental settings . in an experiment on a vertically vibrating granular monolayer of spheres @xcite
both a state at rest on the plate and a chaotically bouncing state are stable . when a small perturbation is applied to the stationary state , the chaotic state
is observed to invade the stable state through a propagating front . in a rayleigh - bnard convection experiment @xcite both straight rolls and spiral defect chaos are stable under some conditions and it is observed that a region of straight rolls is invaded by a region of spiral defect turbulence . for our study
we employ a type of model called a continuum coupled map ( ccm ) introduced in ref .
models of this type ( sec .
ii ) are appropriate to periodically forced systems ( such as that in the experiment of the ref .
@xcite ) . in common with other generic models , like the complex ginzburg - landau equation or the swift - hohenberg equation ,
ccm models are meant to incorporate the minimal basic properties capable of reproducing the phenomena of interest . with this in mind we construct our ccm model to incorporate the essential feature that both a stable steady homogeneous attractor and a spatio - temporally chaotic attractor exist . using our ccm model , we numerically investigate two phenomena : ( i ) the coarsening of the front due to the inherent ` noise ' associated with the spatio - temporal chaos ( sec .
iii ) , and ( ii ) cusp formation induced by initial long wavelength perturbations of the front location from the flat state ( sec .
iv ) . with respect to ( i ) , an important concept used to study various coarsening processes is scaling . for a large number of systems ( e.g. ,
@xcite ) , it is found that the interface width due to roughening , @xmath0 , increases as a power of time , @xmath1 .
the width eventually saturates at a value that increases as a power of the system size , @xmath2 .
these scaling properties are also observed in our model , and we determine and discuss the scaling exponents @xmath3 and @xmath4 that we find . with respect to ( ii ) , we argue that on long scale ( i.e. , long compared to @xmath5 ) , our fronts propagate at constant speed in a direction locally normal to the interface .
we show that this basic property explains the mechanism of cusp formation and the evolution of the shape of the fronts observed in our numerical simulations .
as in reference @xcite , we consider a ccm model which maps a field @xmath6 forward from time @xmath7 to time @xmath8 .
with reference to a system driven periodically in time ( e.g. , as in @xcite ) , we may think of @xmath9 , with @xmath10 as being the system state stroboscopically sampled once each period .
furthermore , we consider @xmath11 to be two - dimensional , and , for simplicity , we take @xmath12 to be a scalar field .
the ccm model mapping @xmath12 to @xmath13 consists of two steps : the first step is a nonlinear local operation in which a one - dimensional map @xmath14 is applied to @xmath15 at each point in space , @xmath16 the second step is a translationally invariant linear operation coupling the dynamics at nearby spatial locations .
the most general such coupling is conveniently expressed in terms of the spatial fourier transform .
if @xmath17 is the spatial fourier transform of @xmath6 , then we write @xmath18 as @xmath19 from which @xmath20 is obtained by inverse fourier transforming . the model
is then specified by the choice of the nonlinear map @xmath21 and the linear spatial coupling @xmath22 .
we make these choices so as to include the minimum properties that we hypothesize are relevant for the investigated phenomena @xcite .
since we desire the simultaneous existence of a stable steady state as well as a spatio - temporally chaotic state , we choose the map @xmath21 to have a stable fixed point attractor and a chaotic attractor . a convenient choice having this property is given by @xmath23 - a { \rm exp}(-1/\sigma^2 ) .
\label{eq : map2}\ ] ] referring to fig .
[ fig : map ] , we see that this map has a stable fixed point at @xmath24 , for @xmath25 .
moreover , any initial point in @xmath26 is attracted to this point .
we also see from fig .
[ fig : map ] that the interval @xmath27 is mapped into itself .
thus there is ( at least ) one attractor in this interval . for the parameter values @xmath28 , and @xmath29 that we investigate ( @xmath30 )
there is one attractor in @xmath31 and it is chaotic .
our choice of the spatial coupling @xmath22 is similarly motivated by a desire for simplicity .
we assume that the coupling is isotropic .
thus we can write @xmath32 as @xmath33 , where @xmath34 .
taking @xmath35 we write @xmath36 , \label{eq : fk}\ ] ] where @xmath37 is a wavenumber - dependent growth / damping rate per period . since we want the spatio - temporal chaos to have a finite spatial correlation scale , such a scale must be reflected in our choice of @xmath37 . denoting this scale by @xmath38 , we make the following simple choice @xcite for @xmath37 , @xmath39 .
\label{eq : couple2}\ ] ] thus @xmath40 ( growth ) for @xmath41 , @xmath37 has its maximum value at @xmath42 , and @xmath37 becomes strongly negative ( damping ) as @xmath43 becomes large .
our numerical implementation of this ccm model employs doubly periodic boundary conditions with periodicity lengths @xmath44 in @xmath45 and @xmath46 in @xmath47 .
the nonlinear map operator is applied at points on a square grid , while the spatial coupling operator ( [ eq : couple1 ] ) employs fast fourier transforming from @xmath48 to @xmath49 and back .
that @xmath50 is an attractor can be seen by introducing an initial perturbation at wavenumber @xmath51exp@xmath52 .
linearization of the ccm model about @xmath53 then shows that this perturbation evolves with time to @xmath54exp@xmath52 , where @xmath55^n , m'(\xi ) \equiv \partial m / \partial \xi$ ] . for the parameters we choose @xmath56 for all @xmath43 ( in particular @xmath57 ) .
thus the homogeneous state @xmath50 is an attractor for the system .
we also find that , as we had anticipated , for other initial conditions there is another attractor which is spatio - temporally chaotic .
figure [ fig : stchaos1 ] shows the properties of the spatio - temporal chaos produced by our model .
figure [ fig : stchaos1](a ) shows the spatial pattern @xmath58 at a representative time .
this picture applies to a time @xmath59 evolved from an initial condition where @xmath60 was chosen randomly with uniform distribution between @xmath61 and @xmath62 .
visually , we observe that the pattern appears to have a characteristic scale of the order of @xmath63 . this is confirmed by the wavenumber power spectrum , fig .
[ fig : stchaos1](b ) .
we note that the only length scales in our model are @xmath63 , the system size @xmath64 , and the grid size @xmath65 , and that , by our choice @xmath66 , we had sought to obtain spatio - temporal chaos with properties independent of @xmath67 and @xmath65 .
figure [ fig : stchaos1](b ) , which evidences variation on the scale @xmath68 , conforms with this expectation .
further discussion of the form observed for @xmath69 is given in the appendix . to characterize the temporal variation of the patterns , fig .
[ fig : stchaos1](c ) shows a plot of the time correlation function @xmath70 defined as @xmath71 where @xmath72 means time average , @xmath73 , and @xmath74 denotes the @xmath75 location of a grid point . as can be seen from fig .
[ fig : stchaos1](c ) the time correlation function decays to zero with increasing time @xmath76 ( where @xmath76 is an integer ) , confirming that the temporal behavior is chaotic .
we have also examined other parameter values for which ( [ eq : map1])-([eq : couple2 ] ) yields spatio - temporal chaos , and we find behavior similar to that in figs . [
fig : stchaos1 ] .
the main objective of our investigation is to characterize the dynamics of the interface between a spatio - temporally chaotic state and a stable steady state . in our first set of simulations we focus on an initially flat interface , @xmath77 at @xmath78 . after we generate an initial spatio - temporally chaotic state ( fig .
[ fig : stchaos1](a ) ) , we create the interface by setting the amplitude @xmath79 of all grid points with @xmath80 to the stable steady state value @xmath24 . during further iterations , the front between the chaotic and steady state moves downward , i.e. , the chaotic state propagates into the stable steady state , and the front coarsens ( see fig .
[ fig : flat1 ] ) . in order to examine the front dynamics for long times and save computational time
, we use a shifting method : on every iterate , we reset @xmath79 to zero in the region adjacent to the bottom ( @xmath81 ) of the periodic box , and , when the front comes close to @xmath82 , we shift the whole system upward in the _ y direction . due to the periodic boundary conditions ,
after this shift there will be a region below the front and above @xmath81 that is in the spatio - temporally chaotic regime , and we then set @xmath83 in this region . _ after an initial transient , the scaling properties of the front are studied using the following definition of the interface width .
first , we calculated the average value of @xmath79 at fixed @xmath84 , @xmath85 .
we then note that the basin boundary between the two attractors of the one dimensional map @xmath21 is at the unstable fixed point , @xmath86 ( fig .
[ fig : map ] ) , and that the average @xmath79 for the spatio - temporally chaotic state ( fig .
[ fig : stchaos1](a ) ) is approximately four times the critical value .
thus , we define a lower boundary of the front , @xmath87 , by @xmath88 and an upper boundary of the front , @xmath89 , by @xmath90 . the width of the front is then defined as @xmath91 . because of the inherent ` noise ' generated in the spatio - temporally chaotic region , the proper quantity to study is the ensemble averaged mean of the front widths .
we calculate ensembles using many different random initial conditions .
our results for the ensemble averaged width @xmath29 of the front are obtained by averaging ( @xmath92 ) over 10 runs for the largest system @xmath93 and over 500 runs for the smallest @xmath94 .
the typical coarsening of the front is shown in fig .
[ fig : flat1 ] .
one observation from our simulations is that the propagation velocity of fronts is constant , except for a few transient initial iterations .
that is , the velocity does not depend on time or system size .
a constant propagation front velocity is also observed in the experiments in refs .
@xcite and @xcite . to minimize the effect of the initial transient , we redefine time as the total increase in the area of the chaotic state . as is typical for a front coarsening problem @xcite , the time and system size dependence of the mean front width @xmath95 can be described by a scaling function @xmath96 , @xmath97 here @xmath98 is constant for @xmath99 and @xmath100 for @xmath101 . for @xmath102 ,
the width saturates at @xmath103 where @xmath104 is the roughness exponent .
barabsi and stanley @xcite have summarized the values of the scaling exponents @xmath105 , @xmath3 , and @xmath4 that are obtained for several experimental systems as well as relevant theoretical results . in fig .
[ fig : kpz](a ) , we show @xmath29 versus @xmath106 for different system sizes .
these data show two characteristic regimes : power law growth , followed by saturation .
the growth exponent , @xmath107 , is calculated by measuring the slopes of straight line fits to the data before saturation , and the roughness exponents @xmath108 is calculated by comparing the saturation widths .
figure [ fig : kpz](b ) shows @xmath109 versus @xmath110 , where @xmath3 and @xmath105 have been adjusted to @xmath111 and @xmath112 ( corresponding to @xmath113 ) .
consistent with eq .
( [ eq : sfunction ] ) , we find collapse of the data in fig .
[ fig : kpz](a ) to a single scaling function .
the exponent values we obtain are roughly consistent with those of both the two dimensional eden model ( @xmath114 ) and the two dimensional kardar - parisi - zhang equation ( @xmath115 ) @xcite .
we now consider the evolution of a front on large length scale .
specifically , we are interested in the case where the front is initially not flat ; that is , the position of the front is initially given by @xmath116 .
furthermore , we assume that , as the front evolves , the scale on which the front position varies , @xmath117 , remains large compared to the front width , @xmath118 to analyze this situation , we consider that a point on the front moves with a normal velocity @xmath119 whose magnitude , @xmath120 , is constant in time .
this assumption also implies that the direction of @xmath119 following the trajectory of a point on the front is constant in time .
this is because the slope of the front , @xmath121 , following a trajectory does not change with time ; i.e. , it depends only on the initial location @xmath122 on the front and not on @xmath106 .
this is illustrated by the construction in fig .
[ fig : cusp1](a ) . as shown in the figure , the trajectory line segments are straight , are all of the same length , @xmath123 , and are normal to both the initial front and to the evolved front . considering an initial front position given by @xmath116 , propagation at a velocity @xmath119 normal to the front then yields @xmath124^{-1/2 } , \\
\label{eq : cusp2 } y(t ) & = & h(x_0 ) - vt [ 1+(h'(x_0))^2]^{-1/2}. \end{aligned}\ ] ] at any given time eqs .
( [ eq : cusp1 ] ) and ( [ eq : cusp2 ] ) specify the front position parametrically with @xmath122 as a parameter . as
an example consider the case of an initial sinusoidal undulation of the front @xmath125 .
as the front propagates , the initial sinusoidal curve becomes distorted so that the maxima become sharp and the minima become broad . as can be seen from fig .
[ fig : cusp1 ] ( a ) this arises because of the converging ( diverging ) of trajectories that originate near maxima ( minima ) .
past a critical time @xmath126 a cusp develops at the maxima of the evolved front @xcite .
the cusp formation time is determined by noting that @xmath127 first becomes zero at @xmath126 . from eq .
( [ eq : cusp1 ] ) we obtain @xmath128 for @xmath129 there are pairs of values of @xmath122 for which the trajectories given by ( [ eq : cusp1 ] ) and ( [ eq : cusp2 ] ) pass through each other . for a given time @xmath106 greater than @xmath130 we refer to the range of @xmath122 for which this occurs as the unphysical range .
the development of the unphysical range is illustrated in fig .
[ fig : cusp1 ] ( b ) where the dashed portion of the curve shows the result of plotting ( [ eq : cusp1 ] ) and ( [ eq : cusp2 ] ) for the unphysical range .
figure [ fig : cusp2 ] shows how cusps develop in time . in fig .
[ fig : cusp2 ] the noisy front curves are from our simulations and the smooth curves are from eqs .
( [ eq : cusp1 ] ) and ( [ eq : cusp2 ] ) with @xmath122 restricted to the physical range .
the two cases shown in fig .
[ fig : cusp2 ] ( namely , @xmath131 and @xmath132 ) illustrate how front roughening becomes of less influence as ( [ eq : nonflat ] ) becomes better satisfied .
for both cases in fig .
[ fig : cusp2 ] the initial sinusoid has amplitude @xmath133 and wavenumber @xmath134 .
thus from ( [ eq : cusp3 ] ) we have @xmath135 . at times
@xmath136 , the roughening is a small effect if @xmath137 is small . since @xmath138 ( assuming @xmath139 ) , we see that @xmath140 , and roughening will be inconsequential for the large scale front evolution if @xmath44 is sufficiently large . the good agreement of the @xmath141 numerical results from our ccm model with the theory , eqs .
( [ eq : cusp1 ] ) and ( [ eq : cusp2 ] ) , confirms that the front does indeed propagate at constant velocity in a direction perpendicular to the interface . in particular , for long length scales
@xmath142 examined in fig .
[ fig : cusp2 ] we see no evidence for curvature dependence of the front velocity . as a comparison
, we have also considered the propagation of a front between two steady homogeneous states . in particular , replacing @xmath143 in eq .
[ eq : map2 ] by @xmath144 we see ( fig .
[ fig : newmap ] ) that there are two stable fixed points , @xmath3 and @xmath4 , and one unstable fixed point @xmath145 . using ( [ eq : newmap ] ) , ( [ eq : couple1 ] ) and([eq : couple2 ] ) with a sinusoidal front with initialization of @xmath4 above the front and at @xmath3 below the front , we find that the front shape evolves smoothly ( with no roughening ) according to eqs .
( [ eq : cusp1 ] ) and ( [ eq : cusp2 ] ) .
thus the cases of a chaotic invading region and a nonchaotic invading region become similar for large @xmath44 ( e.g. , @xmath146 ) . in conclusion ,
we have introduced a continuum coupled map model for the study of the dynamics of a front separating a region of spatio - temporal chaos from a stable steady region .
this model is applicable to periodically forced systems .
we find that the front roughens and that this coarsening obeys a scaling hypothesis , eq .
( [ eq : sfunction ] ) .
we also investigate the large length scale evolution of a nonplanar front .
we find that this evolution is consistent with the hypothesis that , on large scale , the front velocity is constant and normal to the front .
this hypothesis and our numerical simulations indicate the formation of cusp structures in the front .
this work was supported by onr ( physics ) .
the work of scv was supported by a research fellowship from the alfred p. sloan jr .
foundation and by the nsf ( dmr9975533 ) .
we now comment on the specific form that we have found for @xmath147 ( fig . [
fig : stchaos1](b ) ) . in this connection
we note that in the limit of a wildly varying map @xmath143 with lyapunov exponent approaching infinity @xmath148 will be wildly varying in space .
this is because small variation of @xmath15 with @xmath48 are greatly amplified when @xmath21 is applied .
thus , in this limit , the spatial correlation function for @xmath148 will be a delta function , and @xmath149 independent of @xmath49 .
thus from ( [ eq : couple1 ] ) @xmath150 which is plotted in fig .
[ fig : appendix1 ] as the dashed line along with the data from fig .
[ fig : stchaos1](b ) .
it is seen that eq .
( a.1 ) provides a crude indication of the general form of @xmath69 .
w. losert , d. g. w. cooper , and j. p. gollub , phys .
e * 59 , 5855 ( 1999 ) .
i. v. melnikov , d. a. egolf , s. jeanjean , b. b. plapp , and e. bodenschatz , to appear in
stochastic dynamics and pattern formation in biological and complex systems " , edited by s. kim , k. j. lee , t. k. lim , and w. sung , aip conference proceeding 1999 .
s. c. venkataramani , and e. ott , phys .
lett . * 80 , 3495 ( 1998 ) .
barabsi and h. e. stanley , _ fractal concepts in surface growth ( cambridge university press , cambridge , 1995 ) .
this general viewpoint was also adopted in ref .
@xcite where , motivated by experimentally observed phenomena [ f. melo , p. b. umbanhowar , and h. l. swinney , phys .
lett . * 75 , 3838 ( 1995 ) ] , other choices for @xmath21 and @xmath151 were employed . in that case
the objective was to check the hypothesis that period doubling in conjunction with pattern formation at a preferred scale were the essential ingredients necessary to explain the observed bifurcations and the evolution of time periodic patterns occurring in a vertically oscillated granular layer of the order of 10 grains thick .
m. kardar , g. parisi , and y. c. zhang , phys .
lett . * 56 , 889 ( 1986 ) .
cusp formation also occurs for the kardar - parisi - zhang equation ; see fig . 1 of ref
. @xcite . * * _ * * | we study the dynamics of the front separating a spatio - temporally chaotic region from a stable steady region using a simple model applicable to periodically forced systems . in particular , we investigate both the coarsening of the front induced by the inherent ` noise ' of the chaotic region , and the long wavelength dynamics causing the front to develop cusps . ] |
the observation of neutrino oscillation confirms at least two non - zero masses of active neutrinos indicating physics beyond the standard model ( sm ) to generate the neutrino masses .
radiative seesaw models are one of the attractive candidate to generate the neutrino masses where a neutrino mass matrix is induced at loop level and a dark matter ( dm ) candidate can be included as a particle propagating inside a loop diagram for generating neutrino mass .
it is also interesting to include flavor dependent gauge symmetry with which we can obtain predictive structure of neutrino mass matrix @xcite .
one of the interesting flavor dependent @xmath2 gauge symmetry is the @xmath3 which can induce sizable deviation of muon anomalous magnetic dipole moment from sm prediction , @xmath4 , where experimental observation indicates @xmath5 suggesting discrepancy from the sm value .
in addition , some interesting phenomenologies regarding the @xmath3 are investigated , e.g. in refs .
we then apply the @xmath3 gauge symmetry in a radiative seesaw model and investigate prediction in neutrino mass matrix . in this paper
, we construct a radiative seesaw model with @xmath3 gauge symmetry and @xmath6 symmetry in which we introduce exotic @xmath7 doublet leptons with @xmath8 , @xmath6 even singlet scalar field , and @xmath6 odd triplet and singlet scalar fields . in the model , active neutrino mass matrix is generated at one loop level where @xmath6 odd particles propagate inside the loop .
we find that structure of the dirac mass matrix of exotic lepton determines that of the active neutrino mass matrix when exotic leptons have degenerated masses . in that case , we have two zero texture of the neutrino mass matrix which provides some predictions in neutrino oscillation experiments . in addition
, we have dm candidate which is the lightest @xmath6 odd neutral particle .
then we carry out numerical analysis taking into account charged lepton flavor violation ( clfv ) , @xmath4 , and relic density of dm searching for parameter region satisfying all the experimental constraints .
this paper is organized as follows . in sec .
ii , we introduce our model and discuss some phenomenology such as neutrino mass matrix , lepton flavor violation , and some processes induced by @xmath9 interactions .
the numerical analysis is carried out in sec .
iii to search for parameter region satisfying experimental constraints and to obtain some prediction for neutrino mass matrix .
finally we summarize the results in sec . iv .
.field contents of fermions and their charge assignments under @xmath10 . [ cols="^,^,^,^,^,^,^,^,^,^,^ " , ] in this section , we introduce our model and discuss some phenomenology . as extra symmetries , local @xmath3 and discrete @xmath6 symmetries are added . in the fermion sector , we introduce @xmath7 doublet vector like fermions @xmath11^t_{e,\mu,\tau}$ ] , and impose a flavor dependent gauge symmetry @xmath0 as summarized in table [ tab:1 ] .
also @xmath6 odd parity is imposed for this new fermion in order to discriminate the sm model leptons with @xmath7 and forbid the mixing between them .
can not be a dm candidate , because it is ruled out by the direct detection search via z boson portal .
] in the scalar sector , we add an @xmath7 triplet inert scalar @xmath12 real singlet inert scalar @xmath13 , and singlet scalar @xmath14 to the sm higgs @xmath15 as summarized in table [ tab:2 ] .
notice here the higgs doublet @xmath15 ( that spontaneously breaks electroweak symmetry ) , the @xmath16 singlet field @xmath14 ( that spontaneously break @xmath0 symmetry ) , have the vacuum expectation values ( vevs ) , which are respectively symbolized by @xmath17 , @xmath18 , and @xmath6 odd parity is also imposed for the inert scalars @xmath12 and @xmath13 due to forbidding the tree level neutrino masses through vevs .
therefore the lightest neutral scalar boson with @xmath6 odd parity can be a dm candidate .
0_anomaly cancellation _ : the @xmath19 gauge symmetry is anomaly free where the anomaly is canceled within each generation of fermions @xcite .
we then assign @xmath19 charges to first and second generation fermions but charges for third generation fermions are required to be zero .
the triangle anomaly within one generation cancels as follows : @xmath20 ^ 2u(1)_r:\quad 3\left(\frac49x+\frac19(-x)\right)-x=0,\\ & [ u(1)_r]^2u(1)_y:\quad 3\left(\frac23x^2-\frac13 x^2\right)-x^2=0,\\ & [ u(1)_r]^3:\quad 3\left(x^3-x^3\right)-x^3+x^3=0,\\ & u(1)_r:\quad 3\left(x - x\right)-x+x=0.\end{aligned}\ ] ] _ yukawa interactions _ : under these fields and symmetries , the renormalizable lagrangians for quark and lepton sector are given by @xmath21 { \nonumber}\\ & + y_{\delta_1 } \bar l_{l_e}^c ( i\sigma_2)\delta l'_{l_e } + y_{\delta_2 } \bar l_{l_\tau}^c ( i\sigma_2)\delta l'_{l_\mu } + y_{\delta_3 } \bar l_{l_\mu}^c ( i\sigma_2)\delta l'_{l_\tau } { \nonumber}\\ & + y_{e_1 } \varphi^ * \bar l'_{l_e } l'_{r_\mu } + y_{e_2 } \varphi \bar l'_{l_e } l'_{r_\tau } + { \rm c.c . } , \label{eq : lag - quark}\end{aligned}\ ] ] where @xmath22 is the second pauli matrix , and again @xmath23^t$ ] .
0 after two higgs doublet develops vevs , we obtain the quark mass matrix such that @xmath24 which is the same structure as discussed in ref . @xcite . as in the sm ,
the quark mass matrices are diagonalized by unitary matrices @xmath25 and @xmath26 which change quark basis @xmath27 .
then the ckm matrix is given by @xmath28 .
the ckm matrix can be approximated as @xmath29 since @xmath30 correspond to right - handed rotations , and @xmath31 is obtained @xcite .
the details of quark yukawa couplings with two higgs doublet are discussed in ref .
@xcite , and we omit the further discussion here . _
@xmath9 couplings to sm fermions _ : the @xmath9 couplings to the sm fermions are written as @xmath32 where @xmath33 is the gauge coupling constant associated with the @xmath34 .
the coupling matrices for down - type quarks are given by @xmath35 where @xmath36s are elements of ckm matrix and we applied the relation @xmath37 as we discussed above .
we parametrize the scalar fields as @xmath38,\quad \delta = \left [ \begin{array}{cc } \frac{\delta^+}{\sqrt2 } & \delta^{++}\\ \delta^{0 } & -\frac{\delta^+}{\sqrt2 } \end{array}\right ] , \quad \delta_0=\frac{\delta_r + i\delta_i}{\sqrt2},\quad \varphi=\frac{v'+\rho+i z'}{\sqrt2 } \label{component } \ ] ] where @xmath39 gev is vev of the higgs doublet , and @xmath40 , @xmath41 , and @xmath42 are respectively gb which are absorbed by the longitudinal component of @xmath43 , @xmath44 , and @xmath45 boson . then we have two neutral boson mass matrices @xmath46 and @xmath47 in the basis of @xmath48^t$ ] and @xmath49^t$ ] , and these are diagonalized by @xmath50diag[@xmath51 and @xmath52diag[@xmath53 respectively , where the mixing source of @xmath54 arises from the nontrivial quartic coupling @xmath55 and each of mass eigenstate can be written in terms of couplings of higgs potential . here
we define mixing matrices as @xmath56,\quad s_a=\frac{2\lambda_{\phi\varphi}}{m_{h_1}^2-m_{h_2}^2},\quad s_\alpha=\frac{2\sqrt2\lambda_{0}}{m_{h_1}^2-m_{h_2}^2},\end{aligned}\ ] ] where @xmath57 is the short - hand notation of @xmath58 .
notice here that we suppose to be @xmath59 below analysis , which could however be an natural assumption because @xmath60 from the lhc experiment , therefore @xmath61 and @xmath62 .
after the @xmath63 gauge symmetry breaking , _ vector - like fermion mass matrix _ can be written in the basis @xmath64^t$ ] as follows : @xmath65,\label{eq : ml'}\end{aligned}\ ] ] where we simply assume to be positive real for each component and define @xmath66 and @xmath67
. then @xmath68 is diagonalized by orthogonal mixing matrix @xmath69 ( @xmath70 ) as @xmath71,\quad n_{{e,\mu,\tau}}=v n_{{1,2,3}},\label{eq : n - mix}\end{aligned}\ ] ] where @xmath72 is the mass eigenstate . and
@xmath73 which is dirac phase , while the right side of figure represents correlation between @xmath74 and @xmath75 which are majorana phases . here
the pink region in the left side figure is given by @xmath76 $ ] at 90 % confidential level with the best fit value @xmath77 , and the blue points in the right side figure correspond to the parameter sets providing the points in the pink region .
, title="fig:",width=283 ] and @xmath73 which is dirac phase , while the right side of figure represents correlation between @xmath74 and @xmath75 which are majorana phases . here
the pink region in the left side figure is given by @xmath76 $ ] at 90 % confidential level with the best fit value @xmath77 , and the blue points in the right side figure correspond to the parameter sets providing the points in the pink region .
, title="fig:",width=283 ] _ our active neutrino mass matrix _ is given in general at one - loop level by the diagram shown in fig .
[ fig : neutrino ] which is calculated as @xcite @xmath78^t , \label{eq : neut - mass}\\ \epsilon & \equiv \left[\begin{array}{ccc } 1 & 0 & 0 \\ 0 & 0 & 1 \\ 0 & 1 & 0 \end{array}\right],\quad r_k = \frac{s_\alpha c_\alpha}{(4\pi)^2 } \left[\frac{r_{k_2}\ln r_{k_2}}{1-r_{k_2}}-\frac{r_{k_1}\ln r_{k_1}}{1-r_{k_1}}\right],\end{aligned}\ ] ] where @xmath79 , @xmath80 and @xmath81 are diagonal yukawa matrices respectively .
then the neutrino mass is diagonalized by 3 by 3 unitary matrix @xmath82 as @xmath83 , where @xmath84 $ ] is neutrino mass eigenvalues @xcite .
applying eq .
( [ eq : neut - mass ] ) and an arbitral complex anti - symmetric matrix @xmath85 to @xmath83 , one finds a relation @xmath86 = @xmath87 @xcite . in order to obtain some predictions to the neutrino sector here
, we impose an assumption @xmath88 .
then @xmath89 is simplified as @xmath90 \nonumber \\ & = r ( y_{\delta}\epsilon m_{l'}y_s+ y_sm_{l'}\epsilon y_{\delta}){\nonumber}\\ & = r \left[\begin{array}{ccc } 2 m_e y_{s_e}y_{\delta_1 } & m_{e\tau}y_{s_e}y_{\delta_2}+m_{e\mu}y_{s_\mu}y_{\delta_1 } & m_{e\mu}y_{s_e}y_{\delta_3}+m_{e\tau}y_{s_\tau}y_{\delta_1 } \\ m_{e\tau}y_{s_e}y_{\delta_2}+m_{e\mu}y_{s_\mu}y_{\delta_1 } & 0 & m_\mu y_{s_\mu}y_{\delta_3}+m_\tau y_{s_\tau}y_{\delta_2 } \\
m_{e\mu}y_{s_e}y_{\delta_3}+m_{e\tau}y_{s_\tau}y_{\delta_1 } & m_\mu y_{s_\mu}y_{\delta_3}+m_\tau y_{s_\tau}y_{\delta_2 } & 0 \end{array}\right]\label{eq : type - c}\\ & \approx \left[\begin{array}{ccc } 0.022 - 0.75 & 0.028 - 0.039 & 0.030 - 0.040 \\ 0.028 - 0.039 & 0 & 0.023 - 0.75
\\ 0.030 - 0.040 & 0.023 - 0.75 & 0 \end{array}\right ] [ { \rm ev } ] , \label{eq : exp - neut - oscil}\end{aligned}\ ] ] where we have used @xmath91 .
( [ eq : type - c ] ) corresponds to the type c two zero texture that provides several predictions that only an inverted neutrino mass ordering is allowed and specific pattern of @xmath73 in the right side of fig . [ fig : phases ] , therefore the maximum value of @xmath92ev is in favored in our model . ] and specific majorana phases ( @xmath93 ) in the right side of fig . [
fig : phases ] are obtained , where we adapt the recent global neutrino oscillation data @xcite up to 3@xmath75 confidential level in the section of numerical analysis .
we find that the region of @xmath75 is restricted when the dirac phase is within @xmath76 $ ] which is @xmath94 confidence level range with the best fit value @xmath95 .
we obtain these numerical values of the last eq .
( [ eq : exp - neut - oscil ] ) by using their central values of neutrino oscillation data . in the numerical analysis
below , we will show this prediction combined with the other phenomenologies such as lfvs and dm + _ lepton flavor violations(lfvs ) _ arises from the term @xmath81 and @xmath80 at one - loop level , and its form can be given by @xmath96 + 2 m_{\ell_i } ( 2f_2[m_k , m_{\delta^{\pm\pm}}]+f_2[m_{\delta^{\pm\pm}},m_k ] ) \right ) \right.{\nonumber}\\ & - \left . y^\dag_{s_{ki } } y_{s_{jk } } m_{\ell_i } \left(c_\alpha^2 f_2 [ h_1,m_k ] + s_\alpha^2 f_2 [ h_2,m_k ] \right)\right],\\
\quad a_{l_{ij } } & = \frac{1}{(4\pi)^2 } \sum_{k=1,2,3}\left [ \frac{y^\dag_{\delta_{ki } } y_{\delta_{jk}}}{2}\left(m_{\ell_i } f_2 [ m_k , m_{\delta^\pm } ] + 2 m_{\ell_j } ( 2f_2[m_k , m_{\delta^{\pm\pm}}]+f_2[m_{\delta^{\pm\pm}},m_k ] ) \right ) \right.{\nonumber}\\ & - \left . y^\dag_{s_{ki } } y_{s_{jk } } m_{\ell_j } \left(c_\alpha^2 f_2 [ h_1,m_k ] + s_\alpha^2 f_2 [ h_2,m_k ] \right)\right],\\ & f_{2}(m_a , m_b)=\frac{2 m_a^6 + 3 m_a^4 m_b^2 -6 m_a^2 m_b^4 + 6 m_b^6 + 12 m_a^4 m_b^2 \ln(m_b / m_a)}{12(m_a^2-m_b^2)^4},\end{aligned}\ ] ] where @xmath97 is the singly charged component of @xmath98 , @xmath99[gev]@xmath100 is the fermi constant , @xmath101 is the fine structure constant , @xmath102 , @xmath103 , and @xmath104 .
experimental upper bounds are respectively given by @xmath105 , @xmath106 , and @xmath107 .
+ _ new contributions to the muon anomalous magnetic moment ( muon @xmath1 : @xmath4 ) _ arises from yukawa terms @xmath81 with negative contribution and @xmath80 with positive contribution . also another source via additional gauge sector can also be induced by @xmath108_{\mu\mu } , \quad \delta a_{\mu}^{z'}\approx
\frac{g_{z'}^2}{8\pi^2}\int_0 ^ 1 da \frac{2 r a ( 1-a)^2}{r(1-a)^2+a } , \label{eq : g2-zp}\end{aligned}\ ] ] where @xmath109 , and @xmath9 is the new gauge vector boson .
thus we could explain the sizable muon @xmath1 @xmath110)$ ] @xcite , if we can satisfy the constraint of trident process .
notice here that @xmath111
@xcite has to be satisfied due to the trident process , and we use the allowed range @xmath112 in our numerical analysis below . here
we consider the lightest inert boson @xmath113 , and assume to neglect @xmath114 and any higgs couplings to simply evade the constraint of direct detection searches through higgs portal . in this case ,
dominant dm annihilation processes are @xmath115 via yukawa interactions associated with yukawa couplings @xmath80 and @xmath81 , which are shown in fig .
[ fig : dm ] . then the non - relativistic cross section to explain the relic density of dm is @xmath116-wave dominant , and its effective form is given by @xmath117 where @xmath118 .
then the relic density @xmath119 is simply given by @xcite @xmath120 where @xmath121[gev ] is the planck mass , @xmath122 is the total number of effective relativistic degrees of freedom at the time of freeze - out , and @xmath123 is defined by @xmath124 at the freeze out temperature ( @xmath125 ) .
the measured relic density is reported by planck that suggests @xmath126 @xcite . in our numerical analysis
below , however , we will adopt more relaxed value @xmath127 .
in this section , we show a global analysis , where we have fixed some parameters for simplicity . at first , we fix @xmath128 in order to evade the oblique parameters in the triplet boson .
we also fix these three degenerated mass eigenvalues and corresponding rotation matrix used for diagonalizing @xmath68 as follows : @xmath129 , \quad v \approx \left[\begin{array}{ccc } 0.707 & -0.707 & -0.000232 \\ -0.412 & \-0.413 & 0.812 \\
-0.574 & -0.574 & -0.583 \end{array}\right].\end{aligned}\ ] ] then we randomly select the following range of input parameters as @xmath130\ [ { \rm gev}],\ m_{z'}\in [ 10^{-3},10
^ 3]\ [ { \rm gev } ] , \
m_{h_2 } \in [ 1.2m_x,2500]\ [ { \rm gev } ] , { \nonumber}\\ & |y_{s_e}| \in[0.1 , 4\pi],\ |y_{s_\tau}| \in[0.1 , 4\pi],\ |s_\alpha| \in[10^{-5},0.1],\ g_{z ' } \in[10^{-5},10^{-3}],\end{aligned}\ ] ] where the lower bound @xmath131 gev originates from the relic density of dm , yukawa couplings @xmath81 and @xmath132 can be rewritten in terms of @xmath68 , @xmath133 , and neutrino oscillation data in eqs .
( [ eq : type - c ] ) and ( [ eq : exp - neut - oscil ] ) , where we have used the central value of eq .
( [ eq : exp - neut - oscil ] )
. then we explore the allowed region to satisfy all the constraints such as neutrino oscillation , lfvs , muon @xmath1 , and relic density of dm as discussed above . in the numerical analysis
, we firstly find that the yukawa contribution to the muon @xmath1 @xmath134 is at most the order of @xmath135 , therefore @xmath136 . in this case ,
notice here that we have to consider the constraint from neutrino trident production process @xcite .
this is because we take @xmath111 as discussed in the last part of muon @xmath1 . in the left side of fig .
[ fig : dm - mh2-zp ] , we show the allowed region to satisfy all the data in terms of @xmath137 and @xmath138 , and its form directly reflects eq .
( [ eq : g2-zp ] ) . and
this figure provide the upper bound on @xmath139 . while the right side of fig .
[ fig : dm - mh2-zp ] represents the allowed region to satisfy all the data in terms of @xmath140 and @xmath141 , and it suggests all the region that we take are allowed . , we show the allowed region to satisfy all the data in terms of @xmath137 and @xmath138 , and its form directly reflects eq .
( [ eq : g2-zp ] ) . and
this figure provide the upper bound on @xmath139 . while the right side of fig .
[ fig : dm - mh2-zp ] represents the allowed region to satisfy all the data in terms of @xmath140 and @xmath141 , and it suggest all the region that we take are allowed .
, title="fig:",width=264 ] , we show the allowed region to satisfy all the data in terms of @xmath137 and @xmath138 , and its form directly reflects eq .
( [ eq : g2-zp ] ) . and
this figure provide the upper bound on @xmath139 . while the right side of fig .
[ fig : dm - mh2-zp ] represents the allowed region to satisfy all the data in terms of @xmath140 and @xmath141 , and it suggest all the region that we take are allowed .
, title="fig:",width=264 ]
we have proposed an predictive radiative seesaw model at one - loop level with a flavor dependent gauge symmetry @xmath0 , in which we have consider gauge singlet -like bosonic dark matter candidate .
and we have obtained a specific two zero texture ( type - c ) of active neutrino mass matrix that provides us several predictions such as inverted mass ordering , specific pattern of dirac cp phase and majorana phases .
here we have assumed that three of exotic fermion masses are almost degenerated , which is an minimal requirement to obtain such a specific neutrino texture .
we have also carried out a numerical analysis to search for parameter region which satisfy all the experimental data such as constraints from charged lepton flavor violations , sizable muon @xmath1 , relic density of dark matter , as well as neutrino oscillation data . in this analysis , we have firstly found that @xmath142 is negligible compared to the measured value @xmath143 , therefore @xmath136 . in the left side of fig . [
fig : dm - mh2-zp ] , we have shown the allowed region to satisfy all the data in terms of @xmath137 and @xmath138 , and its form directly reflects eq .
( [ eq : g2-zp ] ) . and
this figure provide the upper bound on @xmath139 . while the right side of fig .
[ fig : dm - mh2-zp ] has represented the allowed region to satisfy all the data in terms of @xmath140 and @xmath141 , and it suggests all the region that we have taken are allowed . in our model , we have inert doubly charged higgs boson which decay into dark matter and sm fermions by cascade decay modes .
it will be interesting to search for the signal of `` missing @xmath144 + same sign leptons '' as a signature of the inert higgs triplet as well as our dark matter .
the detailed analysis of the signal is beyond the scope of this paper and it will be studied elsewhere .
here we give the most general higgs potential in a renormalizable theory as @xmath145+m_s^2s^2{\nonumber}\\ & + ( \lambda_0 \phi^t(i\sigma_2)\delta^\dag \phi s+{\rm c.c . } ) + \lambda_\phi |\phi^\dag\phi|^2 + \lambda_\varphi |\varphi^*\varphi|^2 + \lambda_\delta ( { \rm tr}[\delta^\dag\delta])^2 + \lambda_\delta ' { \rm det}[\delta^\dag\delta ] + \lambda_ss^4{\nonumber}\\ & + \lambda_{\phi\varphi}(\phi^\dag\phi ) ( \varphi^*\varphi ) + \lambda_{\phi\delta}(\phi^\dag\phi ) { \rm tr}[\delta^\dag\delta ] + \lambda_{\phi\delta}'\sum_{i=1 - 3}(\phi^\dag\tau_i\phi ) { \rm tr}[\delta^\dag\tau_i\delta ] + \lambda_{\phi s}(\phi^\dag\phi)s^2{\nonumber}\\ & + \lambda_{\varphi \delta}\varphi^*\varphi { \rm tr}[\delta^\dag\delta ] + \lambda_{\varphi s}\varphi^*\varphi s^2 + \lambda _ { \delta s } { \rm tr}[\delta^\dag\delta ] s^2.\end{aligned}\ ] ]
h. o. is sincerely grateful for all the kias members , korean cordial persons , foods , culture , weather , and all the other things . x. g. he , g. c. joshi , h. lew and r. r. volkas , phys .
d * 43 * , 22 ( 1991 ) ; d * 44 * , 2118 ( 1991 )
. s. baek , n. g. deshpande , x. g. he and p. ko , phys .
d * 64 * , 055006 ( 2001 ) [ hep - ph/0104141 ] .
e. ma , d. p. roy and s. roy , phys .
b * 525 * , 101 ( 2002 ) [ hep - ph/0110146 ] .
e. j. chun and k. turzynski , phys .
d * 76 * , 053008 ( 2007 ) [ hep - ph/0703070 ] .
s. baek and p. ko , jcap * 0910 * , 011 ( 2009 ) [ arxiv:0811.1646 [ hep - ph ] ] .
k. harigaya , t. igari , m. m. nojiri , m. takeuchi and k. tobe , jhep * 1403 * , 105 ( 2014 ) [ arxiv:1311.0870 [ hep - ph ] ] .
w. altmannshofer , s. gori , m. pospelov and i. yavin , phys .
d * 89 * , 095033 ( 2014 ) [ arxiv:1403.1269 [ hep - ph ] ] . j. heeck , m. holthausen , w. rodejohann and y. shimizu , nucl .
b * 896 * , 281 ( 2015 ) [ arxiv:1412.3671 [ hep - ph ] ] .
t. araki , f. kaneko , y. konishi , t. ota , j. sato and t. shimomura , phys .
d * 91 * , no . 3 , 037301 ( 2015 ) [ arxiv:1409.4180 [ hep - ph ] ] . s. baek , h. okada and k. yagyu , jhep * 1504 * , 049 ( 2015 ) [ arxiv:1501.01530 [ hep - ph ] ] . t. araki , f. kaneko , t. ota , j. sato and t. shimomura , phys .
d * 93 * , no . 1 , 013014 ( 2016 ) [ arxiv:1508.07471 [ hep - ph ] ] . s. baek , phys .
b * 756 * , 1 ( 2016 ) [ arxiv:1510.02168 [ hep - ph ] ] .
p. ko , t. nomura and h. okada , arxiv:1701.05788 [ hep - ph ] .
p. ko , t. nomura and h. okada , arxiv:1702.02699 [ hep - ph ] .
h. okada and y. orikasa , phys . rev .
d * 94 * , no . 5 , 055002 ( 2016 ) [ arxiv:1512.06687 [ hep - ph ] ] .
w. altmannshofer , s. gori , m. pospelov and i. yavin , phys .
* 113 * , 091801 ( 2014 ) [ arxiv:1406.2332 [ hep - ph ] ] .
m. srednicki , r. watkins and k. a. olive , nucl .
b * 310 * , 693 ( 1988 ) .
p. a. r. ade _ et al .
_ [ planck collaboration ] , astron .
astrophys . * 571 * , a16 ( 2014 ) [ arxiv:1303.5076 [ astro-ph.co ] ] . | we study an predictive radiative seesaw model at one - loop level with a flavor dependent gauge symmetry @xmath0 , in which we consider bosonic dark matter .
we obtain a specific two zero texture with inverse mass matrix that provides us several predictions such as a specific pattern of dirac cp phase .
we also analyze the constraint of lepton flavor violations , muon @xmath1 , relic density of dark matter , and collider physics , have numerical analysis and show allowed region to satisfy all the constraints .
kias - p17013 |
superfluidity is a striking property of quantum fluids at very low temperatures . for bosonic systems , important examples are liquids and clusters of @xmath0he and atomic bose - einstein condensates . in fermionic systems , superfluidity is a more intricate phenomenon as it requires pairing of particles .
fermionic superfluidity is known to occur in atomic nuclei and @xmath1he liquids and it is also at the heart of superconductivity , thus being of great technological importance .
recent advances with ultracold fermi gases have opened up unprecedented possibilities to study the properties of strongly interacting fermionic superfluids @xcite .
early experiments on ultracold fermi gases with resonant interparticle interactions compiled increasing evidence for superfluidity @xcite until the phenomenon was firmly established by the observation of vortex lattices @xcite . here we report on the manifestation of superfluidity in a quenched moment of inertia ( moi ) in a strongly interacting fermi gas that undergoes slow rotation .
the basic idea of a quenched moi as a signature of superfluidity dates back to more than 50 years ago in nuclear physics , where mois below the classical , rigid - body value were attributed to superfluidity @xcite .
the quenching of the moi was also shown in liquid @xmath0he @xcite and has , more recently , served for the discovery of a possible supersolid phase @xcite . here
we introduce the observation of the quenched moi as a new method to study superfluidity in ultracold fermi gases .
the basic situation that underlies our experiments is illustrated in fig .
[ idea ] . at a finite temperature below the critical temperature @xmath2 , the harmonically trapped cloud consists of a superfluid core centered in a collisionally hydrodynamic cloud .
we assume that the trapping potential is close to cylindrical symmetry , but with a slight , controllable deformation that rotates around the corresponding axis with an angular velocity @xmath3 .
the nonsuperfluid part of the cloud is then subject to friction with the trap and follows its rotation with an angular velocity @xmath4 @xcite , which in a steady state ideally reaches @xmath5 .
the corresponding angular momentum can be expressed as @xmath6 , where @xmath7 denotes the moi .
the superfluid core can not carry angular momentum , assuming that vortex nucleation is avoided , and therefore does not contribute to the moi of the system .
thus @xmath7 represents the moi of the whole system .
the case of a rotating system in a steady state , where the normal part carries the maximum possible angular momentum , allows us to distinguish the superfluid quenching of the moi from a non - equilibrium quenching effect as studied in ref .
there the authors investigated the hydrodynamic expansion of a gas with a known angular momentum .
this situation , where the velocity fields of the normal and superfluid components are not in a steady state , can also be discussed in terms of a moi below the rigid - body value .
in contrast to the phenomenon investigated in our present work , the effect of ref .
@xcite is related to irrotational flow and can occur for both the superfluid and the collisionally hydrodynamic normal phase .
, which in an equilibrium state approaches the rotation frequency @xmath3 that is imposed by the trap .
the superfluid core ( sf ) does not carry angular momentum and therefore does not contribute to the moi . ]
our measurements rely on the possibility to determine the total angular momentum @xmath8 of a rotating hydrodynamic cloud by detecting the precession of a radial quadrupole excitation .
this method is well established and has been extensively used in the context of atomic bose - einstein condensates @xcite .
we have recently applied it to a rotating , strongly interacting fermi gas to investigate the slow decay of angular momentum @xcite .
the method works under the general condition that the gas behaves hydrodynamically .
then the precession frequency can be written as @xmath9 @xcite , where @xmath10 corresponds to a moment of inertia as calculated from the density distribution under the assumption that the whole cloud , including the superfluid part , would perform a rigid rotation .
substituting @xmath11 for @xmath8 , we obtain @xmath12 , with @xmath13 for the full moi in a normal system , and @xmath14 for a moi that is quenched because of the superfluid core .
the starting point of our experiments is an optically trapped , strongly interacting fermi gas consisting of an equal mixture of @xmath15li atoms in the lowest two atomic states @xcite .
the broad 834-g feshbach resonance @xcite allows us to control the @xmath16-wave interaction .
if not otherwise stated , the measurements presented here refer to the resonance center . here
a unitarity - limited fermi gas @xcite is realized , which is known to exhibit deep hydrodynamic behavior even well above the critical temperature for superfluidity , see e.g. @xcite .
the cigar - shaped quantum gas is confined in a far red - detuned , single - beam optical dipole trap with additional axial magnetic confinement .
the trap can be well approximated by a harmonic potential with radial oscillation frequencies @xmath17hz and an axial frequency of @xmath18hz .
the fermi energy of the noninteracting gas is given by @xmath19 , where @xmath20 is the total atom number .
the fermi temperature is @xmath21k , with @xmath22 denoting the boltzmann constant .
our scheme to study the rotational properties is described in detail in ref .
it is based on a rotating elliptical deformation of the trap , characterized by a small ellipticity parameter @xcite @xmath23 .
in contrast to our previous work , we use a lower rotation frequency of @xmath24hz @xmath25 .
this low value allows us to avoid a resonant quadrupole mode excitation , which is known as an efficient mechanism for vortex nucleation @xcite . to excite the quadrupole mode @xcite we switch on an elliptic trap deformation for 50@xmath26s .
we detect the resulting oscillation by taking absorption images of the cloud after a variable hold time in the trap and a short free expansion time after release from the trap .
more details on this excitation and detection scheme are given in ref .
@xcite . at this point
it is important to discuss the consequences of residual trap imperfections , still present when we attempt to realize a cylindrically symmetric optical potential . as we showed in previous work @xcite
, we can control the ellipticity down to a level of @xmath271% .
moreover , deviations from perfect cylindrical symmetry may occur because of other residual effects , such as corrugations of the optical trapping potential . as a consequence ,
a certain rotational damping is unavoidable , but damping times can reach typically one second @xcite .
this has two main effects for our observations .
first , our measurements yield precession frequencies slightly below @xmath28 . this is because of a delay time of @xmath29ms between turning off the rotating trap ellipticity and applying the quadrupole mode excitation .
it is introduced to make sure that any possible collective excitation resulting from the rotating trap has damped out when the mode precession is measured .
because of rotational damping during this delay time , the measured precession frequencies @xmath30 are somewhat below @xmath28 . to compensate for this effect
, we directly measure the reduction of @xmath28 that occurs during a 20ms hold time to determine the corresponding damping parameter @xmath31 for each set of measurements , finding day - to - day variations with typical values between @xmath32 and @xmath33 .
the second effect is induced by friction with static ( nonrotating ) trap imperfections when the rotating ellipticity is applied .
this leads to equilibrium values for @xmath4 typically a few percent below @xmath3 , depending on the ratio between the time constants for spin up and damping @xcite .
for this second effect there is no straightforward compensation , and it needs to be explicitly discussed when interpreting the experimental results . versus spin - up time @xmath34 for various values of the final temperature , as characterized by the heating parameter @xmath35 ( see text ) .
the quenching of the moi shows up in the temperature - dependent saturation behavior .
the applied timing sequence to facilitate measurements at constant temperature is illustrated above the graph .
for these sets of measurements @xmath36 . ]
thermometry is performed after the whole experimental sequence .
we damp out the rotation by stopping the trap rotation and keeping the ellipticity @xcite .
we convert the gas into a weakly interacting one by a slow magnetic field ramp to @xmath37 g , and we finally measure the temperature @xmath38 @xcite .
note that the isentropic conversion tends to decrease the temperature such that @xmath38 is always somewhat below the temperature @xmath39 at unitarity @xcite .
the relative statistical uncertainty of the temperature measurement is about @xmath40 in the relevant temperature range .
to discuss our experimental results we introduce a dimensionless precession parameter @xmath41 by normalizing our observable @xmath28 to its maximum possible value of @xmath42 , @xmath43 the maximum possible value of @xmath44 corresponds to a fully rotating , classically hydrodynamic cloud .
values @xmath45 show the presence of at least one of the two effects , namely the incomplete rotation of the normal part ( @xmath46 ) or the superfluid quenching of the moi ( @xmath47 ) .
it is crucial for the interpretation of our experimental results to distinguish between these two effects . our basic idea to achieve this relies on the fact that @xmath48 represents a temperature - dependent _ equilibrium _ property , whereas @xmath49 depends on the dynamics of the spin - up before the system has reached an equilibrium .
experimentally , however , measurements of equilibrium properties at a fixed temperature are not straightforward because of the presence of residual heating leading to a slow , steady temperature increase . in the rotating trap
we always observe some heating , which under all our experimental conditions can be well described by a constant rate @xmath50nk / s = @xmath51/s @xcite . to identify the conditions under which our cloud reaches its equilibrium state of rotation
, we have developed a special procedure based on the timing scheme illustrated on top of fig.[quench ] .
our procedure takes advantage of the constant heating rate @xmath52 to control the final temperature of the gas when @xmath41 is measured .
we apply the trap rotation in two separate stages of duration @xmath53 and @xmath34 . in an intermediate time interval of @xmath54ms @xcite
we damp out the rotation that is induced by the first stage .
the angular momentum disappears , but the heating effect remains @xcite .
the second stage spins up the cloud again and induces further heating .
when @xmath55 is kept constant , we find that the total heating by the two rotation stages is @xmath56 .
as only the second stage leads to a final angular momentum , the equilibrium state reached at a constant temperature can be identified when @xmath57 reaches a constant value for increasing @xmath34 and fixed @xmath58 .
the temperature can be controlled by a variation of the parameter @xmath58 and is obtained as @xmath59 .
the temperature offset @xmath60 is set by the initial cooling and some unavoidable heating during the experimental sequence without trap rotation . under our conditions @xmath61 .
our experimental results for @xmath57 are shown in fig.[quench ] for four different values of the heating parameter @xmath62 in a range between @xmath63 to @xmath64 , which corresponds to a range of @xmath38 between about @xmath65 and @xmath66 .
all four curves show qualitatively the same behavior . within a few 100ms
, @xmath41 rises before reaching a final equilibrium value .
this time - dependent increase of @xmath41 is related to the spin - up dynamics @xcite .
we find that the observed increase and saturation of @xmath57 can be well fit by simple exponential curves ( solid lines ) , and we use these fits to extract the different equilibrium values @xmath67 .
the equilibrium values @xmath67 exhibit an interesting temperature dependence .
the lower three values show a pronounced increase with temperature , @xmath68 , @xmath69 , and @xmath70 for @xmath71 , @xmath72 , and @xmath73 , respectively .
we interpret this increase as a consequence of the decreasing superfluid core and thus the decreasing moi quenching effect . for our highest temperature ( @xmath74 )
we only observe a marginal further increase to @xmath75 .
this indicates that the superfluid core is very small or absent leading to a disappearance of the quenching effect .
the fact that the maximum @xmath67 stays a few percent below @xmath76 can be explained by trap imperfections as discussed in sec . [
sec : procs ] .
let us comment on the possible influence of vortices @xcite .
we can not exclude their presence @xcite , as their nucleation can proceed not only via a resonant quadrupole mode excitation @xcite , but also via a coupling to the thermal cloud @xcite .
vortices would result in additional angular momentum in the rotating cloud and its collective behavior would be closer to the normal case .
this would tend to increase @xmath41 at lower temperatures , counteracting the behavior that we observe . in a second set of experiments , we study the superfluid phase transition in a way which is experimentally simpler , but which requires information on the equilibrium state as obtained from the measurements presented before .
the trap rotation is applied continuously , and we observe the increase of @xmath41 with the rotation time @xmath77 . all other parameters and procedures are essentially the same as in the measurements before .
here the temperature is not constant , but rises according to @xmath78 , where the heating rate @xmath79nk / s is the same as before and @xmath80 is somewhat lower because of the less complex timing sequence . as a function of the rotation time @xmath77 (
filled symbols ) ; the upper scale shows the corresponding temperature @xmath38 . for comparison ,
the crosses show the equilibrium values @xmath81 as obtained from fig .
[ quench ] .
the shaded region indicates the range in which we expect the superfluid phase transition according to previous experiments @xcite .
for this set of measurements @xmath82.,title="fig : " ] + figure [ spinup ] shows how @xmath41 increases with the rotation time @xmath77 ( filled symbols ) ; the upper scale shows the corresponding temperature @xmath38 .
the observed increase of @xmath41 generally results from both factors in eq.([idea ] ) , corresponding to the rising @xmath83 ( spin - up dynamics ) and the rising @xmath84 ( decrease of the superfluid moi quenching ) .
figure [ spinup ] also shows the values @xmath81 as determined from fig .
[ quench ] ( crosses ) , for which we know that the spin - up of the normal component has established an equilibrium with @xmath83 being close to one .
the comparison shows that already for @xmath85s the data set obtained with the simpler procedure follows essentially the same behavior . the small quantitative difference that the crosses are slightly below the open symbols
can be explained by a somewhat stronger influence of trap imperfections in the earlier measurements of sec .
[ sec : equilib ] @xcite or by the uncertainty in the initial temperature @xmath60 . for @xmath86s
, we can assume that the system is in an equilibrium state , which follows the slowly increasing temperature , and we can fully attribute the further increase of @xmath41 to the quenching of the moi .
the superfluid phase transition corresponds to the point where the precession parameter @xmath41 reaches its saturation value .
this is observed for a time @xmath87s , when @xmath88 .
the conversion of this temperature parameter ( measured in the weakly interacting regime after an isentropic change ) to the actual temperature in the unitarity - limit regime @xcite yields a value for the critical temperature @xmath2 of about @xmath89 .
this result is consistent with previous experimental results @xcite , the range of which is indicated by the shaded region in fig . [ spinup ] .
the result is also consistent with theoretical predictions @xcite . for a more precise extraction of @xmath2 from experimental moi quenching data
, a theoretical model would be required that describes the saturation behavior of @xmath84 as @xmath2 is approached .
theoretical predictions are available for the bec limit @xcite and the bcs limit @xcite . in the unitarity limit it should ,
in principle , be possible to extract the moi from spatial profiles of the normal and the superfluid fraction @xcite .
clearly , more work is necessary to quantitatively understand the quenching effect in the strongly interacting regime .
we have demonstrated the quenching of the moment of inertia that occurs in a slowly rotating , strongly interacting fermi gas as a consequence of superfluidity .
this effect provides us with a novel probe for the system as , in contrast to other common methods such as expansion measurements and studies of collective modes , it allows us to distinguish between the two possible origins of hydrodynamic behavior , namely collisions in a normal phase and superfluidity .
we thank l. sidorenkov and m. k. tey for discussions . we acknowledge support by the austrian science fund ( fwf ) within sfb 15 ( project part 21 ) and sfb 40 ( project part 4 ) .
10 url # 1#1urlprefix[2][]#2 giorgini s , pitaevskii l p and stringari s 2008 _ rev .
* 80 * 12151273 inguscio m , ketterle w and salomon c ( eds ) 2008 _ ultra - cold fermi gases _
( ios press , amsterdam ) proceedings of the international school of physics `` enrico fermi '' , course clxiv , varenna , 20 - 30 june 2006 we assume that the normal cloud performs a rigid rotation with an angular velocity @xmath4 .
this can be justified by the internal friction in the non - superfluid component along with the fact that the rotating trap deformation is applied to all regions of the cloud simultaneously . to determine the temparature increase in the rotating trap we measure @xmath38 after variable rotation times .
we find that , in the relevant temperature range , the behavior of @xmath38 is well approximated by a linear increase with time .
this justifies the description in terms of a constant heating rate . in our setup
we can not directly observe vortices by absorption imaging .
the main reason is the technical limitation that our coil system does not allow for fast enough magnetic field ramps as required for increasing the size of vortex cores during expansion @xcite .
this explanation is supported by the fact that we measured a slower decay of angular momentum for the later experiments of fig .
[ spinup ] ( @xmath90 ) than we did for the earlier measurements of fig .
[ quench ] ( @xmath91 ) . between the two sets of measurements the optical setup of the trapping beam
was readjusted , leading to reduced imperfections in the later experiments . | we report on the observation of a quenched moment of inertia as resulting from superfluidity in a strongly interacting fermi gas .
our method is based on setting the hydrodynamic gas in slow rotation and determining its angular momentum by detecting the precession of a radial quadrupole excitation .
the measurements distinguish between the superfluid and collisional origins of hydrodynamic behavior , and show the phase transition . |
the laser geodynamic satellite experiment ( lageos-3 ) is a joint usaf , nasa , and asi proposed program to measure , for the first time , a quasi - stationary property of the earth
its gravitational magnetic dipole moment ( gravitomagnetism ) as predicted by einstein s theory of general relativity .
this gravitomagnetic field causes local inertial frames to be dragged around with the earth at a rate proportional to the angular momentum of the earth , and inversely proportional to the cube of the distance from the center of the earth .
thus the line of nodes of the orbital plane of lageos-3 precesses eastward at @xmath0 .
although in this example the frame dragging effect is small compared to the torque on the orbital plane due to the oblateness of the earth , it is an essential ingredient in the dynamics of accretion disks , binary systems , and other astrophysical phenomena @xcite . today , almost eighty years after einstein introduced his geometric theory of gravity , we have just begun to measure to verify his gravitation theory . of no less stature than the `` tide producing ''
@xmath1 `` electric component '' of gravity is the inertial - frame defining `` magnetic component '' of gravitation @xmath2 . to see this force in action : first , inject a satellite into a polar orbit about an earth - like mass idealized as not spinning with respect to the distant quasars .
the satellite will remain in orbit in a continuous acceleration towards the center - of - mass of the attracting body under the influence of the newtonian @xmath3 force , and its orbital plane will remain fixed in orientation with respect to distant quasars .
second , spin the central body , giving it angular momentum , and follow the trajectory of the satellite . its orbital plane will experience a torque along the central body s rotation axis .
the orbital plane will undergo a precessional motion in the direction of the central body s rotation .
the mass in motion of the central body , or `` mass current '' , produces a dipole gravitational field the gravitomagnetic field . in the case of a satellite orbiting at two earth radii
, the orbital plane will precess about the body axis of the earth at approximately @xmath4 .
this is the lense - thirring effect @xcite .
the lense - thirring force has never been directly measured .
a measurement of this gravitomagnetic force can be compared to the pioneering work of michael faraday on the measurement of the magnetic force between two current - carrying wires
. however , the laboratory setting for this gravity measurement will be the 4-dimensional curved spacetime ( approximately kerr ) geometry enveloping the earth .
the idea behind the lageos gravity measurement is simple . whereas the everitt - fairbanks experiment ( gravity probe - b ) proposes putting a gyroscope into polar orbit @xcite , the ciufolini lageos-3/lageos-1 experiment @xcite proposes the use of the orbital planes themselves as a gyroscope . in 1976
nasa launched the lageos-1 satellite , a totally passive @xmath5 diameter ball of aluminum with @xmath6 retro - reflecting mirrors embedded in its surface .
( there are numerous globally - located laser tracking stations to observe lageos - type satellites . )
lageos-1 was injected into a two earth - radii circular orbit at an inclination of @xmath7 . due to the oblateness of the earth
, the orbital plane rotates at a rate of @xmath8 .
this torquing can only be modeled to @xmath9 which is not accurate enough to measure the @xmath10 gravitomagnetic force .
the idea of ciufolini is to launch another lageos satellite ( lageos-3 ) into an orbit identical to that of lageos-1 , except that its inclination is supplementary ( @xmath11 ) .
this proposed orbital plane will rotate in the opposite direction , _
i.e. _ , @xmath12 .
the intersection of the two ( lageos-1 , lageos-3 ) orbital planes will sweep out a `` tandem - generated gyro plane '' ( fig .
[ figure1 ] ) .
the utilization of two satellites cancels out many of the large precessional effects due to mass eccentricities of the earth , providing a plane inertial enough for a measurement , accurate to five percent ( or better ) , of the `` magnetic component '' of gravity . =
the success of the lageos experiment hinges upon the detection of a @xmath10 eastward drift of the line of nodes of the two satellites .
a strong effort is now underway to model the orbital and spin dynamics of these satellites , and to make an assessment of the uncertainties these will add to the desired measurements .
encapsulated in table [ table1 ] are the five major classes of errors in this experiment : ( 1 ) geopotential ( other than even zonals ) and tides , ( 2 ) earth radiation pressure , ( 3 ) uncertainty in other relativistic effects , ( 4 ) earth and solar - induced thermal forces , and ( 5 ) even zonal geopotentials ( per @xmath13 inclination injection error ) . due to the recent gem - t1 improvements in the earth
s zonal harmonics , the errors due to solid - earth tides have been significantly reduced , and are now potentially smaller than those due to surface effects @xcite .
these surface effects ( _ e.g. _ , yarkowsky thermal drag , neutral and charged particle drag ) cause a change in the nodal precession of the satellite , thus contributing to what will be potentially the largest source of error in the lageos-3 measurement of the gravitomagnetic effect . in modeling these surface forces , in particular the yarkowsky thermal drag ( also referred to as the rubincam effect ) due to the satellite s differential heating and delayed reradiation , the behavior of the spin vector of the satellite is of crucial importance . that the uncertainties induced by the surface forces on lageos are on the order of 4% , out of a 6% experiment , makes such a theoretical modeling of paramount importance . with this in mind , we undertake in this paper a theoretical model of the spin dynamics of lageos - type satellites , and compare our results with observational data .
we must emphasize that we are looking for qualitative , not quantitative , results .
we do not intend to predict the exact magnitude and direction of the spin vector at any particular instant .
instead , we desire a rough estimate of its magnitude , and whether its behavior is predictable or chaotic .
we seek to answer questions of the sort : `` in the asymptotic limit , does the satellite tidally lock , tumble , or have some other behavior ? '' previous studies of the spin dynamics of the lageos-1 satellite @xcite were valid only for spin rates much larger than the orbital frequency . today , however , the spin of the satellite , decaying with a three - year time constant , is rapidly approaching the orbital period .
we therefore require an analysis in the low frequency regime .
unlike the earlier analysis which dealt with orbit - averaged quantities , we will solve numerically the full set of dynamical equations .
as we will show , while the low frequency regime exhibits complex behavior , the asymptotic state of this forced , damped system appears to be tidally locked . in this paper
we will examine the spin - orbit resonance phase and discuss the asymptotic state of the spin of the satellite .
it is the goal of this research to provide an optimum strategy for the measurement of the spin dynamics of the lageos satellites in support of the proposed gravity measurement .
in addition , we use our theoretical model to propose an optimum orbital injection procedure for lageos-3 : it is our opinion that the lageos-3 satellite should be injected into the orbital plane with as large a spin rate as possible .
our results provide the first analysis of the asymptotic spin dynamics of the lageos satellite .
previous calculations were unable to analyze the spin - orbit resonances of the satellite , nor its asymptotic behavior , which will play a crucial role in the experiment .
we demonstrate in this paper that the lageos-1 satellite will be sufficiently predictable to support the gravitomagnetic measurement . .
* the latest error budget for the lageos-3 lense - thirring experiment*. errors in geopotential and tides reflect improvement over gem - t1 .
uncertainty in the thermal forces depends mainly on knowledge of spin - axis motion .
the even zonal geopotential error calculation assumes less than @xmath14 inclination injection error .
the random and stochastic error estimate can accommodate seasonal variations in low degree spherical harmonics of the geopotential .
[ cols="<,<,>",options="header " , ]
it is rather interesting that after 36 years the spin dynamics of passive satellites is once again important to the field of astrodynamics . in 1957
vinti @xcite analyzed the spin dynamics of a non - ferromagnetic spherical satellite in the earth s magnetic field , which was then of critical importance to the alignment of antennas .
today we perform the same analysis on a slightly oblate satellite , of critical importance to the first measurement of the magnetic component of gravity , as predicted by einstein .
there are many factors to consider when analyzing the spin dynamics of an oblate , metallic satellite orbiting in the gravitational and magnetic fields of the earth .
the most prominent effect is the torqueing due to the gravitational field of the earth .
this arises from the oblateness of the satellite , with the oblateness of the earth producing a negligible contribution that can be added to our calculations as an adiabatic correction . if the satellite s ( bulging ) equatorial plane lies in the plane of orbit , no such torques are possible .
however , when the satellite is not placed exactly in such a position , gravitational torques will arise . in an effort to model these torques ,
we consider the situation of an oblate spheroid in orbit around a point mass . as was done by bertotti and iess , we parallel the development in goldstein @xcite .
bertotti and iess s analysis of the effects of gravity on the spin dynamics of the oblate satellite , which resulted in predicting a _ chaotic _ spin dynamics of arbitrarily large amplitude in the obliquity at late times , is not appropriate for small rates of spin .
their prediction is based on the `` hipparcos '' formula for the rate of precession @xmath15 of an oblate satellite in the inhomogeneous gravity field of the earth @xmath16 where @xmath17 is the orbital angular velocity , @xmath18 is the satellite spin rate , @xmath19 is the obliquity angle of the satellite ( the angle between @xmath20 and the normal to the orbital plane ) and @xmath21 is the satellite s oblateness . here
@xmath22 and @xmath23 are the principal moments of inertia ( the principal direction corresponding to @xmath22 is assumed to be that of @xmath20 and coincides , by assumption , with the symmetry axis of the oblate satellite ) .
it is argued by bertotti and iess that , since @xmath24 , the gravitational precession in the asymptotic limit of small @xmath25 becomes very fast and may make the spin dynamics chaotic .
this conclusion is based upon a misunderstanding .
we have shown , via a careful analysis of assumptions underlying the `` hipparcos '' formula , that even in the approximation commonly used in deriving the formula ( averaging of the gravitational potential over the satellite orbit , dipole cutoff of the multipole decomposition , _ etc .
_ ) eqn .
( [ 1 ] ) can only be used when @xmath26 _ i.e. _ , when the spin rate of the satellite is much greater than its orbital angular velocity .
the latter restriction is easy to overcome , and the corrected equation for @xmath15 is @xmath27 this equation imposes a bound on @xmath28 : @xmath29 which makes it clear that the rate of precession can not grow to cause chaoticity of the satellite spin dynamics .
a subsequent qualitative investigation @xcite by chris fuchs has shown that when magnetic forces are included in the picture the precession rate remains bounded , and should be much smaller than @xmath17 .
another conclusion reached in the course of our analysis has been that , when both gravitational and magnetic forces are taken into account , the nutation , although bounded in its amplitude , does not disappear completely even in the asymptotic limit .
the value of these results , however , is limited by the fact that they do not yield the exact bounds , nor do they provide any information on the time scale to reach the asymptotic limit .
however , they do clearly show that chaoticity of the lageos spin dynamics caused by an unbounded growth of gravitationally - induced precession can not occur .
another important factor governing the evolution of the spin vector is the interaction of the metallic core of the satellite with the magnetic field of the earth .
the lageos satellite @xcite is composed of two aluminum hemispheres bolted together , with a brass cylindrical core along its body axis ( the original axis of spin ) ( fig .
[ figure2 ] ) .
the spinning of this metallic object in the magnetic dipole field of the earth ( and the motion through that field ) will cause eddy currents within the satellite , which will in turn cause dissipation through joule heating and a slowdown of the spin , and furthermore will cause torques on the spin vector .
these torques can be understood as the interaction of the magnetic dipole , caused by the induced eddy currents , with the magnetic field of the earth . in modeling this effect ,
we have treated a uniform , spherical object in the orbit of a perfect magnetic dipole .
here we are concerned with a simple qualitative analysis of the spin dynamics of lageos , as the complexity of the true satellite geometry prohibits us from a precise model of the eddy current distribution .
our theoretical model permits us to analyze arbitrary inclinations of the orbital plane to the earth s magnetic dipole axis .
the results presented in this paper correspond to a retrograde @xmath30 orbit . = the problem of a spinning metallic sphere in a constant magnetic field has been treated by landau and lifshitz @xcite , and we avail ourselves of their results . for our purposes , we ignore the torques caused by the changing magnetic field due to the orbit ( as opposed to spin ) of the satellite . these torques can be shown to be negligible until asymptotically late stages of motion , and have no qualitative effects upon the dynamics . the source of many of the difficulties in doing analyses of such orbiting , spinning bodies lies in the involved coordinate systems needed to describe their motion .
thus , it is important at this point to give a brief description of the coordinates we will use in this paper . in our analysis of the spin dynamics of the lageos satellite we found it convenient to introduce the following four coordinate systems : * 1 . * _ the orbit - centered inertial frame ( oci ) @xmath31 .
_ here the @xmath32-axis is oriented along the normal to the orbital plane of the satellite ( @xmath33 coinclination ) .
the @xmath34-axis is defined to be the intersection of the orbital plane and the earth s equatorial plane , and the origin is the center of mass of the earth .
we have assumed here that this frame is inertial , and have not included the secular drag of the line - of - nodes of the orbital plane due to the oblateness of the earth .
this @xmath35 precession can be included at the end of our analysis as an adiabatic correction , and again has no qualitative effects upon the dynamics .
_ the earth - centered inertial ( eci ) reference frame @xmath36_. here the @xmath37-axis is aligned with the body axis of the earth .
the @xmath38-axis lies in the earth
s equatorial plane at zero degrees longitude , and the origin coincides with the center of the earth .
* 3 . * _ the body frame ( non - inertial ) @xmath39 . _
the origin is at the center of the satellite , and the axes correspond to a set of principal axes .
the satellite is assumed to be a slightly oblate ( @xmath40 ) spheroid of brass , and the @xmath41-axis is aligned along its body axis .
the @xmath42 and @xmath43 axes are an arbitrary fixed set of orthogonal axes spanning the equatorial plane of the satellite . in our calculations ,
the body axis is related to the orbit - centered frame through the three euler angles @xmath19 , @xmath44 , and @xmath45 .
the nutation angle @xmath19 is the angle between @xmath41 and @xmath32 , while the angle of precession @xmath44 is the angle between the @xmath34-axis and the line of nodes ( the intersection of the orbital plane and the equatorial plane of the satellite ( @xmath42-@xmath43 plane ) ) .
the spin angle @xmath45 is the angle between the line of nodes and the @xmath42-axis .
* 4 . * _ the landau - lifshitz ( non - inertial ) coordinate system @xmath46 . _
the @xmath47-axis is aligned along the instantaneous angular momentum vector of the satellite ( @xmath48 ) .
the @xmath49-axis is picked so that the instantaneous magnetic field ( @xmath50 ) at the satellite lies in the @xmath51 plane .
note that @xmath47 need not be aligned with the body axis of the satellite , and in fact , during the asymptotic behavior of the satellite they are vastly different .
in particular , the angle ( @xmath52 ) between the symmetry axis of the satellite ( @xmath53 ) and the instantaneous angular momentum vector ( @xmath54 ) can be expressed in terms of the three euler angles as @xmath55 in the early stages of the lageos mission when @xmath56 and @xmath57 this angle is rather small ( @xmath58 ) .
a proper interpretation of the results of the numerical simulation requires a careful distinction between @xmath15 and @xmath59 ( where @xmath44 is the euler angle of the body frame ) .
the two coincide only under the assumption that the @xmath18 component of the satellite s angular velocity is responsible for all of the satellite s energy , or , to put it differently , the angular momentum of the satellite is directed along the body axis of the satellite . as the satellite s spinning motion slows down , this last assumption is violated .
@xmath15 represents only a part of @xmath59 , the other part being caused by the `` tidal locking '' effect . as we shall see ,
this is exactly what happens in the asymptotic phase of the satellite s motion .
nevertheless , both @xmath15 and @xmath59 remain bounded .
the spin dynamics are determined by euler s equations @xmath60 where @xmath61 , @xmath62 , @xmath18 are the components of the satellite s angular velocity in the body frame , @xmath23 , @xmath22 are the principal moments of inertia , and @xmath63 , @xmath64 , @xmath65 are the components of the torques along the satellite s principal axes . after substituting the expressions for @xmath61 , @xmath62 , @xmath18 in terms of the euler angles
@xmath66 the euler equations become @xmath67 the torque components @xmath63 , @xmath64 , @xmath65 are due to gravitational and magnetic forces acting on the satellite @xmath68 gravitational torques in the body frame are given by @xmath69 using the standard dipole approximation , the gravitational potential @xmath70 is @xmath71 where @xmath72 is the direction cosine between ( 1 ) the radial vector from the satellite center of mass to the center of the earth , and ( 2 ) the symmetry axis of the satellite .
it is related to euler s angles via @xmath73 where @xmath74 gives the angular position of the center of mass of the satellite in its orbit about the earth , and @xmath75 is an arbitrary starting position .
the potential is given by @xmath76 and thus @xmath77 equations ( 10 ) and ( 14 ) lead to the following final expressions for the components of the gravitational torque in the body frame @xmath78 as for the magnetic effects , we are interested in the torque components acting on a conducting ball of radius @xmath79 spinning with angular velocity @xmath20 in an external magnetic field @xmath50 .
we take our expressions from landau and lifshitz @xcite , noting that their results are for what we have dubbed the landau - lifshitz frame ( as described above ) , and for our purposes need to be transformed to the body frame .
as already described , the landau - lifshitz frame is determined by the vectors @xmath20 and @xmath50 .
if we introduce an arbitrary set of rectangular coordinates @xmath80 , @xmath81 , @xmath82 , in this frame @xmath20 and @xmath50 are represented as @xmath83 where hats ( @xmath84 ) denote vectors normalized to unity .
the transition between this arbitrary frame and the landau - lifshitz frame is given by @xmath85_x\over \omega\vert\vec\omega\times\vec b \vert & [ ( \vec\omega \times\vec b ) \times \vec\omega ] _
y\over \omega \vert \vec\omega \times \vec b \vert & [ ( \vec\omega \times\vec b ) \times \vec\omega ] _
z\over \omega \vert \vec\omega \times \vec b \vert \\ ( \vec\omega \times \vec b)_x\over \vert\vec\omega \times \vec b\vert & ( \vec\omega \times \vec b)_y\over \vert\vec\omega \times \vec b\vert & ( \vec\omega \times \vec b)_z\over \vert\vec\omega \times \vec b\vert \\ \omega_x\over\omega & \omega_y\over\omega & \omega_z\over\omega \end{array}\right)\ , \left(\begin{array}{c}\hat x \\ \hat y \\
\hat z \end{array}\right ) , \label{17}\ ] ] where @xmath86 .
the components of the magnetic torque in the landau - lifshitz frame are given by @xmath87 where @xmath88 is the volume of the ball , @xmath89 , @xmath90 , @xmath91 are components of the magnetic field in the landau - lifshitz frame , and the real and complex parts of the coefficient of magnetization are @xmath92 , \nonumber
\\ \alpha '' & = & -{9 \delta^2\over 16 \pi a^2 } \left [ 1 - { a\over \delta } { { \rm sinh } \left ( 2 { a\over\delta}\right ) + \sin \left ( 2 { a\over\delta}\right ) \over { \rm cosh } \left ( 2 { a\over\delta}\right ) - \cos \left ( 2 { a\over\delta}\right ) } \right ] , \nonumber \label{19}\end{aligned}\ ] ] with @xmath93 here @xmath94 is the speed of light and @xmath95 is the specific conductivity of the material forming the ball . at small values of @xmath25 ( @xmath96 ) ,
@xmath97 and @xmath98 can be approximated by the expressions @xmath99 components of the magnetic field can be evaluated in the eci frame using the dipole approximation . in the spherical coordinate representation of the eci frame ,
the magnetic field components are @xmath100 where @xmath101 is the magnetic dipole moment of the earth . in the rectangular coordinate representation of the eci frame
( the index 1 is used everywhere for quantities in eci ) @xmath102 transforming these to the oci frame : @xmath103 where @xmath52 is the colatitude angle of the orbit with respect to magnetic north , and the index 2 is used for quantities in the oci .
the satellite s semimajor axis is @xmath104 , and for the earth s magnetic dipole moment we used @xmath105 .
although we do not average the magnetic field in our simulations , it was useful for back of the envelope calculations to note that the the magnetic field , averaged over one orbit , is @xmath106 , with components , @xmath107 the angles @xmath108 and @xmath109 are dependent upon the satellite s position in its orbit .
the satellite s coordinates in the oci frame are @xmath110 where @xmath111 in eci
we have @xmath112 hence , @xmath113 the three euler equations ( eq . [ 8 ] ) , with the magnetic and gravitational torques included ( properly transformed to the body frame ) , give us the vehicle to analyze qualitatively the spin dynamics of the lageos-1 satellite .
we present our results from the numerical integration of these equations in the following two sections .
we solved the euler equations ( eq . [ 8 ] ) using a fourth - order bulirsch - stoer algorithm with adaptive time - stepping @xcite .
the equations were integrated for @xmath114 seconds , as we wanted to ( 1 ) reproduce the experimentally - measured spin rates ( @xmath115 following launch ) , ( 2 ) examine the spin - orbit resonance ( @xmath116 after launch ) , and ( 3 ) reveal the asymptotics of the spin dynamics ( @xmath117 after launch ) . the experimentally - measured exponential decrease in the spin rate imposed a constraint on our theoretical model , linking the `` effective '' radius of the satellite ( @xmath79 ) with the satellite s `` effective '' conductivity ( @xmath95 ) : @xmath118 the satellite was modeled as a @xmath119 radius spheroid of brass ( @xmath120 ) .
lageos i s moment of inertia about the body axis is @xmath121 , while the moment of inertia perpendicular to the body axis is @xmath122 , corresponding to an approximately @xmath123 deviation from sphericity . from the experimental data , we observed a deviation from pure exponential damping of the spin of the satellite at early times .
this is presumably due to the satellite s transition from magnetic opaqueness to transparency in the course of its spin damping . a rapidly rotating conductor , with angular velocity @xmath25 and conductivity @xmath95 ,
will have an associated magnetic skin depth @xmath124 this skin depth starts out considerably smaller than the satellite , but as the satellite s spin is damped down , the skin depth becomes much larger than the satellite s diameter .
although this transition effect will be more pronounced for our idealized spherical brass model satellite than for lageos ( with its additional surface structure ) , it helps account for the structure in the experimental spin - down data ( for which it would have been convenient to have access to the associated error bars ) .
it is for this reason that we began our calculations within the transparent phase , at @xmath125 .
we integrated forward in time to @xmath126 , and backwards to @xmath127 . at @xmath128 , the skin depth of our satellite model , given the experimentally measured spin rate of @xmath129 and a conductivity for brass ( @xmath130 ) ,
is @xmath131 .
we started the integration with the satellite in a retrograde circular orbit , at a radius of @xmath132 and an inclination of @xmath133 . at @xmath128 ,
the satellite was located over the equator and positioned along the @xmath134 axis ( @xmath135 ) .
we inserted the body axis of the satellite into the orbital plane with the angular momentum vector parallel to the oci @xmath136-axis .
the initial spin of the satellite was determined by the experimental measurements , and was set to @xmath137 .
the initial conditions were @xmath138 in running this initial data backward in time from @xmath139 to launch at @xmath140 , we recovered the following angles and angular velocities ( measured in @xmath141 ) : @xmath142 the magnetic field in our simulation was assumed to be a perfect dipole field , of moment @xmath143 and aligned along the @xmath37-axis of the eci frame .
we have identified three distinct phases in the spin dynamics of the lageos satellite , which we shall refer to as ( 1 ) the fast spin phase , ( 2 ) the spin - orbit resonance phase , and ( 3 ) the asymptotic phase .
\a ) _ fast spin phase_. this first phase , from @xmath140 to @xmath144 , is characterized by an exponential decrease in the spin rate ( @xmath145 ) of the satellite ( fig .
[ figure3 ] ) from @xmath146 to @xmath147 .
the body axis of the satellite is aligned with the angular momentum of the satellite in this phase ; hence , all other quantities ( angular momentum , kinetic energy , _ etc . _ ) also decrease exponentially .
the nutation of the satellite ( the angle between the body axis of the satellite and the normal to the orbital plane ) undergoes a steady increase from @xmath148 at @xmath149 to @xmath150 over a period of @xmath151 , indicating a nutation angular velocity of @xmath152 .
the nutation then settles down to a `` pseudo - stable '' state at @xmath153 over the next @xmath154 years and remains at this value ( @xmath155 ) until the onset of the spin - orbit resonance at @xmath156 ( fig .
[ figure6 ] ) . finally , within this twenty five year period
the satellite precesses in a positive sense by 63 revolutions before unwinding when entering the spin - orbit resonance phase .
\b ) _ spin - orbit resonance phase_. the spin dynamics abruptly change at @xmath157 .
this is precisely when the satellite s spin , decreased by magnetic damping , approaches the orbital angular velocity @xmath158 the conductivity of the satelite was chosen to reproduce experimentally observed exponential decay in the spin rate of the satellite , @xmath159 with ( @xmath160 time constant ) .
therefore , the spin - orbit resonance should occur at @xmath161 , which is in agreement with the numerical results .
the resonance phase is marked by a movement of the angular momentum vector to a position orthogonal to the orbital plane , and is furthermore characterized by the beginning of satellite wobble ( _ i.e. _ , the point in time when @xmath52 of eq .
( [ * ] ) becomes nonzero and the body axis becomes misaligned with the instantaneous angular momentum vector ) . from this point
forward it is more illustrative to examine the dynamics of the instantaneous angular velocity and momentum rather than the euler angles .
in addition to the dramatic changes in the spin dynamics , this second phase also gives rise to a reversal in the signs of the precessional velocity ( @xmath162 ) and spin ( @xmath163 ) .
\c ) _ the asymptotic phase_. following the spin - orbit resonance phase , the spin dynamics gradually settled down to an asymptotic regime over the course of @xmath164 . not surprisingly , the satellite becomes tidally locked ( think of the moon ) .
in particular , the asymptotic value of the total angular velocity is equal to the orbital angular velocity , subject to small fluctuations .
we note that in this phase the torques induced from the changing magnetic field due to _
orbital _ motion of the satellite will become important , and should no longer be ignored .
however , their addition will not significantly change dynamics , as the energies at this point are quite low . from our numerical runs , a rough estimate of the asymptotic behavior of the satelite model ( modulo phase and a finite offset in @xmath44 )
is given by @xmath165 the asymptotic values of the other relevant parameters ( fig .
[ figure4 ] ) : @xmath166 = = =
of the five largest sources of error identified in the lageos-3 experiment , the earth and solar - induced surface forces are potentially the most troublesome ( table [ table1 ] ) .
the anisotropic heating of the satellite , and subsequent reradiation , gives a `` thermal rocketing '' perturbation ( referred to as the rubincam effect or the yarkowsky thermal drag ) which tends to degrade the experiment . to model this effect requires , in part , a detailed knowledge of the behavior of the angular momentum of the satellite . toward this end
we have derived , and solved numerically , a simplified set of euler equations that evolve the angular momentum vector for a slightly oblate spheroid of brass orbiting an earth - like mass , idealized as being a perfect sphere and having a perfect polar - oriented dipole magnetic field .
the euler equations included both the tidal gravitational torques and the eddy - current torques , as well as the resistive damping torques , as modeled by complex magnetization coefficients . using this rather simplified model ,
we have identified three phases of the rotational dynamics
a fast spin phase , a spin - orbit resonance phase , and an asymptotic phase ( fig .
[ figure3 ] ) .
we have also identified an error in the previously established model of asymptotic spin dynamics @xcite .
this error has led to confusion and , in attempts to reconcile observed data with theoretical predictions , has led others to hypothesize erroneous models for the moments of inertia of lageos-1 @xcite .
our results have led us to formulate four as yet unresolved questions : ( 1 ) can we obtain the asymptotic solution analytically , and in so doing can we understand the wobbling or slippage of the euler angles with respect to the relatively stable total instantaneous angular velocity ? ; ( 2 ) can we understand why the _ rms _ fluctuations in the gravitational potential energy cause the asymptotic value of the angular momentum vector to be offset from the orbital plane by @xmath167 ? ; ( 3 ) can we understand why the nutation angle ( @xmath19 ) drifts initially at a rate of @xmath168 and reaches a pseudo - stable value of @xmath169 ? ; and ( 4 ) can we understand the fluctuations in the spin rate ( @xmath145 ) over the first @xmath170 , which do not appear to have been detected experimentally ? we are addressing these questions by ( 1 ) introducing a more realistic model of the satellite and earth into our calculations @xcite , and ( 2 ) exploring more of phase space by way of poincare sections . the results presented here provide us with clues that must be pieced together to reveal the physics behind the complex motion we observe .
the current spin dynamics model suggests that we launch lageos-3 , with as large a spin @xmath171 as possible , into an obliquity of @xmath172 ; although , due to the qualitative nature of our results , the precise numbers are far from being conclusive .
we are currently working with the center for space research at the university of texas at austin to determine the impact this revised model of the spin dynamics of lageos will have on the lageos-3 mission ( in particular , how will the rubincam effect alter the line of nodes of the orbital plane ? ) .
in addition , we are working closely with colleagues at the university of texas and the university of maryland to reconcile the experimental measurements of the spin dynamics of lageos-1 with our theoretical model @xcite .
furthermore , we will propose an optimal experimental measurement schedule in support of the proposed lageos-3 mission .
we wish to thank stirling colgate , douglas currie , christopher fuchs , and sara matzner for many helpful discussions .
this work was supported in part by a grant from los alamos national laboratory under ldrd xl31 , by the afosr under the summer faculty research program , and by nsf grants phy 88 - 06567 and phy 93 - 10083 .
99 j. m. bardeen and j. a. peterson , _ app .
j. lett . _
* 195 * , l65 ( 1975 ) ; d. a. macdonald , k. s. thorne , r. h. price and xiao - he zhang , _ astrophysical applications of black hole electrodynamics _ in _ black holes : the membrane paradigm _ , edited by k. s. thorne , r. h. price , and d. a. macdonald ( yale university press , new haven , 1986 ) , ch .
h. thirring and j. lense , _ phys .
z. _ * 19 * , 156 ( 1918 ) ; h. thirring , _ ibid _ * 19 * , 33 ( 1918 ) , _ ibid _ * 22 * , 29 ( 1921 ) .
an english translation is given by b. mashhoon , f. w. hehl , and d. s. theiss , _ gen .
. gravit . _ * 16 * , 711 ( 1984 ) . c. w. f. everitt , in _ experimental gravitation _ , edited by b. bertotti ( academic , new york , 1973 ) ; r. a. vanpatten and c. w. f. everitt , _ phys .
* 36 * ( 1976 ) .
i. ciufolini _ phys .
* 56 * , 278 ( 1986 ) .
j. ries , private communication ( 1993 ) .
d. p. rubincam , _ j. geophys research _ * 92 * , 1278 - 1294 ( 1987 ) ; _ ibid _ * 93 * , 13805 ( 1988 ) ; _ ibid _ * 95 * , 4881 ( 1990 ) ; _ the lageos along track acceleration : a review _ , paper presented at the first william fairbanks meeting on relativistic gravity experiments in space , rome , italy , september 10 - 14 , 1990
. b. bertotti b. and l. iess , _ j. geophys . research _ * 96 * , 2431 ( 1991 ) .
j. p. vinti , _ theory of the spin of a conducting satellite in the magnetic field of the earth _ , defense technical information center , brl-1020 ( 1957 ) .
h. goldstein , _ classical mechanics _ ( addison - wesley , reading , ma , 1981 ) . c. fuchs , _ lagrangian formulation of lageos s spin dynamics _ , final report , 1992 air force summer research program , august 1992 .
c. w. johnson , c. a. lundquist , and j. l. zurasky , _ the lageos satellite _ , paper presented at the international astronautical federation xxvii congress , anaheim , ca , october 10 - 16 , 1976 ) .
l. d. landau and e. m. lifshitz , _ electrodynamics of continuous media _ ( pergamon press , oxford , 1984 ) . w. h. press , s. a. teukolsky , w. t. vetterling , and b. p. flannery , _ numerical recipes in c _
( cambridge university press , cambridge , 1992 ) .
r. scharroo , k. f. wakker , b. a. c. ambrosius , and r. noomen , _ j. geophys
. research _ * 96 * , 729 ( 1991 ) .
r. p. halverson and h. cohen , _ ieee trans .
_ ane-11 , 118 ( 1964 ) .
d. currie , s. habib , r. matzner , and w. a. miller , ( in preparation ) . | lageos is an accurately - tracked , dense spherical satellite covered with 426 retroreflectors .
the tracking accuracy is such as to yield a medium term ( years to decades ) inertial reference frame determined via relatively inexpensive observations .
this frame is used as an adjunct to the more difficult and data intensive vlbi absolute frame measurements .
there is a substantial secular precession of the satellite s line of nodes consistent with the classical , newtonian precession due to the non - sphericity of the earth .
ciufolini has suggested the launch of an identical satellite ( lageos-3 ) into an orbit supplementary to that of lageos-1 : lageos-3 would then experience an equal and opposite classical precession to that of lageos-1 . besides providing a more accurate real - time measurement of the earth s length of day and polar wobble
, this paired - satellite experiment would provide the first direct measurement of the general relativistic frame - dragging effect .
of the five dominant error sources in this experiment , the largest one involves surface forces on the satellite , and their consequent impact on the orbital nodal precession .
the surface forces are a function of the spin dynamics of the satellite .
consequently , we undertake here a theoretical effort to model the spin ndynamics of lageos . in this paper
we present our preliminary results . |
there are certain scientific problems that provide deep connections between many different scientific fields .
the study of the low - energy states of classical interacting many - particle systems is an exemplar of a class of such problems because of its manifest importance in physics , materials science , communication theory , cryptography , mathematics and computer science .
such many - particle systems have been used with great success to model liquids , glasses and crystals when quantum effects are negligible @xcite .
the total potential energy @xmath3 of @xmath4 identical particles with positions @xmath5 in some large volume in @xmath0-dimensional euclidean space @xmath1 can be resolved into separate one - body , two - body , @xmath6 , @xmath4-body contributions : @xmath7 where @xmath8 represents the intrinsic @xmath9-body interaction in excess to the interaction energy for @xmath10 particles . to make the statistical - mechanical problem more tractable , the exact many - body potential ( [ full ] )
is usually replaced by a mathematically simpler form .
for example , in the absence of an external field ( i.e. , @xmath11 ) , often one assumes pairwise additivity , i.e. , @xmath12 pairwise additivity is exact for hard - sphere systems and frequently has served to approximate accurately the interactions in simple liquids , such as the well - known lennard - jones pair potential @xcite , and in more complex systems where the pair potential @xmath13 in ( [ pair ] ) can be regarded to be an _ effective _ pair interaction @xcite .
an outstanding problem in classical statistical mechanics is the determination of the _ ground states _ of @xmath3 , which are those configurations that globally minimize @xmath3 and hence are the states that exist at absolute zero temperature .
while classical ground states are readily produced by slowly freezing liquids in experiments and computer simulations , our theoretical understanding of them is far from complete @xcite .
virtually all theoretical / computational ground - state studies of many - particle systems have been conducted for pairwise additive potentials @xcite .
often the ground states of short - range pairwise interactions are crystal structures in low dimensions @xcite , but long - range interactions exist that can suppress any kind of symmetry leading to disordered ground states in low dimensions @xcite .
moreover , in sufficiently high dimensions , it has been suggested that even short - ranged pairwise interactions possess disordered ground states @xcite .
ground states of purely repulsive pair interactions have profound connections not only to low - temperature states of matter but to problems in pure mathematics , including discrete geometry and number theory @xcite , information theory , and computer science .
as will be explained further below , it is known that the sphere packing problem and the number variance problem ( closely related to an optimization problem in number theory ) can be posed as energy minimizations associated with an infinite number of point particles in @xmath0-dimensional euclidean space @xmath1 interacting via certain repulsive pair potentials .
both of these problems can be interpreted to be optimization problems involving _ point processes _ , which can then be recast as energy minimizations .
a point process in @xmath1 is a distribution of an an infinite number of points in @xmath1 at number density @xmath14 ( number of points per unit volume ) with configuration @xmath15 ; see ref .
@xcite for a precise mathematical definition .
a packing of congruent nonoverlapping spheres is a special point process in which there is a minimal pair separation distance , equal to the sphere diameter .
the sphere packing problem seeks to determine the densest arrangement(s ) of congruent , nonoverlapping @xmath0-dimensional spheres in euclidean space @xmath1 @xcite .
although it is simple to state , it is a notoriously difficult problem to solve rigorously .
indeed , kepler s four - century - old conjecture , which states that the face - centered - cubic lattice in @xmath16 is maximally dense , was only recently proved @xcite . for @xmath2 ,
the packing problem remains unsolved @xcite .
it is well known that the sphere packing problem can be posed as an energy minimization problem involving pairwise interactions between points in @xmath1 ( e.g. , inverse power - law functions in which the exponent tends to infinity ) ; see ref .
@xcite and references therein .
problems concerning the properties and quantification of density fluctuations in many - particle systems continue to provide many theoretical challenges .
of particular interest are density fluctuations that occur on some local length scale @xcite .
it has been shown that the minimal number variance associated with points ( e.g. , centroids of atomic or molecular systems ) contained within some window " can also be formulated as a ground - state problem involving bounded repulsive pair interactions with compact support @xcite . for spherical windows in the large - radius limit , the best known solutions in @xmath1 are usually point configurations that are duals " ( in the sense discussed later in the paper ) to the best known sphere packings in @xmath1 .
the focus of this paper is on two other optimization problems involving point processes in @xmath1 : the _ covering _ and _ quantizer _ problems . roughly speaking
, the covering problem asks for the point configuration that minimizes the radius of overlapping spheres circumscribed around each of the points required to cover @xmath1 .
the covering problem has applications in wireless communication network layouts @xcite , the search of high - dimensional data parameter spaces ( e.g. , search templates for gravitational waves ) @xcite , and stereotactic radiation therapy @xcite .
the quantizer problem is concerned with finding the point configuration in @xmath1 that minimizes a distance error " associated with a randomly placed point and the nearest point of the point process .
it has applications in computer science ( e.g. , data compression ) @xcite , digital communications @xcite , coding and cryptography @xcite , and optimal meshing of space for numerical applications ( e.g. , quadrature and discretizing partial differential equations ) @xcite .
heretofore , the covering and quantizers problems were not known to correspond to any ground - state problems .
we reformulate the covering and quantizer problems as the determination of the ground states of interacting particles in @xmath1 that generally involve single - body , two - body , three - body , and higher - body interactions .
this is done by linking the covering and quantizer problems to certain optimization problems involving the void " nearest - neighbor functions that arise in the theory of random media and statistical mechanics @xcite .
these reformulations , which again exemplifies the deep interplay between geometry and physics , enable one to employ theoretical and numerical optimization techniques to solve these energy minimization problems .
we find that disordered _ saturated _ sphere packings ( roughly , packings in which no space exists to add an additional sphere ) provide relatively thin ( i.e. , economical ) coverings and may yield thinner coverings than the best known lattice coverings in sufficiently large dimensions . in the case of the quantizer problem , we derive improved upper bounds on the quantizer error that utilize sphere - packing solutions .
these improved bounds are generally substantially sharper than an existing upper bound in low to moderately large dimensions .
we also demonstrate that disordered saturated sphere packings yield relatively good quantizers .
our reformulation helps to explain why the known solutions of quantizer and covering problems are identical in the first three space dimensions and why they can be different for @xmath2 . in the first three space dimensions , the best known solutions of the sphere packing and number variance problems are directly related to those of the covering and quantizer problems , but such relationships may or may not exist for @xmath2 , depending on the peculiarities of the dimensions involved .
we begin by summarizing basic definitions and concepts in sec .
[ def ] . because of the connections between the sphere - packing , number - variance , covering and quantizer problems , in sec .
[ problems ] , we formally define each of these problems , summarize key developments , and compare the best known solutions for each of them in selected dimensions .
this includes calculations obtained for the best known number - variance solutions for @xmath17 and 24 .
we then define in sec .
[ near ] the void nearest - neighbor functions and represent them in terms of series involving certain integrals over the @xmath9-particle correlation functions , which statistically characterize an ensemble of interacting points . the special case of a single realization of the point distribution follows from this ensemble formulation , which reveals that quantizer and covering problems can be expressed as ground - state solutions of many - body interactions of the general form ( [ full ] ) .
section [ reform ] specifically gives these ground - state reformulations and shows how some known solutions in low dimensions can be explicitly recovered using the void nearest - neighbor functions . in sec .
[ results - cover ] , we show that _ disordered _ saturated sphere packings provide relatively thin coverings and may yield thinner coverings than the best known lattice coverings in sufficiently large dimensions . in sec .
[ results - quant ] , we derive improved upper bounds on the quantizer error that utilize sphere - packing solutions .
we also show that _ disordered _ saturated sphere packings yield relatively good quantizers . finally , in sec .
[ conc ] , we make concluding remarks and comment on the application of the quantizer problem to the search for gravitational waves .
for a statistically homogeneous point process in @xmath1 at number density @xmath14 ( number of points per unit volume ) , the quantity @xmath18 is proportional to the probability density for simultaneously finding @xmath9 sphere centers at locations @xmath19 in @xmath1 @xcite . with this convention , each
_ @xmath9-particle correlation function _
@xmath20 approaches unity when all of the points become widely separated from one another .
statistical homogeneity implies that @xmath20 is translationally invariant and therefore only depends on the relative displacements of the positions with respect to some arbitrarily chosen origin of the system , _
i.e. _ , @xmath21 where @xmath22 .
as we will see , statistically homogeneous point processes include as special cases periodic point distributions . the _ pair correlation _
function @xmath23 is a particularly important quantity .
if the point process is also rotationally invariant ( statistically isotropic ) , then @xmath24 depends on the radial distance @xmath25 only , _
i.e. _ , @xmath26 .
thus , it follows that the expected number of points @xmath27 found in a sphere of radius @xmath28 from a randomly chosen point of the point process , called the _ cumulative coordination function _ ,
is given by @xmath29 where @xmath30 is the surface area of a @xmath0-dimensional sphere of radius @xmath31 . a _ lattice _
@xmath32 in @xmath1 is a subgroup consisting of the integer linear combinations of vectors that constitute a basis for @xmath1 and thus represents a special subset of point processes . in a lattice @xmath32
, the space @xmath1 can be geometrically divided into identical regions @xmath33 called _ fundamental cells _ ,
each of which contains the just one point specified by the _ lattice vector _ @xmath34 where @xmath35 are the basis vectors for the fundamental cell and @xmath36 spans all the integers for @xmath37 .
we denote by @xmath38 the volume of the fundamental cell . in the physical sciences ,
a lattice is equivalent to a bravais lattice .
unless otherwise stated , we will use the term lattice .
every lattice has a dual ( or reciprocal ) lattice @xmath39 in which the sites of the lattice are specified by the dual ( reciprocal ) lattice vector @xmath40 , where @xmath41 .
the dual fundamental cell @xmath42 has volume @xmath43 .
this implies that the number density @xmath14 of @xmath32 is related to the number density @xmath44 of the dual lattice @xmath39 via the expression @xmath45 .
a _ periodic _
point process is a more general notion than a lattice because it is is obtained by placing a fixed configuration of @xmath4 points ( where @xmath46 ) within one fundamental cell of a lattice @xmath32 , which is then periodically replicated .
thus , the point process is still periodic under translations by @xmath32 , but the @xmath4 points can occur anywhere in the chosen fundamental cell .
common @xmath0-dimensional lattices include the _ hypercubic _ @xmath47 , _ checkerboard _ @xmath48 and _ root _
@xmath49 lattices , defined , respectively , by @xmath50 @xmath51 @xmath52 where @xmath53 is the set of integers ( @xmath54 ) and @xmath55 denote the components of a lattice vector of either @xmath47 or @xmath48 and @xmath56 denote a lattice vector of @xmath49 .
the @xmath0-dimensional lattices @xmath57 , @xmath58 and @xmath59 are the corresponding dual lattices ; see ref .
@xcite for definitions .
the dual lattice @xmath57 is also a hypercubic lattice ( even if the lattice spacing is @xmath60 times the lattice spacing of @xmath47 ) and hence we say that the hypercubic lattice @xmath47 is equivalent ( similar ) to its dual lattice @xmath61 , i.e. , @xmath62 . following conway and sloane @xcite , we say that two lattice are _ equivalent _ or _ similar _ if one becomes identical to the other by possibly a rotation , reflection and change of scale , for which we use the symbol @xmath63 .
in fact , the hypercubic lattice @xmath47 is characterized by the stronger property of _ self - duality_. a _ self - dual _ lattice is one with an _ identical _ dual lattice at density @xmath64 , i.e. , without any rotation , reflection , or change of scale @xcite .
the @xmath49 and @xmath48 lattices can be regarded to be @xmath0-dimensional generalizations of the face - centered - cubic ( fcc ) lattice because this three - dimensional lattice is defined by @xmath65 . in one dimension ,
@xmath66 ( equality meaning self - duality ) are identical to the integer lattice @xmath67 . in two dimensions
, @xmath68 defines the triangular lattice . in three dimensions
, @xmath69 defines the body - centered - cubic ( bcc ) lattice .
the @xmath0-dimensional laminated lattice @xmath70 @xcite is of special interest . in dimensions 8 and 24 , @xmath71 , where @xmath72 is the _ self - dual _ root lattice , and
@xmath73 is the self - dual leech lattice , are remarkably symmetric and believed to be the densest sphere packings in those dimensions @xcite .
thus , @xmath74 and @xmath75 .
the laminated lattice @xmath76 , called the barnes - wall lattice , and the coxeter - todd lattice @xmath77 are thought to be the densest lattice packings in sixteen and twelve dimensions , respectively .
note that for a single periodic point configuration at number density @xmath14 , the radial pair correlation function can be written as @xmath78 where @xmath79 is the coordination number at radial distance @xmath80 ( number of points that are exactly at a distance @xmath81 from a point of the point process ) such that @xmath82 and @xmath83 is a radial dirac delta function . for cases in which each point is equivalent to any other , which includes all lattices and some periodic point processes , the coordination numbers @xmath79 are integers .
for point processes for which the point are generally inequivalent , @xmath79 should be interpreted as the _ expected _ coordination number and hence will generally be a non - integer .
substitution of ( [ period ] ) into ( [ cum ] ) gives the coordination function for such a periodic configuration as @xmath84 where @xmath85 is the smallest integer for which @xmath86 .
consider any discrete set of points with position vectors @xmath87 in @xmath1 .
associated with each point @xmath88 is its _ voronoi cell _ , @xmath89 , which is defined to be the region of space nearer to the point at @xmath90 than to any other point @xmath91 in the set , i.e. , @xmath92 the voronoi cells are convex polyhedra whose interiors are disjoint , but share common faces , and therefore the union of all of the polyhedra is the whole of @xmath1 .
this partition of space is called the _ voronoi _ tessellation .
the vertices of the voronoi polyhedra are the points whose distance from the points @xmath93 is a local maximum .
while the voronoi polyhedra of a lattice are congruent to one another , the voronoi polyhedra of a non - bravais lattice are not identical to one another .
a hole in a lattice is a point in @xmath1 whose distance to the nearest lattice point is a local maximum .
a _ deep _ hole is one whose distance to a lattice point is a global maximum .
the distance @xmath94 to the deepest hole of a lattice is the _ covering radius _ and is equal to the _ circumradius _ of the associated voronoi cell ( the radius of the smallest circumscribed sphere ) . in the case of the @xmath0-dimensional simple cubic ( hypercubic )
lattice @xmath47 , the voronoi cell is a hypercube and there is only one type of hole with covering radius @xmath95 , assuming unit number density @xmath96 .
figure [ reg3d - vor ] shows the voronoi cell in the case @xmath97 as well as the corresponding voronoi cells for the three - dimensional body - centered cubic and face - centered cubic lattices . note that the truncated octahedron is the most spherically symmetric of the three voronoi cells shown in fig .
[ reg3d - vor ] @xcite .
a sphere packing @xmath98 in @xmath0-dimensional euclidean space @xmath1 is a collection of @xmath0-dimensional nonoverlapping congruent spheres . the _ packing density _ or , simply , density @xmath99 of a sphere packing is the fraction of space @xmath1 covered by the spheres . for spheres of diameter @xmath100 and number density @xmath14 ,
the density is given by @xmath101 where @xmath102 is the volume of a @xmath0-dimensional sphere of radius @xmath28 .
a packing is _ saturated _ if there is no space available to add another sphere without overlapping the existing particles .
we denote the packing density of such a packing by @xmath103 .
a _ lattice packing _
@xmath104 is one in which the centers of nonoverlapping spheres are located at the points of @xmath32 .
thus , the density of a lattice packing @xmath105 consisting of spheres of diameter @xmath100 is given by @xmath106 where @xmath38 is the volume of a fundamental cell .
periodic _ packing of congruent spheres is obtained by placing a fixed configuration of @xmath4 sphere centers ( where @xmath46 ) within one fundamental cell of a lattice @xmath32 , which is then periodically replicated without overlaps .
the packing density of a periodic packing is given by @xmath107 where @xmath108 is the number density .
the sphere packing problem seeks to answer the following question : among all packings of congruent spheres , what is the maximal packing density @xmath109 , i.e. , largest fraction of @xmath1 covered by the spheres , and what are the corresponding arrangements of the spheres @xcite ? more precisely , the maximal density is defined by @xmath110 where the supremum is taken over all packings in @xmath1 .
the sphere packing problem is of great fundamental and practical interest , and arises in a variety of contexts , including classical ground states of matter in low dimensions @xcite , the famous kepler conjecture for @xmath97 @xcite , error - correcting codes @xcite and spherical codes @xcite .
the optimal solutions are known only for the first three space dimensions @xcite . for @xmath111 ,
the densest known packings are bravais lattice packings @xcite .
for example , the checkerboard " lattice @xmath48 , which is the @xmath0-dimensional generalization of the fcc lattice ( densest packing in @xmath16 ) , is believed to be optimal in @xmath112 and @xmath113 .
the remarkably symmetric self - dual @xmath72 and leech lattices in @xmath114 and @xmath115 , respectively , are most likely the densest packings in these dimensions @xcite .
table [ pack ] summarizes the densest known packings in selected dimensions . &
packing & packing density , @xmath116 + 1 & @xmath117 & 1 + 2 & @xmath118 & @xmath119 + 3 & @xmath120 & @xmath121 + 4 & @xmath122 & @xmath123 + 5 & @xmath124 & @xmath125 + 6 & @xmath126 & @xmath127 + 7 & @xmath128 & @xmath129 + 8 & @xmath130 & @xmath131 + 9 & @xmath132 & @xmath133 + 10 & @xmath134 & @xmath135 + 12 & @xmath136 & @xmath137 + 16 & @xmath76 & @xmath138 + 24 & @xmath75 & @xmath139 + for large @xmath0 , the best that one can do theoretically is to devise upper and lower bounds on @xmath109 @xcite .
for example , minkowski @xcite proved that the maximal density @xmath140 among all bravais lattice packings for @xmath141 satisfies the lower bound @xmath142 where @xmath143 is the riemann zeta function .
it is seen that for large values of @xmath0 , the asymptotic behavior of the _ nonconstructive _ minkowski lower bound is controlled by @xmath144 .
note that the density of a _ saturated _ packing of congruent spheres in @xmath1 for all @xmath0 satisfies @xmath145 which has the same dominant exponential term as ( [ mink ] ) .
this is a rather weak lower bound on the density of saturated packings because there exists a disordered but _ unsaturated packing construction _ in @xmath1 , known as the ghost " rsa packing @xcite , that achieves the density @xmath144 in any dimension .
we will employ these results in sec .
[ improve ] .
it is also known that there are saturated packings in @xmath1 with densities that exceed the scaling @xmath144 @xcite , as we will discuss in sec .
[ cov ] . in the large - dimensional limit
, kabatiansky and levenshtein @xcite showed that the maximal density is bounded from above according to the asymptotic upper bound @xmath146 we will employ sphere - packing solutions to obtain heretofore unattained results for both the covering and quantizer problems . in particular ,
we obtain coverings and quantizers utilizing disordered saturated packings in secs .
[ results - cover ] and [ sat - quant ] , respectively . in sec .
[ results - quant ] , we use the densest lattice packings to derive improved upper bounds on the quantizer error .
we denote by @xmath147 the variance in the number of points @xmath148 contained within a window @xmath149 .
the number variance @xmath147 for a specific choice of @xmath150 is necessarily a positive number and is generally related to the _ total correlation function _
@xmath151 for a translationally invariant point process @xcite , where @xmath24 is the pair correlation function defined in sec .
[ def ] . in the special case of a spherical window of radius @xmath28 in @xmath1 ,
the number variance is explicitly given by @xmath152 , \label{variance}\end{aligned}\ ] ] @xmath153 is the dimensionless volume common to two spherical windows of radius @xmath28 ( in units of the volume of a spherical window of radius @xmath28 , @xmath154 ) whose centers are separated by a distance @xmath31 .
we will call @xmath153 the _ scaled intersection volume _ , which will play an important role in this paper .
the scaled intersection volume has the support @xmath155 $ ] , the range @xmath156 $ ] , and the following alternative integral representation @xcite : @xmath157 } \sin^d(\theta ) \ ,
d\theta , \label{alpha}\end{aligned}\ ] ] where @xmath158 is the @xmath0-dimensional constant given by @xmath159}. \label{c}\end{aligned}\ ] ] torquato and stillinger @xcite found the following series representation of the scaled intersection volume @xmath153 for @xmath160 and for any @xmath0 : @xmath161 } x^{2n-1 } , \label{series}\end{aligned}\ ] ] where @xmath162 .
for even dimensions , relation ( [ series ] ) is an infinite series because it involves transcendental functions , but for odd dimensions , the series truncates such that @xmath153 is a univariate polynomial of degree @xmath0 . for example , in two and three dimensions , respectively , the scaled intersection volumes are given by @xmath163\theta(2r - r ) \quad ( d=2),\ ] ] @xmath164 \theta(2r - r ) \quad ( d=3),\ ] ] where @xmath165 is the heaviside step function .
figure [ intersection ] provides plots of @xmath153 as a function of @xmath31 for the first five space dimensions . for any dimension
, @xmath153 is a monotonically decreasing function of @xmath31 . at a fixed value of @xmath31 in the interval @xmath166
, @xmath153 is a monotonically decreasing function of the dimension @xmath0 . for spherical windows of radius @xmath28 as a function of @xmath31 for the first five space dimensions .
the uppermost curve is for @xmath167 and lowermost curve is for @xmath168 .
, width=336 ] for large @xmath28 , it has been proved that @xmath169 can not grow more slowly than @xmath170 , where @xmath171 is a positive constant @xcite .
we note that point processes ( translationally invariant or not ) for which @xmath169 grows more slowly than the window volume ( i.e. , as @xmath172 ) for large @xmath28 are examples of _ hyperuniform _ ( or superhomogeneous ) point patterns @xcite .
for hyperuniform point processes in which the number variance grows like the surface area of the window , one has @xmath173
\qquad r \rightarrow \infty,\ ] ] where @xmath174 is a dimensionless constant with @xmath158 defined by ( [ c ] ) , @xmath175 is a dimensionless density , and @xmath176 represents some microscopic " length scale , such as the minimum pair separation distance in a packing or the mean nearest - neighbor distance .
this class of hyperuniform point processes includes all periodic point patterns , quasicrystals that possess bragg peaks , and disordered hyperuniform point patterns in which the pair correlation functions decay exponentially fast to unity @xcite .
it has been shown that finding the point process that minimizes the number variance @xmath169 is equivalent to finding the ground state of a certain repulsive pair potential with compact support @xcite .
specifically , by invoking a volume - average interpretation of the number variance problem valid for a single realization of a point process , torquato and stillinger found @xcite : @xmath177 where @xmath178 .
\label{bn}\ ] ] the asymptotic coefficient @xmath179 defined by ( [ b ] ) for a hyperuniform point pattern is then related to @xmath180 by the expression @xmath181 these results imply that the asymptotic coefficient @xmath179 obtained in ( [ b ] ) involves an average over small - scale fluctuations in the number variance with length scale on the order of the mean separation between points @xcite . in the special case of a ( bravais ) lattice @xmath32
, one can express the rescaled surface - area coefficient as follows : @xmath182^{1 - 1/d}}{v_f^{1 + 1/d } } \sum_{\mathbf{q}\neq\mathbf{0 } } \frac{1}{\vert\mathbf{q}\vert^{d+1 } } , \label{num}\end{aligned}\ ] ] where we recall that @xmath38 is the volume of the fundamental cell of the lattice @xmath32 and @xmath183 represents a lattice vector in the dual ( or reciprocal ) lattice @xmath39 .
the rescaled coefficient @xmath184 renders the result independent of the length scale in the lattice @xcite . finding the lattice that minimizes @xmath179
is directly related to an outstanding problem in number theory , namely , finding the minima of the epstein zeta function @xmath185 @xcite defined by @xmath186 where @xmath187 is a lattice vector of the lattice @xmath32 .
note that the _ dual _ of the lattice that minimizes the epstein zeta function at @xmath188 among all lattices will minimize the scaled asymptotic number - variance coefficient ( [ num ] ) among lattices @xcite .
certain duality relations have been derived that establish rigorous upper bounds on the energies of such ground states and help to identify energy - minimizing lattices @xcite .
because @xmath185 is globally minimized for @xmath167 by the integer lattice @xcite and is minimized for @xmath189 among all lattices by the triangular lattice @xcite , it has been conjectured that the epstein zeta function for @xmath190 is minimized among lattices by the maximally dense lattice packing @xcite .
sarnak and str " ombergsson @xcite have proved that the conjecture can not be generally true , but for @xmath191 , @xmath192 and @xmath193 , the densest lattice packing is a strict local minimum . since as @xmath194 ,
the minimizer of epstein zeta function is the densest sphere packing in @xmath1 for any @xmath0 , it is likely that in the high - dimensional limit the minimizers of this function are non - lattices , namely , disordered sphere packings @xcite . &
structure & scaled coefficient , @xmath184 + 1 & @xmath195 & 0.083333 + 2 & @xmath118 & 0.12709 + 3 & @xmath196 & 0.15560 + 4 & @xmath197 & 0.17488 + 5 & @xmath198 & 0.19069 + 6 & @xmath199 & 0.20221 + 7 & @xmath200 & 0.21037 + 8 & @xmath201 & 0.21746 + 12 & @xmath202 & 0.24344 + 16 & @xmath203 & 0.25629 + 24 & @xmath204 & 0.26775 + in table [ var ] , we tabulate the best known solutions to the asymptotic number variance problem in selected dimensions .
values for the first three space dimensions were given in ref .
@xcite and those for @xmath205 were provided in ref .
the values reported for @xmath206 and 24 were ascertained using efficient algorithms based on alternative number - theoretic representations of the epstein zeta function @xmath185 @xcite for the corresponding densest known lattice packings for @xmath188 and then using the duality relations connecting it to the asymptotic surface - area coefficient for the number variance .
appendix [ num - ep ] provides details for these computations .
surround each of the points of a point process @xmath207 in @xmath1 by congruent overlapping spheres of radius @xmath28 such that the spheres cover the space . the _ covering density _
@xmath208 is defined as follows : @xmath209 where @xmath154 is given by ( [ v1 ] ) . the _ covering problem
_ asks for the arrangement of points with the least density @xmath208 .
we define the covering radius @xmath94 for any configuration of points in @xmath1 to be the minimal radius of the overlapping spheres to cover @xmath1 . figure [ sq - tri ] shows two examples of coverings in the plane .
the covering density associated with @xmath59 at unit number density @xmath96 is known exactly for any dimension @xmath0 @xcite : @xmath210^{d/2}.\ ] ] for the hypercubic lattice @xmath47 at @xmath96 , @xmath211 thus the ratio of the covering density for @xmath59 to that of @xmath47 is given by @xmath212^{d/2}.\ ] ] for large @xmath0 , this ratio becomes @xmath213 and thus we see that @xmath59 provides exponentially thinner coverings than that of @xmath47 in the large-@xmath0 limit .
we note that in this asymptotic limit , @xmath214 until recently , @xmath59 was the best known lattice covering in all dimensions @xmath215 .
however , for most dimensions in the range @xmath216 , schrmann and vallentin @xcite have discovered other lattice coverings that are slightly thinner than those for @xmath59 .
table [ coverings ] provides the best known solutions to the covering problem in selected dimensions . & covering & covering density , @xmath208 + 1 & @xmath217 & 1 + 2 & @xmath118 & 1.2092 + 3 & @xmath196 & 1.4635 + 4 & @xmath218 & 1.7655 + 5 & @xmath219 & 2.1243 + 6 & @xmath220 & 2.4648 + 7 & @xmath221 & 2.9000 + 8 & @xmath222 & 3.1422 + 9 & @xmath223 & 4.3401 + 10 & @xmath224 & 5.2517 + 12 & @xmath225 & 7.5101 + 16 & @xmath226 & 15.3109 + 17 & @xmath227 & 18.2878 + 18 & @xmath228 & 21.8409 + 24 & @xmath75 & 7.9035 + until the present work , there were no known explicit non - lattice constructions possessing covering densities smaller than those of the best lattice coverings in any dimension @xmath0 @xcite . in sec .
[ results - quant ] , we provide evidence that certain disordered point patterns give thinner coverings than the best known lattice coverings beginning at @xmath229 .
there is a fundamental difference between coverings associated with point patterns that have identical voronoi cells ( i.e. , lattices and periodic point patterns in which each point is equivalent ) and those point processes whose voronoi cells are generally different ( e.g. , irregular point processes ) . this salient point
is illustrated in fig .
[ tri - rsa - cov ] and explained in the corresponding caption .
rogers showed that ( possibly nonlattice ) coverings exist with @xmath230 for @xmath141 .
this is a nonconstructive upper bound .
this upper bound provides a substantially thinner covering density than that of the @xmath59 lattice in the large-@xmath0 limit [ cf .
( [ a_n^ * ] ) ] , but it is not known whether this bound becomes sharp in the large-@xmath0 limit .
the best lower bound on the covering density is given by @xmath231 to define @xmath232 , let @xmath233 be a regular simplex with edge length equal to two .
spheres of radius @xmath234 centered at the vertices of @xmath233 just cover @xmath233 .
the quantity @xmath232 is the ratio of the sums of the intersections of these spheres with @xmath233 to the volume of @xmath233 .
thus , in the large-@xmath0 limit , @xmath235 .
is quantized ( rounded - off " ) to the nearest point @xmath90 . left panel : triangular lattice ( best quantizer in @xmath236 @xcite ) .
right panel : irregular point process.,width=211 ] consider a point process @xmath237 in @xmath1 with configuration @xmath238 .
a @xmath0-dimensional _ quantizer _ is device that takes as an input a point at position @xmath239 in @xmath1 generated from some probability density function @xmath240 and outputs the nearest point @xmath90 of the point process to @xmath241 .
equivalently , if the input @xmath241 belongs to the voronoi cell @xmath89 , the output is @xmath90 ( see fig .
[ tri - rsa ] ) for simplicity , we assume that @xmath241 is uniformly distributed over a large ball in @xmath1 containing the @xmath4 points of the point process .
one attempts to choose the configuration @xmath238 of the point process to minimize the mean squared error , i.e. , the expected value of @xmath242 . specifically , the quantizer problem is to choose the @xmath4-point configuration so as to minimize the _ scaled dimensionless error _
( sometimes called the _ distortion _ ) @xcite @xmath243 where @xmath244 is a dimensionless error , @xmath245\ ] ] is the expected volume of a voronoi cell , and @xmath246 is the volume of the @xmath247th voronoi cell .
the scaling factor @xmath248 is included to compare the second moments appropriately across dimensions .
we will denote the minimal scaled dimensionless error by @xmath249 .
note that since there is one point per voronoi cell , the number density @xmath14 is set equal to unity .
if all of the voronoi cells are congruent ( as they are in the case of all lattices and some periodic point processes ) , we have the simpler expression @xmath250 where the centroid of the voronoi cell @xmath251 is the origin of the coordinate system .
lattice quantizer problem _ is to find the lattice for which @xmath252 , given by ( [ gamma2 ] ) , is minimum .
thus , @xmath252 can be interpreted as the scaled , dimensionless _
second moment of inertia _ of the voronoi cell .
& quantizer & scaled error , @xmath252 + 1 & @xmath217 & 0.083333 + 2 & @xmath118 & 0.080188 + 3 & @xmath196 & 0.078543 + 4 & @xmath197 & 0.076603 + 5 & @xmath253 & 0.075625 + 6 & @xmath199 & 0.074244 + 7 & @xmath254 & 0.073116 + 8 & @xmath255 & 0.071682 + 9 & @xmath256 & 0.071626 + 10 & @xmath257 & 0.070814 + 12 & @xmath202 & 0.070100 + 16 & @xmath203 & 0.068299 + 24 & @xmath258 & 0.065771 + the best known quantizers in any dimension @xmath0 are usually lattices that are the duals of the densest known packings ( see the discussion in sec . [ sphere ] and ref .
@xcite ) , except in dimensions 9 and 10 , where the best solutions are still lattices , but are not duals of the densest lattice packings in those dimensions .
although the best known solutions of the quantizer and covering problems are the same in the first three space dimensions , they are generally different for @xmath2 @xcite .
zador @xcite has derived upper and lower bounds on @xmath249 . conway and
sloane @xcite have obtained conjectural lower bounds on @xmath249 .
we defer the discussion of these bounds to sec .
[ results - quant ] , where we derive sharper upper bounds on @xmath249 , among other results .
table [ quant ] provides the best known solutions to the quantizer problem in selected dimensions .
table [ comparison ] lists the best known solutions of the quantizer , covering and number - variance problems and sphere - packing problems in @xmath1 for selected @xmath0 .
it is seen that in the first two space dimensions , the best known solutions for each of these four problems are identical to one another . for @xmath97 ,
the densest sphere packing is the @xmath259 or , equivalently , @xmath260 lattice , which is the dual lattice associated with the best known solutions to the quantizer , covering and number - variance problems , which is the @xmath261 lattice .
thus , for the first three space dimensions , the best known solutions for each of the four problems are lattices , and either they are identical to one another or are duals of one another . however , such relationships may or may not exist for @xmath2 , depending on the peculiarities of the dimensions involved . &
quantizer & covering & variance & packing + 1 & @xmath195 & @xmath195 & @xmath217 & @xmath217 + 2 & @xmath118 & @xmath118 & @xmath118 & @xmath118 + 3 & @xmath196 & @xmath196 & @xmath196 & @xmath120 + 4 & @xmath197 & @xmath218 & @xmath197 & @xmath197 + 5 & @xmath262 & @xmath219 & @xmath198 & @xmath124 + 6 & @xmath199 & @xmath220 & @xmath199 & @xmath126 + 7 & @xmath263 & @xmath221 & @xmath264 & @xmath128 + 8 & @xmath72 & @xmath222 & @xmath72 & @xmath72 + 9 & @xmath256 & @xmath223 & @xmath265 & @xmath132 + 10 & @xmath257 & @xmath224 & @xmath266 & @xmath134 + 12 & @xmath77 & @xmath225 & @xmath267 & @xmath136 + 16 & @xmath268 & @xmath226 & @xmath268 & @xmath76 + 24 & @xmath73 & @xmath73 & @xmath73 & @xmath73 + there is a fundamental difference between the nature of the interactions for the sphere - packing problem and those for the other three problems .
any sphere packing ( optimal or not ) consisting of nonoverlapping spheres of diameter @xmath100 is described by a short - ranged pair potential that is zero whenever the spheres do not overlap ( when the pair separation distance is greater than @xmath100 ) and is infinite whenever the pair separation distance is less than @xmath100 . by contrast , the other three problems are described by
soft " bounded interactions .
in particular , we have seen that the number variance is specified by a bounded repulsive pair potential with compact support [ cf .
( [ volavginterp ] ) ] .
we will see in the subsequent sections that the covering and quantizer problems are described by many - particle bounded interactions but of the more general form ( [ full ] ) , which involves single - body , two - body , three - body , and higher - body interactions .
one simple reason why the optimal solutions of the sphere - packing and number - variance problems are related to one another either directly or via their dual solutions in the first three space dimensions is that they both involve short - ranged repulsive pair interactions only .
the reader is referred to refs . @xcite and @xcite for a comprehensive explanation .
the reasons why the optimal solutions to these two problems are sometimes the optimal solutions for the covering and quantizer problems will become apparent in the subsequent sections .
the explanation for why the optimal covering and quantizer solutions are generally different for @xmath2 is discussed in sec .
[ reform ] .
we note that the leech lattice @xmath73 for @xmath269 is an exceptional case in that it provides the optimal solution to all four different problems .
the remarkably high degree of symmetry possessed by this self - dual lattice @xcite accounts for this unique property .
the only other dimensions where all four optimal solutions are the same are @xmath167 and @xmath189 .
we recall the definition of the void " nearest - neighbor probability density function @xmath270 @xcite : @xmath271 the void " exclusion probability @xmath272 is the _ complementary cumulative distribution function _ associated with @xmath270 : @xmath273 and hence is a monotonically decreasing function of @xmath28 @xcite .
thus , @xmath272 has the following probabilistic interpretation : @xmath274 there is another interpretation of @xmath275 that involves circumscribing spheres of radius @xmath28 around each point in a realization of the point process .
it immediately follows that @xmath272 is the _ expected _ fraction of space not covered by these circumscribing spheres . differentiating ( [ ev - cum ] ) with respect to @xmath28 gives @xmath276 note that these void quantities are different from the particle " nearest - neighbor functions @xcite in which the sphere of radius @xmath28 is centered at an actual point of the point process ( as opposed to an arbitrary point in the space ) .
it is useful to introduce the conditional " nearest - neighbor function @xmath277 @xcite , which is defined in terms of @xmath278 and @xmath279 as follows : @xmath280 where @xmath281 is the surface area of a @xmath0-dimensional sphere of radius @xmath28 [ cf .
( [ area - sph ] ) ] .
thus , we have the following interpretation of the conditional function : @xmath282 therefore , it follows from ( [ derv - ev ] ) and ( [ def - hv2 ] ) that the exclusion probability can be expressed in terms of @xmath283 via the relation @xmath284 .
\label{ev - gv}\end{aligned}\ ] ] it is clear that the void functions have the following behaviors at the origin for @xmath285 @xcite : @xmath286 moments of the nearest - neighbor function @xmath270 arise in rigorous bounds for transport properties of random media @xcite .
the @xmath9th moment of @xmath270 is defined as @xmath287 the void functions can be expressed as infinite series whose terms are integrals over the @xmath9-particle density functions @xcite .
for example , the void exclusion probability functions for a translationally invariant point process are respectively given by @xmath288 where @xmath20 is the @xmath9-particle correlation function and @xmath289 is the heaviside step function defined by ( [ heaviside ] ) .
the corresponding series for @xmath270 is obtained from the series above using ( [ derv - ev ] ) .
note that the series ( [ ev - series ] ) can be rewritten in terms of intersection volumes of spheres : @xmath290 where @xmath291 is the intersection volume of @xmath9 equal spheres of radius @xmath28 centered at positions @xmath292 .
observe that @xmath293 , where @xmath154 is the volume of a sphere of radius @xmath28 [ cf .
( [ v1 ] ) ] and @xmath153 is the scaled intersection volume [ cf .
( [ alpha ] ) ] and ( [ series ] ) ] . in the special case of a poisson point distribution ,
@xmath294 for all @xmath9 , and hence ( [ f - series ] ) immediately yields the well - known exact result for such a spatially uncorrelated point process @xmath295 the use of this relation with definition ( [ derv - ev ] ) gives @xmath296 for a single realization of @xmath4 points within a large volume @xmath297 in @xmath1 , we have @xmath298 this formula assumes that @xmath4 is sufficiently large so that boundary effects can be neglected . the second term in ( [ ev ] )
@xmath299 can be interpreted as a sum over one - body terms , which is independent of the point configuration .
clearly , the ( @xmath300)th term in ( [ ev ] ) can be interpreted as a sum over intrinsic @xmath9-body interactions , namely , @xmath301 .
thus , except for the trivial constant of unity ( the first term ) , @xmath272 can be regarded to be a many - body potential of the general form ( [ full ] ) , which heretofore was not observed .
upper and lower bounds on the so - called _
canonical @xmath9-point correlation function _
@xmath302 ( with @xmath303 and @xmath304 ) for point processes in @xmath1 have been found @xcite .
since the void exclusion probability and nearest - neighbor probability density function are just special cases of @xmath305 , then we also have strict bounds on them for such models .
let @xmath306 represent either @xmath275 or @xmath307 and @xmath308 represent the @xmath309th term of the series for these functions .
furthermore , let @xmath310 be the partial sum .
then it follows that for any of the exclusion probabilities or nearest - neighbor probability density functions , we have the bounds @xmath311 application of the aforementioned inequalities yield the first three successive bounds on the nonnegative exclusion probability : @xmath312 where @xmath313 is the intersection volume of two @xmath0-dimensional spheres of radius @xmath28 whose centers are separated by the distance @xmath314 and @xmath315 is the scaled intersection volume given by ( [ alpha ] ) .
the corresponding first two nontrivial bounds on the nonnegative pore - size density function @xmath270 are as follows : @xmath316 where @xmath317 is the surface area of the intersection volume @xmath318 .
bounds on the conditional function @xmath277 follow by combining the bounds above on @xmath279 and @xmath278 and definition ( [ def - hv2 ] ) .
for example , the following bounds have been found @xcite : @xmath319 and @xmath320 which should only be applied for @xmath28 such that @xmath321 remains positive .
for congruent sphere packings of diameter @xmath100 at packing density @xmath116 , the infinite series expansion for @xmath272 [ cf .
( [ ev - series ] ) ] will truncate after a finite number of terms for a bounded value of the radius @xmath28 .
a spherical region of radius @xmath28 centered at an arbitrary point in the space exterior to the spheres can contain at most @xmath322 sphere centers .
therefore , series truncates after @xmath323 terms , i.e. , @xmath324 for a spherical region of radius @xmath325 , @xmath326 and hence we have the exact expression that applies for @xmath327 @xmath328 for any sphere packing for which ( [ period ] ) applies such that @xmath329 , we have upon use of ( [ ev - packing ] ) the exact result @xmath330 \left(\frac{r}{d}\right)^d , & d/2 \le r \ge d/\sqrt{3 } , \end{array } \right .
\label{ev - packing2}\ ] ] where we have used the fact that @xmath331 .
importantly , this relation applies to the densest known lattice packings of spheres , at least for dimensions in the range @xmath332 .
now we can reformulate the covering and quantizers problems in terms of the void exclusion probability .
in particular , the covering problem asks for the point process in @xmath1 at unit density ( @xmath96 ) that minimizes the support of the radial function @xmath272 .
we define @xmath333 the smallest possible value of the covering radius @xmath94 among all point processes for which @xmath334 , which we call the _ minimal covering radius_. this is indeed a special ground state in which the energy " is identically zero ( i.e. , @xmath335 ) @xcite . depending on the space dimension @xmath0 , this special ground state will involve one - body interactions , the first two terms of ( [ ev ] ) , one- and two - body interactions , the first three terms of ( [ ev ] ) , one- , two- and three - body interactions , the first four terms of ( [ ev ] ) , etc . , and will truncate at some particular level , provided that @xmath272 for the point process has compact support . the minimal covering radius @xmath333 increases with the space dimension @xmath0 and , generally speaking , the highest - order @xmath9-body interaction required to fully characterize the associated @xmath272 increases with @xmath0
note that for a particular point process , twice the covering radius @xmath336 can be viewed as the effective interaction range " between any pair of points , since the intersection volume @xmath337 , which appears in expression ( [ ev ] ) for @xmath272 , is exactly zero for any pair separation @xmath338 ; moreover , for such pair separations @xmath339 can be written purely in terms of the lower - order intersection volume @xmath340 . therefore , because @xmath341 for @xmath342 , the effective interaction range between any @xmath9 points for @xmath342 is still given by @xmath336 .
the quantizer problem asks for the point process in @xmath1 at unit density that minimizes the scaled average squared error @xmath343 defined as @xmath344 we will call the minimal error @xmath249 .
thus , we seek the ground state of the many - body interactions that are involved upon substitution of ( [ ev ] ) into ( [ g ] ) .
again , depending on the dimension , this many - body energy will truncate at some particular level , provided that @xmath272 has compact support . again , as @xmath0 increases , successively higher - order interactions in the expression ( [ ev ] ) must be incorporated to completely characterize @xmath272 .
it is instructive to express explicitly the void exclusion probabilities for some common lattices and use these functions to evaluate explicitly their corresponding covering densities and scaled average squared errors in the first three space dimensions . in the simplest case of one dimension ,
the series expansion for @xmath272 [ cf .
( [ ev ] ) ] for the integer lattice @xmath53 truncates after only one - body terms . at unit number density ( @xmath96 )
, it is trivial to show that @xmath345 where the covering radius @xmath346 , @xmath347 , and the nearest - neighbor distance from a lattice point is unity . using the definitions ( [ covering ] ) and ( [ g ] ) in combination with ( [ integer ] )
yield the covering density and scaled average squared error , respectively , for the optimal integer lattice : @xmath348 @xmath349 let us now determine the void exclusion probabilities for the @xmath350 ( square ) and @xmath351 ( triangular ) and lattices for @xmath28 up to their respective covering radii for which @xmath334 .
for these lattices , the series expression ( [ ev ] ) for @xmath279 truncates after two - body terms . for the square and triangular lattices at @xmath96 , @xmath352 where @xmath353 is the nearest - neighbor distance from a lattice point and the covering radius @xmath94 , equal to one half of the next - nearest - neighbor distance , which we will denote by @xmath354 . for the square and triangular lattices at @xmath96 , @xmath355 and @xmath356 , and @xmath357 and @xmath358 respectively
. figure [ 2d - lattices ] provides plots of @xmath272 for these two @xmath189 lattices . employing the definitions ( [ covering ] ) and ( [ g ] ) in combination with ( [ square ] ) provide the covering density and scaled average squared error , respectively , for the @xmath350 lattice : @xmath359 @xmath349 similarly , the corresponding equations for the @xmath360 lattice yields the covering density and scaled average squared error , respectively , for this optimal structure : @xmath361 @xmath362 for the @xmath350 ( square ) and @xmath351 ( triangular ) lattice have support up to the covering radii @xmath356 and @xmath358 , respectively , at unit number density ( @xmath96 ) . , width=336 ] it is also useful to express explicitly the void exclusion probabilities for the @xmath363 ( simple cubic ) and @xmath261 ( bcc ) lattices for @xmath28 up to their respective covering radii for which @xmath334 .
it turns out that calculating @xmath272 in the case of the simple - cubic lattice is more complicated than that for the bcc lattice because the former involves up through four - body terms , i.e. , @xmath364 .
however , the symmetry of the geometry for @xmath363 enables one to express @xmath364 purely in terms of @xmath365 and @xmath366 .
the calculation of @xmath272 does not involve four - body terms .
for the simple cubic lattice at @xmath96 , @xmath367 where @xmath355 , @xmath368 , @xmath369 and @xmath370 is explicitly given by ( [ v3int ] ) in appendix [ v3-int ] for triangles of side lengths @xmath31 , @xmath371 and @xmath372 .
here we have used the fact that for @xmath373 , @xmath374 . using the definitions ( [ covering ] ) and ( [ g ] ) in combination with ( [ sc ] )
yield the covering density and scaled average squared error , respectively , for the @xmath363 lattice : @xmath375 @xmath349 for the bcc lattice at @xmath96 , @xmath376 where @xmath377 , @xmath378 , @xmath379 , and @xmath380 is the circumradius of a triangle of side lengths @xmath353 , @xmath353 and @xmath354 , the general expression of which is given by ( [ circum ] ) in appendix [ v3-int ] .
employing the definitions ( [ covering ] ) and ( [ g ] ) in combination with ( [ bcc ] ) provide the covering density and scaled mean squared error , respectively , for the @xmath381 lattice : @xmath382 @xmath383 figure [ 3d - lattices ] provides plots of @xmath272 for the simple cubic and bcc three - dimensional lattices .
we see that the best known solutions to the covering and quantizer problems are identical for the first three space dimensions .
however , there is no reason to expect that the optimal solutions for these two problems to be the same in higher dimensions , except for @xmath269 for reasons mentioned in sec .
although both problems involve ground states associated with the many - body interaction " function @xmath272 , such that it possesses compact support for finite @xmath0 , the precise shape of the function @xmath272 for a particular point configuration is crucial in determining its first moment or quantizer error .
by contrast , the best covering seeks to find the point configuration that minimizes the support of @xmath272 without regard to its shape .
for the @xmath363 ( simple cubic ) lattice and @xmath261 ( bcc ) lattice have support up to the covering radii @xmath369 and @xmath379 , respectively , at unit number density ( @xmath96 ) . , width=336 ]
saturated sphere packings in @xmath1 should provide relatively thin coverings . surrounding every sphere of diameter @xmath100 in any saturated packing of congruent spheres in @xmath1 at packing density @xmath103 by spheres of radius @xmath100
provides a covering of @xmath1 , and thus the associated covering density @xmath384 is given by @xmath385 where @xmath386 and @xmath387 are the number density and packing density , respectively , of the saturated packing .
what is the thinnest possible covering associated with a saturated packing ?
it immediately follows that the thinnest coverings among saturated congruent sphere packings in @xmath1 are given by the saturated packings that have the minimal packing density @xmath388 in that space dimension and have covering density @xmath389 we can also bound the packing density @xmath103 of any saturated sphere packing in @xmath1 from above using upper bounds on the covering density . * lemma 1 : * _ the density @xmath103 of any saturated sphere packing in @xmath1 is bounded from above according to _ @xmath390 the proof is trivial in light of the upper bound ( [ cover - up ] ) on the covering density and relation ( [ theta - den ] ) .
the standard random sequential addition ( rsa ) sphere packing is a time - dependent process produced by randomly , irreversibly , and sequentially placing nonoverlapping spheres into a large region @xmath391 @xcite .
initially , this large region is empty of sphere centers and subsequently spheres are added provided each attempted placement of a sphere does not overlap an existing sphere in the packing . if an attempted placement results in an overlap , further attempts are made until the sphere can be added to the packing at that time without violating the impenetrability constraint . for identical @xmath0-dimensional rsa spheres
, the filling process terminates at the _ saturation limit _ at infinitely long times in the infinite - volume limit .
thus , in this limit , the rsa packing is a saturated packing .
the saturation density @xmath103 can only be determined exactly in one dimension , where it is known to be @xmath392 @xcite . in higher dimensions , the saturation density
can only be determined from computer simulations .
earlier work focused on two and three dimensions , where is was found that @xmath393 @xcite and @xmath394 @xcite , respectively .
more recent numerical work has reported rsa saturation densities for the first six space dimensions @xcite .
it is important to emphasize that upper bound ( [ sat - up ] ) provides a very poor estimate of the rsa saturation density ; for example , for @xmath97 , it gives an upper bound greater than unity ( 2.32224 , which is about 6.1 times larger than the value obtained from simulations ) and for @xmath395 , it provides the upper bound @xmath396 , which is about 7.3 times larger than the value obtained from simulations @xcite .
thus , bound ( [ sat - up ] ) could only be realized by rsa saturated packings , if at all , in high dimensions .
therefore , it is useful here to compute the corresponding covering densities for rsa packings . the numerical data for rsa saturation densities for @xmath397 reported in ref .
@xcite was well approximated ( with a correlation coefficient of 0.999 ) by the following form : @xmath398 where @xmath399 and @xmath400 .
this can be used to estimate the rsa saturation densities for @xmath401 . however , even though the upper bound ( [ sat - up ] ) on the packing density of a saturated sphere packing grossly overestimates the rsa value for the first six space dimensions , we include in the fit function for @xmath103 a @xmath402 correction , which will lead to a more conservative estimate of the covering density , as explained shortly @xcite .
including such a term , we find the following fitted function for the saturated rsa packing density : @xmath403 with a correlation coefficient of @xmath404 , where @xmath405 , @xmath406 and @xmath407 , does as well as ( [ linear ] ) for @xmath397 .
since ( [ log ] ) predicts slightly higher densities than ( [ linear ] ) for @xmath408 @xcite , we use it to obtain the corresponding estimate of the rsa covering density , namely , @xmath409 which is a slightly more conservative estimate , since ( [ linear ] ) would yield a slightly thinner covering density . in table
[ rsa ] , we provide estimates of the rsa covering densities for selected dimensions up through @xmath269 .
the covering density for @xmath167 is determined from reyi s exact saturation packing density value @xcite and multiplying it by 2 .
the values for @xmath410 is are obtained from the reported saturated density values in ref .
@xcite and multiplying each density value by @xmath411 .
the values reported for @xmath412 are estimates obtained from the fitting formula ( [ log2 ] ) .
comparing table [ rsa ] to table [ coverings ] for the best known coverings , we see that saturated rsa packings not only provide relatively thin coverings but putatively represent the first non - lattices that yield thinner coverings than the best known lattice coverings beginning in about dimension 17 .
this suggests that saturated rsa packings may be thinner than the previously best known coverings for @xmath413 and probably for some dimensions greater than 24 . & covering density , @xmath384 & packing density , @xmath103 + 1 & 1.4952 & 0.74759 + 2 & 2.1880 & 0.54700 + 3 & 3.0622 & 0.38278 + 4 & 4.0726 & 0.25454 + 5 & 5.1526 & 0.16102 + 6 & 6.0121 & 0.09394 + 7 & 7.0512 & 0.05508 + 8 & 8.0526 & 0.03145 + 9 & 10.0706 & 0.01769 + 10 & 11.0860 & 0.009834 + 12 & 12.1052 & 0.002955 + 16 & 16.2141 & @xmath414 + 17 & 17.2482 & @xmath415 + 18 & 18.2848 & @xmath416 + 24 & 24.5489 & @xmath417 +
using the successive lower and upper bounds on the void exclusion probability function @xmath272 given in the previous section , we can , in principle , derive corresponding bounds on the minimal error @xmath249 .
moreover , one can obtain a variety of upper bounds on @xmath249 using our approach by utilizing the exact form of the void exclusion probability , when it is known , for some point process at unit density .
since a general point process must have an error @xmath343 that is generally larger than the minimal @xmath249 , it trivially follows that @xmath418 to illustrate how we can obtain bounds on @xmath249 using our approach , we begin by rederiving the following bounds due to zador @xcite : @xmath419 consider the lower bound first .
combination of relation ( [ g ] ) and lower bound ( [ ev - bound1 ] ) yields at unit density @xmath420 dr=\frac{1}{(d+2)\pi}\gamma(1+d/2)^{2/d},\ ] ] which is seen to be equal to zador s lower bound .
here @xmath421 is the zero of @xmath422 .
it is clear that a sphere of radius @xmath423 has the smallest second moment of inertia of any solid @xmath0-dimensional solid , and hence establishes the lower bound .
the simplest example of a point process for which @xmath272 is known is the poisson point process [ cf .
( [ poisson ] ) ] .
substitution of ( [ poisson ] ) into ( [ upper ] ) at unit density yields @xmath424 which is seen to be equal to zador s upper bound .
these derivations of the inequalities stated in ( [ zador ] ) appear to be much simpler than the ones presented by zador .
in the large-@xmath0 limit , zador s upper and lower bounds become identical , and hence one obtains the exact asymptotic result @xmath425 the convergence of zador s bounds to the exact asymptotic limit is to be contrasted with the sphere - packing problem in which the best upper and lower bounds on the maximal density become exponentially far apart in the high - dimensional limit . improved upper bounds on @xmath249 can be obtained by considering those point processes corresponding to a sphere packing for which the minimal pair separation is @xmath100 and lower bounds on the conditional function @xmath321 for @xmath426 . in what follows ,
we present two different upper bounds on @xmath249 based on this idea that improve upon zador s upper bound . for any packing of identical spheres with diameter @xmath100
, the following exact relations on the nearest - neighbor quantities apply for @xmath427 @xcite : @xmath428 observe that @xmath321 is a monotonically increasing function of @xmath28 in the interval @xmath429 $ ] . from these equalities
, it immediately follows that @xmath430 consider now the class of sphere packings for which the conditional nearest - neighbor function is bounded from below according to @xmath431 this class of packings includes equilibrium ( gibbs ) ensembles of hard spheres along the disordered fluid branch of the phase diagram @xcite , nonequilibrium disordered sphere packings , such as the ghost " random sequential addition ( rsa ) process @xcite , and a large class of lattice packings of spheres , as will be described below . in the equilibrium cases
, it is known that @xmath321 is a monotonically increasing function of @xmath28 for @xmath285 and thus using this property together with the equality @xmath432 [ cf .
( [ equality ] ) ] means that the lower bound ( [ gv - bound ] ) is obeyed . for one - dimensional equilibrium
rods , " the bound ( [ gv - bound ] ) is sharp ( exact ) @xmath321 for all realizable @xmath433 $ ] . a bound of the type ( [ gv - bound ] ) was used to bound the related particle " mean nearest - neighbor distance from above for different classes of sphere packings for all @xmath0 @xcite . using definition ( [ ev - gv ] ) and inequality ( [ gv - bound ] ) , the void exclusion probability function obeys the following upper bound for @xmath426 : @xmath434\right\ } , \quad r \ge d/2 . \label{ev-1}\ ] ] since any upper bound on the nonnegative function @xmath272 leads to an upper bound on its first moment , we then have upon use of ( [ upper ] ) , ( [ exact ] ) and ( [ ev-1 ] ) , the upper bound @xmath435^{2/d}}{d\pi}\left[\frac{(d+2(1-\phi))}{4(2+d ) } + \frac{(1-\phi)}{2d } \left(\frac{1-\phi } { \phi}\right)^{2/d}\exp\left(\frac{\phi}{1-\phi}\right ) \gamma\left(\frac{2}{d},\frac{\phi}{1-\phi}\right)\right ] , \label{new - upper}\ ] ] where @xmath436 is the incomplete gamma function .
observe that the prefactor multiplying the bracketed expression is @xmath437 , where , in light of ( [ v1 ] ) and ( [ den ] ) , @xmath438^{1/d}/\sqrt{\pi}$ ] , assuming unit number density .
note also that the upper bound ( [ new - upper ] ) depends on a single parameter , namely , the packing density @xmath116 .
thus , there is an optimal packing density @xmath439 $ ] that yields the best ( smallest ) upper bound for any particular @xmath0 , where @xmath440 is the maximal packing density .
since the right side of the inequality is a monotonically decreasing function of @xmath116 for any @xmath0 , then the optimal density @xmath441 is , in principle , given by @xmath440 .
it is noteworthy that the upper bound ( [ new - upper ] ) for the optimal choice @xmath442 may still be valid for a packing even if the bound ( [ ev-1 ] ) , upon which it is based , is violated for @xmath28 of the order of @xmath100 because the exponential tail can more than compensate for such a violation such that the error [ first moment of @xmath272 ] is overestimated .
observe also that because @xmath444 + { \cal o}(1 ) \quad ( d \rightarrow \infty)\ ] ] the upper bound ( [ new - upper ] ) tends to the exact asymptotic result ( [ asympt ] ) of @xmath445 .
before discussing the optimal bounds , it is useful to begin with an application of the upper bound ( [ new - upper ] ) for the _ sub - optimal _ case of a disordered sphere packing , namely , the aforementioned _ ghost _ rsa packing process @xcite , which we now show generally improves on zador s upper bound .
this represents the only exactly solvable disordered sphere - packing model for all realizable densities and in any dimension , as we now briefly describe .
the ghost rsa packing process involves a ( time - dependent ) sequential addition of spheres in space subject to the nonoverlap condition .
not only is an attempted addition of a sphere rejected if it overlaps an existing sphere of the packing , it is also rejected if it overlaps any previously rejected sphere ( called a ghost " sphere ) . unlike the standard rsa packing , the ghost rsa packing does not become a saturated packing in the infinite - time limit .
all of the @xmath9-particle correlation functions for this nonequilibrium model have been obtained analytically for any @xmath0 , time @xmath372 , and for all realizable densities .
for example , one can show that the maximal density ( achieved at infinite time ) is given by @xmath446 and the associated pair correlation function is @xmath447 where @xmath289 is the unit step function , equal to zero for @xmath448 and unity for @xmath449 .
it is straightforward to verify that the upper bound on the exclusion probability @xmath272 for this infinite - time case obtained by using ( [ g2-grsa ] ) in the inequality ( [ ev - bound2 ] ) is always below the upper bound ( [ ev-1 ] ) .
therefore , the upper bound ( [ new - upper ] ) is valid at the maximal density , i.e. , at @xmath450 , we have @xmath451^{2/d}}{d\pi}\left[\frac{(d+2(1 - 1/2^d))}{4(2+d ) } + \frac{2(1 - 1/2^d)^{(2+d)/d}}{d } \exp\left(\frac{1}{2^d-1}\right ) \gamma\left(\frac{2}{d},\frac{1}{2^d-1}\right)\right ] .
\label{upper - grsa}\ ] ] for @xmath167 , 2 and 3 , this upper bound yields @xmath452 , @xmath453 and @xmath454 , respectively , which is to be compared to zador s upper bound , which gives @xmath455 , @xmath456 and @xmath457 , respectively .
we note that in the large-@xmath0 limit , the upper bound ( [ upper - grsa ] ) yields the exact asymptotic result ( [ asympt ] ) , which implies that the upper bound ( [ ev-1 ] ) on @xmath272 becomes exact for ghost rsa packings , tending to the unit step function in this asymptotic limit , i.e. , @xmath458 this asymptotic result implies the following corresponding one for the void nearest - neighbor probability density function : @xmath459 scaled by @xmath460 for the optimal lattice packings for @xmath189 ( @xmath461 ) and @xmath191 ( @xmath462 ) , as obtained from ( [ ev - packing2 ] ) , for @xmath463 ( solid curves ) to the corresponding estimates obtained from ( [ ev-1 ] ) for these cases ( dashed curves ) .
it is only for the case @xmath189 that estimate ( [ ev-1 ] ) is not a rigorous pointwise upper bound on the exact void exclusion probability for @xmath332 and , likely , for @xmath464 .
the exponential tail associated with ( [ ev-1 ] ) more than compensates for the narrow pointwise violation for the special case @xmath189 , resulting in a strict upper bound on the first moment of @xmath272 , i.e. , ( [ new - upper ] ) remains a strict upper bound for @xmath189 .
, width=336 ] we now return to finding the optimal ( smallest ) upper bound ( [ new - upper ] ) for each dimension . for @xmath167 ,
the optimal density is @xmath465 , which produces the sharp bound @xmath466 this bound is exact in this case because the inequality ( [ gv - bound ] ) is exact for all realizable densities for equilibrium hard rods , " including at @xmath467 , which corresponds to the optimal integer lattice packing .
this is to be contrasted with zador s upper bound , which yields 1/2 for @xmath167 and is far from the exact result .
the improved upper bound ( [ new - upper ] ) in the higher dimensions reported in table [ tab - eta-3 ] is obtained by evaluating it at the densities of the densest known lattice packings in these respective dimensions @xcite .
we note that it is only for optimal triangular lattice packing in @xmath236 that the upper bound ( [ ev-1 ] ) on @xmath272 is violated pointwise for a small range of @xmath28 around @xmath468 [ inequality ( [ ev-1 ] ) is obeyed for @xmath469 and in the vicinity of @xmath470 , but the exponential tail associated with ( [ ev-1 ] ) more than compensates for this narrow pointwise violation , resulting in a strict upper bound on the first moment of @xmath272 , i.e. , ( [ new - upper ] ) remains a strict upper bound for @xmath189 . using relation ( [ ev - packing2 ] ) and
lattice coordination properties , it is easily verified that the ( [ ev-1 ] ) is a strict upper bound on the void exclusion probability for the densest known lattice packings for all @xmath426 and @xmath471 as well as @xmath167 , and hence inequality ( [ new - upper ] ) provides a strict upper bound on the scaled error for all of these lattices . for illustration purposes , we compare in fig .
[ ev - bounds ] the exact result for @xmath275 obtained from ( [ ev - packing2 ] ) to the estimate ( [ ev-1 ] ) for the cases @xmath189 and @xmath191 for @xmath463 .
the upper bound ( [ new - upper ] ) is generally appreciably tighter than zador s upper bound for low to moderately high dimensions .
@xmath0 & quantizer & scaled error , @xmath252 & conjectured & improved + & & & lower bound & upper bound + 1 & @xmath217 & 0.083333 & 0.083333 & 0.083333 + 2 & @xmath118 & 0.080188 & 0.080188 & 0.080267 + 3 & @xmath196 & 0.078543 & 0.077875 & 0.079724 + 4 & @xmath197 & 0.076603 & 0.07609 & 0.078823 + 5 &
@xmath198 & 0.075625 & 0.07465 & 0.078731 + 6 & @xmath199 & 0.074244 & 0.07347 & 0.077779 + 7 & @xmath264 & 0.073116 & 0.07248 & 0.076858 + 8 & @xmath255 & 0.071682 & 0.07163 & 0.075654 + 9 & @xmath256 & 0.071626 & 0.070902 & 0.075552 + 10 & @xmath257 & 0.070814 & 0.070405 & 0.074856 + 12 & @xmath202 & 0.070100 & 0.06918 & 0.073185 + 16 & @xmath203 & 0.068299 & 0.06759 & 0.070399 + 24 & @xmath204 & 0.065771 & 0.06561 & 0.067209 + [ tab - eta-3 ] for any saturated packing of identical spheres of diameter @xmath100 , @xmath272 by definition is exactly zero for @xmath28 beyond the diameter , i.e. , @xmath472 in the particular case of saturated rsa packings , the void exclusion probability can computed using the same techniques described in ref .
@xcite for the first six space dimensions .
these results are summarized in fig .
[ rsa - all ] .
the corresponding quantizer errors for these dimensions are listed in table [ rsa - quant ] .
we see that the discrepancies between the saturated rsa quantizer error improves as @xmath0 increases as compared to the best known quantizer error reported in table [ quant ] . for saturated rsa packings of congruent spheres of diameter @xmath100 for the first six space dimensions.,width=288 ] &
quantizer error , @xmath473 + 1 & 0.11558 + 2 & 0.09900 + 3 & 0.09232 + 4 & 0.08410 + 5 & 0.07960 + 6 & 0.07799 + * lemma 2 : * _ saturated sphere packings in @xmath1 possess void nearest - neighbor functions that tend to the following high - dimensional asymptotic behaviors : _
@xmath474 for any saturated packing at packing density @xmath103 , it is clear that @xmath272 is bounded from above for @xmath475 as follows : @xmath476 let @xmath473 denote the scaled dimensionless quantizer error for a saturated packing . combination of the expression for @xmath272 in ( [ exact ] ) and ( [ rsa - bound ] ) yields the following upper bound on @xmath473 : @xmath477^{2/d}}{d\pi}\left[\frac{(d+2(1-\phi_s))}{4(2+d ) } + \frac{3(1-\phi_s)}{8 } \right ] , \label{new - upper2}\ ] ] the fact that this upper bound becomes exact in the high - dimensional limit ( that is , it tends to @xmath478 ) , implies that @xmath272 and @xmath270 for a saturated packing tends to the unit step function and radial delta function , as specified by ( [ lemma2 ] ) .
not surprisingly , the bound ( [ new - upper2 ] ) is not that tight in relatively low dimensions .
we have reformulated the covering and quantizer problems as the determination of the ground states of interacting point particles in @xmath1 that generally involve single - body , two - body , three - body , and higher - body interactions ; see sec .
[ reform ] .
the @xmath9-body interaction is directly related to a purely geometrical problem , namely , the intersection volume of @xmath9 spheres of radius @xmath28 centered at @xmath9 arbitrary points of the system in @xmath1 .
this was done by linking the covering and quantizer problems to certain optimization problems involving the void " nearest - neighbor functions .
this reformulation allows one to employ theoretical and numerical optimization techniques to solve these energy minimization problems .
a key finding is that disordered saturated sphere packings provide relatively thin coverings and may yield thinner coverings than the best known lattice coverings in sufficiently large dimensions . in the case of the quantizer problem , we derived improved upper bounds on the quantizer error that utilize sphere - packing solutions .
these improved bounds are generally substantially sharper than an existing upper bound due to zador in low to moderately large dimensions .
moreover , we showed that disordered saturated sphere packings yield relatively good quantizers .
our reformulation helps to explain why the known solutions of quantizer and covering problems are identical in the first three space dimensions and why they can be different for @xmath2 . in the first three space dimensions , the best known solutions of the sphere packing and number variance problems are directly related to those of the covering and quantizer problems , but such relationships may or may not exist for @xmath2 , depending on the peculiarities of the dimensions involved .
it is clear that as @xmath0 becomes large , the quantizer problem becomes the easiest to solve among the four ground - state problems considered in this paper , since , unlike the other three problems , the asymptotic quantizer error tends to the same limit independent of the configuration of the point process .
the detection of gravitational waves from various astrophysical sources has and will be searched for in the output of interferometric networks @xcite by correlating the noisy output of each interferometer with a set of theoretical waveform templates @xcite .
depending upon the source , the parameter space is generally multidimenisonal and can be as large as @xmath229 or larger for inspiraling binary black holes .
the templates must _ cover _ the space and the challenge is to place them in some optimal fashion such that the fewest templates are used to reduce computational cost without reducing the detectability of the signals .
optimal template placement for gravitational wave data analysis has proved to be highly nontrivial .
one solution proposed for the optimal placement in flat ( euclidean ) space is to simply use the optimal solution to the covering problem @xcite .
however , this requires every point in the parameter space to be covered by at least one template , which rapidly becomes inefficient in higher dimensions when optimal lattice covering solutions are employed .
another approach to template placement consists in relaxing the strict requirement of complete coverage for a given mismatch , and instead require coverage only with a certain confidence @xcite .
such stochastic approaches have involved randomly placing templates in the parameter space , accompanied by a pruning " step in which redundant " templates , which are deemed to lie too close to each other , are removed @xcite .
the pruning step may be a complication that can be avoided , as discussed below .
another approach is to place spherical templates down according to a poisson point process @xcite , which has been claimed to provide good solutions for @xmath479 .
the problem with the latter approach is that there will be numerous multiple overlaps of templates , which only increases as the number density of templates ( intensity of the poisson point process ) is increased in order to cover as much of the space as is computationally feasible .
the results of the present study suggest alternative solutions to the optimal template construction problem .
first , we remark that if the covering of space by the templates is relaxed , then it is possible that the optimal lattice quantizers could serve as good solutions in relatively low dimensions ( @xmath480 ) because the mean square error is minimized .
second , in such relatively low dimensions , we have shown that saturated rsa sphere packings provide both relatively good coverings and quantizers , and hence may be useful template - based constructions for the search of gravitational waves . indeed
, we have shown that saturated rsa sphere packings are expected to become better solutions as @xmath0 becomes large . however , for @xmath481 , it will be computationally costly to create truly saturated packings , which by definition provide coverings of space ( see sec .
[ results - cover ] ) .
however , the existing stochastic approaches do not require complete coverage of space and hence an _ unsaturated _ rsa sphere packing that gets relatively close to the saturation state might still be more computationally efficient than either the random placement / pruning technique @xcite ( because pruning is unnecessary ) or the poisson placement procedure @xcite ( because far fewer spheres need to be added ) .
moreover , when the template parameter space is curved , which occurs in practice @xcite , the rsa sphere packing would be computational faster to adapt than lattice solutions . in future work
, we will explore whether our reformulations of the covering and quantizer problems as ground - state problems of many - body interactions of the form ( [ full ] ) can facilitate the search for better solutions to these optimization tasks .
clearly , a computational challenge in high dimensions will be the determination of the intersection volume @xmath339 of @xmath9 spheres of radius @xmath28 at @xmath9 different locations in @xmath1 for sufficiently large @xmath9 .
however , it is possible that that the series representations ( [ ev ] ) for @xmath272 and bounds on this quantity [ cf . sec .
[ bounds - e ] ] can be used to devise useful approximations of the monotonic function @xmath272 , which should be zero for @xmath482 , where @xmath94 is the bounded covering radius .
such approximation could be employed to evolve an initial guess of the configuration of points within a fundamental cell to useful but sub - optimal solutions , which upon further refinement could suggest novel solutions . in short , the implications of our reformulations to discover better solutions to the covering and quantizer problems in selected dimensions have yet to be fully investigated and deserves future attention .
i am very grateful to yang jiao and chase zachary for their assistance in creating many of the figures for this manuscript and for many useful discussions .
i thank andreas str " ombergsson and zeev rudnick for their generous help in providing efficient algorithms to compute the epstein zeta function for @xmath206 and 24 .
i also thank henry cohn for introducing me to the quantizer problem as well as for many helpful discussions .
this work was supported by the office of basic energy sciences , u.s .
department of energy , under grant no . de - fg02 - 04-er46108 .
here we summarize the steps in computing the asymptotic number - variance coefficient ( [ num ] ) for the lattices @xmath77 , @xmath76 and @xmath73 , which correspond to the densest lattice packings in dimensions @xmath483 and 24 , respectively .
importantly , the sum in ( [ num ] ) converges slowly .
we noted in sec .
[ num - var ] that the _ dual _ of the lattice that minimizes the epstein zeta function @xmath185 [ defined by ( [ epstein ] ) ] at @xmath188 among all lattices will minimize the asymptotic number - variance coefficient ( [ num ] ) among lattices .
we will exploit number - theoretic representations of the epstein zeta function that enable its efficient numerical evaluation and thus efficient computation of the asymptotic number - variance coefficient ( [ num ] ) using the aforementioned duality relation .
first , let us note that epstein zeta function @xmath185 ( [ epstein ] ) can be rewritten as follows : @xmath484 where @xmath79 is the coordination number at a radial distance @xmath485 from some lattice point in the lattice @xmath32 .
[ note that the epstein zeta function defined in this way applies to a general periodic point process provided that @xmath79 is interpreted in the generalized sense discussed in ( [ def ] ) . ]
the quantities @xmath79 and @xmath485 for many well - known lattices in @xmath1 can be obtained analytically using the _
theta series _ for a lattice @xmath32 , which is defined by @xmath486 and is directly related to the quadratic form associated with the lattice @xcite .
this series expression can usually be generated from the simpler functions @xmath487 @xmath488 , and @xmath489 , which are defined by @xcite : @xmath490 specifically , for the @xmath77 , @xmath76 and @xmath73 lattices , the associated theta series are given by @xcite @xmath491 @xmath492\nonumber\\ & = & 1 + 4320q^4 + 61440q^6 + \cdots \end{aligned}\ ] ] @xmath493 ^ 3 - \frac{45}{16}[\theta_2(q)\theta_3(q)\theta_4(q)]^8\nonumber\\ & = & 1 + 196560q^4 + 16773120q^6 + \cdots \end{aligned}\ ] ] where @xmath494 direct evaluation of ( [ zeta ] ) has the same convergence problems that the direct evaluation of the asymptotic number - variance coefficient ( [ num ] ) .
however , we can exploit alternative number - theoretic representations of ( [ zeta ] ) to facilitate its evaluation .
in particular , there is an expression for the epstein zeta function that can be derived using poisson summation @xcite : @xmath495 where @xmath496 and @xmath497 is the complementary incomplete gamma function .
it is important to note that the volumes of the fundamental cells of the lattice and its dual associated with the first and second sums in ( [ f ] ) , respectively , are both taken to be unity here . using the appropriate theta series given above for the lattices corresponding to the densest lattice packings for @xmath17 and 24 and expression ( [ f ] ) for @xmath188
, one finds the corresponding epstein zeta functions to be @xmath498 now since all of these lattices are self - dual ( i.e. , @xmath499 , @xmath500 , @xmath501 ) , we can directly determine the corresponding asymptotic number - variance from ( [ num ] ) by replacing the sum therein with the appropriate evaluation of the epstein zeta function specified by relations ( [ 12])-([24 ] ) .
the asymptotic number - variance values for @xmath17 and 24 reported in table [ var ] were obtained in this fashion .
for @xmath97 , the intersection volume @xmath370 of three identical spheres of radius @xmath28 whose centers are separated by the distances @xmath31 , @xmath371 , and @xmath372 for @xmath502 is given by @xcite @xmath503 where @xmath504 , @xmath505^{1/2 } } , \label{circum}\ ] ] is the circumradius of the triangle with side length lengths @xmath31 , @xmath371 , and @xmath372 and @xmath506 .
h. reiss , h. l. frisch , and j. l. lebowitz , j. chem .
phys . * 31 * , 369 ( 1959 ) .
this paper considered void " nearest - neighbor functions for the special case of identical hard spheres in equilibrium ( gibbs ensemble ) .
s. torquato , b. lu , and j. rubinstein , phys .
a * 41 * , 2059 ( 1990 ) .
this paper considered _ two types _ of nearest - neighbor functions ( void " and particle " quantities ) for a general nonequilibrium " case of identical of spheres with arbitrary interactions _
e.g. _ , spheres with variable interpenetrability that interact with repulsive / attractive forces .
mathematicians usually define a dual bravais lattice to have a fundamental cell volume @xmath507 ( i.e. , without the factor of @xmath508 ) , in which case self - duality is defined with respect to unit density ; see ref .
@xcite .
a laminated lattice @xmath70 in @xmath1 is built up of layers of @xmath509-dimensional lattices in @xmath0-dimensional euclidean space .
note that a general lattice has been denoted by @xmath32 , which should not be confused with the specific laminated lattice @xmath70 . these are common notations for both objects and therefore we adhere to this convention .
s. torquato and y. jiao , nature * 460 * , 876 ( 2009 ) ; s. torquato and y. jiao , phys .
e * 80 * , 041104 ( 2009 ) . in these papers ,
the _ asphericity _ of a nonspherical solid body is defined to be ratio of the circumradius to the inradius of the circumsphere and insphere of the nonspherical particle , respectively , which provides a measure of the spherical asymmetry of a solid body .
an asphericity equal unity corresponds to a perfect sphere .
s. torquato , j. stat . phys .
* 45 * , 843 ( 1986 ) .
canonical @xmath9-point correlation function _
@xmath302 is a _ hybrid _
probablistic function in that it contains the character of a correlation function , probability function , and probability density function , depending upon its arguments @xmath512 and the values of @xmath513 , @xmath514 and @xmath515 . from @xmath516 , which has been explicitly represented as a series involving certain integrals over the @xmath9-particle correlation function @xmath20 , one can derive any of the statistical descriptors that have arisen in the theory of random media and statistical mechanics as well as their generalizations .
the reader is referred to ref .
@xcite for a comprehensive discussion of @xmath516 .
there is an uncountably infinite number of periodic point configurations in @xmath1 in which circumscribed overlapping spheres of radius @xmath94 around each of the points just cover the space and renders @xmath517 such that for @xmath518 .
the point configuration that minimizes the support of @xmath272 , i.e. , minimizes @xmath94 ( called @xmath333 ) is the unique ground - state configuration or the optimal covering .
twice the covering radius @xmath336 for a particular point process determines the effective range " of the interaction associated with @xmath272 ; see the text for further explanation .
note that leading order term in either ( [ linear ] ) or ( [ log ] ) involving @xmath519 is a dominant dimensional contribution for rsa saturation densities in relatively low dimensions for good theoretical reasons @xcite . the fact that this term does not appear in the upper bound ( [ sat - up ] ) is another reason why it could only be realizable by rsa saturated packings in high dimensions .
if true , then the @xmath520 and @xmath521 terms in ( [ sat - up ] ) are high - dimensional asymptotic corrections .
observe that the extrapolation of the conservative fit function ( [ log ] ) to @xmath269 gives a density that is about 3.7 percent larger than that predicted by ( [ linear ] ) .
this percentage difference between the two fit functions decreases as @xmath0 decreases to @xmath522 .
this includes the operational ground - based laser interferometer gravitational - wave observatory ( ligo ) and the space - based laser interferometer space antenna ( lisa ) , which is expected to be operational in the next six years .
see the link `` ' ' astro2010 : the astronomy and astrophysics decadal survey . " | it is known that the sphere packing problem and the number variance problem ( closely related to an optimization problem in number theory ) can be posed as energy minimizations associated with an infinite number of point particles in @xmath0-dimensional euclidean space @xmath1 interacting via certain repulsive pair potentials .
we reformulate the covering and quantizer problems as the determination of the ground states of interacting particles in @xmath1 that generally involve single - body , two - body , three - body , and higher - body interactions .
this is done by linking the covering and quantizer problems to certain optimization problems involving the void " nearest - neighbor functions that arise in the theory of random media and statistical mechanics .
these reformulations , which again exemplifies the deep interplay between geometry and physics , allow one now to employ theoretical and numerical optimization techniques to analyze and solve these energy minimization problems .
the covering and quantizer problems have relevance in numerous applications , including wireless communication network layouts , the search of high - dimensional data parameter spaces , stereotactic radiation therapy , data compression , digital communications , meshing of space for numerical analysis , and coding and cryptography , among other examples . in the first three space dimensions , the best known solutions of the sphere packing and number variance problems ( or their dual " solutions ) are directly related to those of the covering and quantizer problems , but such relationships may or may not exist for @xmath2 , depending on the peculiarities of the dimensions involved .
our reformulation sheds light on the reasons for these similarities and differences .
we also show that disordered saturated sphere packings provide relatively thin ( economical ) coverings and may yield thinner coverings than the best known lattice coverings in sufficiently large dimensions . in the case of the quantizer problem ,
we derive improved upper bounds on the quantizer error using sphere - packing solutions , which are generally substantially sharper than an existing upper bound in low to moderately large dimensions .
we also demonstrate that _ disordered _ saturated sphere packings yield relatively good quantizers .
finally , we remark on possible applications of our results for the detection of gravitational waves . |
although the general picture of synchrotron emission at low energies and inverse compton at high energies is well established , important aspects of blazars are not well understood . in particular , the location of the gamma - ray emission region is not clearly established .
there are models and observations that argue for a location close to central engine ( @xmath0 pc , e.g. , * ? ? ?
* ; * ? ? ?
* ) , and also another set of models and observations pointing to a location further down the jet , at tens of parsecs ( e.g. , * ? ? ?
* ; * ? ? ?
this problem would have a straightforward answer if imaging in gamma - rays was possible , but this is not the case with current technology where we can get angular resolutions of @xmath1 at the highest gamma - ray energies observable with the large area telescope onboard the fermi gamma - ray space telescope ( _ fermi_-lat , * ? ? ?
nonetheless , we have a good idea about the origin of the radio emission , thanks to the submilliarcsecond resolution achievable with very long baseline interferometry ( e.g. , * ? ? ?
we also know that blazars are extremely variable at most wavelengths and one alternative is to use that property to put constraints on the relative location of the radio and gamma - ray emission .
the basic idea of our programme is that if the radio and gamma - ray emission are produced in the same region , we expect to see correlated variability between the light curves of these two bands , while delayed emission will tell us about the relative location of these two emission regions .
this observational programme requires simultaneous monitoring of a large sample of objects in the radio and gamma - ray bands . for the gamma - ray band we can use the excellent capabilities of _ fermi _ , that provide continuous monitoring of the sky with complete coverage every 3 hours .
the complementary resources needed in the radio band are provided by the ovro 40 m monitoring programme described below .
since mid-2007 we have carried out a dedicated long - term monitoring programme at 15 ghz using the owens valley radio observatory 40 meter telescope
. we are currently observing about 1800 blazars twice per week .
the sample includes all the _
fermi_-lat detected blazars north of declination @xmath2 .
a detailed description of the programme can be found in @xcite , and a list of publications and data in our website http://www.astro.caltech.edu / ovroblazars/. one of the main advantages of continuously monitoring a large number of sources is that we can study the variability properties of a large sample , and their variation over different source populations .
the results of such studies are presented in @xcite and @xcite .
similar studies including optical data are presented in @xcite .
a question that is closely related to the existence of correlated radio / gamma - ray variability is the correlation of their mean fluxes .
this question has been studied using single epoch surveys , but monitoring data can overcome several of the difficulties of those studies .
detailed discussions about this subject , and a look over different source classes is presented in @xcite and @xcite .
besides those results , we discovered a 120 - 150 d quasi periodic oscillation in a blazar , the shortest one ever identified @xcite .
we regularly contribute data for multiwavelength campaigns , which makes this programme a valuable result for the community .
variability is one the main characteristics of blazars , so we need to understand it in order to incorporate it in emission models . efforts to incorporate this characteristic in models are starting to appear in the literature ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , but they still require further development to provide testable observational predictions .
a detailed understanding of the multiwavelegth variability properties of blazars is also essential for our understanding of the significance of multiwavelength cross - correlations as explained below .
we can use the power spectral density ( psd ) , but there are other models that we do not consider here but that are discussed elsewhere and in this conference ( e.g. , * ? ? ?
even for the simple psd case , the characterization is complicated by the uneven sampling of the light curves .
we refer the reader to our recent paper @xcite , which describes the monte carlo approach used to for the psd fit .
we characterize the variability using a simple power law psd ( psd @xmath3 ) .
this model has a single parameter , the exponent @xmath4 .
using 4 years of radio data on 1593 sources we can find good constraints for @xmath4 in 238 sources .
we emphasize that the number of sources with good constraints is only a fraction of the total number of sources .
this is because in most cases we need longer time series , or because the amplitude of the variability is not large enough to discriminate between different values of @xmath4 .
the lesson to learn is that getting errors for the psd is as important as finding a best fit value , because it has a direct impact of the uncertainty in the estimation for the significance of a correlation as described in @xcite and briefly below . for the sources with constraints we found a range of values of @xmath4 , with the property that they cluster around 2.3 . using our large sample we can look at variations of the psd over different source populations ( e.g. , bl lac versus fsrq ) , but we did not find any variations of the psd between classes , probably due to the large uncertainties in @xmath4 that we expect to reduce with longer radio light curves ( max - moerbeck et al . in preparation ) .
with a characterization of the variability , we can look at the significance of cross - correlations .
but before , we look at how important is the variability for the to study of the significance .
figure 12 of @xcite illustrates the different characteristics of the variability for time series with @xmath5 and @xmath6 .
for @xmath7 we see that no clear flares are seen in the light curve , but for @xmath8 there appear clear flares similar to the ones observed in blazars .
the problem arises when we look at correlations between those independent light curves . in this case
( see figure 13 of * ? ? ?
* ) , we see that @xmath7 does not produce large amplitude correlations , but that @xmath8 produce large amplitude peaks in the cross - correlation .
the lesson is that , whenever we correlate @xmath9 noise time series we will see correlations , so the most important question is how significant they are . to evaluate the significance of a cross - correlation we use monte carlo simulations .
these simulation take into account the psd , the sampling and also the observational noise . in figure 28 of @xcite
we can see how the results of the significance of correlations vary when we change the model used for the simulated light curves .
this figure shows how important is to have a good characterization of the variability for the significance estimate .
having large errors in the psd power - law exponent is equivalent to having fuzzy significance contours .
unfortunately this is the situation with most sources , and should be indicated when presenting cross - correlation significance results , as it one of the largest sources of uncertainty in these estimates .
we studied the cross - correlation significance between 15 ghz and gamma - ray light curves .
we used 4 years of radio data and 3 years of fermi data for 86 sources @xcite .
of those 86 sources only 41 are variable , have no long term trends and thus a significance analysis is possible . of those 41 only 1
has a larger than @xmath10 significance and 2 more above @xmath11 significance , the level at which we expect to see one chance cross - correlation in a sample of 41 .
from this we find two main results : * only a minority of the sources have significant correlations * in all the cases with correlations , the radio emission lags the gamma - ray emission .
these results demonstrate very clearly the difficulty of measuring statistically robust multiwavelength correlations and the care needed when comparing light curves even when many years of data are used .
we use a simple model to estimate the location of the gamma - ray emission site . in this model
the emission is coming from the moving components we observe with vlbi .
the peak of the gamma - ray flare occurs at @xmath12 , and some time later @xmath13 , when the component propagates to optically thin part of the jet , beyond the radio core , we see the peak of the radio flare .
our cross - correlation result , gives @xmath14 , vlba measurements and variability constraints provide the geometry of the jet and speed of the moving components .
we have enough ancillary data for two sources . for ao 0235
+ 164 we consider the two peaks above @xmath10 seen in the correlation and we get a very weak constraint . for pks 1502 + 106 ,
the constraint is better and locates the gamma - ray emission a few parsecs away the central engine , but the errors are very large ( 22@xmath1515 pc for a conical jet , and 12@xmath159 pc for the collimated jet case ) .
the emission mechanism in the radio band is synchrotron radiation , and the variations can be due to changes in electron density or magnetic field , or to relativistic beaming effects . in this
case polarization observations provide data on the evolution of the magnetic fields in the emission region .
the ovro 40 meter telescope has a completely new receiver , all the way from the front to the backend , that has been working since june 2014 .
this new receiver covers a broader band and will soon provide polarized flux density light curves .
this programme provides the largest data base of single frequency light curves and has produced many results by itself and for the community .
we have used this data to learn about the variability of blazars in the radio band , and how it relates to the gamma - ray variations .
our main result is that only a minority of sources show significant correlations between the radio and gamma - ray bands , with radio lagging the gamma - ray emission . | since mid-2007 we have carried out a dedicated long - term monitoring programme at 15 ghz using the owens valley radio observatory 40 meter telescope ( ovro 40 m ) .
one of the main goals of this programme is to study the relation between the radio and gamma - ray emission in blazars and to use it as a tool to locate the site of high energy emission . using this large sample of objects we are able to characterize the radio variability , and study the significance of correlations between the radio and gamma - ray bands .
we find that the radio variability of many sources can be described using a simple power law power spectral density , and that when taking into account the red - noise characteristics of the light curves , cases with significant correlation are rare .
we note that while significant correlations are found in few individual objects , radio variations are most often delayed with respect to the gamma - ray variations .
this suggests that the gamma - ray emission originates upstream of the radio emission .
because strong flares in most known gamma - ray - loud blazars are infrequent , longer light curves are required to settle the issue of the strength of radio - gamma cross - correlations and establish confidently possible delays between the two .
for this reason continuous multiwavelength monitoring over a longer time period is essential for statistical tests of jet emission models . |
it is well known that sharp propagation of free waves along light cones in even - dimensional flat spacetimes , known as huygens property , is blurred by the presence of a potential .
physically , the spreading of waves inside the light cone is caused by the backscattering off the potential .
if the potential falls off exponentially or faster at spatial infinity , then the backscattered waves decay exponentially in time , while the long range potentials with an algebraic fall - off give rise to tails which decay polynomially in @xmath0 .
the precise description of these tails is an important issue in scattering theory .
there are two main approaches to this problem in the literature . on the one hand , there are mathematical results in the form of various decay estimates .
these results are rigorous , however they rarely give optimal decay rates inside the light cone and provide very poor information about the amplitudes of tails .
the notable exception is the work of strauss and tsutaya @xcite ( recently strengthened by szpak @xcite ) where the optimal pointwise decay estimate for the tail was proved in four dimensions .
unfortunately , to the best of our knowledge , there is no analogous result in higher dimensions . on the other hand ,
there are non - rigorous results in the physics literature based on perturbation theory .
the most complete work in this category was done by ching _
_ @xcite who derived first - order approximations of the tails for radial potentials .
although these results were originally formulated for partial waves in four dimensions , they can be easily translated to spherical waves in higher dimensions .
noticed that there are exceptional potentials for which the first - order tail vanishes , however they did not pursue their analysis to the second order , apart from giving some dimensional arguments .
the main purpose of this paper is to analyze the tails for such exceptional potentials in more detail .
one of the physical motivations behind our work stems from the fact that this kind of potentials arise in the study of linearized perturbations of higher even - dimensional schwarzschild black holes .
the behavior of tails on the schwarzschild background in well known in four dimensions ( see @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) , but not in higher even dimensions ( despite statements to the contrary in the literature @xcite ) .
although our analysis is restricted to the flat background , it sheds some light on the problem of tails on the black hole background because the properties of tails are to some extent independent of what happens in the central region .
the rest of the paper is organized as follows . in section
2 we construct the iterative scheme for the perturbation expansion of a spherically symmetric solution of the linear wave equation with a potential .
this scheme is applied in section 3 to derive the first and second - order approximations of the tails for radial potentials which fall off as pure inverse - power at infinity . in section 4
we discuss the modifications caused by subleading terms in the potential .
section 5 contains numerical evidence confirming the analytic formulae from sections 3 and 4 .
finally , in section 6 we give a heuristic argument to predict the behavior of tails outside schwarzschild black holes in higher even dimensions .
technical details of most calculations are given in the appendix . throughout the paper
we use the succinct notation and summation technics from the excellent book by graham _
in particular , we shall frequently use the following abbreviations @xmath1
we consider the wave equation with a potential in even - dimensional minkowski spacetime @xmath2 @xmath3 the prefactor @xmath4 is introduced for convenience - throughout the paper we assume that @xmath4 is small which allows us to use it as the perturbation parameter .
the precise assumptions about the fall - off of the potential will be formulated below .
we restrict attention to spherical symmetry , i.e. , we assume that @xmath5 and @xmath6 .
then , equation ( [ eqm ] ) becomes @xmath7 we are interested in the late - time behavior of @xmath8 for smooth compactly supported ( or exponentially localized ) initial data .
@xmath9 to determine the asymptotic behavior of solutions we define the perturbative expansion ( born series ) @xmath10 where @xmath11 satisfies initial data ( [ i d ] ) and all @xmath12 with @xmath13 have zero initial data . substituting this expansion into equation ( [ eqs ] )
we get the iterative scheme @xmath14 which can be solved recursively .
the zeroth - order solution is given by the general regular solution of the free radial wave equation which is a superposition of outgoing and ingoing waves @xcite @xmath15 where @xmath16 and @xmath17 , @xmath18 are the retarded and advanced times , respectively . here and in the following , instead of @xmath19
, we use the index @xmath20 defined by @xmath21 ( remember that we consider only _ odd _ space dimensions @xmath19 ) . note that for compactly supported initial data the generating function @xmath22 can be chosen to have compact support as well ( this condition determines @xmath22 uniquely ) . to solve equation ( [ scheme ] ) for
the higher - order perturbations we use the duhamel representation for the solution of the inhomogeneous equation @xmath23 with zero initial data @xmath24 where @xmath25 are legendre polynomials of degree @xmath20 and @xmath26 ( note that @xmath27 within the integration range ) .
this formula can be readily obtained by integrating out the angular variables in the standard formula @xmath28 where @xmath29 is the retarded green s function of the wave operator in @xmath30 dimensions ( see , for example , @xcite ) .
it is convenient to express ( [ a2 ] ) in terms of null coordinates @xmath31 and @xmath32 @xmath33 where now @xmath34 . using this representation
we can rewrite the iterative scheme ( [ scheme ] ) in the integral form @xmath35 this `` master '' equation will be applied below to evaluate the first two iterates for a special class of potentials .
it is natural to expect that for sufficiently small @xmath4 these iterates provide good approximations of the true solution .
in this section we consider the simple case ( below referred to as type i ) when the potential is _ exactly _ @xmath36 for @xmath37 greater than some @xmath38 .
we assume that @xmath39 .
the modifications caused by subleading corrections to the pure inverse - power decay of the potential will be discussed in section 4 .
we wish to evaluate the first iterate @xmath40 near timelike infinity , i.e , for @xmath41 and @xmath42 .
thanks to the fact that @xmath43 has compact support we may interchange the order of integration in ( [ iter ] ) and drop the advanced part of @xmath43 to obtain @xmath44 where we have substituted @xmath45 . plugging ( [ f1 ] ) into ( [ iter1 ] ) , after a long calculation ( see appendix a for the technical details ) , we get @xmath46^{\alpha-1+l } } \nonumber\\ & \times & \sum_{0 \leq n \leq \lfloor ( \alpha-2 ) / 2 \rfloor } ( -1)^{n } \frac { 2^{2n}(l+n ) ! } { n !
( 2l+2n+1 ) ! } \left ( - \frac { \alpha-2 } { 2 } - l - 1 \right)^{\underline{n } } \left ( \frac { \alpha-1 } { 2 } - l - 1 \right)^{\underline{n } } \nonumber\\ & \times & \sum_{n \leq m \leq n+l } ( -1)^{m } \left ( \begin{array}{c } l \\ m - n \end{array } \right ) \frac { \left ( - \frac { \alpha } { 2 } + 1 \right)^{\overline{m } } } { \left ( \frac { \alpha } { 2 } \right)^{\overline{m } } } \left ( \frac { r } { t-\eta } \right)^{2m}. \label{tailf1}\end{aligned}\ ] ] asymptotic expansion of ( [ tailf1 ] ) near timelike infinity yields the following first - order approximation of the tail @xmath47 \ , , \ ] ] where @xmath48 and @xmath49 in general @xmath50 and the tail decays as @xmath51 , however there are nongeneric initial data for which @xmath52 and then the tail decays as @xmath53 ; in particular this happens for time symmetric initial data for which @xmath22 is an odd function
remark 1 .
_ it is easy to check that if the function @xmath8 satisfies equation ( [ eqs ] ) , then the function @xmath54 satisfies the radial wave equation for the @xmath20th multipole @xmath55 the late - time tails for this equation were studied by ching _
@xcite who derived the formula equivalent to ( [ tail1 ] ) via the fourier transform methods .
it follows from ( [ tail1 ] ) that if @xmath56 is an odd integer satisfying @xmath57 , then @xmath40 vanishes identically due to factor @xmath58 in ( [ b ] ) and there is no ( polynomial ) tail whatsoever in the first order .
thus , in order to compute the tail in this exceptional case we need to go the second order of the perturbation expansion . using ( [ iter ] ) and proceeding as above
we get the second iterate @xmath59 where @xmath60 is the outgoing solution of the inhomogeneous equation @xmath61 in general @xmath62 is a sum of the solution of the homogeneous equation and the particular solution of the inhomogeneous equation . the homogeneous part has the form ( [ f1 ] ) ( with a different generating function than @xmath63 , but still compactly supported ) , thus for the same reason as above it gives no contribution to the tail .
the particular solution of the inhomogeneous equation ( [ nh ] ) reads @xmath64 where @xmath65 denotes the solution of the homogeneous equation with @xmath66 and the same generating function @xmath63 as in @xmath11 ( see ( [ f1 ] ) ) .
the formula ( [ f1nul ] ) can be easily derived by the method of undetermined coefficients ( we emphasize that this formula is valid _ only _ for odd @xmath56 satisfying @xmath57 ) . substituting ( [ f1nul ] ) into ( [ iter2 ] ) , after a long calculation ( see appendix a for the technical details ) , we obtain the following asymptotic behavior near timelike infinity @xmath67\,,\ ] ] where the coefficients @xmath68 and @xmath69 are defined in ( [ ab ] ) and @xmath70 here @xmath71 stands for the generalized hypergeometric function @xmath72 @xmath73 & @xmath74 & & & + @xmath75 & @xmath76 & @xmath77 & & + @xmath78 & @xmath79 & @xmath80 & @xmath81 & + @xmath74 & @xmath82 & @xmath83 & @xmath84 & @xmath85 + we remark that the behavior @xmath86 of the tail ( [ tail2 ] ) was conjectured before by ching _
@xcite on the basis of dimensional analysis .
in this section we analyze how the presence of subleading corrections to the pure inverse - power asymptotic behavior of the potential affects the results obtained in section 3 .
we restrict ourselves to the most interesting and common case ( below referred to as type ii ) when near infinity @xmath87 if @xmath88 , then the dominant behavior of the tail is of course the same as in ( [ tail1 ] ) : @xmath89 however , in the exceptional case , when @xmath90 , the situation is more delicate . as we showed above , in this case there is the second - order contribution to the tail given by ( [ tail2 ] ) @xmath91 in contrast to the type i case where the first - order tail vanishes identically , in the type ii case the subleading term in the potential produces the first - order contribution which is given by ( [ tail1 ] ) with @xmath56 replaced by @xmath92 : @xmath93 assuming that @xmath92 is not an odd integer @xmath94 ( otherwise one has to repeat the analysis for the next subleading term in the potential ) .
now , comparing the decay rates in ( [ gtail2 ] ) and ( [ tailf1p ] ) we conclude that the leading asymptotics of the tail is given by the first - order term @xmath95 if @xmath96 ( we call it subtype iia ) , but otherwise , _ i.e. _ for @xmath97 ( subtype iib ) , the second - order term @xmath98 is dominant for @xmath42 . 0.2 cm _ remark 2 . _ in the context of equation ( [ a3 ] ) a formula analogous to ( [ tailf1p ] ) was obtained by hod who studied tails in the presence of subleading terms in the potential ( see subgroup iiib in @xcite ) . however , hod s analysis , restricted to the first - order approximation , was inconclusive because , as we just have shown , without the second - order formula ( [ gtail2 ] ) one is not in position to make assertions about the dominant behavior of the tail .
in order to verify the above analytic predictions we solved numerically the initial value problem ( 4 - 5 ) for various potentials and initial data . our numerical algorithm is based on the method of lines with finite differencing in space and explicit fourth - order accurate runge - kutta time stepping . as was pointed out in @xcite , a reliable numerical computation of tails
requires high - order finite - difference schemes , since otherwise the ghost potentials generated by discretization errors produce artificial tails which might mask the genuine behavior .
the minimal order of spatial finite - difference operators depends on the fall - off of the potential for the cases presented below the fourth - order accuracy was sufficient , but for the faster decaying potentials a higher - order accuracy is needed . to eliminate high - frequency numerical instabilities we added a small amount of kreiss - oliger artificial dissipation all computations
were performed using quadruple precision which was essential in suppressing round - off errors at late times .
the numerical results presented here were produced for initial data of the form @xmath99 as follows from ( [ f1 ] ) the generating function for these data is @xmath100 we considered the following potentials v(r)= & ) + ( 1 + ) & , for various values of @xmath56 and @xmath101 .
the regularizing factor @xmath102 introduces exponentially decaying corrections to the pure inverse - power behavior at infinity but such corrections do not affect the polynomial tails . the numerical verification of the formulae ( [ tail1 ] ) , ( [ tailf1p ] ) , and ( [ tail2 ] ) is shown in tables ii and iii .
the observed decay rates agree perfectly with analytic predictions , while small errors in the amplitudes are due to ( neglected ) higher - order terms in the perturbation expansion .
.the generic case : numerical verification of the analytic formula ( [ tail1 ] ) for the potential ( 31a ) ( @xmath103 ) and initial data ( [ idn ] ) . comparing the second column of this table ( corresponding to @xmath104 ) with the last column of table iii
one can see the discontinuity of the decay rate at @xmath105 ( for @xmath106 and @xmath107 ) . [
cols="^,^,^,^,^,^,^,^ " , ]
consider the evolution of the massless scalar field outside the @xmath30 dimensional schwarzschild black hole @xmath108 where @xmath109 is the round metric on the unit sphere @xmath110 and @xmath111 is odd .
here we use units in which the horizon radius is at @xmath112 . introducing the tortoise
coordinate @xmath113 , defined by @xmath114 , and decomposing the scalar field into multipoles , one obtains the following reduced wave equation for the @xmath115th multipole @xcite @xmath116 note that ( [ eqsch ] ) is the @xmath117 dimensional wave equation on the whole axis @xmath118 . for large positive @xmath113 we have @xmath119 which implies that @xmath120 with @xmath121 for large negative @xmath113 ( near the horizon ) the potential is exponentially small
, so one expects that the backscattering off the left edge of the potential can be neglected .
if so , the decay rate ( but not the amplitude ! ) should follow from the analysis of section 4 . comparing equation ( [ eqsch ] ) for large positive @xmath113 to equation ( [ a3 ] ) with the potential ( [ case2 ] ) and using ( [ expv ] ) we find that @xmath122 and the potential @xmath123 is of the subtype iia with @xmath124 and @xmath125 . thus , applying ( [ tailf1p ] ) we get the first - order tail @xmath126 _ remark 3 .
_ late - time tails outside higher dimensional schwarzschild black holes were studied in @xcite , however in the even - dimensional case the reasoning presented there is not correct , even though the result agrees with ( [ tails ] ) .
the reason is that the analysis of @xcite is based on the application of ching _ et al .
_ conjecture about the decay of the second - order tail @xmath127 which for @xmath122 and @xmath124 gives @xmath128 .
unfortunately , this conjecture does not apply to the problem at hand . for @xmath129
this is evident because the leading term in @xmath123 ( proportional to @xmath130 ) vanishes ( since by ( [ wsp ] ) @xmath131 ) , while the subleading term ( proportional to @xmath132 ) is of generic type . for @xmath133
this follows from the fact that the potential is of the subtype iia .
thus , for all @xmath134 the dominant ( first - order ) contribution to the tail comes from the subleading term in the potential .
the agreement of the decay rate obtained in @xcite with ( [ tails ] ) is accidental and due to the fact that the subdominant term in ( [ expv ] ) ( not considered in @xcite ) is on a borderline between subtypes iia and iib .
admittedly , the handwaving argument leading to ( [ tails ] ) is far from satisfactory .
unfortunately , we have not been able to carry over the analysis from sections 2 - 4 in the case of equation ( [ eqsch ] ) .
there are two difficulties in this respect .
first , in contrast to the spherical case , huygens principle is not valid for the free wave equation in @xmath117 dimensions .
second , there is no natural small parameter in the problem . in
the impressive tour de force work @xcite barack showed how to overcome these difficulties for a restricted class of initial data in four dimensions .
it would be interesting to generalize barack s approach to higher even - dimensional schwarzschild spacetimes .
0.3 cm * acknowledgments : * pb thanks nikodem szpak for helpful discussions and leor barack for clarifying some details of the paper @xcite .
ar thanks prof .
bernd brgmann for hospitality in his group at fsu jena , where a part of this work was done .
this research was supported in part by the mnii grant 1po3b01229 and grant 189/6 .
pr ue/2007/7 . throughout the appendix we use the notation of @xcite
in which the square bracket around a logical expression returns a value @xmath73 if the expression is true and a value @xmath135 if the expression is false : @xmath136 = \left\ { \begin{array}{ccl } 1 & \mbox{if } & condition = \mbox{true } \\ 0 & \mbox{if } & condition = \mbox{false } \end{array } \right.\nonumber\ ] ] in order to derive the asymptotic behavior of the iterates ( [ iter1 ] ) and ( [ iter2 ] ) near timelike infinity ( fixed @xmath37 and @xmath137 ) we need to evaluate the following expression @xmath138 where @xmath139 and @xmath140 from ( [ f1 ] ) and ( [ iter1 ] ) we have @xmath141 and from ( [ iter2 ] ) and ( [ f1nul ] ) we have @xmath142 since @xmath143 has compact support , it is advantageous to begin with integration by parts @xmath144 for @xmath145 as defined in ( [ mu ] ) and for any function @xmath146 the following identity holds @xmath147 hence @xmath148 the sum over @xmath149 can be evaluated explicitly @xmath150 let us define @xmath151 changing the integration variable from @xmath152 to @xmath145 and integrating by parts , we get @xmath153^{\beta-1+l } } \int \limits_{-1}^{+1 } d\mu \ , p_{l } ( \mu ) \frac { d^j } { d\mu^j } \left [ ( 1 - \mu^2)^j \left ( 1 - \frac { r } { t-\eta } \mu \right)^{\beta-2+l - j } \right ] . \label{i}\ ] ] using the identity @xcite @xmath154 and expanding @xmath155 $ ] in taylor series we get @xmath156^{\beta-1+l } } \ , \sum_{n=0}^{\beta-2+l } ( j+n)^{\underline{j } } \ , \int \limits_{-1}^{+1 } d\mu \ , p_{l } ( \mu ) \mu^{n } \nonumber\\ & \times & \sum_{m=0}^{\lfloor ( j+n)/2 \rfloor } \left ( \begin{array}{c } j \\
m \end{array } \right ) \left ( \begin{array}{c } \beta-2+l - j
\\ j+n-2 m \end{array } \right ) ( -1)^{j+n+m } \left ( \frac { r } { t-\eta } \right)^{j+n-2 m } \nonumber\\ & = & \frac { r^{l+1 } ( t-\eta)^{\beta-2+l - l } } { \left [ ( t-\eta)^2 - r^2 \right]^{\beta-1+l } } \ , \sum_{n=0}^{\lfloor ( \beta-2+l - l ) / 2 \rfloor } ( j+l+2n)^{\underline{j } } \ , \int \limits_{-1}^{+1 } d\mu \ , p_{l } ( \mu ) \mu^{l+2n } \nonumber\\ & \times & \sum_{m=0}^{\lfloor ( j+l+2n)/2 \rfloor } \left ( \begin{array}{c } j \\ m \end{array }
\right ) \left ( \begin{array}{c } \beta-2+l - j \\
j+l+2n-2 m \end{array } \right ) ( -1)^{l+m } \left ( \frac { r } { t-\eta } \right)^{2j+2n-2 m } \nonumber\\ & = & \frac { r^{l+1 } ( t-\eta)^{\beta-2+l - l } } { \left [ ( t-\eta)^2 - r^2 \right]^{\beta-1+l } } \ , \sum_{n=0}^{\lfloor ( \beta-2+l - l ) / 2 \rfloor } ( j+l+2n)^{\underline{j } } \ , \ , 2^{l+1 } \frac { ( l+2n ) ! ( l+n ) ! } { n ! ( 2l+2n+1 ) ! }
\nonumber\\ & \times & \sum_{m=0}^{\lfloor ( j+l+2n)/2 \rfloor } \left ( \begin{array}{c } j \\
m \end{array } \right ) \left ( \begin{array}{c } \beta-2+l - j
\\ j+l+2n-2 m \end{array } \right ) ( -1)^{l+m } \left ( \frac { r } { t-\eta } \right)^{2j+2n-2 m } \label{i(2)}\,.\end{aligned}\ ] ] collecting the results of ( [ k - sum ] , [ i ] , [ i(2 ) ] ) and plugging them into ( [ master(2 ) ] ) we get @xmath157^{\beta-1+l } } \sum_{n=0}^{\lfloor ( \beta-2+l - l ) / 2 \rfloor } \frac { ( l+2n ) ! ( l+n ) ! } { n ! ( 2l+2n+1 ) ! } ( -1)^{l+l } l ! \ , s ( \beta , l ) , \label{master(3)}\ ] ] where @xmath158 to evaluate the first iterate @xmath40
we apply the formula ( [ master(3 ) ] ) with @xmath159 and @xmath160 .
then @xmath161 where we shifted the summation index @xmath162 .
next , we interchange the order of summation according to @xmath163 [ j \leq l ] [ m - n \leq j ] [ j \leq 2m+l ] \nonumber\\ & \leftrightarrow & [ -\frac { l}{2 } \leq m < 0 ] [ 0 \leq j \leq l+2 m ] \ , + \ , [ 0 \leq m < n ] [ 0 \leq j \leq
l ] \ , + \ ,
[ n \leq m \leq l+n ] [ m - n \leq j \leq l ] \,,\nonumber\end{aligned}\ ] ] and convert the sum over @xmath115 into a generalized hypergeometric function @xcite . defining @xmath164
we see that @xmath165 iff @xmath166 , thus the sums for @xmath167 $ ] and @xmath168 $ ] do not contribute to ( [ si ] ) and we are left with @xmath169 where we shifted the summation index @xmath170 . defining @xmath171 , we see that @xmath172 and @xmath173 hence @xmath174 where in the last equation we used the identity @xmath175 substituting @xmath176 and @xmath177 into ( [ si(2 ) ] ) we get @xmath178 the last equation can be still simplified due to the identity @xmath179 we have @xmath180 and @xmath181 so finally @xmath182 plugging ( [ si(4 ) ] ) into ( [ master(3 ) ] ) with @xmath159 and @xmath160 we get the expression ( [ tailf1 ] ) .
the calculation in the second order ( @xmath183 and @xmath184 ) is only a slight modification of what we have already done in the first order .
following the same steps which led us from ( [ si ] ) to ( [ si(3 ) ] ) we get @xmath185 the last equation can be simplified due to the identity @xmath186 which for @xmath160 reduces to ( [ gammasid ] ) .
we have @xmath187 and @xmath188 hence @xmath189 plugging ( [ sii(4 ) ] ) into ( [ master(3 ) ] ) we get @xmath190^{2\alpha-3+l } } \\ & \times & \sum_{n=0}^{\alpha-2}\ ! ( -1)^{n } \frac { 2^{2n}(l+n ) ! }
{ n ! ( 2l+2n+1 ) ! } \left ( -\alpha+1-l \right)^{\underline{n } } \left ( \alpha - \frac { 3 } { 2 } - l + q \right)^{\underline{n } } \ , \sum_{m = n}^ { n+l-1-q } \!(-1)^{m } \left ( \begin{array}{c } l-1-q \\ m - n \end{array }
\right ) \frac { \left ( -\alpha + 2 \right)^{\overline{m } } } { \left ( \alpha + q \right)^{\overline{m } } } \left ( \frac { r } { t-\eta } \right)^{2m}.\nonumber\end{aligned}\ ] ] substituting this into ( [ a5 ] ) and expanding in @xmath0 we have @xmath191 \nonumber\\ & \times & \left ( \sum_{q=0}^{l-(\alpha-1)/2 } ( -1)^{q } ( l - p)^{\underline{q } } \,\ , \frac { 2^q \left ( \alpha/2 \right)^{\overline{q } } } { \alpha^{\overline{q } } } \left ( \alpha - \frac { 5 } { 2 } \right)^{\underline{l-1-q } } \left ( \alpha - 2 + l \right)^{\underline{l-1-q } } ( 2\alpha-3)^{\overline{1+q } } \right),\end{aligned}\ ] ] with @xmath68 and @xmath69 defined in ( [ ab ] ) . converting the sum over @xmath192 into the generalized hypergeometric function we get ( [ tail2 ] ) . | we consider the late - time tails of spherical waves propagating on even - dimensional minkowski spacetime under the influence of a long range radial potential . we show that in six and higher even dimensions there exist exceptional potentials for which the tail has an anomalously small amplitude and fast decay . along the way we clarify and amend some confounding arguments and statements in the literature of the subject . |
many real systems can be viewed as a combination of slow and fast motions which leads to complicated double scale equations . already in the 19th century in applications to celestial mechanics it was well understood ( though without rigorous justification ) that a good approximation of the slow motion can be obtained by averaging its parameters in fast variables .
later , averaging methods were applied in signal processing and , rather recently , to model climate weather interactions ( see @xcite , @xcite , @xcite and @xcite ) .
the classical setup of averaging justified rigorously in @xcite presumes that the fast motion does not depend on the slow one and most of the work on averaging treats this case only . on the other hand , in real systems
both slow and fast motions depend on each other which leads to the more difficult fully coupled case which we study here .
this setup emerges , in particular , in perturbations of hamiltonian systems which leads to fast motions on manifolds of constant energy and slow motions across them . in this work
we consider a system of differential equations for @xmath6 and @xmath7 @xmath8 with initial conditions @xmath9 on the product @xmath10 where @xmath11 is a compact @xmath12-dimensional @xmath13 riemannian manifold and @xmath14 are smooth in @xmath15 families of bounded vector fields on @xmath16 and on @xmath17 respectively , so that @xmath18 serves as a parameter for @xmath19 and @xmath20 for @xmath21 .
the solutions of ( [ 1.1.1 ] ) determine the flow of diffeomorphisms @xmath22 on @xmath10 acting by @xmath23 .
taking @xmath24 we arrive at the flow @xmath25 acting by @xmath26 where @xmath27 is another family of flows given by @xmath28 with @xmath29 being the solution of @xmath30 it is natural to view the flow @xmath31 as describing an idealised physical system where parameters @xmath32 are assumed to be constants ( integrals ) of motion while the perturbed flow @xmath22 is regarded as describing a real system where evolution of these parameters is also taken into consideration .
essentially , the proofs of this paper work also in the slightly more general case when @xmath19 and @xmath21 in ( [ 1.1.1 ] ) together with their derivatives depend lipschitz continuously on @xmath33 ( cf .
@xcite ) but in order to simplify notations and estimates we do not consider this generalisation here .
assume that the limit @xmath34 exists and it is the same for `` many '' @xmath35s .
for instance , suppose that @xmath36 is an ergodic invariant measure of the flow @xmath27 then the limit ( [ 1.1.3 ] ) exists for @xmath37almost all @xmath18 and is equal to @xmath38 if @xmath39 does not , in fact , depend on @xmath20 then @xmath40 and @xmath41 are also independent of @xmath20 and we arrive at the classical uncoupled setup . in this case
lipschitz continuity of @xmath19 implies already that @xmath42 is also lipshitz continuous in @xmath20 , and so there exists a unique solution @xmath43 of the averaged equation @xmath44 in this case the standard averaging principle says ( see @xcite ) that for @xmath45-almost all @xmath18 , @xmath46 as the main motivation for the study of averaging is the setup of perturbations described above we have to deal in real problems with the fully coupled system ( [ 1.1.1 ] ) which only in very special situations can be reduced by some change of variables to a much easier uncoupled case where the fast motion does not depend on the slow one .
observe that in the general case ( [ 1.1.1 ] ) the averaged vector field @xmath47 in ( [ 1.1.3 ] ) may even not be continuous in @xmath20 , let alone lipschitz , and so ( [ 1.1.4 ] ) may have many solutions or none at all . moreover
, there may exist no natural well dependent on @xmath48 family of invariant measures @xmath36 since dynamical systems @xmath27 may have rather different properties for different @xmath20 s .
even when all measures @xmath36 are the same the averaging principle often does not hold true in the form ( [ 1.1.5 ] ) , for instance , in the presence of resonances ( see @xcite and @xcite ) .
thus even basic results on approximation of the slow motion by the averaged one in the fully coupled case can not be taken for granted and they should be formulated in a different way requiring usually stronger and more specific assumptions .
if convergence in ( [ 1.1.3 ] ) is uniform in @xmath20 and @xmath18 then ( see , for instance , @xcite ) any limit point @xmath49 as @xmath50 of @xmath51 is a solution of the averaged equation @xmath52 it is known that the limit in ( [ 1.1.3 ] ) is uniform in @xmath18 if and only if the flow @xmath27 on @xmath11 is uniquely ergodic , i.e. it possesses a unique invariant measure , which occurs rather rarely .
thus , the uniform convergence in ( [ 1.1.3 ] ) assumption is too restrictive and excludes many interesting cases . probably , the first relatively general result on fully coupled averaging is due to anosov @xcite ( see also @xcite and @xcite ) .
relying on the liouville theorem he showed that if each flow @xmath27 preserves a probability measure @xmath36 on @xmath11 having a @xmath53 dependent on @xmath20 density with respect to the riemannian volume @xmath54 on @xmath11 and @xmath36 is ergodic for lebesgue almost all @xmath20 then for any @xmath55 @xmath56 where mes is the product of @xmath54 and the lebesgue measure in a relatively compact domain @xmath57 .
an example in appendix to @xcite shows that , in general , this convergence in measure can not be strengthened to the convergence for almost all initial conditions and , moreover , in this example the convergence ( [ 1.1.5 ] ) does not hold true for any initial condition from a large open domain .
such examples exist due to the presence of resonances , more specifically to the `` capture into resonance '' phenomenon , which is rather well understood in perturbations of integrable hamiltonian systems .
resonances lead there to the wealth of ergodic invariant measures and to different time and space averaging .
it turns out ( see @xcite ) that wealth of ergodic invariant measures with nice properties ( such as gibbs measures ) for axiom a and expanding dynamical systems also yields in the fully coupled averaging setup with the latter fast motions examples of nonconvergence as @xmath50 for large sets of initial conditions ( see remark [ rem1.2.12 ] ) . in hamiltonian systems , which are a classical object for applications of averaging methods ,
the whole space is fibered into manifolds of constant energy .
for some mechanical systems these manifolds have negative curvature with respect to the natural metric and their motion is described by geodesic flows there .
hyperbolic hamiltonian systems were discussed , for instance , in @xcite and a specific example of a particle in a magnetic field leading to such systems was considered recently in @xcite .
of course , these lead to hamiltonian systems which are far from integrable .
such situations fall in our framework and they are among main motivations for this work .
this suggests to consider the equation ( [ 1.1.1 ] ) on a ( locally trivial ) fiber bundle @xmath58 with a base @xmath59 being an open subset in a riemannian manifold @xmath60 and fibers @xmath61 being diffeomorphic compact riemannian manifolds ( see @xcite ) . on the other hand , @xmath62 has a local product structure and if @xmath63 is bounded then the slow motion stays in one chart during time intervals of order @xmath64 with @xmath65 small enough .
hence , studying behavior of solutions of ( [ 1.1.1 ] ) on each such time interval separately we come back to the product space @xmath10 setup and will only have to piece results together to see the picture on a larger time interval of length @xmath66 we assume in the first part of this work that @xmath39 is @xmath13 in @xmath20 and @xmath18 and that for each @xmath20 in a closure of a relatively compact domain @xmath67 the flow @xmath27 is anosov or , more generally , axiom a in a neighborhood of an attractor @xmath68 let @xmath69 be the sinai - ruelle - bowen ( srb ) invariant measure of @xmath27 on @xmath70 and set @xmath71 it is known ( see @xcite ) that the vector field @xmath47 is lipschitz continuous in @xmath72 and so the averaged equations ( [ 1.1.4 ] ) and ( [ 1.1.6 ] ) have unique solutions @xmath73 and @xmath74 still , in general , the measures @xmath69 are singular with respect to the riemannian volume on @xmath11 , and so the method of @xcite can not be applied here .
we proved in @xcite that , nevertheless , ( [ 1.1.7 ] ) still holds true in this case , as well , and , moreover , the measure in ( [ 1.1.7 ] ) can be estimated by @xmath75 with some @xmath76 .
the convergence ( [ 1.1.7 ] ) itself without an exponential estimate can be proved by another method ( see @xcite ) which can be applied also to some partially hyperbolic fast motions .
an extension of the averaging principle in the sense of convergence of young measures is discussed in section [ sec1.11 ] .
once the convergence of @xmath51 to @xmath77 as @xmath50 is established it is interesting to study the asymptotic behavior of the normalized error @xmath78.\ ] ] namely , in our situation it is natural to study the distributions @xmath79 as @xmath50 where @xmath54 is the normalized riemannian volume on @xmath11 and @xmath80 is a borel subset in the space @xmath81 of continuous paths @xmath82 $ ] on @xmath83 we will obtain in this work large deviations bounds for @xmath84 which will give , in particular , the result from @xcite saying that @xmath85 exponentially fast in @xmath5 where @xmath86 is the uniform norm on @xmath87 however , the main goal of this work is not to provide another derivation of ( [ 1.1.9 ] ) but to obtain precise upper and lower large deviations bounds which not only estimate measure of sets of initial conditions for which the slow motion @xmath88 exhibits substantially different behavior than the averaged one @xmath89 but also enable us to go further and to investigate much longer exponential in @xmath5 time behavior of @xmath88 .
namely , we will be able to study exits of the slow motion from a neighborhood of an attractor of the averaged one and transitions of @xmath88 between basins of attractors of @xmath89 .
such evolution , which becomes visible only on much longer than @xmath5 time scales , is usually called adiabatic in the framework of averaging . in the simpler case when the fast motion does not depend on the slow one
such results were discussed in @xcite .
still , even in this uncoupled situation descriptions of transitions of the slow motion between attractors of the averaged one were not justified rigorously both in the markov processes case of @xcite and in the dynamical systems case of @xcite .
extending these technique to three scale equations may exhibit stochastic resonance type phenomena producing a nearly periodic motion of the slowest motion which is described in section [ sec1.10 ] below .
these problems seem to be important in the study of climate weather interactions and they were discussed in @xcite and @xcite in the framework of a model describing transitions between steady climatic states with weather evolving as a fast chaotic system and climate playing the role of the slow motion . such `` very long '' time description of the slow motion is usually impossible in the traditional averaging setup which deals with perturbations of integrable hamiltonian systems . in the fully coupled situation we can not work just with one hyperbolic flow but have to consider continuously changing fast motions which requires a special technique . in particular , the full flow @xmath22 on @xmath10 defined above and viewed as a small perturbation of the partially hyperbolic system @xmath31 plays an important role in our considerations .
it is somewhat surprising that the `` very long time '' behavior of the slow motion which requires certain `` markov property type '' arguments still can be described in the fully coupled setup which involves continuously changing fast hyperbolic motions .
it turns out that the perturbed system still possesses semi - invariant expanding cones and foliations and a certain volume lemma type result on expanding leaves plays an important role in our argument for transition from small time were perturbation techniques still works to the long and `` very long '' time estimates .
it is plausible that moderate deviations type results can be proved for @xmath90 when @xmath91 and that the distribution of @xmath92 in @xmath18 converges to the distribution of a gaussian diffusion process in @xmath16 .
still , this requires somewhat different methods and it will not be discussed here . in this regard
we can mention limit theorems obtained in @xcite for a system of two heavy and light particles which leads to an averaging setup for a billiard flow . for the simpler case when @xmath21 does not depend on @xmath72 i.e. when
all flows @xmath93 are the same , the moderate deviations and gaussian approximations results were obtained previously in @xcite . related results in this uncoupled situation concerning hasselmann s nonlinear ( strong ) diffusion approximation of the slow motion @xmath2
were obtained in @xcite .
we consider also the discrete time case where ( [ 1.1.1 ] ) is replaced by difference equations for sequences @xmath94 and
@xmath95 so that @xmath96 where @xmath97 is lipschitz in both variables and the maps @xmath98 are smooth and depend smoothly on the parameter @xmath48 . introducing the map @xmath99 we can also view this setup as a perturbation of the map @xmath100 describing an ideal system where parameters @xmath48 do not change . assuming that @xmath101 are @xmath13 depending on @xmath20 families of either @xmath13 expanding transformations or @xmath13 axiom a diffeomorphisms in a neighborhood of an attractor @xmath70 we will derive large deviations estimates for the difference @xmath102 where @xmath103 solves the equation @xmath104 where @xmath105 and @xmath69 is the corresponding srb invariant measure of @xmath106 on @xmath70 .
the discrete time results are obtained , essentially , by simplifications of the corresponding arguments in the continuous time case which enable us to describe `` very long '' time behavior of the slow motion also in the discrete time case .
since our methods work not only for fast motions being axiom a diffeomorphisms but also when they are expanding transformations we can construct simple examples satisfying conditions of our theorems and exhibiting corresponding effects . in particular , we produce in section [ sec1.9 ] computational examples which demonstrate transitions of the slow motion between neighborhoods of attractors of the averaged system . a series of related results for the case when ordinary differential equations in ( [ 1.1.1 ] ) are replaced by fully coupled stochastic differential equations appeared in @xcite , @xcite@xcite , @xcite , and @xcite .
hasselmann s nonlinear ( strong ) diffusion approximation of the slow motion in the fully coupled stochastic differential equations setup was justified in @xcite . when the fast process does not depend on the slow one
such results were obtained in @xcite , @xcite , and @xcite . especially relevant for our results here
is @xcite and we employ some elements of the probabilistic strategy from this paper .
still , the methods there are quite different from ours and they are based heavily , first , on the markov property of processes emerging there and , secondly , on uniformity and nondegeneracy of the fast diffusion term assumptions which can not be satisfied in our circumstances as our deterministic fast motions are very degenerate from this point of view .
note that the proof in @xcite contains a vicious cycle and substantial gaps which recently were essentially fixed in @xcite .
some of the dynamical systems technique here resembles @xcite but the dependence of the fast motion on the slow one complicates the analysis substantially and requires additional machinery . a series of results on cramer s type asymptotics for fully coupled averaging with axiom a diffeomorphisms as fast motions appeared recently in @xcite@xcite .
observe that the methods there do not work for continuous time axiom a dynamical systems considered here , they can not lead , in principle , to the standard large deviations estimates of our work and they deal with deviations of @xmath2 from the averaged motion only at the last moment and not of its whole path . various limit theorems for the difference equations setup ( [ 1.1.10 ] ) with partially hyperbolic fast motions were obtained recently in @xcite and @xcite .
the study of deviations from the averaged motion in the fully coupled case seems to be quite important for applications , especially , from phenomenological point of view .
in addition to perturbations of hamiltonian systems mentioned above there are many non hamiltonian systems which are naturally to consider from the beginning as a combination of fast and slow motions .
for instance , hasselmann @xcite based his model of weather climate interaction on the assumption that weather is a fast chaotic motion depending on climate as a slow motion which differs from the corresponding averaged motion mainly by a diffusion term .
though , as shown in @xcite , @xcite and @xcite , this diffusion error term does not help in the study of large deviations which are responsible for rare transitions of the slow motion between attractors of the averaged one , the latter phenomenon can be described in our framework and it seems to be important in certain models of climate fluctuations ( see @xcite and @xcite ) .
very slow nearly periodic motions appearing in the stochastic resonance framework considered in section [ sec1.10 ] may also fit into this subject in the discussion on `` ice ages '' .
of course , it is hard to believe that real world chaotic systems can be described precisely by an anosov or axiom a flow but one may take comfort in the chaotic hypothesis @xcite : `` a chaotic mechanical system can be regarded for practical purposes as a topologically mixing anosov system '' . many real systems can be viewed as a combination of slow and fast motions which leads to complicated double scale equations .
already in the 19th century in applications to celestial mechanics it was well understood ( though without rigorous justification ) that a good approximation of the slow motion can be obtained by averaging its parameters in fast variables .
later , averaging methods were applied in signal processing and , rather recently , to model climate weather interactions ( see @xcite , @xcite , @xcite and @xcite ) .
the classical setup of averaging justified rigorously in @xcite presumes that the fast motion does not depend on the slow one and most of the work on averaging treats this case only . on the other hand , in real systems
both slow and fast motions depend on each other which leads to the more difficult fully coupled case which we study here .
this setup emerges , in particular , in perturbations of hamiltonian systems which leads to fast motions on manifolds of constant energy and slow motions across them .
it is natural to view double scale models as describing physical systems considered as perturbations of an idealized one which depends on parameters @xmath2466 assumed to be constants ( integrals ) of motion . in part [ part2 ]
we suppose that the evolution of this idealized sistem is described by certain family of markov processes @xmath2467 on a separable metric space @xmath11 . in the perturbed system parameters start changing slowly in time and we assume that the corresponding slow motion @xmath2468 is described by an ordinary differential equations in @xmath16 having the form @xmath2469 where @xmath2470 is lipschitz continuous and the fast motion @xmath2471 evolves on @xmath11 , it depends , in general , on the slow one and tends to @xmath2472 as @xmath50 .
usually , @xmath2473 is determined by certain equations , in general , coupled with ( [ 2.1.1 ] ) which means that their coefficients depend on the slow motion @xmath2474 .
assume that a nonrandom limit @xmath2475 exists in some sense , it `` essentially '' does not depend on @xmath18 and it depends lipschitz continuously on @xmath20 .
then there exists a unique solution @xmath2476 of the averaged equation @xmath2477 the averaging principle suggests that often @xmath2478 in some sense .
if unperturbed motions @xmath2479 do not depend on the slow variables @xmath20 and @xmath2479 then the averaged principle holds true under quite general circumstances but when the fast motion depends on the slow one ( coupled case ) the situation becomes more complicated and approximation of @xmath2480 by @xmath2481 in the weak or the average sense was justified under some conditions in@xcite and @xcite .
an extension of the averaging principle in the sense of convergence of young measures is discussed in section [ sec2.9 ] below . in this work
we are interested in large deviations bounds for probabilities that the time changed slow motion @xmath2482 belongs to various sets of curves which leads , in particular , to exponential bounds of the form @xmath2483 where @xmath2393 satisfies @xmath2484 when the fast motion do not depend on the slow one such results were obtained in @xcite and @xcite but the coupled case was dealt with much later in @xcite though ( as we indicated this to the author ) the proof there contained a vicious circle and substantial gaps which , essentially , were fixed recently in @xcite .
still , @xcite is rather difficult to follow and we find it useful to provide a precise and consistent exposition of this important result which also deals with a more general case including fast motions being random evolutions whose extreme partial cases are diffusions and finite markov chains with continuous time .
moreover , we go beyond bounded time large deviations and describe the adiabatic behaviour of the slow motion @xmath88 on exponentially large in @xmath5 time intervals such as its exits from a domain of attraction and transitions between attractors of the averaged system ( [ 2.1.6 ] ) .
we observe that essentially the same proof yields the same results for a bit more general case when both @xmath19 in [ 2.1.1 ] and the coefficients of the random evolutions in the next section depend also lipschitz continuously on @xmath33 .
we consider also the discrete time case where ( [ 2.1.1 ] ) is replaced by a difference equation of the form @xmath2485 where @xmath368 is the same as in ( [ 2.1.1 ] ) and the fast motion @xmath2486 is a perturbation of a family @xmath2487 of markov chains parametrized by @xmath48 . for somewhat less general discrete time situation
large diviations bounds were obtained in @xcite by a simpler approach but in our more general situation we can rely only on methods similar to the continuous time case . moreover , unlike @xcite we go farther and study also very long time `` adiabatic '' behaviour of the slow motion similar to the continuous time case and illustrate some of the results by computer simulations for simple models .
the strategy and many of arguments in part [ part2 ] are rather similar to part [ part1 ] where deterministic chaotic fast motions such as anosov and axiom a systems were considered . still , in view of the heavy dynamical systems background and machinery part [ part1 ] is hardly accessible for most of probabilists . by this reason we give full proofs here refering to part [ part1 ] only for proofs of some general results on large deviations , rate functionals and some others which do not rely on the specific dynamical systems setup .
let @xmath107 be a @xmath13 flow on a compact riemannian manifold @xmath11 given by a differential equation @xmath108 a compact @xmath109invariant set @xmath110 is called hyperbolic if there exists @xmath111 and the splitting @xmath112 into the continuous subbundles @xmath113 of the tangent bundle @xmath114 restricted to @xmath115 the splitting is invariant with respect to the differential @xmath116 of @xmath117 @xmath118 is the one dimensional subbundle generated by the vector field @xmath21 , and there is @xmath119 such that for all @xmath120 and @xmath121 @xmath122 a hyperbolic set @xmath123 is said to be basic hyperbolic if the periodic orbits of @xmath124 are dense in @xmath115 @xmath124 is topologically transitive , and there exists an open set @xmath125 with @xmath126 such a @xmath123 is called a basic hyperbolic attractor if for some open set @xmath59 and @xmath127 @xmath128 where @xmath129 denotes the closure of @xmath130 if @xmath131 then @xmath107 is called an anosov flow . [ ass1.2.1 ] the family @xmath132 in ( [ 1.1.2 ] ) consists of @xmath13 vector fields on a compact @xmath12-dimensional riemannian manifold @xmath11 with uniform @xmath13 dependence on the parameter @xmath20 belonging to a neighborhood of the closure @xmath133 of a relatively compact open connected set @xmath57 .
each flow @xmath134 on @xmath11 given by @xmath135 possesses a basic hyperbolic attractor @xmath70 with a splitting @xmath136 satisfying ( [ 1.2.2 ] ) with the same @xmath111 and there exists an open set @xmath137 and @xmath119 such that @xmath138 let @xmath139 be the absolute value of the jacobian of the linear map @xmath140 with respect to the riemannian inner products and set @xmath141 the function @xmath142 is known to be h " older continuous in @xmath143 since the subbundles @xmath144 are h " older continuous ( see @xcite and @xcite ) , and @xmath142 is @xmath53 in @xmath20 ( see @xcite ) .
let @xmath145 satisfy ( [ 1.2.4 ] ) and set @xmath146\}.$ ] a set @xmath147 is called @xmath148separated for the flow
@xmath106 if @xmath149 , @xmath150 imply @xmath151 for some @xmath152 $ ] , where @xmath153 is the distance function on @xmath154 for each continuous function @xmath155 on @xmath145 set @xmath156 @xmath157 if @xmath158 and @xmath159 the latter is monotone in @xmath160 and so the limit @xmath161 exists and it is called the topological pressure of @xmath155 for the flow @xmath162 let @xmath163 denotes the space of @xmath164invariant probability measures on @xmath70 then ( see , for instance , @xcite ) the following variational principle @xmath165 holds true where @xmath166 is the kolmogorov
sinai entropy of the time - one map @xmath167 with respect to @xmath168 if @xmath169 is a h " older continuous function on @xmath70 then there exists a unique @xmath164invariant measure @xmath170 on @xmath171 called the equilibrium state for @xmath172 such that @xmath173 we denote @xmath174 by @xmath69 since it is usually called the sinai
ruelle bowen ( srb ) measure . since @xmath70 are attractors we have that @xmath175 ( see @xcite ) . for any probability measure @xmath176 on @xmath177 define @xmath178 then @xmath179 observe that by the ruelle inequality ( see , for instance , @xcite , theorem s.2.13 ) , @xmath180 , and so in view of assumption [ ass1.2.1 ] for any @xmath181 , @xmath182 it is known that @xmath183 is upper semicontinuous in @xmath176 since hyperbolic flows are entropy expansive ( see @xcite ) .
thus @xmath184 is a lower semicontinuous functional in @xmath176 and it is also convex ( and affine on @xmath163 ) since entropy @xmath185 is affine in @xmath176 ( see , for instance , @xcite ) .
hence , by the duality theorem ( see @xcite , p.201 ) , @xmath186 observe that this formula can be proved more directly .
namely , if we define @xmath184 by it in place of ( [ 1.2.8 ] ) then ( [ 1.2.8 ] ) follows for @xmath181 from theorem 9.12 in @xcite and it is easy to show directly that @xmath184 defined in this way equals @xmath187 for any finite signed measure @xmath176 which is not @xmath106-invariant .
since we assume that the vector field @xmath19 is @xmath53 in both arguments ( here only continuity in @xmath18 is needed ) then for any @xmath188 and @xmath189 we can define @xmath190 and @xmath191 .
then @xmath192 where @xmath193 if @xmath181 satisfying the condition in brackets exists and @xmath194 , otherwise .
since , @xmath195 is convex and continuous the duality theorem ( see @xcite , p.201 ) yields that @xmath196 provided there exists a probability measure @xmath181 such that @xmath197 and @xmath194 , otherwise .
clearly , @xmath198 is convex and lower semicontinuous in all arguments and , in particular , it is measurable .
we set also @xmath199 .
denote by @xmath81 the space of continuous curves @xmath200 $ ] in @xmath67 which is the space of continuous maps of @xmath201 $ ] into @xmath202 for each absolutely continuous @xmath203 its velocity @xmath204 can be obtained as the almost everywhere limit of continuous functions @xmath205 when @xmath206 .
hence @xmath204 is measurable in @xmath207 , and so we can set @xmath208 where @xmath209 provided for lebesgue almost all @xmath210 $ ] there exists @xmath211 for which @xmath212 and @xmath213 otherwise .
it follows from @xcite and @xcite that @xmath214 where @xmath215 is the unique solution of the equation @xmath216 where @xmath217 and the equality @xmath218 holds true if and only if @xmath219 define the uniform metric on @xmath81 by @xmath220 for any @xmath221 set @xmath222 since @xmath223 is lower semicontinuous and convex in @xmath224 and , in addition , @xmath225 if @xmath226 we conclude that the conditions of theorem 3 in ch.9 of @xcite are satisfied as we can choose a fast growing minorant of @xmath223 required there to be zero in a sufficiently large ball and to be equal , say , @xmath227 outside of it . as a result
, it follows that @xmath228 is lower semicontinuous functional on @xmath81 with respect to the metric @xmath229 , and so @xmath230 is a closed set which plays a crucial role in the large deviations arguments below .
we suppose that the coefficients of ( [ 1.1.1 ] ) satisfy the following [ ass1.2.2 ] there exists @xmath231 such that @xmath232 where @xmath233 is the @xmath234 norm of the corresponding vector fields on @xmath235 set @xmath236 and @xmath237 for all @xmath238,\,{{\varepsilon}}>0\}.$ ] clearly , @xmath239 the following is one of the main results of this paper .
[ thm1.2.3 ] suppose that @xmath240 and @xmath241 , @xmath242 are solutions of ( [ 1.1.1 ] ) with coefficients satisfying assumptions [ ass1.2.1 ] and [ ass1.2.2 ] .
set @xmath51 then for any @xmath243 and every @xmath244 there exists @xmath245 such that for @xmath246 @xmath247 and @xmath248 where , recall , @xmath54 is the normalized riemannian volume on @xmath154 the functional @xmath249 for @xmath203 is finite if and only if @xmath250 for @xmath211 and lebesgue almost all @xmath210.$ ] furthermore , @xmath249 achieves its minimum 0 only on @xmath251 satisfying ( [ 1.2.14 ] ) for all @xmath252.$ ] finally , for any @xmath253 there exist @xmath254 and @xmath255 such that for all @xmath246 @xmath256 where @xmath257 is the unique solution of ( [ 1.2.14 ] ) .
observe that ( [ 1.2.18 ] ) ( which was proved already in @xcite by a less precise large deviations argument ) follows from ( [ 1.2.17 ] ) and the lower semicontinuity of the functional @xmath228 and it says , in particular , that @xmath258 converges to @xmath259 in measure on the space @xmath260 with respect to the metric @xmath229 .
it is naturally to ask whether we have here also the convergence for @xmath54-almost all @xmath261 .
an example due to a.neishtadt discussed in @xcite shows that in the classical situation of perturbations of integrable hamiltonian systems , in general , the averaging principle holds true only in the sense of convergence in measure on the space of intitial conditions but not in the sense of the almost everywhere convergence .
this example concerns the simple system @xmath262 with the one dimensional slow motion @xmath263 and the fast motion @xmath264 evolving on the circle while the corresponding averaged motion @xmath265 satisfies the equation @xmath266 .
the resonance occurs here only when @xmath267 but it suffices already to create troubles in the averaging principle .
namely , it turns out that for any initial condition @xmath268 with @xmath269 there exists a sequence @xmath270 such that @xmath271 though , of course , convergence in measure holds true here ( see @xcite ) . recently
( see @xcite and remark [ rem1.2.12 ] ) , such nonconvergence examples were constructed for the difference equations averaging setup ( [ 1.1.10 ] ) with expanding fast motions and there is no doubt that such examples exist also in the continuous time setup ( [ 1.1.1 ] ) when fast motions are axiom a flows as in this paper .
observe also that ( [ 1.2.16 ] ) and ( [ 1.2.17 ] ) remain true ( with the same proof ) if we replace @xmath54 there by @xmath69 but as an example in @xcite shows we can not , in general , replace @xmath54 there by an arbitrary gibbs measure @xmath36 of @xmath27 .
next , let @xmath272 be a connected open set and put @xmath273 where we take @xmath274 if @xmath275 for all @xmath276 the following result follows directly from theorem [ thm1.2.3 ] .
[ cor1.2.4 ] under the conditions of theorem [ thm1.2.3 ] for any @xmath277 and @xmath278 @xmath279,\,{{\gamma}}_0=x,\ , { { \gamma}}_t\not\in v\right\}\nonumber .
\end{aligned}\ ] ] precise large deviations bounds such as ( [ 1.2.16 ] ) and ( [ 1.2.17 ] ) of theorem [ thm1.2.3 ] ( which will be needed uniformly on certain unstable discs ) are crucial in our study in sections [ sec1.7 ] and [ sec1.8 ] of the `` very long '' , i.e. exponential in @xmath5 , time `` adiabatic '' behaviour of the slow motion which can not be described usually in the traditional theory of averaging where only perturbations of integrable hamiltonian systems are considered .
namely , we will describe such long time behavior of @xmath88 in terms of the function @xmath280 under various assumptions on the averaged motion @xmath281 observe that @xmath282 satisfies the triangle inequality @xmath283 for any @xmath284 and it determines a semi metric on @xmath67 which measures `` the difficulty '' for the slow motion to move from point to point in terms of the functional @xmath285 .
introduce the averaged flow @xmath286 on @xmath287 by @xmath288 where , recall , @xmath289 and @xmath290 for any probability measure @xmath176 on @xmath11 . call a @xmath286-invariant compact set @xmath291 an @xmath285-compact if for any @xmath292 there exist @xmath293 and an open set @xmath294 such that whenever @xmath295 and @xmath296 we can pick up @xmath297 $ ] and @xmath298 satisfying @xmath299 it is clear from this definition that @xmath300 for any pair points @xmath301 of an @xmath285-compact @xmath302 and by the above triangle inequality for @xmath282 we see that @xmath303 takes on the same value when @xmath304 is fixed and @xmath20 runs over @xmath302 .
we say that the vector field @xmath19 on @xmath305 is complete at @xmath306 if the convex set of vectors @xmath307 , \,\nu\in{{\mathcal m}}_x\}$ ] contains an open neigborhood of the origin in @xmath16 . in lemma [ lem1.6.2 ]
we will show that if @xmath291 is a compact @xmath286-invariant set such that @xmath19 is complete at each @xmath295 and either @xmath302 contains a dense orbit of the flow @xmath286 ( i.e. @xmath286 is topologically transitive on @xmath302 ) or @xmath300 for any @xmath308 then @xmath302 is an @xmath285-compact .
moreover , to ensure that @xmath302 is an @xmath285-compact it suffices to assume that @xmath19 is complete only at some point of @xmath302 and the flow @xmath286 on @xmath302 is minimal , i.e. the @xmath286-orbits of all points are dense in @xmath302 or , equivalently , for any @xmath292 there exists @xmath309 such that the orbit @xmath310\}$ ] of length @xmath311 of each point @xmath295 forms an @xmath312-net in @xmath302 .
the latter condition obviously holds true when @xmath302 is a fixed point or a periodic orbit of @xmath286 but also , more generally , when @xmath286 on @xmath302 is uniquely ergodic ( see @xcite , @xcite , @xcite ) . among well known examples of uniquely ergodic flows we can mention irrational translations of tori and horocycle
flows on surfaces of negative curvature .
a compact @xmath286-invariant set @xmath291 is called an attractor ( for the flow @xmath286 ) if there is an open set @xmath313 and @xmath314 such that @xmath315 for an attractor @xmath302 the set @xmath316 , which is clearly open , is called the basin ( domain of attraction ) of @xmath302 .
an attractor which is also an @xmath285-compact will be called an @xmath285-attractor . in
what follows we will speak about connected open sets @xmath317 with piecewise smooth boundaries @xmath318 .
the latter can be introduced in various ways but it will be convenient here to adopt the definition from @xcite saying that @xmath318 is the closure of a finite union of disjoint , connected , codimension one , extendible @xmath53 ( open or closed ) submanifolds of @xmath16 which are called faces of the boundary .
the extendibility condition means that the closure of each face is a part of a larger submanifold of the same dimension which coincides with the face itself if the latter is a compact submanifold .
this enables us to extend fields of normal vectors to the boundary of faces and to speak about minimal angles between adjacent faces which we assume to be uniformly bounded away from zero or , in other words , angles between exterior normals to adjacent faces at a point of intersection of their closures are uniformly bounded away from @xmath319 and @xmath320 . the following result which will be proved in section [ sec1.7 ] describes exits of the slow motion from neighborhoods of attractors of the averaged motion .
[ thm1.2.5 ] let @xmath291 be an @xmath285-attractor of the flow @xmath286 whose basin contains the closure @xmath321 of a connected open set @xmath317 with a piecewise smooth boundary @xmath318 such that @xmath322 and assume that for each @xmath323 there exists @xmath324 and an @xmath325invariant probability measure @xmath326 on @xmath327 such that @xmath328,\ ] ] i.e. @xmath329 and the former vector points out into the interior while the latter into the exterior of @xmath317 .
set @xmath330 and @xmath331 .
then @xmath332 takes on the same value @xmath333 and @xmath334 coincides with the same compact nonempty set @xmath335 for all @xmath336 while @xmath337 for all @xmath338 .
furthermore , for any @xmath338 , @xmath339 and for each @xmath340 there exists @xmath341 such that for all small @xmath342 , @xmath343 next , set @xmath344 where @xmath345 dist@xmath346 and @xmath347 if @xmath348 and @xmath349 , otherwise
. then for any @xmath338 and @xmath253 there exists @xmath350 such that for all small @xmath342 , @xmath351 finally , for every @xmath338 and @xmath253 , @xmath352 provided @xmath353 and the latter holds true if and only if for some @xmath277 there exists @xmath354 such that @xmath355 for lebesgue almost all @xmath252 $ ] with @xmath211 then @xmath353 .
theorem [ thm1.2.5 ] asserts , in particular , that typically the slow motion @xmath88 performs rare ( adiabatic ) fluctuations in the vicinity of an @xmath285-attractor @xmath302 since it exists from any domain @xmath313 with @xmath356 for the time much smaller than @xmath357 ( as the corresponding number @xmath358 will be smaller ) and by ( [ 1.2.23 ] ) it can spend in @xmath359 only small proportion of time which implies that @xmath88 exits from @xmath59 and returns to @xmath360 ( exponentially in @xmath5 ) many times before it finally exits @xmath317 .
we observe that in the much simpler uncoupled setup corresponding results in the case of @xmath302 being an attracting point were obtained for a continuous time markov chain and an axiom a flow as fast motions in @xcite and @xcite , respectively , but the proofs there rely on the lower semicontinuity of the function @xmath282 which does not hold true in general , and so extra conditions like @xmath285-compactness of @xmath302 or , more specifically , the completness of @xmath19 at @xmath302 should be assumed there , as well .
it is important to observe that the intuition based on diffusion type small random perturbations of dynamical systems should be applied with caution to problems of large deviations in averaging since the @xmath285-functional of theorem [ thm1.2.3 ] describing them is more complex and have rather different properties than the corresponding functional emerging in diffusion type random perturbations of dynamical systems ( see @xcite ) .
the reason for this is the deterministic nature of the slow motion @xmath88 which unlike a diffusion can move only with a bounded speed , and moreover , even in order to ensure its `` diffusive like '' local behaviour ( i.e. to let it go in many directions ) some extra nondegeneracy type conditions on the vector field @xmath19 are required .
our next result describes rare ( adiabatic ) transitions of the slow motion @xmath88 between basins of attractors of the averaged flow @xmath286 which we consider now in the whole @xmath16 and impose certain conditions on the structure of its @xmath361-limit set .
[ ass1.2.6 ] assumptions [ ass1.2.1 ] and [ ass1.2.2 ] hold true for @xmath362 , the family @xmath363 is a compact set of diffeomorphisms in the @xmath13 topology , @xmath364 for some @xmath231 independent of @xmath365 and there exists @xmath366 such that @xmath367 the condition ( [ 1.2.26 ] ) means that outside of some ball all vectors @xmath368 have a bounded away from zero projection on the radial direction which points out to the origin .
this condition can be weakened , for instance , it suffices that @xmath369 but , anyway , we have to make some assumption which ensure that the slow motion stays ( at least , for `` most '' initial points @xmath261 ) in a compact region where really interesting dynamics takes place .
next , suppose that the @xmath361-limit set of the averaged flow @xmath286 is compact and it consists of two parts , so that the first part is a finite number of @xmath285-attractors @xmath370 whose basins @xmath371 have piecewise smooth boundaries @xmath372 and the remaining part of the @xmath361-limit set is contained in @xmath373 .
we assume also that for any @xmath374 there exist @xmath324 and an @xmath375-invariant measures @xmath376 such that @xmath377\,\ , \mbox{and}\,\ , i=1, ... ,k,\ ] ] i.e. @xmath378 and it points out into the interior of @xmath379 which means that from any boundary point it is possible to go to any adjacent basin along a curve with an arbitrarily small @xmath285-functional .
let @xmath253 be so small that the @xmath380-neighborhood @xmath381 of each @xmath382 is contained with its closure in the corresponding basin @xmath383 . for any @xmath384 set @xmath385 in section [ sec1.8 ] we will derive the following result .
[ thm1.2.7 ] the function @xmath386 takes on the same value @xmath387 for all @xmath388 .
let @xmath389 .
then for any @xmath384 , @xmath390 and for any @xmath340 there exists @xmath341 such that for all small @xmath342 , @xmath391 next , set @xmath392 then for any @xmath384 and @xmath253 there exists @xmath393 such that for all small @xmath342 , @xmath394 now , suppose that the vector field @xmath19 is complete on @xmath395 for some @xmath396 ( which strengthens ( [ 1.2.27 ] ) there ) and the restriction of the @xmath361-limit set of @xmath286 to @xmath395 consists of a finite number of @xmath285-compacts .
assume also that there is a unique index @xmath397 such that @xmath398 .
then for some @xmath399 and all small @xmath342 , @xmath400 finally , suppose that the above conditions hold true for all @xmath401 .
define @xmath402 , @xmath403 and recursively , @xmath404 where @xmath405 , @xmath406 if @xmath407 , and set @xmath408 .
then for any @xmath384 and @xmath340 there exists @xmath341 such that for all @xmath409 and sufficiently small @xmath342 , @xmath410 and for some @xmath399 , @xmath411 generically there exists only one index @xmath412 such that @xmath413 and in this case theorem [ thm1.2.7 ] asserts that @xmath414 arrives ( for `` most '' @xmath261 ) at @xmath415 after it leaves @xmath383 .
if @xmath416 contains more than one index then the method of the proof of theorem [ thm1.2.7 ] enables us to conclude that in this case @xmath417 arrives ( for `` most '' @xmath261 ) at @xmath418 after leaving @xmath383 but now we can not specify the unique basin of attraction of one of @xmath419 s where @xmath420 exits from @xmath383 .
if the succession function @xmath421 is uniquely defined then it determines an order of transitions of the slow motion @xmath88 between basins of attractors of @xmath89 and because of their finite number @xmath88 passes them in certain cyclic order going around such cycle exponentially many in @xmath5 times while spending the total time in a basin @xmath383 which is approximately proportional to @xmath422 . if there exist several cycles of indices @xmath423 where @xmath424 and @xmath425 then transitions between different cycles may also be possible . in the uncoupled case with fast motions being continuous
time markov chains a description of such transitions via certain hierarchy of cycles appeared in @xcite and @xcite without detailed proofs but relying on some heuristic arguments . in our
fully coupled deterministic setup a rigorous justification of the corresponding description seems to be difficult in a more or less general situation though for some specific simple examples ( as , for instance , those which are considered in section [ sec1.9 ] ) this looks feasible while it is not clear whether it is possible to describe in our situation a limiting as @xmath426 behaviour of the slow motion @xmath427 when @xmath33 is small but fixed .
the proof of theorems [ thm1.2.5 ] , [ thm1.2.7 ] and to certain extent also of theorem [ thm1.2.3 ] rely , in particular , on certain `` markov property type '' arguments which enable us to extend estimates on relatively short time intervals to very long time intervals by , essentially , iterating them where the crucial role is played by a volume lemma type result of section [ sec1.3 ] together with the technique of @xmath428-separated sets and bowen s @xmath428-balls on unstable leaves of the perturbed flow @xmath22 .
moreover , the proof of ( [ 1.2.32 ] ) and ( [ 1.2.33 ] ) require certain rough strong markov property type arguments which enable us to study the slow motion at subsequent hitting times @xmath429 of small neighborhoods of attractors of the averaged motion . in order to produce a wide class of systems satisfying the conditions of theorem [ thm1.2.7 ] we can choose , for instance , a vector field @xmath430 on @xmath16
whose @xmath361-limit set satisfies the conditions stated above for the averaged system together with a family of vector fields @xmath431 on @xmath16 ( parametrized by @xmath432 ) such that @xmath433 and then set @xmath434 . as a specific example we can take the flows @xmath435 to be geodesic flows on the manifold @xmath11 with ( changing ) constant negative curvature @xmath20 , @xmath436 to be a one dimensional vector field on @xmath437 and @xmath431 can be just a function @xmath438 on @xmath11 with zero integral with respect to the lebesgue measure there .
in section [ sec1.9 ] we will derive similar results for the discrete time case where differential equations ( [ 1.1.1 ] ) are replaced by difference equations ( [ 1.1.10 ] ) .
namely , recall that a compact subset @xmath123 of a compact riemannian manifold @xmath11 is called hyperbolic if it is @xmath439-invariant and there exists @xmath111 and the splitting @xmath440 into the continuous subbundles @xmath441 of the tangent bundle @xmath114 restricted to @xmath115 the splitting is invariant with respect to the differential @xmath442 of @xmath443 and there is @xmath444 such that for all @xmath120 and @xmath445 the inequalities ( [ 1.2.2 ] ) with @xmath207 replaced by @xmath446 hold true .
a hyperbolic set @xmath123 is said to be basic hyperbolic if the periodic orbits of @xmath447 are dense in @xmath115 @xmath447 is topologically transitive , and there exists an open set @xmath125 with @xmath448 such a @xmath123 is called a basic hyperbolic attractor if for some open set @xmath59 and @xmath449 @xmath450 where @xmath129 denotes the closure of @xmath130 if @xmath131 then @xmath107 is called an anosov flow . if @xmath439 is a @xmath13 endomorphism of @xmath11 and there exists @xmath111 such that @xmath451 for all @xmath452 then @xmath439 is called an expanding map ( or transformation ) of @xmath11
. it will be convenient for our exposition to use the notation of the expanding subbundle @xmath453 also in the case of expanding maps where , of course , @xmath454 . we replace now assumption [ ass1.2.1 ] by the following one .
[ ass1.2.8 ] the family @xmath455 in ( [ 1.1.10 ] ) consists of @xmath13-diffeomorphisms or endomorphisms of a compact @xmath12-dimensional riemannian manifold @xmath11 with uniform @xmath13 dependence on the parameter @xmath20 belonging to a neighborhood of the closure @xmath133 of a relatively compact open connected set @xmath57 .
all @xmath456 are either expanding maps of @xmath11 or diffeomorphisms possessesing basic hyperbolic attractors @xmath70 with hyperbolic splittings satisfying ( [ 1.2.2 ] ) with the same @xmath111 and there exists an open set @xmath137 and @xmath444 satisfying ( [ 1.2.4 ] ) with @xmath446 in place of @xmath207 .
let @xmath457 be the absolute value of the jacobian of the linear map @xmath458 with respect to the riemannian inner products and set @xmath459 the function @xmath142 is known to be h " older continuous in @xmath143 since the subbundles @xmath144 are h " older continuous ( see @xcite ) , and @xmath142 is @xmath53 in @xmath20 ( see @xcite ) .
the topological pressure @xmath460 of a function @xmath155 for @xmath439 is defined similarly to the continuous time ( flow ) case above but now time should run only over integers and the integral @xmath461 should be replaced by the sum @xmath462 ( see @xcite ) . again
the variational principle ( [ 1.2.6 ] ) holds true and if @xmath169 is a h " older continuous function on @xmath70 there exists a unique @xmath463invariant measure @xmath170 on @xmath171 called the equilibrium state for @xmath464 which satisfies ( [ 1.2.7 ] ) .
in particular , @xmath465 is usually called the sinai ruelle
bowen ( srb ) measure . since @xmath70 are attractors we have that @xmath175 ( see @xcite ) and the same holds true in the expanding case , as well .
next , we define @xmath184 , @xmath195 , @xmath466 , @xmath198 , @xmath223 , @xmath249 , and @xmath251 as in ( [ 1.2.8 ] ) and ( [ 1.2.10])([1.2.14 ] ) . in place of assumption [ ass1.2.2 ]
we will rely now on the similar one concerning the equation ( [ 1.1.10 ] ) .
[ ass1.2.9 ] there exists @xmath231 such that @xmath467 where the first @xmath468 is the @xmath53 norm of the corresponding vector fields on @xmath305 and the second expression is the @xmath13 norm ( with respect to the corresponding riemannian metrics ) of the map @xmath469 acting by @xmath470 .
[ thm1.2.10 ] assume that assumptions [ ass1.2.8 ] and [ ass1.2.9 ] are satisfied and that @xmath471 is obtained by ( [ 1.1.10 ] ) . for @xmath472 $ ]
define @xmath473 and set @xmath51 .
then theorem [ thm1.2.3 ] and corollary [ cor1.2.4 ] hold true with the corresponding functionals @xmath474 .
theorems [ thm1.2.5 ] and [ thm1.2.7 ] hold true , as well , under the corresponding assumptions about the family @xmath475 ( with @xmath362 in the case of theorem [ thm1.2.7 ] ) and about the averaged system ( [ 1.1.11 ] ) ( in particular , about its attractors ) in place of the system ( [ 1.1.6 ] ) . in section [ sec1.9 ] we exhibit computations which demonstrate the phenomenon of theorem [ thm1.2.7 ] in the discrete time case for two simple examples where @xmath476 are one dimensional maps @xmath477 ( mod 1 ) and the averaged equation has three attracting fixed points . in the last section [ sec1.10 ] we discuss a stochastic resonance type phenomenon which can be exhibited in three scale systems where fast motions are hyperbolic flows ( hyperbolic diffeomorphisms , expanding transformations ) as above depending on the intermediate and slow motions while the intermediate motion performs rare transitions between attracting fixed points of corresponding averaged systems which under certain conditions creates a nearly periodic motion of the slow one dimensional motion .
[ rem1.2.11 ] computation or even estimates of functionals @xmath249 seem to be quite difficult already for simple discrete ( and , of course , more for continuous ) time examples since this leads to complicated nonclassical variational problems .
this is crucial in order to estimate numbers @xmath387 which according to theorem [ thm1.2.7 ] are responsible for transitions of the slow motion between basins of attractors of the averaged system .
[ rem1.2.12 ] the estimate ( [ 1.2.18 ] ) shows that @xmath420 tends as @xmath50 to @xmath259 uniformly on @xmath201 $ ] in the sense of convergence in measure @xmath54 considered on the space of initial conditions @xmath432 .
a natural question to ask is whether the convergence for almost all ( fixed ) initial conditions also takes place in our circumstances . in @xcite
we give a negative answer to this question , in paricular , for the following simple discrete time example @xmath478 identifying 0 and 1 we view @xmath18 variable as belonging to the circle in order to fit into our setup where the fast motion runs on a compact manifold .
the averaged equation ( [ 1.1.11 ] ) has here zero in the right hand side so the averaged motion stays forever at the initial point .
the discrete time version of ( [ 1.2.18 ] ) asserted by theorem [ thm1.2.10 ] implies that @xmath479 in the sense of convergence in ( the lebesgue ) measure on the circle but we show in @xcite that for each @xmath20 there is a set @xmath480 of full lebesgue measure on the circle such that if @xmath481 then @xmath482 as @xmath50 of the left hand side in ( [ 1.2.36 ] ) is positive , i.e. there is no convergence for lebesgue almost all @xmath18 there .
namely , it turns out that for almost all initial conditions there exists a sequence @xmath483 such that the fast motion @xmath484 stays for a time of order @xmath485 close to an orbit @xmath486 ( mod @xmath487 of the doubling map with @xmath488 being a generic point with respect to a gibbs invariant measure @xmath45 of this map satisfying @xmath489 which prevents ( [ 1.2.36 ] ) .
for readers convenience we exhibit , first , in this section the setup and necessary technical results from @xcite and though their proofs can can be found in @xcite we provide for completness and readers convenience their slightly modified and corrected version also here . any vector @xmath490 can be uniquely written as @xmath491 where @xmath492 and @xmath493 and it has the riemannian norm @xmath494 where @xmath495 is the usual euclidean norm on @xmath16 and @xmath496 is the riemannian norm on @xmath114 .
the corresponding metrics on @xmath11 and on @xmath10 will be denoted by @xmath497 and @xmath498 , respectively , so that if @xmath499 then @xmath500 it is known ( see @xcite ) that the hyperbolic splitting @xmath501 over @xmath70 can be continuously extended to the splitting @xmath502 over @xmath145 which is forward invariant with respect to @xmath503 and satisfies exponential estimates with a uniform in @xmath306 positive exponent which we denote again by @xmath504 i.e. we assume now that @xmath505 provided @xmath506 , @xmath507 , @xmath508 , and @xmath509 moreover , by @xcite ( see also @xcite ) we can choose these extensions so that @xmath510 and @xmath511 will be h " older continuous in @xmath512 and @xmath53 in @xmath20 in the corresponding grassmann bundle . actually , since @xmath145 is contained in the basin of each attractor @xmath70
, any point @xmath513 belongs to the stable manifold @xmath514 of some point @xmath515 ( see @xcite ) , and so we choose naturally @xmath510 to be the tangent space to @xmath516 at @xmath517 now each vector @xmath518 can be represented uniquely in the form @xmath519 with @xmath520 , @xmath521 , @xmath522 and @xmath523 .
we denote also @xmath524 and @xmath525 . for each small @xmath526 set @xmath527 and @xmath528 which are unstable cones around @xmath453 and @xmath529 respectively .
similarly , we define @xmath530 and @xmath531 which are stable cones around @xmath532 and @xmath533 respectively . put @xmath534\}$ ] , where , recall , @xmath22 is the flow determined by the equations ( [ 1.1.1 ] ) .
[ lem1.3.1 ] .
there exist @xmath535 such that if @xmath536 and @xmath537 then @xmath538 and for any @xmath539 @xmath540 @xmath541 let @xmath542 @xmath543 @xmath544 and @xmath545 then @xmath546 and @xmath547 @xmath548 @xmath549 @xmath550 for some @xmath551 independent of @xmath552 and @xmath553 if @xmath554 hence , for @xmath121 @xmath555 and @xmath556 if @xmath557 then @xmath558 and @xmath559 hence , by the above , @xmath560 set @xmath561 choose @xmath562 and @xmath563 .
then we obtain that @xmath564 for all @xmath565 $ ] , and so by continuity of the splitting @xmath566 and by perturbation arguments it follows that @xmath567 for all @xmath568 $ ] provided @xmath33 is small enough . repeating this argument for @xmath569 $ ] , @xmath570 we conclude the proof of the first part of ( [ 1.3.2 ] ) and
its second part follows in the same way .
next , for @xmath557 and @xmath121 @xmath571 choose @xmath572 so small ( for instance , @xmath573 ) that for all @xmath574 and @xmath575 , @xmath576 for all @xmath577.$ ] then , @xmath578 for all such @xmath207 , and so if @xmath33 small enough we have also @xmath579 . using ( [ 1.3.2 ] ) and repeating this argument for @xmath580 in place of @xmath552 we derive ( ii ) for all @xmath581 .
the proof for stable cones @xmath582 is similar . for any linear subspace @xmath583 of @xmath584 denote by @xmath585 absolute value of the jacobian of the linear map @xmath586 with respect to inner products induced by the riemannian metric .
for each @xmath587 set also @xmath588 let @xmath589 be the dimension of @xmath590 which does not depend on @xmath20 and @xmath512 by continuity considerations . if @xmath583 is an @xmath591dimensional subspace of @xmath592 , @xmath593 and @xmath594 then it follows easily from assumption [ ass1.2.2 ] and lemma [ lem1.3.1 ] that there exists a constant @xmath595 independent of @xmath596 and of a small @xmath33 such that for any @xmath597 @xmath598 recall , that an embedded @xmath599 @xmath600dimensional disc @xmath601 in @xmath10 , @xmath602 is the image of an @xmath603-dimensional disc ( ball ) @xmath604 in @xmath605 centered at 0 under a @xmath606 diffeomorphism of a neighborhood of 0 in @xmath605 into @xmath10 and we define the boundary @xmath607 of @xmath601 as the image of the boundary @xmath608 of @xmath604 considered in the corresponding @xmath603-dimensional euclidean subspace of @xmath605 .
denote by @xmath609 the ball in @xmath610 centered at @xmath611 and let @xmath612 , @xmath551 be the set of all @xmath53 embedded @xmath613dimensional closed discs @xmath614 such that @xmath615 , @xmath616 and if @xmath617 then @xmath618 where @xmath619 is the tangent bundle over @xmath601 and @xmath620 is the interior metric on @xmath621 each disc @xmath622 will be called unstable or expanding and , clearly , @xmath623 and if @xmath624 and @xmath625 are small enough and @xmath626 then dist@xmath627 for any @xmath628 let @xmath629 and @xmath630 . set @xmath631\}$ ] and let @xmath632 and @xmath633 be natural projections on the first and second factors , respectively .
[ lem1.3.2 ] let @xmath634 , @xmath635 be as in lemma [ lem1.3.1 ] and @xmath277 .
there exist @xmath636 such that if @xmath637 , @xmath615 , @xmath638 , @xmath639 , @xmath640 and @xmath641 then \(i ) @xmath642 for any @xmath643 and @xmath644,$ ] where @xmath645 is the interior distance on @xmath59 ; \(ii ) @xmath646 and @xmath647 provided @xmath648 ; \(iii ) for all @xmath643 and @xmath649 , @xmath650 ( iv ) @xmath651 provided @xmath652 , where @xmath653 is the induced ( not normalized ) riemannian volume on @xmath621 \(i ) let @xmath654 be a smooth curve on @xmath655 connecting @xmath656 and @xmath657 then @xmath658 is a smooth curve on @xmath659 connecting @xmath660 and @xmath661 since @xmath662 then by ( [ 1.3.3 ] ) , length@xmath663length@xmath664 if @xmath665 then for such @xmath207 and @xmath666 @xmath667 observe that ( i ) is nontrivial only for large @xmath668 , so minimizing in @xmath654 in the above inequality we derive the assertion ( i ) . next , we derive ( ii ) .
its first part follows from ( [ 1.3.2 ] ) . by the definition of @xmath317 , @xmath669 for any @xmath670 .
it remains to show that @xmath671 for any @xmath672 indeed , suppose @xmath673 set @xmath674 next , we conclude via perturbation arguments that @xmath675 provided @xmath33 is small enough .
let @xmath676 it follows from lemma [ lem1.3.1 ] that @xmath677 for all @xmath152.$ ] hence , if @xmath678 then @xmath679 and so @xmath680 in order to derive ( iii ) take an arbitrary smooth curve @xmath654 on @xmath659 connecting @xmath681 and @xmath660
. then @xmath682 it follows that if @xmath683 with @xmath684 and @xmath685 then @xmath686 , @xmath687 @xmath688 and so @xmath689 hence @xmath690 and so @xmath691 minimizing the right hand side here over such @xmath654 we obtain ( iii ) using ( i ) .
finally , ( iv ) follows from ( [ 1.3.5 ] ) , ( i ) , ( ii ) , and the h " older continuity of @xmath692 ( as a function on @xmath693 for each @xmath694 and @xmath695 small enough set @xmath696 and @xmath697 which are local stable and unstable manifolds for @xmath106 at @xmath698 according to @xcite and @xcite these families can be included into continuous families of @xmath699 and @xmath591dimensional stable and unstable @xmath53 discs @xmath700 and @xmath701 respectively , defined for all @xmath261 and such that @xmath700 is tangent to @xmath702 , @xmath703 is tangent to @xmath144 , @xmath704 , and @xmath705 actually , as we noted it above if @xmath261 then @xmath18 belongs to a stable manifold @xmath706 of some @xmath707 and we choose @xmath708 to be the subset of @xmath709 [ lem1.3.3 ] for any @xmath710 small enough and a continuous function @xmath711 on @xmath712 uniformly in @xmath622 , @xmath713 and @xmath714 $ ] , @xmath715 where @xmath653 is the induced riemannian volume on @xmath621 set @xmath716 @xmath717 and @xmath718 by standard transversality considerations we can define a one - to - one map @xmath719 by @xmath720 provided @xmath721 are small and @xmath722 is sufficiently large . by the absolute continuity of the stable foliation arguments ( see , for instance , @xcite , section 3.3 ) we conclude that @xmath723 and its inverse have bounded jacobians .
it follows that it suffices to establish ( [ 1.3.6 ] ) for @xmath724 where @xmath725 uniformly in @xmath726,\,{{\gamma}}_0>0.$ ] set @xmath727 @xmath728 and @xmath729 then @xmath730 and so @xmath731 where @xmath496 is the supremum norm on @xmath732 since @xmath733 by @xcite then given @xmath292 there exist @xmath734 @xmath735 and @xmath736 such that for any @xmath737 we can define a one - to - one map @xmath738 by @xmath739 by standard absolute continuity of the stable foliation considerations ( see @xcite , section 3.3 ) it follows that @xmath740 and its inverse have bounded jacobians which together with the above arguments yield that it suffices to prove lemma [ lem1.3.3 ] only when @xmath741 and so ( see @xcite ) , @xmath742 we observe also that without loss of generality we can assume @xmath654 to be sufficiently small since we can always cover @xmath743 by @xmath744 with @xmath745 @xmath746 and small @xmath747 so proving lemma [ lem1.3.3 ] for all such @xmath748 will imply it for @xmath743 itself .
so now assume that @xmath749 and we claim that for any @xmath292 there exists @xmath750 such that @xmath751 forms an @xmath752net in @xmath70 for any @xmath753 indeed , by topological transitivity there exists @xmath515 whose orbit is dense in @xmath171 and so by standard ergodicity considerations with respect to any ergodic invariant measure with full support on @xmath70 ( take , for instance , the srb measure ) we conclude that for any @xmath754 already @xmath755 is dense in @xmath70 .
then by transversality of @xmath756 and @xmath757 there exists @xmath758 such that @xmath759 for some @xmath760 and so the forward orbit of @xmath512 is dense in @xmath70 , whence our claim holds true . by compactness and structural stability considerations
it follows that we can choose the same @xmath761 for all @xmath261 and @xmath762 for any set @xmath763 put @xmath764\}.$ ] recall , that a finite set @xmath765 is called @xmath766separated for the flow @xmath27 if @xmath767 implies @xmath768 a set @xmath765 is called @xmath766spanning if for any @xmath694 there is @xmath769 such that @xmath770 .
let @xmath751 be an @xmath752net in @xmath70 and @xmath771 be a maximal @xmath766separated subset of @xmath751
. then @xmath772 are disjoint sets . by transversality of @xmath751 and @xmath756
there exists @xmath595 such that for any @xmath694 we can find @xmath773 such that @xmath774 and so for some @xmath775 @xmath776 with some @xmath777 large enough but independent of @xmath778 hence , @xmath771 is @xmath779spanning , and so @xmath780 .
assume , first , that @xmath781 in ( [ 1.3.6 ] ) is h " older continuous in @xmath782 then by standard volume lemma arguments ( see @xcite ) we obtain for @xmath783 and @xmath784 that @xmath785 where @xmath786 does not depend on @xmath787 now ( [ 1.3.6 ] ) follows from the above integral estimates and the uniform in @xmath788separated and @xmath788spanning sets approximation of the topological pressure ( see , for instance , @xcite and @xcite ) . for a general continuous @xmath711 approximate it uniformly by h " older continuous functions and ( [ 1.3.6 ] ) will follow in this case again .
the limit ( [ 1.3.6 ] ) is uniform in @xmath789 and in @xmath611 since @xmath790 uniformly continuous in @xmath789 ( easy to see ) and it is uniformly continuous in @xmath611 ( see @xcite ) and , furthermore , it follows from lemma 5.1 from @xcite that the family @xmath791 is equicontinuous in @xmath611 .
[ prop1.3.4 ] for any @xmath792 with @xmath793 large and @xmath794 small enough there exists a positive function @xmath795 such that @xmath796 and for any @xmath797 @xmath798 , @xmath799 , @xmath800 @xmath615 and @xmath801 satisfying @xmath802 we have @xmath803 where @xmath804 is the inner product , @xmath805 @xmath806 , and @xmath653 is the induced riemannian volume on @xmath621 by ( [ 1.1.1 ] ) and ( [ 1.2.15 ] ) for any @xmath807 @xmath808 where , @xmath809 and , recall , @xmath810 then , by gronwall s inequality @xmath811 hence , @xmath812 recall , that by lemma [ lem1.3.2](iii ) for any @xmath813 , @xmath814 set @xmath815 then @xmath816 where @xmath817 , @xmath818 is the tangent space to @xmath819 at @xmath512 and @xmath820 is the induced riemannian volume on @xmath821 by ( [ 1.3.4 ] ) , ( [ 1.3.5 ] ) , lemma [ lem1.3.2](i ) , and the h " older continuity of the function @xmath822 @xmath823 for some @xmath777 independent of @xmath824 and @xmath825 since @xmath826 by lemma [ lem1.3.2](ii ) , it follows from ( [ 1.2.15 ] ) and the above estimates that @xmath827 where @xmath828 @xmath829 is a constant independent of @xmath830 and @xmath831 with @xmath832 . next , choose @xmath806 and @xmath833 $ ] so that @xmath834 is an integer .
if @xmath835 then ( [ 1.3.8 ] ) follows from ( [ 1.3.9])([1.3.11 ] ) and lemma [ lem1.3.3 ] .
now , let @xmath836 and @xmath837 . then by ( [ 1.2.15 ] ) and lemma [ lem1.3.2](i ) for any @xmath838 @xmath839 and @xmath840 where @xmath841 . integrating the inequalities above we obtain @xmath842 from the estimates ( [ 1.3.9])([1.3.11 ] ) together with lemma [ lem1.3.3 ]
we conclude that for some @xmath843 independent of @xmath844 and @xmath789 , @xmath845 where @xmath846 and @xmath847 , @xmath848 as @xmath849 observe that by ( [ 1.1.1 ] ) , ( [ 1.2.15 ] ) and lemma [ lem1.3.2](iii ) , @xmath850 and so setting @xmath851 we obtain by ( [ 1.2.15 ] ) and @xcite ( see also @xcite and @xcite ) that @xmath852 for some @xmath853 independent of @xmath854 and @xmath831 with @xmath832 provided , say , @xmath855 which we can assume without loss of generality .
a finite set @xmath856 will be called @xmath857separated if @xmath858 @xmath859 implies that @xmath860 let @xmath861 be a maximal @xmath862separated set in @xmath601 and define @xmath863 then for @xmath864 , @xmath865 where @xmath866 and the left hand side of ( [ 1.3.16 ] ) follows from lemma [ lem1.3.2](i ) assuming that @xmath33 is small enough .
observe also that @xmath867 are disjoint for different @xmath868 for @xmath869 set @xmath870 and @xmath871 with @xmath872 then by ( [ 1.3.12])([1.3.16 ] ) and lemma [ lem1.3.2](iii ) , @xmath873 similarly , we obtain @xmath874 emloying these estimates recursively for @xmath875 and estimating @xmath876 by ( [ 1.3.14 ] ) with @xmath877 and with @xmath878 in place of @xmath879 we derive ( [ 1.3.8 ] ) with @xmath880 provided @xmath881 next , under assumption [ ass1.2.6 ] we derive a volume lemma type assertion ( see @xcite ) which will be needed in sections [ sec1.7 ] and [ sec1.8 ] and which will hold true on any time intervals and not just on time intervals of order @xmath5 as in lemma [ lem1.3.2](iv ) . in order to do so we will consider a subset of embedded @xmath13 discs from @xmath882 taking special care of their @xmath13 bounds . namely , let exp@xmath883 be the exponential map which is a diffeomorphism of @xmath884 onto the open @xmath885neighborhood @xmath886 of @xmath18 provided @xmath253 is small enough .
given @xmath887 and @xmath888 set @xmath889 which is a diffeomorphism of @xmath890 onto @xmath891 .
let @xmath892 , @xmath893 , @xmath261 be an embedded @xmath13 disc .
assuming that @xmath894 we can define @xmath895 which is a @xmath13 hypersurface in @xmath896 . if @xmath380 is sufficiently small then the tangent subbubndle @xmath897 over @xmath898 still stays close to @xmath899 , and so we can represent @xmath898 as a parametric set @xmath900 where @xmath901 , @xmath902 and @xmath903 .
we will write that @xmath904 if the parametric representation of the corresponding @xmath898 as above satisfies @xmath905 [ lem1.3.5 ] there exists @xmath906 such that for any @xmath581 we can choose @xmath253 small enough and @xmath907 large enough so that if @xmath908 and @xmath894 then @xmath909 since the differential @xmath910exp@xmath911 of the exponential map at zero is the identity map it follows from the definition of @xmath612 that @xmath912 where @xmath913 ( uniformly in all @xmath601 as above ) as @xmath914 .
let @xmath893 and set @xmath915 which for each fixed @xmath916 and a sufficiently small @xmath253 ( depending on @xmath207 ) defines a diffeomorphism of @xmath890 onto its image . by ( [ 1.3.2 ] ) the tangent subbundle over @xmath917 is contained in @xmath918 , and so for small @xmath253 the tangent subbundle @xmath919 over @xmath920 stays close to @xmath921 where @xmath922 and @xmath923 .
hence , we can represent @xmath924 in a parametric form @xmath925 where @xmath926 and @xmath927 .
fix some @xmath928 so that ( [ 1.3.3 ] ) holds true .
write @xmath929 , so that , in particular , @xmath930 then @xmath931 where @xmath901 , @xmath932 , @xmath48 , @xmath80 is an @xmath933matrix , @xmath19 is an @xmath934matrix with @xmath935 and @xmath936 for all @xmath937 and @xmath938 where @xmath913 as @xmath914 . by ( [ 1.2.15 ] ) , ( [ 1.3.1 ] ) and assumption [ ass1.2.6 ] it follows that there exists a constant @xmath939 such that for any @xmath940 , @xmath48 , @xmath432 , @xmath941 by ( [ 1.3.1 ] ) we can choose @xmath916 large enough and then @xmath33 and @xmath380 small enough so that for all @xmath942 , @xmath943 now @xmath207 is fixed and we can choose @xmath33 and @xmath380 so small that ( [ 1.3.19 ] ) implies that , @xmath944 in order to shorten notations for every vector function @xmath945 , @xmath946 we denote by @xmath947 the jacobi matrix @xmath948 and by @xmath949 we denote the collection @xmath950 .
we set also @xmath951 observe that by assumptions [ ass1.2.1 ] , [ ass1.2.2 ] and [ ass1.2.6 ] for any @xmath916 there exists @xmath952 such that @xmath953 and @xmath954 it follows by ( [ 1.3.17])([1.3.23 ] ) ( with natural product notations ) that @xmath955 and @xmath956 similarly , @xmath957 choosing @xmath958 we obtain that if @xmath959 then @xmath960 and the assertion of lemma [ lem1.3.5 ] follows . the main purpose of the previous result is to derive the following volume lemma type assertion which plays an essential role in section [ sec1.6 ] .
[ lem1.3.6 ] for any @xmath961 there exists @xmath962 such that if @xmath963 and @xmath964 is large enough then for any @xmath916 and @xmath965 satisfying @xmath966 , @xmath967 set @xmath968 and @xmath969 . similarly to lemma [ lem1.3.2](ii ) , @xmath970 , and so by uniformity considerations there exists @xmath971 independent of @xmath972 and @xmath601 as above such that @xmath973 choose @xmath974 so that @xmath975 and set @xmath976
. then for any @xmath977 , @xmath978 and by lemma [ lem1.3.2](i ) , @xmath979 by ( [ 1.3.5 ] ) , ( [ 1.3.22 ] ) , ( [ 1.3.26 ] ) , and ( [ 1.3.27 ] ) together with lemma [ lem1.3.5 ] we conclude that there exists a constant @xmath722 such that @xmath980 now ( [ 1.3.24 ] ) follows from ( [ 1.3.25 ] ) , ( [ 1.3.26 ] ) , and ( [ 1.3.28 ] ) with @xmath981
we will need the following version of general large deviations bounds when usual assumptions hold true with errors .
an upper bound similar to ( [ 1.4.3 ] ) below appeared previously in @xcite . for simplicity we will formulate the result for @xmath982valued random vectors though the same arguments work for random variables with values in a banach space .
the proof is a strightforward modification of the standard one ( cf .
@xcite ) but still we exhibit it here for readers convenience .
[ lem1.4.1 ] let @xmath983 , @xmath984 be uniformly bounded on compact sets functions on @xmath16 and @xmath985 be a family of @xmath982valued random vectors on a probability space @xmath986 such that @xmath987 with probability one for some constant @xmath793 and all @xmath855 . for any @xmath988 and @xmath989 set
@xmath990 ( i ) for any @xmath991 there exists @xmath992 such that whenever for some @xmath993 , @xmath994 and each @xmath831 with @xmath995 , @xmath996 then for any compact set @xmath997 , @xmath998 where @xmath999 ( ii ) suppose that @xmath1000 , @xmath1001 and there exists @xmath994 such that @xmath1002 and @xmath1003 if ( [ 1.4.2 ] ) holds true then for any @xmath253 , @xmath1004 ( iii ) assume that @xmath1005 satisfy ( [ 1.4.5 ] )
. for any @xmath991 there exists @xmath1006 such that whenever for some @xmath1007 and each @xmath831 with @xmath1008 the inequality ( [ 1.4.2 ] ) holds true together with @xmath1009 then for any @xmath1010 , @xmath1011 where @xmath1012 @xmath1013 @xmath1014 , @xmath1015 @xmath1016 and @xmath1017 denotes the closure of @xmath59 .
\(i ) by ( [ 1.4.1 ] ) for any @xmath1018 and @xmath1019 there exists @xmath1020 such that @xmath1021 set @xmath1022 and cover the compact set @xmath1023 by open balls @xmath1024 let @xmath1025 be a finite subcover with a minimal number @xmath446 of elements . observe that @xmath446 does not exceed the maximal number of points in @xmath1026 with pairwise distances at least @xmath1027 and the latter number depends only on @xmath1028 and @xmath1029 . by ( [ 1.4.2 ] ) and ( [ 1.4.9 ] ) for each @xmath1030 , @xmath1031 since @xmath1032 and @xmath1033 then summing these inequalities in @xmath1030 we obtain @xmath1034 since @xmath446 is bounded by a number depending only on @xmath1035 and @xmath793 we can choose @xmath1036 so that @xmath1037 which together with ( [ 1.4.10 ] ) yield ( [ 1.4.3 ] ) .
\(ii ) by ( [ 1.4.2 ] ) and ( [ 1.4.5 ] ) , @xmath1038 and ( [ 1.4.6 ] ) follows .
\(iii ) by ( [ 1.4.5 ] ) and ( [ 1.4.7 ] ) for any @xmath1039 , @xmath1040 where @xmath1041 is the expectation with respect to the probability measure @xmath1042 on @xmath1043 such that @xmath1044 now by ( [ 1.4.2 ] ) and ( [ 1.4.5 ] ) for any @xmath831 with @xmath995 we obtain that @xmath1045 where @xmath1046 and @xmath1047 .
observe that @xmath1048 thus , applying ( i ) on the probability space @xmath1049 we derive that @xmath1050 provided @xmath1007 for a sufficiently large @xmath1036 .
this together with ( [ 1.4.11 ] ) yield ( [ 1.4.8 ] ) .
[ lem1.4.2 ] let @xmath1051 be a nondecreasing sequence of lower semicontinuous functions on a metric space @xmath1052 and let @xmath1053 assume that @xmath285 is also lower semicontinuous and for any compact set @xmath1054 denote @xmath1055 then @xmath1056 by the lower semicontinuity of @xmath1057 and @xmath285 and by compactness of @xmath1058 it follows that there exist @xmath1059 such that @xmath1060 and @xmath1061 .
passing if needed to a subsequence assume that @xmath1062 as @xmath206 .
since @xmath1063 then @xmath1064 assume now that @xmath1065 .
since @xmath1066 then for any @xmath342 there exists @xmath1067 such that @xmath1068 by the lower semicontinuity of @xmath1069 it follows that for @xmath1070 large enough @xmath1071 where we use also that @xmath1072 is a nondecreasing sequence . since ( [ 1.4.19 ] ) holds true for any @xmath1070 large enough and for each @xmath342 we can pass there to the limit so that , first , @xmath1073 and then @xmath50 yielding that @xmath1074 which together with ( [ 1.4.17 ] ) give ( [ 1.4.15 ] ) under the condition @xmath1065 .
if @xmath1075 then @xmath1076 and for any @xmath1077 there exists @xmath1078 such that @xmath1079 for any @xmath1080 . by the lower semicontinuity of @xmath1057
we conclude that @xmath1081 for @xmath1082 large enough which implies that @xmath1083 for all sufficiently large @xmath54 .
hence @xmath1084 and since @xmath80 is arbitrary the left hand side of ( [ 1.4.20 ] ) equals infinity , i.e. again ( [ 1.4.15 ] ) holds trues with both parts of it being equal @xmath187 . in the next section
we will employ the following general result which will enable us to subdivide time into small intervals freezing the slow variable on each of them so that the estimate ( [ 1.3.8 ] ) of proposition [ prop1.3.4 ] becomes sufficiently precise and , on the other hand , we will not change much the corresponding functionals @xmath228 appearing in required large deviations estimates .
this result is certainly not new , it is cited in @xcite as a folklore fact and a version of it can be found in @xcite , p.67 but for readers convenience we give its proof here .
[ lem1.4.3 ] let @xmath1085 be a measurable function on @xmath437 equal zero outside of @xmath201 $ ] and such that @xmath1086 .
for each positive integer @xmath54 and @xmath1087 $ ] define @xmath1088{{\delta}}-c)$ ] where @xmath1089 and @xmath1090 $ ] denotes the integral part . then there exists a sequence @xmath1091 such that for lebesgue almost all @xmath1087 $ ] , @xmath1092 for each @xmath253 there exists a @xmath53 function @xmath711 on @xmath437 equal zero outside of @xmath201 $ ] and such that @xmath1093 define @xmath1094 as above with @xmath711 in place of @xmath1095
. then @xmath1096 we have also @xmath1097 since @xmath1098 it follows from ( [ 1.4.22])([1.4.24 ] ) that @xmath1099 this together with the chebyshev inequality and the borel cantelli lemma yield ( [ 1.4.21 ] ) for some sequence @xmath1091 and lebesgue almost all @xmath1087 $ ] .
[ lem1.5.1 ] let @xmath1100 @xmath1101 @xmath1102 @xmath1103 , @xmath1104 @xmath1105 , @xmath1106 , @xmath1107 and for @xmath252 $ ] , @xmath1108 then @xmath1109 @xmath1110 and @xmath1111 where , recall , @xmath1112 and @xmath1113 if @xmath1114 . by ( [ 1.2.12 ] ) ,
@xmath1115 and ( [ 1.5.2 ] ) follows .
observe , that @xmath1116 and ( [ 1.5.3 ] ) follows in view of ( [ 1.2.15 ] ) .
next , by ( [ 1.2.15 ] ) , @xmath1117 and ( [ 1.5.4 ] ) follows from the gronwall inequality . for any @xmath1118 and @xmath1119 set @xmath1120 and @xmath1121 with @xmath1122 given by ( [ 1.2.10 ] ) .
[ prop1.5.2 ] let @xmath1123 and @xmath1124 be the same as in lemma [ lem1.5.1 ] and assume that @xmath1125 fix also @xmath1126 so that proposition [ prop1.3.4 ] holds true .
\(i ) there exist @xmath1127 and @xmath1128 independent of @xmath1129 such that if @xmath1130 and @xmath1131 then for any @xmath1132 , @xmath1133 where @xmath1134 , @xmath1135 does not depend on @xmath1129 and @xmath1136 in particular , if for each @xmath1137 there exists @xmath1138 such that @xmath1139 and @xmath1140 then ( [ 1.5.6 ] ) holds true with @xmath1141 in place of @xmath1142 , @xmath1137 .
\(ii ) for any @xmath1143 there exist @xmath1144 and @xmath1145 , the latter depending also on @xmath1146 , such that if @xmath1147 and @xmath1148 satisfy ( [ 1.5.8 ] ) and ( [ 1.5.9 ] ) , @xmath1149 , @xmath1150 and @xmath1151 then @xmath1152 with some @xmath1128 depending only on @xmath21 and @xmath1153 .
\(i ) assuming that @xmath625 is small and @xmath1154 is large so that @xmath1155 is still small , we consider for each @xmath240 and @xmath261 closed discs @xmath1156 and @xmath629 with @xmath1157 . for each small @xmath1158
set @xmath1159 @xmath1160 then @xmath1161 provided @xmath1162 . for any pair of compact sets @xmath1163 and @xmath695 a finite set @xmath1164 will be called @xmath1165-separated if @xmath1166 @xmath859 implies that @xmath1167 choose a maximal @xmath1168-separated set @xmath1169 in @xmath1170 ( where maximal means that the set can not be enlarged still remaining @xmath1171-separated ) . then @xmath1172 and , by lemma [ lem1.3.2](i ) for small @xmath33 , @xmath1173 , and @xmath1174 , @xmath1175 set @xmath1176 and @xmath1177 assuming that @xmath1178 .
then for @xmath1179 , @xmath1180 , @xmath1181 by lemma [ lem1.3.2](i ) if @xmath1173 and @xmath33 is small enough then @xmath1182 for any @xmath1183 and using , in addition , assumption [ ass1.2.2 ] and the inequality ( [ 1.5.3 ] ) we obtain that for any @xmath1184 , @xmath1185 hence , if @xmath1186 then for @xmath1187 and @xmath1188 , @xmath1189 provided @xmath33 is small enough , and so , by ( [ 1.5.3 ] ) and ( [ 1.5.4 ] ) , @xmath1190 where we set @xmath1191 . since @xmath1122 is ( lipschitz ) continuous in @xmath1192 there exists @xmath1193 such that @xmath1194 let @xmath1195 and @xmath1196 since @xmath1122 is lipschitz continuous ( and even @xmath53 ) in @xmath789 and @xmath1197 ( see @xcite ) it follows from ( [ 1.5.14 ] ) that @xmath1198 where @xmath1199 depends only on @xmath1200 . since @xmath1201 provided @xmath1202 we derive from lemma [ lem1.3.2](iv ) , proposition [ prop1.3.4 ] , and lemma [ lem1.4.1](i ) that for such @xmath1203 and any @xmath988 , @xmath1204 where @xmath1205 as , first , @xmath50 and then @xmath1206 .
since @xmath1207 are disjoint for different @xmath1208 we obtain from ( [ 1.5.13 ] ) and lemma [ lem1.3.2](iv ) that @xmath1209 employing ( [ 1.5.11 ] ) , ( [ 1.5.17 ] ) and ( [ 1.5.18 ] ) for @xmath1210 with @xmath1211 and @xmath1212 , respectively , and using only ( [ 1.5.17 ] ) for @xmath1213 we derive that @xmath1214 provided @xmath1215 and @xmath1216 with @xmath1217 satisfying ( [ 1.5.7 ] ) and with the same @xmath1218 as in ( [ 1.5.6 ] ) .
let @xmath1219 be a ball on @xmath1220 centered at @xmath512 and having radius @xmath1221 , @xmath1222 in the interior metric on @xmath1220 ( which , recall , is a semi - invariant extension of the family of local unstable manifolds on @xmath70 see section 3 and @xcite ) .
then @xmath1223 if @xmath1154 . recall , that if @xmath625 is small enough then the extended local unstable and stable discs @xmath1224 and @xmath1225 are defined for all @xmath513 and , in fact , by ( [ 1.2.4 ] ) , the compactness arguments and by @xcite such discs can be defined for all @xmath512 from a small neighborhood @xmath59 of @xmath1226 which is still contained in the basin of attraction of each @xmath327 . for each @xmath1227 set @xmath1228 and assume that @xmath625 is small enough so that @xmath1229 . then ( [ 1.5.19 ] ) together with the fubini theorem yield ( [ 1.5.6 ] ) with the box @xmath1230 in place of the whole @xmath145 .
relying on the transversality of unstable and weakly stable submanifolds together with compactness arguments we conclude that there exist an integer @xmath1231 depending only on @xmath625 such that @xmath145 can be covered by @xmath1231 boxes @xmath1232 which yields now ( [ 1.5.6 ] ) in the required form .
\(ii ) we start proving ( [ 1.5.10 ] ) by using ( [ 1.5.12 ] ) in order to conclude similarly to ( [ 1.5.13 ] ) that if @xmath1173 , @xmath1233 , and @xmath1234 then @xmath1235 . hence , @xmath1236 where the last inequality holds true since @xmath1237 are disjoint for different @xmath1238 . using ( [ 1.5.16 ] ) , lemma [ lem1.3.2](iv ) , proposition [ prop1.3.4 ] , and lemma [ lem1.4.1](iii )
we obtain that for any @xmath1233 , @xmath1239 , @xmath1240 and @xmath1241 , @xmath1242 where @xmath1243 @xmath1244 @xmath1245 and @xmath1246 as , first , @xmath50 and then @xmath1206 . by lemma [ lem1.3.2](iv ) and the definitions of @xmath1247 and @xmath1248 , @xmath1249 employing ( [ 1.5.20])([1.5.22 ] ) for @xmath1250 with @xmath1251 , @xmath1252 ... ,@xmath1253 and @xmath1254 , respectively , and using only ( [ 1.5.21 ] ) for @xmath1213 we derive that @xmath1255 for some @xmath1256 provided , say , @xmath1257 and @xmath1258 . since @xmath466 is differentiable in @xmath1192 ( see @xcite ) then @xmath1259 for any @xmath1260 ( see theorems 23.5 and 25.1 in @xcite ) , and so by the lower semicontinuity of @xmath223 in @xmath224 ( and , in fact , also in @xmath20 ) , @xmath1261 this together with lemma [ lem1.4.2 ] yield that @xmath1262 appearing in the definition of @xmath1263 is positive provided @xmath21 is sufficiently large .
in fact , it follows from the lower semicontinuity of @xmath223 that @xmath1262 is bounded away from zero by a positive constant independent of @xmath1264 and @xmath1147 , @xmath1137 if these points vary over fixed compact sets and ( [ 1.5.8 ] ) together with ( [ 1.5.9 ] ) hold true .
now , given @xmath1019 choose , first , sufficiently large @xmath21 as needed and then subsequently choosing small @xmath1265 and @xmath1266 , then small @xmath65 , and , finally , small enough @xmath33 we end up with an estimate of the form @xmath1267 where @xmath1128 and @xmath1268 satisfies ( [ 1.5.7 ] )
. finally , ( [ 1.5.10 ] ) follows from ( [ 1.5.23 ] ) , ( [ 1.5.24 ] ) and the fubini theorem ( similarly to ( i ) ) .
next , we pass directly to the proof of theorem [ thm1.2.3 ] starting with the lower bound . some of the details below are borrowed from @xcite but we believe that our exposition and the way of proof are more precise , complete and easier to follow .
assume that @xmath1269 , and so that @xmath654 is absolutely continuous , since there is nothing to prove otherwise .
then by ( [ 1.2.13 ] ) , @xmath1270 for lebesgue almost all @xmath1271.$ ] by ( [ 1.2.15 ] ) and ( [ 1.3.6 ] ) , @xmath1272 and so if @xmath1270 it follows from ( [ 1.2.12 ] ) that @xmath1273 .
suppose that @xmath1274 and let ri@xmath1275 be the interior of @xmath1275 in its affine hull ( see @xcite ) .
then either ri@xmath1276 or @xmath1275 ( by its convexity ) consists of one point and recall that @xmath1277 for lebesgue almost all @xmath1271 $ ] . by ( [ 1.2.10 ] ) and ( [ 1.5.25 ] ) , @xmath1278 this together with the nonnegativity and lower semi - continuity of @xmath1279 yield that there exists @xmath1280 such that @xmath1281 and by a version of the measurable selection ( of the implicit function ) theorem ( see @xcite , theorem iii.38 ) , @xmath1280 can be chosen to depend measurably in @xmath1271 $ ] .
of course , if ri@xmath1282 then @xmath1275 contains only @xmath1280 and in this case @xmath1283 for lebesgue almost all @xmath1271 $ ] . taking @xmath1284 and @xmath1285
we obtain @xmath1286 observe that @xmath1287 is measurable as a function of @xmath1288 and @xmath224 since it is obtained via ( [ 1.2.12 ] ) as a supremum in one argument of a family of continuous functions , and so this supremum can be taken there over a countable dense set of @xmath1192 s .
hence , the set @xmath1289,\,{{\alpha}}\in{{\mathcal d}}(l_s)\}=\ell^{-1}[0,\infty)$ ] is measurable , and so the set @xmath1290\}$ ] is measurable , as well .
its projection @xmath1291:\ , ( s,{{\alpha}})\in b\,\mbox{for some}\,{{\alpha}}\in{{\mathbb r}}^d\}$ ] on the first component of the product space is also measurable and @xmath317 is the set of @xmath1271 $ ] such that @xmath1275 contains more than one point .
employing theorem iii.22 from @xcite we select @xmath1292 measurably in @xmath1293 and such that @xmath1294 . by convexity and lower semicontinuity of @xmath1279
it follows from corollary 7.5.1 in @xcite that @xmath1295 for each @xmath253 set @xmath1296 then , clearly , @xmath1297 is a measurable function of @xmath1288 , and so @xmath1298 and @xmath1299 are measurable in @xmath1288 , as well . by theorems 23.4 and 23.5 from @xcite for each @xmath1300 there exists @xmath1301 such that ( [ 1.5.27 ] ) holds true . given @xmath1302 take @xmath1303 and for @xmath1271\setminus v$ ] set @xmath1284 .
then @xmath1304 for each @xmath1132 set @xmath1305 if the corresponding @xmath1306 in ( [ 1.5.27 ] ) satisfies @xmath1307 and @xmath1308 , otherwise .
note , that ( [ 1.5.27 ] ) remains true with @xmath1309 in place of @xmath1310 with @xmath1285 if @xmath1308 . as observed above @xmath1311 whenever @xmath1312 , and so @xmath1313 for lebesgue almost all @xmath1271 $ ] .
we recall also that @xmath1314 and @xmath1315 for lebesgue almost all @xmath1271 $ ] .
since @xmath1269 , @xmath1316 , and @xmath1317 as @xmath1318 for lebesgue almost all @xmath1271 $ ] , we conclude from ( [ 1.5.29 ] ) and the above observations that for @xmath21 large enough @xmath1319 next , we apply lemma [ lem1.4.3 ] to conclude that there exists a sequence @xmath1320 such that for each @xmath1321 and lebesgue almost all @xmath1322 , @xmath1323 where @xmath1324{{\delta}}_j - c$ ] , @xmath1090 $ ] denotes the integral part and we assume @xmath1325 and @xmath1326 if @xmath1327 .
choose @xmath1328 $ ] and set @xmath1329 , @xmath1330 where @xmath1331 if @xmath1332 , @xmath1333 , @xmath1334 @xmath1335 and @xmath1336 , @xmath1337 for @xmath1338 and @xmath1339 for @xmath1340 where @xmath1341 .
since @xmath1342 for lebesgue almost all @xmath1271 $ ] then @xmath1343 and , in addition , @xmath1344 by ( [ 1.5.29])([1.5.31 ] ) .
this together with ( [ 1.5.3 ] ) and ( [ 1.5.4 ] ) yield that for @xmath1345 , @xmath1346 provided @xmath1347 where @xmath1348 and @xmath1349 are the same as in lemma [ lem1.5.1 ] , the latter is defined with @xmath1350 , @xmath1338 and @xmath1351 .
choose @xmath1352 so small and @xmath1353 so large that @xmath1354 then by ( [ 1.5.32 ] ) , @xmath1355 by ( [ 1.5.29])([1.5.31 ] ) , @xmath1356 and by the construction above the conditions of the assertion ( ii ) of proposition [ prop1.5.2 ] hold true , so choosing @xmath1353 sufficiently large we derive ( [ 1.2.16 ] ) ( with @xmath1357 in place of @xmath1029 ) from ( [ 1.5.10 ] ) , ( [ 1.5.33 ] ) and ( [ 1.5.34 ] ) provided @xmath33 is small enough .
next , we pass to the proof of the upper bound ( [ 1.2.17 ] ) .
assume that ( [ 1.2.17 ] ) is not true , i.e. there exist @xmath1358 and @xmath240 such that for some sequence @xmath1359 as @xmath1360 , @xmath1361 since @xmath1362 by ( [ 1.2.15 ] ) all paths of @xmath1363 $ ] and of @xmath1364 $ ] given by ( [ 1.5.1 ] ) ( the latter for any measurable @xmath155 ) belong to a compact set @xmath1365 which consists of curves starting at @xmath20 and satisfying the lipschitz condition with the constant @xmath604 .
let @xmath1366 denotes the open @xmath625-neighborhood of the compact set @xmath230 and @xmath1367 .
for any small @xmath1368 choose a @xmath1352-net @xmath1369 in @xmath1370 where @xmath1371 . since @xmath1372 then there exists @xmath1373 and a subsequence of @xmath1374 , for which we use the same notation , such that @xmath1375 denote such @xmath1376 by @xmath1377 , choose a sequence @xmath1378 and set @xmath1379 . since @xmath1370 is compact there exists a subsequence @xmath1380 converging in @xmath81 to @xmath1381 which together with ( [ 1.5.36 ] ) yield @xmath1382 for all @xmath1368 .
we claim that ( [ 1.5.37 ] ) contradicts ( [ 1.5.2 ] ) and the assertion ( i ) of proposition [ prop1.5.2 ] .
indeed , set @xmath1383 by the monotone convergence theorem @xmath1384 similarly to our remark ( before assumption [ ass1.2.2 ] ) in section [ sec1.2 ] it follows from the results of section 9.1 of @xcite that the functionals @xmath1385 and @xmath1386 are lower semicontinuous in @xmath155 and @xmath654 ( see also section 7.5 in @xcite ) .
this together with ( [ 1.5.38 ] ) enable us to apply lemma [ lem1.4.2 ] in order to conclude that @xmath1387 where @xmath1388 .
the last inequality in ( [ 1.5.39 ] ) follows from the lower semicontinuity of @xmath228 .
thus we can and do choose @xmath1132 such that @xmath1389 by the lower semicontinuity of @xmath1390 in @xmath155 there exists a function @xmath1391 on @xmath1392 such that for each @xmath1393 , @xmath1394 next , we restrict the set of functions @xmath155 to make it compact .
namely , we allow from now on only functions @xmath155 for which there exists @xmath1393 such that either @xmath1395 or @xmath1396 for @xmath1397 , @xmath1398 and @xmath1399 where @xmath54 is a positive integer .
it is easy to see that the set of such functions @xmath155 is compact with respect to the uniform convergence topology in @xmath81 and it follows that @xmath1400 in ( [ 1.5.41 ] ) constructed with such @xmath155 in mind is lower semicontinuous in @xmath654 .
hence @xmath1401 now take @xmath1402 satisfying ( [ 1.5.37 ] ) and for any integer @xmath1403 set @xmath1404 , @xmath1405 , @xmath1406 and @xmath1407 @xmath1408 .
define a piecewise linear @xmath1409 and a piecewise constant @xmath1410 by @xmath1411 and @xmath1398 with @xmath1412 . since @xmath1402 is lipschitz continuous with the constant @xmath604 then @xmath1413
if @xmath54 is large enough and @xmath342 is sufficiently small then @xmath1414 where @xmath1415 is the same as in ( [ 1.5.6 ] ) .
since @xmath1416 it follows from ( [ 1.5.44 ] ) and ( [ 1.5.45 ] ) that @xmath1417 and by ( [ 1.5.41 ] ) and the first inequality in ( [ 1.5.45 ] ) we obtain that @xmath1418 hence , by ( [ 1.5.6 ] ) and the second inequality in ( [ 1.5.45 ] ) for all @xmath33 small enough , @xmath1419 provided @xmath1420 ( taking into account that @xmath1421 ) . by ( [ 1.5.2 ] ) and the definition of vectors @xmath1422 for any @xmath1423 , @xmath1424 therefore , @xmath1425 choosing , first , @xmath54 large enough so that @xmath65 satisfies ( [ 1.5.45 ] ) with all sufficiently small @xmath33 and also that @xmath1426 , and then choosing @xmath1352 so small that @xmath1427 , we conclude that ( [ 1.5.47 ] ) together with ( [ 1.5.49 ] ) contradicts ( [ 1.5.37 ] ) , and so the upper bound ( [ 1.2.17 ] ) holds true . since @xmath218 if and only if @xmath1428 satisfying ( [ 1.2.14 ] ) the estimate ( [ 1.2.18 ] ) follows from ( [ 1.2.17 ] ) and the lower semicontinuity of the functional @xmath228 , completing the proof of theorem [ thm1.2.3 ] .
in this section we study essential properties of the functionals @xmath228 which will be needed in the proofs of theorems [ thm1.2.5 ] and [ thm1.2.7 ] in the next sections
. we will start with the following general fact which do not require specific conditions of theorems [ thm1.2.5 ] and [ thm1.2.7 ] .
[ lem1.6.1 ] there exists @xmath758 such that if @xmath1429 then any @xmath1430 from the space @xmath163 of @xmath27-invariant probability measures on @xmath70 can be included into a weakly continuous in @xmath611 family @xmath1431 ( considered in the space of probability measures on @xmath1432 ) for which @xmath1433 is @xmath53 in @xmath611 and the entropy @xmath1434 is continuous in @xmath611 as @xmath1435 .
furthermore , there exists @xmath1436 such that @xmath1437 and for any @xmath340 there exists @xmath1438 such that if @xmath1439 is small enough the structural stability theorem for axiom a flows obtained in @xcite can be applied in order to compare @xmath167 and @xmath1440 but here we will need its more recent form derived in @xcite , @xcite , and @xcite which yields a homeomorphism @xmath1441 and a continuous function @xmath1442 on @xmath327 both with @xmath53 dependence on @xmath611 and such that the conjugate flow @xmath1443 satisfies @xmath1444 where @xmath1445 is the identity map on @xmath70 and @xmath1446 . by the standard direct verification we see that @xmath1447 is an @xmath1448-invariant probability measure .
it is known ( see , for instance , @xcite , theorem 4.2 ) that then the probability measure @xmath1449 on @xmath327 defined by its radon nikodim derivative @xmath1450 is @xmath375-invariant . in our case
this can be seen easily since for any @xmath53 function @xmath169 on @xmath327 , @xmath1451 where the last equality holds true by @xmath1452-invariance of @xmath45 .
now @xmath1453 this together with ( [ 1.2.15 ] ) yield the differentiability of @xmath1454 in @xmath611 taking into account that @xmath1442 and @xmath1455 are @xmath53 in @xmath611 ( see @xcite ) and since the proof of this fact relies on a version of the implicit function theorem ( see @xcite ) which provides derivatives in @xmath611 uniformly in @xmath306 whenever @xmath1435 and @xmath1456 is small enough we derive also ( [ 1.6.1 ] ) .
next , clearly , @xmath1457 . if we knew that @xmath1430 were ergodic then , of course
, @xmath45 would be ergodic , as well , and it would follow from theorem 10.1 in @xcite that @xmath1458 which would yield the differentiability of @xmath1434 in @xmath611 . in the general case we obtain from @xcite that @xmath1459 and so @xmath1460 since by ruelle s inequality ( see , for instance @xcite ) , @xmath1461 we derive both the continuity of @xmath1434 in @xmath611 and the first part of ( [ 1.6.2 ] ) . the second part of ( [ 1.6.2 ] ) follows from its first part in view of ( [ 1.2.8 ] ) taking into account that the function @xmath1462 defined by ( [ 1.2.5 ] ) is h " older continuous in @xmath18 and uniformly lipschitz continuous ( even @xmath53 ) in @xmath20 ( see @xcite ) and that @xmath368 is lipschitz continuous in both variables ( see ( [ 1.2.15 ] ) ) .
the following result gives , in particular , sufficient conditions for a set to be an @xmath285-compact .
[ lem1.6.2 ] ( i ) there exists @xmath1436 and for each @xmath306 where the vector field @xmath19 is complete there exists @xmath1463 such that if @xmath1464 and @xmath1465 then we can construct @xmath298 with @xmath1466 satisfying @xmath1467 it follows that @xmath1468 and @xmath1469 are locally lipschitz continuous in @xmath611 belonging to the open @xmath1456-neighborhood of @xmath20 when @xmath1470 is fixed .
\(ii ) let @xmath291 be a compact @xmath286-invariant set which either contains a dense in @xmath302 orbit of @xmath286 or @xmath300 for any pair @xmath308 .
suppose that @xmath19 is complete at each point of @xmath302 .
then @xmath302 is an @xmath285-compact .
\(iii ) assume that for any @xmath292 there exists @xmath309 such that for each @xmath295 its orbit @xmath310\}$ ] of length @xmath311 forms an @xmath312-net in @xmath302 or , equivalently , that @xmath286 is a minimal flow on @xmath302 .
suppose that @xmath19 is complete at a point of @xmath302 .
then @xmath302 is an @xmath285-compact .
\(i ) fix some @xmath306 . in view of the ergodic decomposition ( see , for instance , @xcite ) any @xmath1471 can be represented as an integral over the space of ergodic measures from @xmath163 . using the specification ( see @xcite and @xcite )
any ergodic @xmath1471 can be approximated ( in the weak sense ) by @xmath27-invariant measures sitting on its periodic orbits , i.e. by measures of the form @xmath1472 where @xmath1473 , @xmath1474 is a periodic orbit of @xmath27 with a period @xmath1475 .
this is done in a standard way by choosing a generic point of an ergodic measure @xmath45 , i.e. a point @xmath512 which satisfies @xmath1476 for any continuous function @xmath711 on @xmath70 , and then approximating the orbit of @xmath512 by periodic orbits of @xmath27 using the specification theorem ( see theorem 3.8 in @xcite ) .
it is well known ( see @xcite ) that there are countably many periodic orbits of @xmath27 which together with the above discussion yield that the closed convex hull @xmath1477 of the set @xmath1478 is a periodic orbit of @xmath1479 coincides with @xmath1480 .
now assume that @xmath19 is complete at @xmath20 .
then @xmath1481\}= \{{{\alpha}}{{\gamma}}_x^{(0)},\,{{\alpha}}\in[0,1]\}$ ] contains an open neighborhood of 0 in @xmath16 .
but then we can find a simplex @xmath1482 with vertices in @xmath1477 such that @xmath1483\}$ ] contains an open neighborhood of 0 in @xmath16 and for some periodic orbits @xmath1484 of @xmath1485 , @xmath1486 where we denote @xmath1487 . by compactness of @xmath1482
it follows also that @xmath1488 now , set @xmath1489 , @xmath1490 and include each @xmath1491 into the weakly continuous in @xmath611 families @xmath1492 constructed in lemma [ lem1.6.1 ] for @xmath611 in some neighborhood of @xmath20 .
if @xmath1493 and @xmath1494 is small enough each simplex @xmath1495 intersects and not at 0 with any ray emanating from @xmath1496 or , in other words , @xmath1497\}$ ] contains an open neighborhood of 0 in @xmath16 and , moreover , @xmath1498 since all @xmath1499 are @xmath375-invariant probability measures provided @xmath1500 we conclude that for any @xmath611 in the @xmath1494-neighborhood of @xmath20 and any vector @xmath552 there exists an @xmath375-invariant probability measure @xmath1501 such that @xmath1502 has the same direction as @xmath552 and @xmath1503 where @xmath604 is the same as in ( [ 1.2.15 ] ) .
it follows that any two points @xmath1504 and @xmath1505 from the open @xmath1494-neighborhood of @xmath20 can be connected by a curve @xmath654 lying on the interval connecting @xmath1504 and @xmath1505 with @xmath1506 , i.e. @xmath1507 with some @xmath1508 $ ] and by ( [ 1.2.9 ] ) , @xmath1509 in view of the triangle inequality for @xmath282 what we have proved yields the continuity of @xmath1468 and @xmath1469 in @xmath611 belonging to the open @xmath1494-neighborhood of @xmath20 when @xmath1470 is fixed . covering @xmath133 by @xmath1494-neighborhoods of points @xmath1429 and choosing a finite subcover we obtain ( i ) with the same constant for all @xmath133 .
next , we derive the sufficient conditions of ( ii ) for the @xmath285-compactness .
first , observe that both assumptions there imply that for any @xmath292 there exist @xmath1510 and @xmath1511 such that @xmath1512 form an @xmath1513-net in @xmath302 and @xmath1514 where @xmath793 is the same as in ( i ) . indeed ,
if there exists a dense orbit of @xmath286 in @xmath302 then a sufficiently long piece of this orbit will work as such @xmath654 with its @xmath285-functional equal 0 .
if @xmath300 for any @xmath308 then we can choose an @xmath1513-net @xmath1515 in @xmath302 and then construct curves @xmath1516 such that @xmath1517 with @xmath1518 .
taking @xmath1519 for @xmath1520 $ ] we obtain the required curve .
now , for each @xmath295 let @xmath1521 be the open @xmath1494-neighborhood of @xmath20 in @xmath16 where the construction of the part ( i ) can be implemented .
since @xmath302 is compact we can choose from the cover @xmath1522 of @xmath302 a finite subcover @xmath1523 of @xmath302 . for any positive @xmath312 such that @xmath1513 is less than the lebesgue number ( see @xcite ) of @xmath1524 we construct @xmath1512 as above and then for any @xmath336 there is @xmath1525 and @xmath1526 such that @xmath1527 , and so by the assertion ( i ) we can connect @xmath611 and @xmath1470 by a curve @xmath1528 with @xmath1529 and @xmath1530 .
it follows that any two points @xmath308 can be connected by a curve @xmath298 with @xmath1531 $ ] and @xmath1532 .
now set @xmath1533 and suppose that @xmath1534 .
let @xmath1535 be smaller than the lebesgue number of the cover @xmath1536 of @xmath1537 and set @xmath1538dist@xmath1539 .
then for any @xmath296 there exists @xmath295 with @xmath1540 , and so @xmath1541 for some @xmath1525 .
hence , by ( i ) there exists a curve @xmath1542 connecting @xmath20 with @xmath611 and such that @xmath1543 and @xmath1544 . by above we can connect any @xmath1545 with @xmath20 by a curve @xmath298 with @xmath1531 $ ] and @xmath1532 and then using @xmath1546 we arrive at a combined curve connecting @xmath1547 with @xmath611 and satisfying the conditions required to ensure that @xmath302 is an @xmath285-compact by taking @xmath1548 .
\(iii ) now assume that for any @xmath292 and each @xmath295 its piece of the @xmath286-orbit of length @xmath311 forms an @xmath312-net in @xmath302 and suppose that @xmath19 is complete at @xmath1549 .
set @xmath1550 where @xmath1551 is the differential of @xmath286 at @xmath611 .
let @xmath295 and @xmath304 with dist@xmath1552 where @xmath1553 . then for some @xmath1554 , @xmath1555 and @xmath1556 , @xmath1557 for some @xmath1558 $ ] , and so @xmath1559 .
in addition , for any @xmath1560 there exists @xmath1561 so that @xmath1562 with @xmath1563 $ ] .
now , by the assertion ( i ) we can connect @xmath1564 with @xmath1565 by a curve @xmath1566 with @xmath1567 and @xmath1568 , then connect @xmath1565 with @xmath1569 by a curve @xmath1570 with @xmath1571 and @xmath1572 .
finally , we can connect @xmath20 with @xmath611 by the curve @xmath1573 with @xmath1574 and such that @xmath1575 for @xmath1576 $ ] , @xmath1577 for @xmath1578 $ ] , @xmath1579 for @xmath1580 $ ] , and @xmath1581 for @xmath1582 $ ] yielding that @xmath302 is an @xmath285-compact . the following assertion which relies on lemma [ lem1.6.1 ]
will be also useful in our analysis .
[ lem1.6.3 ] for any @xmath292 and @xmath277 there exists @xmath1583 such that if @xmath1584 , @xmath1269 , @xmath1585 , and @xmath1586 then we can find @xmath1587 , @xmath1588 with @xmath1589 satisfying @xmath1590 by ( [ 1.2.13 ] ) and the lower semicontinuity of the functionals @xmath1591 there exist measures @xmath1592 $ ] such that @xmath1593 for lebesgue almost all @xmath252 $ ] and @xmath1594 for lebesgue almost all @xmath252 $ ] .
recall also that @xmath204 is measurable in @xmath207 . introduce the ( measurable ) map @xmath1595\times{{\mathcal p}}(\bar{{\mathcal w}})\to{{\mathbb r}}\cup\{\infty\}\times{{\mathbb r}}^d$ ] defined by @xmath1596 . recall that @xmath204 is measurable in @xmath207 , and so another map @xmath1597\to
{ { \mathbb r}}\cup\{\infty\}\times{{\mathbb r}}^d$ ] defined by @xmath1598 is also measurable in @xmath252 $ ] . then @xmath1599 and it follows from the measurable selection in the implicit function theorem ( see @xcite , theorem iii.38 ) that measures @xmath1600 satisfying this condition
can be chosen to depend measurably on @xmath252 $ ] .
now , given @xmath292 we pick up a small @xmath1583 which will be specified later on and employ lemma [ lem1.4.3 ] in the same way as in ( [ 1.5.31 ] ) together with ( [ 1.2.9 ] ) , ( [ 1.2.11 ] ) , and ( [ 1.2.13 ] ) in order to conclude that for all @xmath409 large enough there exists @xmath1601 such that if @xmath1602 then @xmath1603 set @xmath1604 for @xmath1605 and @xmath1606 for @xmath1607
. then @xmath1608 $ ] defines a polygonal line such that @xmath1609 next , set @xmath1610 for all @xmath1611 $ ] and continue the construction of @xmath1546 in the following recursive way .
suppose that @xmath1612 is already defined for all @xmath1613 $ ] and some @xmath1614 .
denote @xmath1615 , @xmath1616 , @xmath1617 and suppose that @xmath1618 where @xmath604 is the same as in ( [ 1.2.15 ] ) and @xmath1456 comes from lemma [ 1.6.1 ] . for @xmath1619 $ ]
define @xmath1612 as the integral curve starting at @xmath1620 of the vector field @xmath1621 with @xmath1622 obtained in lemma [ lem1.6.1 ] for @xmath1623 , i.e. @xmath1612 is the solution of the equation @xmath1624 this definition is legitimate since in view of ( [ 1.2.15 ] ) and our assumption on @xmath1620 the curve @xmath1625 $ ] does not exit the @xmath1456-neighborhood of @xmath1626 . by
( [ 1.6.1 ] ) and the above for all @xmath1619 $ ] , @xmath1627 and so @xmath1628}| \tilde{{\gamma}}_t-\psi_t^{(n)}|\leq|z_j - y_j|(1+ctn^{-1})+\zeta t / n+ckt^2n^{-2}.\ ] ] assuming that @xmath1586 with @xmath1629 small enough and since @xmath1630 we obtain successively from here that for all @xmath1631 , @xmath1632 which enables us to continue our construction recursively for @xmath1631 if @xmath1629 and @xmath1633 are small enough yielding also that @xmath1634 hence , the first part of ( [ 1.6.3 ] ) follows provided @xmath1629 and @xmath1633 are sufficiently small . next ,
observe that @xmath1635}|\tilde{{\gamma}}_t - x_j|\leq |z_j - y_j| |y_j - x_j|+ktn^{-1}\leq ktn^{-1}(1+e^{ct})+\zeta(1 + 2e^{ct})\ ] ] and the right hand side here can be made as small as we wish choosing @xmath1629 small and @xmath446 large .
hence , by lemma [ lem1.6.1 ] we can make @xmath1636}\big\vert i_{\tilde{{\gamma}}_t}(\mu_{x_j\tilde{{\gamma}}_t})-i_{x_j}(\nu_{t_j^{(n)}})\big\vert <
\eta/2t\ ] ] which together with ( [ 1.6.4 ] ) yield the second part of ( [ 1.6.3 ] ) . the following result will enable us to control the time which the slow motion can spend away from the @xmath361-limit set of the averaged motion .
[ lem1.6.4 ] let @xmath1637 be a compact set not containing entirely any forward semi - orbit of the flow @xmath286 .
then there exist positive constants @xmath1638 and @xmath1639 such that for any @xmath1640 and @xmath641 , @xmath1641\big\}\geq a[t / t]\ ] ] where @xmath1642 $ ] denotes the integral part of @xmath1643 .
for each @xmath1640 set @xmath1644 . by the assumption of the lemma @xmath1645 for each @xmath1640
and it follows from continuous dependence of solutions of ( [ 1.1.6 ] ) on initial conditions that @xmath1646 is upper semicontinuous .
hence , @xmath1647 . set @xmath1648 and @xmath1649 for all @xmath1271\}$ ] .
since no @xmath1650 can be a solution of the equation ( [ 1.2.14 ] ) then @xmath1651 for any @xmath1650 .
the set @xmath1652 is closed with respect to the uniform convergence and since the functional @xmath228 is lower semicontinuous we obtain that @xmath1653 this together with ( [ 1.2.13 ] ) yield the assertion of lemma [ lem1.6.4 ] .
untill now we have not used specific assumptions of theorem [ thm1.2.5 ] but some of them will be needed for the following auxiliary result .
[ lem1.6.5 ] let @xmath317 be a connected open set with a piecewise smooth boundary and assume that ( [ 1.2.20 ] ) holds true .
then the function @xmath1654 is upper semicontinuous at any @xmath1655 for which @xmath1656 .
let @xmath1657 be an @xmath285-compact .
\(i ) then for each @xmath1658 the function @xmath303 takes on the same value @xmath1659 for all @xmath295 , and so @xmath1654 takes on the same value @xmath333 for all @xmath295 and the set @xmath1660 coincides with the same ( may be empty ) set @xmath335 for all @xmath295 . furthermore , for each @xmath253 there exists @xmath1661 such that for any @xmath295 we can construct @xmath1662 with @xmath1663 $ ] satisfying @xmath1664 \(ii ) suppose that @xmath1665 and dist@xmath1666 for some @xmath338 and @xmath1667 as @xmath426 .
then @xmath337 and for any @xmath253 there exist @xmath1668 ( depending only on @xmath380 and the function @xmath1218 but not on @xmath20 ) and @xmath1669 with @xmath1670 $ ] satisfying @xmath1671 in particular , if @xmath353 then @xmath1672 and if @xmath302 is an @xmath285-attractor of the flow @xmath286 then @xmath1672 for all @xmath338 .
\(iii ) suppose that for any open set @xmath313 the compact set @xmath1673 does not contain entirely any forward semi - orbit of the flow @xmath286 .
then the function @xmath1659 is lower semicontinuous in @xmath1658 , @xmath1674 as dist@xmath1675 , and @xmath335 is a nonempty compact set .
let @xmath1656 for some @xmath1655 .
then for any @xmath340 there exist @xmath277 and @xmath1676 such that @xmath1677 and @xmath1678 . by lemma [ lem1.6.3 ] for any @xmath292 we can choose @xmath1583 so that if @xmath1679 then there exists @xmath1587 such that @xmath1680 and @xmath1681 .
let @xmath1682 be measures obtained in lemma [ lem1.6.1 ] for @xmath1683 with @xmath176 satisfying the second part of ( [ 1.2.20 ] ) .
since the boundary @xmath318 is piecewise smooth it follows from the continuous dependence of solutions of ordinary differential equations on initial conditions that for all small @xmath292 there exists @xmath1684 as @xmath1685 such that if @xmath1686 $ ] is an integral curve of the vector field @xmath1687 with @xmath1688 then @xmath1689 .
since @xmath1690 we can define @xmath1691 for @xmath1692 $ ] .
now , @xmath1693 and by ( [ 1.2.9 ] ) , @xmath1694 thus we can choose @xmath312 so small that @xmath1695 and the upper semicontinuity of @xmath1654 at @xmath1565 follows . from now on till the end of the proof of this lemma
we assume that @xmath302 is an @xmath285-compact and prove , first , the assertion ( i ) .
it follows from the definition of an @xmath285-compact that @xmath1696 for any pair @xmath1697 , and so @xmath1698 for any such @xmath1699 and each @xmath1658 .
it follows that @xmath1654 takes on the same value @xmath333 for all @xmath295 and all sets @xmath1700 coincide with some , may be empty , set @xmath335 . fix @xmath1549 .
then for each @xmath253 there exists @xmath1701 and @xmath1702 such that @xmath1703 by the definition of an @xmath285-compact there exists @xmath1704 such that for any @xmath336 we can construct @xmath1705 with @xmath1706 $ ] satisfying @xmath1707 defining @xmath1708 by @xmath1709 for @xmath1710 $ ] and @xmath1711 for @xmath1712 $ ] we obtain a curve satisfying ( [ 1.6.5 ] ) with @xmath1713 .
next , we prove ( ii ) assuming that @xmath353 and that dist@xmath1714 for some @xmath338 with @xmath1667 as @xmath426 . by ( i ) , for any @xmath292 there exists @xmath1715 such that for any @xmath336 we can construct @xmath1716 with @xmath1717 $ ] and @xmath1718 satisfying ( [ 1.6.5 ] ) with @xmath1719 . for such @xmath312 and @xmath1720 choose @xmath1629 by lemma [ lem1.6.3 ] so that if @xmath1721 and @xmath336 then in the same way as at the beginning of the proof of this lemma we can construct @xmath1722 with @xmath1684 as @xmath1685 such that @xmath1723 pick up @xmath1724 so that @xmath1725
. then @xmath1726 for @xmath1727 and some @xmath336 .
now construct as above @xmath1546 for such @xmath611 and define @xmath1669 with @xmath1728 setting @xmath1729\,\,\mbox{and}\,\ , \hat{{\gamma}}^x_t=\tilde{{\gamma}}(t-\tilde t)\,\,\mbox{for}\,\ , t\in[\tilde t , \tilde t+t_z+t(\eta)].\ ] ] then @xmath1730 and @xmath1731 $ ] .
choosing @xmath312 so small that @xmath1732 and then taking @xmath1733 we conclude that @xmath1734 satisfies ( [ 1.6.6 ] ) .
since @xmath312 is arbitrary we obtain that @xmath337 .
if @xmath302 is an @xmath285-attractor whose basin contains @xmath321 then we can choose @xmath1667 as @xmath1735 which in view of the continuous dependence of @xmath1736 on @xmath20 will be the same for all @xmath1737 ( though for this lemma @xmath1738 as above depending on @xmath20 would suffice , as well ) , so our conditions are satisfied now for all @xmath338 .
hence , in this case @xmath1654 is finite in the whole @xmath317 , completing the proof of ( ii ) . finally , we prove ( iii ) .
recall , that by the definition of an @xmath285-compact @xmath302 it follows that @xmath300 whenever @xmath308 .
for all @xmath292 let @xmath294 be open sets appearing in the definition of an @xmath285-compact . if @xmath295 and @xmath296 then @xmath1739 .
hence , if dist@xmath1740 as @xmath206 then @xmath1741 .
now , let @xmath1742 and @xmath1743 as @xmath206 .
for each @xmath1744 set @xmath1745dist@xmath1746 and let @xmath1747 . without loss of generality
we will assume that @xmath1748 for some @xmath1749 and all @xmath1750 .
fix @xmath295 . by the definition of the function @xmath282 for any @xmath1583 we can choose @xmath1751 and @xmath1752 , @xmath1753 such that @xmath1754 for each @xmath1755 set @xmath1756 consider @xmath1757 defined by @xmath1758 for @xmath1759 $ ] which stays in @xmath1760int@xmath1761 ( where int@xmath1762 means the interior of a set @xmath1762 ) , and so by lemma [ lem1.6.4 ] we conclude that @xmath1763 provided , say , @xmath1764 where @xmath1765 depends only on @xmath312 . in order to verify the lower semicontinuity of @xmath303 at @xmath1766 we have only to consider the case @xmath1767 and
so we can assume that @xmath1768 for all @xmath1769 .
passing to a subsequence and denoting its members by the same letters we can assume also that @xmath1770 the curves @xmath1771 are lipschitz continuous with a constant @xmath604 from ( [ 1.2.15 ] ) , and so this sequence is relatively compact .
hence , we can choose a uniformly converging subsequence and denoting , again , its members by the same letters we obtain now that @xmath1772 where @xmath1773 with @xmath1774,\,$ ] dist@xmath1775 and @xmath1776 .
each curve @xmath1777 can be extended to a curve in @xmath1778 with @xmath1779 and the same @xmath285-functional by adding to one of its ends a piece of the orbit of the flow @xmath286 .
hence , we can rely on the lower semicontinuity of the functional @xmath228 in order to derive from ( [ 1.6.7 ] ) that @xmath1780 by the definition of an @xmath285-compact there exists @xmath1781 with @xmath1782 $ ] such that @xmath1783 and @xmath1784 .
it follows that @xmath1785 and since @xmath312 and @xmath1629 can be chosen arbitrarily small we conclude that @xmath1786 obtaining the lower semicontinuity of @xmath303 at @xmath1766 .
finally , the lower semicontinuity of @xmath303 in @xmath323 for a fixed @xmath295 implies that @xmath1787 is nonempty and compact and since @xmath1787 is the same for all @xmath295 by ( i ) , the proof of lemma [ lem1.6.5 ] is complete . in this section
we study essential properties of the functionals @xmath228 which will be needed in the proofs of theorems [ thm2.2.5 ] and [ thm2.2.7 ] in the next sections .
the following result which follows from @xcite is a basic step in our analysis of functionals @xmath2235 and our thanks go to r. pinsky who quickly produced on our request @xcite deriving some properties of functionals @xmath2791 needed here . [ lem2.4.1 ] for each @xmath1429 and any vector measure @xmath2543 on @xmath1052 with @xmath2792 , @xmath2793 if and only if each @xmath2794 has density @xmath2795 with respect to the riemannian volume @xmath54 on @xmath1052 such that @xmath2796 where @xmath2797 is the riemannian gradient and @xmath496 is a corresponding norm .
furthermore , there exists @xmath1436 such that for any @xmath1429 and each @xmath45 as above for which ( [ 2.4.1 ] ) holds true , @xmath2798 where @xmath2799 , and if @xmath2800 is another point then @xmath2801 next , using lemma [ lem2.4.1 ] we are able to show that each point where @xmath19 is complete can be connected with close points by curves with small @xmath285-functionals which , in particular , enables us to obtain important examples of @xmath285-compacts .
[ lem2.4.2 ] ( i ) there exists @xmath1436 and for each @xmath1429 where the vector field @xmath19 is complete there exists @xmath1463 such that if @xmath1464 and @xmath1465 then we can construct @xmath298 with @xmath1466 satisfying @xmath1467 it follows that @xmath1468 and @xmath1469 are locally lipschitz continuous in @xmath611 belonging to the open @xmath1456-neighborhood of @xmath20 when @xmath1470 is fixed .
\(ii ) let @xmath291 be a compact @xmath286-invariant set which either contains a dense in @xmath302 orbit of @xmath286 or @xmath300 for any pair @xmath308 .
suppose that @xmath19 is complete at each point of @xmath302 .
then @xmath302 is an @xmath285-compact .
\(iii ) assume that for any @xmath292 there exists @xmath309 such that for each @xmath295 its orbit @xmath310\}$ ] of length @xmath311 forms an @xmath312-net in @xmath302 and suppose that @xmath19 is complete at a point of @xmath302 .
then @xmath302 is an @xmath285-compact .
\(i ) fix some @xmath1429 and assume that @xmath19 is complete at @xmath20 .
then we can find a simplex @xmath1482 with vertices in @xmath2802 such that @xmath1483\}$ ] contains an open neighborhood of 0 in @xmath16 and @xmath1486 for some @xmath2803 with @xmath2804 . by compactness of @xmath1482
it follows that @xmath1488 by ( [ 2.2.1 ] ) there exists a small @xmath2805 such that if @xmath1493 then each simplex @xmath2806 intersects and not at 0 with any ray emanating from @xmath1496 or , in other words , @xmath1497\}$ ] contains an open neighborhood of 0 in @xmath16 and , moreover , @xmath1498 it follows that for any @xmath611 in the @xmath1494-neighborhood of @xmath20 and any vector @xmath552 there exist @xmath2807 with @xmath2808 such that @xmath2809 observe that by ( [ 2.4.2 ] ) and convexity of @xmath2810 , @xmath2811 for some @xmath722 .
hence , any two points @xmath1504 and @xmath1505 from the open @xmath1494-neighborhood of @xmath20 can be connected by a curve @xmath654 lying on the interval connecting @xmath1504 and @xmath1505 with @xmath1506 , i.e. @xmath1507 with some @xmath1508 $ ] and by ( [ 2.2.9 ] ) , @xmath2812 in view of the triangle inequality for @xmath282 what we have proved yields the continuity of @xmath1468 and @xmath1469 in @xmath611 belonging to the open @xmath1494-neighborhood of @xmath20 when @xmath1470 is fixed . covering @xmath133 by @xmath2813neighborhoods of points @xmath1429 and choosing a finite subcover we obtain ( i ) with the same constant @xmath1436 for all points in @xmath133 . [ lem2.4.3 ] for any @xmath292 and @xmath277 there exists @xmath1583 such that if @xmath1584 , @xmath1269 , @xmath1585 , and @xmath1586 then we can find @xmath1587 , @xmath1588 with @xmath1589 satisfying @xmath2814 by ( [ 2.2.9 ] ) , ( [ 2.2.19 ] ) and the lower semicontinuity of the functionals @xmath1591 there exist measures @xmath1592 $ ] such that @xmath1593 for lebesgue almost all @xmath252 $ ] and @xmath1594 for lebesgue almost all @xmath252 $ ] .
recall also that @xmath204 is measurable in @xmath207 .
introduce the ( measurable ) map @xmath1595\times{{\mathcal p}}(\bar{{\mathcal w}})\to{{\mathbb r}}\cup\{\infty\}\times{{\mathbb r}}^d$ ] defined by @xmath1596 .
recall that @xmath204 is measurable in @xmath207 , and so another map @xmath1597\to { { \mathbb r}}\cup\{\infty\}\times{{\mathbb r}}^d$ ] defined by @xmath1598 is also measurable in @xmath252 $ ]
. then @xmath1599 and it follows from the measurable selection in the implicit function theorem ( see @xcite , theorem iii.38 ) that measures @xmath1600 satisfying this condition can be chosen to depend measurably on @xmath252 $ ] .
since @xmath1269 and the @xmath263-functionals are nonnegative then @xmath2815 for lebesgue almost all @xmath252 $ ] ( and , actually , without loss of generality we can assume that @xmath2816 is finite for all @xmath252 $ ] ) .
now let @xmath2817,\ ] ] which in view of ( [ 2.2.1 ] ) determines @xmath1587 . then by ( [ 2.2.1 ] ) , @xmath2818 and by gronwall s inequality @xmath2819 this together with ( [ 2.4.2 ] ) and ( [ 2.4.3 ] ) yields that @xmath2820 for some @xmath722 independent of @xmath1629 and @xmath654 .
exchanging @xmath654 and @xmath1546 , applying the same argument and using the inequality @xmath2821 we conclude that @xmath2822 choosing @xmath1629 small enough we arrive at ( [ 2.4.4 ] ) .
[ lem2.4.4 ] let @xmath1637 be a compact set not containing entirely any forward semi - orbit of the flow @xmath286 . then there exist positive constants @xmath1638 and @xmath1639 such that for any @xmath1640 and @xmath641 , @xmath1641\big\}\geq a[t / t]\ ] ] where @xmath1642 $ ] denotes the integral part of @xmath1643 .
the result is a simple consequence of lower semicontinuity of functionals @xmath474 and the fact that @xmath218 if and only if @xmath654 is a part of an orbit of the flow @xmath286 .
further details of the argument can be found in lemma [ lem1.6.4 ] in part [ part1 ] and in lemma 2.2(a ) , chapter 4 of @xcite .
[ lem2.4.5 ] let @xmath317 be a connected open set with a piecewise smooth boundary and assume that ( [ 2.2.24 ] ) holds true .
then the function @xmath1654 is upper semicontinuous at any @xmath1655 for which @xmath1656 .
let @xmath1657 be an @xmath285-compact .
\(i ) then for each @xmath1658 the function @xmath303 takes on the same value @xmath1659 for all @xmath295 , and so @xmath1654 takes on the same value @xmath333 for all @xmath295 and the set @xmath1660 coincides with the same ( may be empty ) set @xmath335 for all @xmath295 .
furthermore , for each @xmath253 there exists @xmath1661 such that for any @xmath295 we can construct @xmath1662 with @xmath1663 $ ] satisfying @xmath2823 \(ii ) suppose that @xmath1665 and dist@xmath1666 for some @xmath338 and @xmath1667 as @xmath426 .
then @xmath337 and for any @xmath253 there exist @xmath1668 ( depending only on @xmath380 and the function @xmath1218 but not on @xmath20 ) and @xmath1669 with @xmath1670 $ ] satisfying @xmath2824 in particular , if @xmath353 then @xmath1672 and if @xmath302 is an @xmath285-attractor of the flow @xmath286 then @xmath1672 for all @xmath338 . \(iii ) suppose that for any open set @xmath313 the compact set @xmath1673 does not contain entirely any forward semi - orbit of the flow @xmath286 . then the function @xmath1659 is lower semicontinuous in @xmath1658 , @xmath1674 as dist@xmath1675 , and @xmath335 is a nonempty compact set .
in this section we derive theorems [ thm2.2.5 ] relying on certain `` markov property type '' arguments which are substantial modifications of the corresponding arguments from sections 4 and 5 of @xcite . in this and the following section in order to simplify notations we will write @xmath1788 for @xmath1789 (
both introduced in section [ sec1.3 ] ) with some large @xmath964 so that appropriate discs on ( extended ) unstable leaves @xmath1790 and all their @xmath1791-iterates belong to this set .
we start with the following result which will not only yield theorem [ thm1.2.5 ] but also will play an important role in the proof of theorem [ thm1.2.7 ] in the next section .
[ prop1.7.1 ] let @xmath317 be a connected open set with a piecewise smooth boundary @xmath318 such that @xmath1792 .
assume that for each @xmath323 there exist @xmath1793 and an @xmath107-invariant probability measure @xmath176 on @xmath327 so that @xmath1794,\ ] ] i.e. @xmath1795 and it points out into the exterior of @xmath321 .
\(i ) suppose that for some @xmath1796 and any @xmath1658 there exists @xmath1797 such that for some @xmath1798 $ ] , @xmath1799 then for each @xmath338 , @xmath1800 and for any @xmath340 there exists @xmath341 such that for all small @xmath342 ,
@xmath1801 \(ii ) assume that there exists an open set @xmath1762 such that @xmath317 contains its closure @xmath1802 and the intersection of @xmath1803 with the @xmath361-limit set of the flow @xmath286 is empty .
let @xmath1652 be a compact subset of @xmath318 such that @xmath1804 for some @xmath1805 .
then for some @xmath277 and any @xmath1438 there exists @xmath1806 such that for each @xmath338 and any small @xmath342 , @xmath1807 suppose that for some @xmath338 , @xmath1808 then @xmath1809 and for each @xmath277 there exists @xmath1810 such that for all small @xmath342 , @xmath1811 and if the set @xmath1652 from ( [ 1.7.5 ] ) coincides with the whole @xmath318 then @xmath1812 the corresponding to ( [ 1.7.3 ] ) , ( [ 1.7.4 ] ) , ( [ 1.7.8 ] ) and ( [ 1.7.9 ] ) assertions hold true also when @xmath145 and @xmath54 in these estimates are replaced by a disc @xmath622 with @xmath1813 and by @xmath653 , respectively ( ( see ( [ 1.7.21 ] ) and[1.7.22 ] ) , ( [ 1.7.34 ] ) , ( [ 1.7.36 ] ) and ( [ 1.7.37 ] ) below ) .
observe that applying to ( [ 1.5.19 ] ) and ( [ 1.5.23 ] ) the arguments which were used in order to derive theorem [ thm1.2.3 ] from proposition [ prop1.5.2 ] and the latter from proposition [ prop1.3.4 ] and lemma [ lem1.4.1 ] we obtain that ( [ 1.2.16 ] ) and ( [ 1.2.17 ] ) can be written for any disc @xmath1814 in place of the whole @xmath145 , namely , for any @xmath203 with @xmath1815 , @xmath1816 , and @xmath33 small enough @xmath1817 and @xmath1818 which holds true in the same sense as ( [ 1.2.16])([1.2.17 ] ) and ( [ 1.7.10])([1.7.11 ] ) are uniform in @xmath601 as above . in order to prove ( i ) we observe , first , that the assumption ( [ 1.7.1 ] ) above together with lemma [ lem1.6.2](i ) and the compactness of @xmath318 considerations enable us to extend any @xmath1819 slightly so that it will exit some fixed neighborhood of @xmath317 with only slight increase in its @xmath285-functional .
hence , from the beginning we assume that for each @xmath1438 there exists @xmath1820 such that for any @xmath1821 we can find @xmath277 , @xmath1797 and @xmath1798 $ ] satisfying @xmath1822 where @xmath1823dist@xmath1824 .
it follows that for any @xmath1825 , and @xmath1826 , @xmath1827\big\}\\ & = \big\ { v\in d:\,\tau^{{\varepsilon}}_{\phi_{{\varepsilon}}^{kt/{{\varepsilon}}}v}(v)>t,\,\ , \forall\ , k=0,1, ... ,n-1\big\}\subset g^{{\varepsilon}}_{n,{{\delta}}}\nonumber\\ & \stackrel{\mbox{def}}{=}\big\ { v\in d:\,\phi_{{\varepsilon}}^{kt/{{\varepsilon}}}v \not\in\bigcup_{z\in v}a_{{{\delta}}}^{{{\varepsilon}},t}({{\varphi}}^z ) , \forall\ , k=0,1, ... ,n-1\big\}\nonumber \end{aligned}\ ] ] where for any subset @xmath1828 and @xmath1829 , @xmath1830 for @xmath1831 define @xmath1832 and @xmath1833 which are , clearly , compact sets satisfying @xmath1834 let @xmath861 be a maximal @xmath1835- separated set in @xmath1836
. then @xmath1837 and the left hand side of ( [ 1.7.13 ] ) is a disjoint union .
this together with lemma [ lem1.3.6 ] give @xmath1838 by lemma [ lem1.3.2](ii ) , @xmath1839 clearly , for any @xmath1840 , @xmath1841 in view of ( [ 1.7.15 ] ) we can apply ( [ 1.7.10 ] ) which together with the choice of curves @xmath1842 yield that for any @xmath1019 and @xmath33 small enough , @xmath1843 where @xmath1844 . by lemma [ lem1.3.6 ]
it follows that @xmath1845 for some @xmath254 . since @xmath1846 are disjoint for different @xmath1840 we derive from ( [ 1.7.14 ] ) , ( [ 1.7.16 ] ) and ( [ 1.7.18 ] ) that @xmath1847 where @xmath1848 . applying ( [ 1.7.19 ] ) for @xmath869
we obtain that @xmath1849 this together with ( [ 1.7.12 ] ) yield that for any @xmath1438 there exists @xmath1850 such that for all small @xmath342 , @xmath1851 observe that by ( [ 1.7.12 ] ) and ( [ 1.7.20 ] ) , @xmath1852 in the same way as at the end of the proof of proposition [ prop1.5.2](i ) we fix now an initial point @xmath1853 and choose discs @xmath601 to be small balls on the ( extended ) local unstable manifolds @xmath1854 which by means of the fubini theorem and compactness arguments enable us to extend ( [ 1.7.21 ] ) and ( [ 1.7.22 ] ) to the case when @xmath653 is replaced by @xmath54 and @xmath601 by @xmath145 yielding ( [ 1.7.3 ] ) and ( [ 1.7.4 ] ) since @xmath1192 and @xmath1029 in ( [ 1.7.22 ] ) can be chosen arbitrarily small as @xmath50 .
next , we derive the assertion ( ii ) .
let @xmath916 and @xmath446 be the integral part of @xmath1855 where @xmath277 will be chosen later .
let , again , @xmath1826 and @xmath338
. then @xmath1856 let @xmath604 be the intersection of the @xmath361-limit set of the flow @xmath286 with @xmath321 . then @xmath604 is a compact set and by our assumption @xmath1857 .
hence , @xmath1858 and if we set @xmath1859 then @xmath1860 .
now suppose that @xmath1861 for some @xmath864 and @xmath1862 with @xmath338 and @xmath513
. then either there is @xmath1863 $ ] such that @xmath1864 for all @xmath1865 $ ] or there exist @xmath1866 such that @xmath1867 and @xmath1868 while @xmath1869 .
set @xmath1870 and either there is @xmath1871 $ ] so that @xmath1872 for all @xmath1865 $ ] or @xmath1873 and @xmath1874 for some @xmath1875 .
then for any @xmath864 , @xmath1876 where @xmath1844 and @xmath1877 .
let @xmath1878 ( later both discs will be small balls on @xmath1879 ) assuming that @xmath625 is small and @xmath1154 is large so that @xmath1155 is still small . choose a maximal @xmath1880-separated set @xmath1881 in @xmath910 and let @xmath1882 then for @xmath33 small enough , @xmath1883 and for any @xmath1884 , @xmath1885 if @xmath1886 and @xmath1887 then by lemma [ lem1.3.2](iii ) , @xmath1888 is of order @xmath33 , and so for each @xmath292 if @xmath33 is small enough then @xmath1889 .
for each @xmath1890 set @xmath1891 and suppose that for some @xmath292 there is @xmath1892 so that @xmath1893 then @xmath1894 , where @xmath1895 is the same as in theorem [ thm1.2.3 ] , and so @xmath1896 hence , @xmath1897 by lemma [ lem1.3.2](ii ) , @xmath1898 and so applying ( [ 1.7.11 ] ) to @xmath1899 we obtain from ( [ 1.7.27]) ( [ 1.7.29 ] ) that for any @xmath1438 and sufficiently small @xmath33 uniformly in discs @xmath1899 as above , @xmath1900 this together with lemma [ lem1.3.6 ] yield that for each @xmath1901 , @xmath1902 for some @xmath722 depending only on @xmath794 . combining this with ( [ 1.7.24])([1.7.26 ] ) and lemma [ lem1.3.6 ]
we obtain that for any @xmath864 , @xmath1903 for some @xmath1904 depending only on @xmath794 .
next , we will specify @xmath1905 in ( [ 1.7.27 ] ) choosing @xmath1906 .
for each @xmath1821 we can write @xmath1907 where @xmath1908 and @xmath1909 for some @xmath1875 with @xmath1910 dist@xmath1911 and @xmath1912 and there is @xmath1871 $ ] so that @xmath1913 for all @xmath1865\}$ ] . by ( [ 1.7.5 ] ) and the lower semicontinuity of the functional @xmath1914 it follows that for any @xmath1583 we can choose @xmath292 small enough so that @xmath1915 since @xmath1916 is disjoint with the @xmath361-limit set of the flow @xmath286 and the latter is closed
then if @xmath312 is sufficiently small @xmath1917 is also disjoint with this @xmath361-limit set and , in particular , it does not contain any forward semi - orbit of @xmath286 .
hence we can apply lemma [ lem1.6.4 ] which in view of ( [ 1.2.13 ] ) implies that there exists @xmath988 such that for all small @xmath292 , @xmath1918 which is not less than @xmath1919 if we take @xmath1920 .
now , ( [ 1.7.32 ] ) and ( [ 1.7.33 ] ) produce ( [ 1.7.27 ] ) with @xmath1921 , and so ( [ 1.7.30 ] ) follows with such @xmath1905 .
this together with ( [ 1.7.23 ] ) yield that for any @xmath1438 we can choose sufficiently small @xmath1922 and then @xmath292 so that for all @xmath33 small enough @xmath1923 now assume that ( [ 1.7.7 ] ) holds true for some @xmath338 .
recall , that @xmath218 implies that @xmath654 is a piece of an orbit of the flow @xmath286 . since no @xmath203 satisfying @xmath1924}\,\mbox{dist } ( { { \gamma}}_t,\partial v)\leq a(x)/2\ ] ] can be such piece of an orbit we conclude by the lower semicontinuity of @xmath228 that @xmath1925 whenever ( [ 1.7.35 ] ) holds true for some @xmath1926 independent of @xmath654 ( but depending on @xmath20 ) .
hence , by ( [ 1.7.11 ] ) , @xmath1927 provided @xmath33 is small enough and ( [ 1.7.8 ] ) follows .
observe also that any @xmath298 with @xmath1928 and @xmath1929 should contain a piece which either belongs to some @xmath1930 or @xmath1931 , as above , or to satify ( [ 1.7.35 ] ) . by ( [ 1.7.32 ] ) , ( [ 1.7.33 ] ) , and the above remarks it follows that @xmath1932 for such @xmath654 where @xmath1933 depends only on @xmath20 , and so @xmath1934 .
finally , similarly to proposition [ prop1.5.2 ] we fix @xmath1935 , choose discs @xmath601 and @xmath910 to be small balls on the ( extended ) local unstable manifolds @xmath1936 and using the fubini theorem we extend ( [ 1.7.34 ] ) and ( [ 1.7.36 ] ) to the case when @xmath601 and @xmath653 are replaced by @xmath145 and @xmath54 , respectively , yielding ( [ 1.7.6 ] ) .
if @xmath1937 then by ( [ 1.7.6 ] ) and ( [ 1.7.8 ] ) , @xmath1938 and , since @xmath1438 is arbitrary , ( [ 1.7.9 ] ) follows completing the proof of proposition [ prop1.7.1 ] .
now we will derive theorem [ thm1.2.5 ] from proposition [ prop1.7.1 ] .
assume , first , that @xmath353 .
then by lemma [ 1.6.4 ] , @xmath1654 is finite and continuous in the whole @xmath317 .
moreover , since @xmath302 is an @xmath285-attractor the conditions of lemma [ lem1.6.5 ] are satisfied with some @xmath1667 as @xmath426 the same for all points of @xmath317 which yields the conditions of proposition [ prop1.7.1](i ) with @xmath1939 for any @xmath253 . hence , ( [ 1.7.3 ] ) and ( [ 1.7.4 ] ) hold true with @xmath1940 .
since @xmath302 is an @xmath285-attractor of the flow @xmath286 and its basin contains @xmath321 then the intersection of @xmath1941 with the @xmath361-limit set of @xmath286 is empty . by the definition of an @xmath285-attractor for any @xmath292 there
exists an open set @xmath294 such that @xmath1739 whenever @xmath295 and @xmath296 .
hence , by the triangle inequality for the function @xmath282 and lemma [ lem1.6.5 ] for any set @xmath1942 , @xmath1943 if @xmath1937 then by lemma [ lem1.6.5 ] the right hand side of ( [ 1.7.38 ] ) equals @xmath1944 .
assuming that @xmath1665 we can apply proposition [ prop1.7.1](ii ) with such @xmath1919 yielding ( [ 1.7.6 ] ) , ( [ 1.7.8 ] ) and since @xmath292 is arbitrary ( [ 1.2.21 ] ) and ( [ 1.2.22 ] ) follow in this case .
if @xmath1945 then ( [ 1.2.22 ] ) is trivial and by ( [ 1.7.38 ] ) , @xmath1946 for any @xmath1947 and @xmath1948 , and so we can apply proposition [ prop1.7.1](ii ) with any @xmath1919 which sais that the left hand side in ( [ 1.7.9 ] ) equals @xmath187 , and so ( [ 1.2.21 ] ) holds true in this case , as well .
next , we establish ( [ 1.2.23 ] ) . for small @xmath1949 and large @xmath277 which will be specified later on set @xmath1950 , @xmath1951 and @xmath1952 .
then @xmath1953 if @xmath380 is sufficiently small then @xmath1954 is still contained in the basin of @xmath302 with respect to the flow @xmath286 , and so we can choose @xmath1153 ( depending only on @xmath380 ) so that @xmath1955 then for some @xmath988 , @xmath1956 and so if @xmath1957 and @xmath1958 then dist@xmath1959 for any @xmath1960 . relying on ( [ 1.7.11 ] ) we obtain that for any @xmath1961 with @xmath1962 , @xmath1963 provided @xmath33 is small enough .
next , the same arguments which yield ( [ 1.7.34 ] ) and ( [ 1.7.36 ] ) enable us to conclude that if @xmath1438 is small enough then for any @xmath1964 with @xmath1965 , @xmath1966 now let @xmath622 , @xmath893 , @xmath1967 and @xmath1968 with @xmath625 small and @xmath793 large so that @xmath1155 is still small . let @xmath1969 and @xmath1970 be maximal @xmath1971 and @xmath1972separated sets , respectively . then @xmath1973 and since the last union is contained in a small neighborhood of @xmath601 and @xmath1974 are disjoint for different @xmath1975 we obtain using lemma [ lem1.3.6 ] that @xmath1976 similarly , @xmath1977 and @xmath1978 since by lemma [ lem1.3.2](ii ) , @xmath1979 we can apply ( [ 1.7.39])([1.7.41 ] ) together with lemma [ lem1.3.6 ] ( similarly to the proof of ( [ 1.7.30 ] ) ) in order to conclude that for sufficiently small @xmath1192 and any much smaller @xmath33 , @xmath1980 choosing discs @xmath601 to be small balls on the ( extended ) local unstable manifolds @xmath1981
together with the fubini theorem we extend this estimate to @xmath1982 for some @xmath722 depending on @xmath380 but independent of @xmath446 and @xmath33 . finally , ( [ 1.2.22 ] ) and ( [ 1.7.42 ] ) together with the chebyshev inequality yield that for @xmath1983 $ ] , each @xmath338 , a small @xmath1438 and any much smaller @xmath342 , @xmath1984 since @xmath1985 and we can choose @xmath1192 to be arbitrarily small , ( [ 1.7.43 ] ) yields ( [ 1.2.23 ] ) . in order to complete the proof of theorem [ thm1.2.5 ] we have to derive ( [ 1.2.24 ] ) . if @xmath1986 then there is nothing to prove , so we assume that @xmath335 is a proper subset of @xmath318 and in this case , clearly , @xmath1665 .
since @xmath1987 is compact and disjoint with @xmath335 which is also compact then by the lower semicontinuity of @xmath1659 established in lemma [ lem1.6.5](iii ) it follows that @xmath1988 for some @xmath1438 and all @xmath348 .
then by ( [ 1.7.38 ] ) , @xmath1989 for any @xmath1990 and @xmath1991 . hence , applying proposition [ prop1.7.1 ] we obtain that @xmath1992 and @xmath1993 for some @xmath1019 and all @xmath33 small enough yielding ( [ 1.2.24 ] ) and completing the proof of theorem [ thm1.2.5 ] .
[ prop2.5.1 ] let @xmath317 be a connected open set with a piecewise smooth boundary @xmath318 such that @xmath1792 .
assume that for each @xmath323 there exist @xmath1793 and a probability measure @xmath45 with @xmath2825 so that @xmath2826,\ ] ] i.e. @xmath2827 and it points out into the exterior of @xmath321 .
\(i ) suppose that for some @xmath1796 and any @xmath1658 there exists @xmath1797 such that for some @xmath1798 $ ] , @xmath2828 then for any @xmath338 uniformly in @xmath432 , @xmath2829 and for any @xmath340 there exists @xmath341 such that uniformly in @xmath432 for all small @xmath342 , @xmath2830 \(ii ) assume that there exists an open set @xmath1762 such that @xmath317 contains its closure @xmath1802 and the intersection of @xmath1803 with the @xmath361-limit set of the flow @xmath286 is empty .
let @xmath1652 be a compact subset of @xmath318 such that @xmath2831 for some @xmath1805 .
then for some @xmath277 and any @xmath1438 there exists @xmath1806 such that uniformly in @xmath432 for each @xmath338 and any small @xmath342 , @xmath2832 suppose that for some @xmath338 , @xmath2833 then @xmath1809 and for each @xmath277 there exists @xmath1810 such that uniformly in @xmath432 for all small @xmath342 , @xmath2834 and if the set @xmath1652 from ( [ 2.5.5 ] ) coincides with the whole @xmath318 then for all @xmath338 uniformly in @xmath432 , @xmath2835 in order to prove ( i ) we observe , first , that the assumption ( [ 2.5.1 ] ) above together with lemma [ lem2.4.2](i ) and the compactness of @xmath318 considerations enable us to extend any @xmath1819 slightly so that it will exit some fixed neighborhood of @xmath317 with only slight increase in its @xmath285-functional .
hence , from the beginning we assume that for each @xmath1438 there exists @xmath1820 such that for any @xmath1821 we can find @xmath277 , @xmath1797 and @xmath1798 $ ] satisfying @xmath1822 where @xmath1823dist@xmath1824 . employing the markov property
we obtain that for any @xmath2836 , @xmath2837\big\}\\ & = p\big\{\tau^{{\varepsilon}}_{z^{{\varepsilon}}_{x , y}(kt)}(v)>t,\,\ , \forall\ , k=0,1, ... ,n-1\big\}\leq\big(\sup_{w\in v\times{{\bf m } } } p\big\{\tau^{{\varepsilon}}_{w}(v)>t\big\}\big)^n .
\nonumber \end{aligned}\ ] ] from ( [ 2.2.10 ] ) and ( [ 2.5.2 ] ) it follows that @xmath2838 by ( [ 2.5.10 ] ) and ( [ 2.5.11 ] ) , @xmath2839 and @xmath2840 yielding ( [ 2.5.3 ] ) and ( [ 2.5.4 ] ) since @xmath1192 and @xmath1029 in ( [ 2.5.13 ] ) can be chosen arbitrarily small as @xmath2841 .
next , we derive the assertion ( ii ) .
let @xmath916 and @xmath446 be the integral part of @xmath1855 where @xmath277 will be chosen later .
let , again , @xmath2842 with @xmath338 and @xmath432
. then @xmath2843 let @xmath604 be the intersection of the @xmath361-limit set of the flow @xmath286 with @xmath321 . then @xmath604 is a compact set and by our assumption @xmath1857 .
hence , @xmath1858 and if we set @xmath1859 then @xmath1860 .
now suppose that @xmath2844 for some @xmath864 and @xmath2845 with @xmath338 and @xmath2846 .
then either there is @xmath1863 $ ] such that @xmath2847 for all @xmath1865 $ ] or there exist @xmath1866 such that @xmath1867 and @xmath2848 while @xmath2849 . set @xmath1870 and either there is @xmath1871 $ ] so that @xmath1872 for all @xmath1865 $ ] or @xmath1873 and @xmath1874 for some @xmath1875 .
then for any @xmath864 , @xmath2850 for each @xmath1890 set @xmath1891 and suppose that for some @xmath292 there is @xmath1892 so that @xmath2851 then @xmath1894 , where @xmath1895 is the same as in theorem [ thm2.2.2 ] , and so @xmath2852 from ( [ 2.2.10 ] ) and ( [ 2.5.15])([2.5.17 ] ) we obtain that for any @xmath1438 and all sufficiently small @xmath33 , @xmath2853 for some @xmath1904 . next , we will specify @xmath1905 in ( [ 2.5.16 ] ) choosing @xmath1906 . for each @xmath1821 we can write @xmath2854 where @xmath1908 and @xmath1909 for some @xmath1875 with @xmath1910 dist@xmath1911 and @xmath1912 and there is @xmath1871 $ ] so that @xmath1913 for all @xmath1865\}$ ] . by ( [ 2.5.5 ] ) and the lower semicontinuity of the functional @xmath1914 it follows that for any @xmath1583 we can choose @xmath292 small enough so that @xmath2855 since @xmath1916 is disjoint with the @xmath361-limit set of the flow @xmath286 and the latter is closed
then if @xmath312 is sufficiently small @xmath1917 is also disjoint with this @xmath361-limit set and , in particular , it does not contain any forward semi - orbit of @xmath286 .
hence we can apply lemma [ lem2.4.4 ] which in view of ( [ 2.2.9 ] ) implies that there exists @xmath988 such that for all small @xmath292 , @xmath2856 which is not less than @xmath1919 if we take @xmath1920 .
now , ( [ 2.5.20 ] ) and ( [ 2.5.21 ] ) produce ( [ 2.5.16 ] ) with @xmath1921 , and so ( [ 2.5.18 ] ) follows with such @xmath1905 .
this together with ( [ 2.5.14 ] ) yield that for any @xmath1438 we can choose sufficiently small @xmath1922 and then @xmath292 so that for all @xmath33 small enough @xmath2857 and ( [ 2.5.6 ] ) follows .
now assume that ( [ 2.5.7 ] ) holds true for some @xmath338 .
recall , that @xmath218 implies that @xmath654 is a piece of an orbit of the flow @xmath286 .
since no @xmath203 satisfying @xmath2858}\,\mbox{dist } ( { { \gamma}}_t,\partial v)\leq a(x)/2\ ] ] can be such piece of an orbit we conclude by the lower semicontinuity of @xmath228 that @xmath1925 whenever ( [ 2.5.23 ] ) holds true for some @xmath1926 independent of @xmath654 ( but depending on @xmath20 ) .
hence , by ( [ 2.2.11 ] ) , @xmath2859 provided @xmath33 is small enough and ( [ 2.5.8 ] ) follows .
observe also that any @xmath298 with @xmath1928 and @xmath1929 should contain a piece which either belongs to some @xmath1930 or to @xmath1931 , as above , or to satify ( [ 2.5.23 ] ) . by ( [ 2.5.20 ] ) , ( [ 2.5.21 ] ) , and the above remarks it follows that @xmath1932 for such @xmath654 where @xmath1933 depends only on @xmath20 , and so @xmath1934 . if @xmath1937 then by ( [ 2.5.6 ] ) and ( [ 2.5.8 ] ) , @xmath2860 and , since @xmath1438 is arbitrary , ( [ 2.5.9 ] ) follows completing the proof of proposition [ prop2.5.1 ] .
now we will derive theorem [ thm2.2.5 ] from proposition [ prop2.5.1 ] .
assume , first , that @xmath353 .
then by lemma [ lem2.4.5 ] , @xmath1654 is finite in the whole @xmath317 .
moreover , since @xmath302 is an @xmath285-attractor the conditions of lemma [ lem2.4.5 ] are satisfied with some @xmath1667 as @xmath426 the same for all points of @xmath317 which yields the conditions of proposition [ prop2.5.1](i ) with @xmath1939 for any @xmath253 . hence , ( [ 2.5.3 ] ) and ( [ 2.5.4 ] ) hold true with @xmath1940 . since @xmath302 is an @xmath285-attractor of the flow @xmath286 and its basin contains
@xmath321 then the intersection of @xmath1941 with the @xmath361-limit set of @xmath286 is empty . by the definition of an @xmath285-attractor for any @xmath1583 there
exists an open set @xmath2861 such that @xmath2862 whenever @xmath295 and @xmath1947 .
hence , by the triangle inequality for the function @xmath282 and lemma [ lem2.4.5 ] for any set @xmath1942 , @xmath2863 if @xmath1937 then by lemma [ lem2.4.5 ] the right hand side of ( [ 2.5.26 ] ) equals @xmath2864 . assuming that @xmath1665 we can apply proposition [ prop2.5.1](ii ) with such @xmath1919 yielding ( [ 2.5.6 ] ) , ( [ 2.5.8 ] ) and since @xmath1583 is arbitrary ( [ 2.2.25 ] ) and ( [ 2.2.26 ] ) follow in this case .
if @xmath1945 then ( [ 2.2.26 ] ) is trivial and by ( [ 2.5.26 ] ) , @xmath1946 for any @xmath1947 and @xmath1948 , and so we can apply proposition [ prop2.5.1](ii ) with any @xmath1919 which sais that the left hand side in ( [ 2.5.9 ] ) equals @xmath187 , and so ( [ 2.2.25 ] ) holds true in this case , as well .
next , we establish ( [ 2.2.27 ] ) . for small @xmath1949 and large @xmath277 which will be specified later on set @xmath2865 and
define the event @xmath2866 then @xmath2867 if @xmath380 is sufficiently small then @xmath1954 is still contained in the basin of @xmath302 with respect to the flow @xmath286 , and so we can choose @xmath1153 ( depending only on @xmath380 ) so that @xmath1955 then for some @xmath988 , @xmath1956 and so if @xmath1957 and @xmath1958 then dist@xmath1959 for any @xmath1960 . relying on ( [ 2.2.11 ] ) and the markov property we obtain that for any @xmath2868 with @xmath1821 , @xmath2869 provided @xmath33 is small enough .
next , the same arguments which yield ( [ 2.5.22 ] ) and ( [ 2.5.24 ] ) together with the markov property enable us to conclude that if @xmath1438 is small enough then for any @xmath2868 with @xmath1821 , @xmath2870 applying ( [ 2.5.27])([2.5.29 ] ) we conclude that for sufficiently small @xmath1192 and any much smaller @xmath33 , @xmath2871 finally , ( [ 2.2.26 ] ) and ( [ 2.5.30 ] ) together with the chebyshev inequality yield that for @xmath1983 $ ] , each @xmath338 , a small @xmath1438 and any much smaller @xmath342 , @xmath2872 since @xmath1985 and we can choose @xmath1192 to be arbitrarily small , ( [ 2.5.31 ] ) yields ( [ 2.2.27 ] ) . in order to complete the proof of theorem [ thm2.2.5 ] we have to derive ( [ 2.2.28 ] ) .
if @xmath1986 then there is nothing to prove , so we assume that @xmath335 is a proper subset of @xmath318 and in this case , clearly , @xmath1665 . since @xmath1987 is compact and disjoint with @xmath335 which is also compact then by the lower semicontinuity of @xmath1659 established in lemma [ lem2.4.5](iii ) it follows that @xmath1988 for some @xmath1438 and all @xmath348 . then by ( [ 2.5.26 ] ) , @xmath1989 for any @xmath1990 and @xmath1991
hence , applying proposition [ prop2.5.1 ] we obtain that @xmath2873 and @xmath2874 for some @xmath1019 and all @xmath33 small enough yielding ( [ 2.2.28 ] ) and completing the proof of theorem [ thm2.2.5 ] .
in this section we will prove theorem [ thm1.2.7 ] relying , again , on proposition [ prop1.7.1 ] together with `` markov property type '' arguments and at the end of the proof we will apply even some rough `` strong markov property type '' arguments in order to deal with subsequent transitions between basins of attractors .
in view of ( [ 1.2.27 ] ) and lemma [ lem1.6.2]i any curve @xmath298 starting at @xmath1994 and ending at @xmath1995 can be extended into each @xmath1996 with arbitrarily small increase in its @xmath285-functional . hence , @xmath1997 where @xmath1998 .
let @xmath1999 be an open ball of radius at least @xmath2000 centered at the origin of @xmath16 . by assumption [ ass1.2.6 ] the slow motion @xmath420 can not exit @xmath1999 provided @xmath2001 and @xmath261 .
furthermore , it is clear that @xmath1999 contains the @xmath361-limit set of the averaged flow @xmath286 .
assumption [ ass1.2.6 ] enables us to deal only with restricted basins @xmath2002 and though the boundaries @xmath2003 of @xmath2004 may include now parts of the boundary @xmath2005 of @xmath1999 it makes no difference since @xmath88 can not reach @xmath2005 if it starts in @xmath1999 .
set @xmath2006 where @xmath253 is small enough .
we claim that in view of ( [ 1.2.27 ] ) each @xmath383 satisfies conditions of proposition [ prop1.7.1](i ) for any @xmath1438 with @xmath2007 and some @xmath2008 depending on @xmath1192 .
indeed , set @xmath2009 in view of ( [ 1.2.9 ] ) and ( [ 1.2.27 ] ) there exists @xmath907 such that if @xmath312 is small enough and @xmath2010 we can construct a curve @xmath2011 with @xmath2012 , @xmath2013 for some @xmath2014 $ ] and @xmath2015 where @xmath2016 . since @xmath2017 is the basin of @xmath419 there exists @xmath2018 such that @xmath2019 and extending @xmath2020 by the piece of the orbit of @xmath286 we obtain a curve @xmath2021 starting at @xmath611 , entering @xmath2022 and satisfying @xmath2023 .
hence , for @xmath2010 the condition ( [ 1.7.2 ] ) holds true with @xmath2024 and @xmath2025 . since the @xmath361-limit set of the flow @xmath286
is contained in @xmath2026 it follows from assumption [ ass1.2.6 ] and compactness considerations that there exists @xmath2027 such that for any @xmath2028 we can find @xmath2029 $ ] with @xmath2030 . if @xmath2031 then we take @xmath2032 $ ] to satisfy ( [ 1.7.2 ] ) for @xmath2024 and @xmath2033 .
if @xmath2034 then we extend the curve @xmath2035 $ ] as in the above argument which yields a curve @xmath2036 starting at @xmath611 , ending in some @xmath2037 and having its @xmath285-functional not exceeding @xmath2038 . finally , in the same way as in the proof of theorem [ thm1.2.5 ] for any @xmath1438 there exists @xmath2039 such that whenever @xmath2040 we can construct @xmath2041 such that ( [ 1.7.2 ] ) holds true with @xmath2042 and @xmath2043 and , moreover , dist@xmath2044 for some @xmath2045 and @xmath2046 with @xmath2047 .
then in the same way as above we can extend @xmath2020 to some @xmath2048 so that @xmath2049 for some @xmath1373 as above , @xmath2050 and @xmath2051 which gives ( [ 1.7.2 ] ) for all @xmath2052 with @xmath2053 provided @xmath312 is small enough .
hence , proposition [ prop1.7.1](i ) yields the estimates ( [ 1.7.3 ] ) and ( [ 1.7.4 ] ) for @xmath2054 in place of @xmath2055 with @xmath2056 . in order to obtain the corresponding bounds in the other direction
observe that in view of ( [ 1.2.27 ] ) , @xmath2057 since @xmath2058 is contained in the basin of @xmath382 we can apply to @xmath2059 the same estimates as in theorem [ thm1.2.5 ] which together with ( [ 1.8.2 ] ) and the fact that the exit time of @xmath88 from @xmath2059 is smaller than its exit time from @xmath383 provide the remaining bounds yielding ( [ 1.2.28 ] ) and ( [ 1.2.29 ] ) . next , we derive ( [ 1.2.30 ] ) similarly to ( [ 1.2.23 ] ) but taking into account that @xmath373 may contain parts of the @xmath361-limit set of the flow @xmath286 which allows the slow motion @xmath88 to stay long time near these boundaries . still , set @xmath2060 using the same arguments as above we conclude that for any @xmath292 there exists @xmath2018 such that whenever @xmath2061 we can construct @xmath1797 with @xmath2062 and @xmath2063 .
this together with ( [ 1.7.21 ] ) and assumption [ ass1.2.6 ] yield that for any disc @xmath2064 , @xmath2065 for some @xmath2066 and all small @xmath33 . set @xmath2067 @xmath2068 and @xmath2069 where @xmath312 is much smaller than @xmath1192 . then proceeding similarly to the proof of ( [ 1.2.23 ] ) as in ( [ 1.7.40])([1.7.43 ] ) above we arrive at ( [ 1.2.30 ] ) . next , we obtain ( [ 1.2.31 ] ) relying on additional assumptions specified in the statement of theorem [ thm1.2.7 ] .
let @xmath2070 be the same as above and @xmath2071 . since @xmath382 is an @xmath285-attractor it follows from lemma [ lem1.6.5](i ) that @xmath303 and @xmath2072 coincide with the same function @xmath2073 and the same ( in general , may be empty ) set @xmath2074 , respectively , for all @xmath2075 .
by lemma [ lem1.6.2](i ) , our assumption that @xmath19 is complete on @xmath395 implies that @xmath2076 is continuous in a neighborhood of @xmath395 , and so @xmath2074 is a nonempty compact set .
since we assume that @xmath2077 is the unique index @xmath1373 for which @xmath2078 then by ( [ 1.2.27 ] ) , @xmath2079 observe that if @xmath2080 is an @xmath285-compact then either @xmath2081 or @xmath2082 .
denote by @xmath2083 the @xmath361-limit set of the averaged flow @xmath286 .
since @xmath2084 consists of a finite number of @xmath285-compacts it follows that @xmath2085 by the continuity of @xmath2076 in @xmath2086 there exists @xmath988 such that @xmath2087 these considerations enable us to construct a connected open set @xmath1762 with a piecewise smooth boundary @xmath2088 such that @xmath2089 and for @xmath2090 and some @xmath2091 , @xmath2092 provided @xmath2093 .
the idea of this construction is that if @xmath2094 then the slow motion should exit @xmath1762 through the part @xmath1652 of its boundary . somewhat similarly to the proof of proposition [ prop1.7.1](ii ) we will show that for `` most '' initial conditions @xmath18 this can only occur after the time @xmath2095 and , on the other hand , we conclude from ( [ 1.2.29 ] ) that for `` most '' initial conditions @xmath18 the exit time @xmath2054 does not exceed @xmath2096 .
let @xmath2097 be a sufficiently small open neighborhood of @xmath2074 so that , in particular , @xmath2098 and set @xmath2099 for each disc @xmath2100 we can write @xmath2101 where @xmath2102 for some small @xmath1438 , @xmath2103 $ ] , @xmath2104 \big\}$ ] for a sufficiently small @xmath292 , @xmath2105 , @xmath2106\big\}$ ] , @xmath2107 , and @xmath2108 . observe that @xmath2109 satisfies conditions of proposition [ prop1.7.1](i ) with arbitrarily small @xmath2110 , so similarly to ( [ 1.7.21 ] ) ( and taking into account lemma [ lem1.3.6 ] ) we can estimate @xmath2111 similarly to the proof of proposition [ prop1.7.1](ii ) we obtain also that @xmath2112 where we , first , choose @xmath312 small and then @xmath1153 large enough .
next , we estimate @xmath2113 for @xmath2114 by the following markov property type argument .
let @xmath2115 and choose a maximal @xmath2116-separated set @xmath771 in @xmath601 .
let @xmath2117 , @xmath2118 and @xmath2119 .
assume that dist@xmath2120 for any @xmath2121 .
by lemma [ lem1.6.2](i ) , @xmath303 is continuous in @xmath611 when @xmath611 belong to a sufficiently small neighborhood of @xmath395 which together with the definition of @xmath285-compacts yields that @xmath2122 for any @xmath2123 , provided @xmath312 is small enough .
observe that @xmath2124 for any @xmath2125 , provided @xmath33 is sufficiently small .
these together with the arguments similar to the proof of proposition [ prop1.7.1](ii ) yield the estimate @xmath2126 for all @xmath33 small enough .
since @xmath2127 are disjoint for different @xmath2128 we obtain by lemma [ lem1.3.6 ] , @xmath2129 where @xmath2130 . in a similar way we obtain that for each disc @xmath2131 , @xmath2132 provided @xmath33 is small enough . by ( [ 1.8.7])([1.8.9 ] ) together with lemma [ lem1.3.6 ] , @xmath2133 provided @xmath2114 and @xmath33 is small enough . summing in @xmath2134 and @xmath446 we obtain from ( [ 1.8.4])([1.8.6 ] ) and ( [ 1.8.10 ] ) that for a small @xmath1192 and all sufficiently small @xmath33 , @xmath2135 taking discs @xmath601 to be small balls on the ( extended ) local unstable manifolds @xmath2136 and using the fubini theorem as before we obtain ( [ 1.8.11 ] ) for @xmath54 and @xmath145 in place of @xmath653 and @xmath601 , respectively . on the other hand ,
employing proposition [ prop1.7.1](i ) we derive that @xmath2137 for some @xmath1019 and all @xmath33 small enough which together with ( [ 1.8.11 ] ) considered for @xmath54 and @xmath145 in place of @xmath653 and @xmath601 yield ( [ 1.2.31 ] ) . in order to complete the proof of theorem [ thm1.2.7 ] it remains to derive ( [ 1.2.32 ] ) and ( [ 1.2.33 ] ) .
both statements hold true for @xmath1213 in view of ( [ 1.2.29 ] ) and ( [ 1.2.31 ] ) but , in fact , we will use them as the induction base with @xmath653 and @xmath601 in place of @xmath54 and @xmath145 where @xmath622 and @xmath2138 which holds true in view of ( [ 1.8.4])([1.8.11 ] ) together with the corresponding form of proposition [ prop1.7.1 ] . for
such @xmath601 and @xmath409 set @xmath2139 and @xmath2140 as the induction hypotesis we assume that for any @xmath340 there exist @xmath2141 and @xmath1019 such that for all small @xmath33 , @xmath2142 set @xmath2143 where @xmath604 is the same as in ( [ 1.2.15 ] ) so that if @xmath2144 then @xmath2145.\ ] ] choose also @xmath2146 so that for any @xmath2147 , @xmath2148 let @xmath2149 be a maximal @xmath2150-separated set where @xmath2151 as before
. set @xmath2152 then for @xmath33 small enough , @xmath2153 we claim that there exists @xmath1438 such that if @xmath2154 then for all small @xmath33 , @xmath2155 indeed , let @xmath2156\,\,\mbox{and}\,\,{{\gamma}}_t\not\in u_{{{\delta}}/2 } ( { { \mathcal o}}_{{{\iota}}_n(i)})\,\,\mbox{for all}\,\ , t\in[0,na]\big\}$ ] .
then by ( [ 1.8.14 ] ) and the lower semicontinuiti of the functional @xmath2157 we obtain that @xmath2158 since by lemma [ lem1.3.2](ii ) and ( iii ) for any @xmath2159 the distance @xmath2160 has the order of @xmath33 we conclude from ( [ 1.8.13 ] ) and ( [ 1.8.16 ] ) that for each @xmath2161 , @xmath2162 hence , by ( [ 1.7.11 ] ) and lemma [ lem1.3.6 ] it follows that for any @xmath2163 and all @xmath33 small enough , @xmath2164 and since @xmath2165 are disjoint for different @xmath2166 we apply lemma [ 1.3.6 ] once more and obtain ( [ 1.8.15 ] ) .
set @xmath2167 . applying ( [ 1.8.12 ] ) with @xmath1213 to each @xmath2168 and using lemma [ lem1.3.6 ]
we derive also that @xmath2169 for some @xmath1438 and all small @xmath33 . by ( [ 1.8.17 ] ) we can write @xmath2170 observe that for any finite measure @xmath45 , measurable sets @xmath2171 and integers @xmath2172 , @xmath2173 which follows applying @xmath2174 to @xmath2175 and @xmath2176 for @xmath2177 . applying ( [ 1.8.19 ] ) for @xmath2178 and
@xmath2179 it follows from ( [ 1.8.15 ] ) that @xmath2180 i.e. for any @xmath2181 , @xmath2182 this together with ( [ 1.8.12 ] ) and ( [ 1.8.18 ] ) complete the induction step and proves ( [ 1.2.32 ] ) and ( [ 1.2.33 ] ) for @xmath601 and @xmath653 in place of @xmath145 and @xmath54 . finally , as before we complete the proof of theorem [ thm1.2.7 ] by choosing discs @xmath601 to be small balls on the ( extended ) local unstable manifolds @xmath2183 which together with the fubini theorem enables us to extend the estimates to @xmath145 and @xmath54 as required in ( [ 1.2.32 ] ) and ( [ 1.2.33 ] ) . in this section
we will prove theorem [ thm2.2.7 ] relying , again , on proposition [ prop2.5.1 ] together with markov and strong markov property of the markov process @xmath2875 .
in view of ( [ 2.2.31 ] ) and lemma [ lem2.4.2](i ) any curve @xmath298 starting at @xmath1994 and ending at @xmath1995 can be extended into each @xmath1996 with arbitrarily small increase in its @xmath285-functional .
hence , @xmath2876 where @xmath1998 .
let @xmath1999 be an open ball of radius at least @xmath2000 centered at the origin of @xmath16 . by assumption [ ass2.2.6 ]
the slow motion @xmath420 can not exit @xmath1999 provided @xmath2001 and @xmath261 .
furthermore , it is clear that @xmath1999 contains the @xmath361-limit set of the averaged flow @xmath286 .
assumption [ ass2.2.6 ] enables us to deal only with restricted basins @xmath2002 and though the boundaries @xmath2003 of @xmath2004 may include now parts of the boundary @xmath2005 of @xmath1999 it makes no difference since @xmath88 can not reach @xmath2005 if it starts in @xmath1999 .
set @xmath2006 where @xmath253 is small enough .
we claim that in view of ( [ 2.2.31 ] ) each @xmath383 satisfies conditions of proposition [ prop2.5.1](i ) for any @xmath1438 with @xmath2007 and some @xmath2008 depending on @xmath1192 .
indeed , set @xmath2009 in view of ( [ 2.2.31 ] ) and lemma [ lem2.4.1 ] there exists @xmath907 such that if @xmath312 is small enough and @xmath2010 we can construct a curve @xmath2011 with @xmath2877 , @xmath2878 for some @xmath2014 $ ] and @xmath2015 . since @xmath2017 is the basin of @xmath419 there exists @xmath2018 such that @xmath2019 and extending @xmath2020 by the piece of the orbit of @xmath286 we obtain a curve @xmath2021 starting at @xmath611 , entering @xmath2022 and satisfying @xmath2879 .
hence , for @xmath2010 the condition ( [ 2.5.2 ] ) holds true with @xmath2024 and @xmath2880 . since the @xmath361-limit set of the flow @xmath286
is contained in @xmath2026 it follows from assumption [ ass2.2.6 ] and compactness considerations that there exists @xmath2027 such that for any @xmath2028 we can find @xmath2029 $ ] with @xmath2030 . if @xmath2031 then we take @xmath2032 $ ] to satisfy ( [ 2.5.2 ] ) for @xmath2024 and @xmath2033 .
if @xmath2034 then we extend the curve @xmath2035 $ ] as in the above argument which yields a curve @xmath2036 starting at @xmath611 , ending in some @xmath2037 and having its @xmath285-functional not exceeding @xmath2881 . finally , in the same way as in the proof of theorem [ thm2.2.5 ] for any @xmath1438 there exists @xmath2039 such that whenever @xmath2040 we can construct @xmath2041 such that ( [ 2.5.2 ] ) holds true with @xmath2042 and @xmath2043 and , moreover , dist@xmath2044 for some @xmath2045 and @xmath2046 with @xmath2047 .
then in the same way as above we can extend @xmath2020 to some @xmath2048 so that @xmath2049 for some @xmath1373 as above , @xmath2050 and @xmath2882 which gives ( [ 2.5.2 ] ) for all @xmath2052 with @xmath2053 provided @xmath312 is small enough .
hence , proposition [ prop2.5.1](i ) yields the estimates ( [ 2.5.3 ] ) and ( [ 2.5.4 ] ) for @xmath2054 in place of @xmath2055 with @xmath2056 .
in order to obtain the corresponding bounds in the other direction observe that in view of ( [ 2.2.31 ] ) , @xmath2883 since @xmath2058 is contained in the basin of @xmath382 we can apply to @xmath2059 the same estimates as in theorem [ thm2.2.5 ] which together with ( [ 2.6.2 ] ) and the fact that the exit time of @xmath88 from @xmath2059 is smaller than its exit time from @xmath383 provide the remaining bounds yielding ( [ 2.2.32 ] ) and ( [ 2.2.33 ] ) . next , we derive ( [ 2.2.34 ] ) similarly to ( [ 2.2.27 ] ) but taking into account that @xmath373 may contain parts of the @xmath361-limit set of the flow @xmath286 which allows the slow motion @xmath88 to stay long time near these boundaries .
still , set @xmath2060 using the same arguments as above we conclude that for any @xmath292 there exists @xmath2018 such that whenever @xmath2061 we can construct @xmath1797 with @xmath2062 and @xmath2063 .
this together with ( [ 2.5.12 ] ) and assumption [ ass2.2.6 ] yield that @xmath2884 for some @xmath2066 and all small @xmath33
. set @xmath2885 @xmath2886 and @xmath2069 where @xmath312 is much smaller than @xmath1192 . then proceeding similarly to the proof of ( [ 2.2.27 ] ) as in ( [ 2.5.28])([2.5.31 ] ) above we arrive at ( [ 2.2.34 ] ) .
next , we obtain ( [ 2.2.35 ] ) relying on additional assumptions specified in the statement of theorem [ thm2.2.7 ] .
let @xmath2070 be the same as above and @xmath2071 . since @xmath382 is an @xmath285-attractor it follows from lemma [ lem2.4.5](i ) that @xmath303 and @xmath2072 coincide with the same function @xmath2073 and the same ( in general , may be empty ) set @xmath2074 , respectively , for all @xmath2075 .
by lemma [ lem2.4.2](i ) , our assumption that @xmath19 is complete on @xmath395 implies that @xmath2076 is continuous in a neighborhood of @xmath395 , and so @xmath2074 is a nonempty compact set .
since we assume that @xmath2077 is the unique index @xmath1373 for which @xmath2078 then by ( [ 2.2.31 ] ) , @xmath2079 observe that if @xmath2080 is an @xmath285-compact then either @xmath2081 or @xmath2082 .
denote by @xmath2083 the @xmath361-limit set of the averaged flow @xmath286 .
since @xmath2084 consists of a finite number of @xmath285-compacts it follows that @xmath2085 by the continuity of @xmath2076 in @xmath2086 there exists @xmath988 such that @xmath2087 these considerations enable us to construct a connected open set @xmath1762 with a piecewise smooth boundary @xmath2088 such that @xmath2089 and for @xmath2090 and some @xmath2091 , @xmath2887 provided @xmath2093 .
the idea of this construction is that if @xmath2094 then the slow motion should exit @xmath1762 through the part @xmath1652 of its boundary . somewhat similarly to the proof of proposition [ prop2.5.1](ii ) we will show that `` most likely '' this can only occur after the time @xmath2095 and , on the other hand , we conclude from ( [ 2.2.33 ] ) that except for small probability the exit time @xmath2054 does not exceed @xmath2096 .
let @xmath2097 be a sufficiently small open neighborhood of @xmath2074 so that , in particular , @xmath2098 and set @xmath2099 then @xmath2888 where @xmath2102 for some small @xmath1438 , @xmath2103 $ ] , @xmath2889 \big\}$ ] for a sufficiently small @xmath292 , @xmath2890 , @xmath2891\big\}$ ] , @xmath2892 , and @xmath2893 .
observe that @xmath2109 satisfies conditions of proposition [ prop2.5.1](i ) with arbitrarily small @xmath2110 , so similarly to ( [ 2.5.12 ] ) we can estimate @xmath2894 similarly to the proof of proposition [ prop2.5.1](ii ) we obtain also that @xmath2895 where we , first , choose @xmath312 small and then @xmath1153 large enough .
next , relying on the markov property and the arguments similar to the proof of proposition [ prop2.5.1](ii ) we estimate @xmath2896 provided @xmath2114 and @xmath33 is small enough . summing in @xmath2134 and @xmath446
we obtain from ( [ 2.6.4])([2.6.7 ] ) that for a small @xmath1192 and all sufficiently small @xmath33 , @xmath2897 employing proposition [ prop2.5.1](i ) we derive that @xmath2898 for some @xmath1019 and all @xmath33 small enough which together with ( [ 2.6.8 ] ) yield ( [ 2.2.35 ] ) . in order to complete the proof of theorem [ thm2.2.7 ] it remains to derive ( [ 2.2.36 ] ) and ( [ 2.2.37 ] ) .
both statements hold true for @xmath1213 in view of ( [ 2.2.33 ] ) and ( [ 2.2.35 ] ) and we proceed by induction .
set @xmath2899 and @xmath2900 as the induction hypotesis we assume that for any @xmath340 there exist @xmath2141 and @xmath1019 such that for all small @xmath33 , @xmath2901 by ( [ 2.2.35 ] ) and the strong markov property @xmath2902 which implies ( [ 2.2.37 ] ) . similarly , by ( [ 2.2.33 ] ) and the strong markov property @xmath2903 proving ( [ 2.2.36 ] ) and completing the proof of theorem [ thm2.2.7 ] . finally , we prove theorem [ thm2.2.8 ] employing the arguments similar to 2 and 3 in ch .
6 of @xcite .
namely , in order to obtain the upper bound in ( [ 2.2.38 ] ) observe that for any @xmath2904 there are @xmath2905 such that if @xmath2906 and a curve @xmath298 satisfies @xmath2907 and dist@xmath2908 then @xmath2909 . using lemma
[ lem2.4.4 ] and the upper bound of large deviations ( [ 2.2.11 ] ) we can choose @xmath2910 such that for all small @xmath33 and any @xmath2911 , @xmath2912 any path of @xmath88 starting at a point of @xmath2913 and reaching @xmath2914 at time @xmath2915 either spends the time @xmath2916 without touching the set @xmath2917 or arrives at @xmath2918 during the time @xmath2916 .
in the latter case @xmath2919 and by ( [ 2.2.11 ] ) and ( [ 2.6.12 ] ) for any @xmath1345 with @xmath2920 , all @xmath33 small enough and @xmath2046 , @xmath2921 for some @xmath2922 independent of @xmath33 .
any path of @xmath2923 starting at @xmath2924 and reaching @xmath2914 at the time @xmath2915 must first hit at time @xmath2925 the set @xmath2913 , and so ( [ 2.6.13 ] ) together with the markov property yields the upper bound in ( [ 2.2.38 ] ) . in order to derive the lower bound in ( [ 2.2.38 ] ) observe that using the definition of @xmath285-attractors and lemma [ lem2.4.3 ] ( similarly to the proof of lemma [ lem1.6.5](ii ) in part [ part1 ] and see also 2 in ch .
6 of @xcite ) we conclude that for any @xmath2904 there exists @xmath2926 such that if @xmath2606 then for any @xmath2927 there exists a curve @xmath298 such that @xmath2928 for @xmath2929 $ ] , @xmath2930 for @xmath2931 , @xmath2932 and , finally , @xmath2933 . then by ( [ 2.2.10 ] ) for all small @xmath342 , @xmath2934 for some @xmath2922 independent of @xmath33 which together with ( [ 2.6.13 ] ) yields ( [ 2.2.38 ] ) .
for readers convenience we start this section with the setup and necessary technical results from @xcite refering there for the corresponding proofs .
these results are similar to section [ sec1.3 ] and we refer the reader also to @xcite where more details of proofs can be found than in @xcite and though @xcite deals only with the continuous time case the corresponding discrete time proofs can be obtained , essentially , by simplification
. we will discuss below mainly the axiom a case since the corresponding proofs for expanding transformations can be obtained , essentially , by simplification of the same arguments , roughly speaking , by ignoring the stable direction . as in section [ sec1.3 ]
we will use the representations @xmath491 of vectors @xmath490 , the norms @xmath2184 and the distances @xmath497 and @xmath153 on @xmath11 and on @xmath10 , respectively .
it is known ( see @xcite ) that the hyperbolic splitting @xmath2185 over @xmath70 can be continuously extended to the splitting @xmath2186 over @xmath145 which is forward invariant with respect to @xmath2187 and satisfies exponential estimates ( [ 1.3.1 ] ) with a uniform in @xmath306 exponent @xmath2188 moreover , by @xcite ( see also @xcite ) we can choose these extensions so that @xmath510 and @xmath511 will be h " older continuous in @xmath512 and @xmath53 in @xmath20 in the corresponding grassmann bundle . actually , since @xmath145 is contained in the basin of each attractor @xmath70
, any point @xmath513 belongs to the stable manifold @xmath514 of some point @xmath515 ( see @xcite ) , and so we choose naturally @xmath510 to be the tangent space to @xmath516 at @xmath517 now each vector @xmath518 can be represented uniquely in the form @xmath2189 with @xmath2190 , @xmath521 , and @xmath523 . for each small @xmath526 set @xmath2191 and @xmath528 which are cones around @xmath453 and @xmath529 respectively .
similarly , we define @xmath2192 and @xmath528 which are cones around @xmath532 and @xmath533 respectively . the corresponding version of lemma [ lem1.3.1 ] is proved in @xcite and the discrete time versions of lemmas [ lem1.3.2 ] and [ lem1.3.3 ] follow in the same way as in @xcite .
let , again , @xmath612 be the set of all @xmath53 embedded @xmath613dimensional closed discs @xmath614 such that @xmath615 , @xmath616 and if @xmath617 then @xmath618 . for @xmath622 and @xmath2193 set @xmath2194 and let @xmath632 and @xmath633 be natural projections on the first and second factors , respectively .
the same proof as in @xcite yields the following discrete time version of proposition [ prop1.3.4 ] .
[ prop1.9.1 ] for any @xmath792 with @xmath793 large and @xmath794 small enough there exists a positive function @xmath795 satisfying ( [ 1.3.7 ] ) such that for any @xmath2195 @xmath2196 , @xmath799 , @xmath2197 @xmath615 and @xmath2198 we have @xmath2199 where @xmath2200 and @xmath806 .
next , observe that the results of section [ sec1.4 ] above are so general that they work both for the continuous and the discrete time case .
now , we will discuss the discrete time version of lemma [ 1.5.1 ] .
[ lem1.9.1 ] let @xmath1100 @xmath1101 @xmath1102 @xmath1103 , @xmath1104 @xmath1105 and @xmath2201-[{{\varepsilon}}^{-1}t_{j-1}])^{-1 } \sum_{[{{\varepsilon}}^{-1}t_{j-1}]\leq k\leq [ { { \varepsilon}}^{-1}t_j]}b(x , y^{{\varepsilon}}_v(k)).\ ] ] set for @xmath2202,\ , n\in{{\mathbb n}}$ ] , @xmath2203 and for @xmath2204 @xmath2205 then @xmath2206 @xmath2207 and @xmath2208 where , recall , @xmath1112 and @xmath1113 if @xmath1114 .
the proof of ( [ 1.9.5 ] ) and ( [ 1.9.6 ] ) is strightforward using the definitions ( [ 1.9.2])([1.9.4 ] ) in the same way as the proof of ( [ 1.5.2 ] ) and ( [ 1.5.3 ] ) only the integrals in the latter case should be replaced by the corresponding sums in the former one .
the estimate ( [ 1.9.7 ] ) follows in the same way as ( [ 1.5.4 ] ) only the use of the standard gronwall inequality in the latter proof should be replaced by the discrete time version of the gronwall inequality as in lemma 4.20 of @xcite .
now the proof of the discrete time version of proposition [ prop1.5.2 ] and of the remaining part of the proof of large deviations bounds ( [ 1.2.16 ] ) and ( [ 1.2.17 ] ) for the discrete time case proceeds almost verbatim as the corresponding continuous time proofs in section [ sec1.5 ] .
observe that in the discrete time case the functionals @xmath228 are given again by ( [ 1.2.13 ] ) with @xmath184 defined by ( [ 1.2.8 ] ) where @xmath2209 and @xmath2210 is given by ( [ 1.2.34 ] ) .
the property of @xmath263-functionals described in lemma [ lem1.6.1 ] follows directly in the discrete time case via conjugation since we do not have to deal with the time change here .
other auxiliary results of section [ sec1.6 ] are derived in the discrete time case exactly in the same way as there .
the proof of the discrete time versions of theorems [ thm1.2.5 ] and [ thm1.2.7 ] under the corresponding assumptions goes through exactly in the same way as its continuous time counterpart in section [ sec1.6 ] yielding the assertion of theorem [ thm1.2.10 ] .
next , we exhibit computations demonstrating a discrete time version of theorem [ thm1.2.7 ] for simple examples . the maps @xmath106 in both examples have the form @xmath2211 where @xmath2212 and @xmath2213 $ ] but by identifying the end points of the unit interval we view @xmath106 as expanding maps of the circle @xmath2214 . the function @xmath19 from ( [ 1.1.10 ] )
is given in the first example by @xmath2215 hence , we are dealing here with the maps @xmath2216 defined by @xmath2217 all maps @xmath106 preserve the normalized lebesgue measure @xmath2218 on @xmath2214 and it is the srb measure @xmath69 for each @xmath106 in this simple case . the averaged equation ( [ 1.1.11 ] ) for
@xmath2219 has here the form @xmath2220 where @xmath2221 .
the one dimensional vector field @xmath47 has three attracting fixed points @xmath2222 and two repelling fixed points 1 and @xmath2223 .
in order to apply the discrete time version of theorem [ thm1.2.7 ] ( i.e. theorem [ thm1.2.10 ] ) to this example we have to verify that @xmath19 is complete at the fixed points @xmath2224 of the averaged system .
since at these points @xmath106 coincides with the map @xmath2225(mod 1 ) we can take the periodic orbits @xmath2226 and @xmath2227 of the latter and notice that the average of @xmath2228 along the former is @xmath2229 and along the latter @xmath2230 which yields completness of @xmath19 at zeros of @xmath2231 . according to the corresponding part of theorem [ thm1.2.10 ] which is a discrete time version of theorem [ thm1.2.7 ]
the transitions between @xmath2232 and @xmath2233 are determined by @xmath2234 which are obtained via the functionals @xmath2235 given by ( [ 1.2.13 ] ) .
even here these functionals are not easy to compute though their main ingredients the functionals @xmath184 from ( [ 1.2.8 ] ) are given now by the simple formula @xmath2236 and the set of @xmath106-invariant measures can be reasonably described since all @xmath106 s are conjugate to the simple map @xmath2237 .
we plot below the histogram of a single orbit of the slow motion @xmath2238 with @xmath2239 and the initial values @xmath2240 .
the histogram shows that most of the points of the orbit stay near the attractors @xmath2241 and @xmath2233 and @xmath2242 hops between basins of attraction of these points .
the form of the histogram indicates ( according to theorem [ thm1.2.7 ] ) the equality @xmath2243 and in this case theorem [ thm1.2.7 ] ( or its discrete time version ) can not specify whether the slow motion exits from the basin of @xmath2244 to the basin of @xmath2245 or to the basin of @xmath2233 .
observe that theorem [ thm1.2.7 ] is an asymptotical as @xmath50 result and it takes an exponential in @xmath5 time for a typical orbit to exit from the basin of one attractor and to hop to the basin of another one .
hence , the computations should be done for small @xmath33 and exponentially long in @xmath5 orbits which is time consuming , so we put a big coefficient in front of @xmath2246 which makes this exponent smaller .
of course , it is hard to be absolutely sure that @xmath33 in our computations is small enough and the number of iterates is large enough to demonstrate faithfully the real situation in this case but we found that our histograms are rather robust , for instance , their shapes have the same form for @xmath2239 when the number of iterates ranges from @xmath2247 to , at least , @xmath2248 and various initial conditions were checked , as well .
our second example differs from the first one only in @xmath19 which is given now by @xmath2249 here the averaged system has the same attracting fixed points @xmath2250 but one of two repelling fixed points moves from @xmath2223 to @xmath2251 .
this makes the basin of attraction of @xmath2252 smaller while the left interval of the basin of attraction of @xmath2253 becomes larger .
the latter leads to the inequality @xmath2254 which according to the discrete time version of theorem [ thm1.2.7 ] makes it more difficult for the slow motion to exit to the left from the basin of @xmath2244 than to the right . as in the first example in order to apply the latter result we have to check that @xmath19 is complete at all zeros of @xmath2231 but since we did this already for all integer points it remains to verify completness only for @xmath2255 which follows since @xmath2228 equals 1 and @xmath2223 at two fixed points @xmath2256 and @xmath2257 of @xmath2258 , respectively . in the histogram here
we plot @xmath2238 with @xmath2239 and the initial values @xmath2259 . in compliance with the discrete time version of theorem [ thm1.2.7 ]
the histogram demonstrates that the slow motion leaves the basin of @xmath2233 and after arriving at the basin of @xmath2244 it exits mostly to the basin of @xmath2245 , and so the slow motion hops mostly between basins of @xmath2245 and @xmath2244 staying most of the time in small neighborhoods of these points .
theorem [ thm2.2.10 ] follows by a slight modification ( essentially , by simplification ) of the proof of theorems [ thm2.2.2 ] , in particular , the standard gronwall inequality required in the proof of lemma [ lem2.3.4 ] should be replaced by its discrete time version from @xcite .
we have also to check that ( [ 2.2.41 ] ) holds true here which is easier to do than in the continuous time case .
indeed , @xmath2935 where @xmath2936 , @xmath2937 , @xmath2938 . by ( [ 2.2.1 ] ) ,
@xmath2939 since all @xmath2940 which is compact and @xmath2941 , i.e. all @xmath1626 stay also in a compact set , we obtain from our assumptions on transition densities that for all @xmath2942 and @xmath2943 , @xmath2944 for some @xmath1436 independent of @xmath2945 and @xmath20 staying in a compact set .
hence , @xmath2946 where @xmath2947 is obtained from @xmath2948 by replacing @xmath242 in the latter by @xmath2949 .
it follows from standard facts on principal eigenvalues of positive operators ( see , for instance , @xcite and @xcite ) that uniformly in @xmath432 and @xmath2538 the limit @xmath2950 exists and it satisfies the conditions of assumption [ ass2.2.9 ] , and so taking the logarithm in the ineguality above and dividing by @xmath2539 we arrive at ( [ 2.2.41 ] ) . theorem [ thm2.2.12 ] also follows by a slight modification of proofs of theorems [ thm2.2.5 ] and [ thm2.2.7 ] , only we have to derive a result which replaces lemma [ lem2.4.1 ] providing required properties of @xmath263-functionals given by ( [ 2.2.44 ] ) . since
, without loss of generality , we can assume that @xmath2951 for some @xmath1436 and by ( [ 2.2.44 ] ) , @xmath2952 where the supremum is taken over positive continuous functions @xmath2545 .
then @xmath2953 it is easy to see from here that @xmath2793 if and only if @xmath2954 and the density @xmath711 is bounded .
hence , in this case , @xmath2955 since @xmath2956 for some @xmath2957 , we obtain @xmath2958 two last inequalities provide all properties of @xmath263-fuctionals which are needed in order to replace lemma [ 2.4.1 ] and to proceed with arguments of sections [ sec2.4][sec2.6 ] in the discrete time case .
theorem [ thm2.2.10 ] provides , in particular , an approximation of the slow motion by the averaged one in probability but , in general , we do not have convergence in ( [ 2.1.4 ] ) also with probability one ( see @xcite ) .
sometimes , we can derive this almost sure convergence from the upper large deviations bound estimating the derivative in @xmath33 of the slow motion as in the following example .
let @xmath368 be a bounded 1-periodic in @xmath18 function on @xmath2959 with bounded derivatives and let @xmath2960 be a sequence of independent identically distributed ( i.i.d . )
random variables .
define recursively @xmath2961 where @xmath2962 and @xmath2963 , @xmath2964
. then @xmath2965 set @xmath2966 which are sequences of bounded matrices and vectors .
taking into account the equalities @xmath2967 we obtain from ( [ 2.7.2 ] ) by induction ( with the agreement @xmath2968 ) that @xmath2969 since @xmath2970 and @xmath2971 are bounded we obtain that @xmath2972 for some @xmath1436 independent of @xmath446 and @xmath33 .
since @xmath368 is 1-periodic in @xmath18 we can replace the second equality in ( [ 2.7.1 ] ) by
@xmath2973 i.e. we consider now @xmath2974 evolving on the interval @xmath2975 $ ] with 0 and 1 identified which makes it the circle of radius @xmath2976 .
suppose that the distribution of @xmath2977 has a @xmath53 density @xmath2978 with respect to the lebesgue measure which is positive on @xmath2975 $ ] .
now we have the family of markov chains @xmath2979 with transition probabilities @xmath2980
thus we are in the framework of our main model satisfying assumption [ ass2.2.9 ] , and so the assertion of theorem [ thm2.2.10 ] holds true .
let @xmath2635 be the invariant measure of the markov chain @xmath2549 ( which is unique since the doeblin condition is satisfied here ) and assume that @xmath2981 which is , essentially , not a restriction since we always can consider @xmath2722 in place of @xmath368 .
this means that @xmath2982 and we derive from theorem [ thm2.2.10 ] that for any @xmath253 there exists @xmath2560 such that for all small @xmath33 , @xmath2983 set @xmath2723 then @xmath2725 and by the borel
cantelli lemma we obtain that there exists @xmath2726 finite with probability one so that for all @xmath2727 , @xmath2984 by ( [ 2.7.3 ] ) for @xmath2729 and @xmath2730 , @xmath2985 it follows that with probability one , @xmath2986 the conditions above can be relaxed a bit but this method will not already work if , for instance , the second equality in ( [ 2.7.1 ] ) is replaced by @xmath2987 since in this case the derivative @xmath2988 may grow exponentially in @xmath446 and , indeed , we show in @xcite that for the latter example there is no convergence with probability one in ( [ 2.7.10 ] ) provided @xmath253 is small enough .
next , we exhibit two examples of computations which demonstrate adiabatic transitions between attractors of the averaged system via the statistics of proportions of time the slow motion spends in basins of different attractors .
the fast motions @xmath2989 in both examples are given by the second equation in ( [ 2.7.1 ] ) where @xmath2960 are i.i.d .
random variables with the uniform distribution on @xmath2975 $ ] .
the slow motion @xmath2990 is given by the first equation in ( [ 2.7.1 ] ) where in the first example @xmath2991 and in the second example @xmath2992 the markov chains @xmath2549 preserve here the lebesgue measure on @xmath2975 $ ] which is the unique invariant measure for them , and so the averaged equation ( [ 2.1.6 ] ) for @xmath2993 has the right hand side @xmath2994 in the first case and , @xmath2995 in the second case .
the one dimensional vector field @xmath47 has three attracting fixed points @xmath2222 and two repelling fixed points 1 and @xmath2223 , while @xmath2996 has the same attracting fixed points but one repelling fixed point moves now from @xmath2223 to @xmath2251 making the basin of @xmath2252 smaller which makes it easier for the slow motion to escape from there .
it is easy to see that @xmath2997 and @xmath2998 are complete at the fixed points of the averaged system , and so theorem [ thm2.2.12 ] is applicable in this situation . according to the corresponding part of theorem [ thm2.2.12 ]
the transitions between @xmath2232 and @xmath2233 are determined by @xmath2234 which are obtained via the functionals @xmath2235 given by ( [ 2.2.9 ] ) but even here these functionals are not easy to compute .
the functionals @xmath2235 yield non classical variational problems and the effective ways of their computation remain for further research . in the first example we plot above the histogram with @xmath2999 intervals of a single orbit of the slow motion @xmath3000 with @xmath2239 and the initial values @xmath3001 .
the histogram shows that most of the points of the orbit stay near the attractors @xmath2241 and @xmath2233 and @xmath2242 hops between basins of attraction of these points .
the form of the histogram indicates the equality @xmath2243 , which follows also by the symmetry considerations , but in this case theorem [ thm2.2.12 ] can not specify whether the slow motion mostly exits from the basin of @xmath2244 to the basin of @xmath2245 or to the basin of @xmath2233 . in the second example the basin of attraction of @xmath2252 becomes smaller while the left interval of the basin of attraction of @xmath2253 becomes larger .
the latter leads to the inequality @xmath2254 which according to theorem [ thm2.2.12 ] makes it more difficult for the slow motion to exit to the left from the basin of @xmath2244 than to the right .
in the histogram below ( which has again @xmath2999 intervals ) we plot @xmath3000 with @xmath2239 and the initial values @xmath3002 . in compliance with theorem [ thm2.2.12 ]
the histogram demonstrates that the slow motion leaves the basin of @xmath2233 and after arriving at the basin of @xmath2244 it exits mostly to the basin of @xmath2245 , and so the slow motion hops mostly between basins of @xmath2245 and @xmath2244 staying most of the time in small neighborhoods of these points .
still , a complete rigorous explanation of these histograms even for our simple examples requires nontrivial additional arguments .
it is interesting to observe that these histograms have the same form as in section [ sec1.9 ] of part [ part1 ] where randomness is generated by the expanding ( chaotic ) map @xmath3003 instead of adding uniformly distributed random variables as we do it here .
the scheme for the stochastic resonance type phenomenon described below is a slight modification of the model suggested by m.freidlin ( cf .
@xcite ) and it can be demonstrated in the setup of three scale systems @xmath2260 @xmath2261 , @xmath2262 , @xmath2263 with initial conditions @xmath2264 , @xmath2265 and @xmath2266 .
we assume that @xmath2267 , @xmath2268 while @xmath2269 evolves on a compact @xmath12-dimensional @xmath13 riemannian manifold @xmath11 and the coefficients @xmath80 , @xmath19 , @xmath21 are bounded smooth vector fields on @xmath2270 , @xmath16 and @xmath11 , respectively , depending on other variables as parameters .
the solution of ( [ 1.10.1 ] ) determines the flow of diffeomorphisms @xmath2271 on @xmath2272 acting by @xmath2273 .
taking @xmath2274 we arrive at the ( unperturbed ) flow @xmath2275 acting by @xmath2276 where @xmath2277 is another family of flows given by @xmath2278 with @xmath2279 which are solutions of @xmath2280 it is natural to view the flow @xmath31 as describing an idealized physical system where parameters @xmath2281 , @xmath32 are assumed to be constants of motion while the perturbed flow @xmath2282 is regarded as describing a real system where evolution of these parameters is also taken into consideration but unlike the averaging setup ( [ 1.1.1 ] ) we have now two sets of parameters moving with very different speeds . set @xmath2283 @xmath2284 @xmath2285 and pass from ( [ 1.10.1 ] ) to the equations in the new time @xmath2286 assume that the equation ( [ 1.10.1 ] ) satisfy the assumptions similar to assumptions [ ass1.2.1 ] , [ ass1.2.2 ] , [ ass1.2.6 ] together with other corresponding conditions appearing in the setup of theorem [ thm1.2.7 ] ( with @xmath2287 in place of @xmath16 ) , in particular , that @xmath2288 form a compact set of flows in the @xmath13 topology with @xmath13 dependence on @xmath2289 and for all @xmath2289 they are axiom a flows in a neighborhood @xmath145 which contains a basic hyperbolic attractor @xmath2290 for @xmath2277 and @xmath145 itself is contained in the basin of each @xmath2290 . set @xmath2291 where @xmath2292 is the srb measure for @xmath2277 and let @xmath2293 be the solution of the averaged equation @xmath2294 first , we apply averaging and large deviations estimates in averaging from the previous section to two last equations in ( [ 1.10.3 ] ) freezing the slowest variable @xmath512 ( i.e. taking for a moment @xmath2295 ) .
namely , set @xmath2296 and @xmath2297 so that @xmath2298 suppose for simplicity that @xmath2299 ( i.e. both @xmath2300 and @xmath2301 are one dimensional ) and that the solution @xmath2302 of ( [ 1.10.5 ] ) has the limit set consisting of two attracting points @xmath2245 and @xmath2244 , which for simplicity we assume to be independent of @xmath512 , and a repelling fixed point @xmath2303 depending on @xmath512 and separating their basins . as an example of @xmath2304 we may have in mind @xmath2305
let @xmath2306 be the large deviations rate functional for the system ( [ 1.10.6 ] ) defined in ( [ 1.2.13 ] ) and set for @xmath2307 , @xmath2308 ( cf . with @xmath387 in theorem [ thm1.2.7 ] ) .
set @xmath2309 and assume that for all @xmath512 , @xmath2310 which means in view of the averaging principle ( see theorem [ thm1.2.3 ] and the following it discussion ) that @xmath2311 decreases ( increases ) while @xmath2312 stays close to @xmath2245 ( to @xmath2244 ) for `` most '' @xmath18 s with respect to the riemannian volume on @xmath11 restricted to @xmath145 .
the following statement suggests a `` nearly '' periodic behavior of the slowest motion .
[ conj1.10.1 ] suppose that there exist strictly increasing and decreasing functions @xmath2313 and @xmath2314 , respectively , so that @xmath2315 and @xmath2316 for some @xmath1019 while @xmath2317 for @xmath2318 .
assume that @xmath2319 and @xmath50 in such a way that @xmath2320 then for any @xmath2289 there exists @xmath119 so that the slowest motion @xmath2321 converges weakly ( as @xmath914 so that ( [ 1.10.10 ] ) holds true ) as a random process on the probability space @xmath2322 ( where @xmath2323 is the normalized riemannian volume on @xmath145 ) to a periodic function @xmath2324 , @xmath2325 with @xmath2326 the argument supporting this conjecture goes as follows .
set @xmath2327 , @xmath2328 and @xmath2329 which satisfy @xmath2330 since @xmath2331 moves much slower than @xmath2332 we can freeze the former and in place of ( [ 1.10.11 ] ) we can study ( [ 1.10.6 ] ) . applying the arguments of theorem [ thm1.2.7 ] to the pair @xmath2333 from ( [ 1.10.6 ] )
we conclude by ( [ 1.2.30 ] ) that the intermediate motion @xmath2334 most of the time stays very close to either @xmath2245 or @xmath2244 before it exits from the corresponding basin , and so in view of an appropriate averaging principle ( which follows , for instance , from theorem [ 1.2.3 ] ) on bounded time intervals the slowest motion @xmath2335 mostly stays close to the corresponding averaged motion determined by the vector fields @xmath2336 and @xmath2337 given by ( [ 1.10.4 ] ) . when @xmath2334 is close to @xmath2245 the slowest motion @xmath2335 decreases until @xmath2338 where @xmath2339 . in view of ( [ 1.2.29 ] ) and the scaling ( [ 1.10.10 ] ) between @xmath33 and @xmath380 ,
a moment later @xmath2340 becomes less than @xmath625 and @xmath2341 jumps immediately close to @xmath2244 .
there @xmath2342 , and so @xmath2335 starts to grow until it reaches @xmath2343 where @xmath2344 .
a moment later @xmath2345 becomes smaller than @xmath625 and in view of ( [ 1.2.29 ] ) the intermediate motion @xmath2334 jumps immediately close to @xmath2245 .
this leads to a nearly periodic behavior of @xmath2335 .
in order to make these arguments precise we have to deal here with an additional difficulty in comparison with the two scale setup considered in previous sections since now the large deviations @xmath285-functionals from theorem [ thm1.2.3 ] and the @xmath282-functions describing adiabatic fluctuations and transitions of theorems [ thm1.2.5 ] and [ thm1.2.7 ] depend on another very slowly changing parameter .
still , the technique of sections [ sec1.7 ] and [ sec1.8 ] above applied on time intervals where changes in the @xmath512-variable can be neglected should work here but the details of this approach have not been worked out yet . on the other hand ,
when the fast motion @xmath2269 does not depend on the slow motions , i.e. when the coefficient @xmath21 in ( [ 1.10.1 ] ) depend only on the coordinate @xmath18 ( but not on @xmath512 and @xmath20 ) , then the above arguments can be made precise without much effort .
indeed , we can obtain estimates for transition times @xmath2346 and @xmath2347 of @xmath2348 between neighborhoods of @xmath2245 and @xmath2244 as in theorem [ thm1.2.7 ] applying the latter to @xmath2349 and @xmath2350 from ( [ 1.10.6 ] ) with freezed @xmath512-variable .
this is possible since the method of proposition [ prop1.7.1 ] requires us to make large deviations estimates , essentially , only for probabilities @xmath2351 , i.e. on bounded time intervals , and then combine them with the markov property type arguments . during such times the slowest motion @xmath2300 can move only a distance of order @xmath2352 . thus freezing @xmath512 and using the gronwall inequality for the equation of @xmath2 in order to estimate the resulting error we see that the latter is small enough for our purposes .
observe , that it would be much more difficult to justify freezing @xmath512 in the coefficient @xmath21 of @xmath1 , if we allow the latter to depend on @xmath512 , since a strightforward application of the gronwall inequality there would yield an error estimate of an exponential in @xmath5 order which is comparable with @xmath2353 .
still , it may be possible to take care about the general case using methods of sections [ sec1.4 ] and [ sec1.5 ] since we produce large deviations estimates there by gluing large deviations estimates on smaller time intervals where the @xmath20-variable ( and so , of course , @xmath512-variable ) can be freezed .
next , set @xmath2354 where now @xmath2355 does not depend on @xmath33 and @xmath380 .
then by ( [ 1.10.1 ] ) together with the gronwall inequality we obtain that @xmath2356 where @xmath964 is the lipschitz constant of @xmath80 . if @xmath20 belongs to the basin
@xmath382 then according to theorem [ thm1.2.7 ] @xmath2349 , and so also @xmath2301 , stays most of the time near @xmath382 up to its exit from the basin of the latter which yields according to the above inequality that @xmath2300 stays close to @xmath2357 during this time .
but now we can employ the averaging principle for the pair @xmath2358 which sais that @xmath2359 stays close on the time intervals of order @xmath2360 to the averaged motion @xmath2361 defined by @xmath2362 and in view of ( [ 1.10.9 ] ) , @xmath2363 decreases while @xmath2364 increases which leads to the behavior described in conjecture [ conj1.10.1 ] .
a similar conjecture can be made under the corresponding conditions for the discrete time case determined by a three scale difference system of equations of the form @xmath2365 where @xmath80 and @xmath19 are smooth vector functions and @xmath2366 is a smooth map ( a diffeomorphism or an endomorphism ) .
we obtain an example where discrete time versions of conditions of conjecture [ conj1.10.1 ] hold true setting , for instance , @xmath2367 , @xmath2368 and @xmath2369 ( mod 1 ) .
the scheme for the stochastic resonance type phenomenon described below is a slight modification of the model suggested by m.freidlin ( cf .
@xcite ) and it can be demonstrated in the setup of three scale systems @xmath3004 @xmath3005 , @xmath3006 , @xmath3007 with initial conditions @xmath3008 , @xmath2265 and @xmath2266 and the last equation in ( [ 2.8.1 ] ) is a stochastic differential equation coupled with first two ordinary differential equations though together they should be considered as a system of stochastic differential equations ( with a degeneration in the first two ) .
we assume that @xmath3009 , @xmath2268 while @xmath2269 evolves on a compact @xmath12-dimensional @xmath13 riemannian manifold @xmath11 and the coefficients @xmath80 , @xmath19 , @xmath21 are bounded smooth vector fields on @xmath2270 , @xmath16 and @xmath11 , respectively , depending on other variables as parameters .
we suppose also that @xmath3010 is a uniformly positive definite smooth matrix field on @xmath11 . in the same way as in section [ sec2.2 ]
we can generalize the setup taking @xmath2269 to be random evolutions but in order to simplify the notations we restrict ourselves to fast motions @xmath2269 being diffusions .
the solution of ( [ 2.8.1 ] ) determines a markov diffusion process which the triple @xmath3011 . taking @xmath2274 we arrive at the ( unperturbed ) process @xmath3012 where @xmath3013 solves the unperturbed stochastic differential equation @xmath3014 it is natural to view the diffusion @xmath3015 as describing an idealized physical system where parameters @xmath3016 and @xmath32 are assumed to be constants of motion while the perturbed process @xmath3011 is regarded as describing a real system where evolution of these parameters is also taken into consideration but unlike the averaging setup ( [ 2.1.1 ] ) we have now two sets of parameters moving with very different speeds .
let @xmath3017 be the unique invariant measure of the diffusion @xmath3018 .
set @xmath3019 and let @xmath3020 be the solution of the averaged equation @xmath3021 first , we apply averaging and large deviations estimates in averaging from the previous sections to two last equations in ( [ 2.8.1 ] ) freezing the slowest variable @xmath488 ( i.e. taking for a moment @xmath2295 ) .
namely , set @xmath3022 and @xmath3023 so that @xmath3024 suppose for simplicity that @xmath2299 ( i.e. both @xmath3025 and @xmath2301 are one dimensional ) and that the solution @xmath3026 of ( [ 2.8.4 ] ) has the limit set consisting of two attracting points @xmath2245 and @xmath2244 , which for simplicity we assume to be independent of @xmath488 , and a repelling fixed point @xmath3027 depending on @xmath488 and separating their basins .
as an example of @xmath2304 we may have in mind @xmath3028 . let @xmath3029 be the large deviations rate functional for the system of last two equations in ( [ 2.8.1 ] ) defined in ( [ 2.2.9 ] ) and set for @xmath2307 , @xmath3030 ( cf . with @xmath387 in theorem [ thm2.2.7 ] ) .
set @xmath3031 and assume that for all @xmath488 , @xmath3032 which means in view of the averaging principle ( see theorem [ thm2.2.2 ] and the following it discussion ) that @xmath3033 decreases ( increases ) with high probability while @xmath3034 stays close to @xmath2245 ( to @xmath2244 ) .
[ conj2.8.1 ] suppose that there exist strictly increasing and decreasing functions @xmath3035 and @xmath3036 , respectively , so that @xmath3037 and @xmath3038 for some @xmath1019 while @xmath3039 for @xmath2318 .
assume that @xmath2319 and @xmath50 in such a way that @xmath3040 then for any @xmath3041 there exists @xmath119 so that the slowest motion @xmath3042 converges in distribution ( as @xmath914 so that ( [ 2.8.9 ] ) holds true ) to a periodic function @xmath2324 , @xmath2325 with @xmath3043 the argument supporting this conjecture goes as follows .
since @xmath3025 moves much slower than @xmath2301 we can freeze the former and in place of ( [ 2.8.1 ] ) we can study , first , ( [ 2.8.5 ] ) . applying the arguments of theorem [ thm2.2.7 ] to the pair @xmath2333 from ( [ 2.8.5 ] )
we conclude from ( [ 2.2.34 ] ) that the intermediate motion @xmath2301 most of the time stays very close to either @xmath2245 or @xmath2244 before it exits from the corresponding basin , and so in view of an appropriate averaging principle ( which follows , for instance , from theorem [ 2.2.2 ] ) on bounded time intervals the slowest motion @xmath3025 mostly stays close to the corresponding averaged motion determined by the vector fields @xmath2336 and @xmath2337 given by ( [ 2.8.7 ] ) .
when @xmath2301 is close to @xmath2245 the slowest motion @xmath3025 decreases until @xmath3044 where @xmath3045 . in view of ( [ 2.2.33 ] ) and the scaling ( [ 2.8.9 ] ) between @xmath33 and @xmath380 , a moment later @xmath3046 becomes less than @xmath625 and @xmath3047 jumps immediately close to @xmath2244 .
there @xmath3048 , and so @xmath3025 starts to grow until it reaches @xmath3049 where @xmath3050 .
a moment later @xmath3051 becomes smaller than @xmath625 and in view of ( [ 2.2.33 ] ) the intermediate motion @xmath2301 jumps immediately close to @xmath2245 .
this leads to a nearly periodic behavior of @xmath3025 .
in order to make these arguments precise we have to deal here with an additional difficulty in comparison with the two scale setup considered in previous sections since now the large deviations @xmath285-functionals from theorem [ thm2.2.2 ] and the @xmath282-functions describing adiabatic fluctuations and transitions of theorems [ thm2.2.5 ] and [ thm2.2.7 ] depend on another very slowly changing parameter .
still , we can use the technique of sections [ sec2.5 ] and [ sec2.6 ] above applied on time intervals where changes in the @xmath488-variable can be neglected should work here but the details of this approach have not been worked out yet . on the other hand ,
when the fast motion @xmath2269 does not depend on the slow motions , i.e. when the coefficients @xmath1265 and @xmath21 in ( [ 2.8.1 ] ) depend only on the coordinate @xmath18 ( but not on @xmath488 and @xmath20 ) , then the above arguments can be made precise without much effort .
indeed , we can obtain estimates for transition times @xmath2346 and @xmath2347 of @xmath2348 between neighborhoods of @xmath2245 and @xmath2244 as in theorem [ thm2.2.7 ] applying the latter to @xmath2349 and @xmath2350 from ( [ 2.8.5 ] ) with freezed @xmath488-variable .
this is possible since the method of proposition [ prop2.5.1 ] requires us to make large deviations estimates , essentially , only for probabilities @xmath3052 , i.e. on bounded time intervals , and then combine them with the markov property arguments . during such times the slowest motion @xmath3025 can move only a distance of order @xmath2352 . thus freezing @xmath488 and using the gronwall inequality for the equation of @xmath2 in order to estimate the resulting error we see that the latter is small enough for our purposes .
observe , that it would be much more difficult to justify freezing @xmath488 in the coefficients @xmath1265 and @xmath21 of @xmath1 , if we allow the latter to depend on @xmath488 , since a strightforward application of the gronwall inequality there would yield an error estimate of an exponential in @xmath5 order which is comparable with @xmath2353 .
still , it may be possible to take care about the general case using methods of section [ sec2.3 ] since we produce large deviations estimates there by gluing large deviations estimates on smaller time intervals where the @xmath20-variable ( and so , of course , @xmath488-variable ) can be freezed .
next , set @xmath3053 where now @xmath2355 does not depend on @xmath33 and @xmath380 . then by ( [ 2.8.1 ] ) together with the gronwall inequality we obtain that @xmath3054 where @xmath604 is the lipschitz constant of @xmath80 .
if @xmath20 belongs to the basin @xmath382 then according to theorem [ thm2.2.7 ] @xmath2349 , and so also @xmath2301 , stays most of the time near @xmath382 up to its exit from the basin of the latter which yields according to the above inequality that @xmath3025 stays close to @xmath3055 during this time .
but now we can employ the averaging principle for the pair @xmath3056 which sais that @xmath3057 stays close on the time intervals of order @xmath2360 to the averaged motion @xmath3058 defined by @xmath3059 and in view of ( [ 2.8.9 ] ) , @xmath3060 decreases while @xmath3061 increases which leads to the behavior described in conjecture [ conj2.8.1 ] .
a similar conjecture can be made under the corresponding conditions for the discrete time case determined by a three scale difference system of equations of the form @xmath3062 where @xmath80 and @xmath19 are smooth vector functions and @xmath3063 are coupled with @xmath3064 and @xmath3065 perturbations of a parametrized by @xmath488 and @xmath20 appropriate family of markov chains having smooth transition densities similar to those considered in theorem [ thm2.2.12 ] .
this section deals with the averaging principle and a bit with the corresponding large deviations in the sense of convergence of young measures and i thank k.gelfert for asking me about young measures applications in averaging and for indicating to me the paper @xcite .
let @xmath45 belongs to the space @xmath2370 of probability measures on @xmath10 and consider the young measure ( which is a map from a measure space to a space of measures , see @xcite ) @xmath2371 from @xmath2372\times{{\mathbb r}}^d\times{{\bf m}},\ell_t\times\mu)$ ] to @xmath2370 defined by @xmath2373 where @xmath2374 is the lebesgue measure on @xmath201 $ ] , @xmath2375 is the unit mass at @xmath512 , and @xmath2376 are solutions of @xmath2377 on the product @xmath2370 .
we assume that for all @xmath2378 and @xmath2379 the coefficients @xmath21 and @xmath19 satisfy @xmath2380 for some @xmath907 independent of @xmath2381 .
of course , we could require the lipschitz continuity and the boundedness conditions ( [ 1.11.1 ] ) only in some open domain as in section [ sec1.2 ] but we can always extend these vector fields to the whole @xmath16 keeping these properties intact .
suppose that @xmath2382 has a disintegration @xmath2383 such that for each lipschitz continuous function @xmath711 on @xmath11 and any @xmath2378 , @xmath2384 depending only on @xmath964 where @xmath2385 is both a lipschitz constant of @xmath711 and it also bounds @xmath2386 . set @xmath2387 then by ( [ 1.11.1 ] ) and ( [ 1.11.3 ] ) , @xmath2231 is bounded and lipschitz continuous , and so there exists a unique solution @xmath2388 of ( [ 1.1.3 ] ) .
for any bounded continuous function @xmath711 on @xmath10 define @xmath2389 where @xmath2390 . by the definition ( see @xcite ) , the young measures @xmath2371 converge as @xmath50 to the young measure @xmath2391 defined by @xmath2392 @xmath2393 , if for any bounded continuous function @xmath1095 on @xmath201\times{{\mathbb r}}^d\times{{\bf m}}$ ] , @xmath2394 the following result provides a verifiable ( in some interesting cases ) criterion for even stronger convergence .
[ thm1.11.1 ] let @xmath2382 has the disintegration ( [ 1.11.2 ] ) satisfying ( [ 1.11.3 ] ) .
then @xmath2395 for any bounded continuous function @xmath2396 on @xmath201\times{{\mathbb r}}^d\times { { \bf m}}$ ] where @xmath2397 if and only if for each @xmath2398 and any finite collection @xmath2399 of bounded lipschitz continuous functions on @xmath10 there exists an integer valued function @xmath2400 as @xmath50 such that for any @xmath253 and @xmath2401 , @xmath2402 where @xmath2403 and , recall , @xmath2404 .
first , we prove that ( [ 1.11.5 ] ) implies ( [ 1.11.6 ] ) .
let @xmath2399 be bounded lipschitz continuous functions on @xmath2405 and set @xmath2406 if @xmath2407 then by ( [ 1.11.5 ] ) for each @xmath2401 , @xmath2408 choose an integer valued function @xmath2409 as @xmath50 so that @xmath2410 and let @xmath2411 .
set @xmath2412 , @xmath2413 and @xmath2414 , @xmath2415 .
then by ( [ 1.11.7 ] ) , @xmath2416 by ( [ 1.11.1 ] ) , @xmath2417 where @xmath2418 is the lipschitz constant of @xmath2419 .
similarly , by ( [ 1.11.1 ] ) and ( [ 1.11.3 ] ) , @xmath2420 and @xmath2421 it follows from ( [ 1.11.10])([1.11.13 ] ) that @xmath2422 given @xmath253 choose @xmath2423 such that for all @xmath2424 and @xmath2401 , @xmath2425 then by ( [ 1.11.14 ] ) , @xmath2426 by chebyshev s inequality @xmath2427 by ( [ 1.11.9 ] ) the right hand side of ( [ 1.11.15 ] ) tends to 0 as @xmath50 yielding ( [ 1.11.6 ] ) .
next , we derive ( [ 1.11.5 ] ) from ( [ 1.11.6 ] ) .
since @xmath1095 in ( [ 1.11.5 ] ) is a bounded function and @xmath1029 is a probability measure it is easy to see that it suffices to prove ( [ 1.11.5 ] ) when the integration in @xmath20 there is restricted to compact subsets of @xmath16 .
but if we integrate in ( [ 1.11.5 ] ) in @xmath20 running over a compact set @xmath2428 then by ( [ 1.1.1 ] ) and ( [ 1.11.1 ] ) , @xmath2429 i.e. both @xmath2430 and @xmath2431 belong to the @xmath2432neighborhood @xmath2433 of the set @xmath1762 when @xmath1640 and @xmath1271 $ ] . on @xmath201\times
g_{kt}\times{{\bf m}}$ ] we can approximate @xmath1095 uniformly by lipschitz continuous functions .
thus , in place of ( [ 1.11.5 ] ) it suffices to show that for any compact set @xmath2428 and a bounded lipschitz continuous function @xmath1095 on @xmath201\times g_{kt}\times{{\bf m}}$ ] with a lipschitz constant @xmath2434 in all variables , @xmath2435 by ( [ 1.11.2 ] ) , ( [ 1.11.3 ] ) and ( [ 1.11.4 ] ) , @xmath2436 since ( [ 1.11.6 ] ) holds true also for @xmath2437 , it follows from theorem 2.1 of @xcite that @xmath2438 and so we have only to deal with the first absolute value in the right hand side of ( [ 1.11.18 ] ) . as before set @xmath2412 , @xmath2413 , @xmath2414 and fix a large @xmath2398 .
let @xmath2439=[jn / n({{\varepsilon}})]$ ] then by ( [ 1.11.1 ] ) , ( [ 1.11.2 ] ) and ( [ 1.11.3 ] ) , @xmath2440 and @xmath2441 now using ( [ 1.11.20 ] ) , ( [ 1.11.21 ] ) and assuming that @xmath2442 for some constant @xmath2443 we obtain @xmath2444 where @xmath2445 integrating against @xmath45 both parts of ( [ 1.11.22 ] ) over @xmath2446 we obtain @xmath2447 where @xmath2448 by the assumption there exists an integer valued function @xmath2409 as @xmath50 such that ( [ 1.11.6 ] ) holds true for all @xmath2449 and then @xmath2450 as @xmath50 .
hence , letting first @xmath50 , then @xmath2319 and , finally , @xmath2451 we obtain ( [ 1.11.17 ] ) in view of ( [ 1.11.18 ] ) and ( [ 1.11.19 ] ) , completing the proof of theorem [ thm1.11.1 ] .
observe that ( [ 1.11.5 ] ) holding true for all bounded continuous functions is , in principle , stronger than the averaging principle in the form ( [ 1.11.19 ] ) since the latter is equivalent to ( [ 1.11.5 ] ) with @xmath2452 .
in fact , if we require ( [ 1.11.6 ] ) only for one function @xmath2437 then in the same way as in the proof of theorem [ thm1.11.1 ] above we conclude that ( [ 1.11.6 ] ) is equivalent to ( [ 1.11.19 ] ) if we consider the latter over all compacts @xmath2428 ( which was proved earlier in theorem 2.1 of @xcite ) .
still , the main interesting classes of systems , we are aware of , for which ( [ 1.11.5 ] ) holds true are the same for which ( [ 1.11.19 ] ) is satisfied though it is easy to construct examples of ( somewhat degenerate ) right hand sides @xmath21 and @xmath19 in ( [ 1.1.1 ] ) for which ( [ 1.11.19 ] ) holds true but ( [ 1.11.5 ] ) fails ( since in the latter we require convergence for all functions @xmath1095 and in the former only for @xmath2452 ) .
[ cor1.11.2 ] suppose that there exists an integer valued function @xmath2400 as @xmath50 such that @xmath2457 and for any @xmath253 and each bounded lipschitz continuous function @xmath2458 on @xmath10 , @xmath2459 then ( [ 1.11.6 ] ) is also satisfied , and so ( [ 1.11.5 ] ) holds true . in the same way as in @xcite we obtain that ( [ 1.11.24 ] )
holds true in the anosov theorem setup when @xmath36 is an @xmath27-invariant measure which is ergodic for @xmath1029-almost all @xmath20 , where @xmath1029 is the normalized lebesgue measure on a large compact in @xmath16 , and @xmath2460 with @xmath2461 differentiable in @xmath20 and @xmath18 .
furthermore , in the same way as in theorem 2.4 of @xcite or similarly to theorem 2.4 of @xcite we conclude that ( [ 1.11.6 ] ) and ( [ 1.11.24 ] ) hold true under assumptions [ ass1.2.1 ] and [ ass1.2.2 ] .
moreover , employing the method of @xcite the result can be extended to some partially hyperbolic systems .
observe that under assumptions [ ass1.2.1 ] and [ ass1.2.2 ] we can obtain also large deviations bounds in the form ( [ 1.2.16 ] ) and ( [ 1.2.17 ] ) for @xmath2462 with the functional @xmath2463\big\},\,\,\,\bar f_\nu(s , x)= \int f(s , x , y)d\nu(y ) , \end{aligned}\ ] ] where @xmath1095 is a bounded lipschitz continuous vector function .
this can be done deriving first an estimate similar to proposition [ prop1.3.4 ] for @xmath2464 in place of @xmath2465 there , which should follow in the same way as the proof of proposition 4.4 of @xcite , and proceeding similarly to sections [ sec1.4 ] and [ sec1.5 ] above .
of course , analogous results can be obtained in the discrete time setup of difference equations ( [ 1.1.10 ] ) . in this section
we derive the averaging principle and discuss corresponding large deviations in the sense of convergence of young measures adapted to our probabilistic setup . for more detailed information about young measures we refer the reader to @xcite and references there .
let @xmath45 belongs to the space @xmath2370 of probability measures on @xmath10 .
we consider a random young measure @xmath2371 from @xmath2372\times{{\mathbb r}}^d\times{{\bf m}},\ell_t\times\mu)$ ] to @xmath2370 which we define by @xmath2373 where @xmath2374 is the lebesgue measure on @xmath201 $ ] , @xmath2375 is the unit mass at @xmath512 , and @xmath2376 are the same as in ( [ 2.1.1 ] ) .
suppose that @xmath2382 has a disintegration @xmath3066 such that for each lipschitz continuous function @xmath711 on @xmath11 and any @xmath2378 , @xmath3067 depending only on @xmath964 where @xmath2385 is both a lipschitz constant of @xmath711 and it also bounds @xmath2386 .
set @xmath3068 and assume that ( [ 2.2.1 ] ) holds true which together with ( [ 2.9.2 ] ) yields that @xmath2231 is bounded and lipschitz continuous , and so there exists a unique solution @xmath3069 of ( [ 2.1.3 ] ) . for
any bounded continuous function @xmath711 on @xmath10 define @xmath3070 where @xmath2390 . in the spirit of @xcite )
we say that the young measures @xmath2371 converge as @xmath50 to a young measure @xmath2391 defined by @xmath2392 @xmath2393 , if for any bounded continuous function @xmath3071 on @xmath2372\times{{\mathbb r}}^d\times{{\bf m}}$ ] , @xmath3072 the following result provides a verifiable ( in some interesting cases ) criterion for even stronger convergence .
[ thm2.9.1 ] let @xmath2382 has the disintegration ( [ 2.9.1 ] ) satisfying ( [ 2.9.2 ] ) .
then @xmath3073 for any bounded continuous function @xmath2396 on @xmath201\times{{\mathbb r}}^d\times { { \bf m}}$ ] where @xmath2397 if and only if for each @xmath2398 and any finite collection @xmath2399 of bounded lipschitz continuous functions on @xmath10 there exists an integer valued function @xmath2400 as @xmath50 such that for any @xmath253 and @xmath2401 , @xmath3074 where @xmath2403 .
first , we prove that ( [ 2.9.4 ] ) implies ( [ 2.9.5 ] ) .
let @xmath2399 be bounded lipschitz continuous functions on @xmath2405 and set @xmath3075 if @xmath2407 then by ( [ 2.9.4 ] ) for each @xmath2401 , @xmath3076 choose an integer valued function @xmath2409 as @xmath50 so that @xmath3077 and let @xmath2411 .
set @xmath2412 , @xmath2413 and @xmath2414 , @xmath2415 .
then by ( [ 2.9.6 ] ) , @xmath3078 where @xmath3079 and @xmath3080 . by ( [ 2.2.1 ] ) , @xmath3081 where @xmath2418 is the lipschitz constant of @xmath2419 .
similarly , by ( [ 2.2.1 ] ) and ( [ 2.9.2 ] ) , @xmath3082 and @xmath3083 it follows from ( [ 2.9.9])([2.9.12 ] ) that @xmath3084 given @xmath253 choose @xmath2423 such that for all @xmath2424 and @xmath2401 , @xmath2425 then by ( 9.13 ) and the markov property , @xmath3085 where @xmath3086 is the conditional expectation . by chebyshev s inequality @xmath3087 by ( [ 2.9.8 ] ) the right hand side of ( [ 2.9.14 ] ) tends to 0 as @xmath50 yielding ( [ 2.9.5 ] ) .
next , we derive ( [ 2.9.4 ] ) from ( [ 2.9.5 ] ) .
since @xmath1095 in ( [ 2.9.4 ] ) is a bounded function and @xmath1029 is a probability measure it is easy to see that it suffices to prove ( [ 2.9.4 ] ) when the integration in @xmath20 there is restricted to compact subsets of @xmath16 .
but if @xmath20 belongs to a compact set @xmath2428 in view of ( [ 2.1.1 ] ) and ( [ 2.2.1 ] ) the slow motion @xmath3088 , as well as the averaged one @xmath3089 , stays during the time @xmath3090 in a @xmath2432neighborhood @xmath2433 of @xmath1762 .
but on @xmath201\times g_{kt}\times{{\bf m}}$ ] we can approximate @xmath1095 uniformly by lipschitz continuous functions .
thus , in place of ( [ 2.9.4 ] ) it suffices to show that for any compact set @xmath2428 and a bounded lipschitz continuous function @xmath1095 on @xmath201\times g_{kt}\times{{\bf m}}$ ] with a lipschitz constant @xmath2434 in all variables , @xmath3091 by ( [ 2.2.1 ] ) , ( [ 2.9.2 ] ) and ( [ 2.9.3 ] ) , @xmath3092 by ( [ 2.1.1 ] ) , ( [ 2.2.1 ] ) , ( [ 2.9.2 ] ) and ( [ 2.9.3 ] ) , @xmath3093 this together with the gronwall inequality gives @xmath3094
now we see that the integral term in the right hand side of ( [ 2.9.17 ] ) is a particular case of the integral term in the right hand side of ( [ 2.9.16 ] ) with @xmath2452 , and so it suffices to estimate only the latter .
set , again , @xmath2412 , @xmath2413 , @xmath2414 , @xmath2415 and fix a large @xmath2398 .
let @xmath3095=[jn / n({{\varepsilon}})]$ ] then by ( [ 2.2.1 ] ) , ( [ 2.9.1 ] ) and ( [ 2.9.2 ] ) , @xmath3096 and @xmath3097 now using ( [ 2.9.18 ] ) , ( [ 2.9.19 ] ) together with the markov property and assuming that @xmath2442 for some constant @xmath2443 we obtain @xmath3098 where @xmath2445 integrating against @xmath45 both parts of ( [ 2.9.20 ] ) over @xmath2446 we obtain @xmath3099 where @xmath3100 by the assumption there exists an integer valued function @xmath2409 as @xmath50 such that ( [ 2.9.5 ] ) holds true for all @xmath3101 and then @xmath2450 as @xmath50 .
hence , letting first @xmath50 , then @xmath2319 and , finally , @xmath2451 we obtain ( [ 2.9.15 ] ) in view of ( [ 2.9.16 ] ) and ( [ 2.9.17 ] ) , completing the proof of theorem [ thm2.9.1 ] .
observe that ( [ 2.9.4 ] ) holding true for all bounded continuous functions is , in principle , stronger than the averaging principle in the form @xmath3102 since ( [ 2.9.22 ] ) is equivalent to ( [ 2.9.15 ] ) with @xmath2452 .
in fact , if we require ( [ 2.9.5 ] ) only for one function @xmath2437 then it will be equivalent to ( [ 2.9.22 ] ) which follows in the same way as the proof of theorem [ thm2.9.1 ] above .
still , the main interesting classes of systems , we are aware of , for which ( [ 2.9.4 ] ) holds true are the same for which ( [ 2.9.22 ] ) is satisfied though it is easy to construct examples of ( somewhat degenerate ) right hand sides @xmath19 in ( [ 2.1.1 ] ) for which ( [ 2.9.22 ] ) holds true but ( [ 2.9.4 ] ) fails ( since in the latter we require convergence for all functions @xmath1095 and in the former only for @xmath2452 ) .
it follows from @xcite that the assumptions of theorem [ thm2.9.1 ] hold true when the unperturbed fast motions @xmath2472 are diffusion processes on @xmath11 so that @xmath36 is an invariant measure of @xmath2549 on @xmath11 ergodic for @xmath1029-almost all @xmath20 , where @xmath1029 is the normalized lebesgue measure on a large compact in @xmath16 , and @xmath2460 with @xmath2461 differentiable in @xmath20 and @xmath18 .
this can be extended to random evolutions considered in previous sections .
observe that under assumptions of theorem [ thm2.2.2 ] we can obtain also large deviations bounds in the form ( [ 2.2.10 ] ) and ( [ 2.2.11 ] ) for @xmath2462 with the functional @xmath2463\big\},\,\,\bar f_\nu(s , x)= \int f(s , x , y)d\nu(y ) , \end{aligned}\ ] ] where @xmath1095 is a bounded lipschitz continuous vector function .
the proof can be carried out quite similarly to the proof of theorem [ thm2.2.2 ] .
analogous results can be obtained in the discrete time setup of difference equations ( [ 2.1.7 ] ) .
we will assume that right hand side of ( [ 2.1.1 ] ) is bounded and lipschitz continuous , i.e. for some @xmath231 , @xmath2488 where @xmath497 is the metric on @xmath11 .
our large deviations estimates will be derived under the following general assumption on the fast motion which is satisfied , as we explain it below , for random evolutions which are markov processes with switching at random times between a finite number of diffusion processes .
[ ass2.2.1 ] there exist a convex differentiable in @xmath1192 and lipschitz continuous in other variables function @xmath195 defined for all @xmath831 and for @xmath2489 from the closure @xmath133 of a relatively compact open connected set @xmath57 and a positive function @xmath2490
satisfying @xmath2491 such that for all @xmath916 , @xmath2492 and @xmath832 , @xmath2493 where @xmath806 and @xmath804 is the inner product . set @xmath2494 @xmath191 and @xmath2495 . since @xmath2496 then @xmath2497
in view of assumption [ ass2.2.1 ] and standard convex analysis duality results ( see @xcite and @xcite ) @xmath198 is ( strictly ) convex , lower semicontinuous and we have also that @xmath2498 it follows also from assumption [ ass2.2.1 ] that @xmath2499 since @xmath2496 by ( [ 2.2.6 ] ) and @xmath198 is lower semicontinuous then it follows from ( [ 2.2.5 ] ) that there exists a unique @xmath2500 such that @xmath2501 set @xmath2502 .
if @xmath2503 depends lipschitz continuously in @xmath20 then we can define the averaged motion @xmath2504 in this general setup as the solution of the ordinary differential equation @xmath2505 denote by @xmath81 the space of continuous curves @xmath200 $ ] in @xmath67 which is the space of continuous maps of @xmath201 $ ] into @xmath202 for each absolutely continuous @xmath203 its velocity @xmath204 can be obtained as the almost everywhere limit of continuous functions @xmath205 when @xmath206 .
hence @xmath204 is measurable in @xmath207 , and so we can set @xmath2506 define the uniform metric on @xmath81 by @xmath220 for any @xmath221 set @xmath2507 since @xmath223 is lower semicontinuous and convex in @xmath224 and , in addition ,
@xmath225 if @xmath2508 it follows that the conditions of theorem 3 in ch.9 of @xcite are satisfied as we can choose a fast growing minorant of @xmath223 required there to be zero in a sufficiently large ball and to be equal , say , @xmath227 outside of it . as a result we conclude that @xmath228 is lower semicontinuous functional on @xmath81 with respect to the metric @xmath229 , and so @xmath230 is a closed set which plays a crucial role in the large deviations arguments below .
set @xmath2509 .
[ thm2.2.2 ] suppose that ( [ 2.2.1 ] ) and assumption [ ass2.2.1 ] hold true .
set @xmath51 and let @xmath240 .
then for any @xmath243 and every @xmath244 there exists @xmath245 such that for @xmath1151 uniformly in @xmath432 , @xmath2510 and @xmath2511 the main class of markov processes satisfying our conditions which we have in mind consists of random evolutions on @xmath2513 where @xmath1052 is a compact @xmath446-dimensional @xmath13 riemannian manifold and the unperturbed parametric family of markov processes @xmath2472 is the pair @xmath2514 governed by the stochastic differential equations @xmath2515 where @xmath2516 and for all @xmath2517 , @xmath2518 we assume that @xmath2519 are bounded positive @xmath53 functions , @xmath2520 is a @xmath53 field of positively definite symmetric matrices on @xmath1052 , @xmath2521 is a @xmath53 vector field and all functions are defined and satisfy the above properties for @xmath2522 and @xmath20 belonging to an open neighborhood of @xmath133 . here
@xmath2523 is the brownian motion and the equation ( [ 2.2.12 ] ) is written in local coordinats .
observe that the existence and some properties of such markov processes are discussed in @xcite .
the generator @xmath2524 of the markov process @xmath2525 is the operator acting on @xmath13 vector functions @xmath2526 on @xmath1052 by the formula @xmath2527 where @xmath2528 is the elliptic second order differential operator @xmath2529 now , the perturbed fast motion @xmath2530 satisfies @xmath2531 @xmath2532 and @xmath2533 where @xmath2 is given by ( [ 2.1.1 ] ) with @xmath2534 smoothly depending on @xmath20 and @xmath488 , so that the triple @xmath2535 is a markov processes .
the following result which will be proved in section [ sec2.3 + ] claims , in particular , that random evolutions above satisfy assumption [ ass2.2.1 ] [ prop2.2.4 ] for the process @xmath2536 defined by ( [ 2.2.12 ] ) and ( [ 2.2.13 ] ) the limit @xmath2537 exists uniformly in @xmath2538 , @xmath432 and @xmath832 , it is strictly convex and differentiable in @xmath1192 and lipschitz continuous in other variables , and it does not depend under our conditions on @xmath488 and @xmath2539 . in this circumstances the function @xmath198 given by ( [ 2.2.4 ] ) can be represented in the explicit form @xmath2540 where @xmath2541 and the first infinum is taken over the set @xmath2542 of probability measures on @xmath11 , i.e. over the vector measures @xmath2543 with @xmath2544 , and the second one is taken over positive vector functions @xmath2545 on @xmath1052 belonging to the domain of the operator @xmath2524 . clearly , @xmath2546 and , furthermore , @xmath2547 if and only if @xmath45 is the invariant measure @xmath2548 of the markov process @xmath2549 which is unique in our circumstances since the doeblin condition ( see @xcite ) holds true here .
the vector field @xmath2550 is @xmath53 in @xmath20 , and so we can define the averaged motion @xmath2476 by @xmath2551 hence , @xmath218 if and only if @xmath2552 for all @xmath252 $ ] . the processes @xmath1 given by ( [ 2.2.16 ] ) and ( [ 2.2.17 ] ) together with the function @xmath195 satisfy assumption [ ass2.2.1 ] . clearly , if @xmath2553 above then @xmath1 becomes a diffusion process and if all operators @xmath2528 are just zero then we arrive to the case of continuous time markov chains as fast motions which also satify all our assumptions .
we observe also that both proposition [ prop2.2.4 ] and the results below can be extended to the case when @xmath2528 are hypoelliptic operators satisfying natural conditions so that we could rely , in particular , on results of section 6.3 from @xcite .
suppose that the coefficients @xmath2554 and @xmath2555 in ( [ 2.2.12 ] ) and ( [ 2.2.13 ] ) do not depend on @xmath20 .
then @xmath2556 is an ergodic markov process with the unique invariant measure @xmath45 and for any @xmath18 almost surely @xmath2557 and by standard general results on the uncoupled averaging ( see @xcite ) it follows that for any @xmath18 almost surely @xmath2558 in the fully coupled case ( i.e. when @xmath2559 depend on @xmath20 ) theorem [ thm2.2.2 ] implies in the case of fast motions given by ( [ 2.2.16 ] ) and ( [ 2.2.17 ] ) that for each @xmath253 there is @xmath2560 such that for all small @xmath33 , @xmath2561 which means , in particular , that in this case we have in ( [ 2.2.22 ] ) convergence in probability .
examples from @xcite show that , in general , in the fully coupled setup we do not have convergence in ( [ 2.2.22 ] ) with probability one though in some cases such convergence can be derived from ( [ 2.2.22 + ] ) if the derivatives of @xmath2 and @xmath1 in @xmath33 grow subexponentially in @xmath5 on time intervals of order @xmath5 ( see remark [ rem2.3.6 ] ) . in the following assertions
we assume always that the fast motions are obtained by means of ( [ 2.2.12 ] ) and ( [ 2.2.13 ] ) so that we could rely on ( [ 2.2.18])([2.2.20 ] ) though , in principle , it is possible to impose some general conditions on functions @xmath223 which would enable us to proceed with our arguments .
precise large deviations bounds such as ( [ 2.2.10 ] ) and ( [ 2.2.11 ] ) of theorem [ thm2.2.2 ] are crucial in our study in sections [ sec2.5 ] and [ sec2.6 ] of the `` very long '' , i.e. exponential in @xmath5 , time `` adiabatic '' behaviour of the slow motion .
namely , we will describe such long time behavior of @xmath88 in terms of the function @xmath280 under various assumptions on the averaged motion @xmath281 observe that @xmath282 satisfies the triangle inequality @xmath283 for any @xmath284 and it determines a semi metric on @xmath67 which measures `` the difficulty '' for the slow motion to move from point to point in terms of the functional @xmath285 .
introduce the averaged flow @xmath286 on @xmath287 by @xmath2562 where @xmath2563 is the same as in ( [ 2.2.21 ] ) and set @xmath2564 for any probability measure @xmath2543 on @xmath2513 . call a @xmath286-invariant compact set @xmath291 an @xmath285-compact if for any @xmath292 there exist @xmath293 and an open set @xmath294 such that whenever @xmath295 and @xmath296 we can pick up @xmath297 $ ] and @xmath298 satisfying @xmath2565 it is clear from this definition that @xmath300 for any pair points @xmath301 of an @xmath285-compact @xmath302 and by the above triangle inequality for @xmath282 we see that @xmath303 takes on the same value when @xmath304 is fixed and @xmath20 runs over @xmath302 .
we say that the vector field @xmath19 on @xmath305 is complete at @xmath306 if the convex set of vectors @xmath2566 , \mu\in{{\mathcal p}}({{\bf m}}),\ , i_x(\mu)<\infty\}$ ] contains an open neigborhood of the origin in @xmath16 .
it follows by lemma [ lem1.6.2 ] in part [ part1 ] that if @xmath291 is a compact @xmath286-invariant set such that @xmath19 is complete at each @xmath295 and either @xmath302 contains a dense orbit of the flow @xmath286 ( i.e. @xmath286 is topologically transitive on @xmath302 ) or @xmath300 for any @xmath308 then @xmath302 is an @xmath285-compact .
moreover , to ensure that @xmath302 is an @xmath285-compact it suffices to assume that @xmath19 is complete already at some point of @xmath302 and the flow @xmath286 on @xmath302 is minimal , i.e. the @xmath286-orbits of all points are dense in @xmath302 or , equivalently , for any @xmath292 there exists @xmath309 such that the orbit @xmath310\}$ ] of length @xmath311 of each point @xmath295 forms an @xmath312-net in @xmath302 which is equivalent to minimality of the flow @xmath286 on @xmath302 ( see @xcite ) .
the latter condition obviously holds true when @xmath302 is a fixed point or a periodic orbit of @xmath286 but also , more generally , when @xmath286 on @xmath302 is uniquely ergodic ( see @xcite ) .
a compact @xmath286-invariant set @xmath291 is called an attractor ( for the flow @xmath286 ) if there is an open set @xmath313 and @xmath314 such that @xmath315 for an attractor @xmath302 the set @xmath316 , which is clearly open , is called the basin ( domain of attraction ) of @xmath302 .
an attractor which is also an @xmath285-compact will be called an @xmath285-attractor . in
what follows we will speak about connected open sets @xmath317 with piecewise smooth boundaries @xmath318 .
the latter can be introduced in various ways but it will be convenient here to adopt the definition from @xcite saying that @xmath318 is the closure of a finite union of disjoint , connected , codimension one , extendible @xmath53 ( open or closed ) submanifolds of @xmath16 which are called faces of the boundary .
the extendibility condition means that the closure of each face is a part of a larger submanifold of the same dimension which coincides with the face itself if the latter is a compact submanifold .
this enables us to extend fields of normal vectors to the boundary of faces and to speak about minimal angles between adjacent faces which we assume to be uniformly bounded away from zero or , in other words , angles between exterior normals to adjacent faces at a point of intersection of their closures are uniformly bounded away from @xmath319 and @xmath320 . the following result which will be proved in section [ sec2.5 ] describes exits of the slow motion from neighborhoods of attractors of the averaged motion .
[ thm2.2.5 ] let @xmath291 be an @xmath285-attractor of the flow @xmath286 whose basin contains the closure @xmath321 of a connected open set @xmath317 with a piecewise smooth boundary @xmath318 such that @xmath322 and assume that for each @xmath323 there exists @xmath324 and a probability measure @xmath2567 with @xmath2568 such that @xmath2569,\ ] ] i.e. @xmath2570 and the former vector points out into the interior while the latter into the exterior of @xmath317 .
set @xmath330 and @xmath331 .
then @xmath332 takes on the same value @xmath333 and @xmath334 coincides with the same compact nonempty set @xmath335 for all @xmath336 while @xmath337 for all @xmath338 .
furthermore , for any @xmath338 uniformly in @xmath432 , @xmath2571 and for each @xmath340 there exists @xmath341 such that uniformly in @xmath432 for all small @xmath342 , @xmath2572 next , set @xmath344 where @xmath345 dist@xmath346 and @xmath347 if @xmath348 and @xmath349 , otherwise . then for any @xmath338 and @xmath253 there exists @xmath350 such that uniformly in @xmath432 for all small @xmath342 , @xmath2573 finally , for every @xmath338 and @xmath253 , @xmath2574 provided @xmath353 and the latter holds true if and only if for some @xmath277 there exists @xmath354 such that @xmath355 for lebesgue almost all @xmath252 $ ] with @xmath211 then @xmath353 .
theorem [ thm2.2.5 ] asserts , in particular , that typically the slow motion @xmath88 performs rare ( adiabatic ) fluctuations in the vicinity of an @xmath285-attractor @xmath302 since it exists from any domain @xmath313 with @xmath356 for the time much smaller than @xmath357 ( as the corresponding number @xmath358 will be smaller ) and by ( [ 2.2.27 ] ) it can spend in @xmath359 only small proportion of time which implies that @xmath88 exits from @xmath59 and returns to @xmath360 ( exponentially in @xmath5 ) many times before it finally exits @xmath317 .
we observe that in the much simpler uncoupled setup corresponding results in the case of @xmath302 being an attracting point were obtained for a continuous time markov chain as a fast motion in @xcite but the proofs there rely on the lower semicontinuity of the function @xmath282 which does not hold true in general , and so extra conditions like @xmath285-compactness of @xmath302 or , more specifically , the completness of @xmath19 at @xmath302 should be assumed there , as well .
it is important to observe that the intuition based on diffusion type small random perturbations of dynamical systems should be applied with caution to problems of large deviations in averaging since the @xmath285-functional of theorem [ thm2.2.2 ] describing them is more complex and have rather different properties than the corresponding functional emerging in diffusion type random perturbations of dynamical systems ( see @xcite ) .
the reason for this is the deterministic nature of the slow motion @xmath88 which unlike a diffusion can move only with a bounded speed and , moreover , even in order to ensure its `` diffusive like '' local behaviour ( i.e. to let it go in many directions ) some extra nondegeneracy type conditions on the vector field @xmath19 are required .
our next result describes rare ( adiabatic ) transitions of the slow motion @xmath88 between basins of attractors of the averaged flow @xmath286 which we consider now in the whole @xmath16 and impose certain conditions on the structure of its @xmath361-limit set .
[ ass2.2.6 ] assumption [ ass2.2.1 ] holds true for @xmath362 , the families @xmath2575 and @xmath2576 of matrix and vector fields are compact sets in the @xmath53 topology , @xmath2577 for some @xmath231 independent of @xmath365 and there exists @xmath366 such that @xmath2578 the condition ( [ 2.2.30 ] ) means that outside of some ball all vectors @xmath368 have a bounded away from zero projection on the radial direction which points out to the origin .
this condition can be weakened , for instance , it suffices that @xmath369 but , anyway , we have to make some assumption which ensure that the slow motion stays in a compact region where really interesting dynamics takes place . next , suppose that the @xmath361-limit set of the averaged flow @xmath286 is compact and it consists of two parts , so that the first part is a finite number of @xmath285-attractors @xmath370 whose basins @xmath371 have piecewise smooth boundaries @xmath372 and the remaining part of the @xmath361-limit set is contained in @xmath373 .
we assume also that for any @xmath374 there exist @xmath324 and probability measures @xmath2579 such that @xmath2580 and @xmath2581\,\ , \mbox{and}\,\ , i=1, ... ,k,\ ] ] i.e. @xmath2582 and it points out into the interior of @xmath379 which means that from any boundary point it is possible to go to any adjacent basin along a curve with an arbitrarily small @xmath285-functional .
let @xmath253 be so small that the @xmath380-neighborhood @xmath381 of each @xmath382 is contained with its closure in the corresponding basin @xmath383 .
for any @xmath384 set @xmath2583 [ thm2.2.7 ] the function @xmath386 takes on the same value @xmath387 for all @xmath388 .
let @xmath389 .
then for any @xmath384 uniformly in @xmath432 , @xmath2584 and for any @xmath340 there exists @xmath341 such that for all small @xmath342 , @xmath2585 next , set @xmath392 then for any @xmath384 and @xmath253 there exists @xmath393 such that uniformly in @xmath432 for all small @xmath342 , @xmath2586 now , suppose that the vector field @xmath19 is complete on @xmath395 for some @xmath396 ( which strengthens ( [ 2.2.31 ] ) there ) and the restriction of the @xmath361-limit set of @xmath286 to @xmath395 consists of a finite number of @xmath285-compacts .
assume also that there is a unique index @xmath397 such that @xmath398 .
then for any @xmath384 there exists @xmath399 such that uniformly in @xmath432 for all small @xmath342 , @xmath2587 finally , suppose that the above conditions hold true for all @xmath401 .
define @xmath402 , @xmath403 and recursively , @xmath404 where @xmath405 , @xmath406 if @xmath407 , and set @xmath408 .
then for any @xmath384 and @xmath340 there exists @xmath341 such that uniformly in @xmath432 for all @xmath409 and sufficiently small @xmath342 , @xmath2588 and for some @xmath399 , @xmath2589 generically there exists only one index @xmath412 such that @xmath413 and in this case theorem [ thm2.2.7 ] asserts that @xmath414 arrives ( for `` most '' @xmath261 ) at @xmath415 after it leaves @xmath383 .
if @xmath416 contains more than one index then the method of the proof of theorem [ thm2.2.7 ] enables us to conclude that in this case @xmath417 arrives ( for `` most '' @xmath261 ) at @xmath418 after leaving @xmath383 but now we can not specify the unique basin of attraction of one of @xmath419 s where @xmath420 exits from @xmath383 .
if the succession function @xmath421 is uniquely defined then it determines an order of transitions of the slow motion @xmath88 between basins of attractors of @xmath89 and because of their finite number @xmath88 passes them in certain cyclic order going around such cycle exponentially many in @xmath5 times while spending the total time in a basin @xmath383 which is approximately proportional to @xmath422 . if there exist several cycles of indices @xmath423 where @xmath424 and @xmath425 then transitions between different cycles may also be possible . in the uncoupled case with fast motions being continuous time markov chains a description of such transitions via certain hierarchy of cycles appeared without a detailed proof in @xcite and @xcite . in our fully coupled setup
the corresponding description does not seem to be different from the uncoupled situation since its justification relies only on the markov property arguments and estimates of probabilities of transitions of @xmath88 from @xmath2590 to @xmath2022 .
set @xmath2591 .
following @xcite we call a graph consisting of arrows @xmath2592 @xmath2593 an @xmath1525-graph if every point @xmath2594 is the origin of exactly one arrow and the graph has no circles .
let @xmath2595 be the set of all @xmath1525-graphs .
next , choose @xmath253 so small that @xmath2596 and define stopping times @xmath2597 and by induction for @xmath864 , @xmath2598 @xmath2599 define the markov chain @xmath2600 which evolves on the phase space @xmath2601 where @xmath2602 .
[ thm2.2.8 ] let @xmath2603 be the transition probability of the markov chain @xmath2604 .
then for any @xmath1438 there exist @xmath2605 such that if @xmath2606 and @xmath1151 then @xmath2607 whenever @xmath2608 . furthermore ,
if @xmath2609 is an invariant measure of @xmath2604 on @xmath2601 then @xmath2610 where @xmath2611 since total times spent by a markov process in different sets are asymptotically proportional to masses given to these sets by corresponding invariant measures then theorem [ thm2.2.8 ] ( together with theorem [ thm2.2.7 ] ) yields actually that the slow motion @xmath427 spends in a basin @xmath2017 of the attractor @xmath419 a percentage of total time approximately proportional to @xmath2612 which will be illustrated by computational examples in section [ sec2.7 ] .
in fact , this description is effective only if there is a unique @xmath2613 and a graph @xmath2614 such that @xmath2615 is minimal possible among all such sums over all @xmath1373-graphs . in this case
the slow motion spends in @xmath2616 a proportion of time close to one .
[ ass2.2.9 ] there exist a convex differentiable in @xmath1192 and lipschitz continuous in other variables function @xmath195 defined for all @xmath831 and for @xmath2489 from the closure @xmath133 of a relatively compact open connected set @xmath57 and a positive function @xmath2490 satisfying ( [ 2.2.2 ] ) such that for all @xmath2617 , @xmath2492 and @xmath832 , @xmath2618 where @xmath806 and @xmath2619 appears in ( [ 2.1.7 ] ) .
[ thm2.2.10 ] suppose that ( [ 2.2.1 ] ) and assumption [ ass2.2.9 ] are satisfied and that @xmath2620 are given by ( [ 2.1.7 ] ) . for @xmath2621 $ ]
define @xmath2622 and set @xmath2623 .
then theorem [ thm2.2.2 ] and corollary [ cor2.2.3 ] hold true with the corresponding functionals @xmath228 . the main model of markov chains serving as
fast motions @xmath2624 , we have in mind , is obtained in the following way .
we start with a parametric family of markov chains @xmath2625 on a compact @xmath13 riemannian manifold @xmath1052 with transition probabilities @xmath2626 having positive densities @xmath2627 with respect to the riemannian volume @xmath54 , so that @xmath2628 is @xmath53 in @xmath20 and continuous in other variables .
next , we define @xmath2629 and @xmath2630 adding to ( [ 2.1.7 ] ) another equation @xmath2631 [ prop2.2.11 ] let @xmath2632 be as above
. then the limit @xmath2633 exists uniformly in @xmath2489 running over a compact set and in @xmath432 and it satisfies conditions of assumption [ ass2.2.9 ] . in this circumstances
the functionals @xmath228 appearing in the large deviations estimates ( [ 2.2.10 ] ) and ( [ 2.2.11 ] ) again have the form ( [ 2.2.9 ] ) with @xmath223 given by ( [ 2.2.19 ] ) where now @xmath2634 clearly , @xmath2546 and , furthermore , @xmath2547 if and only if @xmath45 is the invariant measure @xmath2635 of the markov chain @xmath2549 which is unique since the doeblin condition ( see @xcite ) holds true here .
the vector field @xmath2636 is @xmath53 in @xmath20 , and so we can define uniquely the averaged motion @xmath2476 by ( [ 2.2.21 ] ) and , again , @xmath218 if and only if @xmath2637 for all @xmath252 $ ] . furthermore , @xmath2624 given by ( [ 2.2.42 ] ) satisfies ( [ 2.2.41 ] ) . the existence of the limit ( [ 2.2.43 ] ) and its properties in our circumstances are well known ( see @xcite , @xcite , @xcite , @xcite , @xcite ) and the fact that ( [ 2.2.41 ] ) holds true here will be explained at the beginning of section [ sec2.7 ] .
[ thm2.2.12 ] let the fast motion @xmath2638 be constructed as above via ( [ 2.2.42 ] ) then with the corresponding definitions of @xmath285-compacts and under similar conditions the conclusions of theorems [ thm2.2.5 ] , [ thm2.2.7 ] and [ thm2.2.8 ] remain true for the corresponding slow motion @xmath427 defined in theorem [ thm2.2.10 ]
. observe , that we can easily produce a wide class of systems satisfying the conditions of theorems [ thm2.2.5 ] , [ thm2.2.7 ] , and [ thm2.2.8 ] or theorem [ thm2.2.12 ] by setting @xmath434 so that @xmath2639 where @xmath2635 is the unique invariant measure of @xmath2549 and the vector field @xmath436 , which becomes now the averaged vector field @xmath2231 , has an @xmath2640-limit set satisfying conditions of the above theorems .
simple examples of this construction will be exhibited in section [ sec2.7 ] for which we also compute historgrams indicating proportions of time the slow motion spends near different attracting points of the averaged motion .
we observe that the functional @xmath228 , which plays a crucial role in the above theorems , seems to be quite difficult to compute since this leads to difficult nonclassical variational problems .
we will need the following version of general large deviations bounds when usual assumptions hold true with errors .
the proof is a strightforward modification of the standard one ( cf .
@xcite ) and its details can be found in part [ part1 ] , lemma [ lem1.4.1 ] .
[ lem2.3.1 ] let @xmath983 , @xmath984 be uniformly bounded on compact sets functions on @xmath16 and @xmath985 be a family of @xmath982valued random vectors on a probability space @xmath986 such that @xmath987 with probability one for some constant @xmath793 and all @xmath855 . for any @xmath988 and @xmath989 set
@xmath2641 ( i ) for any @xmath991 there exists @xmath992 such that whenever for some @xmath993 , @xmath994 and each @xmath831 with @xmath995 , @xmath2642 then for any compact set @xmath997 , @xmath2643 where @xmath2644 ( ii ) suppose that @xmath1000 , @xmath1001 and there exists @xmath994 such that @xmath1002 and @xmath2645 if ( [ 2.3.2 ] ) holds true then for any @xmath253 , @xmath2646 ( iii ) assume that @xmath1005 satisfy ( [ 2.3.5 ] )
. for any @xmath991 there exists @xmath1006 such that whenever for some @xmath1007 and each @xmath831 with @xmath1008 the inequality ( [ 2.3.2 ] ) holds true together with @xmath2647 then for any @xmath1010 , @xmath2648 where @xmath1012 @xmath1013 @xmath1014 , @xmath1015 @xmath1016 and @xmath1017 denotes the closure of @xmath59 .
[ lem2.3.2 ] let @xmath1051 be a nondecreasing sequence of lower semicontinuous functions on a metric space @xmath1052 and let @xmath2649 assume that @xmath285 is also lower semicontinuous and for any compact set @xmath1054 denote @xmath1055 then @xmath2650 we will need also the following general result which will enable us to subdivide time into small intervals freezing the slow variable on each of them so that the estimate ( [ 2.2.3 ] ) of assumption [ ass2.2.1 ] becomes sufficiently precise and , on the other hand , we will not change much the corresponding functionals @xmath228 appearing in required large deviations estimates .
this result is certainly not new , it is cited in @xcite as a folklore fact and a version of it can be found in @xcite , p.67 while for a complete proof we refer the reader to part [ part1 ] , lemma [ lem1.4.3 ] .
[ lem2.3.3 ] let @xmath1085 be a measurable function on @xmath437 equal zero outside of @xmath201 $ ] and such that @xmath1086 .
for each positive integer @xmath54 and @xmath1087 $ ] define @xmath1088{{\delta}}-c)$ ] where @xmath1089 and @xmath1090 $ ] denotes the integral part . then there exists a sequence @xmath1091 such that for lebesgue almost all @xmath1087 $ ] , @xmath2651 [ lem2.3.4 ] let @xmath1100 @xmath1101 @xmath1102 @xmath1103 , @xmath1104 @xmath1105 , @xmath1106 , @xmath1107 and for @xmath252 $ ] , @xmath2652 then @xmath2653 @xmath2654 and @xmath2655 where , recall , @xmath1112 and @xmath1113 if @xmath1114 .
\(i ) there exist @xmath1127 and @xmath1128 independent of @xmath2657 such that if @xmath1130 and @xmath1131 then for any @xmath1132 , @xmath2658 where @xmath1134 , @xmath1135 does not depend on @xmath1129 and @xmath2659 in particular , if for each @xmath1137 there exists @xmath1138 such that @xmath2660 and @xmath2661 then ( [ 2.3.16 ] ) holds true with @xmath1141 in place of @xmath1142 , @xmath1137 .
\(ii ) for any @xmath1143 there exist @xmath1144 and @xmath1145 , the latter depending also on @xmath1146 , such that if @xmath1147 and @xmath1148 satisfy ( [ 2.3.18 ] ) and ( [ 2.3.19 ] ) , @xmath1149 , @xmath1150 and @xmath1151 then @xmath2662 with some @xmath1128 depending only on @xmath21 and @xmath1153 .
\(i ) introduce the events @xmath2663 so that we have @xmath2664 now for @xmath1345 by the markov property @xmath2665 if @xmath2666 then @xmath2667 in view of ( [ 2.3.13 ] ) and ( [ 2.3.14 ] ) satisfies @xmath2668 since @xmath1122 is lipschitz continuous in @xmath789 and @xmath1197 it follows from ( [ 2.3.22 ] ) that @xmath2669 provided @xmath2666 where @xmath1199 depends only on @xmath1200 . in view of assumption [ ass2.2.1 ]
we can estimate from above the probability in the right hand side of ( [ 2.3.22 ] ) by means of lemma [ lem2.3.1](i ) which together with ( [ 2.3.24 ] ) yield that @xmath2670 where @xmath1205 as , first , @xmath50 and then @xmath1206 . applying ( [ 2.3.25 ] ) for @xmath1250 and estimating @xmath2671 by means of lemma [ lem2.3.1](i ) we derive ( [ 2.3.16 ] ) in view of ( [ 2.3.21 ] ) .
\(ii ) in order to obtain ( [ 2.3.20 ] ) we rely on assumption [ ass2.2.1 ] and lemma [ lem2.3.1](iii ) estimating from below the probability in the right hand side of ( [ 2.3.22 ] ) which together with ( [ 2.3.23 ] ) yield @xmath2672 where @xmath1243 @xmath1244 @xmath1245 and @xmath1246 as , first , @xmath50 and then @xmath1206 . employing ( [ 2.3.26 ] ) for @xmath1250 and estimating @xmath2671 by means of lemma [ lem2.3.1](iii ) we obtain from ( [ 2.3.21 ] ) that @xmath2673 for some @xmath1256 provided , say , @xmath1257 and @xmath1258 . since @xmath466 is differentiable in @xmath1192 then @xmath1259 for any @xmath1260 ( see theorems 23.5 and 25.1 in @xcite ) , and so by the lower semicontinuity of @xmath223 in @xmath224 ( and , in fact , also in @xmath20 ) , @xmath1261 this together with lemma [ lem2.3.2 ] yield that @xmath1262 appearing in the definition of @xmath1263 is positive provided @xmath21 is sufficiently large .
in fact , it follows from the lower semicontinuity of @xmath223 that @xmath1262 is bounded away from zero by a positive constant independent of @xmath1264 and @xmath1147 , @xmath1137 if these points vary over fixed compact sets and ( [ 2.3.18 ] ) together with ( [ 2.3.19 ] ) hold true .
now , given @xmath1019 choose , first , sufficiently large @xmath21 as needed and then subsequently choosing small @xmath1265 and @xmath1266 , then small @xmath65 , and , finally , small enough @xmath33 we end up with an estimate of the form @xmath2674 where @xmath1128 and @xmath1268 satisfies ( [ 2.3.17 ] )
. finally , ( [ 2.3.20 ] ) follows from ( [ 2.3.27 ] ) and ( [ 2.3.28 ] ) .
the remaining part of the proof of theorem [ thm2.2.2 ] contains mostly some convex analysis arguments and it repeats almost verbatim the corresponding part of the proof of theorem [ thm1.2.3 ] in part [ part1 ] but for readers convenience we exhibit it also here .
we remark that some of the details below are borrowed from @xcite but we believe that our exposition and the way of proof are more precise , complete and easier to follow .
we start with the lower bound .
assume that @xmath1269 , and so that @xmath654 is absolutely continuous , since there is nothing to prove otherwise . then by ( [ 2.2.9 ] ) , @xmath1270 for lebesgue almost all @xmath1271.$ ] by ( [ 2.2.1 ] ) and assumption [ ass2.2.1 ] , @xmath2675 for some @xmath2676 , and so if @xmath1270 it follows from ( [ 2.2.4 ] )
that @xmath2677 .
suppose that @xmath2678 and let ri@xmath1275 be the interior of @xmath1275 in its affine hull ( see @xcite ) . then either ri@xmath1276 or @xmath1275 ( by its convexity ) consists of one point and recall that @xmath1277 for lebesgue almost all @xmath1271 $ ] . by ( [ 2.2.6 ] ) and ( [ 2.3.29 ] ) ,
@xmath2679 this together with the nonnegativity and lower semi - continuity of @xmath1279 yield that there exists @xmath1280 such that @xmath1281 and by a version of the measurable selection ( of the implicit function ) theorem ( see @xcite , theorem iii.38 ) , @xmath1280 can be chosen to depend measurably in @xmath1271 $ ] .
of course , if ri@xmath1282 then @xmath1275 contains only @xmath1280 and in this case @xmath1283 for lebesgue almost all @xmath1271 $ ] . taking @xmath1284 and @xmath1285
we obtain @xmath2680 observe that @xmath1287 is measurable as a function of @xmath1288 and @xmath224 since it is obtained via ( [ 2.2.4 ] ) as a supremum in one argument of a family of continuous functions , and so this supremum can be taken there over a countable dense set of @xmath1192 s .
hence , the set @xmath1289,\,{{\alpha}}\in{{\mathcal d}}(l_s)\}=\ell^{-1}[0,\infty)$ ] is measurable , and so the set @xmath1290\}$ ] is measurable , as well .
its projection @xmath1291:\ , ( s,{{\alpha}})\in b\,\mbox{for some}\,{{\alpha}}\in{{\mathbb r}}^d\}$ ] on the first component of the product space is also measurable and @xmath317 is the set of @xmath1271 $ ] such that @xmath1275 contains more than one point . employing theorem iii.22 from @xcite we select @xmath1292 measurably in @xmath1293 and such that @xmath1294 . by convexity and lower semicontinuity of @xmath1279 it follows from corollary 7.5.1 in @xcite that @xmath2681 for each @xmath253 set @xmath1296 then , clearly , @xmath1297 is a measurable function of @xmath1288 , and so @xmath1298 and @xmath1299 are measurable in @xmath1288 , as well . by theorems 23.4 and 23.5 from @xcite for each @xmath1300 there
exists @xmath1301 such that ( [ 2.3.31 ] ) holds true .
given @xmath1302 take @xmath1303 and for @xmath1271\setminus v$ ] set @xmath1284 .
then @xmath2682 for each @xmath1132 set @xmath1305 if the corresponding @xmath1306 in ( [ 2.3.31 ] ) satisfies @xmath1307 and @xmath1308 , otherwise .
note , that ( [ 2.3.31 ] ) remains true with @xmath1309 in place of @xmath1310 with @xmath1285 if @xmath1308 . as observed above @xmath1311 whenever @xmath1312 , and so @xmath1313 for lebesgue almost all @xmath1271 $ ] .
we recall also that @xmath1314 and @xmath1315 for lebesgue almost all @xmath1271 $ ] .
since @xmath1269 , @xmath1316 , and @xmath1317 as @xmath1318 for lebesgue almost all @xmath1271 $ ] , we conclude from ( [ 2.3.33 ] ) and the above observations that for @xmath21 large enough @xmath2683 next , we apply lemma [ lem2.3.3 ] to conclude that there exists a sequence @xmath1320 such that for each @xmath1321 and lebesgue almost all @xmath1322 , @xmath2684 where @xmath1324{{\delta}}_j - c$ ] , @xmath1090 $ ] denotes the integral part and we assume @xmath1325 and @xmath1326 if @xmath1327 . choose @xmath1328 $ ] and set @xmath1329 , @xmath1330 where @xmath1331 if @xmath1332 , @xmath1333 , @xmath1334 @xmath1335 and @xmath1336 , @xmath1337 for @xmath1338 and @xmath1339 for @xmath1340 where @xmath1341 . since @xmath2685 for lebesgue almost all @xmath1271 $ ] then @xmath2686 and , in addition , @xmath2687 by ( [ 2.3.33])([2.3.35 ] ) .
this together with ( [ 2.3.13 ] ) and ( [ 2.3.14 ] ) yield that for @xmath1345 , @xmath2688 provided @xmath1347 where @xmath1348 and @xmath1349 are the same as in lemma [ lem2.3.4 ] , the latter is defined with @xmath1350 , @xmath1338 and @xmath1351 .
choose @xmath1352 so small and @xmath1353 so large that @xmath2689 then by ( [ 2.3.36 ] ) , @xmath2690 by ( [ 2.3.33])([2.3.35 ] ) , @xmath2691 and by the construction above the conditions of the assertion ( ii ) of proposition [ prop2.3.5 ] hold true , so choosing @xmath1353 sufficiently large we derive ( [ 2.2.6 ] ) ( with @xmath1357 in place of @xmath1029 ) from ( [ 2.3.20 ] ) , ( [ 2.3.37 ] ) and ( [ 2.3.38 ] ) provided @xmath33 is small enough .
next , we pass to the proof of the upper bound ( [ 2.2.7 ] ) .
assume that ( [ 2.2.7 ] ) is not true , i.e. there exist @xmath1358 and @xmath240 such that for some sequence @xmath1359 as @xmath1360 , @xmath2692 since @xmath2693 by ( [ 2.2.1 ] ) all paths of @xmath1363 $ ] and of @xmath1364 $ ] given by ( [ 2.3.11 ] ) ( the latter for any measurable @xmath155 ) belong to a compact set @xmath1365 which consists of curves starting at @xmath20 and satisfying the lipschitz condition with the constant @xmath604 .
let @xmath1366 denotes the open @xmath625-neighborhood of the compact set @xmath230 and @xmath1367 .
for any small @xmath1368 choose a @xmath1352-net @xmath1369 in @xmath1370 where @xmath1371 .
since @xmath2694 then there exists @xmath1373 and a subsequence of @xmath1374 , for which we use the same notation , such that @xmath2695 denote such @xmath1376 by @xmath1377 , choose a sequence @xmath1378 and set @xmath1379 . since @xmath1370 is compact there exists a subsequence @xmath1380 converging in @xmath81 to @xmath1381 which together with ( [ 2.3.40 ] ) yield @xmath2696 for all @xmath1368 .
we claim that ( [ 2.3.41 ] ) contradicts ( [ 2.3.12 ] ) and the assertion ( i ) of proposition [ prop2.3.5 ] . indeed , set @xmath1383 by the monotone convergence theorem @xmath2697 similarly to our remark in section [ sec2.2 ] it follows from the results of section 9.1 of @xcite that the functionals @xmath1385 and @xmath1386 are lower semicontinuous in @xmath155 and @xmath654 ( see also section 7.5 in @xcite ) .
this together with ( [ 2.3.42 ] ) enable us to apply lemma [ lem2.3.2 ] in order to conclude that @xmath2698 where @xmath1388 .
the last inequality in ( [ 2.3.43 ] ) follows from the lower semicontinuity of @xmath228 .
thus we can and do choose @xmath1132 such that @xmath2699 by the lower semicontinuity of @xmath1390 in @xmath155 there exists a function @xmath2700 on @xmath1392 such that for each @xmath2701 , @xmath2702 next , we restrict the set of functions @xmath155 to make it compact .
namely , we allow from now on only functions @xmath155 for which there exists @xmath1393 such that either @xmath1395 or @xmath1396 for @xmath1397 , @xmath1398 and @xmath1399 where @xmath54 is a positive integer .
it is easy to see that the set of such functions @xmath155 is compact with respect to the uniform convergence topology in @xmath81 and it follows that @xmath1400 in ( [ 2.3.45 ] ) constructed with such @xmath155 in mind is lower semicontinuous in @xmath654 .
hence @xmath2703 now take @xmath1402 satisfying ( [ 2.3.41 ] ) and for any integer @xmath1403 set @xmath1404 , @xmath1405 , @xmath1406 and @xmath1407 @xmath1408 .
define a piecewise linear @xmath1409 and a piecewise constant @xmath1410 by @xmath2704 and @xmath1398 with @xmath1412 .
since @xmath1402 is lipschitz continuous with the constant @xmath2705 then @xmath2706 if @xmath54 is large enough and @xmath342 is sufficiently small then @xmath2707 where @xmath1415 is the same as in ( [ 2.3.16 ] ) . since @xmath1416 it follows from ( [ 2.3.48 ] ) and ( [ 2.3.49 ] ) that @xmath1417 and by ( [ 2.3.45 ] ) and the first inequality in ( [ 2.3.49 ] ) we obtain that @xmath2708 hence , by ( [ 2.3.16 ] ) and the second inequality in ( [ 2.3.49 ] ) for all @xmath33 small enough , @xmath2709 provided @xmath1420 ( taking into account that @xmath1421 ) . by ( [ 2.3.12 ] ) and the definition of vectors @xmath1422 for any @xmath1423 , @xmath2710 therefore , @xmath2711 choosing , first , @xmath54 large enough so that @xmath65 satisfies ( [ 2.3.49 ] ) with all sufficiently small @xmath33 and also that @xmath2712 , and then choosing @xmath1352 so small that @xmath2713 , we conclude that ( [ 2.3.51 ] ) together with ( [ 2.3.53 ] ) contradicts ( [ 2.3.41 ] ) , and so the upper bound ( [ 2.2.7 ] ) holds true , completing the proof of theorem [ thm2.2.2 ] .
[ rem2.3.6 ] in view of examples from @xcite in the fully coupled setup we should not expect convergence ( [ 2.2.22 ] ) in the averaging principle with probability one in spite of exponentially fast convergence in probability ( [ 2.2.22 + ] ) provided by the upper large deviations bound ( [ 2.2.11 ] ) .
still , when derivatives of @xmath2 and @xmath1 in @xmath33 grow not too fast we can derive convergence with probability one from ( [ 2.2.22 + ] ) .
indeed , consider , for instance , the following example @xmath2714 where @xmath2715 is a constant , @xmath2523 is the standard one dimensional brownian motion , @xmath368 satisfies ( [ 2.2.1 ] ) and it is @xmath2716-periodic in @xmath18 and @xmath21 has a bounded derivative in @xmath20 .
set @xmath2717 then @xmath2718 the solution of this linear equation is easy to estimate which yields that for some constant @xmath1436 , @xmath2719 let @xmath2635 be the invariant measure of the diffusion @xmath2720 ( mod 1 ) ( which is unique since the doeblin condition is satisfied here ) and assume that @xmath2721 which does not harm the generality since we always can consider @xmath2722 in place of @xmath368 .
set @xmath2723 where @xmath2724 is the same as in ( [ 2.2.22 + ] ) written for our specific situation .
then @xmath2725 and by the borel
cantelli lemma we obtain that there exists @xmath2726 , finite with probability one , so that for all @xmath2727 , @xmath2728 by ( [ 2.3.55 ] ) for @xmath2729 and @xmath2730 , @xmath2731 it follows that with probability one , @xmath2732 which is what we need since in our case @xmath2733 in view of ( [ 2.3.56 ] ) .
in this section we will prove proposition [ prop2.2.4 ] . observe that @xmath195 obtained by ( [ 2.2.18 ] ) is the principal eigenvalue of the operator @xmath2734 acting on @xmath13 vector functions @xmath2526 on the manifold @xmath1052 by the formula ( see @xcite ) , @xmath2735 where @xmath2489 and @xmath1192 are considered as parameters . according to @xcite
this operator satisfies the strong maximum principle .
thus , the first part of proposition [ prop2.2.4 ] follows from the well known results on operators satisfying the maximum principle ( see @xcite , @xcite and @xcite ) and the results on the principle eigenvalue of positive operators ( see @xcite , @xcite and @xcite ) and of its smooth dependence on parameters which can be derived from the general perturbation theory of linear operators ( see @xcite ) .
now we obtain from ( [ 2.2.18 ] ) that for @xmath2736 uniformly in @xmath2737 @xmath2522 and @xmath832 , @xmath2738 where @xmath2739 as @xmath2740 .
next , we want to compare @xmath2741 in order to do this we introduce auxiliary random evolutions @xmath2742 and @xmath2743 governed by the stochastic differential equations @xmath2744 and @xmath2745 respectively , where @xmath2746 , @xmath2747 and for @xmath2748 , @xmath2749 according to @xcite ( which relies on theorem 2 in 6 , ch .
vii of @xcite ) the distributions in the path space of the processes @xmath2750 and @xmath2751 are absolutely continuous with respect to the distributions in the path space of the processes @xmath2752 and @xmath2753 , respectively , with the densities @xmath2754 and @xmath2755 respectively , where @xmath2756 @xmath2757 and @xmath2758 .
thus , we have to compare @xmath2759 and @xmath2760 observe that by ( [ 2.2.1 ] ) , @xmath2761 let @xmath2762 be both an upper bound for @xmath2763 and their lipschitz constant then we see from ( [ 2.2.1 ] ) and ( [ 2.3+.6])([2.3+.10 ] ) that @xmath2764 employing the witney theorem embed smoothly @xmath1052 as a compact submanifold in an euclidean space @xmath2765 of a sufficiently high dimension @xmath601 and extend the operator @xmath2734 from @xmath1052 to @xmath16 so that its coefficients remain @xmath13 and they vanish outside a relatively compact set containing @xmath1052 ( cf .
now we can view ( [ 2.3+.2 ] ) and ( [ 2.3+.3 ] ) as stochastic differential equations in @xmath16 keeping the same notations for their coefficients and processes there . then using standard martingale moment estimates for stochastic integrals ( see , for instance , @xcite ) together with ( [ 2.3+.10 ] ) and the lipschitz continuity of coefficients in ( [ 2.3+.2 ] ) and ( [ 2.3+.3 ] ) we obtain @xmath2766 for some @xmath595 independent of @xmath2767 and @xmath33 .
hence , by the gronwall inequality @xmath2768 observe also that the distribution of @xmath2769 can be written explicitly as ( see 55 in @xcite ) , @xmath2770 in order to estimate the last expression in the right hand side of ( [ 2.3+.11 ] ) we note that for any random variable @xmath552 , @xmath2771 and so by the cauchy schwarz and the chebyshev s inequalities @xmath2772 now by ( [ 2.3+.11])([2.3+.14 ] ) together with the cauchy
schwarz inequality we obtain that for @xmath2736 uniformly in @xmath2538 , @xmath2522 and @xmath832 , @xmath2773 for another constant @xmath777 independent of @xmath2774 and @xmath33 .
choose @xmath2775 and set @xmath2776 $ ] . by ( [ 2.2.21 ] ) ,
@xmath2777 if @xmath2778 and @xmath2779 then by ( [ 2.3+.1 ] ) , ( [ 2.3+.15 ] ) and the lipschitz continuity of @xmath2780 we obtain that @xmath2781 for some constant @xmath2782 independent of @xmath2783 , @xmath2522 , @xmath2784 and @xmath33 .
observe that by the markov property , @xmath2785 now by ( [ 2.3+.10 ] ) and ( [ 2.3+.17 ] ) applying ( [ 2.3+.18 ] ) for @xmath2786 we obtain that @xmath2787 for some @xmath2788 independent of @xmath2538 , @xmath2789 , @xmath2790 and @xmath33 , which yields ( [ 2.2.3 ] ) completing the proof of proposition [ prop2.2.4 ] . | the work treats dynamical systems given by ordinary differential equations in the form @xmath0 where fast motions @xmath1 depend on the slow motion @xmath2 ( coupled with it ) and they are either given by another differential equation @xmath3 or perturbations of an appropriate parametric family of markov processes with freezed slow variables . in the first case
we assume that the fast motions are hyperbolic for each freezed slow variable and in the second case we deal with markov processes such as random evolutions which are combinations of diffusions and continuous time markov chains .
first , we study large deviations of the slow motion @xmath2 from its averaged ( in fast variables @xmath1 ) approximation @xmath4 the upper large deviation bound justifies the averaging approximation on the time scale of order @xmath5 , called the averaging principle , in the sense of convergence in measure ( in the first case ) or in probability ( in the second case ) but our real goal is to obtain both the upper and the lower large deviations bounds which together with some markov property type arguments ( in the first case ) or with the real markov property ( in the second case ) enable us to study ( adiabatic ) behavior of the slow motion on the much longer exponential in @xmath5 time scale , in particular , to describe its fluctuations in a vicinity of an attractor of the averaged motion and its rare ( adiabatic ) transitions between neighborhoods of such attractors . when the fast motion @xmath1 does not depend on the slow one we arrive at a simpler averaging setup studied in numerous papers but the above fully coupled case , which better describes real phenomena ,
leads to much more complicated problems
. 0.1 cm [ part1 ] |
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.